Diagnostics for Propensity Score Matching: Creating a Balanced Control Group to Assess In Vitro Fertilization Guidelines

• Activity Number: 518
• Type: Contributed
• Date/Time: Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Cassandra K. Wolos*+ and Donald B. Rubin
• Companies: Harvard University and Harvard University
• Keywords: Propensity Scores ; Matching ; Diagnostics ; Covariate Balance ; In Vitro Fertilization

Abstract

This analysis employs propensity score matching to compare the effects of transferring one versus two embryos during in vitro fertilization on pregnancy rates, multiple births, and other outcomes. The data, a subset of the US Assisted Reproductive Technology Surveillance Database, includes 22,735 patients who received either one or two embryos. The goal is to select matching patients from the double embryo group who are as similar as possible to the patients from the single embryo group with respect to background covariates, without access to outcomes. We created matched pairs using propensity scores, under the constraint of exact matching on three vital covariates. This paper focuses on the process of identifying a propensity score model and matching protocol, with a particular emphasis on diagnostics for checking covariate balance.