CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Centro di Studio e Ricerca
sulla Sanità Pubblica
A scientific alliance for
Carry out a Repository for Administrative and Clinical data Knotting
Evaluating management and building evidence from real world health data
Version: November 21, 2013
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 1
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Table of contents
1. Preliminary remarks ………………………………………………
Page
9
2. Investigating real world clinical practice…………….………..
11
2.1. Insufficiency of clinical trials ……………….............................
11
2.2. From clinical trials to observational investigations ……….....
11
2.2.1. The baseline observational design ………………………..
12
2.2.2. The comparative effectiveness principle …………………..
15
2.2.3. Potentiality of observational approach …………………….
16
2.3. Using electronic archives ………………………………………
17
2.4. Warranting good research practice ….……………………….
18
3. Rationale …………………………………...…………………………
23
4. Aims and articulation …………………………………...…………
29
4.1. Strategic target …………………………………………………
29
4.2. General methodological purpose ……………….......................
29
4.3. Clinical and public health purposes …………………………..
30
4.4. Health econometrics and health demand …………………….
4.5. Articulation …………………………………………………….
30
5. Common data framework and repository functioning
(Working Package 1) ……………………………...………………….
33
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 2
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6. Clinical issues (Working Package 2) …………………………..
39
6.1. Cardiovascular disease ….…...………………..….…..……......
39
Project CRACKCV1. Measuring burden of heart failure: prevalence,
incidence and therapeutic approaches …………………….………...............
39
Project CRACKCV2. Measuring the economic burden of acute myocardial
infarction …………………………………………………………………….
44
Project CRACKCV3. Assessing the gap between trials and practice: the
example of hypertensive and lipid lowering drug therapies ……...................
48
Project CRACKCV4. Estimating cost-effectiveness of time-varying drug
therapy using healthcare administrative databases. The case of statin in
secondary prevention ………………………………………………………..
54
Project CRACKCV5. Exploring the impact of prescribing and substituting
generic drugs: the example of cardiovascular therapies …………………….
58
Project CRACKCV6. Utilization, outcomes and costs of implantable
cardioverter defibrillator (ICD) ……………………………………………..
64
Project CRACKCV7. Electrocardiographic waveform analysis to predict the
success of defibrillation in human victims of out-of hospital cardiac arrest
and the association with in hospital outcomes and survival ………………...
68
6.2. Diabetes ….…………...…….……………..….…………………
75
Project CRACKDB1. Assessing adherence, long-term safety and costeffectiveness profiles of drug therapies for type 2 diabetes in clinical
practice ……………………………………………………………………...
74
Project CRACKDB2. Measuring the healthcare burden of diabetes mellitus ..
81
Project CRACKDB3. Assessing the gap between guidelines and practice: the
example of gestational diabetes ……………………………………………..
86
6.3. Oncology …………..……………...…………….…….………...
91
Project CRACKk1. Clinical use, safety and effectiveness of novel high cost
anticancer therapies after marketing approval: a record linkage study ……..
91
4.4. Respiratory diseases ………........…………………………...…
95
Project CRACKRD1. Measuring burden of chronic obstructive pulmonary
disease: prevalence, incidence and therapeutic approaches ………………...
95
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 3
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKRD2. The clinical and economic burden of patients
hospitalized for acute exacerbations of chronic obstructive pulmonary
disease ……………………………………………………………………….
98
Project CRACKRD3. Measuring the healthcare burden associated with
individuals hospitalized for pneumonia ……………………………………..
103
6.5. Gastroenterointestinal diseases ………........………………….
106
Project CRACKGID1. Measuring the burden of ulcerative colitis and Crohn’s
disease: incidence and costs ……………………………………….
6.6. Mental health ………………...…………...……………………
106
111
Project CRACKMH1. Adherence, effectiveness and cost-effectiveness
profiles of care journeys experienced from patients affected by severe
mental disturbance …………………………………………………………..
111
6.7. Geriatrics ……………...……………………..…………………
117
Project CRACKGH1 Drug treatment of elderly patients affected by
cardiovascular disease and other chronic comorbidities ……………………
117
6.8. Paediatrics …………...…...………………………….…………
121
Project CRACKPD1 Hospitalization for pneumonia and empyema: incidence
and association with pneumococcal conjugate vaccines and non-steroidal
antiinflammatory drugs ……………………………………………
6.9. Environmental health ……………..……...……………………
121
127
Project CRACKEH1. Epidemiology of idiopathic pulmonary fibrosis (IPF) in
Lombardy region and its relationship to air pollution ………………………
127
Project CRACKEH2. Cardioresopiratory diseases and environmental
exposure: particulate matter toxicity and molecular risk markers (TOSCA) .
132
Project CRACKEH3. Short-term effect of the exposure to particulate matter
on hospitalizations and pharmaceutical prescriptions in Lombardy ………..
138
Project CRACKEH4. Exposure to particulate matter: short-term effect on
stroke hospitalizations in Milan using ARPA monitoring stations …………
145
6.10. Drug safety …………...……...……………..…………………
150
Project CRACKDS2. MEREaFAPS project: monitoring of events and
adverse drug reactions in emergency department. Evaluation of preventable
150
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 4
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
reactions and costs of Adverse Drug Reactions …………………………….
Project CRACKDS2. Gastrointestinal and cardiovascular safety profiles of
non-steroidal anti-inflammatory drugs (NSAIDs) …......................................
155
6.11. Other issues (miscellaneous) ……………………..…...……...
160
Project CRACKOI1. Estimating prevalence and incidence rates of
rheumatoid arthritis from regional healthcare databases ………………........
160
Project CRACKOI2. Risk/benefit profile of bisphosphonates therapy in
primary/secondary prevention of osteoporotic fractures ……………………
164
7. Health econometrics and health demand (Working
Package 3) …………………………...…………………………………..
169
Project CRACKHE1 CRISP …………………………………………..….......
169
Project CRACKHEP2. Administrative databases as a tool for identifying
healthcare demand and costs in an over-one million population …………...
174
8. Educational issues (Working Package 4) ……………………..
179
9. Scientific board and accredited laboratories ………………...
181
9.1. Scientific board…………………………………………………
181
9.2. Project managers ………………………………………………
182
9.3. Laboratories ……………………………………………………
185
APPENDIX 1. Main characteristics of available healthcare utilization
database covering the entire population of Lombardy Region …………….
187
Health register …………………………………………………………..
188
Hospital discharge database …………………………………………….
189
Emergency room database ………………………………………………
190
Outpatient drug prescriptions database …………………………………
191
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 5
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
File F …………………………………………………………………….
192
Exemptions database ……………………………………………………
193
Outpatients specialistic services database ………………………………
194
Data warehouse DENALI ………………………………………..
195
Mental health services (PSICHE) database ……………………………..
197
Delivery assistance forms database ……………………………………..
198
Vaccination register …………………………………………………..…
200
Pharmacovigilance Database ……………………………………………
201
APPENDIX 2. Methods for controlling misclassification and confounding
203
A2.1. Measurement errors and misclassification ……………………….
204
A2.1.1. Outcome misclassification …………………………………………
204
A2.1.2. Exposure misclassification ………………………………………...
205
A2.1.3. Strategies of accounting for misclassification ……………………..
207
A2.1.3.1. Algebraic methods ………………………………………..
207
A2.1.3.2. Sensitivity analyses ……………………………………….
207
A2.1.3.3. External adjustment ………………………………………
207
A2.2. Confounding and beyond ………………………………………...
209
A2.2.1. Sources of confounding ……………………………………………
210
A2.2.2. Strategies of accounting for confounding: a general guide ………..
210
A2.2.3. Accounting for confounders through study design ………………..
211
A2.2.3.1. Restricting the study cohort ………………………………
211
A2.2.3.2. Matching ………………………………………………….
212
A2.2.3.3. Case-only designs ………………………………………...
213
A2.2.4. Accounting for confounding through data analysis ………………..
214
A2.2.4.1. Stratification and regression modelling …………………..
215
A2.2.4.2. Using proxy measures …………………………………….
215
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 6
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
A2.2.4.3. Sensitivity analysis ……………………………………….
216
A2.2.4.4. Instrumental variable estimation ………………………….
219
A2.2.5. Beyond confounding ………………………………………………
220
A2.2.5.1. Yet on the confounding definition ………………………..
220
A2.2.5.2. Intermediate variables and overadjustment ………………
221
A2.2.5.3. Collider variables …………………………………………
222
A2.References ……………………………………………………………...
223
APPENDIX 3. Curricula vitae and recent pubblications concerning the
CRACK issues of scientific board members ………………………….........
231
Giovanni Corrao ………………………………………………………...
232
Alberico Catapano ……………………………………………………...
234
Giancarlo Cesana ………………………………………………………..
236
Carlo La Vecchia …………………………………………………..........
238
Giorgio Vittadini
241
Giuseppe Mancia ………………………………………………………..
243
APPENDIX 4. Staff members, potentiality and main characteristics of
accredited LABORATORIES affering to the CRACK program …………..
245
PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH ..
246
EPIDEMIOLOGY AND PUBLIC HEALTH REASERCH ……………
247
EPIDEMIOLOGY ………………………………………………………
249
HEALTH ECONOMETRICS AND HEALTH DEMAND …………….
250
PHARMACOVIGILANCE AND DRUG UTILISATION RESEARCH
252
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 7
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
1. Preliminary remarks
The objective of the CRACK program is of implementing a repository through the
integration of different data sources for addressing questions still open in the setting of
diseases relevant for clinics and public health.
We would like to emphasize that the repository does not imply the transferring the data
from the place where they are stored from their owners to a unique warehouse. Rather, the
CRACK program is a proposal for data drawing from several sources according to a
specific protocol approved by health authority and the scientific board. The questions we
would like to answer through the use of the repository are related to diseases’ frequency
and prognosis (prevalence, incidence, survival), healthcare utilization patterns and safety,
effectiveness and cost-effectiveness profiles in ‘real world’ clinical practice.
Both, general methodological issues and specific clinical and public health questions will
be investigated for increasing the value of the health data by exploiting them from the
decision maker point of view. In other words, the main strategic aim of the CRACK
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 8
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
program is of realising a flywheel which allows to health authority of obtaining solid data
and strong evidence for addressing health policies.
The use of the term “flywheel” is justified by the fact that both health authority and
research teams would benefit from the CRACK program. Health authority because a “nocost” instrument for addressing health policies should be obtained. The research teams
because of availability of good-quality health data for scientific purposes. Finally, it is
expected that the CRACK program boosts methodological and applicative research in the
field of observational research (OR). Accordingly, training courses to will be carried out
within CRACK program.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 9
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
2. Investigating real world clinical practice
2.1. Insufficiency of clinical trials
Most of the efficacy and safety data on drugs, and more in general on healthcare services,
is generated from randomized, clinical trials (RCTs). However, high quality scientific
evidence from RCTs may not be generalisable to everyone likely to take the drug, or use
the service [1]. RCTs, in fact, usually have a small sample size that often under-represents
vulnerable patients and they focus on short-term efficacy and safety in a controlled
environment that is often far from routine clinical practice. Moreover, the RCT outcome
sufficient to win marketing approval often fails to answer the more relevant questions are
faced by patients, doctors and public health officiers [2]. Finally, patients enrolled in RCTs
achieve very high and almost optimal compliance, while in the clinical practice compliance
is really impaired [3-10]. Such limitations make it inevitable that epidemiologic
observational research is performed post marketing to define these issues [11].
2.2. From clinical trials to observational investigations
Health insurers, physicians, and patients worldwide need information on the comparative
effectiveness and safety of prescription drugs in routine care [12]. Although randomized
clinical trials (RCTs) are the gold standard to determine a drug’s efficacy against placebo,
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 10
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
it is well recognized that results of such studies may not accurately reflect effectiveness of
therapies delivered in typical practice [13–15]. In addition, clinical decisions usually
involve choices among therapies yet sponsors of drug trials have limited motivation to test
new drugs against existing therapies [16]. This implies that large nonexperimental
observational studies supply data, information and evidence completing our knowledge
based on RCTs.
Nevertheless, controversy sometimes surrounds observational studies [17]. This implies
that methodologic issues concerning study’s design, conduction and interpretation need of
great attention in the epidemiologic framework.
2.2.1. The baseline observational design
A preliminary description of the baseline design in observational epidemiologic
framework, the so called population-based cohort design, will be made in this paragraph.
A cohort is defined by subjects meeting a set of eligibility criteria and by entry and exit
time points. Consider, as hypothetical example, a cohort investigation for studying issues
concerning antidiabetic therapies. Entry into the cohort may be defined by calendar time
(spanned by the study, e.g., any time after January 1, 2004), by age (any age before 40th
birthday), by events (the first use of oral hypoglycaemic medication), or by disease status
(the date of diagnosis of type 2 diabetes). Exit from the cohort may be defined by the first
occurrence of specific calendar time (e.g., December 31, 2010), age (exit at 80th birthday),
events (death; exit from the study; the first switching from oral hypoglycaemic therapy to
insulin), or disease status (first occurrence of coronary heart disease).
The cohort of incident users of oral hypoglycaemic drugs may be illustrated graphically as
in the figure. This figure, based on 6 subjects, is plotted in terms of calendar time, with
subjects ranked according to their date of entry into the cohort, which corresponds to the
first prescription of the considered drug class (e.g. incident users of oral hypoglycaemic
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 11
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
drugs). The restriction to new initiators
of the study drugs (inception cohort)
will mitigate those issues and will also
ensure that patient characteristics are
assessed before the start of the study
drug and can therefore not be the
consequence of the drug, similar to the
principle of RCTs. The advantage of
the so-called new user design has been
summarized [18]. It is important to
stress that the included six incident
users potentially are all the individuals
belonging to the target population who
started therapy during the observational period. This is a first peculiarity
of observational studies with respect to
RCTs. These last, in fact, often select patients from clinical centres of excellence, exclude
patients who are more vulnerable to adverse effects of therapy and those affected by
comorbidities or submitted to co -treatments in the absence, however, of a target population
from which incident users arise. This means that population-based cohort studies are
virtually free from external selection bias (lack of generalizability) and, hence, adequately
describe real data generated from unselected target populations.
Cohort members of incident drug users are followed for recording two families of data. The
first one concerns drug exposure. The figure depicts a strong heterogeneity of drug
exposure for both type (e.g. two classes of oral hypoglycaemic agents are represented in the
figure) and duration (e.g. sporadic, intermittent and continuous exposures are also
represented in the figure). This is the second substantial peculiarity of observational studies
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 12
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
with respect to RCTs. These last, in fact, are based on the minimization of exposure
heterogeneity. Conversely, one main characteristic of observational studies is that they are
aimed to describe heterogeneity of drug exposure observed in real world clinical practice,
including heterogeneity in the compliance to treatment and deviations of guideline-based
clinical recomandations, and identifying components of heterogeneity affecting the
outcome.
The second family of data recorded during follow-up is the outcome onset. Outcome may
be the disease that would be avoided or postponed by the therapy (e.g. switching to insulin
as proxy of disease worsening or macrovascular events avoided for effect a given treatment
regimen) as well as adverse events potentially linked with brief- or long-term drug therapy
(e.g. cancer). This is the third substantial peculiarity of observational studies with respect to
RCTs. These last, in fact, are often characterized by sample size and study duration that do
not allow of investigating rare outcomes and long-term effects of exposure. Conversely,
large population followed for several years from exposure starting are usually submitted to
observational investigation in case of existing electronic databases covering the target
population (cfr. par. 2.3).
Besides this reference design, other ways for observing a given population has been widely
used for epidemiologic porpuses. Among these, the nested case-control design, a direct
derivation of the cohort one, has received great attention owing its higher computational
efficiency with respect to the cohort design [19]. A complete review of observational
designs proposed by the methodological and applicative literature, however, lies outside the
objective of this report. The very clear review by Suissa [20] is a sutable introductive
reader on this issue.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 13
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
2.2.2. The comparative effectiveness principle
There is a last substantial difference between observational studies and RCTs which we are
choosing to emphatize in a separate paragraph owing its general implications.
As pointed out by Cochran about 40 years ago [21], RCTs on the efficacy of drugs for their
regulatory approval study the extent to which an intervention does more good than harm
under ideal circumstances (“Can it
work?”). For most conditions, however, physicians have a choice of
two or more medications that can
prevent, cure, avoid progression of,
and reduce suffering from diseases.
For physicians, it is therefore not a
question of whether to prescribe a
drug [22] but which drug of several
alternatives [23]. In such situations,
physicians need to understand their
comparative effectiveness and safety. In fact, effectiveness assesses whether an intervention does more good than harm when provided under usual circumstances of health-care
practice (“Does it work in practice?”).
Hence, in the absence of enough head-to-head effectiveness trials, Comparative
Effectiveness Research (CER) tries to solve the issue of limited generalizability to routine
care and the lack of an active comparison group by studying post-marketing drug use data,
often from large health-care utilization databases, and associate such use with relevant
health outcomes [16].
It is important to be explicit about the definition of comparative effectiveness as it is
applied in this report. With respect to the term comparative, this report will focus on the
majority of circumstances when comparison can be made between two or more active
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 14
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
treatments rather than comparisons made between an active treatment and “no treatment”
[12]. Consistently, it should be observed that the reference observational design above
described is conceived for head-to-head comparison (rather than for active comparison)
since it aims of evaluating whether difference in healthcare exposure (e.g. heterogeneity in
oral hypoglycaemic agent prescribing or in compliance with the prescribed therapy), affects
the outcome. With respect to the term effectiveness, this report will focus on the benefits of
therapies, other than harms (as extensively examined in the field of pharmacoepidemiology) and costs (as extensively examined in pharmacoeconomics and health
services research).
Although large pharmacoepidemiologic studies have the advantage of being representative
of routine care, they suffer from several methodological issues discussed below (cfr. par 3).
2.2.3. Potentiality of observational approach
Observational studies are suitable for studying several aspects of the impact of healthcare
in routine clinical practice. Two items have been considered while the reference
observational design was described (cfr. par. 2.2.1): exposure course (to drugs, or more in
general to healthcare service) and outcome onset (related to hypothesized effectiveness or
safety of therapy). Because of both exposure course and outcome onset generate costs for
the National Health System (NHS), at least three objectives may be identified by this
process:
(i)
profile of pharmacoutilization, or more in general of healthcare utilization,
including the number of current (prevalent) users, new (incident) users, duration
of use, persistence and adherence, to name a few [24];
(ii)
risk-effectiveness profile of a given therapy, for investigating on both safety
and effectiveness (that is efficacy in routine clinical practice) measuring they on
comparable scale [1];
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 15
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
(iii)
cost-effectiveness profile of a given therapeutic course, for investigating the
additional cost that would be accrued to avoid one outcome event as a
consequence of implementing a given therapeutic intervention in the target
population [25].
2.3. Using electronic archives
Routinely collected and electronically stored information on healthcare utilization in
everyday clinical practice has proliferated over the past several decades. Large
computerized databases with millions of observations of the use of drugs, biologics,
devices, and procedures along with health outcomes may be useful in assessing which
treatments are most effective and safe in routine care without long delays and the
prohibitive costs of most RCTs [26].
Databases collecting health information can be classified into two broad categories: those
that collect information for administrative purposes, such as filling claims for payment
(denoted as administrative or healthcare utilization (HCU) databases), and those that serve
as the patient’s medical record and are therefore a primary mean by which physicians track
health information on their patients (denoted as medical record (MR) databases). A major
advantage of HCU data is that they reflect real world clinical practice for large and
unselected populations [27]. Nevertheless, studies based on HCU data have been criticized
for the incompleteness of the patients’ information such as markers of clinical disease
severity, lifestyle habits, and socio-economic status, among others. In contrast, although
MR data are richer of clinical and lifestyle information, they often suffer from the fact that
any given practitioner provides only a piece of the care which a patient receives, and
specialist and hospital cares are unlikely to be recorded in a common MR database [28].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 16
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data quality issues, as well as the selection of general practitioners who carefully take care
to their patients, are other potential limitations of studies based on MR data.
2.4. Warranting good research practices
There is controversy on how to best design and analyze nonrandomized studies on
comparative treatment effects using databases, including HCU and MR databases, patient
registries, and other routinely collected health-care data. Challenges of conducting
epidemiologic and health services research studies from secondary data sources include
concerns about the adequacy of study design, the relevance of the population and
timeframe available for study, approaches to minimize confounding in the absence of
randomization, and the specificity of clinical outcome assessment. Such threats to validity
limit the usefulness of these studies and adoption of findings into policy and practice. With
proper research design and application of an array of traditional and newer analytic
approaches, such concerns can be addressed to improve our understanding of treatment
effects.
This report is based upon the assumption that, for optimizing the validity of findings from
observational studies designed to inform health-care policy decisions, researchers employ a
priori hypotheses in written protocol and data analysis plans before study implementation,
that they follow reporting standards that make transparent to readers if, why, and how their
analytic plans evolved, as well as provide a justification of the suitability of the database to
test their hypotheses [12].
Although we recognize that exploratory analyses and data mining of large datasets are
often used to generate hypotheses regarding the effectiveness and comparative
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 17
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
effectiveness of treatments, stricter criteria for the design and execution of studies as well
as transparency in their reporting are required to justify the conclusion that such findings
are robust enough to warrant changes in clinical practice or to influence policy decisions.
Thus, the objective of this report is to lay out good research practices for comparative
therapeutic effectiveness studies using secondary databases.
We do not seek to be complete in our discussion of analytic options, nor will we fully
explain all methods, but rather focus on the issues surrounding the most relevant designs
and analytic techniques for secondary databases, and describe several applications in the
public health issue.
References
[1] Dieppe P, Bartlett C, Davey P, et al. Balancing benefits and harms: the example of non-steroidal antiinflammatory drugs. Br Med J 2004;329:31–4
[2] Schneeweiss S, Avorn J. A review of uses of HEALTHCARE utilization databases for epidemiologic
research on therapeutics. J Clin Epidemiol 2005;58:323–37
[3] Fitz-Simon N, Bennett K, Feely J. A review of studies of adherence with antihypertensive drugs using
prescription databases. Therapeutics and Clinical Risk Management 2005;1:93-106
[4] Mazzaglia G, Mantovani L, Sturkenboom MC, et al. Patterns of persistence with antihypertensive
medications in newly diagnosed hypertensive patients in Italy: a retrospective cohort study in primary care. J
Hypertens 2005;23:2093-100
[5] Van Wijk BLG, Klungel OH, Heerdink ER, et al. Rate and determinants of 10-year persistence with antihypertensive drugs. J Hypertens 2005;23:2101-7
[6] Burke TA, Sturkenboom MC, Lu SE, et al. Discontinuation of hypertensive drugs among newly
diagnosed hypertensive patients in UK general practice. J Hypertens 2006;24:1193-200
[7] Elliott WJ, Plauschinat CA, Skrepnek GH, et al. Persistence, adherence, and risk of discontinuation
associated with commonly prescribed antihypertensive drug monotherapies. J Am Board Fam Med
2007;20:72-80
[8] Corrao G, Zambon A, Parodi A, et al. Discontinuation of and changes in drug therapy for hypertension
among newly treated patients: a population-based study in Italy. J Hypertens 2008;26:819-24
[9] Vrijens B, Vincze G, Kristanto P, et al. Adherence to prescribed antihypertensive drug treatments:
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 18
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
longitudinal study of electronically compiled dosing histories. Br Med J 2008;336:1114-7
[10] Perreault S, Dragomir A, Blais L, et al. Impact of adherence to statins on chronic heart failure in primary
prevention. Br J Clin Pharmacol 2008;66:706-16
[11] Black N. Why we need observational studies to evaluate the effectiveness of HEALTHCARE. Br Med J
1996;312:1215–8
[12] Berger ML, Mamdani M, Atkins D, et al. Good research practices for comparative effectiveness
research: defining, reporting and interpreting nonrandomized studies of treatment effects using secondary
data sources: The ISPOR Good Research Practices for Retrospective Database Analysis Task Force Report—
Part I. Value in Health 2002;12:1044-52
[13] Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy
of research designs. N Engl J Med 2000;342:1887–92
[14] Concato J. Observational versus experimental studies: what’s the evidence for a hierarchy? NeuroRx
2004;1:341–7
[15] Avorn J. In defense of pharmacoepidemiologic studies: embracing the yin and yang of drug research. N
Engl J Med 2007;357:2219–21
[16] Schneeweiss S. Developments in post-marketing comparative effectiveness research. Clin Pharmacol
Ther 2007;82:143–56
[17] Sørensen HT, Lash TL, Rothman KJ. Beyond randomized controlled trials: a critical comparison of trials
with nonrandomized studies. Hepatology 2006;44:1075-82
[18] Ray WA. Evaluating Medication Effects Outside of Clinical Trials: New-User Designs. Am J Epidemiol
2003;158:915–920
[19] Essebag V, Platt RW, Abrahamowicz M, et al. Comparison of nested case– control and survival analysis
methodologies for analysis of timedependent exposure. BMC Med Res Methodol 2005;5:5
[20] Suissa S. Novel Approaches to Pharmacoepidemiology Study Design and Statistical Analysis. In:
Pharmacoepidemiology (4th edn). Strom BL (ed). Wiley, New York, 2005;811-29
[21] Cochrane, A. Effectiveness and Efficiency: Random Reflection on Health Services. Nuffiled Provincial
Trust; London: 1972
[22] van Luijn JCF, Gribnau FWJ, Leufkens HGM. Availability of comparative trials for the assessment of
new medicines in the European Union at the moment of market authorization. Br J Clin Pharm 2007;63:159–
62
[23] Pisano, DJ.; Mantus, D. FDA Regulatory Affairs: A Guide for Prescription Drugs, Medical Devices, and
Biologics. CRC Press; Boca Rotan, FL: 2004
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 19
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[24] Hallas J, Gaist D, Bjerrum L. The waiting time distribution as a graphical approach to epidemiologic
measures of drug utilization. Epidemiology 1997;8:666–70
[25] Berger M, Teutsch S. Cost-effectiveness analysis: from science to application. Med Care
2005;43(Suppl.):S49–53
[26] Strom BL. Overview of automated databases in pharmacoepidemiology. In: Pharmacoepidemiology (4th
edn). Strom BL (ed). Wiley, New York, 2005;219-22
[27] Muhajarine N, Mustard C, Roos LL, et al. Comparison of survey and physician claims data for detecting
hypertension. J Clin Epidemiol 1997;50:711–8
[28] Schneeweiss S, Wang P. Association between SSRI use and hip fractures and the effect of residual
confounding bias in claims database studies. J Clin Psychopharmacol 2004;24:632-8
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 20
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 21
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
3.
Rationale
Our approach is based on the following points.
Several important research questions can be answered by means of HCU data covering
very large and well defined populations. For example, in our setting we can easily
recognize patients who start with drug therapy (e.g. for treatment of heart failure or type 2
diabetes) from HCU databases covering residents in (or more properly National Health
Service beneficiaries of) the Lombardy region (including about 10 million inhabitants). The
primary advantages of using these data are that they are comprehensive, cost efficient, and
free of the usual biases associated with survey methods such as recall bias, non-response,
and subject attrition [1].
As an exercise for evaluating appropriateness of HCU databases in answering specific
study questions, imagine now an ideal cohort study with prospectively collected studyspecific data items. This exercise will reveal that in HCU database many factors are not
assessed at all [2], measured factors may be misclassified or missing [3], and some patients
are less likely to show up in the databases than expected. This implies that coded
information needs to be understood and analyzed as a set of proxies that indirectly describe
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 22
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
the health status of patients through the lenses of healthcare providers operating under the
constraints of a specific healthcare system [4].
We firstly must consider that several levels of proxies are involved when using HCU data.
For example, the health status of a patient can be assessed through the dispensing of a drug
that was prescribed by a physician who had made a diagnosis in a patient who visited
his/her practice and complained about symptoms [4]. This chain of proxies is influenced by
issues of access to care, severity of the condition, diagnostic ability of the physician, which
are generally unknown from HCU database. This strongly jeopardizes the possibility of
interpreting conventional measures of disease frequency such as those concerning
prevalence and incidence. In most cases, however, we do not need measuring frequency,
but rather to know that on average an increasing number of medications used by a patient is
just as predictive for clinical outcomes.
Under this point of view, typical confounding problems in assessment drug-outcome
association are systematically introduced by using HCU data. For example, cardiac drugs
are preferably prescribed at, and/or are preferably taken by, patients with certain
characteristics (e.g. more severe symptoms, poorer prognosis, major access to care, etc…).
By omitting sources of selective prescribing, biased estimates of the drug-outcome
association are systematically generated [5].
Finally, unverifiable assumptions and gross approximations must to be necessarily made
when using HCU data. For example, assumption has to be made that the proportion of days
covered by a prescription corresponds to the proportion of days of drug use, which may not
be invariably the case. Moreover, because no data are available on the dose prescribed by
the physician, we must use some metric of average daily dose (e.g., defined daily dose [6]).
This necessarily generates typical misclassification problems in drug exposure assessment
[4].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 23
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Detailed medical records are becoming more and more available from increasing use of
computerization in medical care. Among structured electronic MR databases available in
Italy, we here remind those denoted Health Search and ULNet databases fed by
approximately 900 and 220 general practitioners (GPs) respectively. In addition, the Italian
MR archive specifically oriented at peadiatric population, the so called PEDIANET
database, may be be used for particular applications. Finally, also disease registries,
surveys, and systematic reviews of literature (meta-analyes) may be useful sources of
data/information. Each of these sources allows of answering some specific study questions
in a more suitable way than HCU data, or, in other cases, for integrating lacking
information from HCU database. According with the aim of the CRACK program,
however, methods for data integration are of main interest. In fact, we should reason upon
the virtual absence of selection bias from HCU data, so that such data should be considered
as the main source for firstly answering specific study questions. With the aim of
overcoming data incompleteness, however, external (secondary) sources, such as those
from the network of GPs attending the same populations as that covered by the HCU
database, should be used, if available.
For example, a combination of two drugs may be used as first step treatment strategy [7]. A
drug combination, however, is more likely prescribed to high risk patients and to those with
poor prognosis. By omitting sources of selective prescribing (e.g. because these data are
not available from the HCU database used for investigating this issue) biased estimates of
the drug-outcome association are systematically generated. This problem can be
investigated by sensitivity analysis [8]. The basic concept of sensitivity analyses is to make
informed assumptions about potential confounders and to quantify their effects on the
observed drug-outcome association. If additional data sources can be identified (e.g. from
MR database), these assumptions can be substituted by empirical estimates and then they
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 24
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
may be used for external adjustment of the drug-outcome association [9-12]. Another issue
where the application of data integration would be useful is that of correcting for exposure
misclassification [13-15].
All these considerations taken together suggest that the approaches based on HCU and MR
data should not be considered mutually exclusive or even competing, but rather should play
complementary roles. Accordingly, the rationale underlying the current program is to carry
out an integrated approach for studying management of several relevant conditions/diseases
in real world clinical practice in Italy.
References
[1] Strom BL. Overview of automated databases in pharmacoepidemiology. In: Pharmacoepidemiology (4th
edn). Strom BL (ed). Wiley, New York, 2005;219-22
[2] Schneeweiss S, Wang P. Association between SSRI use and hip fractures and the effect of residual
confounding bias in claims database studies. J Clin Psychopharmacol 2004;24:632-8
[3] Wilchesky M, Tamblyn RM, Huang A. Validation of diagnostic codes within medical services claims. J
Clin Epidemiol 2004;57:131-41
[4] Schneeweiss S, Avorn J. A review of uses of HEALTHCARE utilization databases for epidemiologic
research on therapeutics. J Clin Epidemiol 2005;58:323–37
[5] MacMahon S, Collins R. Reliable assessment of the effects of treatment on mortality and major
morbidity. II. Observational studies. Lancet 2001;357:455-62
[6] WHO Collaborating Centre for Drug Statistics Methodology. ATC index with DDD. Oslo, Norway:
WHO;2003
[7] Mancia G, Laurent S, Agabiti-Rosei E, et al. Reappraisal of European guidelines on hypertension
management: a European Society of Hypertension Task Force document. J Hypertens 2009:27:2121-58
[8] Greenland S. Basic methods for sensitivity analyses of biases. Int J Epidemiol 1996;25:1107-16
[9] Kriebel D, Zeka A, Eisen EA, et al. Quantitative evaluation of the effects of uncontrolled confounding by
alcohol and tobacco in occupational cancer studies. Int J Epidemiol 2004;33:1040-5
[10] Schneeweiss S. Sensitivity analysis and external adjustment for unmeasured confounders in
epidemiologic database studies of therapeutics. Pharmacoepidemiol DS 2006;15:291-303
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 25
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[11] Corrao G, Parodi A, Nicotra F, et al. Cardiovascular protection by initial and subsequent combination of
antihypertensive drugs in daily life practice. Hypertension 2011;58:566-72
[12] Corrao G, Nicotra F, Parodi A, et al. External adjustment for unmeasured confounders improved drugoutcome association estimates based on HEALTHCARE utilization data. J Clin Epidemiol 2012;65:1190-9
[13] Cole SR, Chu H, Greenland S. Multiple-imputation for measurement-error correction. Int J Epidemiol
2006;35:1074-81
[14] Messer K, Natarajan L, et al. Maximum likelihood, multiple imputation and regression calibration for
measurement error adjustment. Stat Med 2008;27:6332-50
[15] Kipnis V, Midthune D, Freedman LS, et al. Regression calibration with more surrogates than
mismeasured variables. Stat Med 2012;31:2713-32
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 26
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 27
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
4. Aims and articulation
4.1. Strategic target
The main strategic aim of the CRACK program is of realising a flywheel which allows to
the regional authorities of obtaining solid data and strong evidence for addressing health
policies. It is also expected that the CRACK program boosts methodological and
applicative research in the field of observational research. Accordingly, training courses to
will be carried out within the CRACK program.
4.2. General methodological purpose
The main methodological aim of the CRACK program is to carry out a repository of
administrative (HCU) and clinical (MR) electronic archives available in Italy (more
specifically in the Lombardy Region) reporting data of patients affected by selected
diseases. The repository will be useful for recognizing the sources of systematic uncertainty
(mainly misclassification and confounding) when HCU and MR data are used separately,
and implementing methods for controlling and/or minimizing the effect of such biases by
means of integrating HCU and MR data.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 28
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
4.3. Clinical and public health purposes
As practical fields of using the electronic repositories we will select several
conditions/diseases particularly relevant for clinicians and public health officers such as
heart failure, hypertension, dyslipidaemia, type 2 diabetes, cancer, chronic obstructive
pulmonary, psychiatric, geriatric and pediatric diseases, among others.
The CRACK program will investigate management of all (and separately for) these
clinical issues in real world clinical practice through of: (i) assessing the care journeys of
patients who start drug therapy for a given condition/disease (including heterogeneity in
therapeutic profiles and refill compliance with drug therapies) according to their sociodemographic and clinical features; (ii) estimating the association between care journeys of
patients who start drug therapy for a given condition/disease and their risk of experiencing
selected clinical outcomes; (iii) providing for cost-effectiveness estimates of care journeys.
4.4. Health econometrics and health demand
A second application field will concern of using the electronic repositories for providing
comprehensive analyses of demand/supply profiles and measuring the level of the
healthcare quality and expenditure.
4.5. Articulation
Some details on common data framework and repository functioning (Working Package 1;
cfr. par 5), using the CRACK structure for elucidating management of selected
condition/disease in real world clinical practice (Working Package 2; cfr. par 6), and for
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 29
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
investigating issues concerning health econometrics and health demand (Working Package
3; cfr. par 7), and educational projects concerning observational studies using electronic
database (Working Package 4; cfr. par 8), will be supplied in the following paragraphs.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 30
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 31
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
5. Methodological issues (Working Package 1)
Records of all subjects receiving at least one prescription for therapeutics commonly used
to treat the considered conditions/diseases will be identified from the prescription drug
database of both outpatients and inpatients (HCU database) as well as from available MR
databases. When feasible, two time windows will be considered including more recent
periods (e.g. 2008-2010 so of studying relatively news data), as well as less recent periods
(e.g. 2005-2007 so to have sufficient data for studying long-term course of included
patients).
For patients who will be identified from a HCU database, data on drug prescriptions,
hospitalizations, specialist and laboratory benefits and disease exemptions (and others
where appropriate, e.g. Mental Health, Emergency Rooms, Delivery Assistance,
Vaccination data, among others), will be extracted from the corresponding HCU database
from date of their first recorded prescription backward for at least the three-year period
preceding it and forward for the last date available. Some details on health database
available covering the entire population of Lombardy Region are reported in Appendix 1.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 32
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Deterministic and probabilistic record linkage procedures will be implemented with the aim of recognizing patients who are recorded within or between databases. In order to preserve
privacy, identification codes
will automatically converted to anonymous codes directly from
data owner
(Lombardy Region), and the inverse process will be prevented by deletion of the
conversion table. In this way, the accredited laboratories will receive anomimized
information that preserves any realistic possibilities of subject recognizing.
Patients will be also identified from available MR databases (ULNet and PEDIANET)
considering the same time window as that of HCU records. In addition, data from
population-based disease-registries and surveys will be used, when available, and
systematic reviews of the literature (meta-analyses), when suitable, for obtaining additional
information on the issue of interest.
Other than a simple data container, the repository will be conceived as a a database
allowing of: (i) collection and standardization information deriving from different,
heterogeneous sources; (ii) organization such an integrated database with an approach
protocol driven for extracting relevant fields, selecting records according to some
predefined inclusion or exclusion criteria and allowing the implementation of several
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 33
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
epidemiological designs; (iii) analyze data according to conventional ed emerging
statistical models.
In particular, statistical methods of adjusting for confounders measured (e.g. models based
on Propensity Score) and unmeasured (e.g. models based on Monte Carlo Sensitivity
Analysis, Instrumental Variables, and Propensity Score Calibration [1-11]), emerging
designs for confounding control (e.g. case-only studies, such as the case-crossover or selfcontrolled case-series analysis, or the case-time-control design, a variant for handling bias
from temporal trends in exposure [12-15]) as well as statistical methods of adjusting for
misclassification (i.e. regression calibration and SIMEX models) [2,16-18], will be
available
for
analyses.
Statistical
and
epidemiologic
methods
for
controlling
misclassification and confounding are reported in Appendix 2.
After a validation process, during which the database will be tested, the integrated data will
be made available for answering specific study questions (cfr. par 6, Working Package 2)
after protocol approval by the scientific board. Scientific board members were experienced
in medicine, epidemiology, biostatistics and public health, and were drawn from academia
(cfr. par. 6.1 and Appendix 3).
Staff members, potentiality and main characteristics of accredited laboratories affering to
the CRACK program are reported in par. 8 and Appendix 4. Conformity to:
•
Italian rules for treatment of sensible data [19], and
•
Good Research Practices for Designing and Analyzing Retrospective Databases
recommend by the International Society for Pharmacoeconomics and Outcomes
Research (ISPOR) [5,6,20]
are guaranteed by the accredited laboratories.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 34
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
References
[1] Essebag V, Platt RW, Abrahamowicz M, et al. Comparison of nested case– control and survival analysis
methodologies for analysis of timedependent exposure. BMC Med Res Methodol 2005;5:5
[2] Lohr KN. Emerging methods in comparative effectiveness and safety. Symposium overview and
summary. Medical Care 2007;45:S5-S8
[3] Brookhart MA, Wang P, Solomon DH, et al. Evaluating short-term drug effects using a physician-specific
prescribing preference as an instrumental variable. Epidemiology 2006;17: 268-75
[4] Sturmer T, Glynn RJ, Rothman KJ, et al. Adjustments for unmeasured confounders in
pharmacoepidemiologic database studies using external information. Medical Care 2007;45:S158-S165
[5] Cox E, Martin BC, Van Staa T, et al. Good research practices for comparative effectiveness research:
approaches to mitigate bias and confounding in the design of nonrandomized studies of treatment effects
using secondary data sources: The International Society for Pharmacoeconomics and Outcomes Research
Good Research. Practices for retrospective database analysis task force report—Part II. Value Health
2009;12:1053–61
[6] Johnson ML, Crown W, Martin BC, et al. Good research practices for comparative effectiveness research:
analytic methods to improve causal inference from nonrandomized studies of treatment effects using
secondary data sources. The International Society For Pharmacoeconomics and Outcomes Research Good
Research. Practices for retrospective database analysis task force report—Part III. Value Health
2009;12:1062–73
[7] Brookhart MA, Sturmer T, Glynn RJ, et al. Confounding control in healthcare database research.
Challenges and potential approaches. Medical Care 2010;48:S114-S120
[8] Schneeweiss S, Gagne JJ, Glynn RJ, et al. Assessing the comparative effectiveness of newly marketed
medications: methodological challenges and implications for drug development. Clin Pharmacol Ther
2011;90:777-90
[9] Sox HC, Goodman SN. The methods of comparative effectiveness research. Annu Rev Public Health
2012;33:425–45
[10] Bradbury BD, Gilbertson DT, Brookhart MA, et al. Confounding and control of confounding in
nonexperimental studies of medications in patients with CKD. Advances in Chronic Kidney Disease
2012;19:19-26
[11] Garbe E, S Kloss, Suling KM, et al. High-dimensional versus conventional propensity scores in a
comparative effectiveness study of coxibs and reduced upper gastrointestinal complications. Eur J Clin
Pharmacol DOI 10.1007/s00228-012-1334-2
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 35
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[12] Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute
events. Am J Epidemiol 1991;133:144–53
[13] Farrington CP. Control without separate controls: evaluation of vaccine safety using case-only methods.
Vaccine 2004;22:2064–70
[14] Lumley T, Levy D. Bias in the case - crossover design: implications for studies of air pollution.
Environmetrics 2000;11:689–704
[15] Suissa S. The case-time-control design. Epidemiology 1995;6:248–53
[16] Spiegelman D, Schneeweiss S, McDermott A. Measurement Error Correction for Logistic Regression
Models with an "Alloyed Gold Standard". Am J Epidemiol 1997;145:184–96
[17] Spiegelman D, McDermott A, Rosner B. Regression calibration method for correcting measurement
error bias in nutritional epidemiology. Am J Clin Nutr l997:65:1179S-86S
[18] Weatherley BD, Nelson JJ, Heiss G, et al. The association of the ankle-brachial index with incident
coronary heart disease: the Atherosclerosis Risk In Communities (ARIC) study, 1987–2001. BMC
Cardiovascular Disorders 2007,7:3 doi:10.1186/1471-2261-7-3
[19] Virone MG. EHR and data protection issues in Italy. Stud Health Technol Inform 2012;180:741-5
[20] Berger ML, Mamdani M, Atkins D, et al. Good research practices for comparative effectiveness
research: defining, reporting and interpreting nonrandomized studies of treatment effects using secondary
data sources: the ISPOR Good Research Practices for Retrospective Database Analysis Task Force Report—
Part Ivhe_600. Value Health 2009;12:1044–52
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 36
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 37
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6. Clinical issues (Working Package 2)
6.1. Cardiovascular disease
Project CRACKCV1
Measuring burden of heart failure: prevalence, incidence and therapeutic
approaches
Background
Heart failure (HF) has been defined as an abnormality of cardiac structure or function
leading to failure of the heart to deliver oxygen at a rate commensurate with the
requirements of the metabolizing tissues, despite normal filling pressures (or only at the
expense of increased filling pressures) [1]. On a clinical point of view, HF has been defined
as a syndrome in which patients have typical symptoms (e.g. breathlessness, ankle
swelling, and fatigue) and signs (e.g. elevated jugular venous pressure, pulmonary crackles,
and displaced apex beat) resulting from an abnormality of cardiac structure or function [2].
Approximately 1-2% of the adult population in developed countries has HF, with a
prevalence rising to ≥ 10% among persons 70 years of age or older [3]. However, the
overall prevalence of HF is increasing because of the ageing of the population, the success
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 38
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
in prolonging survival in patients suffering coronary events, and the success in postponing
coronary events by effective prevention in those at high risk or those who have already
survived a first event (secondary prevention). The outlook is, in general, gloomy, although
some patients can live for many years: overall 50% of patients are dead within 4 years.
Forty percent of patients admitted to a hospital with HF are dead or readmitted within 1
year [4]. The objective of the pharmacological treatment of patients affected by HF is to
reduce mortality, improve quality of life and prevent organ damage and hospitalizations.
The cornerstone of HF therapy is the association between Diuretics, ACE/ARBs and Betablockers to which can be added Aldosteron-antagonists, H-ISDN and Digoxin in case of
lack of control with first-step therapy.
Open questions
Although many RCTs and epidemiologic studies have already been conducted, there are
still some knowledge gaps regarding the treatment and the diagnosis of HF, for example: (i)
the prevalence estimates vary considerably between studies, owing to a lack of uniformity
in the definition and assessment of heart failure and the absence of a gold standard for heart
failure. In addition, non-cardiac conditions, although very frequent, (chronic obstructive
pulmonary disease, obesity or a poor physical condition) are not commonly quantified and
mimic heart failure making the identification more confused. [3]; (ii) the long-term safety,
mainly for patients suffering of comorbidities, are unknown although they have great
clinical and public health relevance; (iii) the treatment of acute heart failure remains largely
opinion-based with little good evidence to guide therapy; (iv) as far as pharmacological
treatments are concerned, current guidelines stress several uncertainty sources such as
those concerning efficacy and safety of digoxin (in modern era of pharmacological and
device therapy), hydralazine and ISDN (in non-black patients), renin inhibitors and dual
neprilysin/angiotensin (alternative or in addition to ACE inhibition), new oral
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 39
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
anticoagulants, clopidogrel and other novel antiplatelet agents (compared with aspirin in
patients in sinus rhythm).
Problems which will be faced
The following four main problems will be faced from the project.
Question 1. HCU database does not allow of recognizing all patients affected by HF and
supply hence lower prevalence and incidence data with respect to MR database. False
positive patients might also be identified from HCU databases. These last, moreover, lack
of clinical information for prognostic stratification of the included patients.
Objective. To verify (i) diagnostic accuracy of HCU data using MR data as gold
standard; (ii) the error which should be made by measuring prognosis by means of
HCU data.
What will be made? HCU and MR data comparison for identifying the best
diagnostic and prognostic models.
Question 2. Therapeutic persistence and adherence need to be accurately measured.
However, because of we do know how defined daily dose (DDD) accurately measure the
prescribed daily dose (PDD), potentially rough approximations are introduced from HCU
data.
Objective. To measure the error introduced by DDD using PDDR data as gold
standard.
What will be made? DDD and PDD comparison by regressing data from MR
archive.
Question 3. It is unknown whether HF patients are treated according to guidelines in the
current clinical practice.
Objective. To measure the deviation of clinical practice from guidelines
recommendations.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 40
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
What will be made? HCU and MR data comparison for identifying the best model
enforceable to HCU data for accurately measure deviations from guidelines
recommendations.
Question 4. The relationship between therapeutic journey (e.g. persistence and adherence
with and type of the prescribed therapy) and the onset of clinical outcomes (e.g. hospital
admission for myocardial infarction) is influenced by factors characterizing both the patient
and its physician. Many of these factors are not recorded in HCU database so that the
corresponding estimates are potentially biased.
Objective. To obtain information about the therapeutic approach of physicians who
operate in Lombardy Region according to clinical and lifestyle patients’ features.
What will be made? External adjustment of drug – outcome association derived
from
HCU data for patients’ features – type of drug prescribed relationships
obtained from MR data.
Management
Project manager (clinics): Giuseppe Mancia1 and Guido Grassi2
Project manager (primary care): Ovidio Brignoli and Alessandro
Filippi 3
Project manager (epidemiology and biostatistics): Giovanni Corrao4,5
Data management and analysis: Arianna Ghirardi and Giulia
Seagafredo5
1
2
Italian Institute for Auxology
Division of Internal Medicine, San Gerardo Hospital, Monza, Dept of Health Sciences, University of
Milano-Bicocca, Milan, Italy
Dept of health Sciences, University of Milano-Bicocca, Milan, Italy
3
Health Search, Italian College of General Practitioners, Florence (G.M., E.S.), Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 41
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
4
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
5
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] Dickstein K, Cohen-Solal A, Filippatos G, et al; ESC Committee for Practice Guidelines (CPG). ESC
guidelines for the diagnosis and treatment of acute and chronic heart failure 2008: the Task Force for the
diagnosis and treatment of acute and chronic heart failure 2008 of the European Society of Cardiology.
Developed in collaboration with the Heart Failure Association of the ESC (HFA) and endorsed by the
European Society of Intensive Care Medicine (ESICM). Eur J Heart Fail 2008;10:933-89
[2] McMurray JJ, Adamopoulos S, Anker SD, et al. ESC Guidelines for the diagnosis and treatment of acute
and chronic heart failure 2012: The Task Force for the Diagnosis and Treatment of Acute and Chronic Heart
Failure 2012 of the European Society of Cardiology. Developed in collaboration with the Heart Failure
Association (HFA) of the ESC. Eur J Heart Fail 2012;14:803-69
[3] Mosterd A, Hoes AW. Clinical epidemiology of heart failure. Heart 2007;93:1137-46
[4] LaRosa JC, He J, Vupputuri S. Effect of statins on risk of coronary disease. A Meta-analysis of
randomized controlled trials. J Am Med Ass 1999;282:2340-6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 42
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKCV2
Measuring the economic burden of acute myocardial infarction
Background
Cardiovascular disease is still the leading cause of death and among the major causes of
disease burden [1]. Its management requires a large resource consumption, mainly due to
its disability consequences [2, 3] in a progressively aging population and to the
introduction of new diagnostic and therapeutic technologies, which are more effective but
also more expensive. In 2003, the estimated annual total cost of coronary care in the
European Union was around €23 billion mainly determined by inpatient care (62%). In
Italy the estimated annual cost was about €3 billion [4]. Acute myocardial infarction (AMI)
in an important contributor of healthcare burden associated with coronary heart disease.
Two multinational studies estimated the direct cost of AMI hospitalization in different
countries, including Italy. One of the two studies included only 27 Italian cases and it
estimated an average cost of €5,988 for a hospitalization in the period 1999-2001. The cost
was lower than in the other investigated European and North-American countries [5]. The
other study included 1,788 Italian cases and it reported an average hospitalization cost of
€7,450 in 2005, which was more elevated than in the rest of Europe [6]. Things seem to
have changed a lot in very few years. Actually, the results of the two studies are not
comparable because of differences in sample selection. Moreover the two samples may not
be representative of prevalence and practice of care in the country. Therefore, any
consideration should be supported by more systematic and extensive analyses of Italian
costs.
Healthcare administrative databases can provide a complete source of real world data.
Epidemiological and economic evaluations with these databases are less expensive and
involve with continuity population with a large number of subjects. Several American
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 43
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
studies used insurance administrative claims to study the healthcare economic burden of
CVD [7-10].
Aims
The aim of the present work was to determine, using healthcare administrative databases,
the economic burden of hospitalized first AMI events registered in 2003 in Lombardy.
Methods
Data were extracted from healthcare administrative databases of the region of Lombardy,
organized in a data warehouse, called DENALI, to facilitate processing and analysis. The
study population included all subjects who had a first hospitalization for AMI during 2003
and 5 years of continuous eligibility in the area of Lombardy prior to the event. All HDs
with ICD-9-CM code 410.xx in at least one of six discharge diagnosis, excluding ICD-9CM code 410.x2 (subsequent episode of care within eight weeks of the acute event), were
identified. The first hospitalization during 2003 defined the index event. Subjects with an
AMI HD (ICD-9-CM code 410.xx) in the 5 years prior to the first hospitalization were
excluded. The population was followed from the date of admission of the index event until
the 31st December 2005 recording vital status, death or emigration outside the area of
Lombardy, and healthcare resource utilizations: total and average annual costs from the
perspective of the Italian National Health Service (NHS) were quantified using charges, i.e.
the amount of money the Italian NHS reimbursed to providers of care, and presented by
type of resource. The average cost per person per year was computed for each individual as
the ratio between charges and the length of individual observation period, expressed in
years.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 44
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Impact
The measurement of the epidemiological and economic burden of diseases could be a
useful contribution to assess the allocation of healthcare and research interventions and to
evaluate their potential costs and benefits. In particular, the contemporaneous evaluation of
the epidemiological and economical aspects of a disease could help to set priorities and to
maximize management strategies without sacrificing safety, efficacy or effectiveness of
care.
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Marco Ferrario2
Data management and analysis: Carla Fornari and Fabiana Madotto3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Epidemiology and Preventive Medicine Research Centre, Department of Experimental Medicine, University
of Insubria, Varese, Italy
3
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] Global status report on non-communicable diseases 2010. I. World Health Organization 2011. ISBN 978
92 4 068645 8 (PDF)
[2] Hodgson TA, Meiners MR. Cost-of-Illness Methodology: A Guide to Current Practices and Procedures.
Milbank Mem Fund Q Health Soc 1982;60:429-62.
[3] Rice DP. Cost of illness studies: what is good about them? Inj Prev 2000;6:177–9.
