Share this article on:

Commentary: The Growing Role of Epidemiology in Drug Safety Regulation

Hamburg, Margaret A.

doi: 10.1097/EDE.0b013e31822837ba
Comparative Effectiveness Research

From the US Food and Drug Administration, Silver Spring, MD.

Correspondence: Jeff Ventura, Room 5253, Building 32, 10903 New Hampshire Ave, Silver Spring, MD 20993. E-mail:

For the US Food and Drug Administration (FDA), having the capacity to monitor drug safety in the postmarket environment under conditions of actual use is as essential to ensuring patient safety as evaluating initial clinical trial data in the premarket setting. The increased availability of electronic health care data represents an opportunity to further develop this capacity. The role of FDA's epidemiologists, alongside other experts in pharmacovigilance, statistics, and clinical medicine, is critical in using these data for detecting, quantifying, and characterizing these serious drug safety concerns. Understanding the role of epidemiology, FDA has increased its investment in, and reliance on, this discipline over the past decade.

However, using epidemiology to examine and apply these data to drug safety is challenging. FDA must have enough epidemiologists who possess the skills to analyze, interpret, and leverage postmarket data to complement the rich clinical-trial information existing within the Agency.

In the context of premarket review of human drugs, FDA relies on extensive preclinical testing, human pharmacology studies, and clinical trials. The clinical trials, which are designed primarily to demonstrate efficacy, are also able to characterize and quantify the most common, but not necessarily the most serious or clinically important, adverse effects of a medicine. For over 4 decades, FDA has relied on spontaneous reports of adverse events from healthcare professionals, patients, and consumers to identify previously unknown risks of medicines after a drug is brought to market. Although such reports are important in the identification of rare, serious, drug-induced events, they are not sufficient to understand more subtle risks of a medicine, especially when those risks are related to clinical events that are common in the population or a manifestation of the underlying disease being treated.

To address these challenges, FDA has strengthened its epidemiology program. Through epidemiology, FDA quantifies adverse effects of medicines in real-world situations. For example, using population-based administrative healthcare claims databases, FDA epidemiologists have shown that use of cerivastatin results in hospitalized cases of rhabdomyolysis more frequently than other statins, and that this risk increases dramatically in the presence of fibrates.1 Although spontaneous reports identified rhabdomyolysis as a drug-induced effect,2 they were not capable of quantifying this rare risk. Calculating an incidence rate for this risk using spontaneous reporting of adverse drug events is impossible, given the voluntary nature of these reports and the fact that underreporting is a well-known phenomenon.3

Similarly, FDA epidemiologists have used population-based data to demonstrate that osteonecrosis of the jaw occurs very infrequently in users of oral bisphosphonates4 and to compare the development of adverse cardiovascular effects in users of rosiglitazone and pioglitazone.5

In addition to quantifying risks, FDA epidemiologists are also interested in understanding how medicines are used. The Agency understands that postmarketing drug safety involves characterizing the specific risks of a medicine in the context of actual use. For instance, FDA epidemiologists have demonstrated that 13% of metoclopramide users take that medicine for longer than the recommended 12-week course of treatment.6 Because longer use of metoclopramide is associated with increased risk for tardive dyskinesia, new labeled warnings were issued based on these findings.

FDA epidemiologists have used similar epidemiologic data to monitor the impact and efficacy of our regulatory actions. For instance, the frequency of use of cisapride with interacting drugs that increase its blood level was measured before and after the Agency warned against this type of use. It was determined that the FDA warnings had little impact on preventing the contraindicated concomitant use,7 a finding that led to the withdrawal of cisapride from the US market. In another example, topical calcineurin inhibitors were approved for the treatment of atopic dermatitis only for children more than 2 years of age. FDA epidemiologists were able to use large prescription claims databases to identify considerable use of these products in children less than 2 years of age—a population in which their safe and effective use had not been studied. After the Agency took regulatory actions, the data were examined again to assure that the use in this population had decreased.8

Epidemiologic analysis of certain public health databases has helped in targeting interventions that are part of FDA risk-reduction measures, such as Risk Evaluation and Mitigation Strategies. For example, FDA epidemiologists have used data from the National Survey on Drug Use and Health and the Drug Abuse Warning Network (a national drug surveillance system) to help understand patterns of opioid use, abuse, and misuse. This information is being used to develop an evaluation and intervention program for long-acting and sustained-release opioid products. As the use of risk evaluation and mitigation strategies increases, so too will FDA's use of epidemiologic analyses to determine the effectiveness of these programs.

Under the FDA Amendments Act, FDA may, when certain conditions are met, require drug manufacturers to conduct studies of the safety of their medicines. Many of these postmarketing requirements will be epidemiologic studies. In these cases, FDA epidemiologists work to develop the specific postmarketing study requirements, review the study protocol and final study results, and participate with other FDA staff in regulatory decision making.

