Estimating the economic and clinical impact of asthma disease management programs traditionally has relied on non-experimental designs and employed matching or stratification methods with limited success. Selecting similar comparison subjects is problematic since subjects must be compared across numerous pretreatment factors. In cases where treatment and comparison subjects differ greatly on observed characteristics, conclusions may be particularly sensitive to an incorrectly specified model used for matching. A propensity score method constructs matched samples of treated-control pairs, addresses program selection bias, and reduces bias in estimates of treatment effects. To investigate the program effects of an asthma care support program delivered to high-risk asthmatics (persons with a previous inpatient admission, emergency department [ED] visit, or observation visit), we conducted a matched-cohort study on 196 participants. Using administrative claims data and selected clinical indicators, we analyzed hospitalization, ED, and physician office visit rates to estimate effects of program enrollment. Total hospitalizations, asthma-related hospitalizations, bed days, and ED visits for participants were lower and statistically different from that of the matched-cohort group during the program period, suggesting the beneficial effects of monitoring, education, and counseling activities for participants. Where controlled randomized clinical trials cannot be performed because of ethical, cost, or feasibility issues, the use of propensity scores provides an alternative for estimating treatment effects using observational data. This study employs a propensity score-matching methodology to select a subset of comparison units most comparable to treatment units, and documents the beneficial outcomes of participation in an asthma care support program.