[4] Leal J, Luengo-Fernández R, Gray A, et al. Economic burden of cardiovascular diseases in the enlarged
European Union. Eur Heart J 2006;27:1610-9.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 45
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[5] Kauf TL, Velazquez EJ, Crosslin DR, et al. The cost of acute myocardial infarction in the new
millennium: evidence from a multinational registry. Am Heart J 2006;151:206-12.
[6] Tiemann O. Variations in hospitalisation costs for acute myocardial infarction - a comparison across
Europe. Health Econ 2008;17:S33-S45.
[7] Etemad LR. McCollam PL. Total First-year costs of acute coronary syndrome in a managed care setting. J
Manag Care Pharm 2005;11:300-6.
[8] Menzin J, Wygant G, Hauch O, et al. One-year costs of ischemic heart disease among patients with acute
coronary syndromes: findings from a multi-employer claims database. Curr Med Res Opin 2008;24:461-8.
[9] Turpie AG. Burden of disease: medical and economic impact of acute coronary syndromes. Am J Manag
Care 2006;12:S430-S434.
[11] Nichols GA, Bell TJ, Pedula KL, et al. Medical care costs among patients with established
cardiovascular disease. Am J Manag Care 2010;16:86-93.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 46
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKCV3
Assessing the gap between trials and practice: the example of
hypertensive and lipid lowering drug therapies
Background
Large randomized clinical trials (RCTs) have shown that antihypertensive and lipidlowering drugs reduce cardiovascular morbidity and mortality in patients with hypertension
and dyslipidaemia respectively [1,2], even in those without established cardiovascular
disease (CVD) [3]. However, large scale studies consistently showed that in real life poor
adherence with antihypertensive drug treatment, as well as with lipid lowering agents, are
associated with increased CV risk [4-17]. Evidence is available that poor compliance to
treatment is related to poor blood pressure and low-density lipoprotein cholesterol goal
attainment [18-21]. Furthermore, several studies have reported a relationship between
compliance and cardiovascular risk [22-40]. As a consequence, poor adherence to chronic
therapies represents a major clinical and public health issue which needs to be adequately
faced.
These findings also suggest that interventions aimed at enhancing adherence would be
effective in achieving the full benefits of medications [41]. On the other hand, interventions
of this type necessarily lead to a growth in the expenditure for supporting the cost of
incremented drug use. The impact of the expenditure is particularly important for a public
health perspective since the economic resources available are limited and it’s important to
allocate them in the best way to maximize the level of population health [42]. It would
therefore be suitable to jointly evaluate cost and effectiveness of treatment in order to
quantify the additional cost needed to increase the effectiveness of treatment-related
enhancing adherence. To the best of our knowledge, only few cost-effectiveness
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 47
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
investigations have been carried out concerning interventions enhancing adherence in the
setting of primary prevention of CV outcomes [43-45].
Aims
The project has the following aims: (i) evaluation of adherence and its temporal trend in the
period 2003-2010; (ii) analysis of the factors potentially responsible for poor adherence;
(iii) detection of the existing relationship between therapeutic adherence and morbidity and
mortality; (iv) evaluation of cost-effectiveness profile of interventions enhancing
adherence.
Methods
A cohort study will be conducted including all information on drug prescription and
clinical outcome evaluated during follow-up for the incident users of lipid-lowering drugs
using the electronic archives of Lombardy Region. As adherence determinants, treatment,
patients and general practitioner characteristics will be investigated. The outcomes of
interest will be major cardiovascular events and death for any cause recorded during
follow-up. The incidence of adverse events will be assessed based on the presence of
prescriptions of specific drugs and / or hospitalization for specific diseases. Finally, the cost
of a CV hospitalization avoided by interventions enhancing adherence will be estimated by
means of cost-effectiveness models [44,45].
Management
Project manager (clinics): Giuseppe Mancia1 and Guido Grassi2
Project manager (pharmacology): Alberico L Catapano3
Project manager (epidemiology and biostatistics): Giovanni Corrao4,5
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 48
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data management and analysis: Davide Soranna and Buthaina
Ibrahim5
1
2
Italian Institute for Auxology
Division of Internal Medicine, San Gerardo Hospital, Monza, Dept of Health Sciences, University of
Milano-Bicocca, Milan, Italy
3
Dept Pharmacological Sciences, and Centre for Pharmacoepidemiology and Pharmacoutilization, University
of Milano, Milan, Italy
4
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
5
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] Mancia G, Laurent S, Agabiti-Rosei E, et al. Reappraisal of European guidelines on hypertension
management: a European Society of Hypertension Task Force document. J Hypertens 2009:27:2121-58
[2] LaRosa JC, He J, Vupputuri S. Effect of statins on risk of coronary disease. A Meta-analysis of
randomized controlled trials. J Am Med Ass 1999;282:2340-6
[3] Brugts JJ, Yetgin T, Hoeks SE, et al. The benefits of statins in people without established cardiovascular
disease but with cardiovascular risk factors: meta-analysis of randomized controlled trials. Br Med J
2009;338:b2376
[4] Fitz-Simon N, Bennett K, Feely J. A review of studies of adherence with antihypertensive drugs using
prescription databases. Therapeutics and Clinical Risk Management 2005;1:93-106
[5] Mazzaglia G, Mantovani L, Sturkenboom MC, et al. Patterns of persistence with antihypertensive
medications in newly diagnosed hypertensive patients in Italy: a retrospective cohort study in primary care. J
Hypertens 2005;23:2093-100
[6] Van Wijk BLG, Klungel OH, Heerdink ER, et al. Rate and determinants of 10-year persistence with antihypertensive drugs. J Hypertens 2005;23:2101-7
[7] Burke TA, Sturkenboom MC, Lu SE, et al. Discontinuation of hypertensive drugs among newly
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 49
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
diagnosed hypertensive patients in UK general practice. J Hypertens 2006;24:1193-200
[8] Elliott WJ, Plauschinat CA, Skrepnek GH, et al. Persistence, adherence, and risk of discontinuation
associated with commonly prescribed antihypertensive drug monotherapies. J Am Board Fam Med
2007;20:72-80
[9] Corrao G, Zambon A, Parodi A, et al. Discontinuation of and changes in drug therapy for hypertension
among newly treated patients: a population-based study in Italy. J Hypertens 2008;26:819-24
[10] Vrijens B, Vincze G, Kristanto P, et al. Adherence to prescribed antihypertensive drug treatments:
longitudinal study of electronically compiled dosing histories. Br Med J 2008;336:1114-7
[11] Caro JJ, Speckman JL, Salas M, et al. Effect of initial drug choice on persistence with antihypertensive
therapy: the importance of actual practice data. CMAJ 1999;160:41-6
[12] Degli Esposti E, Sturani A, Di Martino M, et al. Long-term persistence with antihypertensive drugs in
new patients. J Hum Hypertens 2002;16:439-44
[13] Bourgault C, Sénécal M, Brisson M, et al. Persistence and discontinuation patterns of antihypertensive
therapy among newly treated patients: a population-based study. J Hum Hypertens 2005;19:607-13
[14] Kamal-Bahl SJ, Burke T, Watson D, et al. Discontinuation of lipid modifying drugs among
commercially insured United States patients in recent clinical practice. Am J Cardiol 2007;99:530-4
[15] Helin-Salmivaara A, Lavikainen P, Korhonen MJ, et al. Long-term persistence with statin therapy: A
nationwide register study in Finland. Clin Ther 2008;30:2228-40
[16] Poluzzi E, Strahinja P, Lanzoni M, et al. Adherence to statin therapy and patients' cardiovascular risk: a
pharmacoepidemiological study in Italy. Eur J Clin Pharmacol 2008;64:425-32
[17] Vinker S, Shani M, Baevsky T, et al. Adherence with statins over 8 years in a usual care setting. Am J
Manag Care 2008;14:388-92
[18] Bramley TJ, Gerbino PP, Nightengale BS, et al. Relationship of blood pressure control to adherence with
antihypertensive monotherapy in 13 managed care organizations. J Manag Care Pharm 2006;12:239-45
[19] Krousel-Wood M, Thomas S, Muntner P, et al. Medication adherence: a key factor in achieving blood
pressure control and good clinical outcomes in hypertensive patients. Curr Opin Cardiol 2004;19:357-62
[20] Breekveldt-Postma NS, Penning-van Beest FJ, Siiskonen SJ, et al. Effect of persistent use of
antihypertensives on blood pressure goal attainment. Curr Med Res Opin 2008;24:1025-31
[21] Schultz JS, O'Donnell JC, McDonough KL, et al. Determinants of compliance with statin therapy and
low-density lipoprotein cholesterol goal attainment in a managed care population. Am J Manag Care.
2005;11:306–12[22] Perreault S, Dragomir A, Blais L, et al. Impact of adherence to statins on chronic heart
failure in primary prevention. Br J Clin Pharmacol 2008;66:706-16
[23] Newby LK, LaPointe NM, Chen AY, et al. Long-term adherence to evidence-based secondary
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 50
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
prevention therapies in coronary artery disease. Circulation 2006;113:203-12
[24] Rasmussen JN, Chong A, Alter DA. Relationship between adherence to evidence-based
pharmacotherapy and long-term mortality after acute myocardial infarction. J Am Med Assoc 2007;297:17786
[25] Ho PM, Magid DJ, Shetterly SM, et al. Medication nonadherence is associated with a broad range of
adverse outcomes in patients with coronary artery disease. Am Heart J 2008;155:772-9
[26] Breekveldt-Postma NS, Penning-van Beest FJ, Siiskonen SJ, et al. The effect of discontinuation of
antihypertensives on the risk of acute myocardial infarction and stroke. Curr Med Res Opin 2008;24:121-7
[27] Nelson MR, Reid CM, Ryan P, et al. Self-reported adherence with medication and cardiovascular disease
outcomes in the Second Australian National Blood Pressure Study (ANBP2). Med J Aust 2006;185:487-9
[28] Sokol MC, McGuigan KA, Verbrugge RR, et al. Impact of medication adherence on hospitalization risk
and healthcare cost. Med Care 2005;43:521-30
[29] Mazzaglia G, Ambrosioni E, Alacqua M, et al. Adherence to antihypertensive medications and
cardiovascular morbidity among newly diagnosed hypertensive patients. Circulation 2009;120:1598-605
[30] Kettani FZ, Dragomir A, Côté R, et al. Impact of a better adherence to antihypertensive agents on
cerebrovascular disease for primary prevention. Stroke 2009;40:213-20
[31] Perreault S, Dragomir A, White M, et al. Better adherence to antihypertensive agents and risk reduction
of chronic heart failure. J Intern Med 2009;266:207-18
[32] Perreault S, Dragomir A, White M, et al. Adherence level of antihypertensive agents in coronary artery
disease. Br J Clin Pharmacol 2010;69:74-84
[33] Dragomir A, Côté R, Roy L, et al. Impact of adherence to antihypertensive agents on clinical outcomes
and hospitalization costs. Med Care 2010;48:418-25
[34] Corrao G, Parodi A, Nicotra F, et al. Better compliance to antihypertensive medications reduces
cardiovascular risk. J Hypertens 2011;29:610-8
[35] Bouchard MH, Dragomir A, Blais L, et al. Impact of adherence to statins on coronary artery disease in
primary prevention. Br J Clin Pharmacol 2007;63:698-708
[36] Perreault S, Dragomir A, Blais L, et al. Impact of adherence to statins on chronic heart failure in primary
prevention. Br J Clin Pharmacol 2008;66:706-16
[37] Liberopoulos EN, Florentin M, Mikhailidis DP, et al. Compliance with lipid-lowering therapy and its
impact on cardiovascular morbidity and mortality. Expert Opin Drug Saf 2008;7:717-25
[38] Shalev V, Chodick G, Silber H, et al. Continuation of statin treatment and all-cause mortality. A
population-based cohort study. Arch Intern Med 2009;169:260-8
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 51
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[39] Corrao G, Conti V, Merlino L, et al. Adherence to statin therapy and risk of nonfatal ischemic heart
disease in daily clinical practice. Clin Ther 2010;32:300-10
[40] Degli Esposti L, Saragoni S, Batacchi P, et al. Adherence to statin treatment and health outcomes in an
Italian cohort of newly treated patients: results from an administrative database analysis. Clin Ther
2012;34:190-9
[41] McDonald HP, Garg A, Haynes RB. Interventions to enhance patient adherence to medication
prescriptions: scientific review. JAMA 2002;288:2868-79
[42] Drummond MF, O’Brien B, Stoddart GL, et al. Methods for the economic evaluation of HEALTHCARE
programmes, (2nd edn). Oxford, New York: Oxford University Press, 1997
[43] van Luijn JCF, Gribnau FWJ, Leufkens HGM. Availability of comparative trials for the assessment of
new medicines in the European Union at the moment of market authorization. Br J Clin Pharm 2007;63:159–
62
[44] Corrao G, Scotti L, Zambon A, et al. Cost-effectiveness of enhancing adherence to therapy with statins
in the setting of primary cardiovascular prevention. Evidence from an empirical approach based on
administrative databases. Atherosclerosis 2011;217:479-85
[45] Glick H, Heyse JF, Thompson D, et al. A model for evaluating the cost-effectiveness of cholesterollowering treatment. Int J Technol Assess HEALTHCARE 1992;8:719–34
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 52
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKCV4
Estimating cost-effectiveness of time-varying drug therapy using
healthcare administrative databases. The case of statin in secondary
prevention
Background
Healthcare economic evaluations are an important scientific evidence that could help
decision makers in the allocation of healthcare resources by comparing costs and
consequences of health interventions. In detail, pharmacoeconomic evaluations compare
efficacy and costs related to drugs. Randomized clinical trials (RCTs) are usually used to
obtain efficacy data and in the last years some clinical trials also incorporate measures of
cost. Although RCTs have high theoretical validity, they are characterized by high
adherence and persistence to treatment. Therefore results may be not generalizable to real
practice in the general population where patients aren’t fully compliant or persistent to drug
therapy. Suboptimal compliance or failure to persist with therapy for the prescribed
duration reduce the therapeutic potential of drug treatment and account for differences
between efficacy and clinical effectiveness [1]. Moreover noncompliance and
nonpersistence to drug therapy work on two opposite directions on healthcare resources use
and consequently costs [2]: the cost of therapy decreases but costs associated with the
treated disease may increase as a result of reducing clinical effectiveness. Given that
compliance and persistence affect both health outcomes and costs, these concept should be
included to accurately estimate the cost-effectiveness of drug therapies in observational real
world data [3]. That’s why in the last years the effect of suboptimal compliance and failure
to persist with drug therapy has become an important issue in healthcare economic
evaluations.
Observational studies provide important data to estimate cost-effectiveness in the real-life
medical practice and administrative databases could be a useful tool, though their well-
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 53
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Known limitations. Consequently research in this field is problematic and the development
of economic modeling techniques is challenging [1-6]. To date, the work is sparse and
although a range of approaches have been applied further research in this field is required.
Aim
The main object of this work was to identify appropriate models for cost-effectiveness
estimates of time-varying chronic drug therapy in real-word practice using healthcare
administrative databases. In detail marginal structural models (MSMs) were used to
evaluate cost-effectiveness of statin therapy and medication adherence in the secondary
prevention of acute myocardial infarction (AMI).
Methods
This was an observational longitudinal study based on healthcare administrative databases
of Lombardy, previously organized in a data warehouse with probabilistic matching
between databases because of lack of unique identifiers. The cohort study included patients
hospitalized in 2003 for their first episode of AMI and it was followed until December 31,
2008, collecting data on healthcare services and vital status. At baseline and during followup chronic cardiovascular drug therapies suggested in the secondary prevention of AMI [7]
were investigated: statin therapy (which is the therapy of interest) and therapies with
antithrombotic agents, ace inhibitors, beta blocking agents and calcium channel blockers.
During follow-up persistence and adherence to statin were measured as time-dependent
variables. The main outcome observed during follow-up was all-cause mortality but
hospitalizations for cardiovascular events, such as AMI recurrences, Angina pectoris, heart
revascularization procedures, stroke or transient ischemic attack were also investigated.
Cost were quantified using charges. We adopted a net-benefit regression approach with
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 54
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
related cost-effectiveness acceptability curves, using direct medical costs and gained lifeyears as outcome. MSMs accounted for the dynamic interactive effects between treatment
and the time-varying confounders [8-11] i.e. non-fatal cardiovascular events and others
cardiovascular drug therapies.
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Data management and analysis: Carla Fornari and Fabiana Madotto1,2
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
Revicki DA, Frank L. Pharmacoeconomic evaluation in the real world. Effectiveness versus efficacy
studies. Pharmacoeconomic 1999;15:423-34
[2]
Hughes D, Cowel W, Koncz T, et al. Methods for integrating medication compliance and persistence
in pharmacoeconomic evaluations. Value Health 2007;10:498-509
[3]
Hiligsmann M, Boonen A, Rabenda V, at al. The importance of integrating medication adherence into
pharmacoeconomic analyses: the example of osteoporosis. Expert Rev Pharmacoecon Outcomes Res
2012;12:159-66
[4]
Hughes Da, Bagust A, Haycox A, et al. The impact of non-compliance on the cost-effectiveness of
pharmaceuticals: a review of the literature. Health Econ 2001;10:601-15
[5]
Cleemput I, kesteltoolK, De Geest S. A review of the literature on the economics of noncompliance.
Room for methodological improvement. Health Policy 2002;59:65-94
[6]
Rosen AB, Spaulding AB, Greenberg D, et al. Patient adherence: a blind spot in cost-effectiveness
analyses. Am J Manag Care 2009;15:626-32
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 55
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[7]
European guidelines on cardiovascular disease prevention in clinical practice: executive summary.
European Society of Cardiology. Eur J Cardiovasc Prev Rehabil 2007;14(Supp.2):E1-40
[8]
Robins JM, Hernán MA, Brumback B. Marginal structural models and causal inference in
epidemiology. Epidemiology 2000;11:550–60
[9]
Hernán MA, Brumback B, Robins JM. Marginal structural models to estimate the causal effect of
zidovudine on the survival of HIV-positive men. Epidemiology 2000;11:561–70
[10]
Shortreed SM, Forbes AB. Missing data in the exposure of interest and marginal structural models: a
simulation study based on the Framingham Heart Study. Stat Med 2010;29:431-43
[11]
Yu AP, Yu YF, Nichol MB. Estimating the effect of medication adherence on health outcomes among
patients with type 2 diabetes--an application of marginal structural models. Value Health 2010;13:1038-45
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 56
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKCV5
Exploring the impact of prescribing and substituting generic drugs: the
example of cardiovascular therapies
Background
Appropriate and cost-effective prescribing is a major goal for all participants in the
healthcare system. Due to their low cost profile, there is a strong global focus on generic
drugs. Increased use of these drugs has been promoted by governments as a means of
contrasting the growth in the public pharmaceutical budget, and most Western countries
have implemented generic prescribing and/or generic substitution in order to achieve this
[1–5]. Interventions specifically aimed to increase generic drug use have been encouraged
[6-14] based on the principle that generic drugs can substantially reduce costs without
compromising quality when they are used in appropriate clinical settings [15].
Generic manufacturers normally submit applications to the regulatory authorities based
upon the safety and efficacy data of the equivalent branded product. They have to
demonstrate that the pharmacokinetics of the same molar dose of their product is within
acceptable, predefined limits. This proof of bioequivalence is an important issue affecting
both generic formulations and different brands of a particular drug [16]. However, the
evidence extending bioequivalence of generic and branded medicines to therapeutic
equivalence is rather poor, but overall supportive. A systematic review and meta-analysis
of clinical equivalence of generic and brand-name drugs used in cardiovascular disease
identified 47 articles providing evidence on this topic, of which 38 were randomized,
controlled trials [15]. Overall, there was no evidence of superiority of brand name
compared with generic drugs. There are, however, several clinical areas or drug types
where strong evidence is not available [16].
Few previous studies have explicitly evaluated the relationship between generic drugs and
medication adherence. Two studies of a plan’s switch to a generic-only formulary found
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 57
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
significant reductions in the overall use of prescriptions, including decreases in the
essential use of ACE inhibitors and statins by patients with diabetes and coronary artery
disease, and increases in self-reported financial burden [17,18]. Conversely, a recent study
of a tiered pharmacy benefit found adherence was 12.6% higher for patients initiated on
generic medications [19]. These studies may not be directly comparable, though, since
switching to generics may be a distinct behaviour from initiating generics. Nevertheless,
none of these analyses accounted for the independent role of copayments, or evaluated
whether their findings remained constant across different medical conditions. A recent
study found the use of generic drug therapy was inconsistently associated with improved
adherence, and the effects were generally small [20]. Rather, a $0 copayment was the
strongest and most consistent predictor of adequate adherence in the study conditions,
regardless of the use of generics or brands.
Aims
The project aims of comparing adherence, safety, effectiveness and cost-effectiveness
profiles of generics and brand-name drugs for treatment of hypertension, dyslipidaemia and
type 2 diabetes mellitus.
Methods
A cohort study will be conducted including all information on drug prescription and
clinical outcome evaluated during follow-up for the incident users of antihypertensive,
lipid-lowering and antidiabetic drugs using the electronic archives of Lombardy Region. As
adherence determinants, treatment, patients and general practitioner characteristics will be
investigated. The outcomes of interest will be major cardiovascular events and death for
any cause recorded during follow-up. The incidence of adverse events will be assessed
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 58
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
based on the presence of prescriptions of specific drugs and / or hospitalization for specific
diseases. Monte Carlo Sensitivity Analysis using external information from GP databases
and Instrumental Variables approach [21-31], as well as case-crossover and case-timecontrol designs [32-35] will be used for unmeasured confounding adjustment (e.g. patients
profile and in physicians’ propensity of prescribing generics, among others). Financial
aspects will be considered for assessing the association between copayments amount
charged for brand-name drugs and level of adherence to drug therapy. Finally, the cost of a
CV hospitalization avoided by generics and band-name drug treatments will be compared
by means of cost-effectiveness models [36,37].
Management
Project manager (clinics): Giuseppe Mancia1, Enrico Agabiti Rosei2,
Gianfranco Gensini3
Project manager (pharmacology): Alberico L Catapano4
Project manager (epidemiology and biostatistics): Giovanni Corrao5,6
Data management and analysis: Davide Soranna6
1
2
3
4
Italian Institute for Auxology
Dept of Clinical and Experimental Sciences, University of Brescia, Brescia, Italy
School of Medicine, University of Firenze, Florence, Italy
Dept of Pharmacological Sciences, and Centre for Pharmacoepidemiology and Pharmacoutilization,
University of Milano, Milan, Italy
5
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
6
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 59
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
References
[1] Kanavos P, Mossialos E. Outstanding regulatory aspects in the European pharmaceutical market.
Pharmacoeconomics 1999;15:519–33
[2] Garattini L, Tediosi F. A comparative analysis of generics markets in five European countries. Health
Policy 2000;51:149–62
[3] McManus P, Birkett DJ, Dudley J, et al. Impact of the minimum pricing policy and introduction of brand
(generic) substitution into the pharmaceutical benefits scheme in Australia. Pharmacoepidemiol Drug Saf
2001;10:295–300
[4] Haas JS, Phillips KA, Gerstenberger EP, et al. Potential savings from substituting generic drugs for brandname drugs: medical expenditure panel survey, 1997–2000. Ann Intern Med 2005;142:891–7
[5] Simoens S. Trends in generic prescribing and dispensing in Europe. Expert Rev Clin Pharmacol
2008;1:497–503
[6] Håkonsen H, Horn AM, Toverud EL. Price control as a strategy for pharmaceutical cost containment—
what has been achieved in Norway in the period 1994–2004? Health Policy 2009;90:277–85
[7] Avorn J, Soumerai SB. Improving drug-therapy decisions through educational outreach. A randomized
controlled trial of academically based “detailing”. N Engl J Med 1983;308:1457–63
[8] Fretheim A, Aaserud M, Oxman AD. Rational prescribing in primary care (RaPP): economic evaluation
of an intervention to improve professional practice. PLoS Med 2006;3:e216
[9] Scott AB, Culley EJ, O’Donnell J. Effects of a physician office generic drug sampling system on generic
dispensing ratios and drug costs in a large managed care organization. J Manag Care Pharm 2007;13:412–9
[10] Tootelian DH, Royer J, Johnson RC. Providing incentives to control HEALTHCARE costs and remain
competitive in the marketplace: a pilot study. Health Mark Q 1997;15:87–99
[11] Andersson K, Bergström G, Petzold MG, et al. Impact of a generic substitution reform on patients’ and
society’s expenditure for pharmaceuticals. Health Policy 2007;81:376–84
[12] Johnsrud M, Lawson KA, Shepherd MD. Comparison of mail-order with community pharmacy in plan
sponsor cost and member cost in two large pharmacy benefit plans. J Manag Care Pharm 2007;2:122–34
[13] Schneeweiss S, Walker AM, Glynn RJ, et al. Outcomes of reference pricing for angiotensin-converting
enzyme inhibitors. N Engl J Med 2002;346:822–9
[14] Fischer MA, Schneeweiss S, Avorn J, et al. Medicaid prior-authorization programs and the use of
cyclooxygenase-2 inhibitors. N Engl J Med 2004;351:2187–94
[15] Rector TS, Finch MD, Danzon PM, et al. Effect of tiered prescription copayments on the use of preferred
brand medications. Med Care 2003;41:398–406
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 60
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[16] Kesselheim AS, Misono AS, Lee JL, et al. The clinical equivalence of generic and brand-name drugs
used in cardiovascular disease: a systematic review and meta-analysis. JAMA 2008;300:2514–26
[17] Duerden MG, Hughes DA. Generic and therapeutic substitutions in the UK: are they a good thing? Br J
Clin Pharmacol 2010;70:335–41
[18] Tseng CW, Brook RH, Keeler E, et al. Effect of generic only drug benefits on seniors’ medication use
and financial burden. Am J Manag Care 2006;12:525–32
[19] Christian-Herman J, Emons M, George D. Effects of generic-only drug coverage in a Medicare HMO.
Health Aff (Millwood) 2004;(Suppl Web Exclusives):W4-455–68
[20] Shrank WH, Hoang T, Ettner SL, et al. The implications of choice: prescribing generic or preferred
pharmaceuticals improves medication adherence for chronic conditions. Arch Intern Med 2006;166:332–7
[21] Essebag V, Platt RW, Abrahamowicz M, et al. Comparison of nested case– control and survival analysis
methodologies for analysis of timedependent exposure. BMC Med Res Methodol 2005;5:5
[22] Lohr KN. Emerging methods in comparative effectiveness and safety. Symposium overview and
summary. Medical Care 2007;45:S5-S8
[23] Brookhart MA, Wang P, Solomon DH, et al. Evaluating short-term drug effects using a physicianspecific prescribing preference as an instrumental variable. Epidemiology 2006;17: 268-75
[24] Sturmer T, Glynn RJ, Rothman KJ, et al. Adjustments for unmeasured confounders in
pharmacoepidemiologic database studies using external information. Medical Care 2007;45:S158-S165
[25] Cox E, Martin BC, Van Staa T, et al. Good research practices for comparative effectiveness research:
approaches to mitigate bias and confounding in the design of nonrandomized studies of treatment effects
using secondary data sources: The International Society for Pharmacoeconomics and Outcomes Research
Good Research. Practices for retrospective database analysis task force report—Part II. Value Health
2009;12:1053–61
[26] Johnson ML, Crown W, Martin BC, et al. Good research practices for comparative effectiveness
research: analytic methods to improve causal inference from nonrandomized studies of treatment effects
using secondary data sources. The International Society For Pharmacoeconomics and Outcomes Research
Good Research. Practices for retrospective database analysis task force report—Part III. Value Health
2009;12:1062–73
[27] Brookhart MA, Sturmer T, Glynn RJ, et al. Confounding control in healthcare database research.
Challenges and potential approaches. Medical Care 2010;48:S114-S120
[28] Schneeweiss S, Gagne JJ, Glynn RJ, et al. Assessing the comparative effectiveness of newly marketed
medications: methodological challenges and implications for drug development. Clin Pharmacol Ther
2011;90:777-90
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 61
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[29] Sox HC, Goodman SN. The methods of comparative effectiveness research. Annu Rev Public Health
2012;33:425–45
[30] Bradbury BD, Gilbertson DT, Brookhart MA, et al. Confounding and control of confounding in
nonexperimental studies of medications in patients with CKD. Advances in Chronic Kidney Disease
2012;19:19-26
[31] Garbe E, S Kloss, Suling KM, et al. High-dimensional versus conventional propensity scores in a
comparative effectiveness study of coxibs and reduced upper gastrointestinal complications. Eur J Clin
Pharmacol DOI 10.1007/s00228-012-1334-2
[32] Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute
events. Am J Epidemiol 1991;133:144–53
[33] Farrington CP. Control without separate controls: evaluation of vaccine safety using case-only methods.
Vaccine 2004;22:2064–70
[34] Lumley T, Levy D. Bias in the case - crossover design: implications for studies of air pollution.
Environmetrics 2000;11:689–704
[35] Suissa S. The case-time-control design. Epidemiology 1995;6:248–53
[36] Corrao G, Scotti L, Zambon A, et al. Cost-effectiveness of enhancing adherence to therapy with statins
in the setting of primary cardiovascular prevention. Evidence from an empirical approach based on
administrative databases. Atherosclerosis 2011;217:479-85
[37] Glick H, Heyse JF, Thompson D, et al. A model for evaluating the cost-effectiveness of cholesterollowering treatment. Int J Technol Assess HEALTHCARE 1992;8:719–34
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 62
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKCV6
Utilization, outcomes and costs of implantable cardioverter defibrillator
(ICD)
Background
The implantable cardioverter defibrillator (ICD) was developed to prevent sudden cardiac
death (SCD) in subjects with left ventricular systolic dysfunctions or heart failure. The
extension of treatment indications, clinical experience and improvements in technical
capability of the device contributed to the exponential increase in the number of ICD
implanted [1-3] in the last decades, both in US and Europe. At the beginning (the 80’s),
implantation was recommended only in secondary prevention of SCD, that is in subjects
resuscitated from at least two episodes of cardiac arrest [4] and, given the high fatality rate
of ventricular tachyarrhythmia and cardiac arrest, in the first decade after its approval we
witnessed a low number of ICDs implanted [1]. From the middle 90’s lots of randomized
trials_ENREF_11 [5-11] provided evidence that ICD improves survival in certain high-risk
groups of patients compared to the best antiarrhythmic therapy. For this reason, from the
early two thousands the use of ICD grew rapidly: the number of first ICD implanted in the
American population was 577 per person-years in 2006 e in Europe it was 115 [2].
The treatment with ICD is costly, both as regard the initial expenditure connected with
implant procedure and device and the subsequent costs for check-up, pharmacological
therapies, device replacement and possible complications [12]. The high cost of ICD
requires a continuous monitoring of its real effectiveness and appropriateness and the costeffectiveness of ICD was little investigated in settings outside randomized clinical
trial_ENREF_26 [13] and, given the careful selection of trial participants, there is the need
to ascertain the cost-effectiveness in real practice of this therapy.
To obtain clinical, epidemiological and economic information about ICD implantation
activity in a population is very expensive, since it requires data collection on a large
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 63
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
number of people for enough time to capture long-term outcomes (e.g. mortality,
replacement device). Administrative healthcare databases can be a practical solution to
obtain useful information on a large population with less effort.
Aims
This project estimates the epidemiological and economic impact of ICD therapy in Italy,
which has a universal healthcare system (HS). In detail, we used administrative healthcare
databases of Lombardy to evaluate: (i) the time trends in ICD therapy from 2004 to 2008;
(ii) the patient profiles; (iii) the healthcare economic impact of a cohort of subjects with a
first implant performed during 2005-2007.
Methods
We obtained data for the current analysis from the data warehouse DENALI, which
collects and organizes the administrative healthcare datasets of the publicly funded national
HS in Lombardy.
We identified ICD implantations according to ICD-9-CM diagnosis and procedure codes
reported on hospital discharges and we classified them into “first ICD” and “ICD
replacement”. We evaluated the trend in the annual number of first implants from 2004 to
2008 and to assess temporal trends, the estimates were age and sex standardized using as
reference the 2001 Italian population. Moreover, we estimated the annual replacement rates
using as denominator the time at risk of people with a first ICD implanted.
In order to assess the economic burden of this treatment and to evaluate demographic and
clinical characteristics of patients, we selected subjects who underwent a first ICD
implantation between 2005 and 2007 and we followed them from discharge to 31st
December 2008. Cox proportional hazards models were performed to examine the effects
of baseline covariates on mortality and ICD replacement during follow-up. At last, direct
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 64
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
healthcare cost was analysed from the perspective of the HS system and it was quantified
using charges, the amount of money the Lombardy HS reimbursed to providers of care. We
estimated the mean annual per-capita costs covered by HS after a first ICD implanted,
using the Bang and Tsiatis method which deals with the problem of right censoring during
the follow-up [14].
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Felice Achilli 2
Data management and analysis: Carla Fornari and Fabiana Madotto1,3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
3
Cardiology Dept, A. Manzoni Hospital, Lecco, Italy
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] Camm AJ, Nisam S. The utilization of the implantable defibrillator – a European enigma. Eur Heart J
2000;21:1998-2004
[2] Camm AJ, Nisam S. European utilization of the implantable defibrillator: Has 10 years changed the
'enigma'? Europace 2010;12:1063-9
[3] Hlatky MA, Saynina O, McDonald KM, et al. Utilization and outcomes of the implantable cardioverter
defibrillator, 1987 to 1995. Am Heart J 2002;144:397-403
[4] Mirowski M, Reid P, Mower M et al. Termination of malignant ventricular arrhythmias with an implanted
automatic defibrillator in human beings. N Engl J Med 1980;303:322–4
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 65
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[5] Moss AJ, Hall WJ, Cannom DS, et al. Improved survival with an implanted defibrillator in patients with
coronary disease at high risk for ventricular arrhythmia. Multicenter automatic defibrillator implantation trial
investigators. N Engl J Med 1996;335:1933-40
[6] Moss AJ, Zareba W, Hall WJ, et al. Prophylactic implantation of a defibrillator in patients with
myocardial infarction and reduced ejection fraction. N Engl J Med 2002;346:877-83
[7] Buxton AE, Lee KL, Fisher JD, Josephson et al. A randomized study of the prevention of sudden death in
patients with coronary artery disease. Multicenter unsustained tachycardia trial investigators. N Engl J Med
1999;341:1882-90
[8] Kadish A, Dyer A, Daubert JP, et al. Prophylactic defibrillator implantation in patients with nonischemic
dilated cardiomyopathy. N Engl J Med 2004;350:2151-8
[9] Bristow MR, Saxon LA, Boehmer J, et al. Cardiac-resynchronization therapy with or without an
implantable defibrillator in advanced chronic heart failure. N Engl J Med 2004;350:2140-50
[10] Hohnloser SH, Kuck KH, Dorian P, et al. Prophylactic use of an implantable cardioverter-defibrillator
after acute myocardial infarction. N Engl J Med 2004;351:2481-8
[11] Bardy GH, Lee KL, Mark DB, et al. Amiodarone or an implantable cardioverter-defibrillator for
congestive heart failure. N Engl J Med 2005;352:225-37
[12] Boriani G, Biffi M, Martignani C, et al. Expenditure and value for money: The challenge of implantable
cardioverter defibrillators. QJM 2009;102:349-56
[13] Chan PS, Nallamothu BK, Spertus JA, Masoudi et al. Impact of age and medical comorbidity on the
effectiveness of implantable cardioverter-defibrillators for primary prevention. Circ Cardiovasc Qual
Outcomes 2009;2:16-24
[14] Bang H, Tsiatis AA. Estimating medical costs with censored data. Biometrika 2000;2:329-43
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 66
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKCV7
Electrocardiographic waveform analysis to predict the success of
defibrillation in human victims of out-of hospital cardiac arrest and the
association with in hospital outcomes and survival
Background
Cardiovascular disease (CVD) is a major public health problem. 30% of the number of
deaths occurred in the world during 2008 was due to CVD, which is the major contributor
of non-communicable diseases [1]. WHO future projections state that CVD will still be the
leading cause of death and among the 15 major causes of disease burden in 2030 [2].
Cardiac arrest (CA) is related with CVD because it is a possible outcome of the disease and
in many individuals it is also the first expression of disease. Epidemiology of in hospital or
out-of hospital CA is difficult to estimate because of sudden and often unrecognized nature
of CA, but the available estimates confirm that it remains a substantial public health
problem. In the United States, there were about 295,000 out-of hospital CAs assessed in
emergency medical services in 2008 [3]. In Europe, at the end of the ‘90s, CA affected
about 700,000 individuals per year. The UK reported an out-of-hospital CA incidence of
123 cases per 100,000 person-year, followed by Germany with 115 cases and Norway and
Finland with 51 and 80 cases respectively [4]. The incidence of CA with any initial rhythm
is not decreasing overtime in US [3].
On the other hand, the treatment of CA is still a difficult challenge: survival is still low
despite efforts in research and implementation because of many variables that could
influence the success of the emergency medical intervention [5]. Initial rhythm is a major
determinant of outcome: shockable rhythms like ventricular fibrillation (VF) and pulseless
ventricular tachycardia (VT) have a better outcome when compared with no shockable
rhythms, asystole and pulseless electrical activity (PEA). The median survival rate to
hospital discharge with out-of-hospital CA with any first recorded rhythm is about 7.9%
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 67
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
and increased to 21.0% for VF rhythms [3]. So it is important to optimize treatment
strategies to improve survival of CA. Research concentrates mainly on shockable rhythms,
for which the use of automated external defibrillators (AEDs) is appropriate.
Options for effective treatment of CA are substantially conditioned by a few early
interventions: high quality cardiopulmonary resuscitation (CPR) and rapid defibrillation
[5]. The best coordination of this two interventions is still uncertain in best practice
guidelines in resuscitation. CPR is an attempt to restore spontaneous circulation by a
combination of rescue breathing and chest compressions and it can revert some of the
process that make the fibrillating heart refractory to shock. Recent studies suggested that it
is useful before defibrillation in prolonged attack [6]. Electrical defibrillation is a passage
of current at high voltage to restore a normal regular rhythm by means of automated, semiautomated or manual external defibrillator. Many studies have demonstrated that repetitive
shocks and the amount of power delivered may be injurious to the ischemic myocardium
and so they may compromise clinical outcomes after return of spontaneous circulation and
long-term prognosis [7-10]. Moreover CPR pauses are mandatory prior to attempting
defibrillation for rhythm analysis, but interruptions compromise the ultimate success of
resuscitation process, because myocardial perfusion is decreased of 40% [11-14]. That’s
why it is important to identify the best timing of defibrillation.
The development of a non-invasive and real-time predictor of the success of defibrillation
has become an important issue. Coronary perfusion pressure (CPP) is currently recognized
as the best single indicator, but it’s an invasive method not feasible in out-of hospital
settings [15]. End tidal CO2 is a non-invasive surrogate of CPP but it requires additional
instrumentation [16]. Alternatively, electrocardiogram [ECG] waveform analysis is a noninvasive and real time approach available during CA. Several experimental studies and
retrospective human studies have revealed that the study of ECG waveform may predict the
success of defibrillation. Many ECG parameters derived from ECG waveform has been
suggested concerning amplitude, frequency, bispectral analysis, amplitude spectrum area
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 68
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[AMSA], wavelets nonlinear dynamics, N[α] histograms and combination of parameters
[17-19]. Actually the AMSA is recognized as one of the most efficient parameters and it is
highly correlated with CPP[20-24]. It is a combination of mean amplitude and dominant
frequency and it is not influenced by artifacts due to CPR.
Despite several studies have demonstrated the efficacy of AMSA as predictor of success of
defibrillation in experimental setting and in human retrospective data. Best practice need
more solid evidence. It is necessary to validate the predictive accuracy of this parameter on
an enough large population of human beings and to identify a threshold value of
defibrillation success. Moreover all the studies used only initial restoration of spontaneous
circulation as outcome and they didn’t investigate the effect on survival at hospital
discharge or long-term survival. Other issue not already studied are the effect of possible
confounders like pharmacological therapies on AMSA and the validation of AMSA in a
prospective study.
Aims
The main aim of this translational study is to confirm the efficacy of AMSA to predict the
likelihood that any electrical shock would restore a perfusing rhythm in human victims of
out-of-hospital CA with shockable rhythms using the emergency events registered in
Lombardy in the years 2008-2010. Moreover this study will investigate the association
between AMSA values and in hospital outcomes or prolonged survival.
Methods
Electrocardiographic (ECG) data recorded by AEDs available on emergency vehicles of the
emergency system of Lombardy (AREU) during out-of-hospital CA events occurring in
Lombardy in the biennium 2008-2009 will be analyzed. Among these events, only VF/VT
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 69
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
cardiac arrests receiving defibrillations (DFs) will be included. Data concerning the
emergency events will be extrapolated from the administrative database of the emergency
medical service. A 4.1 seconds ECG window ending at 0.5 sec before defibrillation shock
will be analyzed to derive AMSA and others ECG parameters. The success of a
defibrillation shock will be defined using the standard recognized classification: presence
of spontaneous rhythm ≥ 40 beats per minute starting within 60 seconds from the shock.
The predictive power of AMSA to discriminate among successful and not successful DFs
will be analyzed using ROC curves and the properties of AMSA will be compared to that
of other ECG parameters. A threshold values of AMSA able to discriminate between
successful and not successful DFs will be estimated. Than data from the emergency system
will be matched to that of administrative healthcare electronic database of Lombardy
region, with respect to privacy policies. Data on in-hospital part of the emergency event
and on long-term survival will be extrapolated. The association between ECG derived
parameters and in-hospital outcomes will be investigated with appropriate regression
models. In a second phase of the study, the established threshold value of AMSA will be
validated using out-of-hospital CA registered in Lombardy in 2010 by the emergency
system.
Management
Project manager (clinics): Antonio Pesenti 1, Roberto Latini 2
Project manager (epidemiology and biostatistics): Giancarlo Cesana3
Data management and analysis: Giuseppe Ristagno2, Carla Fornari3,
Tommaso Mauri1
1
2
3
Dept of Perioperative Medicine and Intensive Care, San Gerardo Hospital, Monza, Italy
Dept of Cardiovascular Research, Mario Negri Institute, Milan, Italy
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 70
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
4
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
Global status report on non-communicable diseases 2010. I. World Health Organization 2011. (ISBN
978 92 4 068645 8 (PDF)
[2]
Mathers CD, Loncar D. Projections of Global Mortality and burden of Disease from 2002 to 2030.
Plos Med 2006;3(11):e442
[3]
Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke statistics - 2011 update: a report
from the American Heart Association. Circulation 2011;123:e18–209
[4]
Wollard M. Public access defibrillation: a shocking idea? J Public Health Med 2001;23:98-102
[5]
Scapigliati S, Ristagno G, Cavaliere F. The best timing for defibrillation in shakable cardiac arrest.
Minerva Anestesiol. 2012 (epub ahead of print)
[6]
Deakin CD, Nolan JP. European Resuscitation Council Guidelines for Resuscitation 2005. Section3.
Electrical therapies: automated external defibrillators, defibrillation, cardioversion and pacing. Resuscitation
2005;67(S1):S25-S37
[7]
Yamaguchi H, Weil MH, Tang W, et al. Myocardial dysfunction after electrical defibrillation.
Resuscitation 2002;54:289-96
[8]
Xie J, Weil MH, Sun S, et al. High energy defibrillation increases the severity of post resuscitation
myocardial dysfunction. Circulation 1997;96:683-88
[9]
Gaba DM, Talner NS. Myocardial damage following transthoracic direct current countershock in
newborn piglets. Pediatr Cardiol 1982;2:281–88
[10]
Doherty PW, McLaughlin PR, Billingham M et al. Cardiac damage produced by direct current
countershock applied to the heart. Am J Cardiol 1979;43:225–32
[11]
Steen S, Liao Q, Pierre L, et al. The critical importance of minimal delay between chest
compressions and subsequent defibrillation: a haemodynamic explanation. Resuscitation. 2003; 58:249-58
[12]
Yu T, Weil MH, Tang W, et al. Adverse outcome of interrupted precordial compression during
automated defibrillation. Circulation 2002;106:368-72
[13]
Eftestol T, Sundek, Steen Pa. effects of interrupting compressions on the calculated probability
defibrillation success during out-of-hospital cardiac arrest. Circulation 2002;105:2270-3
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 71
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[14]
Sato Y, Weil MH, Sun S, et al. Adverse effects of interrupting precordial compression during
cardiopulmonary resuscitation. Crit Care Med 1997;25:733-6
[15]
Paradis NA, Martin GB, Rivers EP et al. Coronary perfusion pressure and the return of spontaneous
circulation in human cardiopulmonary resuscitation. JAMA 1990;263:1106-13
[16]
Von Planta M, Von Planta I, Weil MH et al. End tidal carbon dioxide as an haemodynamic
determinant of cardiopulmonary resuscitation in the rat. Cardiovascular Res 1989;23:364-8
[17]
Amann A, Rheinberger K, Achleitner U. Algorithms to analyze ventricular fibrillation signals. Curr
Opin Crit Care 2001;7:152-6
[18]
Neurarter A, Eftestol T, Kramer-Johansen J, et al. Prediction of countershock success using single
features from
multiple ventricular fibrillation frequency bands and features combinations using neural
networks. Resuscitation 2007;3:253-6
[19]
Noc M, Harry M, Wanchun T, et al. Electrocardiographic prediction of the success of cardiac
resuscitation. Critical Care Medicine 1999;27(4):708-14
[20]
Nakagawa Y, Sato Y, Kojima T, et al. Electrical defibrillation outcome prediction by waveform
analysis of ventricular fibrillation in cardiac arrest out of hospital patients. Tokai J Exp Clin Med
2012;37(1):1-5
[21]
Ristagno G, Gullo A, Berlot G, et al. Prediction of successful defibrillation in human victims of out-
of hospital cardiac arrest: a retrospective electrocardiographic analysis. Anaesth Intensive Care
2008;36(1):46-50
[22]
Young C, Bisera J, Gehman S, et al. Amplitude spectrum area: measuring the probability of
successful defibrillation as applied to human data. Crit Care Med 2004;32(suppl 9):s3656-8
[23]
Marn-Pernat A, Weil MH, Tang W, et al. Optimizing timing of ventricular defibrillation. Crit Care
Med 2001;29(12):2360-5
[24]
Povoas HP, Bisera J. Electrocardiographic waveform analysis for predicting the success of
defibrillation. Crit Care Med 2000;28(11):N210-1
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 72
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.2. Diabetes
Project CRACKDB1
Assessing adherence, long-term safety and cost-effectiveness profiles of
drug therapies for type 2 diabetes in clinical practice
Background
The term Type 2 Diabetes Mellitus (D2TM) describes a metabolic disorder with
heterogeneous aetiologies which is characterized by chronic hyperglycaemia as common
phenotype and disturbances of carbohydrate, fat and protein metabolism resulting from
defects in insulin secretion, insulin action, or both [1].
According to the International Diabetes Federation it is estimated that in 2010, 284 million
people had T2DM of whom 55 million were in Europe [2]. The WHO estimates a
worldwide prevalence of 2.8% in 2000 that is going to double and reach the 4.4% in 2030
[3]. In Italy, prevalence ranges around 4.5-5.0%. A recent survey conducted in Turin
reported a D2TM prevalence of 4.9% [4], while update evaluation based on the ISTAT
(Italian National Institute of Statistics) GP’s database and drug consumption archives
reports a prevalence of over 5% [5].
Chronic complications of T2DM have been classified into vascular and nonvascular
complications. The vascular complications of D2TM are further classified into
microvascular (retinopathy, neuropathy, nephropathy) and macrovascular complications
(coronary heart disease, peripheral arterial disease, cerebrovascular disease) [6].
It is well established that the risk of microvascular and macrovascular complications is
related to glycaemia, as measured by glycated haemoglobin (HbA1c). Hence, the main goal
of T2DM treatment is to control hyperglycaemia then reducing its vascular complications.
There are several treatment options to reduce blood glucose levels. For decades treatment
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 73
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
options were based on lifestyle changes, metformin, sulfonylureas, and insulin. By the end
of the 90’s and during the last decade, new compounds with different mechanisms of action
have been developed and authorized for marketing (i.e.: thiazolidinediones, meglitinides,
glucagon-like peptide 1 receptor agonists, dipeptidyl-peptidase 4 inhibitors and amilyn
analogs) [7]. The progressive nature of T2DM and its associated glycaemia tends to lead to
increases in the dose and the use of combinations of non-insulin blood glucose lowering
drugs or the addition of insulin over time to meet the goals for glycaemia control. With this
regard, due to paucity of long-term comparative-effectiveness trials available, uniform
recommendations on the best agent to be combined with metformin cannot be made.
Although safety issues associated with blood glucose lowering drugs are not new, recently
the safety of these treatments has been questioned. It has been reported that some of them
increase the risk or modify the prognosis of diseases such as cancer, cardiovascular or
pancreatic diseases [8]. Particular attention has to be paid to pancreatic outcomes: druginduced pancreatitis is considered a rare diagnosis and it is estimated that less than 2% of
pancreatitis cases are induced by a drug [9]. Health authorities have received reports of
acute pancreatitis in patients taking exenatide and sitagliptin.
According to studies based on prescription archives of Germany, Scotland and other
European northern countries [10-12], the use of retard insulin (mainly glargine insulin) is
suspected to increase the risk of cancer, mainly breast cancer, with respect to both, general
population and diabetics who use other antidiabetic agents. Preliminary data of an
epidemiologic investigation performed by the Takeda Pharmaceutical Company, suggests
a possible association between the use of pioglitazone and the risk of bladder cancer [13].
All these evidence, however, are still open to the discussion because the increased cancer
risk associated with the use of these agents might rather have been generated by patients’
characteristics, such as glycemic control, body mass index, smoking, and other cancer risk
factors.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 74
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Open questions
Although a big effort was spent to test these type of drugs there still is a significant need
for high-quality comparative-effectiveness research, not only regarding glycemic control,
but also with respect to other clinical outcomes (morbidity and life-limiting complications)
and costs. Another issue needing more data and evidence is the concept of durability of
effectiveness (often ascribed to beta cell preservation), which would serve to stabilize
metabolic control and decrease the future treatment burden for patients. Finally, safety
concerns need to be carefully evaluated.
Objective
The main aims of the project is of assessing, further quantify and understando profiles of (i)
pharmacoutilization of blood glucose lowering agents, insulin and insulin analogs; (ii) costeffectiveness of different treatment strategies; (iii) cardio, cerebrovascular, pancreatic and
oncologic safety of blood glucose lowering agents, insulin and insulin analogs in patients
affected by T2DM. A second aim of the project is to provide integrated decision models
that may facilitate clinical and decision making. Key innovations will be made in several
areas, which will allow progressing substantially beyond the current state of the art.
Subjects and Methods
To achieve these objectives two ways will be followed. The first one concerns of creating
the conventional repository of HCU (Lombardy) and MR (UlNet) database mainly for
assessing pharmacoutilization and cost-effectiveness profiles. The second way concerns of
adhering to the SAFEGUARD consortium. SAFEGARD (Safety Evaluation of Adverse
Reactions in Diabetes) is an international network of HCU and MR databases covering
populations from six countries (please see below the list of participants to the
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 75
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
SAFEGUARD project). The main aim of the SAFEGUARD project is to assess and further
quantify and understand safety concerns of blood glucose lowering, insulin and insulin
analogs agents, in patients affected by T2DM. Before making data available to the
consortium members, they will be aggregated and encrypted making impossible to trace
back the patients included in the study. The elaborated data provided by these databases
will be used to conduct different nested case-control studies to evaluate the association
between use of each specific agent, and the risk of myocardial infarction, heart failure,
ventricular arrhythmia/sudden cardiac death, ischemic stroke, hemorrhagic stroke, acute
pancreatitis, pancreatic cancer and bladder cancer. Moreover these data will be used for the
construction of different decision models, one for each outcome of interest.
Impact
The decision models will support the general practitioner in the choice of the treatment to
assign to T2DM newly diagnosed patients according to their characteristics (e.g, age,
gender, clinical profile, severity of T2DM, ecc...). Moreover the results of this project will
increase the knowledge on the diabetes treatment related adverse events focusing in
particular on rare events such as acute pancreatitis and pancreatic and bladder cancer and
on the new generation diabetic therapies.
