Estimation of Treatment Effect of Asthma Case Management Using Propensity Score Methods

Objective: To estimate the treatment effect from participating in an asthma intervention that was part of the National Asthma Control Program. Study Setting: Data on children who participated in asthma case management (N=270) and eligible children who did not participate in case management (N=2,742) were extracted from a claims database. Study Design: We created 81 measures of health care utilization and 40 measures of neighborhood characteristics that could be related to participation in the program. The participation model was selected using the cross-validation-based Deletion Substitution and Addition (DSA) algorithm. We used optimal full matching for the vector of Mahalanobis’ distances and propensity scores to estimate the difference between participants and non-participants in the probability of a range of asthma outcomes. Principal Findings: Compared to non-participants, participants were more likely to have vaccinations for pulmonary illness, use controller medications, and have a refill for rescue medication. There was no statistically significant difference in the number of nebulizer treatments or ED visits between the two groups. We find that the asthma program had no significant effect on overall asthma control. Conclusion: We are not able to discern whether the lack of an effect in overall control is due to the effectiveness of the program, heterogeneity of effects or barriers outside the program’s control. We discuss how current programs could be modified to better inform future research and program design