Blended Payments and Treatment Completion

Background: Blending two or more health financing strategies has been shown to facilitate care integration for people with medical comorbidities, improving both access to care and quality of care in chronic care initiatives. However, limited knowledge is available about how blended payment models impact mental healthcare and substance use treatment services. In addition, there is a gap in the literature addressing how financing models might have differing impacts on agencies based on the extent to which they integrate mental health and substance use treatment. This study assessed a blended payment model that combined capitated per-member per-month Medicaid payments for substance use treatment with Medicaid fee-for-service payments for mental health treatment provided in community mental health centers to identify the impact on completion of treatment by people with co-occurring disorders.

Methods: A triple differences model, expanding on a traditional difference-in-difference model, was used to analyze electronic health record data from nine community mental health agencies pre-and-post blended payment initiation. The outcome variable was a binary variable representing treatment completion. Predictor variables included a binary Medicaid recipient variable, a binary time variable, and a binary variable indicating whether an agency provided fully integrated mental health and substance use treatment or not. Age, race, and gender were held constant in the analysis. The community mental health agency providing services was modeled as a conditional fixed effect. Parallel trends for treatment completion based on insurance type and program integration were assessed and confirmed graphically.

Results: The sample consisted of 4277 individuals, of which 553 completed mental health and substance use disorder treatment during the study period. In the initial difference-in-differences model, Medicaid recipients had decreased odds of completing treatment compared to non-Medicaid recipients post-blended payment initiation (OR = 0.62, 95% CI: 0.41 to 0.92, p = .017). However, when the model was adjusted for level of agency integration through the construction of a triple differences model, there was no difference in the odds of completing treatment between Medicaid and non-Medicaid recipients (OR = 0.76, 95% CI: 0.33 to 1.78, p = .535). After the initiation of the blended payment model, all individuals with co-occurring disorders receiving care in co-located mental health and substance use treatment programs had increased odds of completing treatment compared to individuals receiving care in fully integrated programs (OR = 1.38, 95% CI: 1.16 to 1.64, p < .001).

Conclusions and Implications: Individuals utilizing Medicaid for insurance and experiencing co-occurring disorders have complex behavioral health and social needs that may impede treatment completion. Financial incentives that support co-occurring care integration may help reduce disparate treatment completion rates between Medicaid and non-Medicaid utilizers, impacting integration of services at agencies with less integrated treatment structures. Prior research suggests that financial incentives alone do not improve quality, however this study provides evidence that in some contexts, financial incentives may be sufficient for shaping treatment outcomes through system improvement.