Medicaid Value-Based Payments and Health Care Use for Patients With Mental Illness

Abstract

Importance: Medicaid patients with mental illness comprise one of the most high-need and complex patient populations. Value-based reforms aim to improve care, but their efficacy in the Medicaid program is unclear.

Objective: To investigate if New York state's Medicaid value-based payment reform was associated with improved utilization patterns for patients with mental illness.

Design, setting, and participants: This retrospective cohort study used a difference-in-differences analysis to compare changes in utilization between Medicaid beneficiaries whose outpatient practices participated in value-based payment reform and beneficiaries whose practices did not participate from before (July 1, 2013-June 30, 2015) to after reform (July 1, 2015-June 30, 2019). Participants were Medicaid beneficiaries in New York state aged 18 to 64 years with major depression disorder, bipolar disorder, and/or schizophrenia. Data analysis was performed from April 2021 to July 2023.

Exposure: Beneficiaries were exposed to value-based payment reforms if their attributed outpatient practice participated in value-based payment reform at baseline (July 1, 2015).

Main outcomes and measures: Primary outcomes were the number of outpatient primary care visits and the number of behavioral health visits per year. Secondary outcomes were the number of mental health emergency department visits and hospitalizations per year.

Results: The analytic population comprised 306 290 individuals with depression (67.4% female; mean [SD] age, 38.6 [11.9] years), 85 105 patients with bipolar disorder (59.6% female; mean [SD] age, 38.0 [11.6] years), and 71 299 patients with schizophrenia (45.1% female; mean [SD] age, 40.3 [12.2] years). After adjustment, analyses estimated a statistically significant, positive association between value-based payments and behavioral health visits for patients with depression (0.91 visits; 95% CI, 0.51-1.30) and bipolar disorder (1.01 visits; 95% CI, 0.22-1.79). There was no statistically significant changes to primary care visits for patients with depression and bipolar disorder, but value-based payments were associated with reductions in primary care visits for patients with schizophrenia (-1.31 visits; 95% CI, -2.51 to -0.12). In every diagnostic population, value-based payment was associated with significant reductions in mental health emergency department visits (population with depression: -0.01 visits [95% CI, -0.02 to -0.002]; population with bipolar disorder: -0.02 visits [95% CI, -0.05 to -0.001]; population with schizophrenia: -0.04 visits [95% CI, -0.07 to -0.01]).

Conclusions and relevance: In this cohort study, Medicaid value-based payment reform was statistically significantly associated with an increase in behavioral health visits and a reduction in mental health emergency department visits for patients with mental illness. Medicaid value-based payment may be effective at altering health care utilization in patients with mental illness.

Figure 1.. Differential Change in Outcomes for VBP Patients in Each Year

The y-axis displays the range in values for the coefficients of the association between value-based payment (VBP) (A-C) and the mean change in the number of visits for each outcome (D-F). The colored dots are the estimates from the adjusted model for every year of the study, with July 1, 2013, to June 30, 2014, set as the reference year. The estimates are centered between years to indicate the midpoint between the start and end of each study year (July 1). The vertical red line represents the implementation of VBP on July 1, 2015. ED indicates emergency department.

Figure 2.. Mean Number of Visits per Year for VBP and Non-VBP Groups, Stratified by Diagnosis

The y-axis is the average mean outcome per year for value-based payment (VBP) and non-VBP patient populations stratified by diagnosis. The x-axis represents each year of the study period. The 2 pre-VBP–period year averages are represented by the data points above “Jul 2014” (the average mean outcome for July 1, 2013-June 30, 2014) and the data points above “Jul 15” (the average mean outcome for July 1, 2014-June 30, 2015). The vertical red line indicates the implementation of VBP on July 1, 2015. Error bars represent 95% CIs. ED indicates emergency department.