Management
Project manager (clinics): Giuseppe Mancia1 and Guido Grassi2
Project manager (epidemiology and biostatistics): Giovanni Corrao
and Antonella Zambon,3,4 Carlo La Vecchia5,6
Project manager (primary care): Ovidio Brignoli and Alessandro
Filippi 7
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 76
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data management and analysis: Federica Nicotra, Giulia Segafredo,
Federica Nicotra, Lorenza Scotti and Andrea Arfè,4 and Cristina
Bosetti6
1
2
Italian Institute for Auxology
Division of Internal Medicine, San Gerardo Hospital, Monza, Dept of Health Sciences, University of
Milano-Bicocca, Milan, Italy
3
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
4
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
5
Dept of Clinical Sciences and Community Health, Unit of Medical Statistics and Biometrics, University of
Milano, Milan, Italy
6
Laboratory of EPIDEMIOLOGY, Dept of Epidemiology, Mario Negri Institute for Pharmacologic Research,
Milan, Italy
7
Health Search, Italian College of General Practitioners, Florence (G.M., E.S.), Italy
List of participants to the SAFEGUARD project
•
•
•
•
•
•
•
Erasmus University Medical Center - EMC (Netherlands): Miriam Sturkenboom, Bruno Stricker,
Martijn Schuemie, Gianluca Trifirò, Sabine Straus, Silvana Romio, Peter Rijnbeek, Gwen Masclee,
Ingrid Leal
Synapse Research Management Partners S.L. – SYNAPSE (Spain): Robert Fabregat, Eva Molero
PHARMO Coöperation UA - PHARMO (Netherlands): Ron M.C. Herings, Fernie J.A. Penning-van
Beest, Huub M.P.M. Straatman, Myrthe P.P. van Herk-Sukel, Irene D. Bezemer
Fondazione Scientifica SIMG-ONLUS – FSIMG (Italy): Ovidio Brignoli, Giampiero Mazzaglia,
Iacopo Cricelli
University of Bath- UBATH (UK): Corinne de Vries, Annie Hutchison, Alison Nightingale, Cormac
Sammon, Julia Snowball
Agencia Española de Medicamentos y Productos Sanitarios – AEMPS (Spain): Dolore Montero,
Miguel Gil, Gema Requena, Francisco de Abajo, José Luis Alonso-Lebrero
Consorzio Mario Negri Sud – CMNS (Italy): Antonio Nicolucci, Giorgia De Berardis, Fabio
Pellegrini
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 77
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
•
•
•
•
•
•
•
Drug Safety Research Trust – DSRU (UK): Saad Shakir, Lorna Hazell, Miranda Davies, Deborah
Layton
Univerzita Karlova v Praze – CUNI (Czech Republic): Martin Haluzík, Miloš Mráz, Štěpán Svačina
VU University Medical Center – VUA (The Netherlands): Michaela Diamant, Erik Serné, Richard G
IJzerman, Mark Smits
The Brigham and Women’s Hospital, Harvard Medical School – BWH (USA): Sebastian
Schneeweiss, Jerry Avorn, Jeremy Rassen, John Seeger
University of Milano-Bicocca – UNIMIB (Italy): Giovanni Corrao, Antonella Zambon, Federica
Nicotra, Lorenza Scotti, Andrea Arfè
Universitaet Bremen - UNI-HB (Germany): Edeltraut Garbe, Tania Schink, Niklas Schmedt
RTI Health Solutions - RTI-HS (Spain): Susana Pérez-Gutthann, Cristina Varas-Lorenzo, Manel
Pladevall, Carla Franzoni, Nuria Riera
References
[1] Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus, Abbreviated Report of a
WHO
Consultation,
WHO
2011,
WHO/NMH/CHP/CPM/11.1.
Reported
in:
http://www.who.int/diabetes/publications/report-hba1c_2011.pdf (October 2012)
[2] Kannel WB, McGee DL. Diabetes and cardiovascular disease. The Framingham study. JAMA
1979;241:2035-8
[3] Associazione Medici Diabetologi (AMD) - Società Italiana di Diabetologia (SID). Standard italiani per la
cura del diabete mellito, 2009-2010. Infomedica – Formazione & Informazione Medica Editore, Torino, 2010
[4] Gnavi R, Karaghiosoff L, Costa G, et al. Socioeconomic differences in the prevalence of diabetes in Italy:
the population-based Turin study. Nutr Metab Cardiovasc Dis, 2008;8:678–82
[5] Cricelli C. Mazzaglia G, Samani F, et al. Prevalence estimates for chronic diseases in Italy: exploring the
differences between self-report and primary care database. J Public Health Med 2003;25:254-7
[6] Inzucchi SE, Bergenstal RM., Buse JB, et al. Management of hyperglycaemia in type 2 diabetes: a patientcentered approach. Position statement of the American Diabetes Association (ADA) and the European
Association for the Study of Diabetes (EASD), Diabetologia, 2012;55:1577-96
[7] Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycaemia in type 2 diabetes: a patientcentered approach. Position statement of the American Diabetes Association (ADA) and the European
Association for the Study of Diabetes (EASD). Diabetologia 2012;55:1577-96
[8] Giovannucci E, Harlan DM, Archer MC, Bergenstal RM, Gapstur SM, Habel LA, Pollak M, Regensteiner
JG, Yee D. Diabetes and cancer: a consensus report. Diabetes Care 2010;33:1674-85
[9] Nitsche CJ, Jamieson N, Lerch MM, Mayerle JV. Drug induced pancreatitis. Best Pract Res Clin
Gastroenterol 2010;24:143-55
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 78
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10] Hemkens LG, Grouven U, Bender R, et al. Risk of malignancies in patients with diabetes treated with
human insulin or insulin analogues: a cohort study. Diabetologia 2009;52:1732-44
[11] Jonasson JM, Ljung R, Talback M, et al. Insulin glargine use and short-term incidence of malignancies-a
population-based follow-up study in Sweden. Diabetologia 2009;52:1745-54
[12] Colhoun HM; SDRN Epidemiology Group. Use of insulin glargine and cancer incidence in Scotland: a
study from the Scottish Diabetes Research Network Epidemiology Group. Diabetologia 2009;52:1755-65
[13] Peck, Peggy (September 17, 2010). FDA Says It Will Review Pioglitazone Safety. MedPage Today.
http://www.medpagetoday.com/Endocrinology/Diabetes/22274. Retrieved 18 September 2010
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 79
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKDB2
Measuring the healthcare burden of diabetes mellitus
Background
Diabetes mellitus is one of the most common non-communicable disease and it is
undoubtedly one of the most challenging healthcare problems in the 21st century because
of its increasing prevalence, its chronic nature and the high risk of major complications.
The world prevalence of diabetes among adults (20-79years-old) was estimated to be 6.4%
in 2010 and to rise to 7.7% by 2030 [1], this increase in prevalence was predicted to be
much greater in developing than in developed countries (69% vs 20%) [2]. A 2012 update
estimated a prevalence of 8.3% in 2011, rising to 9.9% by 2030 [3]. The international
diabetes federation (IDF) also estimated that globally half of those who have diabetes, are
unaware of their condition. Most of these have type 2 diabetes. Type 2 diabetes is the
predominant form and accounts for at least 90% of cases. In Italy, the prevalence of
diabetes rose from 3.0% in 2000 to 4.2% in 2007 and projection in subjects aged ≥ 30 years
indicates that the prevalence will rise continuously reaching 11.1% by 2030 [4]. The
increasing prevalence of diabetes could be attributable to increasing incidence of the
disease or decreasing mortality. Many studies have tried to understand the balance between
incidence and mortality of the disease but it is still not clear [4-10]. Incidence of the disease
may be increasing due to changes in the ratio of diagnosed and undiagnosed cases or
changes in life-style as a consequence of progressive urbanization and improved health
status, including high-energy diets and reduced physical activity [5]. On the other hand, the
population demographic changes, as ageing of the population, younger age at onset and
better prognosis of the disease could influence the mortality rate in patients with diabetes.
Many studies in developed countries show a decreasing mortality rate and a stable
incidence rate [7-10]. Others studies observed also an increased incidence in younger ages
and in developing countries [5,6]. In Italy, the incidence of diabetes remained stable
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 80
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
between 2000 and 2007 with a rate of 4 per 1,000 person-year, and overall mortality
declined of 6.7% at a rate slightly higher than that of the general population [4].
Diabetes Mellitus can lead to long-term complications which can be very serious and even
fatal, if not prevented, and have the potential to reduce the quality of life [11]. Actually
cardiovascular disease is the major cause of death in diabetes, accounting in most
populations for 50% or more of all diabetes fatalities. Moreover diabetes may cause renal
failure, eye disease characterised by damage to the retina and it can ultimately lead to
ulceration and amputation of the toes, feet and lower limbs. The cost of complications in
many countries accounts for over 60% of total healthcare expenditure on diabetes [12].
More recent studies have confirmed that a substantial proportion of the costs of diabetes
arise from treating long-term complications, particularly cardiovascular and renal disease
[13].
Healthcare economic evaluations on chronic diseases, such as diabetes, are essential in
public health surveillance for planning health services, prevention strategies [14]. The
burden of diabetes and its complications has been examined by both American and
European studies, but the estimates of costs are difficult to compare because of quite
different approaches [15-20]. Many studies validated the use of administrative databases in
identifying cases of diabetes [21]. The American Diabetes Association show healthcare
expenditure to be as much as four times higher for people with diabetes than for people
without the condition [19]. The London school of Economics processed recently an
analysis about the trend of the diabetic disease in five European countries (Germany,
France, Italy, UK, Spain) and found that diabetes accounts for the greatest part of total
healthcare costs in Germany (16.7%) and for the lowest part (5.6%) in Italy [22]. A recent
report of population-based study in Turin compared direct costs of diabetic and nondiabetic people matching for age, sex, type of diabetes and treatment: in diabetic people
57.2% of direct costs were due to hospitalizations (vs 62.9% among non-diabetic people),
25.6% to drugs (vs 21.1%), 11.9% to outpatient care (vs 14.3%), 0.9% to emergency care
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 81
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
(vs 1.6%) and 4.4% to consumable goods related to diabetes care (nearly absent in nondiabetic people) [17]. However a limited number of diabetic people was examined for a
short time.
Aims
The aims of this study were to estimate the prevalence of patients affected by diabetes
mellitus in the general population of Lombardy in 2000 and to estimate the direct medical
costs associated to diabetic patients followed up until the 31 December 2008 using
administrative databases.
Methods
The cohort study was identified using administrative healthcare electronic databases of
Lombardy region previously organised with a probabilistic linkage in a data warehouse
(DENALI). All individuals who during the year 2000 had an hospital discharge with a
IDC-9-CM code 250.xx, and/or two consecutive prescriptions of drugs for diabetes (ATC
code A10XXXX) within one year, and/or an exemption from co-payment healthcare costs
specific for diabetes mellitus, were selected and followed up until the 31 December 2008.
Prevalence, mortality and mean annual HEALTHCARE costs (hospitalizations, drugs and
outpatient examinations/visits) from the National Health Service’s perspective were
estimated.
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Data management and analysis: Gianluca Furneri and Luciana
Scalone2
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 82
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Fondazione CHARTA (Center for Health Associated Research and Technology Assessment), Milano
References
[1] Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010 and 2030.
Diabetes Res and Clin Practice 2010;87:4-14
[2] Nolan CJ, Damm P, Prentki M. Type 2 diabetes across generations: from pathophysiology to prevention
and management. Lancet 2011;378:169-81
[3] International Diabetes Federation. IDF Diabetes Atlas, 5th ed. Brussels, Belgium: International Diabetes
Federation, 2011. http://www.idf.org/diabetesatlas
[4] Monesi L, Baviera M, Marzona I, et al. Prevalence, incidence and mortality of diagnosed diabetes:
evidence from an Italian population-based study. Diabet Med 2012;29:385-92
[5] Colagiuri S, Borch –Johnsen K, Glumer C, et al. There really is an epidemic of type 2 diabetes?
Diabetologia 2005;48:1459-63
[6] Wareham NJ, Forouhi NG. Is there really an epidemic of diabetes? Diabetologia 2005;48:1454-55
[7] Green A, Stovring H,Andersen M, et al. The epidemic of type 2 diabetes is a statistical artefact.
Diabetologia 2005;48:1456-8
[8] Stovring H, Andersen M, Beck-Nielsen H et al . Rising prevalence of diabetes : evidence from a Danish
pharmaco-epidemiological database. Lancet 2003;362:537-8
[9] Carstensen B, Kristensen JK, Ottosen P, et al. The Danish national diabetes register: trends in incidence,
prevalence and mortality. Diabetologia 2008;51: 2187-96
[10] Ringborg A, Lindgren P, Martinell M, et al. Prevalence and incidence of type 2 diabetes and its
complications 1996-2003- estimates from a Swedish population-based study. Diabet Med 2008;25:1178-86
[11] Cosgrove P, Engelgau M, Ishrat Islam. Cost effective approaches to diabetes care and prevention.
Diabetes Voice. 2002;47(4):13-6
[12] Gruber W, Leese B, Songer B, et al. The economics of diabetes and diabetes care. International Diabetes
Federation and World Health Organisation, 1997
[13] Herman WH. The Economics of Diabetes Prevention. Med Clin North Am 2011;95(2):373–8
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 83
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[14] Brocco S, Visentin C, Fedeli U, et al. Monitoring the occurrence of diabetes mellitus and its major
complications: the combined use of different administrative databases. Cardiovascular Diabetology 2007;6:111
[15] Koster I, von Ferber L, Ihle P, et al. The cost burden of diabetes mellitus: the evidence from Germanythe CoDiM Study. Diabetologia 2006;49:1498-504
[16] Jonsson B. Revealing the cost of type II diabetes in Europe. Diabetologia 2002;45:S5-S12
[17] Bruno G, Picariello R, Petrelli A, et al. Direct costs in diabetic and non-diabetic people: the population
based Turin study; Italy. Nutr Metab Cardiovasc Dis 2012;22(8):684-90
[18] Oliva J, Lobo F, Molina B et al. Direct HEALTHCARE costs of diabetic patients in Spain. Diabetes
Care 2004;27:2616-21
[19] American Diabetes Association. Economic costs of diabetes in the U.S. in 2007. Diabetes Care
2008;31:596-615
[20] Henriksson F, Agardh CD, Berne C, et al. Direct medical costs for patients with type 2 diabetes in
Sweden. J Intern Med 2000;248:387-96
[21] Saydah SH, Geiss LS, Tierney ED, et al. Review of the performance of methods to identify diabetes
cases among vital, administrative and survey data. Ann Epidemiol 2004;14:507-16
[22] Kavanos P, Van den Aardweg S, Schurer W. Diabetes expenditure, burden of disease and management
in
5
EU
countries.
LSE
Health,
London
School
of
Economics.
January
2012.
http://www2.lse.ac.uk/LSEHealthAndSocialCare/research/LSEHealth/MTRG/LSEDiabetesReport26Jan2012.
pdf
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 84
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKDB3
Assessing the gap between guidelines and practice: the example of
gestational diabetes
Background
Gestational diabetes mellitus (GDM) is defined as “any degree of glucose intolerance with
onset or first recognition during pregnancy” [1]. The diagnostic criteria for GDM were
defined more than 40 years ago, and were derived from the criteria to diagnose diabetes in
non-pregnant individuals [2]. The new guidelines for the diagnosis of GDM proposed by
the IADPSG (International Association of Diabetes and Pregnancy Study Groups, 2008)
are based on the data of the HAPO (Hyperglycemia and Adverse Pregnancy Outcomes,
2008) study [2,3] and recommend: i) the universal screening of all pregnant women at 24
weeks of pregnancy; ii) the use of a single oral glucose tolerance test (OGTT) with 75
grams of glucose (i.e., the same test used for non-pregnant individuals); ii) the diagnosis of
GDM when at least one glucose value is above the following thresholds: baseline 92 mg/dl;
60 minutes 180 mg/dl; 120 minutes 152 mg/dl. It is hoped that these new guidelines will
increase adherence to GDM screening, since the proposed procedure (one 75 g OGTT at 24
weeks of pregnancy) is simpler that that used in the past (a 50 g OGTT followed by a 100 g
OGTT).
Data from the HAPO study suggest that even degrees of maternal glucose intolerance less
severe than that of overt diabetes mellitus increase the risk of adverse outcomes [3]. In fact,
GDM, unless promptly diagnosed and treated, is associated with an increased risk of
adverse pregnancy outcomes, both for the mother and the child (e.g., preeclampsia,
prematurity, caesarean section, macrosomia, neonatal hypoglycemia). It complicates
approximately 1-14% of all pregnancies [4].
Besides complicating pregnancy, GDM affects women and their children well beyond
delivery.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 85
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
It has been observed that mothers with GDM have a higher risk of developing type 2
diabetes in the years following childbirth, for this reason current guidelines recommend
that women diagnosed with GDM should test their glucose tolerance with a 75 g OGTT 612 weeks after delivery. However, reports coming from different countries report a low
compliance with post-partum testing (<50%), this implies that several type 2 diabetes cases
are not recognized and not adequately treated. Furthermore, children of mothers with GDM
have a higher risk of developing obesity and other chronic non-transmissible diseases
during their life [3-5].
In Italy there are no available information on screening practice for GDM in the general
population. The few published reports or articles are analysis on population of a specific
area or hospital-based patient cohorts [6-8].
Objective
To estimate current adherence to GDM screening and to monitor changes overtime as the
new guidelines are implemented using the HCU databases of Lombardy Region
(hospitalizations, Certificates of Delivery Care – CEDAP, disease exemptions, specialist
and laboratory benefits, drug prescriptions archives). Furthermore, adherence to glucose
tolerance testing after delivery among women who were diagnosed with GDM will be
estimated. Specific aims: 1) to estimate the proportion of pregnant women who are
screened for GDM and, among those, the incidence of GDM in the Lombardy region; 2) to
assess differences in the probability of being screened for GDM according to factors such
as maternal age, province of residence, country of birth (Italian-born vs. non Italian-born),
residence in urban or rural areas, etc.; 3) to study the association between diagnosis of
GDM and subsequent maternal outcomes (caesarean delivery, length of hospital admission
for childbirth and HEALTHCARE costs during pregnancy) and child outcomes (neonatal
hypoglycemia, preterm delivery, intensive neonatal care, fetal macrosomia, and length of
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 86
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
hospitalization after birth); 4) to estimate the proportion of women with a pregnancy
complicated by GDM who have their glucose tolerance checked in the post-partum period,
and the proportion of women who are diagnosed with type 2 diabetes in the year after
childbirth.
Subjects and Methods
The target population will be represented by all fertile women resident in the Lombardy
Region in the period 2005-2011. All women who gave birth to a child in a public or private
hospitals affiliated with the NHS will be identified. To identify GDM incident cases, all
women with pre-gestational diabetes will be excluded. Among the remaining women, those
who were screened for GDM will be identified. The proportion of screened women, overall
and stratified by several patients’ characteristics, will be estimated. Among screened
women all new cases of GDM will be identified and logistic regression models will be
fitted to compute hazard ratios (HRs), and corresponding 95% CI, estimating the
association between: i) selected patients’ characteristic and diagnosis of GDM; ii) GDM
and the screening of glucose tolerance or the diagnosis of type 2 diabetes in the year after
the childbirth; iii) GDM and each specific maternal or fetal outcome. A sensitivity analysis
using the Monte-Carlo approach [9] will be conducted in order to evaluate the robustness
of our findings with regard to potential biases introduced by unmeasured confounders.
Several scenarios in which the prevalence of the unmeasured confounder (e.g., severity of
diabetes, obesity/overweight, smoking, etc…) will change across adherence categories will
be created, in order to determine how the HR changes after adjusting for the unmeasured
confounders.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 87
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Impact
The results will provide solid data to plan information campaign for the general public and
educational programs for general practitioners and hospital- and clinic-based obstetricians
and midwifes on the importance of screening pregnant women for GDM and testing
glucose tolerance in the post-partum period among women who had a pregnancy
complicated with GDM.
Management
Project manager (clinics): Marina Scavini1
Project manager (epidemiology and biostatistics): Antonella
Zambon2
Data management and analysis: Federica Nicotra3
1
2
Diabetes Research Institute (HSR-DRI), San Raffaele Scientific Institute, Milan, Italy
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
3
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] International Association of Diabetes and Pregnancy Study Groups Consensus Panel, Metzger BE, Gabbe
SG, Persson B, et al. International association of diabetes and pregnancy study groups recommendations on
the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care 2010;33:676-82
[2] HAPO Study Cooperative Research Group, Metzger BE, Lowe LP, Dyer AR, et al. Hyperglycemia and
adverse pregnancy outcomes. N Engl J Med 2008;358:1991-2002
[3] Mulla WR, Henry TQ, Homko CJ. Gestational diabetes screening after HAPO: has anything changed?
Curr Diab Rep 2010;10:224-8
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 88
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[4] Huidobro A, Prentice A, Fulford T, et al. Gestational diabetes, comparison of women diagnosed in second
and third trimester of pregnancy with non GDM women: Analysis of a cohort study. Rev Med Chil
2010;138:316-21
[5] Di Cianni G, Benzi L, Casadidio I, et al. Screening of gestational diabetes in Tuscany: results in 2000
cases. Ann Ist Super Sanita 1997;33:389-91
[6] Fedele D, Lapolla A. A protocol of screening of gestational diabetes mellitus. Ann Ist Super Sanita
1997;33:383-7
[7] Di Cianni G, Volpe L, Lencioni C, et al. Prevalence and risk factors for gestational diabetes assessed by
universal screening. Diabetes Res Clin Pract 2003;62:131-7
[8] Di Cianni G, Volpe L, Casadidio I, et al. Universal screening and intensive metabolic management of
gestational diabetes: cost-effectiveness in Italy. Acta Diabetol 2002;39:69-73
[9] Steenland K, Greenland S. Monte Carlo sensitivity analysis and Bayesian analysis of smoking as an
unmeasured confounder in a study of silica and lung cancer. Am J Epidemiol 2004;160:384-92
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 89
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.3. Oncology
Project CRACKK1
Clinical use, safety and effectiveness of novel high cost anticancer
therapies after marketing approval: a record linkage study
Background
Increased understanding of the molecular events involved in cancer development has led to
the identification of several novel approaches to anticancer therapy [1]. Given the paucity
of phase III trials evaluating these new agents' effectiveness and the excessive reliance on
surrogate end-points, the added value of these drugs has been questioned [2]. Concerns
have also been raised on their toxicity [3,4]. It is unclear whether the frequencies of serious
adverse events reported in clinical trials hold in clinical practice, in particular in patients
excluded from clinical trials [5,6]. Previous work has shown that the Lombardy Region
administrative databases can provide valuable information on these issues [7-10].
Objective
To investigate the utilization of 17 new targeted high cost drugs in Lombardy oncology
practice between 2004 and 2010, namely bevacizumab, bortezomib, cetuximab, docetaxel,
ibritumomab
tiuxetan,
irinotecan,
oxaliplatin,
paclitaxel,
pemetrexed,
rituximab,
trastuzumab, fotemustine, alemtuzumab, temsirolimus, nelarabine, and panitumumab.
Specific objectives are: (i) To monitor the use of selected high cost anticancer therapies in a
large unselected population (ii) To evaluate adherence to the Italian Medicines Agency
(AIFA, Agenzia Italiana del Farmaco) indications of use (iii) To estimate the incidence of
serious adverse events in clinical practice and investigate their predictors (iv) To
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 90
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
investigate the potential of administrative databases for estimating effectiveness of
anticancer drugs.
Subjects and Methods
For each drug, we will perform a retrospective analysis to assess the pattern of drug
prescription using a computerized record linkage of several regional health service
databases: the File F registry (in which the administration of 17 novel anticancer drugs
reimbursed by the National Health Service is mandatory recorded), the Regional hospital
discharge forms (Scheda di Dimissione Ospedaliera, SDO) database, the drug prescription
database, the outpatients' services database, and the Registry Office database. Subjects who
received at least one prescription of the selected drugs from 2004 to 2010 and resident in
Lombardy are eligible. For complications warranting hospitalization that have been related
to each drug, we will search these conditions in the main and secondary diagnoses in the
patients' SDOs after the first drug administration. We will rely on previously developed
algorithms to identify patients with selected chronic conditions (e.g. diabetes), and evaluate
predictors of adverse events.
Using the Registry Office database we will also evaluate survival. We will use various
methods (propensity score, Monte Carlo sensitivity analysis, etc.) to define a comparable
control group for the drug under study in order to verify the feasibility of an effectiveness
study [11-13].
Impact
The project will provide a detailed description of the use of novel high cost anticancer
therapies in clinical practice in the Lombardy Region, which includes 9.5 million
inhabitants. It will identify their licensed and off-label use, and evaluate the adherence to
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 91
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
the regulatory approval indications. Moreover, the project will provide relevant information
on serious adverse drug reactions in clinical practice, whereas, at present, the frequency of
adverse events is available from clinical trials only. In particular, information on serious
adverse events in subjects generally excluded from clinical trials will be investigated. This
project will also allow evaluating strengths and limitations of administrative databases in
various aspects of pharmacoepidemiology. In conclusion, this study will provide relevant
information both at clinical and regulatory level about these novel anticancer drugs, which
will help clinicians and regulatory agencies in their choices.
Management
Project manager: Carlo La Vecchia1,2
Project manager (biostatistics): Eva Negri2
Data management and analysis: Marta Rossi3
1
Dept of Clinical Sciences and Community Health, Unit of Medical Statistics and Biometrics, University of
Milano, Milan, Italy
2
Laboratory of EPIDEMIOLOGY, Dept of Epidemiology, Mario Negri Institute for Pharmacologic Research,
Milan, Italy
3
Laboratory of EPIDEMIOLOGICAL METHODS, Dept of Epidemiology, Mario Negri Institute for
Pharmacologic Research, Milan, Italy
References
[1] Apolone G, Joppi R, Bertele V, et al. Ten years of marketing approvals of anticancer drugs in Europe:
regulatory policy and guidance documents need to find balance between different pressures. Br J Cancer
2005;93:504-9
[2] Garattini S, Bertele V. Efficacy, safety, and cost of new anticancer drugs. Br Med J 2002;325:269-71
[3] Albini A, Pennesi G, Donatelli F, et al. Cardiotoxicity of anticancer drugs: the need for cardio-oncology
and cardio-oncological prevention. J Natl Cancer Inst 2010;102:14-25
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 92
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[4] Zangari M, Fink LM, Elice F, et al. Thrombotic events in patients with cancer receiving antiangiogenesis
agents. J Clin Oncol 2009;27:4865-73
[5] Ratner M, Gura T. Off-label or off-limits? Nat Biotechnol 2008;26:867-75
[6] Stafford RS. Regulating off-label drug use--rethinking the role of the FDA. N Engl J Med 2008;358:14279
[7] Coloma PM, Schuemie MJ, Trifirò G, et al; EU-ADR Consortium. Combining electronic healthcare
databases in Europe to allow for large-scale drug safety monitoring: the EU-ADR Project.
Pharmacoepidemiol Drug Saf 2011;20:1-11
[8] Corrao G, Romio SA, Zambon A et al. Multiple outcomes associated with the use of metformin and
sulphonylureas in type 2 diabetes: a population-based cohort study in Italy. Eur J Clin Pharmacol
2011;67:289-99
[9] Bonifazi M, Rossi M, Moja L et al. Bevacizumab in clinical practice: prescribing appropriateness relative
to national indications and safety. Oncologist 2012;17:117-24
[10] Corrao G, Parodi A, Nicotra F, et al. Better compliance to antihypertensive medications reduces
cardiovascular risk. J Hypertens 2011;29:610-8
[11] Schneeweiss S, Avorn J. A review of uses of HEALTHCARE utilization databases for epidemiologic
research on therapeutics. J Clin Epidemiol 2005;58:323–37
[12] Cole SR, Chu H, Greenland S. Multiple-imputation for measurement-error correction. Int J Epidemiol
2006;35:1074-81
[13] Greenland S. Sensitivity analysis, Monte Carlo risk analysis, and Bayesian uncertainty assessment. Risk
Anal 2001;21:579-83
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 93
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.4. Respiratory diseases
Project CRACKRD1
Measuring burden of chronic obstructive pulmonary disease: prevalence,
incidence and therapeutic approaches
Background
According to the Global Initiative for obstructive lung diseases (GOLD), chronic
obstructive pulmonary disease (COPD) is defined as a preventable and treatable disease
with some significant extra-pulmonary effects that may contribute to the severity in
individual patients. The pulmonary component is characterized by airflow limitation, which
is not fully reversible. The airflow limitation is usually progressive and associated with an
abnormal inflammatory response of the lungs to noxious particles or gases, such as
cigarette smoke [1]. The GOLD definition has become globally accepted for the diagnosis
of COPD and some crucial components of this definition have been incorporated by the
European Respiratory Society (ERS) and by the American Thoracic Society (ATS) [2].
COPD is a major health epidemic, which has important consequences for patients and
community, and still receives insufficient attention from the HEALTHCARE professionals
and scientists [3,4]. COPD is a leading cause of chronic morbidity (affects 210 million
people) and mortality (causes 3 million deaths per year) worldwide [5], and according to
the World Health Organization, it is the fifth most common cause of death and the 10th
most burdensome disease [6].
None of the available medications have been proven to change the long-term decline in
lung function of patients affected by COPD. Therefore, pharmacotherapy is mainly used to
attenuate symptoms and to prevent exacerbations of the disease. According to the GOLD
guidelines, bronchodilators are the mainstay for symptomatic management of COPD [2].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 94
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Bronchodilator treatments include β2-agonists, anticholinergics, and methylxanthines, used
alone or in combination.
Tasks
The main aims from this workpackage are the following: (i) recognizing the source of
systematic uncertainty (mainly misclassification and confounding) when HCU and MR
data are separately used, and implementing methods for controlling and/or minimizing the
effect of such biases by means of integrating HCU and MR data; (ii) measuring clinical
burden of such diseases: observed prevalence of patients under treatment (with respect to
that expected according to available data), incidence (i.e. new diagnoses and new users of
drugs specific for treatment of chronic respiratory diseases), adherence (percentage of
observational time covered by drug availability) and persistence (cumulative survival
without experiencing discontinuity episodes) of pharmacologic therapies; (iii) estimating
the extension of compliance (adherence/persistence with drug therapy) – outcome
(hospitalization for exacerbations, death) association; (iv) providing for cost-effectiveness
profile of enhancing adherence/persistence with drug therapy in clinical practice.
Management
Project manager (clinics): Alberto Pesci1
Project manager (epidemiology and biostatistics): Giovanni Corrao
and Antonella Zambon2,3
Project manager (primary care): Ovidio Brignoli and Alessandro
Filippi 4
Data management and analysis: Andrea Arfè3
1
Division of Pneumology, San Gerardo Hospital, Monza, Dept of Health Sciences, University of Milano-
Bicocca, Milan, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 95
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
2
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
3
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
4
Health Search, Italian College of General Practitioners, Florence (G.M., E.S.), Italy
References
[1] Rabe KF, H.S., Anzueto A, Barnes PJ, et al. Global strategy for the diagnosis, management, and
prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med
2007;176:532-55
[2] Celli BR, MacNee W: Standards for the diagnosis and treatment of patients with COPD: a summary of the
ATS/ERS position paper. Eur Respir J 2004;23: 932-46
[3] Barnes PJ, Kleinert S. COPD--a neglected disease. Lancet 2004;364:564-5
[4] Burney P, Suissa S, Soriano JB, et al.. The pharmacoepidemiology of COPD: recent advances and
methodological discussion. Eur Respir J Suppl 2003;43:1s-44s
[5] Halbert RJ, Isonaka S, George D, et al.. Interpreting COPD prevalence estimates: what is the true burden
of disease? Chest 2003;123:1684-92
[6] World Health Organization (WHO). Global Surveillance, Prevention and Control of Chronic Respiratory
Diseases: a Comprehensive Approach. Geneva, Switzerland. World Health Organization 2007
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 96
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKRD2
The clinical and economic burden of patients hospitalized for acute
exacerbations of chronic obstructive pulmonary disease
Background
Chronic Obstructive Pulmonary Disease (COPD) is a complex heterogeneous clinical
syndrome characterised by limited expiratory flow in the airways, with a consequent
lengthening of expiration time, associated to a variable extent with chronic bronchitis and
pulmonary emphysema [1]. Chronic bronchitis is defined as a clinical condition
characterised by the presence of persistent chronic cough, hyper production of mucus, or
both for a minimum of three months per year for at least two consecutive years, while other
causes of persistent cough should be excluded [2].
World Health Organization projections predict an increase in COPD burden in terms of
proportion of total disability-adjusted life-years lost (DALYs), from rank eleventh in 2002
to rank fourth in 2030 and that COPD will be the fourth leading cause of death by 2030 [34]. There is consolidated evidence that COPD represents half of all deaths due to diseases
of the respiratory system that together represent the third cause of death in Italy after
cardiovascular disease and cancer [5]. Moreover, the overall COPD mortality rate is 30 per
100,000 inhabitants, increasing progressively with age [6-7].
The natural course of COPD is affected by the presence of exacerbations. Based on
healthcare utilization, an exacerbation of COPD can be classified as: mild, when the patient
has an increase in respiratory symptoms that can be controlled by the patient with an
increase of the usual medication; moderate, when the patient has an increased need for
medication, namely systemic steroids and /or antibiotics; or severe, when the
patient/caregiver recognizes obvious and/or rapid deterioration in condition, requiring
hospitalization [8]. Exacerbations can have serious consequences with respect to quality of
life, lung function, and mortality [9].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 97
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
The notable clinical consequences of COPD in terms of both mortality and morbidity, as
well as the high incidence of this disease throughout the world, translate into a considerable
economic impact on the whole of society. The literature on COPD of the last 20 years
proves its high social and economic burden in several countries, particularly in
industrialized countries [10]. A common issue is the increase of cost as disease severity
moves from moderate to severe [11-14] or in relation to age and comorbidities [11].
Moreover exacerbations are the main cost drivers, some studies estimated also the
economic and social cost related to type of exacerbations [15]. In Italy the latest published
estimates refer to years 2002-2003 and are based on ad-hoc collected data on small samples
[16-18]. Direct medical burden of COPD are analysed in relation to disease severity, only
one study estimates also indirect costs [18]. Results are difficult to compare because of
different healthcare resources and methods used.
Aims
The main aim of our study was to analyse direct medical cost related to the treatment of
patient in an advanced stage of COPD in the general population of Lombardy.
As a secondary aim, the study wanted to analyse differences in the size and structure of the
population as well as in the allocation of HEALTHCARE resources during different spans
of time, namely 2003-2005 and 2006-2008.
Methods
Data were extracted from the data warehouse DENALI, which includes several databases
collected by the regional healthcare system of Lombardy as regards vital status, hospital
discharges (HDs), pharmaceutical and outpatient claims for all residents within the region.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 98
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Specifically, we built two cohorts of patients in an advanced stage of COPD: the first one
consisted of all subjects aged more than 18 years who were hospitalized at least once for a
severe exacerbation of the disease during 2003, while the second one consisted of a similar
selection applied to 2006. The two cohorts were followed up respectively until 2005 and
2008, gathering data about drug prescriptions, hospitalizations and outpatient claims, with
the relative charges.
Each cohort was divided into three groups in relation to the history of exacerbations during
the follow-up period: subjects with severe exacerbations (identified through hospital
admissions), subjects with only moderate exacerbations (identified through drug
prescriptions) and subjects without exacerbations.
For the whole cohorts and separately for each exacerbation group we evaluated
demographic characteristics and patients’ health condition at baseline [19-20]. Cox
regression models [21] were used to analyze survival and to estimate the time to death and
to the first severe COPD exacerbation in the study population. Finally, health resources
consumption and cost were evaluated on three main categories: hospitalizations, drug
packages and outpatient claims. For each category we calculated both the global cost of the
population and the mean annual cost per capita, as well as the global number of claims
requested by the whole population and the mean annual number of claims per capita. We
used the an inverse probability weighted partitioned estimator proposed by Bang and
Tsiatis [22], in order to avoid an underestimation of the costs due to right censoring [23].
As conclusion of the analysis we compared in a graphic way the results obtained in 20032005 with the ones obtained in 2006-2008.
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Francesco Blasi and Stefano Aliberti22
Data management and analysis: Fabiana Madotto and Carla Fornari3
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 99
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Respiratory Medicine Section, Dipartimento Toraco-Polmonare e Cardiocircolatorio, University of Milan,
IRCCS Fondazione Ospedale Maggiore Policlinico Cà Granda Milan, Milan, Italy
3
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
American Thoracic Society. Definitions, Epidemiology, Pathophysioloy, Diagnosis, and Staging. Am J
Respir Crit Care Med 1995;152:S77-S120
[2]
American Thoracic Society. Chronic bronchitis, asthma, and pulmonary emphysema: a statement by
the Committee on Diagnostic Standards for Nontuberculous Respiratory Diseases. Am Rev Respir Dis
1962;85:762-68
[3]
Mathers CD, Loncar D. Projections of Global Mortality and Burden of Disease from 2002 to 2030.
Plos medicine 2006;3:2011-26
[4]
The global burden of disease:2004 update. WHO Library Cataloguing-in-Publication Data. ISBN 978
924 1563710
[5]
Desideri M, Viegi G, Carrozzi L, et al. Mortality rates for respiratory disorders in Italy (1979-1990),
Monaldi Arch Chest Dis 1997;52:212-6
[6]
ISTAT. Cause di morte, anno 1993. Roma, Istituto Nazionale di Statistica
[7]
Viegi G, Pedreschi M, Pistelli F, et al. Epidemiologia delle malattie polmonari nell’anziano. Scientifica
Press srl, Firenze, 1996. (Epidemiology of lung diseases in the elderly. Scientifica Press Srl, Florence, 1996)
[8]
Cazzola M, MacNee W, Martinez FJ, et al. Outcomes for COPD pharmacological trials: from lung
function to biomarkers. Eur Respir J 2008;31:416–69
[9]
Kanner RE, Anthonisen NR, Connett JE; Lung Health Study Research Group. Lower respiratory
illnesses promote FEV1 decline in current smokers but not ex-smokers with mild chronic obstructive
pulmonary disease: results from the Lung Health Study. Am J Respir Crit Care Med 2001;164:358–64
[10]
Dal Negro R. Optimizing economic outcomes in the management of COPD. Int J Chron Obstruct
Pulmon Dis 2008;3:1-10
[11]
Halpin DMG and Miravitlles M. Chronic Obstructive Pulmonary Disease. The Disease and Its Burden
to Society. Proc Am Thorac Soc 2003;3:619–23
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 100
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[12]
Starkie HJ, Briggs AH, Chambers MG. Pharmacoeconomics in COPD: lessons for the future. Int J
Chron Obstruct Pulmon Dis 2008;3:71–88
[13]
Detournay B, Pribil C, Fournier M, et al. The SCOPE Study: Health-Care Consumption Related to
Patients with Chronic Obstructive Pulmonary Disease in France. Value in Health 2004;7:168-74
[14]
Oostenbrink JB, Rutten-van Molken MPMH. Resource use and risk factors in high-cost exacerbations
of COPD. Respiratory Medicine 2004;98:883–91
[15]
Toy EL, Gallagher KF, Stanley EL, et al. The Economic Impact of Exacerbations of Chronic
Obstructive Pulmonary Disease and Exacerbation Definition: A Review. COPD 2010;7:214–28
[16]
Lusuardi M, Lucioni C, De Benedetto F, et al. GOLD severity stratification and risk of hospitalisation
for COPD exacerbations. Monaldi Arch Chest Dis 2008;69:164-9
[17]
Kolevaa D, Motterlinia N, Banfib P, et al. Healthcare costs of COPD in Italian referral centres: A
prospective study. Respiratory Medicine 2007;101:2312–20
[18]
Dal Negro RW, Tognellaa S, Tosatto R, et al. Costs of chronic obstructive pulmonary disease (COPD)
in Italy: The SIRIO study. Respiratory Medicine 2008;102:92–101
[19]
Quan H, Sundarajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM
and ICD-10 administrativa data. Med Care 2005;43:1130-9
[20]
Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in
longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83
[21]
Cox DR. Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B
(Methodological). 1972;34:187-220
[22]
Bang H, Tsiatisa AA. Estimating medical costs with censored data. Biometrika 2000;87:329-43
[23]
Lin DY, Feuer EJ, Etzioni R, et al. Estimating medical costs from incomplete follow-up data.
Biometrics 1997;53:419-34
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 101
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKRD3
Measuring the healthcare burden associated with individuals hospitalized
for Pneumonia
Background
Pneumonia is a lung infection and can be caused by microbes, including bacteria, viruses,
or fungi. The lungs are particularly susceptible to infection because they interact with the
outside environment [1]. Healthy individuals can develop pneumonia, but it occurs with
increased frequency in individuals whose immune systems are deficient [2]. It was
estimated that 4 million cases of community-acquired pneumonia occur annually in the
United States, of which 20% to 25% are severe enough to warrant hospitalization [3].
Infants, very young children and the elderly are highly vulnerable. In children aged less
than 5 years the incidence was estimated to be 0.05 episodes per child-year in developed
countries [4]. On the basis of data from Finland, it was estimated that the age-specific
incidence was 15.4 and 34.2 cases per 1000 person-years among those aged 60–74 years
and
75 years, respectively [5].
The human and economic burden of pneumonia is enormous. In the United States alone,
total medical expenditures and indirect costs (lost work and productivity) attributed to
pneumonia amounted to more than $40 billion in 2005 [2].
Aims
The aim of this observational study was to estimate impact, clinical characteristics,
outcome and economic consequences of pneumonia hospitalized cases in the general
population using healthcare administrative databases.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 102
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Methods
The cohort study was identified using administrative healthcare electronic databases of
Lombardy region previously organised with a probabilistic linkage in a data warehouse
(DENALI). All individuals with an hospital discharge with pneumonia diagnosis ( IDC-9
CM codes 480.XX -484.XX) between 01 January 2000 to 31 December 2008 were selected
and followed up until the 31 December 2009. The entry date was the admission date of the
first hospitalization for pneumonia during the recruitment period. During the follow-up
period vital status and healthcare resources use (hospitalizations, pharmaceutical and
outpatients claims) were recorded. Incidence (x 100,000 person-years), patient’s profile,
mortality and mean monthly healthcare costs
from the National Health Service’s
perspective were estimated.
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Data management and analysis: Roberta Ciampichini2, Paolo A
Cortesi1
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Fondazione CHARTA (Center for Health Associated Research and Technology Assessment), Milano
References
[1] Scharaufnagle DE. Pneumonia. In :Breathing in America: diseases, progress and hope. 2010 by the
American Thoracic Society
[2] Mizgerd JP. Acute lower respiratory tract infection. N Engl J Med 2008;358:716–27
[3] Centers
for
Disease
Control
and
Prevention
Web
site.
Pneumonia.
Available
at:
http://www.cdc.gov/nchs/FASTATS/pneumonia.htm.
[4] Rudan I, Boschi-Pinto C, Biloglav Z, Mulholland K, Campbell H. Epidemiology and etiology of
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 103
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
childhood pneumonia. Bull World Health Organ 2008;86: 408-16
[5] Loeb M. Pneumonia in older persons. Clin Infect Dis 2003;37:1335-9
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 104
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.5. Gastrointestinal diseases
Project CRACKGD1
Measuring the burden of ulcerative colitis and Crohn’s disease: incidence
and costs
Background
Crohn’s disease (CD) and ulcerative colitis (UC) are two major chronic inflammatory
bowel diseases (IBD) and are characterized by a chronic or recurrent course with
alternating acute attacks, which require pharmacological treatment [1], and remission
periods.
Several studies aimed at estimating the incidence and prevalence of CD and UC revealed
strong geographical differences, which may be due to genetic and environmental factors
but also to poor diagnostic methods and lack of disease awareness among both physicians
and patients [2-7]. Given the low mortality rates and the early onset of the disease, which is
usually diagnosed in the age range 20-40, IBD prevalence is continuously growing [4],
especially in Europe and North America [5,8]. Furthermore, during the 20th century,
incidence rates experienced a consistent growth, maybe due to better diagnostic tools and
easier access to health services, together with a growing disease awareness [5,9].
In Italy, a few research groups estimated incidence and prevalence of CD and UC [5,10]. A
major study estimated an annual incidence rate of 5.2 (× 100,000 citizens) for UC and 2.3
(× 100,000 citizens) for CD, without a significant difference between Northern and
Southern region. Moreover, higher IBD incidence rates were observed in people aged 20 to
40 [5,11,12]. As far as gender is concerned, higher UC incidence rates were detected
among men, while no differences were observed for CD [5,11]. So far, three studies on the
prevalence of UC and CD were carried out in different Italian areas and they estimated a
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 105
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
prevalence varying form 121 UC cases and 40 CD cases (× 100,000 citizens) in Florence
during 1992; 93 and 45 cases (× 100,000 citizens) in Belluno during 2008 and 142 e 322
cases (× 100,000 citizens) in Sicily during 1979-2002 [5].
The observed increase in disease prevalence necessitates a careful economic analysis,
leading to a correct allocation of healthcare resources [12]. Several studies evaluated both
direct and indirect IBD costs [1,8,13-17], estimating high expenditures per capita. For
instance, a 10-year follow-up study on direct costs of IBD [18] estimated that in Europe,
during 2004, UC patients had a mean annual cost of €1,524, while CD patients required an
expenditure of €2,548, regardless for age and gender. Total costs were higher during the
first year after the diagnosis and mean expenditures showed high variability among
countries, with the highest value in Denmark (€3,705) and the lowest in Norway (€888). In
Italy, the estimated mean annual cost was €1,539, 50% of which was attributable to
pharmacological treatments, 40% to hospitalizations and the remainder 10% to outpatient
visits.
The most recent IBD study carried out in Lombardy dates back to the early 90’s and
estimated mean annual incidence rate of 7 for CU and 3.4 for MC (× 100,000 citizens) [5],
therefore current information on both epidemiology and costs of the disease are lacking.
Given the so far described background, it is desirable to update previous studies in order to
obtain a clear description of the present context.
Aims
The main aim of our study was to evaluate health resources consumption and costs in
subjects affected by IBD in Lombardy region, consequently estimating the burden of
disease in terms of direct costs for the Italian National healthcare system.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 106
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Methods
We carried out an observational study based on healthcare administrative databases of
Lombardy region. Specifically, data were extracted from a data warehouse (DENALI)
which organizes several databases of the regional healthcare system.
We enrolled two study cohorts, one of patients with recently diagnosed UC and the other
with CD. Subjects were identified on the basis of issuing of an exemption from co-payment
healthcare costs between January 1st, 2003 and December 31st, 2009, for CD or UC. The
two cohorts were followed until December 31st, 2009 to assess the epidemiology of both
diseases, in terms of incidence rates, and the consumption of healthcare resources
(hospitalizations, pharmaceutical prescriptions and outpatient services) with the related
costs incurred by the healthcare system. Mean expenditures where estimated through Bang
and Tsiatis [19] inverse probability weighted partitioned estimator, thus overcoming
underestimation caused by right censoring [20].
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Silvio Danese and Gionata Fiorino2
Data management and analysis: Fabiana Madotto and Carla Fornari3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
3
Division of Gastroenterology, Humanitas Clinical and Research Center, Rozzano, Milan, Italy
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 107
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
References
[1]
Park KT, Bass D. Inflammatory bowel disease-attributable costs and cost-effective strategies in the
United States: a review. Inflamm Bowel Dis 2011;17:1603-9
[2]
Gismera CS, Aladrén BS. Inflammatory bowel diseases: a disease (s) of modern times? Is incidence
still increasing? World J Gastroenterol 2008;14:5491-8
[3]
Karlinger K, Gyorke T, Mako E, et al. The epidemiology and the pathogenesis of inflammatory
bowel disease. Eur J Radiol 2000;35:154-7
[4]
Cosnes J, Gower-Rousseau C, Seksik P, et al. Epidemiology and natural history of inflammatory
bowel diseases. Gastroenterology 2011;140:1785-94
[5]
Molodecky NA, Soon IS, Rabi DM, et al. Increasing incidence and prevalence of the inflammatory
bowel diseases with time, based on systemic review. Gastroenterology 2012;142:46-54
[6]
Lakatos PL. Recent trends in the epidemiology of inflammatory bowel diseases: up or down? World
J Gastroenterol 2006;12:6102-8
[7]
Loftus EV Jr. Clinical epidemiology of inflammatory bowel disease: incidence, prevalence, and
environmental influences. Gastroenterology 2004;126:1504-17
[8]
Fedorak RN, Wong K, Bridges R. Canadian digestive health foundation public impact series.
Inflammatory bowel disease in Canada: incidence, prevalence, and direct and indirect economic impact. Can
J Gastroenterol 2010;24:651-5
[9]
Sonnenberg A. Time trends of mortality from Crohn’s disease and ulcerative colitis. Int J Epidemiol
2007;36:890-9
[10]
Masala G, Bagnoli S, Ceroti M, et al. Divergent patterns of total and cancer mortality in ulcerative
colitis and Crohn’s disease patients: the Florence IBD study 1978–2001. Gut BMJ Journal 2004;53(9):130913.
[11]
Tragnone A, Corrao G, Miglio F, et al. Incidence of inflammatory bowel disease in Italy: a
nationwide population-based study. Gruppo Italiano per lo Studio del Colon e del Retto (GISC). Int J
Epidemiol 1996;25:1044-52
[12]
Bernstein CN, Blanchard JF, Rawsthorne P, et al. Epidemiology of Crohn’s disease and ulcerative
colitis in a central Canadian province: a population-based study. Am J Epidemiol 1999;149(:916-24
[13]
Odes S. How expensive is inflammatory bowel disease? A critical analysis. World Journal of
Gastroenterology 2008;14:6641-7
[14]
Gibson TB, Ng E, Ozminkowski RJ, Wang S, et al. The direct and indirect cost burden of Crohn’s
disease and ulcerative colitis. J Occup Environ Med 2008;50:1261-72
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 108
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[15]
Kappelman MD, Rifas-Shiman SL, Porter CQ, et al. Direct HEALTHCARE costs of Crohn’s disease
and ulcerative colitis in US children and adults. Gastroenterology 2008;135:1907-13
[16]
Hillson E, Dybicz S, Waters HC, et al. HEALTHCARE expenditures in Ulcerative Colitis: the
perspectives of a self-insured employer. J Occup Environ Med 2008;50:969-77
[17]
Longobardi T, Jacobs P, Bernstein CN. Work losses related to inflammatory bowel disease in the
United States: results from the national health interview survey. Am J Gastroenterol 2003;98:1064-72
[18]
Odes S, Vardi H, Friger M, et al. Cost analysis and cost determinants in a European inflammatory
bowel disease inception cohort with 10 years of follow-up evaluation. Gastroenterology 2006;131:719-28
[19]
Bang H, Tsiatis AA. Estimating medical costs with censored data. Biometrika 2000;87:329-43
[20]
Lyn DY, Fuer ER, Wax Y. Estimating medical costs from incomplete follow-up data. Biometrics
53:113-28
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 109
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.6. Mental health
Project CRACKMH1
Adherence, effectiveness and cost-effectiveness profiles of care journeys
experienced by patients affected by severe mental disturbance
Background
Recent epidemiological research has clearly demonstrated the considerable burden that
mental disorders impose on individuals, communities, and health services throughout the
world [1]. In 2001 the World Health Organization (WHO) reported that neuropsychiatric
conditions had an aggregate prevalence of about 10% among adult population. Further
surveys conducted in developed as well as developing countries have shown that, during
their entire lifetime, more than 25% of individuals develop one or more mental or
behavioural disorders [2-4]. Globally, about 450 million people worldwide suffer from
some form of mental disorder or brain condition, and one in four people meet criteria at
some point in their life [5]. Besides the high prevalence of mental disorders, the most
recent estimates from the Global Burden of Disease Study indicate that neuropsychiatric
disorders contribute to more than 10% of lost years of healthy life and over 30% of all
years lived with disability [1]. The study showed in particular that unipolar depressive
disorders place an enormous burden on society ranking as the fourth leading cause of
burden among all diseases and accounting for over 50 million lost years of healthy life
worldwide [5]. While considerable progress has been made in the development of
psychopharmacologic agents and in the availability of effective regimens of psychotherapy
employed to treat patients with psychiatric illness [6-9], only a small fraction of this
population is adequately treated [10,11]. Treatment discontinuation is a serious problem in
outpatient community mental HEALTHCARE. Several studies report that between 22 %
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 110
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
and 63 % of patients with either a new episode or first-ever treatment dropped out after
only one service contact [12-15]. Also, a high percentage of patients dropped out of
treatment within the first 3 months, leading to a percentage rate between 18 % and 50 % of
continuing patients [13,14,16]. On the other hand, even adherence to prescribed treatment
is a serious problem. Several studies have evaluated that in schizophrenia, the proportion of
nonadherent patients can exceed 60% [17-19], while in bipolar disorder the proportion
ranges from 20% to 60% [20-22]. Another set of patients will never start or will completely
stop therapy within the first year, and only a minority will continue taking drugs as
prescribed [23]. Both nonadherence and discontinuity to psychiatric care or treatment
regimes has a important impact on psychiatric disease course, relapse, and rehospitalization
[24], high utilization of medical resources, with increased emergency room visits and
psychiatric hospitalizations [25,26]. It has been reported that patients who partially or
entirely do not adherent to therapy were 2.5 and 3 times more likely to require
rehospitalization, respectively, than patients who fully adhere to therapy [27]. This can be
an important impact on the cost of care, as well as significant detriments to the patient’s
long-term functional adaptation, including social adaptation and job reintegration [28-30].
