What you see isn’t what you get: Prevalence and Exposure to Ghost Networks in Medicare Advantage

• In Event: Expanding Coverage, Ensuring Access: Challenges in Public Insurance

Abstract

Changes in Medicare payment policy, low (or zero dollar) monthly premiums, and availability of supplemental benefits not covered by traditional Medicare (TM) have fueled explosive growth in the Medicare Advantage (MA) program over the last decade. By 2023, MA enrolled over 50% of all Medicare beneficiaries and continues to grow. Many MA plans use provider networks as a cost-containment strategy. However, enrollees may find themselves in a “ghost network” if the advertise provider network is different from the realized network of providers who actually deliver care. MA enrollees can face difficulty finding in-network care when providers have closed their panels to new patients, are licensed but clinically inactive, or have moved. Marketing practices, use of brokers, and a limited enrollment period may leave some vulnerable patients in plans that do not meet their needs. Ghost networks have the potential to harm older adults by disrupting continuity of care, incurring high cost out-of-network care, and causing distress. In this study we quantify the size and extent of primary care ghost networks and describe characteristics of beneficiaries with greater exposure who may be disproportionately at risk of potential harms.

We used 2019-2022 national MA Enrollment and Encounter files and comprehensive provider network data from Ideon (formerly Vericred). Through this linkage, we identified in-network primary care providers and their billing data to compare advertised vs. realized networks by MA plan. Enrollment files were used to determine characteristics of beneficiaries with greater exposure to plans with poorer access (i.e., larger ghost networks). The ghost network was defined as the proportion of advertised providers, by contract and plan, with fewer than 11 total encounters by year. Both the extensive margin (proportion of plans with any phantom providers) and intensive margin (the share of ghost providers in each network) were calculated. Among plans with any ghost providers, enrollee characteristics were compared for plans above and below the median. We used logistic regression models to assess whether beneficiary characteristics were associated likelihood of being enrolled in a network with a larger or smaller share of “ghost” providers.

We identified 432,146 primary care physicians and advanced practice providers who were advertised as “in-network” in one or more of the 3659 MA plans nationwide. Of those providers, we found that 38.8% billed fewer than 11 total services for MA beneficiaries, rendering them “ghosts.” On average, MA plan enrollees were 74.6% non-Hispanic White and nearly 20% of enrollees were under age 65 (qualifying for Medicare through disability entitlement). Plans with larger ghost networks (poorer access) had significantly larger shares of enrollees who were from distressed communities (vs. not distressed using the Distressed Community Index) and who were male (vs. female), p<0.001. Preliminary results also signal that plans with larger shares of enrollees whose race is not Non-Hispanic White are more likely to have larger ghost networks.

Ghost networks are a common occurrence in Medicare Advantage. It is unclear whether enrollment in plans with larger ghost networks is a feature of distressed communities, or a consequence. 

Authors

• Mika Hamer, University of Maryland, College Park
  Presenting Author

• Rachel Smith, University of Maryland
  Non-Presenting Co-Author