Propensity Score Methods for Analyzing Observational Data Like Randomized Experiments: Challenges and Solutions for Rare Outcomes and Exposures

• Published in: American journal of epidemiology Vol. 181; no. 12; pp. 989 - 995
• Main Authors: Ross, Michelle E, Kreider, Amanda R, Huang, Yuan-Shung, Matone, Meredith, Rubin, David M, Localio, A Russell
• Format: Journal Article
• Language: English
• Published: United States Oxford Publishing Limited (England) 15-06-2015
• Subjects: Adolescent; Antipsychotic Agents - adverse effects; Causality; Child; Clinical outcomes; Clinical trials; Confounding Factors (Epidemiology); Data Interpretation, Statistical; Diabetes; Diabetes Mellitus, Type 2 - chemically induced; Epidemiologic Research Design; Female; Humans; Intention to Treat Analysis; Longitudinal Studies; Male; Matched-Pair Analysis; Medicaid; Models, Statistical; Observational Studies as Topic; Propensity Score; Psychotropic drugs; Randomized Controlled Trials as Topic; Side effects; United States > US; randomized experiments; confounding; propensity score matching; inverse probability of treatment weighting; discrete-time failure analysis; marginal effects; observational study
• ISSN: 0002-9262; 1476-6256
• DOI: 10.1093/aje/kwu469

Randomized controlled trials are the "gold standard" for estimating the causal effects of treatments. However, it is often not feasible to conduct such a trial because of ethical concerns or budgetary constraints. We expand upon an approach to the analysis of observational data sets that mimics a sequence of randomized studies by implementing propensity score models within each trial to achieve covariate balance, using weighting and matching. The methods are illustrated using data from a safety study of the relationship between second-generation antipsychotics and type 2 diabetes (outcome) in Medicaid-insured children aged 10-18 years across the United States from 2003 to 2007. Challenges in this data set include a rare outcome, a rare exposure, substantial and important differences between exposure groups, and a very large sample size.