Open questions
Despite decades of investigating factors contributing to patient’s attrition from mental
health treatment, obstacles to the delivery and success of treatments remain poorly
understood [31]. Even studies aiming at identifying predictors of treatment discontinuation,
however, are largely inconsistent [32]. Therefore further researches are needed. Lastly,
there is an ever-increasing interest and demand for an economic analysis of mental
HEALTHCARE, due to the fact that it remains a relative paucity of completed mental
health economic evaluations from both developed and developing countries [33].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 111
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Aims
The project has the following aims: (i) to identify variables predicting discontinuation of
service utilization, relapse, rehospitalization among newly diagnosed patients affected by
schizophrenia; (ii) evaluation of the cost-effectiveness of the different mental health
treatment programs.
Methods
A cohort study will be conducted to estimate the potential predictors of treatment dropout
in newly diagnosed patients affected by schizophrenia and resident in Lombardy region
between 1/01/2006 and 31/12/2007. We will include in the analysis all information on
patient socio-demographic features, drug prescriptions, treatment adherence as well as the
characteristics of the outpatient mental health centre which provide care treatment using the
HCU databases of Lombardy Region. We will estimate patterns of dropout, factors related
to relapses and the risk of re-hospitalizations by survival regression model. In addition, Cox
frailty regression models will be used to take into account heterogeneity among patients
[34]. Finally, the cost of hospitalization avoided by different mental health treatment
programs will be compared by means of cost-effectiveness models [35].
Impact
Reflecting and increasing awareness of the importance of treatment adherence on
psychiatric populations, could hold the potential for reducing morbidity and suffering of
patients and their families, in addiction to decreasing the cost of re-hospitalization.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 112
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Management
Project manager (clinics): Antonio Lora1, Emiliano Monzani2,
Caterina Viganò3
Project manager (epidemiology and biostatistics): Giovanni
Corrao,4,5 and Giancarlo Cesana4
Data management and analysis: Buthaina Ibrahim, Elisabetta Thea
Scogniamiglio5
1
Dept. of Mental Health, Local Health Unit of Lecco, Regional Health Service, Lombardy Region, Lecco,
Italy
2
Dept. of Mental Health, Hospital Niguarda Cà Granda, Regional Health Service, Lombardy Region, Milan,
Italy
3
Operative Unit f Psychiatry II, Hospital Sacco, Regional Health Service, Lombardy Region and Dept. of
Biomedical and Clinical Sciences, University of Milano, Milan, Italy
4
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
5
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] World Health Organization. The world health report 2001. Mental health: new understanding, new hope.
Geneva: World Health Organization; 2001
[2] Regier DA, Boyd JH, Burke JD, et al. One-month prevalence of mental disorders in the United States.
Based on five Epidemiologic Catchment Area sites. Arch Gen Psychiatry 1988;45:977–86
[3] Wells KB, Hays RD, Burham MA, et al. Detection of depressive disorder for patients receiving prepaid or
fee-for-service: results from the medical outcomes study. JAMA 1989;262:3298-302
[4] Almeida-Filho, Mari Jde J, Coutinho E, et al. Brazilian multicentric study of psychiatric morbidity:
methodological features and prevalence estimates. Br J Psychiatry 1997;171:524-9
[5] Cross-national comparisons of the prevalences and correlates of mental disorders. WHO International
Consortium in Psychiatric Epidemiology. Bull World Health Organ 2000;78:413–26
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 113
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[6] Ustun B. Ayuso-Mateos JL. Chatterji S, et al. Global burden of depressive disorders in the year 2000. Br J
Psychiatry 2004;184:427–32
[7] Mulrow CD, Williams JW, Trivedi M, et al. Treatment of Depression: Newer Pharmacotherapies:
Evidence Report/Technology Assessment 7: Report AHCPR/PUB-99-E014. Rockville, Md, National Center
for Health Services Research 1999
[8] Lydiard RB, Brawman-Mintzer O, Ballenger JC. Recent developments in the psychopharmacology of
anxiety disorders. J Consult Clin Psychol 1996; 64:660-8
[9] Raphael J. Leo, MD, Kuldip Jassal, et al. Nonadherence Among Psychiatric Patients with
Psychopharmacologic Treatment. Primary Psychiatry 2005;12:33-9
[10] Kessler RC, Zhao S, Katz SJ, et al. Past-year use of outpatient services for psychiatric problems in the
National Comorbidity Survey. Am J Psychiatry 1999; 156:115-23
[11] Regier DA, Narrow WE, Rae DS, et al. The de facto US mental health and addictive disorders service
system: epidemiologic catchment area prospective 1-year prevalence rates of disorders and services. Arch
Gen Psychiatry 1993; 50:85-94
[12] Brekke JS,Ansel M, Long J, et al. Intensity and continuity of services and functional outcomes in the
rehabilitation of persons with schizophrenia. Psychiatr Serv 1999;50:248–56
[13] Lavik NJ Utilisation of mental health services over a given period. Acta Psychiatr Scand 1983;67:404–
13
[14] Morlino M, Martucci G, Musella V, et al. Patients dropping out of treatment in Italy. Acta Psychiatr
Scand 1995;92: 1–6
[15] Tansella M, Micciolo R, Biggeri A, et al. Episodes of care for first-ever psychiatric patients. A long-term
case-register evaluation in a mainly urban area. Br J Psychiatry 1995;167: 220–7
[16] Lerner Y, Zilber N, Barasch M, et al. Utilisation patterns of community mental health services by newly
referred patients. Soc Psychiatry Psychiatr Epidemiol 1993;28:17–22
[17] Valenstein M, Copeland LA, Blow FC, et al. Pharmacy data identify poorly adherent patients with
schizophrenia at increased risk for admission. Med Care 2002;40:630-9
[18] Gilmer TP, Dolder CR, Lacro JP, et al. Adherence to treatment with antipsychotic medication and
HEALTHCARE costs among Medicaid beneficiaries with schizophrenia. Am J Psychiatry 2004;161:692-9
[19] Velligan DI, Diamond PM, Mintz J, et al. The use of individually tailored environmental supports to
improve medication adherence and outcomes in schizophrenia. Schizophr Bull 2008;34:483–93
[20] Perlick DA, Rosenheck RA, Kacynski R, et al. Medication nonadherence in bipolar disorder: a patientcentered review of research findings. Clin Approaches Bipolar Disord 2004;3:54–6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 114
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[21] Lingam R, Scott J. Treatment non-adherence in affective disorders. Acta Psychiatr Scand 2002;105:164–
72
[22] Colom F, Vieta E, Tacchi MJ, et al. Identifying and improving non-adherence in bipolar disorders.
Bipolar Disord 2005;7:24–31
[23] Morris LS, Schulz RM. Patient compliance--an overview. J Clin Pharm Ther 1992;17:283-95
[24] Killaspy H, Banerjee S, King M, et al. Prospective controlled study of psychiatric out-patient nonattendance. Br J Psychiatry 2000;176: 160–5
[25] Weiden PJ, Olfson M. Cost of relapse in schizophrenia. Schizophr Bull 1995;21:419-2
[26] Johnson DA, Pasterski JM, Ludlow JM, et al. The discontinuance of maintenance neuroleptic therapy in
chronic schizophrenic patients: drug and social consequences. Acta Psychiatr Scand 1983;67:339-52
[27] Lieberman JA, Koreen AR, Chakos M, et al. Factors influencing treatment response and outcome of
first-episode schizophrenia: implications for understanding the pathophysiology of schizophrenia. J Clin
Psychiatry 1996;57(suppl 9):5-9
[28] Swartz MS, Swanson JW, Hiday VA, et al. Taking the wrong drugs: the role of substance abuse and
medication noncompliance in violence among severely mentally ill individuals. Soc Psychiatry Psychiatr
Epidemiol 1998;339(suppl 1):75-80
[29] Hunt GE, Bergen J, Bashir M. Medication compliance and comorbid substance abuse in schizophrenia:
impact on community survival 4 years after a relapse. Schizophr Res 2002;54:253-64
[30] Barrett MS, Chua WJ, Crits-Christoph P, et al. Early withdrawal from mental health treatment:
implications for psychotherapy practice. Psychotherapy (Chic) 2008;45:247-67
[31] Berghofer G, Schmidl F, Rudas S, et al. Predictors of treatment discontinuity in outpatient mental
HEALTHCARE. Soc Psychiatry Psychiatr Epidemiol 2002;37:276-82
[32] Shah A, Jenkins R. Mental health economic studies from developing countries reviewed in the context of
those from developed countries. Acta Psychiatr Scand 2000;101:87-103
[33] Hougaard P. Frailty models for survival data. Lifetime Data Anal 1995;1:255–73
[34] Hougaard P. Analysis of multivariate survival data. New York: Springer-Verlag;2000
[35] Corrao G, Nicotra F, Parodi A, et al. External adjustment for unmeasured confounders improved drugoutcome association estimates based on HEALTHCARE utilization data. J Clin Epidemiol 2012;65:1190-9
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 115
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.7. Geriatrics
Project CRACKGH1
Drug treatment of elderly patients affected by cardiovascular disease and
other chronic comorbidities
Background
The increase in the number of elderly people in western countries is a profound
demographic revolution with a relevant impact on HEALTHCARE services of the global
economy. In Italy, recent ISTAT estimates (Report 2010: riporta almeno l’indirizzo web)
show that the number of people aged 65+ is 43% higher than that of younger individuals.
This phenomenon is associated with an increased prevalence of chronic diseases as
hypertension, heart disease, diabetes and chronic obstructive pulmonary disease and of
other conditions, such as urinary incontinence, delirium, falls and cognitive impairment,
which cannot be ascribed to a specific organ system pathology (geriatric syndrome) [1].
Both comorbidities and geriatric syndromes contribute to the increased risk of disability,
mortality, institutionalization, death, and healthcare costs. In light of the complex general
condition of geriatric patients, physicians are faced with several concerns. Besides
accounting for the severity of the comorbidities and geriatric syndromes in the diagnostic
process, also the assessment of the benefit/risk ratio of any drug prescription can be
particularly difficult. Although the potential benefits of pharmacological therapy are
unquestionable, the hazards and negative outcomes of medications in older people are also
relevant issues.
In this context the definition of “Inappropriate Prescribing” (IP) in the geriatric population
is still debated. In fact, the aim of avoiding those medications whose risks outweigh their
benefits in the elderly patient has stimulated the development of different criteria to
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 116
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
identify inappropriate use of pharmacotherapy [2-5]. However, these criteria have not been
implemented for the Italian geriatric population and their impact has not been exhaustively
validated towards ‘hard' end-points [6-8].
Aims
The Project aims to (i) identify the main IP indicators among elderly patients who suffer
from cardiovascular diseases and other chronic comorbidities; (ii) evaluate the relationship
between inappropriate treatment and ‘hard' end-points (one-year acute cardiovascular
events/all-cause hospitalization/all-cause mortality) in cohorts of institutionalized and
community-dwelling elderly patients.
Methods
This study will include patients of 65+ year-old, affected by cardiovascular disease and
other chronic comorbidities, enrolled from three different settings: (i) hospital admissions,
(ii) general practice and (iii) nursing homes. Data of each setting are recruited, respectively,
from (i) HCU databases of Lazio, Tuscany, Lombardy Regions of around 4 million older
individuals (please see below the list of participants to the AIFA-ELDERLY project), (ii) a
repository of 225 General Practitioners from Caserta of almost 20,000 older individuals
(iii) a database of about 30 nursing home of Umbria, Lazio, Puglia of about 2,000
institutionalized patients. All patients will be followed until the end of 2012.
Exposure to IP medications will be defined on the basis of a systematic review of the
medical literature and consensus meetings, as suggested by CRIME methodology [9]. The
outcome evaluations related to IP treatments will be investigated in terms of (i) acute
cardiovascular events; (ii) all-cause hospitalization; (iii) all-cause mortality.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 117
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Impact
Improving knowledge on the inappropriate prescribing among older individuals who suffer
from cardiovascular disease and concurrent chronic morbidities could help physicians in
the choice of medications for these patients. In addition the new IP indicators could be used
to design and implement effective medication utilization programs aimed to influence
prescribing patterns.
Management
Project manager (epidemiology and biostatistics): Giovanni Corrao,
Antonella Zambon1
Data management and analysis: Arianna Ghirardi2
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
List of participants to the AIFA-ELDERLY project
•
•
•
•
•
•
•
University of Florence ( Italy): Alessandro Mugelli, Francesco Lapi, Ersilia Lucenteforte, Alfredo
Vannacci, Roberto Bonaiuti
Research Institute San Raffaele Pisana of Rome ( Italy): Cristiana Vitale, Stefano Bonassi
Catholic University “Sacro Cuore” of Rome (Italy): Roberto Bernabei, Graziano Onder, Federica
Mammarella
Local Health Authority Rome ( Italy): Marina Davoli, Ursula Kirchmayer, Silvai Cascini
Regional Agency for Healthcare Services of Tuscany (Italy): Francesco Cipriani, Giampiero
Mazzaglia, Emiliano Sessa
University of Milano-Bicocca (Italy): Giovanni Corrao, Antonella Zambon, Arianna Ghirardi
University of Messina (Italy): Achille Patrizio Caputi, Vincenzo Arcoraci, Gianluca Trifirò
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 118
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
References
[1] Maggi S, Noale M, Gallina P, et al; ILSA Working Group. Metabolic syndrome, diabetes, and
cardiovascular disease in an elderly Caucasian cohort: the Italian Longitudinal Study on Aging. J Gerontol A
Biol Sci Med Sci 2006;61:505-10
[2] Fick DM, Cooper JW, Wade WE, et al. Updating the Beers criteria for potentially inappropriate
medication use in older adults: results of a US consensus panel of experts. Arch Intern Med 2003;163:271624
[3] Hanlon JT, Schmader KE, Samsa GP, et al. A method for assessing drug therapy appropriateness. J Clin
Epidemiol 1992;45:1045-51
[4] Gallagher PF, Barry PJ, Ryan C, et al. Inappropriate prescribing in an acutely ill population of elderly
patients as determined by Beers' Criteria. Age Ageing 2008;37:96-101
[5] Laroche ML, Charmes JP, Merle L. Potentially inappropriate medications in the elderly: a French
consensus panel list. Eur J Clin Pharmacol 2007;63:725-31
[6] Maio V, Yuen EJ, Novielli K, et al. Potentially inappropriate medication prescribing for elderly
outpatients in Emilia Romagna, Italy: a population-based cohort study. Drugs Aging 2006;23:915-24
[7] Fialová D, Topinková E, Gambassi G, et al; AdHOC Project Research Group. Potentially inappropriate
medication use among elderly home care patients in Europe. JAMA 2005;293:1348-58
[8] Jano E, Aparasu RR. Healthcare outcomes associated with beers' criteria: a systematic review. Ann
Pharmacother 2007;41:438-47
[9] Fusco D, Lattanzio F, Tosato M, et al. Development of CRIteria to assess appropriate Medication use
among Elderly complex patients (CRIME) project: rationale and methodology. Drugs Aging 2009;26 Suppl
1:3-13
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 119
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.8. Paediatrics
Project CRACKPD1
Hospitalization for pneumonia and empyema: incidence and association
with pneumococcal conjugate vaccines and non-steroidal
antiinflammatory drugs
Background
Large randomized clinical trials (RCTs) have shown that pneumococcal conjugate vaccines
protect against pneumonia [1–4].
Following the 7-valent pneumococcal conjugate vaccine (PCV7) introduction in the United
States, rates of pneumonia hospitalizations and ambulatory visits among children aged <2
years decreased through 2004 [5–7]. However, invasive pneumococcal disease caused by
serotypes not covered by PCV7 has increased during recent years [8].
Empyema is a relatively rare complication of pneumonia [9]. Up to half of patients who
suffer from pneumonia will develop a parapneumonic effusion, which can be secondarily
infected and characterized by loculation and reduced pleural fluid pH and glucose
(complicated parapneumonic effusion). Empyema refers to the extreme end of that
spectrum, and is defined by the presence of pus or bacteria in the pleural cavity [10].
Marked increases in the incidence of paediatric empyema have been reported by a number
of North American, Australian, European and Asian studies [9,11-18]. The reasons for this
increase are unknown but may be related to invasive pneumococcal disease (meningitis,
bacteraemia, and pneumonia) caused by emergent non-vaccine replacement serotypes,
particularly serotypes 1, 3, and 19A after the introduction of PCV7 [19–25]. However, this
theory is controversial because several studies identified an increase in empyema incidence
before the introduction of PCV7 [13,26-28]. The recent introduction of the 13-valent
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 120
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
pneumococcal conjugate vaccine (PCV13) offers the opportunity of investigating the
impact of the new vaccine on the incidence of paediatric empyema.
Among the risk factors of empyema, the use of nonsteroidal antiinflammatory drugs
(NSAID), mainly of ibuprofen, has received growing attention [29-31].
Aims
This study aims (i) to measure the prevalence of children vaccinated with pneumococcal
conjugate PCV7 and PCV13 during the years from 2007 to 2012; (ii) to measure the
temporal trend in incidence of hospitalization for paediatric pneumonia and empyema
during the period 2007-2012; (iii) to investigate risk factors of paediatric pneumonia and
empyema including use of pneumococcal conjugate vaccines PCV7 and PCV13, and drugs
such as NSAIDs, among others; (iv) to explore the safety profile of both PCV7 and PCV13
vaccines.
Methods
Data will be retrieved from the Certificate of Delivery Care (CEDAP – CErtificato di
Assistenza al Parto) archive concerning deliveries occurred in Lombardy Region in the
years 2007-2011. The CEDAP certificate must be completed by the midwife, or
supervising physician, at birth. Socio-demographic, anthropometric and clinical data of the
mother, father and child born are recorded.
Five birth-cohort concerning children born in the years 2007, ‘08, ’09, ’10 and ’11 will be
constituted. Data on exposure to pneumococcal conjugate vaccines, antibiotics and
NSAIDs, as well as hospitalization and access to emergency room with diagnosis of
pneumonia and empyema experienced by each cohort member from delivery until
December, 31 2012 will be drawn out from the specific regional databases (i.e. vaccination
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 121
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
register, outpatient drug prescription, hospital discharge and emergency room databases).
Additional information on exposure to other drugs and on the hospitalization with other
selected diagnoses will be defined in a subsequent stage and will be retrieved from the
same sources of data. The annual rate of vaccination with pneumococcal conjugate
vaccines and hospitalization for pneumonia and empyema will be calculated through the
density method approach.
Several case-control studies, nested in each birth-cohort, will be carried out using common
harmonized definitions of the outcome of interest (e.g. hospitalization or access to
emergency room with diagnosis of pneumonia and empyema. In each study cases will be
children who experienced the outcome of interest during follow-up. The index date will be
defined as the date of the first outcome onset. Each event case will be matched to up to 100
controls selected from the same birth cohort on sex, age, follow-up length and index date.
Conditional logistic regression models will be fitted to estimate odds ratios (ORs), and
corresponding 95% CI, of the association between exposure to pneumococcal conjugate
vaccines, antibiotics and NSAIDs (and other drugs which will be afterwards defined) and
the outcome onset.
Sensitivity analyses will be performed by including data from PEDIANET network, to
account for measuring and confounding errors.
Management
Project manager (clinics): Gian Vincenzo Zuccotti1
Project manager (primary paediatric care): Luigi Cantarutti 2
Project manager (epidemiology): Giovanni Corrao3,4
Data management and analysis: Antonella Zambon, Arianna Ghirardi
and Anna Cantarutti4
1
2
Dept of Pediatrics, University of Milan, Luigi Sacco Hospital, Milan, Italy
Family Pediatrician Pedianet Project, Padova, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 122
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
3
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health
University of Milano-Bicocca, Milan, Italy
4
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] Black S, Shinefield H, Fireman B, Lewis E, Ray P, Hansen JR. Efficacy, safety and immunogenicity of
heptavalent pneumococcal conjugate vaccine in children. Pediatr Infect Dis J 2000;19:187-95
[2] Klugman KP, Madhi SA, Huebner RE, Kohberger R, Mbelle N, Pierce N. A trial of a 9-valent
pneumococcal conjugate vaccine in children with and those without HIV infection. N Engl J Med
2003;349:1341-8
[3] Cutts FT, Zaman SM, Enwere G, et al. Efficacy of nine-valent pneumococcal conjugate vaccine against
pneumonia and invasive pneumococcal disease in The Gambia: randomised, double-blind, placebo controlled
trial. Lancet 2005;365:1139-46
[4] Lucero MG, Nohynek H, Williams G, et al. Efficacy of an 11-valent pneumococcal conjugate vaccine
against radiologically confirmed pneumonia among children less than 2 years of age in the Philippines:a
randomized, double-blind, placebo-controlled trial. Pediatr Infect Dis J 2009;28:455-62
[5] Grijalva CG, Nuorti JP, Arbogast PG, Martin SW, Edwards KM, Griffin MR. Decline in pneumonia
admissions after routine childhood immunisation with pneumococcal conjugate vaccine in the USA: a time
series analysis. Lancet 2007;369:1179-86
[6] Zhou F, Kyaw MH, Shefer A, Winston CA, Nuorti JP. HEALTHCARE utilization for pneumonia in
young children after routine pneumococcal conjugate vaccine use in the United States. Arch Pediatr Adolesc
Med 2007; 161(12):1162–1168.
[7] Nelson JC, Jackson M, Yu O, et al. Impact of the introduction of pneumococcal conjugate vaccine on
rates of community acquired pneumonia in children and adults. Vaccine 2008;26:4947-54
[8] Hicks LA, Harrison LH, Flannery B, et al. Incidence of pneumococcal disease due to non-pneumococcal
conjugate vaccine (PCV7) serotypes in the United States during the era of widespread PCV7 vaccination,
1998–2004. J Infect Dis 2007;196:1346-54
[9] Grijalva CG, Nuorti JP, Zhu Y, Griffin MR. Increasing Incidence of Empyema Complicating Childhood
Community-Acquired Pneumonia in the United States. Clin Infect Dis 2010;50:805-13
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 123
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10] Lisboa T, Waterer GW, Lee G. Pleural infection: Changing bacteriology and its implications.
Respirology 2011;16:598–603
[11] Strachan R, Jaffé A. Australian Research Network in Empyema. A.ssessment of the burden of paediatric
empyema in Australia. J Paediatr Child Health 2009;45:431–6
[12] Byington CL, Hulten KG, Ampofo K et al. Molecular epidemiology of pediatric pneumococcal
empyema from 2001 to 2007 in Utah. J Clin.Microbiol 2010;48:520-5
[13] Li ST, Tancredi DJ. Empyema hospitalizations increased in US children despite pneumococcal conjugate
vaccine. Pediatrics 2010;125: 26-33
[14] Hendrickson DJ, Blumberg DA, Joad JP et al. Five-fold increase in pediatric parapneumonic empyema
since introduction of pneumococcal conjugate vaccine. Pediatr Infect Dis J 2008;27:1030–2
[15] Munoz-Almagro C, Jordan I, Gene A et al. Emergence of invasive pneumococcal disease caused by
nonvaccine serotypes in the era of 7-valent conjugate vaccine. Clin Infect Dis 2008;46:174-82
[16] Goldbart AD, Leibovitz E, Porat N et al. Complicated community acquired pneumonia in children prior
to the introduction of the pneumococcal conjugated vaccine. Scand J Infect Dis 2009;41:182-7
[17] Finley C, Clifton J, Fitzgerald JM et al. Empyema: an increasing concern in Canada. Can Respir J
2008;15: 85–9
[18] Roxburgh CS, Youngson GG, Townend JA et al. Trends in pneumonia and empyema in Scottish
children in the past 25 years. Arch Dis Child 2007;93:316–8
[19] Byington CL, Korgenski K, Daly J, Ampofo K, Pavia A, Mason EO. Impact of the pneumococcal
conjugate vaccine on pneumococcal parapneumonic empyema. Pediatr Infect Dis J 2006;25:250–4
[20] Calbo E, Diaz A, Canadell E, Fabrega J, Uriz S, Xercavins M, et al. Invasive pneumococcal disease
among children in a health district of Barcelona: early impact of pneumococcal conjugate vaccine. Clin
Microbiol Infect 2006;12:867-72
[21] Fletcher M, Leeming J, Cartwright K, Finn A; South West of England Invasive Community Acquired
Infection Study Group. Childhood empyema: limited potential impact of 7-valent pneumococcal conjugate
vaccine. Pediatr Infect Dis J 2006;25:559-60
[22] Obando I, Arroyo LA, Sanchez-Tatay D, Moreno D, Hausdorff WP, Brueggemann AB. Molecular
typing of pneumococci causing parapneumonic empyema in Spanish children using multilocus sequence
typing directly on pleural fl uid samples. Pediatr Infect Dis J 2006;25:962–3
[23] Roxburgh CS, Youngson GG, Townend JA, Turner SW. Trends in pneumonia and empyema in Scottish
children in the past 25 years. Arch Dis Child 2008;93:316–8
[24] Grijalva CG, Nuorti JP, Zhu Y, Griffi n MR. Increasing incidence of empyema complicating childhood
community-acquired pneumonia in the United States. Clin Infect Dis 2010;50:805-13
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 124
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[25] Hendrickson DJ, Blumberg DA, Joad JP, Jhawar S, McDonald RJ. Five-fold increase in pediatric
parapneumonic empyema since introduction of pneumococcal conjugate vaccine. Pediatr Infect Dis J
2008;27:1030–2
[26] Playfor SD, Smyth AR, Stewart RJ. Increase in incidence of childhood empyema. Thorax 1997;52:932
[27] Eastham KM, Freeman R, Kearns AM, Eltringham G, Clark J, Leeming J, et al. Clinical features,
aetiology and outcome of empyema in children in the north east of England. Thorax 2004;59:522–5
[28] Singleton RJ, Hennessy TW, Bulkow LR, Hammitt LL, Zulz T, Hurlburt DA, et al. Invasive
pneumococcal disease caused by nonvaccine serotypes among Alaska Native children with high levels of 7valent pneumococcal conjugate vaccine coverage. JAMA 2007;297:1784–92
[29] Voiriot G , Dury S , Parrot A, et al. Nonsteroidal antiinflammatory drugs may affect the presentation and
course of community-acquired pneumonia. Chest 2011;139:387-94
[30] Byington CL , Spencer LY, Johnson TA, et al. A n epidemiological investigation of a sustained high rate
of pediatric parapneumonic empyema: risk factors and microbiological associations. Clin Infect Dis
2002;34:434-40
[31] François P, Desrumaux A , Cans C, et al. Prevalence and risk factors of suppurative complications in
children with pneumonia. Acta Paediatr 2010;99:861-6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 125
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.9. Environmental health
Project CRACKEH1
Epidemiology of idiopathic pulmonary fibrosis (IPF) in Lombardy region
and its relationship to air pollution
Background
Idiopathic pulmonary fibrosis (IPF) is a chronic and progressive fibrotic lung disease with a
severe prognosis and a median survival of 2-5 years from the time of diagnosis. The
epidemiological study of IPF is very challenging, as the pathology is a rare disease and
only few research projects have contributed to the limited available literature [1].
Over the last 6 years, a few studies conducted in the United States estimated an incidence
rate (per 100,000 person-years) of IPF between 6.8 and 16.3 in 2000[2] and between 8.8
and 17.4 in 1997-2005[3]. As regards the prevalence, the ratio (per 100,000 persons) was
between 14.0 and 42.7 in 2000 and it increased up to 37.9-63.0 in 2005 [2,3]. The number
of new cases was expected to increase over time, doubling by 2030 on the basis of agespecific incidence rates and projections of the future demographic characteristics of the
population [3]. In Europe, a similar temporal trend in incidence rate was observed in UK
and Czech Republic [4-7], although the increase was less marked than in the US, while in
Norway IPF incidence appeared to be stable [6].
Some studies have checked on the final stages of the disease, assessing survival, cause of
death and mortality rate [2, 3, 8-10]. In detail, estimated median survival was between 3.5
and 4.4 years [3] in the US and 3.1 years in the UK [10], while the cause of death was often
the disease itself [3]. Mannino and Olson estimated the mortality rate of IPF (per 1 million
population) from 1979 to 2003 in US using the death certificate records [8-9]. During the
25 year-period, the age-adjusted mortality rates increased significantly in both sexes: in
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 126
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
1979 the rates were 48.6 and 21.4 in male and female respectively and in 2003 they
increased up to 61.2 and 54.5. Indeed, the growing trends observed in mortality and
incidence can be affected by diagnostic transfer and greater awareness, which might have
played a role since the ‘90s, when the high resolution computed tomography has been
introduced as a routinely used diagnostic tool [1, 2].
Analyses on subgroups are consistent in identifying the elderly as the most affected [2-3, 8,
10] and with poorest prognosis [11] demographic sub-group; furthermore, men have been
shown to be more susceptible to IPF [2-5, 10], even if the increase in IPF mortality seems
to be more pronounced among women [9]. Differences have been detected among races [8]
and geographical areas [8, 10], but the evidence is not sufficient to come to an exhaustive
conclusion.
The epidemiology of IPF may be affected by environmental factors: cigarette smoking and
occupational exposure to metal and wood dusts seems to be associated with increased
incidence of the disease [2,12]; exposure to viruses [2], as EBV, and to drugs
(chemotherapics, anti-arrhythmics, antibiotics, anticonvulsants and anti-inflammatory
agents) [1] may be further risk factors.
So far, in Italy only two research groups, back in the ‘90s, studied IPF and they focused on
its clinical aspects [13-15]; as a consequence, information about the epidemiology of the
disease and its relationship with environmental factors is still lacking. Given the fact that
IPF is a rare disease, healthcare administrative databases can be a useful tool to bridge the
gap, as they collect information on large populations thus increasing the probability of
detecting disease cases.
Aims
The main purpose of the study is to assess the incidence and prevalence of IPF in
Lombardy and to analyze the progress of this disease, in terms of mortality and relapses.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 127
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
The secondary aim is to evaluate the existence of a long-term association between
environmental factors and this pathology.
Methods
In order to achieve the aforementioned purposes, data will be obtained from the data
warehouse DENALI, which collects and organizes the healthcare administrative datasets of
the Lombardy Regional Health System (demographic characteristics, hospital discharges,
outpatient claims and pharmaceutical prescriptions).
Incident and prevalent cases of IPF will be identified through their accesses to the health
services from 2000 to 2010, namely hospital admissions and outpatient visits performed in
Lombard hospitals and reporting a diagnosis of IPF, and their demographic characteristics
will be analyzed. Annual incidence rates of IPF will be calculated both for the whole region
and separately for different regional areas, mapping them in order to identify potential
spatial clusters [16] within the region. Our estimates will be then compared with those
reported in literature.
For each subject with IPF information on hospital discharges, outpatient claims and
pharmaceutical prescriptions following the diagnosis will be gathered and survival analysis
will be applied to estimate the probabilities of mortality and hospitalization during followup.
In order to assess the possible long-term effects of environmental factors on subjects with
IPF, health data will be integrated with information on exposure to environmental pollution
before the onset [17]. Environmental data will be provided by the Regional Agency for
Environmental Protection of Lombardy (data available at http://www.arpalombardia.it).
The relationship between IPF and air pollution will be analyzed using the most suitable
statistical model among the following: Cox regression model [18], mixed model [19] and
generalized estimating equation regression [20].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 128
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Sergio Harari2
Data management and analysis: Fabiana Madotto and Sara Conti3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
3
Dept of Pneumology, San Giuseppe Hospital, Milano, Italy
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
Demedts M, Wells AU, Antó JM, et al. Interstitial lung diseases: an epidemiological overview. Eur
Respir J Suppl 2001;32:2s-16s
[2]
Raghu G, Weycker D, Edelsberg J, et al. Incidence and prevalence of idiopathic pulmonary fibrosis.
Am J Respir Crit Care Med 2006;174:810-6
[3]
Fernández Pérez ER, Daniels CE, Schroeder DR, et al. Incidence, prevalence, and clinical course of
idiopathic pulmonary fibrosis: a population-based study. Chest 2010;137:129-37
[4]
Gribbin J, Hubbard RB, Le Jeune I, et al. Incidence and mortality of idiopathic pulmonary fibrosis and
sarcoidosis in the UK. Thorax 2006;61:980-5
[5]
Kolek V. Epidemiology of cryptogenic fibrosing alveolitis in Moravia and Silesia. Acta Univ Palacki
Olomuc Fac Med 1994;137:49-50
[6]
Von Plessen C, Grinde O, Gulsvik A. Incidence and prevalence of cryptogenic fibrosing alveolitis in a
Norwegian community. Respir Med 2003;97:428-35
[7]
Hodgson U, Laitinen T, Tukiainen P. Nationwide prevalence of sporadic and familial idiopathic
pulmonary fibrosis: evidence of founder effect among multiplex families in Finland. Thorax 2002;57:338-42
[8]
Mannino DM, Etzel RA, Parrish RG. Pulmonary fibrosis deaths in the United States, 1979-1991. An
analysis of multiple-cause mortality data. Am J Respir Crit Care Med 1996;153:1548-52
[9]
Olson AL, Swigris JJ, Lezotte DC, et al. Mortality from pulmonary fibrosis increased in the United
States from 1992 to 2003. Am J Respir Crit Care Med 2007;176:277-84
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 129
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10]
Navaratnam V, Fleming KM, West J, et al. The rising incidence of idiopathic pulmonary fibrosis in the
U.K. Thorax 2011; 66:462-7
[11]
Ley B, Collard HR, King TE Jr. Clinical course and prediction of survival in idiopathic pulmonary
fibrosis. Am J Respir Crit Care Med 2011;183:431-40
[12]
Hubbard R, Lewis S, Richards K, et al. Occupational exposure to metal or wood dust and aetiology of
cryptogenic fibrosing alveolitis. Lancet. 1996;347:284-9
[13]
Thomeer MJ, Costabe U, Rizzato G, et al. Comparison of registries of interstitial lung diseases in three
European countries. Eur Respir J Suppl 2001;32:114s-118s
[14]
Rizzato G, Bariffi F. Inchiesta epidemiologica sulle interstiziopatie polmonari in Italia: dati e risultati a
macchia di leopardo. L’Internista 1999;7:20-4
[15]
Agostini C, Albera C, Bariffi F,et al. First report of the Italian register for diffuse infiltrative lung
disorders (RIPID). Monaldi Arch Chest Dis 2001;56:364-8
[16]
Elliott P, Wartenberg D. Spatial epidemiology: current approaches and future challenges. Environ
Health Perspect 2004;112:998-1006
[17]
Biggeri A, Bellini P, Terracini B. Meta-analysis of the italian studies on short-term effects of air
pollution 1996-2002. Epidemiol Prev 2004;28:4-100
[18]
Turner MC, Krewski D, Pope CA 3rd, et al. Long-term ambient fine particulate matter air pollution
and lung cancer in a large cohort of never-smokers. Am J Respir Crit Care Med 2011;184:1374-81
[19]
Schwartz J, Alexeeff SE, Mordukhovich I, et al. Association between long-term exposure to traffic
particles and blood pressure in the Veterans Administration Normative Aging Study. Occup Environ Med
2012;69:422-7
[20]
Weuve J, Puett RC, Schwartz J, et al. Exposure to particulate air pollution and cognitive decline in
older women. Arch Intern Med 2012;172:219-27
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 130
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKEH2
Cardioresopiratory diseases and environmental exposure: particulate
matter toxicity and molecular risk markers (TOSCA)
Background
Given its high population density and industrialization, in concurrence with unfavorable
climatic conditions, the Po Valley and especially the metropolitan area of Milan are
continuously facing air quality issues and the need for interventions to reduce toxic
emissions into the atmosphere. During the last decades, though there was a lowering of
industrial pollution due to both the improvement of productive processes and the decrease
of heavy industry, the area was characterized by high concentrations of airborne particulate
matter (PM), one of the criteria pollutant identified by the Clean Air Act [1], therefore
leading to the conception of the multidisciplinary project TOSCA (Particulate matter
toxicity
and
molecular
risk
markers,
http://www.polaris.unimib.it/index.php/it/progetti-di-ricerca),
details
funded
available
by
at
CARIPLO
foundation and involving research groups with expertise in biological, clinical and
epidemiological fields. The main aim of the TOSCA was to evaluate the toxicity of PM in
an urban environment, especially focusing on the area of Milan, and the “Laboratory of
Epidemiology and Healthcare Research” contributed with an investigation of the
epidemiological implication of the exposure.
The choice of focusing on PM descends from the evidences gathered through decades of
research, which have identified it as an important and modifiable determinant of respiratory
and cardiovascular diseases [2-3]. Exposure to PM has been shown to induce the activation
of alveolar macrophages [4-5], mediated by reactive oxygen species (ROS) [6] and calcium
[7], to diminish the clearance of activated macrophages [8], and to cause damage of the
respiratory epithelium [9]. Main consequences are asthma exacerbation, especially in
children, and worsening of chronic obstructive pulmonary disease (COPD) and pneumonia
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 131
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10-11]. The systemic inflammatory response and the production of ROS have been related
also to atherogenesis, plaque destabilization and rupture, which causes acute cardiovascular
and cerebrovascular events, such as myocardial infarction and stroke [12-16]. Mediators of
the same process have been identified responsible of vessel and cardiac remodeling [1718]. Finally, exposure to PM has been associated with disorders of autonomic function of
the vessels, like acute vasoconstriction and arterial blood pressure changes, and of the
heart, including increased heart rate, decreased heart variability, increased electrical
instability and increased cardiac arrhythmias [19-23].
Since many epidemiological studies showed that high exposure to PM are associated with a
short-term increase in mortality and morbidity (in terms of hospital admissions) for many
health outcomes, such as asthma, COPD, respiratory tract infections, cerebrovascular
diseases, ischemic heart diseases, heart failure and arrhythmias [24-32], it was significant
to similarly investigate the relationship between exposure to PM and cardiorespiratory
health in the Po Valley.
Aims
The main aim of the epidemiological study conducted within TOSCA project was to
evaluate the short-term cardiorespiratory health effects of particulate matter (PM10) on a
sample (around 500,000 subjects) of the Lombard population.
The analysis was structured in three sections, with the purpose of exploring the relationship
from three different perspectives: the effect of exposure to PM10 on the daily frequency of
hospitalization due to cardiorespiratory causes, the effect of exposure to PM10 on the daily
consumption of cardiorespiratory drugs, and the evaluation of pharmacological treatments
(cardiovascular or respiratory) as effect modifiers of the relationship between PM10 and
cardiorespiratory hospitalizations.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 132
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Methods
We conducted an ecological study on the population domiciled in the cities of Bergamo,
Lodi, Mantova, Monza, Sesto San Giovanni, Saronno and Sondrio.
Healthcare data were obtained from the data warehouse DENALI, which gathers and links
different healthcare administrative databases (DB) on health services delivered to all
Lombardy residents, together with vital status and demographic information. From
DENALI, we extracted all emergency hospitalizations occurred in subjects domiciled in the
seven cities of interest and reporting a diagnosis of respiratory or cardiovascular disease.
For the same population, we furthermore obtained information on all prescriptions of some
selected respiratory and cardiovascular treatments.
Environmental data, namely meteoclimatic and air quality data, were provided from the
Regional Environmental Protection Agency of Lombardy, which manages a network of
fixed environmental monitoring stations (data available at http://www.arpalombardia.it)
distributed all over the regional territory.
The relationship between exposure to PM10 and hospital admissions or drug consumption
was investigated using both a time-series [33] and a time-stratified case-crossover approach
[34-37], respectively fitting semiparametric generalized additive models [38-39] and
conditional logistic regression [40]. The modification of the effect of PM10 on hospital
admissions, due to pharmacological treatment preceding the admission itself, was
examined through a case-only analysis, which required the building of a logistic regression
model.
Immediate and delayed effects of the pollutant were examined through single and
distributed lag models [41].
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Giovanni De Vito2
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 133
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data management and analysis: Sara Conti3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
3
Dept of Pneumology, San Giuseppe Hospital, Milano, Italy
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
Clean Air Act, EPA, http://www.epa.gov/air/caa/
[2]
Bernstein JA, Alexis N, Barnes C, et al. Health effects of air pollution. J Allergy Clin Immunol
2004;114:1116-23
[3]
Brunekreef B, Holgate ST. Air pollution and Health. Lancet 2002;360:1233-42
[4]
Driscoll KE, Maurer JK, Higgins J, et al. Alveolar macrophage cytokine and growth factor production
in a rat model of crocidolite-induced pulmonary inflammation and fibrosis. J Toxicol Environ Health
1995;46:155–69
[5]
Bouthillier L, Vincent R, Goegan P, et al. Acute effects of inhaled urban particles and ozone: lung
morphology, macrophage activity, and plasma endothelin-1. Am J Pathol 1998;153:1873-84
[6]
MacNee W, Donaldson K. Mechanism of lung injury caused by PM10 and ultrafine particles with
special reference to COPD. Eur Respir J Suppl 2003;40:47s-51s
[7]
Brown DM, Donaldson K, Stone V. Effects of PM10 in human peripheral blood monocytes and J774
macrophages. Respir Res 2004;5:29
[8]
Brown JS, Zeman KL, Bennett WD. Ultrafine particle deposition and clearance in the healthy and
obstructed lung. Am J Respir Crit Care Med 2002;166:1240-7
[9]
Gualtieri M, Mantecca P, Corvaja V, et al. Winter fine particulate matter from Milan induces
morphological and functional alterations in human pulmonary epithelial cells (A549). Toxicol Lett
2009;188:52-62
[10]
Delfino RJ, Quintana PJ, Floro J, et al. Association of FEV1 in asthmatic children with personal and
microenvironmental exposure to airborne particulate matter. Environ Health Perspect 2004;112:932-41
[11]
Donaldson K, Gilmour I, MacNee W. Asthma and PM10. Respir Res 2000;1:12–5
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 134
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[12]
Donaldson K, Stone V, Seaton A, et al. Ambient particle inhalation and the cardiovascular system:
potential mechanisms. Environ Health Perspect 2001; 09(Suppl 4):523-7
[13]
Frampton MW. Systemic and cardiovascular effects of airway injury and inflammation: ultrafine
particle exposure in humans. Envrion Health Perspect 2001;109:529-32
[14]
Mills NL, Donaldson K, Hadoke PW, et al. Adverse cardiovascular effects of air pollution. Nat Clin
Pract Cardiovasc Med 2009;6:36-44
[15]
Dockery DW. Epidemiologic evidence of cardiovascular effects of particulate air pollution. Environ
Health Perspect 2001;109(Suppl 4):483-6
[16]
Bai N, Khazaei M, van Eeden SF, et al. The pharmacology of particulate matter air pollution-induced
cardiovascular dysfunction. Pharmacol Ther 2007;113:16-29
[17]
Ying Z, Yue P, Xu X, et al. Air pollution and cardiac remodeling: a role for RhoA/Rho-kinase. Am J
Physiol Heart Circ Physiol 2009;296:H1540-50
[18]
Baccarelli A, Cassano PA, Litonjua A, et al. Cardiac autonomic dysfunction: effects from particulate
air pollution and protection by dietary methyl nutrients and metabolic polymorphisms. Circulation
2008;117:1802-9
[19]
Ren C, Park SK, Vokonas PS, et al. Air pollution and homocysteine: more evidence that oxidative
stress-related genes modify effects of particulate air pollution. Epidemiology 2010;21:198-206
[20]
Brook RD, Brook JR, Urch B, et al. Inhalation of fine particulate air pollution and ozone causes acute
arterial vasoconstriction in healthy adults. Circulation 2002;105:1534-6
[21]
Bartoli CR, Wellenius GA, Diaz EA, et al. Mechanisms of inhaled fine particulate air pollution-
induced arterial blood pressure changes. Environ Health Perspect 2009;117:361-6
[22]
Chan CC, Chuang KJ, Shiao GM, et al. Personal exposure to submicrometer particles and heart rate
variability in human subjects. Environ Health Perspect 2004;112:1063-7
[23]
Zanobetti A, Stone PH, Speizer FE, et al. T-wave alternans, air pollution and traffic in high-risk
subjects. Am J Cardiol 2009;104:665-70
[24]
Dominici F, Peng RD, Bell ML, et al. Fine particulate air pollution and hospital admission for
cardiovascular and respiratory diseases. JAMA 2006;295:1127-34
[25]
Zanobetti A, Schwartz J. Air pollution and emergency admissions in Boston, MA. J Epidemiol
Community Health 2006;60:890–5
[26]
Zanobetti A, Schwartz J, Dockery DW. Airborne particles are a risk factor for hospital admissions for
heart and lung disease. Environ Health Perspect 2000;108:1071-7
[27]
Zanobetti A, Schwartz J. The effect of particulate air pollution on emergency admissions for
myocardial infarction: a multicity case-crossover analysis. Environ Health Perspect 2005;113:978-82
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 135
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[28]
Medina-Ramón M, Zanobetti A, Schwartz J. The effect of ozone and PM10 on hospital admissions for
pneumonia and chronic obstructive pulmonary disease: a national multicity study. Am J Epidemiol
2006;163:579-88
[29]
Wellenius GA, Schwartz J, Mittleman MA. Particulate air pollution and hospital admissions for
congestive heart failure in seven United States cities. Am J Cardiol 2006;97:404-8
[30]
Vedal S, Rich K, Brauer M, et al. Air pollution and cardiac arrhythmias in patients with implantable
cardioverter defibrillators. Inhal Toxicol 2004;16:353-62
[31]
Rich KE, Petkau J, Vedal S, et al., A case-crossover analysis of particulate air pollution and cardiac
arrhythmia in patients with implantable cardioverter defibrillators. Inhal Toxicol 2004;16:363-72
[32]
Peters A, Liu E, Verrier RL, Schwartz J, et al. Air pollution and incidence of cardiac arrhythmia.
Epidemiology 2000;11:11-7
[33]
Schwartz J, Spix C, Touloumi G, et al. Methodological issues in studies of air pollution and daily
counts of deaths or hospital admissions. J Epidemiol Community Health 1996;50:S3-11
[34]
Bateson TF, Schwartz J. Control for seasonal variation and time trend in case-crossover studies of
acute effects of environmental exposures. Epidemiology.1999;10:539-44
[35]
Lumley T, Levy D. Bias in the case – crossover design: implications for studies of air pollution.
Environmetrics 2000;11:689–704
[36]
Maclure M. the case-crossover design: a method for studying transient effects on the risk of acute
events. Am J Epidemiol 1991;133:144-53
[37]
Maclure M, Mittleman MA. Should we use a case-crossover design? Annu Rev Public Health
2000;21:193-221
[38]
Hastie T, Tibshirani R. Generalized Additive Models. Statistical Science 1986;1:297-318
[39]
Hastie TJ , Tibshirani RJ, The Generalized Additive Models, Chapman et Hall, 1990
[40]
Marshall RJ, Jackson RT. Analysis of case-crossover designs. Stat Med 1993;12:2333-41
[41]
Schwartz J. The distributed lag between air pollution and daily deaths. Epidemiology 2000;11:320-6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 136
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKEH3
Short-term effect of the exposure to particulate matter on hospitalizations
and pharmaceutical prescriptions in Lombardy
Background
Since the 1970’s, when the Clean Air Act was promulgated [1], the scientific community
devoted great energy to the assessment of the health risks and to quantify the effects of the
exposure to global air pollution and to specific pollutants, such as particulate matter (PM)
[2-5].
In detail, the short-term health effects of PM, that is the consequences attributable to
pointwise exposures to high concentrations of pollutant, have been widely investigated and
include the exacerbation of respiratory distress and cardiovascular damage [6-7]: evidences
from toxicological studies underline the effects of PM on oxidative stress and inflammatory
response [8-10]; while several epidemiological studies, both Italian [11-16] and
international [17-21], support the existence of a relationship between the inhalation of
polluted air and the rate of hospital admission, or mortality, due to cardiorespiratory
diseases. It should be however noted that most of the epidemiological studies focused on
urban areas, but it might be significant to explore the effects outside those areas, in order to
highlight potential differences in the pollutant effect.
Furthermore, even if health outcomes as hospital admissions and mortality seem to
adequately identify acute events potentially caused by a short-term exposure to PM, they
only describe severe outcomes, usually concerning a relatively small susceptible
subpopulation [22], and neglect mild effects of the exposure, which have been rarely
examined through alternative outcomes such as increased symptoms, emergency room
visits and outpatient clinic visits [23-28]. A few panel studies used drug consumption as an
health indicator of such mild outcomes: they investigated the association between PM
concentration and the frequency of prescription or sale of drugs to control respiratory
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 137
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
symptoms, such as caught and bronchospasm, for instance beta-agonists [29-34],
bronchodilators and inhalant corticosteroids [35-36]. All of these studies considered highly
selected sample of subjects, usually asthmatic children, and their results are therefore not
generalizable to the whole population. To our knowledge, only two population-based
studies were conducted in this field: one in Italy [37] and one in Alaska [38], with the least
one considering only people less than 20 years old; consistent evidences are consequently
still lacking. Moreover, only respiratory treatments were investigated, but it would be
suitable to analyze cardiovascular treatment such as antiarrhythmics and arterial
vasodilators, because they are almost exclusive treatments of arrhythmic episodes and
ischemic heart disease, which have been shown to be exacerbated by exposure to PM
[21,39].
In conclusion, it is desirable to gain further insight into mild health effects of PM, because
they may concern a much wider population as compared with severe effects. Drug
consumption, as traced in health administrative databases, might be a good surrogate for
the outcome of interest, as it potentially detects more events than other indicators such as
emergency room visits and it is cheaply retrievable.
Aims
Aim of the present study is to investigate the relationship between short-term exposure to
particulate matter and health in Lombardy, both in urban and rural areas.
Focus of the study will be PM related cardiorespiratory morbidity, in terms of severe and
mild outcomes, respectively identified through hospital admissions and drug prescriptions.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 138
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Methods
An ecological study will be carried out on the whole population domiciled in Lombardy
during 2000-2008.
Healthcare data, namely all cardiorespiratory emergency hospitalizations and drug
prescriptions occurred during 2000-2008 in subjects domiciled within the region, together
with their vital status, will be extracted from the health administrative databases of
Lombardy Region.
Environmental data regarding daily PM concentration and meteoclimatic conditions (e.g.
temperature and relative humidity) will be provided from a regional network of
environmental monitoring stations managed by the Regional Environmental Protection
Agency of Lombardy (http://www.arpalombardia.it). Concentrations will be eventually
modeled or averaged in order to obtain the exposure throughout the region, even where a
fixed monitor is not available.
The association between exposure to PM and hospital admissions or drug consumption will
be explored using two different approaches: time-series [40], which focuses on aggregate
data and explores the relationship between the variation of the daily frequency of hospital
admissions or drug prescription and the exposure to PM, and time-stratified case-crossover
approach [41-44], which investigates the association between the variation of the individual
risk of event (hospitalization or prescription) and the exposure to PM. Immediate and
delayed effects of the pollutant will be examined through single and distributed lag models
[45].
Finally, a comparison between the results originated by the two approaches will be carried
out in order to highlight differences and similarities.
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Sergio Harari2
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 139
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data management and analysis: Sara Conti3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
3
Dept of Pneumology, San Giuseppe Hospital, Milano, Italy
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
Clean Air Act, EPA, http://www.epa.gov/air/caa/
[2]
Dockery DW, Pope CA 3rd, Xu X, et al. An association between air pollution and mortality in six U.S.
cities. N Engl J Med 1993;329:1753-9
[3]
Katsouyanni K, Touloumi G, Spix C, et al. Short-term effects of ambient sulphur dioxide and
particulate matter on mortality in 12 European cities: results from time series data from the APHEA project.
