Propensity Score Methods for Bias Reduction in Observational Studies of Treatment Effect

A challenge to the use of observational data to study treatment effects is the issue of confounding. Noncomparability of exposed and nonexposed subjects can lead to biased estimation of the treatment effect. The propensity score is a balancing score that can be used to form matched groups, or pairs, that are not systematically different and enable nonbiased comparisons between groups. This article reviews propensity score methods with an illustrative example of the application of propensity score matching in an observational study of an uncommon disease (systemic sclerosis).