HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:
Sign In to gain access to subscriptions and/or personal tools.

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Request Reprints Citing Articles Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Etminan, M. Articles by Samii, A. Search for Related Content PubMed PubMed Citation Articles by Etminan, M. Articles by Samii, A.

Challenges and Opportunities for Pharmacoepidemiology in Drug-Therapy Decision Making

From the Division of Clinical Epidemiology, Royal Victoria Hospital and McGill Health Center, Montreal, Canada (Dr Etminan); the Center for Clinical Epidemiology and Evaluation, Vancouver Hospital, Vancouver, Canada (Dr Etminan, Mr FitzGerald); Department of Medicine, Queen's University, Kingston, Ontario, Canada (Dr Gill); Faculty of Medicine, University of British Columbia, Vancouver, Canada (Mr FitzGerald); and the Department of Neurology, University of Washington School of Medicine, Seattle, Washington, (Dr Samii).

Address for reprints: Dr Mahyar Etminan, Center for Clinical Epidemiology and Evaluation, Vancouver Hosptial, 7th Floor, 828 West 10th Avenue, Vancouver, B.C., Canada (metminan{at}shaw.ca).

	ABSTRACT

TOP ABSTRACT IMPACT OF PHARMACOEPIDEMIOLOGIC... ADVANCES IN PHARMACOEPIDEMIOLOGY DISCORDANCE AMONG... IMPROVING METHODS OF REPORTING... REFERENCES

 
Pharmacoepidemiology is a relatively new and evolving science that attempts to quantify mainly adverse drug events and patterns of drug use in a large population. The strength of pharmacoepidemiology over randomized trials is the ability to quantify rare adverse events that may occur over long periods. Recently, discordance in the results of pharmacoepidemiologic studies has made it difficult for clinicians and policy makers to make informed drug-therapy decisions. This commentary addresses the strength of pharmacoepidemiology and the advances in the methodology of pharmacoepidemiologic studies over the years. We also discuss the potential problem of discordant results and urge pharmacoepidemiologists to develop good practice guidelines for the conduct of pharmacoepidemiologic studies.

Key Words: Pharmacoepidemiology • drug safety • case-control studies • cohort studies

	IMPACT OF PHARMACOEPIDEMIOLOGIC STUDIES ON CLINICAL PRACTICE

TOP ABSTRACT IMPACT OF PHARMACOEPIDEMIOLOGIC... ADVANCES IN PHARMACOEPIDEMIOLOGY DISCORDANCE AMONG... IMPROVING METHODS OF REPORTING... REFERENCES

 
In recent years, advances in pharmacoepidemiology have helped bring to attention serious ADEs secondary to different drug classes. Perhaps one of the first pharmacoepidemiologic studies that led to the withdrawal of a particular drug class was a case-control study linking appetite-suppressant drugs to cardiac-valve regurgitation.¹ The Food and Drug Administration recently asked makers of the new atypical antipsychotic drugs (risperidone, olanzapine, and quetiapine) to put warning labels on their products as a result of another case-control study that found an increase in the risk of diabetes in users of olanzapine.² Finally, observational studies played an important role in highlighting the increased risk of cardiac events with rofecoxib.

One of the limitations of RCTs is their failure to address adherence to drug therapy, mostly because of the controlled environment inherent in the nature of the design of such studies. Availability of large administrative databases has allowed clinicians to explore adherence to drug therapy in a real clinical setting. One example was the degree of adherence to statin therapy in older adults. Despite numerous RCTs showing the efficacy of these drugs in preventing cardiac events and their relatively safe adverse events profile, data on adherence of these drugs were lacking. A recent cohort study using the Ontario linked databases showed that adherence to statins among older adults is relatively low.⁴

	ADVANCES IN PHARMACOEPIDEMIOLOGY

TOP ABSTRACT IMPACT OF PHARMACOEPIDEMIOLOGIC... ADVANCES IN PHARMACOEPIDEMIOLOGY DISCORDANCE AMONG... IMPROVING METHODS OF REPORTING... REFERENCES

