Considerations on Sampling for Medical Chart Review

• Activity Number: 663
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Haijun Ma*+ and Jason Legg and Jaekwang Kim
• Companies: Amgen, Inc. and Amgen, Inc. and Iowa State University
• Keywords: medical chart review ; fractional imputation ; matching ; healthcare data

Abstract

In pharmacovigilance studies using healthcare data, outcomes are often identified using algorithms based on diagnosis codes and further confirmed through medical chart review, especially when the diagnosis codes lack specificity for a desired condition. For studies with long-term follow-up of a large population, the cost and efforts can be prohibitive to confirm all algorithm-identified potential cases through medical chart review. Sampling designs and estimators need to be developed to ensure cost efficiency while providing consistent estimators of treatment effects. A commonly used approach is to identify matched sets and only review potential cases in the matched sets. In this talk we will discuss an approach of random sampling and fractional imputation to tackle such problems. A simulation study will be used to illustrate the advantages of this approach versus the traditional matching approach. Assumptions behind the two approaches will also be discussed.