Joint Matching on Propensity Scores and Disease Risk Scores in Observational Studies of Drug Effects

• Activity Number: 156
• Type: Topic Contributed
• Date/Time: Monday, August 2, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Robert Glynn*+
• Companies: Brigham and Women's Hospital
• Keywords: matching ; propensity score ; disease risk score ; cohort study

Abstract

Observational studies based on large, administrative databases can provide real-world evaluation of the benefits and risks of medications in individuals with characteristics that frequently are exclusions in clinical trials. In analyses of such studies, both indications for and barriers to treatment can confound relationships between drug use and outcomes and these variables can influence treatment choices and disease risk in different ways. Moreover, the propensity to treat and the risk of the outcome are two important dimensions for consideration of modification of treatment effects. Matching on both disease risk and propensity to treat can provide a straightforward approach both to control confounding and to allow evaluation of potential effect modification along either dimension. We illustrate these issues with examples from pharmacoepidemiologic studies in older individuals.