The Impact of Psychiatrist Network Breadth on Mental Health Care Use in Medicare Advantage

• In Event: Policies Affecting Medicare Beneficiaries

Abstract

Over half of Medicare beneficiaries are now enrolled in Medicare Advantage (MA), the privately administered component of Medicare. MA plans use provider networks to manage utilization and costs, restricting enrollees to a subset of local providers. Prior work has documented that psychiatrist networks in MA are substantially narrower than those in other insurance markets, with nearly two-thirds of MA psychiatrist networks containing fewer than 25% of local providers — far narrower than comparable networks in Medicaid managed care or ACA marketplace plans. Yet the causal impact of these narrow networks on enrollees' access to mental health services has not been established. Identifying this effect is challenging because enrollees may select plans based on their networks, generating selection bias in naive comparisons.

We address this identification problem by exploiting quasi-random variation in network breadth generated by MA plan exits. When a plan terminates or reduces its service area, displaced enrollees most often re-enroll in a different MA plan and, by regression to the mean, tend to select plans with more typical network breadth. Enrollees previously in broad-network plans that exit thus experience narrower networks in the following year, and vice versa. We instrument for plan-county-level psychiatrist network breadth using the interaction of prior-year breadth and an indicator for plan termination, following prior work examining plan-driven mortality in MA. Network breadth is derived from MA encounter data as the share of local mental health providers observed seeing at least one plan enrollee, adjusted for county fixed effects and plan enrollment, which addresses measurement concerns with provider directory data. We use 2018-2019 Medicare claims and enrollment data for approximately 8 million MA enrollees, identifying plan terminations using the CMS plan crosswalk files and defining terminations to include both full plan exits and service area reductions. Approximately 50,000 enrollees experienced a plan termination in our sample period.

We validate our encounter-based network measure, showing it is uncorrelated with measures of MA encounter data missingness and correlates closely with a prescription-based alternative. Our first stage is strong. A one-unit increase in prior-year breadth interacted with termination status is associated with a 0.85 unit decrease in subsequent plan breadth among displaced enrollees (F-statistic = 41,930,000). Further, we find evidence supporting the balance and fallback conditions required for identification. Using 2SLS, we find that a 1-unit increase in psychiatrist network breadth increases the probability that an enrollee sees a mental health specialist by approximately 10%, with effects concentrated among access to psychiatrists specifically. We also examine effects on primary care visits with mental health diagnoses and prescription outcomes, and present heterogeneity analyses across enrollee subgroups.

These findings provide the first causal evidence that the breadth of psychiatrist networks in MA impact mental health care utilization, with implications for federal network adequacy regulations and ongoing efforts to strengthen mental health parity in Medicare.

Author

• Mark Katz Meiselbach, Johns Hopkins University
  Presenting Author