What Happens When Employers Switch from a "Carve-Out" to a "Carve-In" Model of Managed Behavioral Health?

Background: Since the introduction and soaring popularity of the managed behavioral healthcare (BH) "carve-out" model in the 1980s, policymakers have been concerned with their impact on access. In carve-outs, B and medical benefits are administered separately. Earlier literature found they reduced intensity of service use while maintaining penetration rates. Recently it has become more common for employers to drop existing carve-out contracts, partly due to the Mental Health Parity and Addiction Equity Act (MHPAEA), which placed a greater administrative burden on carve-outs for parity compliance. Although prior studies focused exclusively on the impact of moving from carve-in to carve-out models, it is now more policy-relevant to understand the effects of the move from carve-out to carve-in, which may not be symmetric. Moreover, the natural experiment resulting from MHPAEA implementation may attenuate concerns about selection bias. Study Aims: This study examines how specialty BH care patterns change when employees and dependents are moved from a "carve-out" plan to a "carve-in" plan. Methods: Linked insurance claims, eligibility, plan and employer data from 2008-14 were obtained for three Optum (R) employers who dropped their carve-out contracts but retained their carve-in plans. A longitudinal "difference-in-differences" study design was used to compare changes in BH services use over time among individuals who were: (i) moved to carve-in plans when the employer dropped its carve-out contract (N=177,653); and (ii) enrolled in carve-in plans before and after the transition (N=58,658). Outcomes included total and inpatient expenditures, broken down by plan, patient, and total; outpatient visits for assessment, individual psychotherapy, family psychotherapy, and medication management; and days of structured outpatient care, day treatment, residential care, and acute inpatient care. We pooled person-year observations and estimated regressions including individual fixed effects, year dummies and interactions between indicators for post-transition period and whether transitioned from carve-out to carve-in. Results: Relative to individuals continuously in carve-in plans, those who were transitioned experienced significant increases in inpatient utilization (beta=.02; p=.05) and patient inpatient costs (beta=2.35; p=.01) and decreases in day treatment (beta=-0.01; p=.02). Our conclusions proved robust against potential biases due to differing secular time trends and differential changes in benefits resulting from MHPAEA. Discussion: The increased inpatient utilization associated with switching from carve-out to carve-in plans is consistent with previous literature. Carve-outs may use day treatment to reduce inpatient care so that increased inpatient utilization post-transition reduced demand for day treatment. Limitations include possible selection bias at the employer level; lack of data on medication and generalist use, quality, clinical endpoints and quality of life; and potential lack of generalizability. Implications for Health Care Provision and Use: The reduction in the use of carve-out contracts by private employers associated with MHPAEA implementation likely did not have a net negative impact and may have actually increased access to care among former carve-out enrollees in need of inpatient services. Implications for Health Policies: Policymakers should consider and evaluate possible unintended consequences of legislation designed to improve access to care. Implications for Further Research: Future work should replicate these analyses with a more representative sample.