Air Pollution and Health: a European Approach. Br Med J 1997;314:1658-63
[4]
Pope CA. Epidemiology of fine particulate air pollution and human health: biologic mechanisms and
who's at risk? Environ Health Perspect 2000;108:713–23
[5]
Samet JM, Zeger SL, Dominici F, et al. The National Morbidity, Mortality, and Air Pollution Study
(NMMAPS). Part 2. Morbidity and Mortality from Air Pollution in the United States. Boston, MA, Health
Effects Institute: 2000
[6]
Hogg JC, van Eeden S. Pulmonary and systemic response to atmospheric pollution, Respirology
2009;14:336-46
[7]
Mills NL, Donaldson K, Hadoke PW, et al. Adverse cardiovascular effects of air pollution. Nat Clin
Pract Cardiovasc Med 2009;6:36-44
[8]
Donaldson K, Stone V, Seaton A, et al. Ambient particle inhalation and the cardiovascular system:
potential mechanisms. Environ Health Perspect 2001;109:523-7
[9]
Sinden N, Stockley R. Systemic inflammation and comorbidity in COPD: a result of ‘overspill’ of
inflammatory mediators from the lungs? Review of the evidence. Thorax 2010;65:930-6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 140
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10]
Valavanidis A, Fiotakis K, Vlachogianni T. Airborne particulate matter and human health:
toxicological assessment and importance of size and composition of particles for oxidative damage and
carcinogenic mechanisms. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 2008;26:339-62
[11]
Baccini M, Biggeri A, Grillo P, et al. Health impact assessment of fine particle pollution at the regional
level. Am J Epidemiol 2011;174:1396-405
[12]
Biggeri A, Bellini P, Terracini B, et al. Meta-analysis of the Italian studies on short-term effects of air
pollution. Epidemiol Prev 2001;25(Suppl 2):1-71
[13]
Biggeri A, Bellini P, Terracini B. Meta-analysis of the Italian studies on short-term effects of air
pollution--MISA 1996-2002. Epidemiol Prev 2004;28(Suppl 4-5):4-100
[14]
Colais P, Serinelli M, Faustini A, et al. Air pollution and urgent hospital admissions in nine Italian
cities. Results of the EpiAir Project. Epidemiol Prev 2009;33(6 Suppl 1):77-94
[15]
Nuvolone D, Balzi D, Chini M, et al. Short-term association between ambient air pollution and risk of
hospitalization for acute myocardial infarction: results of the cardiovascular risk and air pollution in Tuscany
(RISCAT) study. Am J Epidemiol 2011;174(1):63-71
[16]
Tramuto F, Cusimano R, Cerame G, et al. Urban air pollution and emergency room admissions for
respiratory symptoms: a case-crossover study in Palermo, Italy. Environ Health 2011;10:31
[17]
Daniels M, Dominici F, Samet J, et al. Estimating particulate matter-mortality dose-response curves
and threshold levels: an analysis of daily time-series for the 20 largest US cities. Am J Epidemiol
2000;152:397-406
[18]
Medina-Ramón M, Zanobetti A, Schwartz J. The effect of ozone and PM10 on hospital admissions for
pneumonia and chronic obstructive pulmonary disease: a national multicity study. Am J Epidemiol
2006;163:579-88
[19]
Wellenius G, Schwartz J, Mittleman M. Particulate air pollution and hospital admissions for congestive
heart failure in seven united states cities. The American Journal of Cardiology. 2006;97:404-8
[20]
Zanobetti A, Schwartz J, Dockery DW. Airborne particles are a risk factor for hospital admissions for
heart and lung disease. Environ Health Perspect 2000;108:1071-7
[21]
Zanobetti A and Schwartz J. The effect of particulate air pollution on emergency admissions for
myocardial infarction: a multicity case-crossover analysis. Environ Health Perspect 2005;113:978-82
[22]
Pope CA 3rd. Epidemiology of fine particulate air pollution and human health: biologic mechanisms
and who's at risk? Environ Health Perspect 2000; 108(Suppl 4):713-23
[23]
Arbex MA, de Souza Conceição GM, Cendon SP, et al. Urban air pollution and chronic obstructive
pulmonary disease-related emergency department visits. J Epidemiol Community Health 2009;63:777-83
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 141
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[24]
Chardon B, Lefranc A, Granados D, et al. Air pollution and doctors' house calls for respiratory diseases
in the Greater Paris area (2000-3). Occup Environ Med 2007;64:320-4
[25]
Chen L, Mengersen K, Tong S. Spatiotemporal relationship between particle air pollution and
respiratory emergency hospital admissions in Brisbane, Australia. Sci Total Environ 2007;373:57-67
[26]
Janssen N, Brunekreef B, Van Vliet P, et al. The relationship between air pollution from heavy traffic
and allergic sensitization, bronchial hyperresponsiveness, and respiratory symptoms in Dutch schoolchildren.
Environ Health Perspect 2003;111:1512-8
[27]
Santos UP, Terra-Filho M, Lin CA, et al. Cardiac arrhythmia emergency room visits and
environmental air pollution in Sao Paulo, Brazil. J Epidemiol Community Health 2008;62:267-72
[28]
Schwartz J, Slater D, Larson T, et al. particulate air pollution and hospital emergency room visits for
asthma in Seattle. Am J Respir Crit Care Med 1993;147:826-31
[29]
Escamilla-Nuñez MC, Barraza-Villarreal A, Hernandez-Cadena L, et al. Traffic-related air pollution
and respiratory symptoms among asthmatic children, resident in Mexico City: the EVA cohort study. Respir
Res 2008;9:74
[30]
Gent JF, Koutrakis P, Belanger K, et al. Symptoms and medication use in children with asthma and
traffic-related sources of fine particle pollution. Environ Health Perspect 2009;117:1168-74
[31]
Pope CA 3rd, Dockery DW, Spengler JD, et al. Respiratory health and PM10 pollution. A daily time
series analysis. Am Rev Respir Dis 1991;144:668-74
[32]
Roemer W, Hoek G, Brunekreef B. Pollution effects on asthmatic children in Europe, the PEACE
study. Clin Exp Allergy 2000;30:1067-75
[33]
Segala C, Fauroux B, Just J, et al. Short-term effect of winter air pollution on respiratory health of
asthmatic children in Paris. Eur Respir J 1998;11:677-85
[34]
Slaughter JC, Lumley T, Sheppard L, et al. Effects of ambient air pollution on symptom severity and
medication use in children with asthma. Ann Allergy Asthma Immunol 2003;91:346-53
[35]
Hiltermann TJ, Stolk J, van der Zee SC, et al. Asthma severity and susceptibility to air pollution. Eur
Respir J 1998;11:686-93
[36]
von Klot S, Wölke G, Tuch T, et al. Increased asthma medication use in association with ambient fine
and ultrafine particles. Eur Respir J 2002;20:691-702
[37]
Vegni FE, Castelli B, Auxilia F, Wilkinson P. Air pollution and respiratory drug use in the city of
Como, Italy. Eur J Epidemiol 2005;20:351-8
[38]
Chimonas MA, Gessner BD. Airborne particulate matter from primarily geologic, non-industrial
sources at levels below National Ambient Air Quality Standards is associated with outpatient visits for
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 142
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
asthma and quick-relief medication prescriptions among children less than 20 years old enrolled in Medicaid
in Anchorage, Alaska. Environ Res 2007;103:397-404
[39]
Dockery DW, Luttmann-Gibson H, Rich DQ, et al. Association of air pollution with increased
incidence of ventricular tachyarrhythmias recorded by implanted cardioverter defibrillators. Environ Health
Perspect 2005;113:670-4
[40]
Schwartz J, Spix C, Touloumi G, et al. Methodological issues in studies of air pollution and daily
counts of deaths or hospital admissions. Journal of Epidemiology and Community Health 2002;50:S3-S11
[41]
Bateson TF, Schwartz J. Control for seasonal variation and time trend in case-crossover studies of
acute effects of environmental exposures. Epidemiology 1999;10:539-44
[42]
Lumley T, Levy D. Bias in the case – crossover design: implications for studies of air pollution.
Environmetrics 2000;11:689-704
[43]
Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute
events. Am J Epidemiol 1991;133:144-53
[44]
Maclure M, Mittleman MA. Should we use a case-crossover design? Annu Rev Public Health
2000;21:193-221
[45] Schwartz J. The distributed lag between air pollution and daily deaths. Epidemiology 2000;11:320-6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 143
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKEH4
Exposure to particulate matter: short-term effect on stroke
hospitalizations in Milan using ARPA monitoring stations
Background
Exposure to air pollution is considered an important risk factor for mortality and morbidity
in the whole world. During 2004, urban air pollution ranked among the 10 leading causes
of death and was associated with 2.8% and 2.5% of all deaths respectively in middle and
high-income countries [1]. Increasing epidemiological evidence suggests that, among air
pollutants, atmospheric particulate matter (PM) has an important role in the aforementioned
association and, given that potentially the whole population is exposed to such a pollutant,
its adverse health effects are continuously debated [2].
Ten years after the implementation of the first European directive in the field of air quality
evaluation and management [3], in Lombardy Region, especially in the metropolitan area
of Milan, airborne PM concentrations remain high; therefore, investigation of the potential
epidemiological consequences of the exposure is still required.
Over the last decades, the vast majority of the studies which have explored adverse health
responses to PM have focused on pulmonary and cardiovascular pathologies [4-5], but
more recently researches called attention to the relationship between exposure to PM and
cerebrovascular diseases. In detail, several studies detected a positive effect of PM on the
risk of stroke, both short-term [6-17] and long-term [18-21], but there were also a few
researches which detected no significant associations [22-26]. It should be stressed that the
distinction between ischemic and haemorragic event has proved to be significant, as
association between PM and the risk of stroke appeared to be greater for ischemic accidents
[14,15]; further investigation should therefore distinguish between these two leading causes
of disease.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 144
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Given the overall inconsistency of the results, further analyses are needed to strengthen the
evidence and the city of Milan, with its air quality issues, seems to be the suitable scenario
to perform a new study.
Aim
The study has the main aim of assessing the association between short-term fluctuations in
the concentration of PM10 and hospital admissions for cerebrovascular diseases in the city
of Milan. The secondary aim is to evaluate potential differences in the effect of PM10 on
the risk of ischemic and hemorrhagic stroke.
Methods
We will carry out an observational study on the resident population of Milan during 20002007.
Demographic characteristics of the study population and health data regarding
cerebrovascular emergency hospitalizations occurred during the study period, will be
extracted from the datawarehouse DENALI, which links several health administrative
databases of Lombardy Region.
The
Regional
Environmental
Protection
Agency
of
Lombardy
(http://www.arpalombardia.it), which gathers environmental data registered through a fixed
network of monitoring stations, will provide information regarding daily PM10 and ozone
(O3) concentrations, together with meteoclimatic conditions.
A time-stratified case-crossover study design [27-28], requiring conditional logistic
regression modeling [29-30], will be used to estimate the effect of particulate air pollution
on hospitalizations for all cerebrovascular diseases and separately for subtype of ischemic
and hemorrhagic stroke.
In addition, we will assess the simultaneous effects of PM10 and O3 through a bi-pollutant
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 145
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
model restricted summer months, in line with what has been proposed in literature [31].
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Data management and analysis: Fabio Argentino and Sara Conti2
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
Narayan KM, Ali MK, Koplan JP. Global noncommunicable diseases--where worlds meet. N Engl J
Med 2010;363:1196-8
[2]
Mateen FJ, Brook RD. Air pollution as an emerging global risk factor for stroke. JAMA
2011;305:1240-1
[3]
Council Directive 1999/30/EC of 22 April 1999 relating to limit values for sulphur dioxide, nitrogen
dioxide and oxides of nitrogen, particulate matter and lead in ambient air
[4]
Brook RD, Franklin B, Cascio W, et al. Air pollution and cardiovascular disease: a statement for
healthcare professionals from the Expert Panel on population and prevention science of the American Heart
Association. Circulation 2004;109:2655-71
[5]
Miller KA, Siscovick DS, Sheppard L, et al. Long-term exposure to air pollution and incidence of
cardiovascular events in women. N Engl J Med 2007;356:447-58
[6]
Andersen ZJ, Olsen TS, Andersen KK, et al. Association between short-term exposure to ultrafine
particles and hospital admissions for stroke in Copenhagen, Denmark. Eur Heart J 2010;31:2034-40
[7]
Hong YC, Lee JT, Kim H, et al. Air pollution: a new risk factor in ischemic stroke mortality. Stroke
2002;33:2165-9
[8]
Lisabeth LD, Escobar JD, Dvonch JT, et al. Ambient air pollution and risk for ischemic stroke and
transient ischemic attack. Ann Neurol 2008;64:53-9
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 146
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[9]
Maheswaran R, Haining RP, Brindley P, et al. Outdoor air pollution and stroke in Sheffield, United
Kingdom: a small-area level geographical study. Stroke 2005;36:239-43
[10]
Oudin A, Strömberg U, Jakobsson K, et al. Estimation of short-term effects of air pollution on stroke
hospital admissions in southern Sweden. Neuroepidemiology 2010;34:131-42
[11]
Vidale S, Bonanomi A, Guidotti M, et al. Air pollution positively correlates with daily stroke
admission and in hospital mortality: a study in the urban area of Como, Italy. Neurol Sci 2010;31:179-82
[12]
Villeneuve PJ, Johnson JY, Pasichnyk D, et al. Short-term effects of ambient air pollution on stroke:
Who is most vulnerable? Sci Total Environ 2012;430:193-201
[13]
Wellenius GA, Burger MR, Coull BA, et al. Ambient air pollution and the risk of acute ischemic
stroke. Arch Intern Med 2012;172:229-34
[14]
Wellenius GA, Schwartz J, Mittleman MA. Air pollution and hospital admissions for ischemic and
hemorrhagic stroke among medicare beneficiaries. Stroke 2005;36:2549-53
[15]
Yorifuji T, Kawachi I, Sakamoto T, et al. Associations of outdoor air pollution with hemorrhagic
stroke mortality. J Occup Environ Med 2011;53:124-6
[16]
Zanobetti A, Schwartz J. The effect of fine and coarse particulate air pollution on mortality: a national
analysis. Environ Health Perspect 2009;117:898-903
[17]
Tsai SS, Goggins WB, Chiu HF, et al. Evidence for an association between air pollution and daily
stroke admissions in Kaohsiung, Taiwan. Stroke 2003;34:2612-6
[18]
Johnson JY, Rowe BH, Villeneuve PJ. Ecological analysis of long-term exposure to ambient air
pollution and the incidence of stroke in Edmonton, Alberta, Canada. Stroke 2010;41:1319-25
[19]
Lipsett MJ, Ostro BD, Reynolds P, et al. Long-term exposure to air pollution and cardiorespiratory
disease in the California teachers study cohort. Am J Respir Crit Care Med 2011;184:828-35
[20]
Miller KA, Siscovick DS, Sheppard L, et al. Long-term exposure to air pollution and incidence of
cardiovascular events in women. N Engl J Med 2007;356:447-58
[21]
Yorifuji T, Kashima S, Tsuda T, et al. Long-term exposure to traffic-related air pollution and the risk
of death from hemorrhagic stroke and lung cancer in Shizuoka, Japan. Sci Total Environ 2012;443C:397-402
[22]
Le Tertre A, Medina S, Samoli E, et al. Short term effects of particulate air pollution on cardiovascular
diseases in eight European cities. J Epidemiol Community Health 2002;56:773-9
[23]
Maheswaran R, Pearson T, Smeeton NC, et al. Outdoor air pollution and incidence of ischemic and
hemorrhagic stroke: a small-area level ecological study. Stroke 2012;43:22-7
[24]
O'Donnell MJ, Fang J, Mittleman MA, et al. Fine particulate air pollution (PM2.5) and the risk of acute
ischemic stroke. Epidemiology 2011;22:422-31
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 147
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[25]
Poloniecki JD, Atkinson RW, de Leon AP, et al. Daily time series for cardiovascular hospital
admissions and previous day’s air pollution in London, UK. Occup Environ Med 1997;54:535-40
[26]
Ueda K, Nagasawa SY, Nitta H, et al. Exposure to particulate matter and long-term risk of
cardiovascular mortality in Japan: NIPPON DATA80. J Atheroscler Thromb 2012;19:246-54
[27]
Levy D, Lumley T, Sheppard L, et al. Referent Selection in Case-Crossover Analyses of Acute Health
Effects of Air Pollution. Epidemiology 2001;12:186-92
[28]
Lumley T, Levy D. Bias in the Case-Crossover Design: Implications for studies of air pollution.
Environmetrics 2000;11:689-704
[29]
Lu Y, Zeger SL. On the equivalence of case-crossover and time series methods in environmental
epidemiology. Biostatistics 2007;8:337-44
[30]
Marshall RJ, Jackson RT. Analysis of case-crossover designs. Stat Med 1993;12:2333-41
[31]
Sarnat JA, Schwartz J, Catalano PJ, et al. Gaseous pollutants in particulate matter epidemiology:
confounders or surrogates? Environ Health Perspect 2001;109:1053-61
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 148
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.10. Drug safety
Project CRACKDS1
MEREaFAPS project: monitoring of events and adverse drug reactions
in emergency department. Evaluation of preventable reactions and costs
of Adverse Drug Reactions
Background
The adverse drug reactions (ADRs) constitute a significant burden on public health.
According to the literature the 5% of all hospital admissions are due to ADRs and the 5%
of all hospitalized patients suffer from such reactions. The ADRs are the fifth most
common cause of hospital death. To reduce the incidence of predictable and avoidable
ADRs would represent one of the most effective strategies to manage the risk of clinical
patients and to contain health care costs resulting from iatrogenic disease. In recent years,
several studies conducted in the United States and Italy have tried to assess the impact of
ADR in the accesses to the emergency department (PS): PS accesses due to adverse events
ranging between 0.86 and 5.9%, with a higher incidence in the age group over 65. From the
point of view of the adverse reactions, the PS is a good observation point. These
observations have stimulated the Pharmacovigilance Centre of Lombardy Region, to
finance a Pharmacovigilance project of the Emergency Units, with the acronym of
MEREAFaPS (Monitoring Epidemiologic Reactions and Adverse Events of Drugs in First
Aid) coordinated by Niguarda Cà Granda Hospital, in order to collect uniform,
comprehensive reports of ADRs registred in PS, conveying it in a single database. Since
2006, several PS of Lombardy (at the date 35 PS of 23 Hospitals) are involved in the
project.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 149
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
The present study can thus help to estimate the impact of adverse drug reactions collected
in PS participating in the project MEREAFaPS in the population of Lombardy, and to
assess its economic impact.
Objective
The main objectives of this project are to evaluate:
• ADR reporting rate per 1000 total and serious access in PS (ADR PS / PS accesses)
• The rate of hospitalization in PS ADR (ADR admissions in PS / PS ADR)
• comparison between hospitalization rate by ADR in PS (defined above) and the rate of
hospitalization access in PS (admissions PS / PS accesses) answers the question "who has
access to ADR in PS has a different probability of hospitalization compared to general
population in accessing PS? "
• the rates of total and serious ADRs by age (five-year classes and% severity)
• the percentage of preventable ADR and the cost of the onset of adverse reactions based
avoidable iatrogenic
• descriptive table of the fatal cases, for SOC_PT, ATC (fifth level, active substance) and
preventability
• the distribution of ADR in PS, SOC and PT for seriousness and age
• distribution of ADRs to ATC (third level) - for seriousness and age
Study Design
Observational retrospective cohort analysis on the ADR reports in MEREaFAPS database
from 1/1/2010 to 31/12/2011.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 150
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Source of data
The analysis will be conducted using the card reporting of suspected ADRs specially
designed and created for the project MEREAFaPS. ADR will be assessed in the study
period (January 1, 2010 and December 31, 2011). The variables necessary for the
development of this project which will be extracted from the DataBase MEREaFAPS are:
patient registry, adverse reaction, concurrent conditions, diseases or works carried out
earlier, previous reactions to medications, laboratory examinations and / or laboratory SM /
PA suspects, SM / PA concomitant other substances used, follow-up, data signal.
Study description
Within the database, all drugs are classified by ATC code (Anatomical Therapeutic
Chemical Classification System), while the diagnosis of PS and all clinical events were
recorded and coded in MedDRA. For patients who have not been hospitalized after a visit
to PS, the costs of PS is estimated based on the average cost of access in PS. For patients
admitted in hospital, the cost of hospitalization following an ADR will be estimated by the
data present in the SDO (Discharge sheet) linked to MEREaFAPS.
Statistical analysis
Quantitative variables are expressed as mean ± standard deviation (median and range in the
case of non-normal distribution), any comparisons will be made using Student's t test
(Mann Whitney where necessary). Qualitative variables are expressed by absolute and
relative frequencies, any comparison will be made using the chi-square test (Fisher's exact
test where appropriate). Incidence rates are reported as year / person.
P-values less than 0.05 are considered statistically significant, and all analyzes will be
performed using STATA 12.0 SE.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 151
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
List of participants to the MEREAFAPS project
1. A.O. San Carlo Borromeo
2. A.O. Niguarda Cà Granda
3. A.O. Fatebenefratelli
4. A.O. Bolognini
5. Ospedali civili di Brescia
6. A.O. Istituti Ospedalieri di Cremona
7. A.O. Carlo Poma
8. A.O. San Gerardo
9. A.O. Lecco
10. A.O. San Matteo
11. A.O. Luigi Sacco
12. A.O. San Raffaele
13. A.O. di Valtellina Valchiavenna
14. A.O. Ospedali Riuniti di Bergamo
15. A.O. Sant'Anna Como
16. A.O. della provincia di Pavia
References
[1] NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis, and Therapy. Osteoporosis
prevention, diagnosis, and therapy. JAMA 2001;285:785-95
[2] Cummings SR, Melton LJ. Epidemiology and outcomes of osteoporotic fractures. Lancet 2002;359:17617
[3] Kai MC, Anderson M, Lau EM. Exercise interventions: defusing the world's osteoporosis time bomb. Bull
World Health Organ 2003;81:827-30
[4] Wood J, Bonjean K, Ruetz S, et al. Novel antiangiogenic effects of the bisphosphonate compound
zoledronic acid. J Pharmacol Exp Ther 2002;302:1055-61
[5] Siris ES, Harris ST, Rosen CJ, et al. Adherence to bisphosphonate therapy and fracture rates in
osteoporotic women: relationship to vertebral and nonvertebral fractures from 2 US claims databases. Mayo
Clin Proc 2006;81:1013-22
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 152
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[6] Solomon DH, Avorn J, Katz JN, et al. Compliance with osteoporosis medication. Arch Intern Med 2005
;165 :2414-9
[7] Miller RG, Bolognese M, Worley K, et al. Incidence of Gatrointestinal events among bisphosphonate
patients in an observational setting. Am J Manag Care 2004;10:S207-S15
[8] De Groen PC, et al. Esophagitis associated with the use of alendronate. N Engl J Med 1996;335: 1016-21
[9] Suissa S. Immeasurable time bias in observational studies of drug effects on mortality. Am J Epidemiol
2008; 168: 329-35
[10] Schneeweiss S. Sensitivity analysis and external adjustment for unmeasured confounders in
epidemiologic database studies of therapeutics. Pharmacoepidemiol DS 2006;15:291-303
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 153
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKDS2
Gastrointestinal and cardiovascular safety profiles of non-steroidal antiinflammatory drugs (NSAIDs)
Background
Non-steroidal anti-inflammatory drugs (NSAIDs) are widely used in medical practice for
the symptomatic treatment of acute pain, fever, and chronic inflammatory and degenerative
joint diseases such as rheumatic arthritis (RA) and osteoarthritis (OA). However, traditional
non-steroidal anti-inflammatory drugs (tNSAIDs) have been associated with a 3- to 5-fold
increased risk in serious upper gastrointestinal complications such as bleedings,
obstructions and perforations [1]
It is estimated that in the European Union thousands of gastrointestinal complications are
most likely caused by the use of NSAIDs every year. A new class of NSAIDs, the
“Coxibs”, has been developed specifically to minimize the risk of gastrointestinal events.
Clinical studies have shown that use of coxibs has a 50% lower risk of upper GI
complications than tNSAIDs. [2,3]
Marketing authorisation was granted for rofecoxib and celecoxib as first coxibs at the end
of the 1990s in the EU. In the following years marketing authorisations were granted for
the “second generation” coxibs etoricoxib, parecoxib/valdecoxib and lumiracoxib.
However, data from a large clinical study2 as well as epidemiological data [4,5] raised
concerns since they may increase the risk of cardiovascular events, among which
myocardial infarction and ischemic stroke. Despite numerous evaluation studies several
questions have yet remained unanswered, which hampers adequate treatment decision
making around the use of individual NSAIDs. Single studies often are too small to look at
all individual NSAIDs, mostly because only a particular set of NSAIDS is used in one
country. By combining data from different countries, it should be possible to reach an
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 154
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
adequate sample size to evaluate risk of the different NSAIDs and to estimate the risk in
specific subgroups, such as in children.
The EU project “Safety Of non-Steroidal anti-inflammatory drugs” (SOS) is coordinated by
the Erasmus University Medical Center (Netherlands), and is a consortium of 11 leading
research institutions who work together to address the above mentioned issues.
Objective
The SOS project aims to estimate the relative risks of cardiovascular and gastrointestinal
events associated with the current and recent use of individual NSAIDs, dose, duration,
concomitant drug use, and co-morbidity compared to past use.
Subjects and Methods
Data is being retrieved from 7 European databases covering different study periods (please
see below the list of participants to the SOS project). In particular, SISR database includes
data on NSAIDs users retrieved from the HCU databases of Lombardy Region. In each
database data are being aggregated and used for further analysis. Several case-control
studies, nested in a NSAIDs new-users adult cohort (18 years), are being carried out in each
database using common harmonized definitions for each cardiovascular or gastrointestinal
event and relevant comorbidities. In each study the cases are all patients with an event
hospitalization during follow-up. The index date is being defined as the date of the first
event hospitalization. Event cases are being matched to up to 100 controls by sex, age,
follow-up length and index date. Conditional logistic regression models are being applied
to compute Odds ratios (ORs), and corresponding 95% CI, estimating the association
between: i) current use of individual NSAIDs vs. past use of any NSAID; ii) high dose vs.
low dose of the same drug or drug group during current use period; iii) different duration
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 155
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
of use the same drug or drug group. Estimates are being adjusted for individual
characteristics, comorbidities and concomitant drugs use. Pooled NSAID-specific ORs are
being obtained by a random effects meta-analysis approach to account for the DBs
heterogeneity. Separate models are being built for children since the indications and
dosages differ and little is known on the safety of NSAIDs in this group as they are often
prescribed off-label.
To evaluate the influence of several methodological issues related to the conduction of
observational studies based on healthcare databases several additional analyses are being
performed with the initial new-users cohort. Firstly, to take into account the lack of
relevant potential confounders in administrative databases several methods are being
applied to evaluate the impact of measured and unmeasured confounding on the results of
the cohort study (e.g., the rule-out, the Monte Carlo Sensitivity Analysis and the
instrumental variables approaches). To evaluate the bias due to misclassification of the
duration of therapy of an NSAID (calculated as the sum of the durations of all prescriptions
received by a patient during the follow-up) a regression calibration method is being applied.
Finally, with the aim to explore the relationship between the short-term risk of a first event
hospitalization and NSAID therapy discontinuation and restarting, as well as their
determinants, a multi-state model is being implemented.
Impact
This study will be the largest observational study on NSAID safety that has ever been
performed and will permit the unique possibility to investigate the effects of several
individual NSAIDs. Finally the knowledge generated will allow the development of
decision models, which are needed in clinical practice to decide the type of NSAID that
would yield the lowest gastrointestinal and cardiovascular risk for an individual patient.
Decision models for regulatory agencies will focus on the public health risk.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 156
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Management
Project manager (epidemiology and biostatistics): Giovanni Corrao,
Antonella Zambon1
Data management and analysis: Federica Nicotra, Arianna Ghirardi,
Andrea Arfè, Lorenza Scotti 2
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
List of participants to the SOS project
•
•
•
•
•
•
•
•
•
•
•
•
Erasmus University Medical Center - EMC (Netherlands): Miriam Sturkenboom, Johan van der Lei,
Mees Mosseveld, Ewout Steyerberg, Bruno Stricker, Rene Schade, Martijn Schuemie, Silvana
Romio
Fundació Institut Mar d'Investigacions Mèdiques - FIMIM (Spain): Eva Molero, Montse Camprubí
University of Nottingham - UNOTT (UK): Julia Hippisley-Cox, Carol Anne Charlotte Coupland, Jill
Harris
Università di Milano-Bicocca - UNIMIB (Italy): Giovanni Corrao, Antonella Zambon, Federica
Nicotra, Rino Belloco, Gianluca Baio, Arianna Ghirardi, Davide Soranna, Andrea Arfè, Lorenza
Scotti
Research Triangle Institute - RTI-HS (Spain): Susana Pérez-Gutthann, Cristina Varas-Lorenzo, Jordi
Castellsague, Nuria Riera
Universitaet Bremen - Uni-HB (Germany): Edeltraut Garbe, Tania Schink, Bianca Kollhorst
The Research Institute of the McGill University Health Centre RI-MUHC (Canada): James Brophy
Azienda Ospedaliera di Padova - AOPD (Italy): Carlo Giaquinto, Francesco Zulian, Marta Balzarin
PHARMO Coöperation UA - PHARMO (Netherlands): Ron MC Herings, Huub Straatman
Université Victor-Segalen Bordeaux II - UB2 (France); (i) Pharmacology Department, University of
Victor Segalen Bordeaux2, France: Nicholas Moore, Annie Fourrier, Antoine Pariente, Francesco
Salvo, Francoise Haramburu; (ii) LESIM, University of Victor Segalen Bordeaux2, France: Paul
Avillach, Gayo Diallo, Fleur Mougin, Frantz Thiessard ); LERTIM, Marseilles, France: Marius
Fieschi, Michel Joubert, Jean-Charles Dufour
Azienda Sanitaria Locale della provincia di Cremona - ASL/OSSIFF (Italy): Salvatore Mannino,
Marco Villa
Società Servizi Telematici - SOSETE (Italy): Luigi Cantarutti, Silvia Girotto, Silvia Faggion
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 157
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
References
[1] Hernandez-Diaz S, Rodriguez LA. Association between nonsteroidal anti-inflammatory drugs and upper
gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch
Intern Med 2000;160:2093-9
[2] Bombardier C, Laine L, Reicin A et al. Comparison of upper gastrointestinal toxicity of rofecoxib and
naproxen in patients with rheumatoid arthritis. VIGOR Study Group. N Engl J Med 2000;343:1520-8
[3] Silverstein FE, Faich G, Goldstein JL et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal antiinflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS study: A randomized controlled
trial. Celecoxib Long-term Arthritis Safety Study. JAMA 2000;284:1247-55
[4] Ray WA, Stein CM, Hall K, Daugherty JR, Griffin MR. Non-steroidal anti-inflammatory drugs and risk
of serious coronary heart disease: an observational cohort study. Lancet 2002;359:118-23
[5] Solomon DH, Glynn RJ, Levin R, Avorn J. Nonsteroidal anti-inflammatory drug use and acute
myocardial infarction. Arch Intern Med 2002;162:1099-104
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 158
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
6.11. Other issues (miscellaneous)
Other areas may be investigated by means of the proposed approach. Among these, the
following issues will be of main, even not exclusive, interest.
Project CRACKOI1
Estimating prevalence and incidence rates of rheumatoid arthritis from
regional healthcare databases
Background
Rheumatoid Arthritis (RA) is a chronic autoimmune systemic disease that is associated
with disability, increased mortality and relevant costs to society.
The prevalence and incidence rates in most industrialized countries respectively vary from
0.3% to 1% and from 20 to 300 per 100,000 person-year. Incidence of juvenile RA has
been reported to be 20-50 per 100,000 per year. Prevalence, as well as incidence rates, rises
with age reaching a peack around at the age 70 years, and declining afterwards. About
twice as many women as men are affected by RA [196].
Disability starts early in the course of the disease and progressively worsens over time. In
the developed countries at least 50% of patients are unable to maintain a full-time
employment within 10 years from disease onset [196].
RA is associated with reduced life expectancy. The excess mortality is apparent within the
first few years of disease and increases with disease duration. Most of the excess deaths are
attributable to infection, cardiovascular (in particular coronary heart disease) and
respiratory disease as well as to lung cancer and non-Hodgkin's lymphoma [197].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 159
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data about RA's morbidity and mortality in Italian population are unsystematic and limited
to specific geographical areas, primarily due to the high costs of population studies [198203]. Studies based on the record linkage of healthcare databases could be fill this
knowledge gap, even though RA is a disease with a high degree of diagnostic uncertainty
[204-211].
Aim
The project has two main aims: i) to identify a valid and accurate diagnostic algorithm for
the identification of RA cases; ii) to estimate the prevalence and incidence of RA in the
Lombardy adult population by means record linkage of HCU databases of Lombardy
Region.
Methods
A random sample of RA patients seen at Rheumatology Department, IRCCS San Matteo di
Pavia Hospital, will be identified. A sample of control patients withour RA will be selected
from the same hospital where RA patients were visited.
For each selected patient all variables recorded in the ASL Pavia healthcare databases (e.g.
exemptions, hospital discharges, outpatient prescriptions, etc…) will be retrieved. Based on
these variables, several algorithms will be proposed for RA diagnosis. The diagnostic
accuracy of these algorithms will be estimated by applying them to a set of patients with
known RA diagnosis. The external validity of the most accurate algorithm will be also
assessed. The validated algorithm will be applied to the Lombardy HCU databases to
estimate the prevalence and the incidence of RA in the last years.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 160
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Impact
The project will provide useful measures of the RA frequency in the Lombardy adult
population, as well as a precious and validated tool to measure and monitor Lombardy’s
RA burden by exploiting the information available in HCU databases.
Management
Project manager (clinics): Carlo Alberto Scirè1
Project manager (epidemiology and biostatistics): Antonella
Zambon2
Data management and analysis: Federica Nicotra3
1
2
Unit of Rheumatology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
3
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1] Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bullettin of the World Health
Organization 2003;81: 646-56
[2] Naz SM, Symmons DP. Mortality in established rheumatoid arthritis. Best Pract Res Clin Rheumatol
2007;21:871-83
[3] Della Rossa A, Neri R, Talarico R, et al. Diagnosis and referral of rheumatoid arthritis by primary care
physician: results of a pilot study on the city of Pisa, Italy. Clin Rheumatol 2010;29:71-81
[4] Salaffi F, Sarzi-Puttini P, Girolimetti R, et al. Health-related quality of life in fibromyalgia patients: a
comparison with rheumatoid arthritis patients and the general population using the SF-36 health survey. Clin
Exp Rheumatol 2009;27:S67-74
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 161
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[5] Benucci M, Cammelli E, Manfredi M, et al; Associazione Medici-Scandicci. Early rheumatoid arthritis in
Italy: study of incidence based on a two-level strategy in a sub-area of Florence (Scandicci-Le Signe).
Rheumatol Int 2008;28:777-81
[6] Salaffi F, De Angelis R, Grassi W; MArche Pain Prevalence; INvestigation Group (MAPPING) study.
Prevalence of musculoskeletal conditions in an Italian population sample: results of a regional communitybased study. I. The MAPPING study. Clin Exp Rheumatol 2005;23:819-28
[7] Marotto D, Nieddu ME, Cossu A, et al. Prevalence of rheumatoid arthritis in North Sardinia: the Tempio
Pausania's study. Reumatismo 2005;57:273-6
[8] Cimmino MA, Parisi M, Moggiana G, et al. Prevalence of rheumatoid arthritis in Italy: the Chiavari
Study. Ann Rheum Dis 1998;57:315-8
[9] Gabriel SE. The sensitivity and specificity of computerized databases for the diagnosis of rheumatoid
arthritis. Arthritis Rheum 1994;37:821-3
[10] Katz JN, Barrett J, Liang MH, et al. Sensitivity and positive predictive value of Medicare Part B
physician claims for rheumatologic diagnoses and procedures. Arthritis Rheum 1997;40:1594-600
[11] MacLean CH. Evaluating the quality of care in rheumatic diseases. Curr Opin Rheumatol 2001;13:99103
[12] Losina E, Barrett J, Baron JA, et al. Accuracy of Medicare claims data for rheumatologic diagnoses in
total hip replacement recipients. J Clin Epidemiol 2003;56:515-9
[13] Singh JA, Holmgren AR, Noorbaloochi S. Accuracy of Veterans Administration databases for a
diagnosis of rheumatoid arthritis. Arthritis Rheum 2004;51:952-7
[14] Pedersen M, Klarlund M, Jacobsen S, et al. Validity of rheumatoid arthritis diagnoses in the Danish
National Patient Registry. Eur J Epidemiol 2004;19:1097-103
[15] Thomas SL, Edwards CJ, Smeeth L, et al. How accurate are diagnoses for rheumatoid arthritis and
juvenile idiopathic arthritis in the general practice research database? Arthritis Rheum 2008;59:1314-21
[16] Kim SY, Solomon DH. Use of administrative claims data for comparative effectiveness research of
rheumatoid arthritis treatments. Arthritis Res Ther 2011;13:129
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 162
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKOI2
Risk/benefit profile of bisphosphonates therapy in primary/secondary
prevention of osteoporotic fractures
Background
Osteoporosis is a systemic skeletal disorder characterized by low bone density and microarchitectural deterioration of bone tissue, with subsequent increase in bone fragility and
susceptibility to fracture [1]. Currently, osteoporosis is clinically recognized by the
occurrence of fractures, that typically occur after only moderate trauma and are frequently
related with vertebral, hip and wrist porosity. The disease has became a clinical and public
health concern because, with the progressive aging of the population, osteoporotic fractures
are one of the most common causes of disability and an important contributor to medical
costs in many regions of the world [2]. The WHO estimates that osteoporosis currently
affects more than 75 million people in Europe, Japan and the USA alone, with an estimated
lifetime risk for wrist, hip and vertebral fractures of around 15%, very similar to that of
coronary heart disease [3]. Bisphosphonates (BPs), such as zoledronic acid, pamidronate,
risedronic acid, and alendronic acid, are agents whose main pharmacological action is to
inhibit the bone resorption trough their effects on osteoclasts [4]. Long-term adherence to
therapy is required for optimal therapeutic benefit in patients with osteoporosis. A recent
study, conducted with claims databases in a "real-world setting", has confirmed that greater
refill compliance (adherence) to BPs was significantly associated with fewer fractures after
24 months of follow-up [5]. Despite the proven efficacy, the adherence to BP therapy
remains still suboptimal [6]. Failure to be complaint may be due to multiple factors. A
major reason is because osteoporosis is an asymptomatic, chronic condition, and the
benefits of treatment are not immediately apparent to the patients. Poor tolerability is
another factor limiting adherence. Although oral BPs have generally demonstrated a safety
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 163
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
profile similar to placebo in clinical trials, patients in clinical practice fairly often
experience gastrointestinal (GI)-related adverse events [7,8].
Objective
The Project aims to assess the association between the use of oral BPs, prescribed for the
primary/secondary prevention of osteoporotic fractures, and the occurrence of Upper
GastroIntestinal Complications (UGIC) and osteoporotic fractures.
Subjects and Methods
A nested case-control study will be conducted on the basis of data retrieved from the HCU
databases of 13 Italian territorial units participating at the AIFA-BEST (Bisphosphonates
Efficacy-Safety Tradeoff) Project, covering about 17 million of beneficiaries of NHS
residents in units. Territorial units were four Regions and nine Local Health Authorities
(please see below list of participants to the AIFA-BEST project). All patients who (i)
received at least one dispensation of BPs and (ii) were hospitalized for osteoporotic fracture
in the period July 1, 2003-December 31, 2005 will be identified. The outcomes of interest
will be hospitalization for (i) UGIC and (ii) osteoporotic fractures during follow-up. Each
patient enrolled in the study will be followed until the earliest date among those of outcome
onset (hospital admission for UGIC/osteoporotic fractures) or censoring (death, emigration
or December 31, 2007). With the aim of avoiding the so called immeasurable time bias [9],
some approach for the control of this bias will be applied. Rule-out method [10] will be
used for unmeasured confounding adjustment.
Management
Project manager (epidemiology and biostatistics): Giovanni Corrao,
Antonella Zambon1
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 164
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data management and analysis: Arianna Ghirardi2
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
List of participants to the AIFA-BEST project
•
•
•
•
•
•
•
•
•
•
•
Regional Agency for Healthcare Services of Tuscany (Italy): Giampiero Mazzaglia, Francesco
Cipriani, Emiliano Sessa, Francesco Lapi
University of Messina (Italy): Achille Patrizio Caputi, Vincenzo Arcoraci
University of Milano-Bicocca (Italy): Giovanni Corrao, Antonella Zambon, Lorenza Scotti, Arianna
Ghirardi
University of Bologna (Italy): Nicola Montanaro, Carlo Piccinni, Alberto Vaccheri
Erasmus University Medical Center - (The Netherlands): Miriam Sturkenboom
University of Florence (Italy): Pierangelo Geppetti, Mauro Di Bari
University of Turin (Italy): Dario Gregori
University Plotecnica delle of Marche (Italy): Flavia Carle, Rosaria Gesuita
Regional Agency of Healthcare services of Abruzzo (Italy): Tommaso Staniscia, Angelo Menna
University di Roma “La Sapienza” (Italy): Anna Rita Vestri
University of L’Aquila (Italy): Marco Valenti
References
[1] NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis, and Therapy. Osteoporosis
prevention, diagnosis, and therapy. JAMA 2001;285:785-95
[2] Cummings SR, Melton LJ. Epidemiology and outcomes of osteoporotic fractures. Lancet 2002;359:17617
[3] Kai MC, Anderson M, Lau EM. Exercise interventions: defusing the world's osteoporosis time bomb. Bull
World Health Organ 2003;81:827-30
[4] Wood J, Bonjean K, Ruetz S, et al. Novel antiangiogenic effects of the bisphosphonate compound
zoledronic acid. J Pharmacol Exp Ther 2002;302:1055-61
[5] Siris ES, Harris ST, Rosen CJ, et al. Adherence to bisphosphonate therapy and fracture rates in
osteoporotic women: relationship to vertebral and nonvertebral fractures from 2 US claims databases. Mayo
Clin Proc 2006;81:1013-22
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 165
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[6] Solomon DH, Avorn J, Katz JN, et al. Compliance with osteoporosis medication. Arch Intern Med 2005
;165 :2414-9
[7] Miller RG, Bolognese M, Worley K, et al. Incidence of Gatrointestinal events among bisphosphonate
patients in an observational setting. Am J Manag Care 2004;10:S207-S15
[8] De Groen PC, et al. Esophagitis associated with the use of alendronate. N Engl J Med 1996;335: 1016-21
[9] Suissa S. Immeasurable time bias in observational studies of drug effects on mortality. Am J Epidemiol
2008; 168: 329-35
[10] Schneeweiss S. Sensitivity analysis and external adjustment for unmeasured confounders in
epidemiologic database studies of therapeutics. Pharmacoepidemiol DS 2006;15:291-303
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 166
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 167
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
7. Health econometrics and health demand (Working Package 3)
Project CRACKHED1
CRISP
Concept
The aim of this research project is to provide a comprehensive analysis of the main
characteristics of the demand and the supply in the healthcare sector. Providing healthcare
services to a society with an increasing share of elderly people and controlling the level of
the healthcare quality and expenditure are the most important challenges of welfare states
across western countries.
The project will study the performances of the healthcare system and its financial stability.
Particular attention will be paid to the effects on the health outcomes, the appropriateness
of healthcare services provided, and agents' strategic behavior on the demand side and the
supply side. In order to achieve the crucial objective of "improving the health and the wellbeing of all individuals during all their lifelong", we will draw suggestions for a better
provision of healthcare services within the countries.
In particular the comparative effectiveness will take into account the principal methods
applied in this type of analysis. The experience gained in Italy and in particular in
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 168
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Lombardy Region for the evaluation of effectiveness will be the core of the study. The
investigation of this topic move from the administrative data collected form the healthcare
systems regarding the patients’ discharges.
On the demand side, the research unit will investigate different issues by taking into
account the interaction between different factors: the technological progress in healthcare;
the incidence and the prevalence of chronic diseases related to the individual lifestyle, and
in particular to risky health behaviors, such as over-eating, smoking, drinking and the lack
of physical exercise; socio-demographic characteristics, epidemiological trends, and
individual preferences. All these factors will be tested as determinants of the demand for
healthcare, with a special focus on patients' mobility. Finally, the project will analyze the
impact of the implemented healthcare reforms on the appropriateness of the healthcare
services.
On the supply side, the research project will analyze several issues concerning this topic:
the impact of different schemes of public funding on the performance of the supply and
possible opportunistic behaviors (e.g., patients' selection at the hospital level, readmissions,
etc.); the impact of different types of networks among GPs on their activity and efficiency
(e.g. in terms of per-patient pharmaceutical expenditure, outpatient discharges, inpatient
admissions); the relationship arising between healthcare services provided by primary care
professionals and those provided by hospitals (secondary care) and nursing homes, and the
effects on the healthcare expenditure; the rising trade-off between costs containment and
health outcomes, taking appropriate measures of outcomes into account.
Finally, the research project will provide some evidence at the aggregate level by
investigating the Lombardy regional healthcare systems, its expenditure and performances.
This research activity will identify health treatments that may be included in the so-called
Essential Levels of Care, that should be provided to the entire population. Moreover, the
project will study costs associated with those levels, taking regional aspects and the impact
of different methods of financing the regional expenditure into account.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 169
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
The project is organized in three main areas of research:
1. Regional systems performances, health expenditures and Essential Levels of Care
2. Determinants of the demand of healthcare and impacts on expenditures
3. Determinants of the supply of healthcare and impacts on expenditures.
Objectives
The research project aims at studying the main changes that have taken place in the
European healthcare sector, and the main problems regarding its financial sustainability.
The analysis will highlight the different characteristics of the regional healthcare systems,
and will identify some Essential Levels of Care that should be provided. The project will
focus on some of the aspects that are more problematic for modern national healthcare
systems, such as:
- the definition of indicators measuring healthcare outcomes including also patients' and
relatives' psychological implications of the received treatments
- the evaluation of the productivity of the different institutions acting in primary, secondary
and long-term care
- the modeling and the empirical testing for strategic and opportunistic behaviors both in
the demand and the supply side of the healthcare sector
- a comprehensive analysis of different performances achieved in the Europeans regions
- the determinants of patients' mobility
- the identification of uniform costs levels for the same treatment even if provided in local
contexts with different socio-economic conditions, i.e., the so-called standard of costs.
The results of our projects will be important both to understand the workings of our
healthcare systems and to propose new regulatory mechanisms.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 170
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Since the results of our research will mostly be of interest for the scientific and academic
community and for policy makers in regional healthcare sectors, we plan to present our
results at national and international conferences, to publish our results in scientific journals
and to organize events where to disclose and discuss the policy implications of our
analyses. Indeed, the assessment of the regulatory tools to manage demand and supply is of
great interest for policy-makers as it can be used to inform the choice of the most effective
policy instruments. From a scientific and health policy perspective, the aim of our project is
to answer to several questions.
Management
Project manager (statistitian): Giorgio Vittadini1
Project manager (economist): Gianmaria Martini2
Data management and analysis: Paolo Berta1
1
2
CRISP – University of Milano-Bicocca, Milan, Italy
Dept of Engineering, University of Bergamo, Italy
References
[1]
Donabedian, A., 1988, The quality of care: How can it be assessed?, The Journal of the American
Medical Association, 260(12), 1743-1748
[2]
Goldstein, H., Spiegelhalter, D.J., 1996, League tables and their limitations: statistical issues in
comparisons of institutional performances, Journal of the Royal Statistical Society, 159(3), 385-443.
[3]
Kumbhakar, S.C., Lovell, C.A.K., 2000. Stochastic Frontier Analysis, (Cambridge, U.K., Cambridge
University Press)
[4]
Leyland, A. H., Goldstein, H. (eds.), 2001, Multilevel Models of Health Statistics, J. Wiley, London
[5]
McKay, N.L., Deily, M.E., 2008, Cost inefficiency and hospital health outcomes, Health Economics,
17, 833-848.
[6]
Vittadini, G., Berta, P., Martini, G. and G. Callea, The effect of a law limiting upcoding on hospital
admissions: Evidence from Italy, Empirical Economics, 42(2), 563-582.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 171
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[7]
Berta P, Callea G, Martini G, Vittadini G (2010) The effects of upcoding, cream skimming and
readmissions on the Italian hospitals efficiency: a population–based investigation. Econ Model 27:812–821
[8]
Dafny LS (2005) How do hospitals respond to price changes?. Am Econ Rev 95:1525–1547
[9]
Silverman E, Skinner J (2004) Medicare upcoding and hospital ownership. J Health Econ 23:369–389
[10]
Ash AS, Fienberg SE, Louis TA, Normand ST, Stukel TA, Utts J, Statistical Issues in Assessing
Hospital Performance, Commissioned by the Committee of Presidents of Statistical Societies, (2012)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 172
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Project CRACKHED2
Administrative databases as a tool for identifying healthcare demand and
costs in an over-one million population
Background
Healthcare demand continues rising in most Western countries and the challenge is how to
sustain the high costs resulting. Furthermore the recent global financial crisis is having an
impact on healthcare systems, especially in countries offering a universal coverage.
In particular, in Italy the slow economic growth is limiting both public and private health
expenditures and is making difficult to meet the health needs and demand of population
[1].In this country, policy-makers are responding to crisis through interventions aimed at
increasing the efficiency of public spending and rationalizing costs. These actions may be
improved by a more accurate knowledge of healthcare demand, which sometimes can
provide a direct estimation of individual needs [2].
Managing demand for healthcare has been defined as the process of identifying where,
how, why and by whom demand arises and then deciding on the best methods of dealing
with it, so that the most efficient, appropriate ad equitable approach can be developed and
applied [3].
A study conducted by Lynn et al. suggests that the analysis of healthcare demand can be
efficiently obtained by segmenting the target population on health prospects and priorities
and identifying homogeneous groups with specific provision of services [4]. Population
size and annual healthcare costs for each segment were estimated obtaining data from
various sources that, unfortunately, used different definitions and time periods. Indeed, one
crucial point in constructing accurate and precise demand models is related to the type and
amount of relevant data available [5-6].
Administrative databases of the National Health Systems may constitute a fundamental
source of readily available and relatively inexpensive large amounts of good quality data
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 173
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
(demographic, clinical, economic) referring to the general population [7-9]. In this way,
these datasets allowed us to identify and monitor health needs and demand of the
population and consequently optimize the use of economic resources available.
Aims
The aim of this study was to build a segmentation model that can be applied on healthcare
administrative databases in order to assess demand and the main costs borne by Italian
health system for the care of specific and distinguished groups of a general population.
Methods
The target population included subjects living in an area of northern Italy registered at one
Local Healthcare Unit in 2005 (1,031,684 subjects). On the basis of clinical judgments and
literature, we identified eight different segments: subjects unknown to HS, maternity and
infancy, elderly, people with one chronic disease (CD), people with more CDs, people with
probable or not severe CDs, subjects with acute event, healthy people. To describe these
groups and their health demand, we used demographic and healthcare demand data
(hospital admissions, drug’s prescriptions, medical specialist visits, and diagnostic tests)
from administrative databases available at the Lombardy health system.
These datasets and demographic characteristics of people under the care of the Lombardy
RHS were organized into a data warehouse named DENALI. Since the quality of data
reported in these databases was not optimal, a probabilistic record linkage was
implemented to match the anonymized data of the different datasets belonging to the same
individuals [10-12].
Management
Project manager (epidemiology and biostatistics): Giancarlo Cesana1
Project manager (clinics): Michele A Riva2
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 174
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Data management and analysis: Carla Fornari and Fabiana Madotto3
1
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
2
3
Division of Occupational and Environmental Medicine, San Gerardo Hospital, Monza, Italy
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
References
[1]
De Belvis AG, Ferrè F, Specchia ML, et al. The financial crisis in Italy: implications for the healthcare
sector. Health Policy 2012;106:10-6
[2]
Wright J, Williams R, Wilkinsondia JR. Development and importance of health needs assessment.
BMJ 1998;316:1310-3
[3]
Pencheon D. Managing demand: matching demand and supply fairly and efficiently. BMJ
1998;316:1665-7
[4]
Lynn J, Straube BM, Bell KM, et al. Using population segmentation to provide better HEALTHCARE
for all: the “Bridges to Health” model. Milbank Q 2007;85:185-208
[5]
Sears JM, Krupski A, Joesch JM, et al. The use of administrative data as a substitute for individual
screening scores in observational studies related to problematic alcohol or drug use. Drug Alcohol Depend
2010;111:89-96
[6]
Crane SJ, Tung EE, Hanson GJ, et al. Use of an electronic administrative database to identify older
community dwelling adults at high-risk for hospitalization or emergency department visits: the elders risk
assessment index. BMC Health Serv Res 2010;10:338
[7]
Shahian DM, Silverstein T, Lovett AF, et al. Comparison of clinical and administrative data sources
for hospital coronary artery bypass graft surgery report cards. Circulation 2007;115:1518-27
[8]
Suissa S, Garbe E. Primer: administrative health databases in observational studies of drug effects-
advantages and disadvantages. Nat Clin Pract Rheumatol 2007;3:725-32
[9]
Schmitt J, Maywald U, Schmitt NM, et al. Cardiovascular comorbidity and cardiovascular risk factors
in patients with chronic inflammatory skin diseases: a case-control study utilising a population-based
administrative database. Ital Journal of Public Health 2008;5:187-93
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 175
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10]
Fellegi IP, Sunter AB. A theory for record linkage. J Am Stat Assoc 1969;64:1183-210
[11]
Newcombe HB, Kennedy JM, Axford SJ, et al. Automatic linkage of vital records. Science
1959;30:954-9
[12]
Newcombe HB, Kennedy JM. Making maximum use of the discriminatory power of identifying
information. In: Communications of ACM 1962;5:563-6
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 176
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 177
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
8. Educational issues (Working Package 4)
Residential courses mainly devoted to ASL staff employed for HCU DB management and
health researchers, will be performed with the aim of training them at the following topics:
•
Observational methods: measures, sources of random and systematic uncertainty,
stuy’s designs, meta-analysis
•
Biostatistics for epidemiologists: basic methods and advanced techniques outlines,
sample size and confidence intervals
•
Informative systems for HCU DB management and analysis
•
Writing research protocols and technical reports.
The course will be duration of fourthy hours, including frontal lessons and laboratory
practices.