 
Pharmacoepidemiology has come a long way since the early 1980s when reserpine was linked to an increased risk of breast cancer based on a poorly conducted hospital-based case-control study in which inappropriate selection of controls gave rise to a spuriously high odds ratio.⁵ In the past 20 years, advances in both epidemiology and biostatistics have allowed for novel methods of controlling for confounding and bias as well as addressing methods of quantifying ADEs with an irregular risk pattern. For example, optimal methods of control selection have allowed for conducting more valid case-control studies.⁶ Sophisticated computer software has allowed for complex statistical analyses that control for the time-dependent nature of prescription drugs that may have otherwise biased the results of cohort studies.⁷ Use of propensity scores, a method of predicting the probability of a subject's receiving a drug based on the baseline covariate information for that subject, has to some extent mimicked the process of randomization in retrospective cohort studies.⁸ Finally, the case-crossover design, a relatively novel study design, has allowed for exploring the association between transient drug exposures with acute events.⁹

	DISCORDANCE AMONG PHARMACOEPIDEMIOLOGIC STUDIES

TOP ABSTRACT IMPACT OF PHARMACOEPIDEMIOLOGIC... ADVANCES IN PHARMACOEPIDEMIOLOGY DISCORDANCE AMONG... IMPROVING METHODS OF REPORTING... REFERENCES

 
What has probably undermined the validity of pharmacoepidemiologic studies the most has been discordance in drug efficacy among studies addressing the same question. Although the first study that hinted at the potential increase in the risk of myocardial infarction (MI) associated with the use of rofecoxib, other pharmacoepidemiologic studies failed to find this association. Ray et al conducted a retrospective cohort study using the Tennessee Medicaid Programme.³ The authors found an increase in the risk of MI with high-dose rofecoxib use (25 mg or greater) but not with low-dose rofecoxib use (less than 25 mg). Mamdani et al used the Ontario linked databases to answer the same questions.¹⁰ The authors did not find any increase in the risk of MI with regular rofecoxib use. The discrepancy may lie in the method of defining exposure in the 2 studies. Ray et al stratified exposure by dose (low vs high), whereas Mamdani et al looked at only current use. This discrepancy may also be the reason a pooled analysis of data from RCTs of rofecoxib failed to find an association.¹¹ Given that ADEs may be dose dependent, Mamdani et al may have missed this association. This is a classic example in pharmacoepidemiology, whereby inadequate assessment of drug dose and duration may compromise validity of a drug safety study.

Two relatively recent examples hypothesized that statins can lower the risk of fractures in older adults.^12,13 Both studies were published in JAMA, used a case-control design, and also used the General Practice Research Databases as their primary source of data. Surprisingly, one study found a protective association,¹² whereas the other found no association.¹³ Finally, a recent cohort study showed a protective association with the use of inhaled corticosteroids in preventing morbidity and mortality in patients with chronic obstructive pulmonary disease.¹⁴ The results of this study were soon contradicted and accredited to a bias that may have arisen in the study design.¹⁵

An important issue is whether the discordance in pharmacoepidemiologic studies is attributable mainly to the limitations of the science or to the poor quality in conducting and reporting pharmacoepidemiologic studies. It is now evident that even RCTs are not immune to potentially biased results when robust methodology is not incorporated in either the study design or the reporting of the results. The Woman's Health Initiative, the largest RCT that looked at the effect of estrogen on cardiovascular disease and cancer in postmenopausal women, has been recently criticized for possible bias. This bias is thought to have been introduced as a result of the unblinding of approximately 40.5% of hormone replacement users and 7% of placebo users.¹⁶ At least part of the inconsistency in pharmacoepidemiologic literature may be attributed to the lack of standards in the conducting and the reporting of pharmacoepidemiologic studies.

