FDA Issues First Sentinel Initiative Project Report: Statistical Methods

Published on: 

ePT--the Electronic Newsletter of Pharmaceutical Technology

The first project report on FDA's Sentinel website, discusses the evaluation of existing methods for safety signal identification.

Just one year ago, in May 2008, the US Food and Drug Administration launched the Sentinel Initiative to transform the agency’s ability to track the safety of drugs, biologics, and medical devices once they reach the market. The national electronic system is meant to complement existing FDA systems for tracking adverse-event reports. The agency is now beginning to issue project reports about how to develop and implement the system, which will be completed in stages. The reports will be available on the agency’s Sentinel website and under docket number FDA-2009-N-0192. The first project report discusses the evaluation of existing methods for safety signal identification.

Ideally, the Sentinel system will allow FDA to “query large participating data sources quickly and securely for relevant product safety information,” according to an agency notice in the May 7, 2009, Federal Register. “FDA would send questions to participating data holders, who in turn would, in accordance with existing privacy and security safeguards, evaluate their data and send summary results to FDA for agency review.” The system will facilitate the development of active surveillance methodologies related to signal detection, signal strengthening, and signal validation. Eight contracts have been awarded through the Sentinel Initiative to help in this process.

The first contract project to be released identifies, describes, and evaluates signal-detection methods that could be used in the Sentinel program. The report analyzes the signal’s use for: postmarket surveillance; sensitivity, or ability to appropriately identify important safety signals; specificity, or ability to appropriately identify the absence of important safety signals; ability to minimize both false-positive and false-negative safety signals; robustness of the methodology to perform across different types of data systems for detection of different types of safety signals; flexibility of the methodology to adapt to different situations and to change thresholds for different safety issues as desired, and barriers that exist to using each identified methodology.

The goal of the report, according to the website, is to make further recommendations for signal-detection methods. Comments on the overall docket availability or project report can be submitted to the FDA Division of Dockets Management online.