Safety
Appendix D
Table of Contents: Managing the Risks From Medical Product Use: Creating a Risk Management Framework
To improve methodologies for evaluation of computerized medical product safety surveillance data, FDA's centers have an opportunity to adapt proactive screening methods from other disciplines and settings (e.g., biostatistics, quality improvement). These methods may allow more extensive and efficient monitoring of postmarketing adverse event reports and other data. Such techniques as exploratory data analysis, control charts, and time series modeling contrasts with the reactive mode, in which FDA responds to an inquiry or other external action by evaluating adverse event reports.
Analysis of Premarketing Data
Each center is striving to improve the analysis of premarketing safety data to focus the efforts of postmarketing reviewers. This knowledge transfer is becoming more streamlined and efficient. Residual safety concerns are identified (e.g., elevated liver enzymes) so that postmarketing ADR reports with similar syndromes (e.g., liver failure) will receive greater scrutiny. In this way, the complete safety profile of the product can be based on both pre- and postmarketing safety data.
Information Systems
The growth and development of computerized information systems supporting medical practice in well-defined populations will play an important role in drug safety surveillance as pharmacoepidemiological research systems transcend the limitations of passive surveillance from spontaneously submitted case reports. They come close to achieving active surveillance and greatly enhance the capability of objective epidemiological analyses. One new area may be the generation of signals from these large linked databases, which has been controversial and has had very limited success. Recent developments suggest promise for proactive systematic signal searching, although its complete role in generating an early alert has not been delineated. Systems featuring consistent ascertainment of outcome events may also be valuable for screening well-defined populations to evaluate moderate risks for relatively common or long-latency diseases impossible to distinguish from background noise in the spontaneous surveillance systems.
Another role for large information systems in pharmacoepidemiology is to help understand reporting rates by developing background incidence rates for diseases or syndromes in a population. For the relatively modest cost of conducting additional detailed chart reviews, product-related studies of a given syndrome could be extended to the descriptive epidemiology of the condition in the general population.