We agree that for drug-related questions that can be addressed by RCTs, clinicians must use results of pharmacoepidemiologic studies only for hypothesis generating. However, RCTs cannot answer all important drug-related questions including rare ADEs or the effect of prescription drugs on the risk of motor vehicle crashes.¹⁷ In such circumstances, clinicians may have no choice but to incorporate results of pharmacoepidemiologic studies into their clinical decision making.

	IMPROVING METHODS OF REPORTING FOR PHARMACOEPIDEMIOLOGIC STUDIES

TOP ABSTRACT IMPACT OF PHARMACOEPIDEMIOLOGIC... ADVANCES IN PHARMACOEPIDEMIOLOGY DISCORDANCE AMONG... IMPROVING METHODS OF REPORTING... REFERENCES

 
We believe that methods of reporting for pharmacoepidemiologic studies must be improved, and this initiative must be taken by medical journals that frequently publish these studies. A group of epidemiologists based in the United Kingdom is developing guidelines for methods reporting observational studies. The Standards for the Reporting of Observational Studies in Epidemiology (STROBE)¹⁸ is meant to set standards for the reporting of observational studies (cohort studies, case-control studies, cross-sectional studies) in the hopes that STROBE like Consolidated Standards of Reporting Trials (CONSORT) may improve the quality of the reporting of observational studies.

We believe that standards for the reporting of pharmacoepidemiologic studies are necessary. But these standards must be specific to pharmacoepidemiology as opposed to observational studies as a whole. For example, STROBE does not address reporting of the more complex epidemiologic study designs such as the case-crossover⁹ and case-time-control¹⁹ studies, both of which have played an integral role in exploring ADEs. A case-crossover study is essentially a case-control study in which the cases act as their own controls. Because the cases act as their own controls, the degree of between-person confounding that may be present in a classic case-control study (in which controls are simply those who do not have the disease) is minimized. This design is most suitable for situations in which a certain drug is used transiently (for example, a benzodiazepine that is used on an as-needed basis) and in which the risk of an adverse event associated with that drug is not constant but rises sharply and then falls over time. An example of this design is a relatively recent study that explored the use of short-acting benzodiazepines and risk of car accidents.²⁰ The case-time-control study is similar to a case-crossover study in which a set of controls are used to adjust for any temporal changes that may affect prescription drug use.²¹ Both these designs are relatively new and require methodological and analytical considerations that may be different from those in classic cohort and case-control studies.

Pharmacoepidemiology is emerging as a subspecialty within epidemiology, mostly because of the need for monitoring prescription drug exposures in large populations over time. Differences in prescribing behavior, comorbidity, and limitations of large administrative databases have identified biases that are either unique to pharmacoepidemiologic studies or may present themselves more frequently in these studies. One example is confounding by indication, which may occur if the outcome thought to be associated with a specific drug may actually be the result of the disease or condition in question. Another example is a relatively new bias referred to as immortal time bias in cohort studies, which arises when the time dependency of prescription drug use in a large cohort is not adequately controlled for.⁷

Pharmacoepidemiology will continue to play a major role in drug-therapy decision making. We urge pharmacoepidemiologists to develop stricter standards in conducting and reporting pharmacoepidemiologic studies so that results of these studies may help clinicians make more informed drug-therapy decisions.

DOI: 10.1177/0091270005283285

	REFERENCES

TOP ABSTRACT IMPACT OF PHARMACOEPIDEMIOLOGIC... ADVANCES IN PHARMACOEPIDEMIOLOGY DISCORDANCE AMONG... IMPROVING METHODS OF REPORTING... REFERENCES

 

1. Jick H, Vasilakis C, Weinrauch LA, Meier CR, Jick SS, Derby LE. A population-based study of appetite-suppressant drugs and the risk of cardiac-valve regurgitation. New Engl J Med. 1998;339: 719-724.[Abstract/Free Full Text]

2. Koro CE, Fedder DO, L'Italien GJ, et al. Assessment of independent effect of olanzapine and risperidone on risk of diabetes among patients with schizophrenia: population-based nested case-control study. BMJ. 2002;325: 243.[Abstract/Free Full Text]

3. Ray WA, Stein CM, Daugherty JR, Hall K, Arbogast PG, Griffin MR. COX-2 selective non-steroidal anti-inflammatory drugs and risk of serious coronary heart disease. Lancet. 2004;360: 1071-1073.

