Modeling Strategies for Patient-Centered Comparative Effectiveness Research

• Activity Number: 227
• Type: Invited
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Douglas P. Landsittel* and Sally Morton and Joyce Chang and Elan Cohen and Andrew Topp
• Companies: University of Pittsburgh and University of Pittsburgh and University of Pittsburgh and University of Pittsburgh and University of Pittsburgh
• Keywords: observational ; causal inference ; propensity scores ; instrumental variables ; CER

Abstract

Comparative effectiveness research (CER) seeks to evaluate which treatment works best when, and for whom, under which circumstances. CER often relies on observational data; associated limitations, such as treatment self-selection or other forms of bias, may necessitate use of methods such as propensity scores and instrumental variables. While there is substantial literature on these methods, there are also many variations and a lack of clear guidance on when to use which method. For instance, there are still no clear answers to the most fundamental questions, such as "Which method should I trust more if instrumental variables and propensity scores lead to different results?" This presentation summarizes preliminary results from a study funded by the Patient-Centered Outcomes Research Institute that aims to 1) conduct a systematic review, 2) conduct simulations to address gaps, and 3) develop a decision tool for recommending optimal methods for a given data set. The presentation will present findings of the systematic review and present a preliminary version of the Decision Tool for Observational Data Analysis Methods for Comparative Effectiveness Research, or DecODe CER.