Advanced courses for might be planned on request.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 178
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 179
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
9. Scientific board and accredited laboratories
9.1. Scientific board
•
Giovanni Corrao
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health,
University of Milano-Bicocca, Milan, Italy
•
Alberico Catapano
Dept of Pharmacological Sciences and Centre for Pharmacoepidemiology and Pharmacoutilization,
University of Milano, Milan, Italy
•
Giancarlo Cesana
Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health,
and Centre of Public Health (CESP), University of Milano-Bicocca, Milan, Italy
•
Carlo La Vecchia
Dept of Clinical Sciences and Community Health,, Unit of Medical Statistics and Biometrics,
University of Milano - Dept of Epidemiology, Mario Negri Institute for Pharmacologic Research,
Milan, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 180
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
•
Giorgio Vittadini
Dept of Statistics and Quantitative Methods, and Centro di Ricerca Interuniversitario per I Servizi di
Pubblica Utilità (CRISP), University of Milano-Bicocca, Milan, Italy
•
Giuseppe Mancia
Italian Institute for Auxology, Milan, Italy
Curriculum vitae and selected pertinent publications carried out from scientific board
members in the last two years are reported in Appendix 3.
9.2. Project managers
•
Felice Achilli 1
•
Enrico Agabiti Rosei 2
•
Stefano Aliberti 3
•
Paolo Berta 4
•
Francesco Blasi 3
•
Ovidio Brignoli 5
•
Luigi Cantarutti 6
•
Alberico L Catapano 7
•
Giancarlo Cesana 8,9
•
Valentino Conti 10
•
Giovanni Corrao 8,11
•
Silvio Danese 12
•
Giovanni De Vito 13
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 181
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
•
Marco Ferrario 14
•
Alessandro Filippi 5
•
Gionata Fiorino 15
•
Guido Grassi 16
•
Sergio Harari 13
•
Carlo La Vecchia 17,18,19
•
Roberto Latini 20
•
Olivia Leoni 10
•
Antonio Lora 21
•
Gianmaria Martini 22
•
Emiliano Monzani 23
•
Eva Negri 18,19
•
Alberto Pesci 24
•
Antonio Pesenti 25
•
Michele A Riva 26
•
Alma Luisa Rivolta 10
•
Marina Scavini 27
•
Carlo Alberto Scirè 28
•
Mauro Venegoni 10
•
Caterina Viganò 29
•
Giuseppe Vighi 10
•
Giorgio Vittadini 4,8
•
Antonella Zambon 8,11
•
Gian Vincenzo Zuccotti 30
1 Dept of Cardiology, A. Manzoni Hospital, Lecco, Italy
2 Dept of Clinical and Experimental Sciences, University of Brescia, Brescia, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 182
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
3 Respiratory Medicine Section, Dipartimento Toraco-Polmonare e Cardiocircolatorio, University of Milan,
IRCCS Fondazione Ospedale Maggiore Policlinico Cà Granda Milan, Milan, Italy
4 Laboratory of HEALTH ECONOMETRICS AND HEALTH DEMAND, Centro di Ricerca
Interuniversitario per I Servizi di Pubblica Utilità (CRISP), University of Milano-Bicocca, Milan, Italy
5 Health Search, Italian College of General Practitioners, Florence (G.M., E.S.), Italy
6 Family Pediatrician Pedianet Project, Padova, Italy
7 Dept of Pharmacological Sciences and Centre for Pharmacoepidemiology and Pharmacoutilization,
University of Milano, Milan, Italy
8 Dept of Statistics and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, and
Centre of Public Health, University of Milano-Bicocca, Milan, Italy
9 Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH, Dept of Statistics and
Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of Milano-Bicocca,
Milan, Italy
10 Pharmacovigilance Regional Center of Lombardy, Direzione Generale Sanità, Regione Lombardia,
Milano, Italy
11 Laboratory of PHARMACOEPIDEMIOLOGY AND HEALTHCARE RESEARCH, Dept of Statistics
and Quantitative Methods, Unit of Biostatistics, Epidemiology and Public Health, University of MilanoBicocca, Milan, Italy
12 Division of Gastroenterology, Humanitas Clinical and Research Center, Rozzano, Milan, Italy
13 Dept of Pneumology, San Giuseppe Hospital, Milano, Italy
14 Epidemiology and Preventive Medicine Research Centre, Department of Experimental Medicine,
University of Insubria, Varese, Italy
15 Division of Gastroenterology, Humanitas Clinical and Research Center, Rozzano, Milan, Italy
16 Division of Internal Medicine, San Gerardo Hospital, Monza, Dept of Health Sciences, University of
Milano-Bicocca, Milan, Italy
17 Dept of Clinical Sciences and Community Health,, Unit of Medical Statistics and Biometrics, University
of Milano
18 Dept of Epidemiology, Mario Negri Institute for Pharmacologic Research, Milan, Italy
19 Laboratory of EPIDEMIOLOGY, Dept of Epidemiology, Mario Negri Institute for Pharmacologic
Research, Milan, Italy
20 Dept of Cardiovascular Research, Mario Negri Institute, Milan, Italy
21 Dept of Mental Health, Local Health Unit of Lecco, Regional Health Service, Lombardy Region, Lecco,
Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 183
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
22 Dept of Engineering, University of Bergamo, Italy
23 Dept. of Mental Health, Hospital Niguarda Cà Granda, Regional Health Service, Lombardy Region,
Milan, Italy
24 Division of Pneumology, San Gerardo Hospital, Monza, Dept of Health Sciences, University of MilanoBicocca, Milan, Italy
25 Dept of Perioperative Medicine and Intensive Care, San Gerardo Hospital, Monza, Italy, Dept of Health
Sciences, University of Milano-Bicocca, Milan, Italy
26 Division of Occupational and Environmental Medicine, San Gerardo Hospital, Monza, Italy
27 Diabetes Research Institute (HSR-DRI), San Raffaele Scientific Institute, Milan, Italy
28 Unit of Rheumatology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy
29 Operative Unit f Psychiatry II, Hospital Sacco, Regional Health Service, Lombardy Region and Dept. of
Biomedical and Clinical Sciences, University of Milano, Milan, Italy
30 Dept of Pediatrics, Hospital Sacco, Regional Health Service, Lombardy Region and Dept. of Biomedical
and Clinical Sciences, University of Milano, Milan, Italy
9.3. Laboratories
•
Laboratory
of
PHARMACOEPIDEMIOLOGY
AND
HEALTHCARE
RESEARCH
Dept. Statsitics and Quantitative Methods, Unit Biostatistics, Epidemiology and Public Health,
University Milano-Bicocca, Milan, Italy
Responsible Antonella Zambon
•
Laboratory of EPIDEMIOLOGY AND PUBLIC HEALTH RESEARCH
Dept. Statsitics and Quantitative Methods, Unit Biostatistics, Epidemiology and Public Health,
Research Centre of Public Health, University of Milano-Bicocca, Milan, Italy
Responsible Carla Fornari
•
Laboratory of EPIDEMIOLOGY
Dept of Epidemiology, Mario Negri Institute for Pharmacologic Research, Milan, Italy
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 184
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Responsible Eva Negri
•
Laboratory of HEALTH ECONOMETRICS AND HEALTH DEMAND
Centro di Ricerca Interuniversitario per I Servizi di Pubblica Utilità (CRISP), University of MilanoBicocca, Milan, Italy
Director Mario Mezzanzanica
•
Laboratory of PHARMACOVIGILANCE AND DRUG UTILISATION
RESEARCH
Pharmacovigilance Regional Center of Lombardy, Direzione Generale Sanità, Regione Lombardia
Head Alma Lisa Rivolta (Director)
Staff members, potentiality and main characteristics of accredited laboratories affering to
the CRACK program are reported in Appendix 4.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 185
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Appendix 1
Main characteristics of available HEALTHCARE utilization
databases covering the entire population of Lombardy Region
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 186
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
HEALTH REGISTER
Complete coverage and good quality data since 1997
The list of Lombardy Region residents who enjoy access to the Regional Health
Services, including their anagraphic/demographic data, is reported. The Health Register
is a historical archive: this means that individuals who stopped assistance (because of
death or emigration) or started assistance (because of birth or immigration) during the
entire covered time window are recorded.
Main fields
Anonymized identification code
Gender
Date of birth
Date of entry (starting healthcare assistance from RHS)
Date of exit (stopping healthcare assistance from RHS)
Cause of exit (eg. death, emigration, etc…)
General practitioner
Municipality of residence
ASL of residence
RHS: Regional Health Service; ASL: Azienda Sanitaria Locale (Local Health Unit)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 187
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
HOSPITAL DISCHARGE DATABASE
Complete coverage and good quality data since 1997
All hospital discharges occurring in any public or private hospital of the Lombardy
Region are recorded.† In particular, diagnoses and procedures performed during
hospitalization, including those performed in day-hospital or day-surgery, are recorded.
Main fields
Anonymized identification code
Hospital identification code
Date of hospital admission
Ward of admission
Type of admission (ordinary, day hospital, rehabilitation, …)
Date of hospital discharge
Ward of discharge
Type of hospital discharge (voluntary, transfer, death, …)
Primary discharge diagnosis (ICD-9 code)
Up to 5 secondary discharge diagnoses (ICD-9 codes)
Primary procedure (ICD-9 code)
Up to 5 secondary procedures (ICD-9 codes)
DRG code
Total Reimboursment
DRG Diagnosis-Related Group
†
From the year 2010 hospital discharges concerning residents in Lombardy but occurred in hospitals out of
the Region are recorded
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 188
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
EMERGENCY ROOM DATABASE
Complete coverage and good quality data since 2010
Emergency room visits occurring in any public or private hospital of the Lombardy
Region are recorded. For each visit, diagnoses, procedures and outcome are recorded
irrespectively whether the visit leads to death, hospital admission or discharge.
Main fields
Anonymized identification code
Hospital identification code
Hour and date of emergency room admission
Hour and date of emergency room discharge
Appropriateness level evaluated by the physician after the visit (from white -not critical-to
red -very critical- and black -death- codes)
Outcome of emergency room visit (discharge, hospital admission, death, …)
Primary discharge diagnosis (ICD-9 code)
Secondary discharge diagnoses (ICD-9 codes)
Primary procedure (ICD-9 code)
Secondary procedures (ICD-9 codes)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 189
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
OUTPATIENT DRUG PRESCRIPTIONS DATABASE
Complete coverage and good quality data since: 1997
All drug prescriptions reimbursable by the National Health Service dispensed by
territorial pharmacies of Lombardy Region are recorded. For each prescription, the
dipensend drug type and number of packages are recorded.
Main fields
Anonymized identification code
Date of prescription fill
Drug AIC code
Drug ATC code
Prescribing physician code
Number of dispensed drug packages
Number of pills per drug package
Number of DDDs per drug package
Nominal value (in Euros) of a drug package
AIC code: Autorizzazione Immisione in Commercio (Authorization of Market Immission) the code uniquely
identifies a drug package; ATC code: Anatomical Therapeutic Chemical code; DDD: Defined Daily Dose
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 190
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
FILE F
Complete coverage and good quality data since 2006
Drugs directly dispensed to outpatients (with the exception of emophillic inpatients)
by the hospital pharmacies of Lombardy Region are recorded. Only dispensations of
specific drugs (e.g. anti-neoplastic agents, drugs dispensed in day-hospital regimen)
are recorded.
Main fields
Anonymized identification code
Municipality of residence
Hospital identification code
Date of drug administration
Drug AIC code
Prescriber physician anonymized identification code
Administered dose of drug
Measurement unit for dose
Nominal value (in Euros) of the administered drug (per unit and total)
Type of drug in File F (eg. anti-neoplastic agents, drugs dispensed in a day-hospital
regime, ...)
AIC code: Autorizzazione Immisione in Commercio (Authorization of Market Immission) code; uniquely
identifies a drug package
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 191
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
EXEMPTIONS DATABASE
Complete coverage and good quality data 2010
Patients affected by specific chronic conditions can request the exemption to copayment
(the so called “ticket”) for specialistic procedures related to the disease. Such
exemptions are granted only when the diagnosis is certified by a public hospital or
ambulatory. All exemptions granted to residents in Lombardy Region are recorded.
Main fields
Anonymized identification code
Exemption code (identifing the disease for which the exemption was granted)
Starting date
Stopping date (only for exemptions of limited duration)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 192
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
OUTPATIENTS SPECIALISTIC SERVICES DATABASE
Complete coverage and good quality data since 2010
All visits performed in ambulatories of Lombardy Region are recorded. For each visit,
the performed procedures are recorded. Although the field for diagnoses recording is
indicated, the corresponding information is systematically lacking. Emergency room
procedures for visits unleading to hospital admission are also included.
Main fields
Anonymized identification code
Ambulatory procedure code (ICD-9 code)
Nominal value (in Euros) of the procedure
Copayment amount (in Euros)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 193
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
DATAWAREHOUSE DENALI
Complete coverage and good quality data since 2000
The data warehouse DENALI hosts administrative health-care data collected by the
Region of Lombardy from 2000 to 2010. The data warehouse includes a dynamic cohort
of 13,132,723 individuals who resided in Lombardy from 2000 to 2010, and health care
records for those who received health care for any reason at a regional health care
provider. The data warehouse includes almost 3 billion health-care events of the types
described in the previous pages of this Appendix. In accordance with the Lombardy
Privacy Policy, the data warehouse DENALI does not contain any information that can
lead to the identification of individual patients.
The data has been collected from a variety of data sources, and has been loaded on the
data warehouse using DENALI-ETL, a custom made ETL system. The core functions
of DENALI-ETL support the validation and cleaning of the data sources to ensure high
data quality on the data warehouse. Validation and data cleaning are performed off-line.
The data cleaning is performed in the following steps:
1. Definition of the Meta Schema of the data warehouse
2. Analysis of the source data to identify inconsistencies and errors
3. Mapping of the source data to consistent data types and conventions across the data
warehouse
4. De-duplication and consolidation of the Registry of Vital Records to identify Vital
Records belonging to the same individual that had been split due to multiple
identification codes assigned to the same individual
5. Linkage of the health care events to the Registry of Vital Records
The linkage is performed in three phases:
a) The first phase identifies all the health care events that can unequivocally be linked to
the same individual described by a set of Records of Vital Statistics
b) The second phase utilizes a Probabilistic Matching Approach to identify clusters of
“Best Matching” Vital Statistics Records to which health care records may be linked
c) The third phase utilizes graphical tools to filter “Likely” matches out of the clusters
of “Best Matching” Vital Records
The de-duplication of the Registry of Vital Statistics and the Probabilistic Linkage of
health care events to the Registry of Vital Statistics are performed by DENALI-ETL
using a variety of Machine Learning techniques
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 194
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
The Lombardy Privacy Policy is strictly enforced during all the phases of data cleaning
The cleaned data are finally loaded on the data warehouse and are equipped with control
structures that optimize epidemiological queries.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 195
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
MENTAL HEALTH SERVICES (PSICHE) DATABASE
Complete coverage and good quality data since 2005
Socio-demographic, diagnostic and clinical data on Lombardy Region patients assisted
by Psychiatric Operative Units (Unità Operative di Psichiatria) of Lombardy Region is
arecorded. All contacts of each patient with mental-health services, including
Departments of Mental Health, Residential Communities, etc. are recorded.
Main fields
Anonymized identification code
Marital status (eg. married, single, etc.)
Living conditions (eg. alone, in a family, in a community, etc.)
Employment status (eg. employed, unemployed, etc.)
Educational level (eg. primary school, secondary school, etc.)
Reference Psychiatric Operative Unit code
Psychiatric visit date
Psychiatric diagnosis code (ICD-10 code)
Types of professionals at the visit (physicians, nurses, psychologists, social workers,
educators)
Type of visit (eg. group therapy, psychological counseling, etc.)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 196
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
DELIVERY ASSISTANCE FORMS DATABASE
Complete coverage and good quality data since 2005
Assistance to Delivery Forms (CEDAP – CErtificato di Assistenza al Parto) concerning
deliveries occurred in Lombardy Region are recorded. The CEDAP certificate must be
completed by the midwife, or supervising physician, at the moment of the birth.
Mother’s, father’s and child’s socio-demographic, antropometric and clinical data are
recorded.
Main fields
Mother’s anonymized identification code
Number of mother’s previous deliveries (and date of last delivery, if occurred)
Number of mother’s previous miscarriages or voluntary abortions
Number mother’s physician visits during pregnancy
Number of pre-natal screenings
Number of ultrasound screenings
Gestational age (in weeks)
Presence of fetal growth anomalies
Codes identifying whether the pregnancy was achieved by medically-assisted reproduction
(and used techniques)
Codes identifying whether labor induction (and used techniques) was carried-out
Fetal presentation
Delivery type (spontaneous or cesarean)
Health-care professionals presence at the birth
Birth outcome (eg. single birth, twin birth, …)
Newborn’s Gender
Newborn’s characteristics (weight, length, cranial circumference)
Newborn’s vital status
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 197
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Newborn’s Apgar score
Presence of malformations affecting the newborn
Presence of fetal diseases affecting the newborn
Socio-demographic characteristics of both the parents (e.g. place of delivery, marital
status, nationality, etc…)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 198
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
VACCINATIONS REGISTER
Complete coverage and good quality data since 2006
Childhood vaccinations concerning children residents in the Lombardy Region and born
from 2000 and afterwards are recorded.
Main fields
Anonymized identification code
Vaccine code
Vaccine description
ASL where vaccination was performed
Vaccination date
ASL: Azienda Sanitaria Locale (Local Health Unit)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 199
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Pharmacovigilance Database
Complete coverage and good quality data since 2001
In Italy, spontaneous ADRs are collected through the National Network of
Pharmacovigilance (Rete Nazionale di Farmacovigilanza), an extensive network
throughout the national territory that includes the Regional Authorities and the
Autonomous Provinces of Trento and Bolzano, the Regional Centres of
Pharmacovigilance, more than two hundred Local Health Authorities, about one
hundred Hospitals, 43 Research Institutes and more than eight hundreds Pharmaceutical
Companies and AIFA.
The RNF is also operating in connection with the European network for
pharmacovigilance EudraVigilance that collects in a single database all data provided at
national level by the EU countries.
Main fields
Initials, age and gender of the patient
Hospital identification code
Hour and date of ADR
One or more ADRs
Seriousness of ADRS
Kind of seriousness (death, disability, hospitalization, etc.)
Outcome of the ADR
One or more drugs suspected
Indication for prescription of any suspected drug
Date of beginning and end, dosage, route of administration of any suspected drug
One or more drugs concomitants
Indication for prescription of any concomitant drug
Name and coordinates of reporter
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 200
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 201
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Appendix 2
Methods for controlling misclassification and confounding
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 202
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
A2.1. Measurement errors and misclassification
Errors can occur in measuring both exposure and outcome. These errors lead to classification bias, that is (i)
identifying subjects as having experienced the disease outcome when they have not (or being exposed to drug
when they are not exposed), or (ii) not having experienced outcome when they experienced it (or being not
exposed when they are exposed).
Classification bias is further categorized as differential or nondifferential. Differential misclassification is
present when the likelihood of misclassification is different between exposed or outcome groups. An example
of differential misclassification for drug exposure is when patients who are exposed have a lower likelihood
of outcome misclassification because to receive medication they have to enter the health-care system, which
increases their likelihood of recording a diagnosis. Those not exposed are much more likely to be
misclassified as not having the disease, which is an artefact of not entering the health care system [1].
Nondifferential misclassification occur when the likelihood of misclassification is the same across the
exposed or outcome groups. Examples of nondifferential misclassifications are those caused by coding errors
of diagnosis or medical procedure reported in hospital discharge database, medicaments reported in drug
prescription database, or identifier code (e.g. in fiscal or health individual code) due to accidental mistyping.
Upcoding, assigning codes of higher reimbursement value over codes with lesser reimbursement value, is an
additional source of error at the coder level [2].
The direction of misclassification on measures of association will depend on its type [3]. When the
misclassification is differential, association measures would be biased either toward or away from the null
[4]. The effect of nondifferential misclassification varies by the factor that is misclassified. When outcome
variables are subject to nondifferential misclassification, association measures are typically biased toward the
null [5]. When nondifferential affects variables used to define cohorts (by excluding prevalent disease at
baseline), association measures could be biased away from the null, with the degree of bias varying by
disease incidence and prevalence [6]. Therefore, the effect of misclassification on study conclusions even
when a single variable alone is misclassified is unpredictable.
A2.1.1. Outcome misclassification
A specific disease coding must be used for extracting patients experiencing a given outcome and for
characterizing their clinical profile from HCU data. The International Classification of Diseases, 9th Revision,
Clinical Modification: ICD-9-CM (http://icd9cm.chrisendres.com/), and the ICD-10 version in few countries
(http://apps.who.int/classifications/icd10/browse/2010/en), are the classification system of disease and
medical procedures, common to most HCU databases. Each disease coding system necessarily implies a loss
of information as a simple consequence of classification, as opposed to nomenclature. Many authors have
discussed the qualitative loss that occurs with ICD-9-CM coding [7-12]. Besides problems with billing
considerations distorting coding, and the errors of coding caused thereby, they note that coding diagnosis
categories loses essential information about the true conditions from which patients suffer [13].
To understand the effect of outcome misclassification on association estimates, it is important to note that a
lack of specificity of the outcome measurement is worse than a lack of sensitivity in most situations. If
specificity of the outcome assessment is 100%, then relative risk estimates are unbiased [14]. Given this, the
literature on misclassification of HCU data diagnoses is not quite as depressing as it first seems. A recent
comprehensive study on the misclassification of HCU data diagnoses using medical records review as the
gold standard revealed that the sensitivity of claims diagnoses is often less than moderate, whereas their
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 203
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
specificity is usually
95% or greater (Table)
[15].
A
high
specificity
of
Sensitivity
Specificity Sensitivity
Specificity
diagnostic coding in
Hypertension
60.6
87.7
65
99.9
HCU data can be
COPD
53.4
87.9
91
98.8
expected because if a
Diabetes
62.6
97.2
88
99.4
diagnosis is coded and
Renal failure
18.6
99.0
88
99.4
recorded it is likely
Chronic liver disease
27.6
99.8
100
100
that this diagnosis was
Any cancer
44.8
95.0
91
100
made, particularly in
Peptic ulcer disease
27.6
94.6
92
100
hospital
discharge
Congestive heart failure
41.5
96.1
85
99
summaries
[16].
AMI
25.4
96.8
94
100
Under this point of
Neutropenia
97
view, HCU data may
Stevens-Johnson syndrome
95
be assumed to be
suitable for association studies (e.g. for studying the effect of healthcare on the onset of a given outcome),
rather than for measuring diseases’ frequency (e.g. incidence and prevalence).
Since chronic diseases usually require multiple contacts with the health system, a single diagnostic code may
be insufficient to accurately identify cases [17]. This explains the widespread use of diagnostic algorithms for
identifying patients experiencing a given outcome [18]. The use of hospital charts for identifying cases,
furthermore, limits the possibility of detecting all the outcomes (e.g. those that do not require hospital
admission), thus introducing a bias due to the selective inclusion of more severe outcomes. Alternative
techniques have been specifically developed for detecting less severe outcomes. For example, in the situations
where a certain drug A is suspected of causing an outcome that itself is treated by a drug B, we only need of
detecting patients who experience outcome from drug prescription database. In this way, the effect of
antibacterials on the risk of arrhythmias has been recently investigated in a study that estimated the
association between the use of drugs belonging to the classes of antibacterials (exposure) and antiarrhythmic
(outcome) [19]. Yet, the well known relationship between the use of statins and the risk of diabetes [20]
might be investigated by studying the strength of the association between statin use (exposure) and the risk of
starting an antidiabetic therapy (outcome). It should be considered, however, that patients who statins were
prescribed are those at whom diabetes may be more likely discovered. In other words, we cannot exclude that
a detection bias occurs [21] with the consequence that the exposure-outcome relationship would be amplified.
Ambulatory care billing
diagnoses
Hospital discharge
diagnoses
A2.1.2. Exposure misclassification
Drug coding system, i.e. the Anatomical Therapeutic Chemical (ATC) classification system of WHO
(http://www.whocc.no/atc_ddd_index/), must be used for extracting patients who use a given medicament, as
well as those who use other drugs as proxies of their clinical profile, from HCU data.
Pharmacists fill drug prescriptions with little room for interpretation and are reimbursed by Health
Authorities on the basis of detailed, complete, and accurate claims submitted electronically [22-24].
Pharmacy dispensing information is therefore expected to provide highly accurate data, also because filing of
an incorrect report about the dispensed drugs has legal consequences [25]. These data are usually seen as the
gold standard of drug exposure information compared with self reported information [26] or prescribing
records in outpatient medical records [27].
The time-window during which patients are considered exposed to a given drug is often a key variable in
assessing exposure [1]. This variable, at least, requires that the information for calculating the days covered
by each prescribed drug canister is available, or may be approximated. Within most USA prescription
databases, the “days supply” is usually included as a data field within each prescription claim (e.g., 60 tablets
of a medication that is taken twice daily would yield a 30-day supply), along with the dates that the
prescription was dispensed [28]. In several non-USA databases (e.g. in Italy), information about date
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 204
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
dispensation and dispensed quantity are available, but not those about the prescribed drug dose. This compels
for approximating the number of days covered by each dispensed canister by means of the quantity of the
drug contained in the dispensed canister, plus some metric of average daily dose, e.g., defined daily dose
(DDD) established as the typical adult’s daily maintenance dose [29]. It should be emphasized, however, that
the DDD is a unit of measurement, and does not necessarily reflect the recommended or Prescribed Daily
Dose (PDD). Discrepancies between PDD (which may by assumed as a gold standard of exposure duration),
and DDD (which is likely affected by measurement error) necessarily introduce classification bias.
Drug therapy from
hospital setting
Fill of a 30d
supply
15d
30d
Fill of a 30d
supply
20d
True drug exposure pattern
of a sample patient
20d
30d
Drug exposure pattern
according to pharmacy records
45d
30d
Drug exposure pattern according to
pharmacy records plus 50% rule
starting prescription
time patient is truly exposed
discontinuing use
time patient is classified as exposed
Figure 1 illustrates some typical misclassification problems in drug exposure assessment [30]. If the
calculated days supply is too short or if patients decide to stretch a prescription by using a lower dose (e.g.,
tablet splitting), some period will be classified as unexposed when it truly is exposed. Most chronically
administered drugs are used for longer periods, resulting in multiple refills. A patient can thus be classified as
being unexposed intermittently despite continuous exposure. Many investigators therefore extend the
calculated days supply by some fraction (e.g., 50%) to avoid this misclassification. However, this strategy
aggravates another misclassification that can occur if a patient discontinues drug use without finishing the
supply. The right balance between improved sensitivity of drug exposure assessment and specificity depends
on how well the days supply is calculated; this depends on the type of drug and how regularly it is taken.
Over-the-counter (OTC) medications present a scenario in which misclassification is particularly problematic
[31]. Measurements based on HCU data will underestimate the use of OTC products and lead to
misclassification of exposure to those medications [32,33]. The inability to measure exposure during the
observation period can also be problematic if the available data do not fully capture all sources of exposure.
The use of OTC medication as an exposure is but one example of not being able to accurately capture all
exposures, but this can occur in other circumstances. For example, in most databases, hospital discharge files
do not contain drug use information. Therefore, every hospital stay represents a period with missing drug
exposure information by design, thus generating a particular form of exposure misclassification labelled
immeasurable time bias [34]. Furthermore, exposure may be misclassified because of protopathic bias, i.e. the
drug use could be attenuated or interrupted owing to the onset of early symptoms of outcome disease [35]. It
should be emphasized that both the sources of bias likely generate differential exposure misclassification
because mainly affect patients who experience the outcome.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 205
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
A2.1.3. Strategies of accounting for misclassification
A2.1.3.1. Algebraic methods
The impact of even modest misclassification can be profound but is rarely quantified. If the sensitivity and
specificity of an exposure or outcome measurement are known, simple algebraic methods can be applied for
unmatched [5] and matched analyses [35]. All these methods are, however, applied only to unadjusted
associations (i.e., crude 2×2 tables), which is an unrealistic analysis in pharmacoepidemiology and healthcare
research. Other techniques use the positive predictive value (PPV) of the outcome measure to be determined
in a separate validation study [36,37]. This is an interesting approach for database studies because typically
PPVs are easier to estimate in internal validation studies than sensitivity and specificity. This approach
requires that (i) a sample of records are randomly selected from the HCU database; (ii) the corresponding
medical documentations are traced at the provider (i.e. the hospital) where the records were generated; (iii)
medical documentations are reviewed by a trained physician, possibly blinded with respect to the diagnostic
code as reported in the HCU database. With the aim of protecting the privacy of patients, however, the use of
methods for data anonymization, coupled with oversight by institutional review boards, can make this
approach unfeasible.
A2.1.3.2. Sensitivity analyses
Because of the lack of detailed information make compelling the use of approximations for defining both
outcome and exposure, the effect of such approximations can be explored by sensitivity analyses. Sensitivity
analysis tests different values or combinations of factors that define a critical measure in an effort to
determine how those differences in definition affect the results and interpretation [38]. For example, the
influence of diagnostic criteria for outcome definition (e.g. using alternative ICD-9 codes or algorithms for
combining diagnostic codes), or the length of the time-window for exposure definition (e.g. extending the
calculation of days supply by means of adding different fractions to calculation based on DDD) are
commonly used techniques for investigating robustness of findings by varying criteria adopted in principal
analyses. Yet, the application of various methods for correcting immeasurable time bias [34], protopathic bias
[39], and detection bias [40] would be used when suitable.
Finally, a sensitivity approach for estimating the effect of over-the-counter drugs on the exposure-outcome
association has been recently proposed [41].
A2.1.3.3. External adjustment
Methods that use individual validation data for accounting measurement errors have been available for some
time [42].Among these, regression calibration (RC) may play an interesting role. RC is an intuitive, no
iterative statistical method useful for adjusting point and interval estimates obtained from regression models
for measurement-error bias [43].
To illustrate the RC method, consider a HCU-based study measuring the effect of X’ (e.g. the length of the
time-window of exposure to a given drug measured as a continuous variable) and a dichotomous outcome Y
(Y=1 for event, 0 for no event). The effect of X’ on the risk of Y is usually obtained from the logisticregression model
logit[ P(Y = 1 | X ' )] = β 0 + β ' X '
As above specified, however, X’ is likely affected by measurement error since it is approximated from DDD.
When a gold standard assessment is available, a validation study conducted in a random subsample of
patients included in the main (HCU) study can be used to validate the usual exposure measure (X’) against its
gold standard (X). For example, data obtained by a sample of physicians operating in the same area as the
studying target population, may be informative about the drug doses prescribed to patients. Assuming
prescribed dose as gold standard (X), and considering that X’ is also measurable from this validating data
source, the error term may be easily obtained by the linear-regression model
E ( X X ') = α + γX '
Rosner et al [44] proposed to estimate logarithm relative risk by substituting β’ with
βˆ = βˆ ' / γˆ
and
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 206
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
()
var βˆ =
( )
ˆ2
1
ˆ ' + β ' var(γˆ )
var
β
γˆ 4
γˆ 2
The RC method is therefore an appropriate approach when information about the true exposure is available
from a validation study and the assumption of a linear relationship between observed and true exposure
measures is not markedly violated.
RC was originally proposed by Willett in a study aimed of measuring the effects fat and fiber and the
incidence of breast cancer [45], and has been widely used in the field of nutritional epidemiology [46,47] and
other fields [48,49]. Strangely, despite RC does not involve particular computational complexities, it is not as
popular in the field of pharmacoepidemiology and healthcare research.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 207
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
A2.2. Confounding and beyond
Issues surrounding misclassification is not the only bias that researchers are faced with when using electronic
databases. Confounding also comes into play. Confounding occurs when the estimate of exposure-outcome
relationship is biased by the effect of one or several confounders. A confounder is an independent (causal)
risk factor for the outcome of interest that is associated with the exposure of interest in the population, but
that is not an intermediate step in the causal pathway between the exposure and the outcome [50,51]. For
example, low-density lipoprotein level would be expected to be a confounder in a study of the effects of statin
treatment on the risk of cardiovascular events. Low-density lipoprotein level may lead a physician to
prescribe a statin and also be an independent risk factor for cardiovascular events.
A2.2.1. Sources of confounding
Patient’s exposure to a medical therapy is determined by system-, physician-, and patient-level factors that
may often interact in complex and poorly understood ways [52]. For example, a physician’s treatment
decision may be based on an evaluation of the patient’s health status and prognosis. Patients may initiate and
remain adherent to a new therapeutic regimen because of their disease risk and the benefits of treatment.
Treatment initiation and adherence may also depend on a patient’s physical and cognitive abilities. Patient
and physician factors determining use of a treatment may directly affect health outcomes, or be related to
them through indirect pathways. From this process, several sources of bias can result.
Confounding by indication or severity. A common, pernicious and often intractable form of confounding
endemic to pharmacoepidemiologic and healthcare research studies is confounding by indication of treatment:
physicians’ tendency to prescribe medications to and perform procedures on patients who are most likely to
benefit [52]. Because it is often difficult to assess medical indications and underlying disease severity and
prognosis, confounding by indication often makes medications appear to cause outcomes they are meant to
prevent [53,54]. For example, statins, lipid-lowering drugs, reduce risk of cardiovascular (CV) events in
patients with CV risk factors. Thus these drugs tend to be prescribed to patients perceived to be at increased
CV risk. Incomplete control of CV risk factors can make statins appear to cause rather than prevent CV
events [52].
Confounding by contraindication. When an adverse event is known to be associated with a therapy,
confounding by contraindication is possible. For instance, women with a family history of venous thrombosis
may avoid postmenopausal hormone therapy [55].
Confounding by functional status. Patients who are functionally impaired (defined as having difficulty
performing daily activities of living) may be less able to visit a physician or pharmacy and therefore may be
less likely to collect prescriptions and receive healthcare services [52]. This phenomenon could exaggerate
the benefit of prescription medications, vaccines, and screening tests. For example, functional status appeared
to be a strong confounder in studies of both the effect of non-steroidal anti-inflammatory drugs (NSAIDs) and
the influenza vaccine on all-cause mortality in the elderly [56-58].
Confounding by cognitive impairment. A similar form of confounding could result from differences in
cognitive functioning. Depression may be considered as a example of such a bias because evidence exists that
depression is a strong risk factor for several outcomes, such as CV disease [59], and its presence is
accompanied by a less frequent use of healthcare services, e.g. lower compliance to treatment [60].
The healthy user and healthy adherer bias. Patients who initiate a preventive medication may be more
likely than other patients to engage in other healthy, prevention-oriented behaviours [52]. For example,
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 208
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
patients who start a preventive medication may be more likely to seek out preventive healthcare services,
exercise regularly, moderate their alcohol consumption, and avoid unsafe and unhealthy activities. Incomplete
adjustment for such behaviours can make use of preventive medications spuriously associated with reduced
risk of a wide range of adverse health outcomes.
Similarly, patients who adhere to treatment also may be more likely to engage in other healthy behaviours
[61,62]. Strong evidence of this “healthy adherer” effect comes from a meta-analysis of randomized
controlled trials where adherence to placebo was found to be associated with reduced mortality risk [63]. This
is clearly not an effect of the placebo but is rather due to characteristics of patients who take a medication as
prescribed. The healthy adherer bias is also evident in studies that reported associations between statin
adherence and an increased use of preventive healthcare services and a decreased risk of accidents [64,65].
The healthcare access bias. Patients may vary substantially in their ability to access healthcare [52]. For
example, patients who live in rural areas may have to drive long distances to receive specialized care. Other
obstacles to accessing healthcare include cultural factors (e.g., trust in medical system), economic factors
(e.g., ability to pay), immigration status, and institutional factors (e.g., restrictive formularies), all of which
may have some direct or indirect effects on study outcomes.
A2.2.2. Strategies of accounting for confounding: a general guide
Confounding by indication, as well as other sources of confounding, is not an insurmountable problem [66].
This belief is based on two assumptions. The first is that the magnitude of confounding may be small because
physicians' treatment decisions may have little relationship to patients' pre-treatment prognostic
characteristics. Evidence in support of this assumption comes from studies of the phenomenon of practice
variation [67-71]. The reported magnitude of practice variation seems so large that some researchers have
inferred that it could not arise from variability in patients' characteristics (e.g., illness rates, insurance
coverage, or preferences) [72] and, therefore, physicians pay little attention to individual patients' clinical
characteristics when making decisions [73]. This assumption, however, doesn’t take into account that
unbalancing of measured features among drug user categories has been repeatedly reported. For example,
newer sedative-hypnotics are preferentially prescribed to frail elderly patients more likely to experience falls
and hip fractures to avoid such outcomes. The construct of frailty is difficult to measure in HCU databases
[74] and led to an overestimation of the association of newer sedative hypnotics with hip fractures when
compared with users of traditional benzodiazepines or non-users [75]. Generalizing, then, because of both
magnitude and direction of confounding implications are often unpredictable, the use of adequate tools for
taking into account for confounding is always need.
The second assumption is that HCU databases contain sufficient and sufficiently accurate information about
patients' pre-treatment prognostic differences so that one can make valid corrections for these differences. For
example, Roos et al stated "administrative data has been shown to do nearly as well for risk adjustment as
data that rely on physiological measures and physician judgment of health status." [76] All researchers,
however, have not been comfortable with the role of observational outcomes studies using HCU databases
[77-79]. Controversy has focused on accuracy and sufficiency of information available in HCU databases to
control for difference in populations receiving different treatments and, more broadly, receiving care from
different hospitals, providers, or health care systems.
Strategies to adjust confounding vary depending on whether the potential confounders are measured in a
given database. If confounders are measured, then usual (basic) strategies of accounting for confounding
include those concerning study’s design (e.g. restricting and matching), and those concerning data analysis
(standardization, stratification and regression). These techniques well described in standard epidemiology
texts [80] and can be directly applied to database studies with the usual caveats. However, the degree to
which these clever devices achieve the goal of fully control for confounding remains unpredictable, since
unknown, unmeasured, or immeasurable confounders may strongly impact the findings and the consequent
direction of bias may be unpredictable. Residual confounding refers to factors that have been incompletely
controlled, so that confounding effects of these factors may remain in the observed treatment-outcome effect.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 209
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Strategies of accounting for residual confounding include those concerning study’s design (e.g. only-case
designs) and data analysis (e.g. sensitivity analysis and instrumental variables approach).
The implications of these techniques when they are applied to HCU based studies, need to be discussed. The
purpose of the following paragraphs is to show some usual and emerging techniques of adjusting for
measured and unmeasured confounders in the framework of observational studies based on HCU data.
A2.2.3. Accounting for confounders through study design
A2.2.3.1. Restricting the study cohort
The basic idea of restricting a study cohort is to make its population more homogeneous regarding measured
patient factors. A study may want to restrict its population to the oldest patients and thus minimize the
influence of age. Restriction will by definition reduce the cohort size, but population based HCU databases
are of such massive size that some restriction to improve the validity of findings will usually not impair
precision meaningfully.
There are many different approaches to restriction in specific studies [81] and it is therefore difficult to
provide generic advice that fits specific study designs. However, several guiding principles can be identified
that should be considered in HCU database study on effectiveness and safety of medical interventions [82].
Three restrictions are generally worth considering in comparative effectiveness research [83].
Restricting to users and choosing a comparison group. Choosing a comparison group is a complex and
sometimes subjective issue. The ideal comparison should comprise patients with identical distributions of
measured and unmeasured risk factors of the study outcome [1]. Selecting comparison drugs that have the
same perceived medical indication for head-to-head comparisons of active drugs (i.e. comparing the effects of
two active and competing therapies that are prescribed under the assumption of identical effectiveness and
safety [30]) will reduce confounding by selecting patients with the same indication [1]. For this reason,
excluding nonusers patients and, consequently, actively comparing patients exposed to drugs with the same
indication (i.e. the basic of comparative effectiveness research) has become increasingly popular in the field
of observational studies using HCU data [84]. However, new competitors within a class are often marketed
for better efficacy, slightly expanded indications, or better safety influencing physicians’ prescribing
decisions [85]. In this way, new sources of confounding by indication can arise.
Restricting to new users. As mentioned above, the basic cohort design identifies all patients in a defined
population who were treated with the study medication during a defined study period. Such a cohort will
consist of prevalent (ongoing) and incident (new) drug users; depending on the average duration (chronicity)
of use, such cohorts may be composed predominantly of prevalent users and few new users. The estimated
average treatment effect will therefore underemphasize effects related to drug initiation and will
overemphasize the effects of long term use [86]. Further, prevalent users of a drug have by definition
persisted in their drug use, which may correlate with higher educational status and health-seeking behaviour,
particularly if the study drug is treating a non-symptomatic condition, e.g., BP lowering agents for treatment
of hypertension, statin for treatment of hyperlipidaemia or hormone replacement therapy [74,87].
Consequently, the restriction to new initiators of the study drugs (inception cohort) will mitigate those issues
and will also ensure that patient characteristics are assessed before the start of the study drug.
Support for eliminating prevalent users in observational studies is supplied from a recent study that evaluated
the effect of excluding prevalent users of statins in a meta-analysis of observational studies of persons with
CV disease [88]. The pooled, multivariate adjusted mortality hazard ratio for statin use was 0.77 in 4 studies
that compared incident users with nonusers, 0.70 in 13 studies that compared a combination of prevalent and
incident users with nonusers, and 0.54 in 13 studies that compared prevalent users with nonusers. The
corresponding hazard ratio from 18 RCTs was 0.84. It appears that the greater the proportion of prevalent
statin users in observational studies, the larger the discrepancy between observational and randomized
estimates. The advantage of the new user design has been summarized by Ray [86].
Restricting to adherent patients. Patients dropping out of RCTs for reasons related to the study drug may
cause bias. Non-informative discontinuation causes bias toward the null in intention-to-treat analyses. The
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 210
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
medical profession and regulatory agencies accept such bias because its direction is known and trial results
are considered conservative regarding the drug’s efficacy. Discontinuation of treatment may also be
associated with study outcomes through lack of perceived treatment effect and drug intolerance.
RCTs try to minimize bias from non-adherence by frequently reminding patients and by run in phases before
randomization to identify and exclude non-adherent patients. In routine care, adherence to drugs is
substantially lower than in RCTs. It has been recently shown that in 36-37% of patients who start therapy for
hypertension, hyperlipidaemia or type 2 diabetes, initial treatment is not followed by another specific
dispensation [89]. The study also showed that patients at whom an isolated prescription was dispensed had
clinical features (e.g. cotreatments and comorbidities), as well as rate of hospitalization for CV events,
intermediate between those of patients at whom the considered medicaments were more regularly prescribed
and those of individuals at whom such medicaments were never been dispensed. Therefore, isolated users
would be considered a heterogeneous category of individuals including those who would have needed
continuous drug therapy and those in whom the lack of prescription renewal may be considered a later
correction of an inappropriate initial drug treatment. Similarly, it has been consistently shown that only 45%,
50% and 40% of patients who respectively start therapy for hypertension, hyperlipidaemia or type 2 diabetes,
refill their prescriptions after one year [90-92].
Several studies based on HCU data start follow-up after the second or third refill of the study drug in new
user cohorts with the aim of excluding patients who are least adherent. External validity (generalizability) is
of concern by this restricting criterion. It should be considered, however, that to make a prescribing decision,
physicians must assume that patients will take a drug as directed. If clinicians knew beforehand that a patient
would not take a prescribed medication, they would not ponder the appropriateness of the drug in the first
place [83].
A2.2.3.2. Matching
Matching is one of the techniques used to avoid confounding through study design. In a cohort study this is
done by ensuring that once an exposed subject entered into the study at a given data (e.g. because that day
he/she experienced the first prescription of the drug of interest), one or more individuals belonging to the
same population as that generated exposed ones are included on condition that up to that data they did not
have experienced exposure, and that, with respect to the exposed subject, they presented the same features
which we think may confound the relation of interest. A good example of a matched cohort study was
presented by Ludvigsonn et al. [93] who investigated the relation between celiac disease (CD) and risk of
renal disease. In this study, 14,336 CD patients and 69,875 patients without CD were matched for gender,
age, calendar year, and county. As a result of this matching these variables had an equal distribution among
both groups, therefore these variables had no effect as confounder. Greenland & Morgenstern showed that
matching can reduce the efficiency of a cohort study, even when it produces no sample-size reduction and
even if the matching variable is a confounder [94]. This perhaps explains because this technique is not as
popular in the field of observational cohort study [95].
In a nested case-control study, a case (i.e. a subject who had experienced the outcome at a given data), is
always matched with one or more controls (i.e. individuals who did not experienced the outcome from the
cohort entry till that data). For this design the matching is needed for ensuring that members of each casecontrol(s) set, have the same length of observational time-window. In addition, however, case and controls
might be matched for other variables/features which we think may confound the relation of interest. For
example, once a case has been identified, one or more controls may be included a condition that they entered
into the cohort at the same time, were still at risk of experiencing the outcome, and had the same age as the
index case. In this way, we ensure that case and controls are balanced, other than for duration of observational
period, also for age. In this way age cannot confound the investigated association.
Because of its easiness and applicability, nested case-control studies are often designed taking into account of
matching for confounding adjustment. This technique, however, has at least two weaknesses. First, once
matching has been performed, the effect of matching variables on the outcome risk cannot be measured.
Second, overmatching is always in ambush when using this technique [96,97]. This phenomenon can be
explained by a theoretical example. Suppose we would be able to match for all the variables affecting
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 211
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
outcome. Both cases and controls would become almost completely similar resulting in an odds ratio of
approximately 1 [98]. More in general, overmatching is introduced when at least one of these three conditions
happens: (i) matching variable is not an independent risk factor for the outcome, thus reducing the efficiency
of the analysis [99]; (ii) matching variable is not associated with exposure (or matching variable is a proxy of
exposure), thus resulting in an obscured exposure-outcome relation; (iii) matching variable is on the causal
exposure-outcome pathway. A motivating example is given by studies investigating the impact of gestation
length and plurality on short-term outcome of in vitro fertilization (IVF)-children. Because of the high
number of multiple and preterm births is an inherent part of current IVF practice, matching the control group
by gestation length and/or the number of multiples births may yield misleading results on the total health
impact of IVF, and therefore it should be avoided [100].
Therefore, matching should be considered in case-control studies only if the matching factor is well known to
be an independent risk factor for disease and unlikely to be on the causal pathway between the exposure of
interest and the disease. However, given our little a priori knowledge on the possible effect of all the external
variables of interest, and owing the large sample size of studies based on HCU data, other techniques would
be like better for accounting confounders.
A2.2.3.3. Case-only designs
Although cohort and nested case-control studies are widely accepted designs for the evaluation of the risks
and benefits of post-licensure medications, these designs are vulnerable to confounding. In the late 1980’s,
alternative methods relying only on cases (i.e. without controls), termed case-only designs, were introduced to
attempt to accounting for unmeasured confounders [101]. Case-only designs are attractive because the cases
are self-matched, which eliminates time-invariant confounders. They are generally less expensive, shorter in
time, and simpler to carry out than conventional designs [102]. Among existing case-only designs, five have
been used in pharmacoepidemiology: the case-crossover design [103], the self-controlled case series design,
originally called case series analysis [104], the case-time-control design [105], the screening method [106]
and the prescription sequence symmetry analysis also called the symmetry principle [107]. Of those designs,
the first two are those more frequently used and will be briefly described following.
Case-crossover design. The case-crossover (CC) design was introduced by Maclure in 1991 to study the
short-term effects of intermittent exposures on the risk of acute outcomes [103]. In a CC study, case-subjects
are their own self-matched controls by using pre-defined time period(s). Probability of exposure is compared
between risk (or hazard) time-period(s) preceding the event of interest (where the exposure is considered
associated to the event), and control period(s) defined in the observation period. Definition of these time
periods (duration, etc.) depends on the studied outcome and exposure. The odds ratio for the outcome is
estimated with a conditional logistic regression. Only discordant pairs, i.e. cases exposed only in the risk
period or only in the control period, contribute to the analysis.
In a CC analysis, confounding by constant characteristics is implicitly eliminated. On the other hand, a bias
due to exposure time trend is introduced. Case-time control (CTC) design adjusts the CC estimate on a timetrend in the exposure, by means of estimating it from a group of control subjects [105].
Self-Controlled Case Series. The self-controlled case-series (SCCS) design was introduced by Farrington in
1995 to assess post-licensure adverse events related to vaccines, and more generally associations between
acute outcome and transient exposure [104]. The SCCS design focuses on the relative incidence of an
outcome between risk (post-exposure) and control time periods. Risk periods are defined during and/or after
an exposure, when people are hypothesized to be at greater risk of the event, whereas control periods include
all time period with baseline risk, which may occur both before and after the exposure [108]. The statistics of
this incidence rate ratio relies on a conditional Poisson regression. The tutorial by Whitaker et al [109] is a
very useful guide for this method.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 212
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Design
Direction
Rationale
Comparison of
between-period
outcome incidence rates
SELFCONTROLLED
CASE SERIES
Period uncovered by
drug exposure
Period covered by drug
exposure
Outcome onset
Design
CASECROSSOVER
Direction
Comparison of
between-period number
of couples disagreeing
for drug exposure
wash-out
At risk period
Rationale
Reference period
Outcome onset
Figure 2 depicts the two principal case-only designs. It clarifies that, while the CC design relates to a case
control study (by comparing the probability of exposure between case and control periods), the SCCS relates
to a cohort study (by comparing the probability of events between exposed and non exposed periods). It
should be emphasized, however, that both the designs are particularly suited for studying the effect of
intermittent/transient exposures on the risk of acute outcomes [110]. Thus, for studying chronic or cumulative
effects of long term exposures other methods for accounting residual confounding are better suitable.
A2.2.4. Accounting for confounding through data analysis
1
2
Proxy of unmeasured
confounder
Measured
confounder
Measured
confounder
Exposure
E
Outcome
D
RRED
Exposure
E
3
4
Unmeasured confounder
C
OREC
Exposure
E
Measured
confounder
RRED
RRCD
Outcome
D
Outcome
D
RRED
Unmeasured
confounder
Instrumental
variable
IV
Measured
confounder
Exposure
E
RRED
Figure 3 depicts causal graphs
illustrating
data
modelling
techniques of accounting for
confounding [83,111]. Ideally,
we would be able to fully assess
all the features which make
unbalanced compared groups
and then taking into account for
these features by means of
stratification and regression
modelling
(Figure
5.1).
However, as above specified,
most non-randomized studies
using HCU data with limited
patient information will not be
able to fully measure and adjust
Outcome
D
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 213
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
for such confounders, being therefore unable to show an outcome effect of exposure because of residual
confounding. This issue may be faced by means of several devices, such as resorting to proxy variables
(Figure 5.2 and par.A2.2.4.2), applying some techniques of sensitivity analysis, e.g. external adjustment
models (Figure 5.3 and par.A2.2.4.3), or using instrumental variables (Figure 5.4 and par.A2.2.4.4).
A2.2.4.1. Stratification and regression modelling
Stratification, similar to restriction, identifies patient sub groups based on measured patient factors [112]. In
contrast to restriction, stratification does not discard the “unwanted” population but provides treatment effect
estimates for all strata and combines them into one weighted summary effect measure [83]. In the absence of
effect measure modification, e.g., the treatment effects are the same in old and young patients, and under the
assumption that all confounding factors were measured; stratified analyses will provide unbiased treatment
effects. The large size of HCU databases may permit many such subgroup analyses with substantial numbers
of subjects and is an attractive alternative to wholesale restriction [83].
Regression analyses use statistical modelling to make stratified analyses more efficient by assuming that
parametric statistical distributions fit the data [113]. Study design, conditioning the forms of study outcome,
exposure of interest, and included covariates, will determine the regression model to be used. For cohort
designs, modelling time-to-event data with variable follow-up and censoring of study outcomes, the common
methodology of analysis is the Cox proportional hazards regression. In particular, this approach can easily
handle exposures and study covariates whose values vary over time. When time-varying covariates are
affected by time-varying treatment, marginal structural models may be required. A number of excellent
textbooks describe analyzing time-to-event data [114,115]. For matched study designs (e.g., nested casecontrol design), conditional logistic regression may be considered. Finally, if the study outcome is binary
with fixed follow up and is rare, Poisson regression with robust standard errors can be used to estimate
relative risks and get correct confidence intervals [116,117].
There are a number of analysis options that must be considered, which depend on the study question and
particulars of the study design. For example, repeated outcomes, such as asthma attacks leading to emergency
room admissions, can multiply the apparent number of subjects, resulting in falsely narrow standard errors
[30]. Generalized estimating equations (GEE) are a frequently-used approach to account for correlated data
[118]. In this way, events are used as the unit of analysis and standard errors are corrected for correlation of
covariate information within subjects [30].
Multilevel (also known as hierarchical, random effects, or mixed effects) regression modelling is a suitable
approach to handle for clustering of patients within health care providers and facilities when analyzing
clustered data [119-122]. Failure to use analytical methods that account for clustering can result in misleading
conclusions. In a recent study, results from data consisting of AMI patients nested within treating physicians
nested within hospitals with and without the use of multilevel modelling were contrasted [123]. The 95% CIs
for hospital effects were much wider when multilevel logistic models compared with conventional logistic
regression models, that ignored clustering, were used. Furthermore, substantially fewer statistically
significant associations between patient outcomes and hospital characteristics were found when multilevel
regression models compared with conventional regression models were used. When evaluating research using
HCU data, readers need to carefully assess whether the statistical methods accounted for any clustering that
may have been present in the data.