4. Jackevicius CA, Mamdani M, Tu JV. Adherence with statin therapy in elderly patients with and without acute coronary syndromes. JAMA. 2002;288: 462-467.[Abstract/Free Full Text]

5. Horwitz RI, Feinstein AR. Exclusion bias and the false relationship of reserpine and breast cancer. Arch Intern Med. 1985;145: 1873-1875.[Abstract]

6. Lubin JH, Gail MH. Biased selection of controls for case-control analyses of cohort studies. Biometrics. 1984;40: 63-75.[CrossRef][ISI][Medline] [Order article via Infotrieve]

7. Suissa S. Effectiveness of inhaled corticosteroids in chronic obstructive pulmonary disease: immortal time bias in observational studies. Am J Respir Crit Care Med. 2003;168: 49-53.[Abstract/Free Full Text]

8. Wang J, Donnan P. Propensity score methods in drug safety studies: practice, strengths and limitations. Pharmacoepidemiol Drug Saf. 2001;10: 341-344.[CrossRef][ISI][Medline] [Order article via Infotrieve]

9. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991;133: 144-153.[Abstract/Free Full Text]

10. Mamdani M, Rochon P, Juurlink DN, et al. Effect of selective cyclooxygenase 2 inhibitors and naproxen on short-term risk of acute myocardial infarction in the elderly. Arch Intern Med. 2003;163: 481-486.[Abstract/Free Full Text]

11. Konstam MA, Weir MR, Reicin A, et al. Cardiovascular thrombotic events in controlled, clinical trials of rofecoxib. Circulation. 2001;104: 2280-2288.[Abstract/Free Full Text]

12. Meier CR, Schlienger RG, Kraenzlin ME, Schlegel B, Jick H. HMG-CoA reductase inhibitors and the risk of fractures. JAMA. 2000;283: 3205-3210.[Abstract/Free Full Text]

13. van Staa TP, Wegman S, de Vries F, Leufkens B, Cooper C. Use of statins and risk of fractures. JAMA. 2001;285: 1850-1855.[Abstract/Free Full Text]

14. Sin DD, Tu J. Inhaled corticosteroids and the risk of mortality and readmission in elderly patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2001;164: 580-584.[Abstract/Free Full Text]

15. Fan VS, Bryson CL, Curtis JR, et al. Inhaled corticosteroids in chronic obstructive pulmonary disease and risk of death and hospitalization: time-dependent analysis. Am J Respir Care Med. 2003:168: 1488-1494.[Abstract/Free Full Text]

16. Garbe E, Suissa S. Hormone replacement therapy and acute coronary outcomes: methodological issues between randomized and observational. Maturitas. 2004;19: 8-13.

17. Etminan M, Hemmelgarn B, Delaney JAC, Suissa S. Lithium use and the risk of injurious motor vehicle crashes in older adults: a case-control study. BMJ. 2004;328: 558-559.[Free Full Text]

18. Von Elm E, Egger M. The scandal of poor epidemiological research. BMJ. 2004;329: 868-869.[Free Full Text]

19. Suissa S. The case-time-control design. Epidemiology. 1995;6: 248-253.[ISI][Medline] [Order article via Infotrieve]

20. Barbone F, McMahon AD, Davey PG, et al. Association of road-traffic accidents with benzodiazepine use. Lancet. 1998;352: 1331-1336.[CrossRef][ISI][Medline] [Order article via Infotrieve]

21. Suissa S, Edwardes MD, Boivin JF. External comparisons from nested case-control designs. Epidemiology. 1998;9: 72-78.[ISI][Medline] [Order article via Infotrieve]

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Request Reprints Citing Articles Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Etminan, M. Articles by Samii, A. Search for Related Content PubMed PubMed Citation Articles by Etminan, M. Articles by Samii, A.