Mini-Sentinel: Analytic Methods for Using Laboratory Test Results in Active Database Surveillance

The objective of this project was to select and test statistical methods appropriate for use when analyzing cross-sectional and longitudinal observational healthcare administrative, claims, and clinical data, with specific attention to clinical laboratory results when missing data are expected. Recognizing that different uses of laboratory results data (e.g., confounding adjustment, detecting outcomes, cohort identification) in medical product safety surveillance could optimally employ different methods to handle the missing results data, we evaluated more than one method for use in the Mini-Sentinel Distributed Database environment.

To select and evaluate the performance of statistical methods with intent to potentially recommend techniques for including laboratory test results data in Mini-Sentinel safety surveillance, we had the following Specific Aims:

• Specific Aim 1. Summarize the literature on the use of clinical laboratory test results databases in administrative claims and electronic health records (EHR) database studies of medical product safety.
• Specific Aim 2. Utilize the Mini-Sentinel Distributed Database resources to evaluate analytic methods for incorporating laboratory test results data into medical product safety analyses.
• Specific Aim 3. Develop recommendations and detailed specifications for incorporating laboratory results table data into Mini-Sentinel safety analyses.

Please contact the Sentinel Operations Center if you are interested in acquring a copy of this project's accompanying Access Database.