Approaches for such longitudinal data are described in detail in a number of textbooks [124,125].
A2.2.4.2. Using proxy measures
It is not uncommon for a researcher to be aware of an important confounding variable and to lack of data on
that variable. A measured proxy can sometimes stand in for an unmeasured confounder.
Researchers routinely adjust their analyses using proxy confounders. For example, the most common
confounder of treatment effects is patient’s age. Although age itself does not cause outcomes, old age is
associated with many disease conditions that may be incompletely recorded in the available data, but that are
associated with the outcome and may be a determinant of pharmacologic interventions, and therefore, age can
be considered an implicit proxy confounder [126].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 214
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
In a study of statin use and CV outcomes, Seeger et al found that certain healthcare utilization variables, such
as frequency of lipid tests ordered and physician visits, were strong predictors of statin initiation and appeared
also to be strong confounders [127]. In fact, although frequency of lipid testing does not directly affect CV
risk, it could be viewed as a proxy for concern about disease risk. Therefore, frequency of testing may be
associated with other risk modifying behaviours or to underlying risk.
In several HCU databases, the number of proxies describing the cross-sectional and longitudinal health status
can quickly rise to several hundreds, making it difficult to fit multivariate regression models for a limited
number of observed outcomes even in large studies [128]. The most popular methods to efficiently adjust
large number of proxies in database studies are of constructing exposure propensity score and comorbidity
score.
Exposure propensity score. In a cohort study, there are often substantial differences in the prevalence of
measured patient factors between drug exposure groups that may lead to confounding, if these factors are also
independent risk factors for the study outcome. Such factors need to be adjusted in further analyses. Instead
of considering each factor individually, it is possible to combine all patient characteristics into a single
exposure propensity score (EPS), which is the estimated probability of treatment, given all covariates (i.e. the
conditional probability of being treated given an individual’s covariates [129,130]) and is commonly
calculated using logistic regression models.
The more formal definition offered by Rosenbaum & Rubin [128] for the EPS for subject I (i = 1, . . . , N) is
the conditional probability of assignment to a treatment (Zi = 1) versus comparison (Zi = 0) given observed
covariates, xi:
E ( xi ) = pr ( Z i = 1 | X i = xi )
The underlying approach to propensity scoring uses observed covariates X to derive a “balancing score” β(X)
such that the conditional distribution of X given β (X) is the same for treated (Z = 1) and untreated (Z = 0)
[129,130]. After a EPS has been developed, there are three main applications: matching [130], stratification
[131], and regression adjustment [132,133]. When EPS is utilized in these standard applications, treatment
effects are unbiased where measured covariates are nearly equally balanced across comparison groups
[134,135]. This transparent balancing of confounders promotes confidence in interpreting the results
compared to other statistical modelling approaches [133]. For these reasons, PS methodology has become
increasingly popular to efficiently adjust large numbers of proxies in database studies.
Comorbidity score. Health status, as measured by disease history, has long been recognized as a major class
of potential confounder in most observational studies. Over the last three decades, a variety of methods have
been developed that might permit more uniform comorbidity adjustment across epidemiological studies. Six
distinct comorbidity scores useful for studies based on HCU data have been identified by a literature search
and tested for their predictive performance [136]. Comorbidity measuring instruments include the DartmouthManitoba method [137-139], the Chronic Disease Score [140], and its extended version [141], and the score
proposed by Deyo et al [142], D’Hoore et al [143,144], and Ghali et al. [145]. All these scores reduce the
number of covariates by summarizing the suffered diseases recorded as ICD-9 code in a specific HCU
database, in a single measure. Under the assumption that the score entirely captures information about clinical
profile of all included patients, which may be unrealistic in most practical settings, an analysis adjusting for
the score produces exposure effect estimates unbiased by among patients heterogeneity in clinical profile.
A2.2.4.3. Sensitivity analysis
In basic sensitivity analyses on residual confounding, we try to understand how the strength of an unmeasured
confounder and imbalance among drug exposure categories affects the observed or apparent RR. The
observed exposure-outcome relative risk (RR’) can be expressed as the ‘true’ relative risk times the “bias
factor” which is an expression of the imbalance of a binary confounding factor among exposed (PC1) and
unexposed (PC0) [146]:
RR' = RR ⋅
PC1 (RRCD − 1) + 1
PC 0 (RRCD − 1) + 1
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 215
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
where RRCD is the strength of the association between confounder and disease outcome.
Several approaches are available to obtain a quantitative estimate in the presence of assumed imbalance of the
confounder prevalence in the exposure or outcome groups [1]. Schneeweiss [30,83,146,147] describes two
families of approaches investigating the impact of residual confounders in the field of HCU databases: (1)
identifying the strength of residual confounding that would be necessary to explain an observed drug-outcome
association (rule-out approach); (2) external adjustment given additional information on single binary
confounders from survey data using algebraic solutions and a Monte Carlo sampling procedure (Monte Carlo
sensitivity analysis) or considering the joint distribution of multiple confounders from external sources of
information (propensity score calibration).
The rule out approach. The approach is directed to assess the extent of confounding necessary to fully
explain the observed findings, that is, the observed point estimate would move to the null. The hope is that
unmeasured possible confounders can be ruled out because they cannot possibly be strong enough
confounders to explain the observed association [147]. This approach was also called target-adjustment
sensitivity analysis [148].
The approach consists in finding all combinations of OREC (i.e. the confounder-exposure odds ratio measuring
the strength of the exposure-confounder association) and RRCD (i.e. the confounder-outcome relative risk
measuring the strength of the confounder-outcome association) necessary to move the observed point
estimate of RR’ to 1. It should be observed that OREC and RRCD are respectively the left and right sides of the
confounding triangle in Figure 3.3 [111].
Schneeweiss [146] showed that OREC is a function of the prevalence of the confounder among exposed (PC1)
and the marginal probabilities of exposure PE and confounder PC:
OREC =
PC1 (1 − PC − PE + PC1 )
(PC − PC1 )(PE − PC1 )
(1)
while, assuming no underlying true exposure-disease association (RR=1), Walker [149] showed that the
apparent relative risk (RR’) is a function of PC1, PE, PC, and the confounder-disease association RRCD:
RR' =
PC1 (RRCD − 1) + PE
1 − PE
(PC − PC1 )(RRCD − 1) − PE + 1 PE
(2)
Since the primary interest is to explore the relationship between OREC and RRCD for a given RR’, we need to
solve Equation (2) for PC1 and substitute the derived term in Equation (1). In this way, we may calculate the
couple of OREC - RRCD values annulling the true exposure-disease association.
An example has been recently provided by
Huybrechts et al. [150]. The authors compared
180-day mortality among patients using
conventional
and
atypical
antipsychotic
medication (APM). Compared with patients
receiving atypical medications, those on
conventional APM had a 1.34-fold higher
mortality (RR’ adjusted for EPS). Figure 4
displays the strengths of the associations
between confounder - mortality (RRCD) (x axis)
and exposure confounder (OREC) (y axis) that
would be required to fully explain the observed
increased mortality associated with conventional
antipsychotic medication use if in truth no such
association existed. For an unmeasured
confounder present in 25% of the population (PC
= 0.25) increasing the risk of mortality of 3-fold (RRCD = 3), patients exposed to the hypothetical confounder
should have a 3-fold increased odds of exposure to conventional APM (OREC = 3) to fully explain the
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 216
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
observed association. For confounders present in 5% or 1% of the population, RRCD and OREC greater than
4.5 and 9, respectively, would be needed. Since it is unlikely that a so strong single confounder has been
missed from available data, residual confounding would have a relatively small impact. Consequently,
propensity score methods based on claims alone likely provided likely unbiased estimates. However, the
authors underlined that it is conceivable that several weaker confounders may have acted together and explain
the apparent effect of conventional APM.
Rule-out approach is very useful when little or nothing is known of possible confounders of the investigated
association and of their effect. However, in the case of reliable additional data sources can be identified,
external adjustment of the drug-outcome association is always advisable [151].
External adjustment methods. If additional information is available, for example, a detailed survey in a
sample of the main database study, alternative approaches to sensitivity analyses can be used to correct for
confounders unmeasured in the main study [146]. If internal validation studies are not feasible or are too
costly, external data sources can be used under certain assumptions. For example, structured electronic MR
databases fed by general practitioners (GPs) operating in the same area and covering a sample of the same
population of the principal database, may be used for measuring a wide variety of characteristics not captured
in HCU data, such as drug indications, lifestyle habits, body mass index, blood pressure measures and
laboratory findings, among others. Thus medical records can be used for external adjustment of unmeasured
confounders in a variety of drug studies using HCU data.
An example has been recently provided by Corrao et al. [152,153]. The authors observed that compared with
antihypertensive patients starting BP lowering therapy on fixed-dose combination, those on extemporaneous
combination had CV risk increased of 15%. From a clinical point of view, this is a rather puzzling finding
since the reason why two antihypertensive drugs had different effects according whether they are given in two
distinct pills or in a unique pill is unclear. Several uncontrolled factors may influence the decision of the
physician of starting therapy. For example, one can imagine that patients with more severe hypertension or
worse clinical profile need a more aggressive therapy to reach quickly BP control, and that this aggressive
therapy is often obtained by dispensing extemporaneous combination of two or more agents, rather than fixed
combinations. Because severity of hypertension and clinical profile are independent predictors of the study
outcome, failing to their control can lead to confounding bias. Quantitative assessment of such a bias can
provide more realistic estimates of the relationship between exposure and outcome.
The authors applied the following four step procedure. One, they assessed the exposure-confounder
association (that is, do patient’s clinical characteristics affect the choice of prescribing a given
antihypertensive drug regimen?) drawing out the corresponding data from an Italian network of general
practitioners, the so called Health Search/Cegedim Strategic Database.
Two, some assumptions were made for the confounder-outcome association (that is, do patient’s clinical
characteristics affect the CV risk?).
Three, these two types of external information were used to correct estimates generated from the main study
according with the approach from Steenland & Greenland [154] who proposed to estimate the bias factor
measuring the extension of the residual bias that would result from a failure control for a generic confounder:
J
∑p
Bias =
j =1
j1
(3)
J
∑p
j =1
⋅ RR j
j0
⋅ RR j
where j indexes a generic confounder’s category with j = 0, 1 for mild / moderate, severe hypertension; or j =
0, 1, 2 for increasing number of chronic comorbidities. In equation (3) the risk ratio for the confounderoutcome association (RRj) is weighted for the proportion of patients belonging to the same confounder’s
category among those who started with fixed-dose combination (pj0) or with another therapeutic regimen (pj1).
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 217
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
The effect of starting with a given regimen on the CV risk, estimated from the main study, was separately
adjusted for severity of hypertension and chronic comorbidities simply by dividing the original estimates for
the bias factor (equation 3).
Four, the Monte-Carlo Sensitivity Analysis (MCSA), an expanded version of ordinary sensitivity analysis,
was used with the aim of taking into account for random uncertainty of the estimates obtained with external
adjustment through a Monte-Carlo sampling procedure [154].
Figure 5 displays the CV odds ratios associated with the initial treatment regimen observed from the HCU
data (white squares) and after MCSA adjustment for severity of hypertension and CDS (black circles).
Patients
on
extemporaneous
1.15 (1.03 to 1.29)
Observed
combination were at higher CV risk
than
those
on
fixed-dose
combination.
However,
evidence
that
1.14
(1.02
to
1.28)
Adjustment for
Scenario 1
severity of
patients
on
extemporaneous
1.13 (1.01 to 1.26)
hypertension
Scenario 2
combination are at higher CV risk
than those on fixed-dose combination
1.11 (0.99 to 1.25)
Scenario 3
was annulled after adjustment for
CDS, even when a relatively weak
confounder - outcome association
1.12
(1.01
to
1.25)
Scenario 1
Adjustment for
chronic disease
was imposed (scenario 2). This
1.09 (0.97 to 1.21)
Scenario 2
score
happens because the large difference
in clinical profile of patients starting
Scenario 3
1.00 (0.89 to 1.12)
BP-lowering therapy with fixed-dose
or extemporaneous combination. As
Odds ratio
a matter of fact, according with data
from Health Search used for external
adjustment, patients on extemporaneous combination had higher prevalence of severe hypertension and worse
clinical profile than those on fixed-dose combination. Medical record data can be used to assess confounding
bias unmeasured from HCU database by means of MCSA. A SAS code useful for any application of this
technique has been supplied by the authors [153].
A main limitation of MCSA is that it is not helpful if several confounders are unmeasured and the joint effect
of such confounders is unknown. External adjustment methods, however, were recently extended to a
multivariate adjustment for unmeasured confounders that use a new technique of propensity score calibration
(PSC) [155]. In a validation study for each subject, the full database record is available along with detailed
survey information. The goal is to compute within the validation population an error-prone exposure PS using
only database information, as well as an improved exposure propensity score that also includes survey
information for each subject. The error component in the database PS in the validation study is then
quantified and can be used to correct the PS in the database main study, using established regression
calibration techniques [44]. PSC implicitly takes into account the joint effect of unmeasured confounders that
are measured only in the validation study, as well as the relation between measured and unmeasured
confounders. PSC can, therefore, elegantly overcome major limitations of the algebraic approach to external
adjustment described above, although it may not perform well in situations were the surrogacy assumption of
regression calibration is violated [44,156].
A2.2.4.4. Instrumental variable estimation
To overcome the inability to control for residual confounding by unobserved factors, an analytic approach,
known in economics as instrumental variable (IV) estimation [157], can provide unbiased estimates of causal
effects in non randomized studies [158] by mimicking random assignment of patients into groups of different
likelihood for treatment [159].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 218
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
An IV is a factor that is related to treatment, but unrelated to observed and unobserved confounders, and also
unrelated to the outcome (Figure 3.4), other than through its relationship to treatment, both key assumptions
for valid IV estimation. In the analysis, the unconfounded instrument substitutes the actual treatment status
that may be confounded [83]. The instrument effect on the study outcome will be estimated and then rescaled
by its correlation with the actual exposure [83]. The more strongly an IV is related to the actual treatment, the
less any residual confounding will be weighted and precision of the IV estimation will improve [160].
IV estimation had not been used for the evaluation of medicine until Brookhart et al [161] introduced
physician prescribing preference as a promising instrument for comparative effectiveness research. The basic
idea is that there is a distribution of physician’s preference for one drug over another that is largely
independent of patient characteristics. One way to define a physician-prescribing preference instrument is to
categorize physicians into strong preferers of drug A if they prescribed it in 90% or more of their patients,
whereas nonpreferring doctors prescribe it in only 10% or less of cases. A variety of implementations of
physician-prescribing preference is possible, including the choice of drug that a physician used for the most
recent patient [161,162]. In a study on the comparative effectiveness of selective COX-2 inhibitors relative to
non-selective NSAIDs, the last new NSAID prescription written by a physician was used to determine the IV
status of the next patient. If the last patient received celecoxib, then for the next patient the physician is
classified as a “celecoxib prescriber” [162]. This approach takes into account that NSAID-prescribing
preference may change within the study period. The analysis is conducted with two-stage regression models
and adjustment of standard errors for the fact that patients cluster in physicians’ clinics [163].
A2.2.5. Beyond confounding
A2.2.5.1. Yet on the confounding definition
The confounding hypothesis suggests that a third variable explains the relationship between exposure and
outcome [164-168]. At least one definition of a confounder effect, however, specifically requires that the third
variable not be an “intermediate” variable, as mediators are termed in epidemiological literature [169].
Consider a hypothetical example in which we are
interested in assessing if periodontal disease (exposure)
causes CVD (outcome) [170]. We also have measurements
of C-reactive protein (CRP, i.e. the external third variable),
a marker of chronic inflammation that is associated with
periodontal disease and CVD. In the univariate analysis,
Periodontal
Cardiovascular
we find that there are statistically significant associations
disease
disease
between periodontal disease and CVD, periodontal disease
and CRP, and CRP and CVD. In the multivariate analysis,
when we include CRP and periodontal disease in the same
model to predict CVD risk, we observe a null association
2
C-reactive
between
periodontal disease and CVD, and a positive
protein
association between CRP and CVD. According to Figure
6.1, raised CRP is a marker of a hyper-inflammatory trait,
which causes both increased bone destruction in
periodontal disease and atherosclerotic changes in CVD,
Periodontal
Cardiovascular
disease
disease
but there is no true association between periodontal disease
and CVD. If we believe this is true then we would
conclude that periodontal disease does not cause CVD. The crude association we observed was through the
backdoor path via CRP, which was closed when we adjusted for it in the multivariate analysis. According to
1
C-reactive
protein
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 219
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Figure 6.2, periodontal disease causes chronic low-grade infection and raises CRP levels and causes increased
CVD risk. If we believe that this is true, then we would conclude that periodontal disease causes CVD that is
mediated via chronic inflammation (CRP). We can thus draw completely opposite conclusions with the same
statistical data depending upon which causal pathway we believe in.
This example clarifies that it is not sufficient that a variable is associated with both exposure and outcome to
be considered confounder. It is also necessary that the third external variable is not an intermediate factor
between exposure and outcome [171]. If this is the case, adjusting for the effect of the external variable (i.e.
the intermediate variable or mediator) could substantially bias the estimated association between exposure
and outcome.
In this paragraph, based on an intuitive approach, the inadequacy of conventional criteria for appropriately
identifying confounders, especially in the case where overadjustment from mediation and collider
stratification bias occur [101,172-175], will be faced.
A2.2.5.2. Intermediate variables and overadjustment
One reason why an investigator may begin to explore third variable effects is to elucidate the causal process
by which the exposure affects the outcome, a meditational hypothesis [164,176]. In examining a mediational
hypothesis, the relationship between exposure and outcome is decomposed into two causal paths [177]. One
of these paths links the exposure to the outcome directly (the direct effect), and the other links the
independent variable to the dependent variable through a mediator (the indirect effect). An indirect or
mediated effect implies that exposure causes the mediator (intermediate variable), which, in turn causes the
outcome [178,179].
An example has been recently provided by
Roumie et al. [180]. To determinate if incident
use of oral antidiabetic drugs (OADs) is
associated with 12-month systolic blood
pressure (BP) and if this is mediated through
body mass index (BMI) the authors included a
cohort of veterans with hypertension who
initiated metformin (n = 2057) or sulfonylurea
(n = 1494) between 1 January 2000 and 31
December 2007. Figure 7 shows that
sulfonylurea users had a 1.33 mmHg higher
12-month systolic BP than metformin users.
BP- mean differences in 12-months (mmHg)
However, in a model adjusting for BMI
change, the difference in 12-month systolic BP between sulfonylurea and metformin users became
insignificant (p = 0.72), while one BMI unit change was associated with an increase in 12-month systolic BP
of 1.07mmHg (p<0.0001). These findings (i) strengthen the hypothesis that use of sulfonylurea increases
systolic BP; (ii) suggest that the effect of OAD on BP change is likely “mediated” by the favourable effect of
metformin on reducing BMI; (iii) demonstrate that a biased estimate of the effect of OAD on BP reduction is
employed when adjusting for the mediator (intermediate variable). In other word, the exposure-outcome
association is obscured by the so called effect of “overadjustment” [181].
Rothman & Greenland [80] discussed overadjustment in the context of intermediate variables. They clarified
that intermediate variables, if controlled in an analysis, would usually bias results towards the null. Thus
rather than adjusting for, the assessment of mediation allows to move beyond the simple identification of
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 220
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
exposure–disease associations, toward an explanation of these relationships. The reasons for assessing
mediation in epidemiology are compelling, and can be directly linked to extant mediational effects. Mediation
analysis is very useful for opening the “black box” between exposure and disease in epidemiologic studies
[182].
Investigators are sometimes interested in separating total causal effect into direct and indirect effects, i.e. that
explaining the exposure-outcome directly, and through the mediator, respectively. The goal of mediation
analysis is to assess direct and indirect exposure on the outcome effects. The classical method of mediation
analysis for two stages [183] involves the fitting of a succession of linear regression models. Structural
equations model (SEM) based methods have also been proposed for mediation analysis [184,185]. In the
linear case, it is straightforward to estimate mediation effects in the context of multiple stages via the product
of coefficients approach [186]. Some recent research has focused on mediation for binary or mixed types of
variables [187-190]. Finally, a method applicable to multiple stages of mediation and mixed variable types
using generalized linear models has been recently proposed [191].
A2.2.5.3. Collider variables
Colliders are the result of two independent causes having a common effect [113]. When we include a
common effect of two independent causes in our model, the previously independent causes become
associated thus opening a backdoor path between the treatment and outcome. This phenomenon can be
explained intuitively if we think of two causes (sprinklers being on or it is raining) of a lawn being wet. If we
know the lawn is wet and we know the value of one of the other variables (it is not raining) then we can
predict the value of the other variable (the sprinkler must be on). Therefore, conditioning on a common effect
induces an association between two previously independent causes, i.e., sprinklers being on and rain [113].
Consider a hypothetical study to compare rates of acute liver failure between new users of CCB and diuretics
using HCU data. As illustrated in Figure 8, CCB are mainly prescribed to older patients, while younger
hypertensive manly receives diuretics. On the other hand, older patients more likely receive treatment for
erectile dysfunction and also have a long history of alcohol abuse. Finally, among the considered factors, only
alcohol abuse truly causes acute liver failure.
Acute liver
Nevertheless, antihypertensive treatment and liver
Alcohol abuse
disease (D)
disease should result associated when adjusting for
treatment of erectile dysfunction.
Treatment for
erectile
dysfunction
Age
Antihypertensive
therapy (E)
The introduced bias is known as colliderstratification bias [192], or bias due to
conditioning on a collider [193]. The term
‘conditioning’ refers to restriction, stratification or
regression adjustment that is the techniques above
described for controlling measured confounders
[194].
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 221
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
References
[1] Cox E, Martin BC, Van Staa T, et al. Good research practices for comparative effectiveness research:
approaches to mitigate bias and confounding in the design of nonrandomized studies of treatment effects
using secondary data sources: The International Society for Pharmacoeconomics and Outcomes Research
Good Research. Practices for retrospective database analysis task force report—Part II. Value Health
2009;12:1053-61
[2] O'Malley KJ, Cook KF, Price MD, et al. Measuring Diagnoses: ICD Code Accuracy. Health Serv Res
2005;40:1620-39
[3] Van Walraven C, Bennett C, Forster AJ. Administrative database research infrequently used validated
diagnostic or procedural codes. J Clin Epidemiol 2011;64:1054-9
[4] Chyou PH. Patterns of bias due to differential misclassification by case-control status in a case-control
study. Eur J Epidemiol 2007;22:7-17
[5] Copeland KT, Checkoway H, McMichael AJ, Holbrook RH. Bias due to misclassification in the
estimation of relative risk. Am J Epidemiol 1977;105:488-95
[6] Pekkanen J, Sunyer J, Chinn S. Nondifferential disease misclassification may bias incidence risk ratios
away from the null. J Clin Epidemiol 2006;59:281-9
[7] Slee VN, Slee D, Schmidt HJ. The tyranny of the diagnosis code. N C Med J 2005;66:331-7
[8] Hsia DC, Krushat WM, Fagan AB, et al. Accuracy of diagnostic coding for Medicare patients under the
prospective-payment system. N Engl J Med 1988;318:352-5
[9] Cimino JJ. Formal descriptions and adaptive mechanisms for changes in controlled medical vocabularies.
Methods Inf Med 1996;35:202-10
[10] Vardy DA, Gill RP, Israeli A. Coding medical information: classification versus nomenclature and
implications to the Israeli medical system. J Med Syst 1998;22:203-10
[11] Feinstein AR. ICD, POR, and DRG. Unsolved scientific problems in the nosology of clinical medicine.
Arch Intern Med 1988;148:2269-74
[12] Cimino JJ. An approach to coping with the annual changes in ICD9-CM. Methods Inf Med 1996;35:220
[13] Hogan WR, Slee VN. Measuring the Information Gain of Diagnosis vs. Diagnosis Category Coding.
AMIA Annu Symp Proc 2010;2010:306-10
[14] Kelsey JL, Whittemore AS, Evans AS, Thompson WD. Methods in observational epidemiology. 2nd
edition. New York: Oxford University Press;1996
[15] Wilchesky M, Tamblyn RM, Huang A. Validation of diagnostic codes within medical services claims. J
Clin Epidemiol 2004;57:131-41
[16] Romano PS, Mark DH. Bias in the coding of hospital discharge data and its implications for quality
assessment. Med Care 1994;32:81–90
[17] Antoniou T, Zagorski B, Loutfy MR, et al. (2011) Validation of case-finding algorithms derived from
administrative data for identifying adults living with human immunodeficiency virus infection. PLoS One
2011;6:e21748
[18] Hux JE, Ivis F, Flintoft V, et al. Diabetes in Ontario: determination of prevalence and incidence using a
validated administrative data algorithm. Diabetes Care 2002;25:512–6
[19] Corrao G, Botteri E, Bagnardi V, et al. Generating signals of drug-adverse effects from prescription
databases and application to the risk of arrhythmia associated with antibacterials. Pharmacoepidemiol Drug
Saf 2005;14:31-40
[20] Navarese EP, Buffon A, Andreotti F, et al. Meta-Analysis of Impact of Different Types and Doses of
Statins on New-Onset Diabetes Mellitus. Am J Cardiol 2013; doi:10.1016/j.amjcard.2012.12.037
[21] Delgado-Rodríguez M, Llorca J. Bias. J Epidemiol Community Health 2004;58;635-41
[22] Stergachis AS. Record linkage studies for postmarketing drug surveillance: data quality and validity
considerations. Drug Intell Clin Pharm 1988;22:157-61
[23] Levy AR, O’Brien BJ, Sellors C, Grootendorst P,Willison D. Coding accuracy of administrative drug
claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 2003;10:67-71
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 222
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[24] McKenzie DA, Semradek J, McFarland BH, Mullooly JP, McCamant LE. The validity of Medicaid
pharmacy claims for estimating drug use among elderly nursing home residents: the Oregon experience. J
Clin Epidemiol 2000;53:1248-57
[25] Strom BL. Overview of automated databases in pharmacoepidemiology. In:Strom BL, editor. In:
Pharmacoepidemiology (4th edn). Strom BL (ed). Wiley, New York, 2005;219-22
[26] West S, Savitz DA, Koch G, Strom BL, Guess HA, Hartzema A. Recall accuracy for prescription
medications: self report compared with database information. Am J Epidemiol 1995;142:1103–12
[27] West S, Strom BL, Freundlich B, Normand E, Koch G, Savitz DA. Completeness of prescription
recording in outpatients medical records from a health maintenance organization. J Clin Epidemiol
1994;47:165–71
[28] Peterson AM, Nau DP, Cramer JA, et al. A checklist for medication compliance and persistence studies
using retrospective databases. Value in Health 2007;10:3-12
[29] WHO Collaborating Centre for Drug Statistics Methodology. ATC index with DDD. Oslo, Norway:
WHO; 2003
[30] Schneeweiss S, Avorn J. A review of uses of health care utilization databases for epidemiologic research
on therapeutics. J Clin Epidemiol 2005;58:323–37
[31] de Abajo FJ, Garcia Rodriguez LA. Risk of upper gastrointestinal bleeding and perforation associated
with low-dose aspirin as plain and enteric-coated formulations. BMC Clin Pharmacol
2001;1:doi:10.1186/472-6904-1-1
[32] Ilkhanoff L, Lewis JD, Hennessy S, et al. Potential limitations of electronic database studies of
prescription non-aspirin nonsteroidal anti-inflammatory drugs (NANSAIDs) and risk of myocardial infarction
(MI). Pharmacoepidemiol Drug Saf 2005;14: 513-22
[33] Ulcickas Yood M, Watkins E, Wells KE, et al. Using prescription claims data for drugs available overthe-counter (OTC). Pharmacoepidemiol Drug Saf 2000;9:S37
[34] Suissa S. Immeasurable time bias in observational studies of drug effects on mortality. Am J Epidemiol
2008;168:329-35
[35] Greenland S. The effect of misclassification in matched-pair case-control studies. Am J Epidemiol
1982;116:402–6
[36] Marshall RJ. Validation study methods for estimating exposure proportions and odds ratios with
misclassified data. J Clin Epidemiol 1990;43:941–7
[37] Brenner H, Gefeller O. Use of positive predictive value to correct for disease misclassification in
epidemiologic studies. Am J Epidemiol 1993;138:1007–15
[38] Motheral B, Brooks J, Clark MA, et al. A Checklist for Retrospective Database Studies—Report of the
ISPOR Task Force on Retrospective Databases. Value in Health 2003;6:90-7
[39] McMahon AD. Observation and experiment with the efficacy of drugs: a warning example from a cohort
of nonsteroidal anti-inflammatory and ulcer-healing drug users. Am J Epidemiol 2001;154:557–62
[40] Sjolander A, Humphreys K, Palmgren J. On informative detection bias in screening studies. Stat Med
2008;27:2635-50
[41] Ulcickas Yood M, Campbell UB, Rothman KJ, et al. Using prescription claims data for drugs available
over-the-counter (OTC). Pharmacoepidemiol Drug Saf 2007;16:961-8
[42] Thürigen D, Spiegelman D, Blettner M, et al. Measurement error correction using validation data: a
review of methods and their applicability in case-control studies. Statistical Methods in Medical Research
2000;9:447–74
[43] Rosner B, Willett WC, Spiegelman D. Correction of logistic regression relative risk estimates and
confidence intervals for systematic within-person measurement error. Stat Med 1989;8:1051-69
[44] Rosner B, Spiegelman D, Willett WC. Correction of logistic regression relative risk estimates and
confidence intervals for random within person measurement error. Am I Epidemiol 1992:136:1400-13
[45] Willett WC, Hunter DI, Stampfer MI, et al. Dietary fat and fiber in relation to risk of breast cancer: an
eight year follow-up. JAMA l992:268:2037-44
[46] Freedman LS, Schatzkin A, Midthune D, et al. Dealing with dietary measurement error in nutritional
cohort studies. J Natl Cancer Inst 2011;103:1086-92
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 223
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[47] Beasley JM, LaCroix AZ, Neuhouser ML, et al. Protein intake and incident frailty in the Women's
Health Initiative observational study. J Am Geriatr Soc 2010;58:1063-71
[48] Van Roosbroeck S, Li R, Hoek G, et al. Traffic-related outdoor air pollution and respiratory symptoms in
children: the impact of adjustment for exposure measurement error. Epidemiology 2008;19:409-16
[49] Strand M, Vedal S, Rodes C, et al. Estimating effects of ambient PM(2.5) exposure on health using
PM(2.5) component measurements and regression calibration. J Expo Sci Environ Epidemiol 2006;16:30-8
[50] Grayson DA. Confounding confounding. Am J Epidemiol 1987;126:546–53
[51] Weinberg CR. Towards a clearer definition of confounding. Am J Epidemiol 1993;137:1–8
[52] Brookhart MA, Sturmer T, Glynn RJ, et al. Confounding control in healthcare database research
challenges and potential approaches. Med Care 2010;48(suppl 1):S5-8
[53] Walker AM. Confounding by indication. Epidemiology 1996;7:335-6
[54] Bosco JL, Silliman RA, Thwin SS, et al. A most stubborn bias: no adjustment method fully resolves
confounding by indication in observational studies. J Clin Epidemiol 2010;63:64-74
[55] Bruce M. Psaty BM, Siscovick DS. Minimizing Bias Due to Confounding by Indication in Comparative
Effectiveness Research The Importance of Restriction. JAMA 2010;304:897-8
[56] Sturmer T, Schneeweiss S, Avorn J, et al. Adjusting effect estimates for unmeasured confounding with
validation data using propensity score calibration. Am J Epidemiol 2005;162:279-89
[57] Jackson LA, Jackson ML, Nelson JC, et al. Evidence of bias in estimates of influenza vaccine
effectiveness in seniors. Int J Epidemiol 2006;35:337-44
[58] Jackson LA, Nelson JC, Benson P, et al. Functional status is a confounder of the association of influenza
vaccine and risk of all cause mortality in seniors. Int J Epidemiol 2006;35:345-52
[59] Lett HS, Blumenthal JA, Babyak MA, et al. Depression as a risk factor for coronary artery disease:
evidence, mechanisms, and treatment. Pychosom Med 2004;66:305-15
[60] DiMatteo MR, Lepper HS, Croghan TW. Depression is a risk factor for noncompliance with medical
treatment. Meta-analysis of the effects of anxiety and depression on patient adherence. Arch Intern Med
2000;160:2101-7
[61] White HD. Adherence and outcomes: it’s more than taking the pills. Lancet 2005;366:1989-91
[62] Brookhart MA, Patrick AR, Dormuth C, et al. Adherence to lipid lowering therapy and the use of
preventive health services: an investigation of the healthy user effect. Am J Epidemiol 2007;166:348-54
[63] Simpson SH, Eurich DT, Majumdar SR, et al. A meta-analysis of the association between adherence to
drug therapy and mortality. BMJ 2006;333:15
[64] Lewis JD, Brensinger C. Agreement between GPRD smoking data: a survey of general practitioners and
a population-based survey. Pharmacoepidemiol Drug Saf 2004;13:437-41
[65] Dormuth CR, Patrick AR, Shrank WH, et al. Statin adherence and risk of accidents. A cautionary tale.
Circulation 2009;119:2051-7
[66] Poses RM, Smith WR, McClish DK, et al. Controlling for confounding by indication for treatment: are
administrative data equivalent to clinical data? Med Care 1995;33(Suppl):AS36-46
[67] Chassin MR, Kosecoff J, Park RE, et al. Does inappropriate use explain geographic variation in the use
of health care services: a study of three procedures. JAMA 1987;258:2533-7
[68] Maynard C, Fisher L, Alderman EL, et al. Institutional differences in therapeutic decision making in the
coronary artery surgery study (CASS). Med Decis Making 1986;6:127-35
[69] Perrin JM, Homer CJ, Berwick DM, et al. Variations in the rates of hospitalization of children in three
urban communities. N Engl J Med 1989;320:1183-7
[70] Wennberg JE, Mulley AG, Hanley D, et al. An assessment of prostatectomy for benign urinary tract
obstruction: geographic variations and the evaluations of medical care outcomes. JAMA 1988;259:3027-30
[71] Goyert GL, Bottoms SF, Treadwell MC, et al. The physician factor in cesarean birth rates. N Engl J Med
1989;320:706-9
[72] Wennberg JE. Small area analysis and the medical care outcome problem. In: Sechrest L, Perrin E,
Bunker J, eds. Conference proceedings: research methodology- strengthening causal interpretations of nonexperimental data. Washington, DC: U.S. Department of Health and Human Services, Public Health Service,
Agency for Health Care Policy and Research, May 1990
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 224
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[73] Wennberg JE. Population illness rates do not ex- plain population hospitalization rates. Med Care
1987;25:354-9
[74] Glynn RJ, Knight EL, Levin R, Avorn J. Paradoxical relations of drug treatment with mortality in older
persons. Epidemiology 2001;12:682-9
[75] Schneeweiss S, Wang P. Association between SSRI use and hip fractures and the effect of residual
confounding bias in claims database studies. J Clin Psychopharm 2004;13:695–702
[76] Roos LL, Fisher ES, Sharp SM, et al. Postsurgical mortality in Manitoba and New England. JAMA
1990;263:2453-8
[77] Byar DP. Problems with using observational data-bases to compare treatments. Stat Med 1991;10:663-6
[78] Dambrosia JM, Ellenberg JH. Statistical considerations for a medical data base. Biometrics 1980;36:32332
[79] Feinstein AR. Para-analysis, faute de mieux, and the perils of riding on a data barge. J Clin Epidemiol
1989;42:929-35
[80] Rothman KJ, Greenland S. Modern epidemiology. 2nd edition. Philadelphia: Lippincott Williams &
Wilkins; 1998
[81] Perrio M, Waller PC, Shakir SAW. An analysis of the exclusion criteria used in observational
pharmacoepidemiological studies. Pharmacoepidemiol Drug Saf 2006;16:329–36
[82] Schneeweiss S, Patrick AR, Sturmer T, et al. Increasing levels of restriction in pharmacoepidemiologic
database studies of elderly and comparison with randomized trial results. Med Care 2007;45(Suppl):S131–42
[83] Schneeweiss S. Developments in post-marketing comparative effectiveness research. Clin Pharmacol
Ther 2007;82:143–56
[84] Schneeweiss S, Gagne JJ, Glynn RJ, et al. Assessing the comparative effectiveness of newly marketed
medications: methodological challenges and implications for drug development. Clin Pharmacol Ther
2011;90:777-90
[85] Petri H, Urquhart J. Channeling bias in the interpretation of drug effects. Stat Med 1991;10:577-81
[86] Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol
2003;158:915–20
[87] Glynn RJ, Schneeweiss S, Wang P, Levin R, Avorn J. Selective prescribing can lead to over-estimation
of the benefits of lipid lowering drugs. J Clin Epidemiol 2006;59:819-28
[88] Danaei G, Tavakkoli M, Hernan MA. Bias in observational studies of prevalent users: lessons for
comparative effectiveness research from a meta-analysis of statins. Am J Epidemiol 2012;175:250–62
[89] Corrao G, Zambon A, Parodi A, et al. Incidence of cardiovascular events in Italian patients with early
discontinuations of antihypertensive, lipid-lowering, and antidiabetic treatments. Am J Hypertens
2012;25:549-55
[90] Corrao G, Zambon A, Parodi A, et al. Discontinuation of and changes in drug therapy for hypertension
among newly-treated patients: a population-based study in Italy. J Hypertens 2008;26:819-24
[91] Corrao G, Conti V, Merlino L, et al. Results of a retrospective database analysis of adherence to statin
therapy and risk of nonfatal ischemic heart disease in daily clinical practice in Italy. Clin Ther 2010;32:30010
[92] Corrao G, Romio SA, Zambon A, et al. Multiple outcomes associated with the use of metformin and
sulphonylureas in type 2 diabetes: a population-based cohort study in Italy. Eur J Clin Pharmacol
2011;67:289-99
[93] Ludvigsson JF, Montgomery SM, Olen O, et al. Coeliac disease and risk of renal disease - a general
population cohort study. Nephrol Dial Transplant 2006;21:1809-15
[94] Greenland S, Morgenstern H. Matching and efficiency in cohort studies. Am J Epidemiol 1990;131:1519
[95] Klein-Geltink JE, Rochon PA, Dyer S, et al. Readers should systematically assess methods used to
identify, measure and analyze confounding in observational cohort studies. J Clin Epidemiol 2007;60:766-72
[96] Miettinen OS. Matching and design efficiency in retrospective studies. Am J Epidemiol 1970;91:111-8
[97] Kalish LA. Matching on a non-risk factor in the design of case-control studies does not always result in
an efficiency loss. Am J Epidemiol 1986;123:551-4
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 225
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[98] de Graaf MA, Jager KJ, Zoccali C, et. Matching, an appealing method to avoid confounding? Nephron
Clin Pract 2011;118:c315-8
[99] Smith PG, Day NE. The design of case-control studies: the influence of confounding and interaction
effects. Int J Epidemiol 1984;13:356-65
[100] Gissler M, Hemminki E, The danger of overmatching in studies of the perinatal mortality and
birthweight of infants born after assisted conception. Eur J Obstet Gyn Repr Biol 1996;69:73-5
[101] Nordmann S, Biard L, Ravaud P, et al. Case-only designs in pharmacoepidemiology: a systematic
review. Plos One 2012;7:e49444
[102] Maclure M, Fireman B, Nelson JC, et al. W should case-only designs be used for safety monitoring of
medical products? Pharmacoepidemiol Drug Saf 2012;21:50-61
[103] Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute
events. Am J Epidemiol 1991;133:144–53
[104] Farrington CP. Relative incidence estimation from case series for vaccine safety evaluation. Biometrics
1995;51:228–35
[105] Suissa S. The case-time-control design. Epidemiology 1995;6: 248–53
[106] Farrington CP. Estimation of vaccine effectiveness using the screening method. Int J Epidemiol
1993;22:742–6
[107] Petri H, de Vet HC, Naus J, et al. Prescription sequence analysis: a new and fast method for assessing
certain adverse reactions of prescription drugs in large populations. Stat Med 1988;7:1171–5
[108] Pariente A, Fourrier-Réglat A, Ducruet T, et al. Antipsychotic use and myocardial infarction in older
patients with treated dementia. Arch Int Med 2012;172:648-53
[109] Whitaker HJ, Farrington CP, Spiessens B, et al. Tutorial in biostatistics: the self-controlled case series
method. Stat Med 2006;25:1768–97
[110] Grosso A, Douglas I, Hingorani A, et al. Post-marketing assessment of the safety of strontium ranelate;
a novel case-only approach to the early detection of adverse drug reactions. Br J Clin Pharmacol
2008;66:689-94
[111] Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology
1999;10:37-48
[112] Rothwell PM. Subgroup analysis in randomized controlled trials: importance, indications, and
interpretation. Lancet 2005;365:176-86
[113] Agency for Healthcare Research and Quality (AHRQ). Developing a protocol for observational
comparative
effectiveness
research
(OCER):
a
user’s
guide.
Available
at:
http://www.effectivehealthcare.ahrq.gov/index.cfm/research-DRAFT-COPY_AllChapters.pdf
[114] Klein JP, Moeschberger ML. Survival Analysis: Techniques for Censored and Truncated Data. New
York: Springer-Verlag;1997
[115] Hosmer DW, Lemeshow S, May S. Applied Survival Analysis. 2nd edition. New Jersey: Wiley;2008
[116] Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J
Epidemiol 2004;159:702-6
[117] Lumley T, Kronmal R, Ma Shuangge. Relative risk regression in medical research: models, contrasts,
estimators, and algorithms. UW Biostatistics Working Paper Series. University of Washington. Paper
293;2006
[118] Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika
1986;73:13–22
[119] Snijders PJ, Boskers R. Statistical treatment of clustered data. Multilevel analysis: an introduction to
basic and advanced multilevel modeling. Thousand Oaks, CA: Sage Publications, Inc.;1999:13-37
[120] Austin PC, Goel V, van Walraven C. An introduction to multilevel regression models. Can J Public
Health 2001;92:150-4
[121] Raudenbush SW, Bryk AS. Hierarchical linear models: applications and data analysis methods.
Thousand Oaks, CA: Sage Publications, Inc.;2002
[122] Goldstein H. Multilevel statistical models. 2nd ed. London, UK: Edward Arnold; 1995
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 226
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[123] Austin PC, Tu JV, Alter DA. Comparing hierarchical modeling with traditional logistic regression
analysis among patients hospitalized with acute myocardial infarction: should we be analyzing cardiovascular
outcomes data differently? Am Heart J 2003;145:27-35
[124] Diggle PJ, Heagerty P, Liang K-Y, Zeger SL. Analysis of Longitudinal Data. 2nd edition. New York:
Oxford University Press;2002
[125] Fitzmaurice GM, Laird NM, Ware JH. Applied Longitudinal Analysis. New Jersey:Wiley;2004
[126] Bradbury BD, Gilbertson DT, Brookhart MA, et al. Confounding and control of confounding in
nonexperimental studies of medications in patients with CKD. Adv Chron Kidney Dis 2012;19:19-26
[127] Seeger JD, Kurth T, Walker AM. Use of propensity score technique to account for exposure-related
covariates: an example and lesson. Med Care 2007;45:S143–8
[128] Takahashi Y, Nishida Y, Asai S. Utilization of health care databases for pharmacoepidemiology. Eur J
Clin Pharmacol 2012;68:123-9
[129] Rosenbaum P, Rubin DB. The central role of the propensity score in observational studies for causal
effects. Biometrika 1983;70:41–55
[130] D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a
non-randomized control group. Stat Med 1998;17:2265-81
[131] Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med
1997;127:757-63
[132] Weitzen S, Lapane KL, Toledano AY, et al. Principles for modeling propensity scores in medical
research: a systematic literature review. Pharmacoepidemiol Drug Saf 2004;13:841-53
[133] Shah BR, Laupacis A, Hux JE, et al. Propensity score methods gave similar results to traditional
regression modeling in observational studies: a systematic review. J Clin Epidemiol 2005;58:550-9
[134] McWilliams JM, Meara E, Zaslavsky AM, et al. Use of health services by previously uninsured
Medicare beneficiaries. N Engl J Med 2007;357:143-53
[135] Fu AZ, Liu GG, Christensen DB, et al. Effect of second-generation antidepressants on mania- and
depression-related visits in adults with bipolar disorder: a retrospective study. Value Health 2007;10:128-36
[136] Schneeweiss S, Maclure M. Use of comorbidity scores for control of confounding in studies using
administrative databases. Int J Epidemiol 2000;29:891-8
[137] Roos LL, Sharp SM, Cohen MM, et al. Risk adjustment in claims-based research: the search for
efficient approaches. J Clin Epidemiol 1989;42:1193-206
[138] Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM
administrative data: differing perspectives. J Clin Epidemiol 1993;46:1075-9
[139] Roos LL, Stranc L, James RC, et al. Complications, comorbidities, and mortality: improving
classification and prediction. Health Serv Res 1997;32:229-38
[140] Von Korff M, Wagner EH, Saunders K. A chronic disease score from automated pharmacy data. J Clin
Epidemiol 1992;45:197-203
[141] Clark DO, Von Korff M, Saunders K, et al. A chronic disease score with empirically derived weights.
Med Care 1995;33:783-95
[142] Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM
administrative databases. J Clin Epidemiol 1992;45:613-9
[143] D'Hoore W, Sicotte C, Tilquin C. Risk adjustment in outcome assessment: the Charlson comorbidity
index. Methods Inf Med 1993;32:382-7
[144] D'Hoore W, Bouckaert A, Tilquin C. Practical considerations on the use of the Charlson comorbidity
index with administrative data bases. J Clin Epidemiol 1996;49:1429-33
[145] Ghali WA, Hall RE, Rosen AK, et al. Searching for an improved clinical comorbidity index for use
with ICD-9-CM administrative data. J Clin Epidemiol 1996;49:273-8
[146] Schneeweiss S. Sensitivity analysis and external adjustment for unmeasured confounders in
epidemiologic database studies of therapeutics. Pharmacoepidemiol Drug Saf 2006;15:291–303
[147] Schneeweiss S, Glynn RJ, Tsai EH, et al. Adjusting for unmeasured confounders in
pharmacoepidemiologic claims data using external information: the example of COX2 inhibitors and
myocardial infarction. Epidemiology 2005;16:17–24
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 227
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[148] Phillips CV. Quantifying and reporting uncertainty from systematic errors. Epidemiology 2003;14:459–
66
[149] Walker AM. Observation an inference. Epidemiology Resources Inc.: Newton, 1991;120-4
[150] Huybrechts KF, Brookhart MA, Rothman KJ, et al. Comparison of different approaches to confounding
adjustment in a study on the association of antipsychotic medication with mortality in older nursing home
patients. Am J Epidemiol 2011;164:1-11. DOI: 10.1093/aje/kwr213
[151] Kriebel D, Zeka A, Eisen EA, et al. Quantitative evaluation of the effects of uncontrolled confounding
by alcohol and tobacco in occupational cancer studies. Int J Epidemiol 2004;33:1040-5
[152] Corrao G, Parodi A, Nicotra F, et al. Cardiovascular protection by initial and subsequent combination
of antihypertensive drugs in daily life practice. Hypertension 2011;58:566-72
[153] Corrao G, Nicotra F, Parodi A, et al. External adjustment for unmeasured confounders improved drugoutcome association estimates based on health care utilization data. J Clin Epidemiol 2012;65:1190-9
[154] Steenland K, Greenland S. Monte Carlo Sensitivity Analysis and Bayesian Analysis of smoking as an
unmeasured confounder in a study of silica and lung cancer. Am J Epidemiol 2004;160:384-90
[155] Sturmer T, Schneeweiss S, Avorn J, et al. Correcting effect estimates for unmeasured confounding in
cohort studies with validation studies using propensity score calibration. Am J Epidemiol 2005;162:279–89
[156] Sturmer T, Schneeweiss S, Glynn RJ. Performance of propensity score calibration (PSC). Am J
Epidemiol 2005;161(suppl): S75
[157] Bowden, RJ.; Turkington, DA. Instrumental Variables. Cambridge University Press;
Cambridge,UK:1984
[158] Angrist JD, Imbens GW, Rubin DB. Identification of causal effects using instrumental variables. J Am
Stat Soc 1996;91:444-55
[159] Angrist JD, Krueger AB. Instrumental variables and the search for identification: from supply and
demand to natural experiments. J Econ Perspect 2001;15:69-85
[160] Murray MP. Avoiding invalid instruments and coping with weak instruments. J Econ Perspect
2006;20:111-32
[161] Brookhart MA, Wang PS, Solomon DH, Schneeweiss S. Evaluating short-term drug effects in claims
databases using physician-specific prescribing preferences as an instrumental variable. Epidemiology
2006;17:268-75
[162] Schneeweiss S, Solomon DH, Wang PS, et al. Simultaneous assessment of short-term gastrointestinal
benefits and cardiovascular risks of selective COX-2 inhibitors and non-selective NSAIDs: an instrumental
variable analysis. Arthritis Rheum 2006;54:3390-8
[163] Greene WH. Econometric Analysis. 3rd edn. Prentice Hall; Upper Saddle River, NJ:1997:740-2
[164] MacKinnon DP, Krull JL, Lockwood CM. Equivalence of the Mediation, Confounding and
Suppression Effect. Prev Sci 2000;1:173-85
[165] Susser, M. Causal thinking in the health sciences: Concepts and strategies of epidemiology. Oxford
University Press; New York: 1973
[166] Breslow NE, Day NE. Statistical methods in cancer research. Volume I—The Analysis of Case-Control
Studies. International Agency for Research on Cancer; Lyon: 1980. IARC Scientific Publications No. 32
[167] Meinert CL. Clinical trials: Design, conduct, and analysis. Oxford University Press; New York: 1986
[168] Robins JM. The control of confounding by intermediate variables. Stat Med 1989;8:679–701
[169] Last JM. A dictionary of epidemiology. Oxford Unversity Press; New York: 1988
[170] Merchant AT, Pitiphat W. Directed acyclic graphs (DAGs): an aid to assess confounding in dental
research. Community Dent Oral Epidemiol 2002;30:399-404
[171] Weinberg CR. Toward a clearer definition of confounding. Am J Epidemiol 1993;137:1-8
[172] Hernán MA, Hernández-Díaz S, Werler MM, et al. Causal knowledge as a prerequisite for confounding
evaluation: an application to birth defects epidemiology. Am J Epidemiol 2002;155:176-84
[173] Hernán MA, Hernández-Díaz S, Robins JM. A structural approach to selection bias. Epidemiology
2004;15:615-25
[174] Pearl J. Causal diagrams for empirical research. Biometrika 1995;82:669–710
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 228
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[175] Hernández-Díaz S, Schisterman EF, Hernán MA. The birth weight “paradox” uncovered? Am J
Epidemiol 2006;164:1115-20
[176] James LR, Brett JM. Mediators, moderators and tests for mediation. J Appl Psychol 1984;69:307-21
[177] Alwin DF, Hauser RM. The decomposition of effects in path analysis. Am Soc Rev 1975;40:37-47
[1788] Holland PW. Causal inference, path analysis, and recursive structural equations models. Soc Methodol
1988;18:449-84
[179] Sobel ME. Effect analysis and causation in linear structural equation models. Psychometrika
1990;55:495-515
[180] Roumie CL, Liu X, Choma NN, et al. Initiation of sulfonylureas versus metformin is associated with
higher blood pressure at one year. Pharmacoepidemiol Drug Saf 2012;21:515-23
[181] Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic
studies. Epidemiology 2009;20:488-95
[182] Hafeman DM, Schwartz S. Opening the black box: a motivation for the assessment of mediation. Int J
Epidemiol 2009;38:838-45
[183] Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research:
conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173-82
[184] Ditlevsen S, Christensen U, Lynch J, et al. The mediation proportion: a structural equation approach for
estimating the proportion of exposure effect on outcome explained by an intermediate variable. Epidemiology
2005;16:114-20
[185] MacKinnon DP, Lockwood CM, Hoffman JM, et al. A comparison of methods to test mediation and
other intervening variable effects. Psychol Meth 2002;7:83–104
[186] Taylor AB, MacKinnon D, Tein JY. Test of the three-path mediated effect. Organiz Res Meth
2008;11:241-69
[187] Huang B, Sivaganesan S, Succop P, et al. Statistical assessment of mediational effects for logistic
mediational models. Stat Med 2004;23:2713-28
[188] Schluchter MD. Flexible approaches to computing mediated effects in generalized linear models:
generalized estimating equations and bootstrapping. Multivar Behav Res 2008;43:268-88
[189] Li Y, Schneider JA, Bennett DA. Estimation of the mediation effect with a binary mediator. Stat Med
2007;26:3398-414
[190] Eskima N, Tabata M, Zhi G. Path analysis with logistic regression models: effect analysis of fully
recursive causal systems of categorical variables. J Japan Stat Soc 2001;31:1–14
[191] Albert JM, Nelson S. Generalized causal mediation analysis. Biometrics 2011;67:1028-38
[192] Hernan MA, Hernandez-Diaz S, Robins JM. A structural approach to selection bias. Epidemiology
2004;15:615-25
[193] Greenland S. Quantifying biases in causal models: classical confounding vs. collider-stratification bias.