These and other applications of epidemiology at FDA require high-quality, population-based data with highly specified standardized data structures and terminologies (such as the Clinical Data Interchange Standards Consortium's Study Data Tabulation Model and Analysis Data Model) as well as the breadth and depth of scientific expertise to understand, analyze, and interpret the data. Thus, much of the epidemiologic work at FDA is performed in collaboration with other federal agencies (Veterans Affairs, Department of Defense, Centers for Medicare and Medicaid Services, Agency for Healthcare Research and Quality, and Centers for Disease Control and Prevention), as well as with nonfederal partners. Through various contracts, memoranda of understanding, and interagency agreements, FDA has built multiple intragovernmental and public-private partnerships for conducting epidemiologic safety studies. FDA has recently issued for public comment a draft guidance on best practices for conducting and reporting epidemiologic studies of drug safety that use electronic healthcare data.

Through similar partnerships, FDA is also developing innovative methods for postmarket drug safety surveillance through the Sentinel Initiative. This initiative seeks to use healthcare data from multiple sources (federal and private payer claims databases and de-identified patient records) to assess drug safety signals in real time.9 The Observational Medical Outcomes Partnership—a public-private partnership designed to help improve the monitoring of drugs for safety—has done some fundamental work in identifying methodologic challenges in making inferences about drug safety by using real-time epidemiologic analyses.10

FDA also recently awarded a contract to researchers at The Johns Hopkins School of Medicine to advance patient-centered outcomes research. Using clinical data converted to standard structure and terminologies as well as data from other sources, the researchers will work with FDA scientists to define questions and analytic approaches to learn how we might determine potential strategies to improve health outcomes. Combining newly developed analytic tools and methodologies with a deeper understanding of standardized data and clinical interventions has the potential to better inform regulatory review and decision-making processes for marketed drugs.

Tools, technologies, and standards, however, are only half of the equation. FDA encourages its academic stakeholders to continue to produce highly trained and motivated pharmacoepidemiologists who possess the ability—and imagination—to use these new systems and technologies to maximize the power of the agency's information.

Using epidemiology to evaluate the vast amount of electronic healthcare data now available in the United States represents an opportunity for FDA to substantially improve medical product safety and regulatory decision making. With this greater understanding of how drugs perform once they are on the market and in wider use, FDA may advance a regulatory path that takes full account of real-world experiences to inform our benefit/risk decision making and that better meets the medical needs of diverse populations.

Back to Top | Article Outline


MARGARET A. HAMBURG is the Commissioner of the US Food and Drug Administration. Dr. Hamburg earned her MD from Harvard and conducted neuroscience research at Rockefeller University. She later focused on AIDS research as Assistant Director of the National Institute of Allergy and Infectious Diseases. She served 6 years as the New York City Commissioner of Health, where her aggressive approach to the city's tuberculosis epidemic led to an 86% decline in drug-resistant TB.

Back to Top | Article Outline


1.Graham DJ, Staffa JA, Shatin D, et al. Incidence of hospitalized rhabdomyolysis in patients treated with lipid-lowering drugs. JAMA. 2004;292:2585–2590.
2.Chang JT, Staffa JA, Parks M, Green L. Rhabdomyolysis with HMG-CoA reductase inhibitors and gemfibrozil combination therapy. Pharmacoepidemiol Drug Saf. 2004;13:417–426.
3.McAdams M, Staffa JA, Dal Pan GJ. Estimating the extent of reporting to FDA: a case study of statin-associated rhabdomyolysis. Pharmacoepidemiol Drug Saf. 2008;17:229–239.
4.Lo JC, O'Ryan FS, Gordon NP, et al. Predicting Risk of Osteonecrosis of the Jaw with Oral Bisphosphonate Exposure (PROBE) Investigators. Prevalence of osteonecrosis of the jaw in patients with oral bisphosphonate exposure. J Oral Maxillofac Surg. 2010;68:243–253.
5.Graham DJ, Ouellet-Hellstrom R, MaCurdy TE, et al. Risk of acute myocardial infarction, stroke, heart failure, and death in elderly Medicare patients treated with rosiglitazone or pioglitazone. JAMA. 2010;304:411–418.
6.Kaplan S, Staffa JA, Dal Pan GJ. Duration of therapy with metoclopramide: a prescription claims data study. Pharmacoepidemiol Drug Saf. 2007;16:878–881.
7.Smalley W, Shatin D, Wysowski DK, et al. Contraindicated use of cisapride: impact of Food and Drug Administration regulatory action. JAMA. 2000;284:3036–3039.
8.Moeny D, La Grenade L, Brinker A. Changes in topical calcineurin use following regulatory actions in the US. Pharmacoepidemiol Drug Saf. 2009;18(suppl 1):S180.
9.Behrman RE, Benner JS, Brown JS, McClellan M, Woodcock J, Platt R. Developing the Sentinel System—a national resource for evidence development. N Engl J Med. 2011;364:498–499.
10.Stang PE, Ryan PB, Racoosin JA, et al. Advancing the science for active surveillance: rationale and design for the Observational Medical Outcomes Partnership. Ann Intern Med. 2010;153:600–606.
© 2011 Lippincott Williams & Wilkins, Inc.