Medicare administrative claims data offer a valuable informational source for the surveillance of adverse drug events, but there are substantial challenges in exploiting these data to screen for potential patient safety problems. The merits of these data include large samples covering the spectrum of the elderly and disabled and data accessible at relatively low costs. Medicare claims provide a broad picture of the health services received by individuals, with considerable diagnosis information recorded across many provider types. As Medicare Part D records become available, the claims data will also incorporate the use of pharmaceuticals. In sharp contrast to clinical trials and experimental data, one can draw on claims data to assess potential safety problems on a rapid turnaround basis, investigating both current and recent past experiences to uncover adverse outcomes. There are two central challenges in using claims data for the surveillance of drug safety:
- The development of frameworks to organize claims data into structures that link relevant interventions, outcomes and patient characteristics for the study of pharmaceuticals under investigation; and
- The formulation of statistical approaches to analyze these data to detect adverse events.
This project will develop a data system and empirical framework for identifying and capturing adverse drug events using elements contained in the Medicare claims files. The analysis will exploit information available in Part B claims on drugs, to produce a framework that can be readily adapted to incorporate Part D claims when they come on line next year.