Epidemiology 2003;14:300–6
[194] Cole SR, Platt RW, Schisterman EF, et al. Illustrating bias due to conditioning on a collider. Int J
Epidemiol 2010;39:417-20
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 229
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Appendix 3
Curricula vitae and recent pubblications concerning the
CRACK issues of scientific board members
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 230
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Giovanni Corrao
Curriculum vitae
Prof. Giovanni Corrao was born in 1953 in Turin.
ACADEMIC APPOINTMENTS. He became research assistant at the Faculty of Medicine and Surgery,
University of Turin, in 1984, where he was until 1988. In 1988 he became Associated Professor at the Faculty
of Medicine and Surgery, University of L'Aquila, where he stayed until 1993. He is currently full professor in
Medical Statistics, at the Department of Statistics, University of Milan-Bicocca, Milan where since October
2009 is the dean of the faculty of Statistical Sciences.
TEACHING EXPERIENCE. Prof. Corrao is the holder of the teaching of “Medical statistics” (Degree in
Statistics and information management), “Methodology of Clinical and Epidemiological research” and
“Pharmacoepidemiology” (Degree in Biostatistic and Experimental Statistic). Since academic year 2004/05
he’s holder of the teaching of “Biostatistics” of the Degree in Biology at the Univerity of Milan-Bicocca.
Moreover, since academic year 2009/10 he’s holder of the teaching of “Medical Statistics” at the Postgraduate school of Public health Class at the Faculty of Medicine and Surgery at the Catholic University of
Sacred Heart of Rome. Among the current activities there is the teaching of Pharmacoepidemiology at the
Master in “ Research and pre-clinical and clinical development of Drugs” (University of Milan-Bicocca),
“Pharmacovigilance“ (University of Milan), in "Health Services Management" (Università of Siena).
RESEARCH ACTIVITY. His main research interest covers several methodological issues in clinical research
and in the design,analysis and interpretation of statistical models for both observational and experimental
studies. He is involved in many collaborative research projects and his expertise is requested for both the
statistical and epidemiological aspects of the study. More recently he has focused on statistical methods and
epidemiological designs in pharmacoepidemiology research and the use of electronic health database for
pharmacoepidemiological studies.
He is author of 185 publications registered in PUBMED. In the years 2007-2009 he has been elect president
of the Italian Society of Medical Statistics and Clinical Epidemiology (SISMEC).
He is currently both in the editorial board of or reviewer for the following Journals International Journal of
Epidemiology, Journal of Epidemiology and Community Health, Journal of Epidemiology and Biostatistics,
Annals of Epidemiology, Current Drug Safety, Drug Safety, British Medical Journal.
He is the Founding Editor of the Journal Epidemiology, Biostatsitics and Public Health.
Recent publications on CRACK issues
[1] Corrao G, Nicotra F, Parodi A, Zambon A, Soranna D, Heiman F, Merlino L, Mancia G. External
adjustment for unmeasured confounders improved drug-outcome association estimates based on
HEALTHCARE utilization data. J Clin Epidemiol 2012;65:1190-9
[2] Pradelli D, Soranna D, Scotti L, Zambon A, Catapano A, Mancia G, La Vecchia C, Corrao G. Statins and
primary liver cancer: a meta-analysis of observational studies. Eur J Cancer Prev 2012 [Epub ahead of print]
[3] Bellelli G, Mazzola P, Corsi M, Zambon A, Corrao G, Castoldi G, Zatti G, Annoni G. The combined
effect of ADL impairment and delay in time from fracture to surgery on 12-month mortality: an observational
study in orthogeriatric patients. J Am Med Dir Assoc 2012;13:664.e9-664.e14
[4] Soranna D, Scotti L, Zambon A, Bosetti C, Grassi G, Catapano A, La Vecchia C, Mancia G, Corrao G.
Cancer risk associated with use of metformin and sulfonylurea in type 2 diabetes: a meta-analysis. Oncologist
2012;17:813-22
[5] Lapi F, Cipriani F, Caputi AP, Corrao G, Vaccheri A, Sturkenboom MC, Di Bari M, Gregori D, Carle F,
Staniscia T, Vestri A, Brandi M, Fusco V, Campisi G, Mazzaglia G; on behalf of the Bisphosphonates
Efficacy-Safety Tradeoff (BEST) study group. Assessing the risk of osteonecrosis of the jaw due to
bisphosphonate therapy in the secondary prevention of osteoporotic fractures. Osteoporos Int 2012 [Epub
ahead of print]
[6] Bosetti C, Rosato V, Polesel J, Levi F, Talamini R, Montella M, Negri E, Tavani A, Zucchetto A,
Franceschi S, Corrao G, La Vecchia C. Diabetes mellitus and cancer risk in a network of case-control studies.
Nutr Cancer 2012;64:643-51
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 231
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[7] Lapi F, Nicotra F, Scotti L, Vannacci A, Thompson M, Pieri F, Mugelli N, Zambon A, Corrao G, Mugelli
A, Rubino A. Use of antidepressant serotoninergic medications and cardiac valvulopathy: a nested casecontrol study in the health improvement network (THIN) database. Br J Clin Pharmacol 2012;74:536-44
[8] Coloma PM, Trifirò G, Schuemie MJ, Gini R, Herings R, Hippisley-Cox J, Mazzaglia G, Picelli G,
Corrao G, Pedersen L, van der Lei J, Sturkenboom M; EU-ADR Consortium. Electronic healthcare databases
for active drug safety surveillance: is there enough leverage? Pharmacoepidemiol Drug Saf 2012;21:611-21
[9] Corrao G, Zambon A, Parodi A, Merlino L, Mancia G. Incidence of cardiovascular events in Italian
patients with early discontinuations of antihypertensive, lipid-lowering, and antidiabetic treatments. Am J
Hypertens 2012;25:549-55
[10] Corrao G, Nicotra F, Parodi A, Zambon A, Heiman F, Merlino L, Fortino I, Cesana G, Mancia G.
Cardiovascular protection by initial and subsequent combination of antihypertensive drugs in daily life
practice. Hypertension 2011;58:566-72
[11] Mantovani LG, Fornari C, Madotto F, Riva MA, Merlino L, Ferrario MM, Chiodini V, Zocchetti A,
Corrao G, Cesana G. Burden of acute myocardial infarction. Int J Cardiol 2011;150:111-2
[12] Corrao G, Scotti L, Zambon A, Baio G, Nicotra F, Conti V, Capri S, Tragni E, Merlino L, Catapano AL,
Mancia G. Cost-effectiveness of enhancing adherence to therapy with statins in the setting of primary
cardiovascular prevention. Evidence from an empirical approach based on administrative databases.
Atherosclerosis 2011;217:479-85
[13] Imberti D, Bianchi C, Zambon A, Parodi A, Merlino L, Gallerani M, Corrao G. Venous
thromboembolism after major orthopaedic surgery: a population-based cohort study. Intern Emerg Med
2012;7:243-9
[14] Mancia G, Parodi A, Merlino L, Corrao G. Heterogeneity in antihypertensive treatment discontinuation
between drugs belonging to the same class. J Hypertens 2011;29:1012-8
[15] Coloma PM, Schuemie MJ, Trifirò G, Gini R, Herings R, Hippisley-Cox J, Mazzaglia G, Giaquinto C,
Corrao G, Pedersen L, van der Lei J, Sturkenboom M; EU-ADR Consortium. Combining electronic
healthcare databases in Europe to allow for large-scale drug safety monitoring: the EU-ADR Project.
Pharmacoepidemiol Drug Saf 2011;20:1-11
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 232
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Alberico Catapano
Curriculum vitae
Alberico L. Catapano
Born 23 June 1952, Milan (Italy).
Current position Full Professor of Pharmacology, University of Milan. (since 2000) Director of the
Laboratory of Lipoprotein Metabolism (since 1977) Director of the Center for the Study, Prevention and
Therapy of Atherosclerosis of the Italian Society for the study of atherosclerosis, at the “Bassini” Hospital.
(since 1987) Director of Center of Epidemiology end Prevention Pharmacology, University of Milan (since
2004).
Research interest Studies on the “in vitro” and “in vivo” role of lipoproteins in atherosclerosis and on the
links between lipoproteins and inflammation. Clinical and experimental pharmacology of hypolipemidemic
drug. The scientific activity has resulted in more than 300 research papers and reviews of which approx 220
in International peer reviewer journals.
Membership of scientific societies American Heart Association European Society of Cardiology International
Society and Federation of Cardiology International Society of Atherosclerosis Italian Society for the Study of
Atherosclerosis Italian Society of Cardiology (Coordinator of the ATVB Group) Italian Society of
Pharmacology President of the Italian Society for Clinical and Experimental Therapeutics (SITeCS) Società
Italiana di Biochimica Clinica e Biologia Molecolare Clinica President-elect of the European Atherosclerosis
Society (EAS).
Recent publications on CRACK issues
[1] Poli A, Casula M, Tragni E, Brignoli O, Filippi A, Cricelli C, Catapano AL. Reaching LDL-c targets in
high-risk patients requires high-efficacy cholesterol-lowering drugs in more than 50% of cases. The results of
the CHECK study. Pharmacol Res 2011;64:393-6
[2] Catapano AL, Norata GD, Pirillo A. Therapy and clinical trials: aggressive statin therapy versus combined
and emerging approaches. Curr Opin Lipidol 2011;22:324-5
[3] Catapano AL, Chapman MJ, Wiklund O, Taskinen M. The new joint EAS/ESC guidelines for the
management of dyslipidaemias. Atherosclerosis 2011:217:1-1
[4] Catapano AL, Reiner Z, De Backer G, Graham I, Taskinen M, Wiklund O, Agewall S, Alegria E,
Chapman MJ, Durrington P, Erdine S, Halcox J, Hobbs R, Kjekshus J, Perrone Filardi P, Riccardi G, Storey
RF, Wood D. ESC/EAS Guidelines for the management of dyslipidaemias The Task Force for the
management of dyslipidaemias of the European Society of Cardiology (ESC) and the European
Atherosclerosis Society (EAS). Atherosclerosis 2011:217:1-44
[5] Corrao G, Scotti L, Zambon A, Baio G, Nicotra F, Conti V, Capri S, Tragni E, Merlino L, Catapano AL,
Mancia G. Cost-effectiveness of enhancing adherence to therapy with statins in the setting of primary
cardiovascular prevention. Evidence from an empirical approach based on administrative databases.
Atherosclerosis 2011;217:479-85
[6] Reiner Z, Catapano AL, De Backer G, Graham I, Taskinen M, Wiklund O, Agewall S, Alegria E,
Chapman MJ, Durrington P, Erdine S, Halcox J, Hobbs R, Kjekshus J, Filardi PP, Riccardi G, Storey RF,
Wood D. ESC/EAS Guidelines for the management of dyslipidaemias: The Task Force for the management
of dyslipidaemias of the European Society of Cardiology (ESC) and the European Atherosclerosis Society
(EAS). Eur Heart J 2011;32:1769-818
[7] Filippi A, Casula M, Tragni E, Brignoli O, Cricelli C, Poli A, Catapano AL. Blood pressure and
antihypertensive therapy according to the global cardiovascular risk level in Italy: the CHECK Study. Eur J
Cardiovasc Prev Rehabil 2010;17:562-68
[8] Catapano AL. Pitavastatin - pharmacological profile from early phase studies. Atheroscler Suppl
2010;11:3-7
[9] Corrao G, Conti V, Merlino L, Catapano A.L., Mancia G. Results of a retrospective database analysis of
adherence to statin therapy and risk of nonfatal ischemic heart disease in daily clinical practice in Italy. Clin
Ther 2010;32:300-10
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 233
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[10] F. Pellegatta, L. Grigore, Catapano A.L. (2010). Therapy and clinical trials: new insights. Curr Opin
Lipidol 2010:21:394-5
[11] Hobbs FD, Jukema JW, Da Silva PM, McCormack T, Catapano AL. Barriers to cardiovascular disease
risk scoring and primary prevention in Europe. Qjm-An Int J Med 2010;103:727-39
[12] Thompson GR, Catapano AL, Saheb S, Atassi-Dumont M, Barbir M, Eriksson M, Paulweber B,
Sijbrands E, Stalenhoef AF, Parhofer KG. Severe hypercholesterolaemia : therapeutic goals and eligibility
criteria for LDL apheresis in Europe. Curr Opin Lipidol 2010;21:492-8
[13] Lorenz MW, Bickel H, Bots ML, Breteler MM, Catapano AL, Desvarieux M, Hedblad B, Iglseder B,
Johnsen SH, Juraska M, Kiechl S, Mathiesen EB, Norata GD, Grigore L, Polak J, Poppert H, Rosvall M,
Rundek T, Sacco RL, Sander D, Sitzer M, Steinmetz H, Stensland E, Willeit J, Witteman J, Yanez D,
Thompson SG, Prog-Imt Study Group. Individual progression of carotid intima media thickness as a
surrogate for vascular risk (PROG-IMT): Rationale and design of a meta-analysis project. Am Heart J
2010;159:730-6
[14] Atella V, Brady A, Catapano AL, Critchley J, Graham IM, Hobbs FD, Leal J, Lindgren P, Vanuzzo D,
Volpe M, Wood D, Paoletti R. Bridging science and health policy in cardiovascular disease: focus on lipid
management: A Report from a Session held during the 7th International Symposium on Multiple Risk Factors
in Cardiovascular Diseases: Prevention and Intervention--Health Policy, in Venice, Italy, on 25 October,
2008. Atheroscler Suppl 2009;10:3-21
[15] Catapano A.L. (2009). Perspectives on low-density lipoprotein cholesterol goal achievement. Curr Med
Res Opin 2009:25:431-47
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 234
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Giancarlo Cesana
Curriculum vitae
Name
Birthdate
Birthplace
Citizenship
Academic Training
Dec 73
1975
1981
Giancarlo Cesana
August 16, 1948
Carate Brianza, Milan, Italy
Italian
MD Licensure
University of Milan
Specialization in Occupational Medicine
University of Milan
Specialization in Medical Psychology
University of Milan
Academic and Work Appointments
1975-1985
Assistant Professor of Occupational Medicine at Clinica del Lavoro Luigi
Devoto, University of Milan
1986-2000
Associate Professor of Occupational Medicine at the University of Milan: at Clinica
del Lavoro Luigi Devoto until 1991 and then at Institute of Biomedical Science San
Gerardo, Monza. From 1991 to 2005 dr. Cesana Had been the director of the Research
Center on Chronic Degenerative Diseases
1986-now
Director of the Research Centre on Chronic Degenerative Diseases, now Research
Centre on Public Health
2000-2007
Full Professor of Occupational Medicine, University of Milano Bicocca, at San
Gerardo Hospital, Monza, where dr. Cesana had been the Chief of the Occupational
Medicine Unit from 1997 to 2003
2005-2008
Director of the Department of Preventive and Clinical an Preventive Medicine,
Representative of the Department Directors at the Academic Senate, University of
Milano Bicocca
2007-now
Full Professor of Hygiene and Public Health, since 2003 dr. Cesana has been also
Professor of Hystory of Medicine
2008-2009
Scientific Director of San Giuseppe Hospital, Milan
2009-now
President of “Fondazione Ca’ Granda. Ospedale Maggiore Policlinico”, Milan
Research
After some research on ergonomics and work physiology, his main interest has been devoted to the study of
stress psychophysiology, applied to shiftwork, aging and work related diseases. To go deeper into these studies
he has expanded his interest to the epidemiology of coronary disease and risk, with particular attention to the
etiological role of socio-occupational factors. The attention to the economic and cultural determinants of health
has been completed with the study of the history of medicine, which prof. Cesana teaches in the schools of
medicine and biostatistics.
He has been involved, with the role of principal investigator, in various national
and international projects: W.H.O MONICA Project (a worldwide project of MONItoring trends in
CArdiovascular diseases along ten years); FATMA Project (a collaborative study on disease factors -FATtori di
Malattia- coordinated by the Italian National Research Council); RIFLE Project (another collaborative Italian
study on RIsk Factors and Life Expectancy, coordinated by the Italian National Health Institute); PAMELA
Project (an observational study on ambulatory monitoring of blood pressure -Pressioni AMbulatoriali E Loro
Associazioni- in a population sample of Northern Italy); HEART AT WORK network (a BIOMED concerted
action of the European Union directed to the speculative evaluation of the results of several European studies on
psycho-social factors and cardiovascular disease and risk), CUORE Project (longitudinal study of several Italian
cohorts, coordinated by the Italian National Health Institute), CAMUNI project (epidemiological observatory of
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 235
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
cardiovascular disease in Lombardia Region, DENALI Project (a datawarehouse for the epidemiological analysis
of health administrative data of Lombardia Region)
He has received several grants from public and private institutions
He is author/co-author of about 300 papers, of which about 200 in peer reviewed journals, and 5 books.
Recent publications on CRACK issues
[1] Vishram JK, Borglykke A, Andreasen AH, Jeppesen J, Ibsen H, Jørgensen T, Broda G, Palmieri L,
Giampaoli S, Donfrancesco C, Kee F, Mancia G, Cesana G, Kuulasmaa K, Sans S, Olsen MH; On behalf of
the MORGAM Project. Impact of Age on the Importance of Systolic and Diastolic Blood Pressures for Stroke
Risk: The MOnica, Risk, Genetics, Archiving, and Monograph (MORGAM) Project. Hypertension
2012;60:1117-23
[2] Moia M, Mantovani LG, Carpenedo M, Scalone L, Monzini MS, Cesana G, Mannucci PM. Patient
preferences and willingness to pay for different options of anticoagulant therapy. Intern Emerg Med. 2012
[Epub ahead of print]
[3] Bonzini M, Ferrario MM, Bertù L, Bono G, Vidale S, Veronesi G, Chambless L, Cesana GC. Temporal
trends in ischemic and hemorrhagic strokes in Northern Italy: results from the cardiovascular monitoring unit
in Northern Italy population-based register, 1998-2004. Neuroepidemiology 2012;39:35-42
[4] Veronesi G, Ferrario MM, Chambless LE, Borsani A, Fornari C, Cesana G. The effect of
revascularization procedures on myocardial infarction incidence rates and time trends: the MONICA-Brianza
and CAMUNI MI registries in Northern Italy. Ann Epidemiol 2012;22:547-53
[5] Corrao G, Nicotra F, Parodi A, Zambon A, Heiman F, Merlino L, Fortino I, Cesana G, Mancia G.
Cardiovascular protection by initial and subsequent combination of antihypertensive drugs in daily life
practice. Hypertension 2011;58:566-72
[10] Mantovani LG, Fornari C, Madotto F, Riva MA, Merlino L, Ferrario MM, Chiodini V, Zocchetti A,
Corrao G, Cesana G. Burden of acute myocardial infarction. Int J Cardiol 2011;150:111-2
[11] Grassi G, Padmanabhan S, Menni C, Seravalle G, Lee WK, Bombelli M, Brambilla G, Quarti-Trevano F,
Giannattasio C, Cesana G, Dominiczak A, Mancia G. Association between ADRA1A gene and the metabolic
syndrome: candidate genes and functional counterpart in the PAMELA population. J Hypertens
2011;29:1121-7
[12] Ferrario MM, Veronesi G, Chambless LE, Sega R, Fornari C, Bonzini M, Cesana G. The contribution of
major risk factors and job strain to occupational class differences in coronary heart disease incidence: the
MONICA Brianza and PAMELA population-based cohorts. Occup Environ Med 2011;68:717-22
[13] Corrao G, Parodi A, Nicotra F, Zambon A, Merlino L, Cesana G, Mancia G. Better compliance to
antihypertensive medications reduces cardiovascular risk. J Hypertens 2011;29:610-8
[14] Veronesi G, Ferrario MM, Chambless LE, Sega R, Mancia G, Corrao G, Fornari C, Cesana G. Gender
differences in the association between education and the incidence of cardiovascular events in Northern Italy.
Eur J Public Health 2011;21:762-7
[15] Fornari C, Donfrancesco C, Riva MA, Palmieri L, Panico S, Vanuzzo D, Ferrario MM, Pilotto L,
Giampaoli S, Cesana G. Social status and cardiovascular disease: a Mediterranean case. Results from the
Italian Progetto CUORE cohort study. BMC Public Health 2010;10:574
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 236
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Carlo La Vecchia
Curriculum vitae
DATE OF BIRTH
Feb. 27, 1955; Place of birth: Milano, Italy; Citizienship: Italian; Languages: English, French (and Italian).
CURRENT STATUS
- Head, Department of Epidemiology, Istituto di Ricerche Farmacologiche "Mario Negri", Milan, Italy (2007-)
- Head, Laboratory of Epidemiology, Istituto di Ricerche Farmacologiche "Mario Negri", Milan, Italy (1989-)
- Associate Professor of Epidemiology, Istituto di Statistica Medica e Biometria, Università di Milano (1992-)
- Adjunct Professor of Epidemiology University of Lausanne, Switzerland (2002-2008)
- Adjunct Professor of Medicine, School of Medicine, Vanderbilt University, Nashville, TN, (2006-2009).
- Senior Fellow, International Agency for Research on Cancer (IARC/WHO), Lyon, France (2007-8).
ADDRESS
Istituto di Ricerche Farmacologiche"Mario Negri" Via La Masa 19 - 20156 Milan (Italy) (Tel. +39-0239014.1; Fax +39-02-33200231);
Istituto di Statistica Medica e Biometria, Università di Milano, Via Venezian 1 - 20133 Milan (Italy)
WORK AND RELATED EXPERIENCE:
• 1979 March to date - Researcher at the Istituto di Ricerche Farmacologiche "Mario Negri", Milan.
• 1981-1983 -Research Fellow at the Dept. of Community Medicine and Medical Practice, Univ. of Oxford.
• from 1987 to 1992 - Associate Professor of Epidemiology, University of Lausanne.
• 1996-2001 - Adjunct Associate Professor of Epidemiology, Harvard Schoold of Public Health, Boston, Ma.
Awards:
• 1981-1983 - Two-year Research Training Fellowship by the Italian Labor Ministry and the EEC
• 1991 - European Visiting Professor to the Royal Society of Medicine, London.
• 1993 - Glaxo Prize for medical publications.
• 2006 – Commendatore della Repubblica Italiana for scientific achievements.
MISCELLANEA
• Registered Journalist, Milan (Elenco pubblicisti No. 52412).
• Honorary Senior Lecturer in Oral Medicine, Eastman Dental Institute, University College London (19962001).
• Visiting Lecturer in the Department of Epidemiology, Harvard School of Public Health, Boston, Mass.,
U.S.A. (1994-95).
• Member of the UICC - American Cancer Society Fellowship Committee (1991-95).
• Member of the "Steering Committee" of the International cooperative network of case-control of the
SEARCH Programme of the International Agency for Research on Cancer, IARC/OMS (1989-91).
• Temporary advisor at the International Agency for Research on Cancer, IARC/WHO and at the WHO,
Geneva (1989--).
• Member of the Executive Committee of the European Society for Human Reproduction (ESHRE) (199195).
• Member of the Steering Committee, Collaborative Group on Hormonal Factors and Breast Cancer.
Chairman of the Advisory Committee, Collaborative Group of Hormonal Factors and Cervical Cancer.
Oxford, UK
• Member of the Scientific Committee, Air Quality Project, Regione Lombardia (1998-2000).
• Member of the CPMP Expert Group on Oral Contraceptives and Cardiovascular Risks, EMEA, London,
1998-2001.
• Consultant, U.S. Surgeon General's Report on Smoking and Women's Health (1997-2000).
• Working Group Member, International Agency for Research on Cancer (IARC), Monograph 51 on Coffee,
Tea, Mate, Methylxanthines and Methylglyoxal, 1990.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 237
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
• Working Group Chairman, IARC Monographs on the Evaluation of Carcinogenic Risks to Human.
Monographs 72 and Member Monographs 91 on Hormonal Contraception and Post-menopausal Hormonal
Therapy, 1998 and 2005, and 95 on Alcohol, 2007.
• Member, Scientific Board of the Italian Group of Epidemiologic Studies in Dermatology (GISED, 2001--).
• Member, Scientific Committee, International Head and Neck Cancer Epidemiology (INHANCE)
consortium (IARC/ NCI) (2006--).
• Chairman, Società Italiana della Riproduzione (2002-2003).
• Member Board of Trustees, International Health Foundation, Utrecht (2001-2003).
• Member, Scientific Committee, Foundation for the Advancement of the Mediterranean Diet, Barcelona,
Spain (2002--).
• Member, National Oncology Commission, 2007-2008.
• Editor: European Journal of Public Health (1993-2003) - Journal of Epidemiology and Biostatistics (19962002)
• Associate Editor: European Journal of Cancer Prevention (2004--); Cancer Letter (2008--).
• Member of the Editorial board of the following journals:
Alimentazione e Prevenzione (2000--); American Journal of Epidemiology (1991-97); Archives of Medical
Science (2007--); Asian Pacific Journal of Cancer Prevention (2000--); Cancer Causes and Control (1991-96);
Current Cancer Therapy Reviews (2005 --); Dermatology Research and Practice (2007--); Digestive and
Liver Disease (2001--); Economia Politica del Farmaco (2004-); European Journal of Cancer (1991-95);
European Journal of Cancer Prevention (1991--); European Journal of Clinical Nutrition (1996-2007);
European Journal of Nutrition (1998--); In Scope Oncology & Haematology (2004--); International Journal of
Cancer (2000-2008); Journal of Nephrology (1992--); Maturitas (2008--); Nutrition and Cancer (2000--);
Oncology (1994-1995); Open Cancer Journal (2007---); Oral Oncology (2003--); Revisiones en Ginecologia
y Obstetricia (2000--); Revista Española de Nutrición Comunitaria (1996--); Revue d'Epidémiologie et de
Santé Publique (1991--); Sozial und Praeventivmedizin (1990--2001); The Lancet, edizione italiana (2005--);
Tumori (1993--).
MAIN FIELDS OF INTEREST
Cancer epidemiology (case-control studies on cancers of the breast, female genital tract, digestive sites,
urinary organs, lymphoreticular malignancies, etc.); epidemiological studies on the risk related to diet,
tobacco, oral contraceptive use and occupational or environmental exposure to toxic substances; coordination
of clinical trials. Analysis of temporal trends and of geographical distribution of mortality from cancer,
cardiovascular diseases, perinatal and other selected conditions.
PUBLICATIONS
Over 1,750 publications; of these, over 1,400 included in PUBMED/MEDLINE, with over 42,000 quotations.
Recent publications on CRACK issues
[1] La Vecchia C, Bosetti C. Urological cancer: Aspirin and the risk of prostate cancer mortality. Nat Rev
Clin Oncol 2012;9:616-7
[2] Pradelli D, Soranna D, Scotti L, Zambon A, Catapano A, Mancia G, La Vecchia C, Corrao G. Statins and
primary liver cancer: a meta-analysis of observational studies. Eur J Cancer Prev 2012 [Epub ahead of print]
[3] Kane EV, Bernstein L, Bracci PM, Cerhan JR, Costas L, Dal Maso L, Holly EA, La Vecchia C, Matsuo
K, Sanjose S, Spinelli JJ, Wang SS, Zhang Y, Zheng T, Roman E, Kricker A; for the InterLymph
Consortium. Postmenopausal hormone therapy and non-Hodgkin lymphoma: a pooled analysis of InterLymph
case-control studies. Ann Oncol 2012 [Epub ahead of print]
[4] Hu J, La Vecchia C, Augustin LS, Negri E, de Groh M, Morrison H, Mery L; the Canadian Cancer
Registries Epidemiology Research Group. Glycemic index, glycemic load and cancer risk. Ann Oncol 2012
[Epub ahead of print]
[5] Kane EV, Roman E, Becker N, Bernstein L, Boffetta P, Bracci PM, Cerhan JR, Chiu BC, Cocco P, Costas
L, Foretova L, Holly EA, La Vecchia C, Matsuo K, Maynadie M, Sanjose S, Spinelli JJ, Staines A, Talamini
R, Wang SS, Zhang Y, Zheng T, Kricker A; InterLymph Consortium. Menstrual and reproductive factors,
and hormonal contraception use: associations with non-Hodgkin lymphoma in a pooled analysis of
InterLymph case-control studies. Ann Oncol 2012;23:2362-74
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 238
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
[6] Soranna D, Scotti L, Zambon A, Bosetti C, Grassi G, Catapano A, La Vecchia C, Mancia G, Corrao G.
Cancer risk associated with use of metformin and sulfonylurea in type 2 diabetes: a meta-analysis. Oncologist
2012;17:813-22
[7] Bosetti C, Rosato V, Polesel J, Levi F, Talamini R, Montella M, Negri E, Tavani A, Zucchetto A,
Franceschi S, Corrao G, La Vecchia C. Diabetes mellitus and cancer risk in a network of case-control studies.
Nutr Cancer 2012;64:643-51
[8] Bosetti C, Rosato V, Gallus S, Cuzick J, La Vecchia C. Aspirin and cancer risk: a quantitative review to
2011. Ann Oncol 2012;23:1403-15
[9] Pelucchi C, Chatenoud L, Turati F, Galeone C, Moja L, Bach JF, La Vecchia C. Probiotics
supplementation during pregnancy or infancy for the prevention of atopic dermatitis: a meta-analysis.
Epidemiology 2012;23:402-14
[10] Bosetti C, Rosato V, Gallus S, La Vecchia C. Aspirin and urologic cancer risk: an update. Nat Rev Urol
2012;9:102-10
[11] Bonifazi M, Rossi M, Moja L, Scigliano VD, Franchi M, La Vecchia C, Zocchetti C, Negri E.
Bevacizumab in clinical practice: prescribing appropriateness relative to national indications and safety.
Oncologist 2012;17:117-24
[12] Pelucchi C, Serraino D, Negri E, Montella M, Dellanoce C, Talamini R, La Vecchia C. The metabolic
syndrome and risk of prostate cancer in Italy. Ann Epidemiol 2011;21:835-41
[13] Bosetti C, Bravi F, Talamini R, Montella M, Negri E, La Vecchia C. Aspirin and risk of endometrial
cancer: a case-control study from Italy. Eur J Cancer Prev 2010;19:401-3
[14] Parazzini F, Pelucchi C, Talamini R, Montella M, La Vecchia C. Use of fertility drugs and risk of
endometrial cancer in an Italian case-control study. Eur J Cancer Prev 2010;19:428-30
[15] Rondanelli M, Giacosa A, Opizzi A, Pelucchi C, La Vecchia C, Montorfano G, Negroni M, Berra B,
Politi P, Rizzo AM. Effect of omega-3 fatty acids supplementation on depressive symptoms and on healthrelated quality of life in the treatment of elderly women with depression: a double-blind, placebo-controlled,
randomized clinical trial. J Am Coll Nutr 2010;29:55-64
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 239
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Giorgio Vittadini
Curriculum vitae
1979 - Bachelors with honours (110/110 cum laude) in Economics, Catholic University of the
Sacred Heart, Milan, Italy
Language: Italian (Native), English (Fluent)
Current Positions
Since 2000 - Full Professor of Statistics, of Methodological Statistics. Faculty of Statistics,
University of Milano-Bicocca.
Since 2005 - Chair of the PhD course on “Statistics”
Previous positions
1999/2000 - Associate Prof. of Methodological Statistics, Faculty of Political Sciences,
University of Milan
1998/1999 - Associate Prof. of Methodological Statistics, Faculty of Statistics, Univ. of
Milano -Bicocca
1993 - Associate Professor of Methodological Statistics, Faculty of Political Sciences,
University of Milan
1987 - PhD Methodological Statistics, University of Trento, Italy
1986 - Assistant Professor of Methodological Statistics, Faculty of Economies, University of
Brescia, Italy
1997/2005 - Scientific Director of the International Interuniversity Centre of Research for
Public Utility Services – CRISP
Current courses
- Degree in Statistics at the University of Milano-Bicocca: Multivariate Analysis, Quality
Control .
- Phd course in Statisics: Multivariate Analysis.
National Statistical Societies
- Associated member of the Italian Society of Statistics (SIS).
International Statistical Societies
- Associated member of the American Statistical Association.
- Associated member of the International Federation of Classification Society.
Editorial Board member
- Member of Editorial Board of Statistical Methods and Application (International Review,
edited by Springer and Verlag promoted by Italian Society of Statistics)
- 1997/2005 - Scientific Director of the International Interuniversity Centre of Research for
Public Utility Services – CRISP
- Since 1997 - Director of the “Non Profit Review” (edited by Maggioli Publisher)
Research Activities
Methodological research: Multivariate Analysis (Factorial methods, Cluster Analysis,
Categorical Data Analysis, Non Symmetrical Data Analysis, Multiway Data Analysis;
Computational Statistics; Symbolic Data Analysis, Data Mining, Structural Equation
Modelling
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 240
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Applied Research: Economics (econometric models), Medical Statistics, Total Quality
(Customer Satisfaction, Multivariate Control Charts), Policy evaluation and analysis, Human
Capital.
Recent publications on CRACK issues
[1] Berta, P., Seghieri C., Vittadini G. (2013), Comparing health outcomes among hospitals: the experience of
the Lombardy Region, Health Care Management Science, DOI: 10.1007/s10729-013-9227-1[2] Fattore, M.,
Pelagatti, M.M., Vittadini, G. (2012), Inconsistencies of the Pls-pm approach to structural equation models with
formative-reflective schemes, Electron. J. App. Stat. Anal., Vol. 5, Issue 3, pp. 334-338, e-ISSN 2070-5948, DOI
10.1285/i20705948v5n3p333
[3] P.G. Lovaglio, G. Vittadini (2012), Multilevel Dimensionality-Reduction Methods, Statistical Methods &
Applications, Springer Verlag – DOI 10.1007/s10260-012-0215-2, published on-line: 27.9.2012
[4] S. Ingrassia, S.C. Minotti, G. Vittadini (2012), Local Statistical Modeling via a Cluster-Weighted Approach
with Elliptical Distributions, Journal of Classification, vol 29; 363-401, Springer Verlag-DOI 10.1007/s00181012-9114-3 . http://www.springerlink.com/openurl.asp?genre=article&id=doi:10.1007/s00357-012-9114-3
[5] G. Vittadini, P. Berta, G. Martini, G. Callea (2012), The effect of a law limiting upcoding on hospital
admissions: Evidence from Italy, special issue of Empirical Economics on Health Econometrics, Springer
Verlag-DOI 10.1007/s00181-012-0548-6
[6] P.G. Lovaglio, G. Vittadini (2012), The Balanced Scorecard in Health Care: a Multilevel Latent Variable
approach, Journal of Modelling in Management, vol. 7, n. 1, ISSN. 1746-5664
[7] F. Moscone, E. Tosetti, G. Vittadini (2011), Social Interaction in Patients’ Hospital Choice: Evidence from
Italy, Journal of the Royal Statistical Society Series A (Statistics in Society), 175,Part 2, pp. 453-472
[8] P.G. Lovaglio, G. Vittadini (2010), Balanced Scorecard Health System: a latent variable approach, Statistica
Applicata, vol. 3-4/2008, pp.273-292
[9] P. Berta, G. Callea, G. Martini, G. Vittadini (2010), The Effects of Upcoding, Cream Skimming and
Readmissions on the Italian Hospitals Efficiency Modelling: a Population-based Investigation, Economic
Modelling, Volume 27, Issue 4, Pages 789-890, July 2010, ISSN 0264-9993,
http://www.sciencedirect.com/science/journal/02649993/27/4
[10] E. Monzani, A. Erlicher, A. Lora, P.g. Lovaglio, G Vittadini (2008), Does community care work? A model
to evaluate the effectiveness of mental healthservices , International Journal of Mental Health Systems 2008,
2:10 , http://www.ijmhs.com/content/2/1/10/abstract
[11] G. Vittadini, S.C. Minotti, M. Fattore, P.G. Lovaglio (2007), On the relationships among latent variables
and residuals in PLS path modeling: The formative-reflective scheme, in Computational Statistics & Data
Analysis, Elsevier B.V., Vol. 51, Issue 12, pp.5828-5866
[12] G. Vittadini, S. C. Minotti(2005), A methodology for measuring the relative effectiveness of health
services, IMA Journal of Management Mathematics (Special Healthcare Issue), 16(3), pp. 239-254.
http://imaman.oxfordjournals.org/papbyrecent.dtl.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 241
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Giuseppe Mancia
Curriculum vitae
Giuseppe Mancia is Professor of Medicine, Head of the Division of Internal Medicine, Chairman of the
Department of Medicine and Director of the Post Graduate School of Cardiology at the University MilanBicocca and the S. Gerardo Hospital, Monza, Milan.
He trained at the University of Siena, Medical School where he graduated in 1964. Ph.D. in Physiology in
1970. Investigator in the National Research Council in 1965-1969. Assistant Professor of Medicine,
University of Milan, School of Medicine, in 1969-1973. Post-graduate Fellow of the US Public Health
Service and Research Associate, Mayo Clinic and Foundation in 1972-1974. Resident in Cardiology, Virginia
Commonwealth University in 1974. Associate Professor of Medicine in 1973-1985, and Full Professor of
Medicine at the University of Milan School of Medicine (1986-1998).
He has served as President (1988-1990) and Secretary (1984-1988) of the International Society of
Hypertension (ISH), President of the European Society of Clinical Investigation (1980-1982), Chairman of
the Working Group on Hypertension and the Heart of the European Society of Cardiology (ESC,1994-1996),
member of the Executive Scientific Committee of ESC (1986-1991), President of the Italian Society of
Hypertension (1997-1999) and President of the European Society of Hypertension (ESH) (1999-2001). He is
member ex- officio (only European) of the Executive Council of the American Society of Hypertension. He
has been ISH officer for the Forum of National Societies, member of the Committee on Cardiovascular Drugs
and Therapy of the World Heart Federation and the ESC Task Force of the guidelines on Prevention of
Cardiovascular Diseases (1993, 1998, 2003), chairman of the Committee (2003 and 2007) for the ESH/ESC
Guidelines on Hypertension and member of the Writing Committee of the WHO/ISH Guidelines in
Hypertension (1999). He is currently Chairman of the Educational Board of ESH and the WHO/ISH Liaison
Committee.
Giuseppe Mancia has received several Awards, among which the Heymans Lecture and Award of the
International Society of Pharmacology, the Wright International Lecture of the High Blood Pressure Council
of Australia, the Volhard Award and the Tigersted Award of the International Society of Hypertension, the
Folkow Award of the European Society of Hypertension, the Spinoza Honorary Chair of the University of
Amsterdam, the Honorary Professorship of the University of Cordoba, the Life Achievement Award of the
Italian Society of Hypertension. He was given the Degree in Medicine Honoris Causa by the University of
Gdansk (2008). He was given the Pickering Lecture of the British Hypertension Society, the Merck-Frost
Lecture of the Canadian Medical Society, the Robert Tigersted Lecture of the Finnish Hypertension Society,
the Salam Memorial Lecture for Educational Achievement (Lebanon Cardiology Society) and the Brian
Bronte Steward Memorial Lecture of the University of Glasgow. He was appointed Lecturer of the Year by
the Belgian Universities and Hypertension Leagues (1992) and Honorary member of the Academy of
Sciences of Cordoba. He was conferred Talal Zein Foundation Award, the First International Recordati Prize
(2000), the 2001 Invernizzi Award for Medicine and the Gold Medal of the Lorenzini Foundation (2008). He
is honorary member of several national hypertension Society.
He has been invited to give state-of-the-art, keynote or special lectures at more than 400 international
meetings. He has also been guest lecturer at meetings of many national societies of hypertension, cardiology
and internal medicine.
Giuseppe Mancia's research interests are pathophysiology, diagnosis and therapy of hypertension, heart
failure, coronary and other cardiovascular diseases. His expertise includes ambulatory blood pressure
monitoring, neural control of the circulation, large artery mechanics and clinical trials. He has published more
than 1100 original papers, reviews and editorials in peer-review journals. He has edited more than 70 special
issues or supplements to cardiovascular or internal medicine journals, and various books among which the
Manual of Hypertension (2002) and the ESH Textbook of Hypertension (2008). He is deputy Editor of the
Journal of Hypertension. His papers have received more than 30000 citations in the international scientific
literature.
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 242
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Recent publications on CRACK issues
[1] Falcone C, Buzzi MP, Bozzini S, Boiocchi C, D'Angelo A, Schirinzi S, Choi J, Ochan Kilama M,
Esposito C, Torreggiani M, Mancia G, Investigators T. Relationship between s RAGE and eotaxin-3 with
CRP in hypertensive patients at high cardiovascular risk. J Nephrol 2012 [Epub ahead of print]
[2] Mancia G, Parati G, Bilo G, Gao P, Fagard R, Redon J, Czuriga I, Polák M, Ribeiro JM, Sanchez R,
Trimarco B, Verdecchia P, van Mieghem W, Teo K, Sleight P, Yusuf S. Ambulatory Blood Pressure Values
in the Ongoing Telmisartan Alone and in Combination with Ramipril Global Endpoint Trial (ONTARGET).
Hypertension 2012;60:1400-6
[3] Corrao G, Nicotra F, Parodi A, Zambon A, Soranna D, Heiman F, Merlino L, Mancia G. External
adjustment for unmeasured confounders improved drug-outcome association estimates based on
HEALTHCARE utilization data. J Clin Epidemiol 2012;65:1190-9
[4] Grassi G, Seravalle G, Brambilla G, Bombelli M, Dell'oro R, Gronda E, Mancia G. Novel
antihypertensive therapies: renal sympathetic nerve ablation and carotid baroreceptor stimulation. Curr
Hypertens Rep 2012;14:567-72
[5] Weber MA, Julius S, Kjeldsen SE, Jia Y, Brunner HR, Zappe DH, Hua TA, McInnes GT, Schork A,
Mancia G, Zanchetti A. Cardiovascular outcomes in hypertensive patients: comparing single-agent therapy
with combination therapy. J Hypertens 2012;30:2213-22
[6] Pradelli D, Soranna D, Scotti L, Zambon A, Catapano A, Mancia G, La Vecchia C, Corrao G. Statins and
primary liver cancer: a meta-analysis of observational studies. Eur J Cancer Prev 2012 [Epub ahead of print]
[7] Vishram JK, Borglykke A, Andreasen AH, Jeppesen J, Ibsen H, Jørgensen T, Broda G, Palmieri L,
Giampaoli S, Donfrancesco C, Kee F, Mancia G, Cesana G, Kuulasmaa K, Sans S, Olsen MH; On behalf of
the MORGAM Project. Impact of Age on the Importance of Systolic and Diastolic Blood Pressures for Stroke
Risk: The MOnica, Risk, Genetics, Archiving, and Monograph (MORGAM) Project. Hypertension
2012;60:1117-23
[8] Giannattasio C, Cairo M, Cesana F, Alloni M, Sormani P, Colombo G, Grassi G, Mancia G. Blood
pressure control in italian essential hypertensives treated by general practitioners. Am J Hypertens
2012;25:1182-7
[9] Mancia G. Additional drug treatment in resistant hypertension: need for randomized studies. J Hypertens
2012;30:1514-5
[10] Mancia G, Facchetti R, Parati G, Zanchetti A. Visit-to-visit blood pressure variability, carotid
atherosclerosis, and cardiovascular events in the European Lacidipine Study on Atherosclerosis. Circulation
2012;126:569-78
[11] Torsello A, Bresciani E, Ravelli M, Rizzi L, Bulgarelli I, Ricci G, Ghiazza B, Del Puppo M, Mainini V,
Omeljaniuk RJ, Tamiazzo L, Mancia G, Magni F, Locatelli V. Novel domain-selective ACE-inhibiting
activity of synthetic growth hormone secretagogues. Pharmacol Res 2012;66:317-24
[12] Mancia G. Reporting blood pressure effects of antihypertensive treatment in scientific papers: are
guidelines needed? J Hypertens 2012;30:1307-9
[13] Jardine MJ, Hata J, Woodward M, Perkovic V, Ninomiya T, Arima H, Zoungas S, Cass A, Patel A,
Marre M, Mancia G, Mogensen CE, Poulter N, Chalmers J; ADVANCE Collaborative Group. Prediction of
kidney-related outcomes in patients with type 2 diabetes. Am J Kidney Dis 2012;60:770-8
[14] van Dieren S, Kengne AP, Chalmers J, Beulens JW, Cooper ME, Grobbee DE, Harrap S, Mancia G,
Neal B, Patel A, Poulter N, van der Schouw YT, Woodward M, Zoungas S. Effects of blood pressure
lowering on cardiovascular outcomes in different cardiovascular risk groups among participants with type 2
diabetes. Diabetes Res Clin Pract 2012;98:83-90
[15] Soranna D, Scotti L, Zambon A, Bosetti C, Grassi G, Catapano A, La Vecchia C, Mancia G, Corrao G.
Cancer risk associated with use of metformin and sulfonylurea in type 2 diabetes: a meta-analysis. Oncologist
2012;17:813-22
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 243
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Appendix 4
Staff members, potentiality and main characteristics of
accredited laboratories affering to the CRACK program
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 244
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Laboratory of
PHARMACOEPIDEMIOLOGY AND HEALTHCARE
RESEARCH
Location
Dept. of Statistics and Quantitative Methods, Unit
Biostatistics, Epidemiology and Public Health, University
Milano-Bicocca Building U7 – Second floor – Offices 2062,
64, 65, 66
Via Bicocca degli Arcimboldi, 8 20126 Milano, Italy
Staff
Scientific responsible
Giovanni Corrao (full professor of medical statistics at the
University of Milano-Bicocca);
☛ [email protected]; ✈ 02.64485854
Laboratory head
Antonella Zambon (associate professor of medical statistics at
the University of Milano-Bicocca);
☛ [email protected]; ✈ 02.64485814
Research staff
Federica Nicotra (post-doc fellowship)
Lorenza Scotti (post-doc fellowship)
Silvana Antonietta Romio (PhD in Statistics)
Arianna Ghirardi (PhD student)
Giulia Segafredo (PhD student)
Andrea Arfè (MSc student)
Davide Soranna (MSc student)
Buthaina Ibrahim (MSc student)
Technical staff
Riccardo Giani
Administrative staff
Gianpiero Latino
Computing
environments
Software access
Other
7 Windows PCs (32 bit)
1 UNIX server (32 bit)
1 Windows Server (32 bit)
1 Windows Server (64 bit)
Several statistical analysis packages (SAS, STATA, SPSS)
Microsoft Office
Web access to many biomedical and statistical academic
journals
Access to University libraries
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 245
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Laboratory of EPIDEMIOLOGY and PUBLIC HEALTH
RESEARCH
Location
Staff
Centre on Public Health. University of Milano-Bicocca. Villa
Serena, 6° piano.
Via Pergolesi 33, 20900 Monza (MB) Italy
Scientific responsible
Giancarlo Cesana (full professor of hygiene at the University
of Milano-Bicocca);
☛ [email protected]; ✈ 039.2333097/8
Laboratory head
Carla Fornari (PhD in Epidemiology and Biostatistics –
graduate technician at the University of Milano-Bicocca);
☛ [email protected]; ✈ 039.2333097/8
Research staff
Fabiana Madotto (post-doc research fellow)
Sara Conti (PhD Student)
Michele Riva (MD, PhD student)
Alessandra Lafranconi (MD, MSc, resident)
External Collaboration Staff
Luciana Scalone (PhD)
Paolo Cortesi (Phd)
Technical staff
Efuture s.r.l.
Administrative staff
Carla Ponti, Enzo Mauri
Computing
environments
Software access
2 Windows Laptops (64 bit)
1 Windows PC (64 bit)
6 Windows PCs (32 bit)
2 Windows Server (32 bit)
Statistical packages (SAS, SPSS)
ArcGIS
Microsoft Office
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 246
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Other
Access to University libraries with direct web-access to
electronic resources (journals)
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 247
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Laboratory of
EPIDEMIOLOGY
Location
Dept. of Epidemiology, Istituto di Ricerche Farmacologiche
“Mario Negri”
Via La Masa,19 - 20156 Milano, Italy
Staff
Scientific responsible
Carlo La Vecchia (Head Dept Epidemiology; full professor of
Medical Statistics at the University of Milano);
☛ [email protected]; ✈ 0239014527
Laboratory head
Eva Negri (Head Laboratory of Epidemiological Methods);
☛ [email protected]; ✈ 0239014525
Research staff
Cristina Bosetti (Head Unit Cancer Epidemiology)
Marta Rossi (senior epidemiologist/biostatistician)
Technical staff
Matteo Franchi (young biostatistician)
Valentina Rosato (young biostatistician)
Administrative staff
Ivana Garimoldi
Computing
environments
Software access
Other
Over 10 Windows PCs
1 Windows Server
Statistical analysis packages (SAS, STATA,..)
Microsoft Office
Endnote
Web access to many biomedical and statistical academic
journals
Institute library
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 248
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Laboratory of
HEALTH ECONOMETRICS AND
HEALTH DEMAND
Location
Staff
Interuniversity Research Centre on Public Services
University of Milano-Bicocca
Viale dell’Innovazione 10, 20126 Milano Itally.
U9 Buikding 2nd floor
Scientific responsibles
Mario Mezzanzanica (associate professor of Informative
Systems at the University of Milano-Bicocca)
☛ [email protected]
Giorgio Vittadini (full professor of Statistics at the University
of Milano-Bicocca)
☛ [email protected]
Research staff
Piergiorgio Lovaglio (Associate Professor at the University of
Milano-Bicocca);
☛ [email protected]
Mirko Cesarini (Assistant Professor at the University of
Milano-Bicocca)
☛ [email protected]
Roberto Boselli (Assistant Professor at the University of
Milano-Bicocca)
☛[email protected]
Fulvia Pennoni (Assistant Professor at the University of
Milano-Bicocca)
☛ [email protected]
External Collaboration Staff
Isabella Romeo (Research Fellow at the University of MilanoBicocca);
Laura Mariani (Research Fellow at the University of MilanoBicocca);
Claudia Graziani (Temporary Researcher at the University of
Milano-Bicocca); senior analyst
Gloria Ronzoni (Temporary Researcher the University of
Milano-Bicocca); senior analyst
Silvia Dusi (Temporary Researcher at the University of
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 249
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Milano-Bicocca); junior analyst
Giorgio Colombo (Temporary Researcher the University of
Milano-Bicocca); junior analyst
Technical staff
Paolo Berta (Senior consultant – statistical analysis – at the
University of Milano-Bicocca)
Matteo Fontana (Senior computer engineer at the University
of Milano-Bicocca)
Administrative staff
Lucia Valsecchi (Administrative manager at the University of
Milano-Bicocca)
☛ [email protected]
Paolo Savino (Accountant at the University of MilanoBicocca)
☛ [email protected]
Computing
environments
Software access
4 servers- Virtual Enviroment – Vmware based
2 smart array storages
30 notebook (Windows, Linux and MacOs)
Monitors, scanners, printers
Statistical packages (SAS (SAS Forecast, SAS Miner, SAS
Social Media, SAS Analitycs), SPSS, R, OXmetrics)
MySQL, Pentaho – BI Open source Platform, Talend (ETL
Open Source), SAS Visual Analitycs
Touchgraph Navigator
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 250
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Laboratory of
PHARMACOVIGILANCE AND DRUG UTILISATION
RESEARCH
Location
Staff
Pharmacovigilance Regional Center of Lombardy
Direzione Generale Sanità
Regione Lombardia
Via Rosellini 17, 20126 Milano, Italy
Head of the Pharmacovigilance Regional Centre
Alma Lisa Rivolta (Director)
☛ [email protected] ✈ 02.67653348
Research staff
Alfredo Cocci, Farmacista
☛ [email protected] ✈ 02 67654455
Valentino Conti, Biostatistico
☛ [email protected] ✈ 02 67653641
Olivia Leoni, MD
☛ [email protected] ✈ 02 67658830
Lucrezia Magistro, MD
☛ [email protected] ✈ 02 67653885
Giuseppe Monaco, MD
☛ [email protected] ✈ 02 67653844
Stefania Scotto, MD
☛ [email protected] ✈ 02 67658829
Mauro Venegoni, MD
☛ [email protected] ✈ 02 67658815
Giuseppe Vighi, MD
☛ [email protected] ✈ 02 67655094
Administrative staff
Daniela Beretta
☛ [email protected]
Computing
environments
Software access
Other
10 Windows PCs (32 bit)
Several statistical analysis packages (SAS, STATA, SPSS)
Microsoft Office
Web access to many biomedical and statistical academic
journals
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 251
CRACK
Dept of Statistics & Quantitative Methods
Unit of Biostatistics Epidemiology & Public Health
program
Laboratory of Pharmacoepidemiology & Healthcare Research
Access to University libraries
Il documento è sottoposto ad aggiornamento periodico nei suoi contenuti programmatici e nella descrizione
dei progetti in corso. Autore Giovanni Corrao - 252