Using Clinical Databases to Monitor Drug Side Effects and Performance Interview with:
Leo Anthony Celi, MD, MS, MPH Massachusetts Institute of Technology Cambridge, MA 02139
Leo Anthony Celi, MD, MS, MPH

Massachusetts Institute of Technology
Cambridge, MA 02139 What are the main findings of the study?

Dr. Celi: The main take home point from the paper is that we know little about how drug perform in the real world. Which patients truly benefit? Which patients are harmed? How do drugs interact with different acute (such as critical illness) and chronic conditions? These questions are almost never answered during pre-marketing research due to cost. We need a better system of following the life cycle of drugs post-marketing. Clinical databases provide us with this opportunity. Were any of the findings unexpected?

Dr. Celi: Note that the findings of the study would need to be validated in other databases given that the analysis was only performed using a clinical database from a single center. Retrospective analyses are quite complicated to perform. A significant caveat is that the analysis may be marred by residual confounding and bias by indication. The goal is to identify patient records in the database that are as similar as possible to the patient in terms of the variables that can confound the relationship between the drug and the outcome, and then to compare the outcomes of those who receive the drug versus those who did not. Residual confounding means that the outcome difference might not be due to the drug but rather due to something inherent to those patients who receive the drug, or their condition. What should clinicians and patients take away from your report?

Dr. Celi: Obtaining new information about a drug’s benefits and risks should not stop after market authorization. In recent years, highly-publicized post-marketing crises like those of Vioxx and Avandia have raised FDA’s awareness of the shortcomings of the Adverse Event Reporting System (AERS), a passive surveillance system for collecting spontaneous reports of drug safety by providers and researchers. We propose modernizing and extending this surveillance system through systematic aggregation of safety and efficacy information. What recommendations do you have for future research as a result of this study?

Dr. Celi: Electronic Medical Records that include detailed clinical information provide researchers an opportunity to accumulate safety and efficacy evidence, to discover patient subpopulations that experience increased efficacy or unanticipated delayed adverse effects, and to uncover interactions between simultaneous treatments as drugs become used in wider, more diverse patient populations than those possible during pre-market approval clinical studies. We recommend the FDA partner with representative clinics and hospitals across the country for access to their clinical databases, to better represent minority and vulnerable patient populations. Such an active national surveillance system would allow drugs to be monitored longitudinally over the entire market life, providing the FDA timely access to new information for evaluating a drug’s benefit-risk profile.


SSRI use prior to ICU admission is associated with increased hospital mortality

Ghassemi M, Marshall J, Singh N, Stone DJ, Celi L.
Ssri Use Prior To Icu Admission Is Associated With Increased Hospital Mortality
Published online December 26, 2013.