The Breadth and Characteristics of Effective Medicare Advantage Networks

Background: Medicare Advantage (MA) plans use network management tools to restrict enrollees’ access to providers. Little is known about the extent of MA plan networks, its relationship to market factors, the demographics of patients in narrow vs broad networks, and whether plans offer meaningful access to providers. One reason for the paucity of evidence is that there is no comprehensive, historical source of provider affiliation with MA plans, and plan directories are known to be error prone.

Methods: We used 2014 Medicare Part D claims and the Physician Compare Database to measure “effective” prescriber-based networks for MA plans. Our primary measure was the share of prescribers in a county that are in network for a given contract-plan combination. Providers that did not match a prescription drug event were excluded. We allowed for multiple locations of a provider listed in Physician Compare, so long as that location was within a plan’s service area. We considered providers to be in-network if they had at least one prescription drug event for a given contract-plan-county combination.

Results: Out of 10.4 million MA enrollees, 37% (3.9 million) were enrolled in narrow network plans in 2014, those including less than 30% of all prescribers in the market. The mean contract-plan-county included 26% of county-level prescribers in-network. Broad networks, defined as including 70% or more of providers, plans were more prominent in less populous areas. Per-capita income was higher ($41,810) in counties served by narrow networks versus broad networks ($35,590). Lastly, HMO plans tended to have about the same share of providers in-network (27%) than PPOs (26%). In the coming months, we will conduct additional analyses focused on primary care providers that will illuminate changes in network extent over time (using data from 2011-2015), and association with market factors.

Conclusions: The extent of MA plan networks is an important regulatory issue, with recent lawsuits based in part on inadequate access to providers. Beyond regulatory concerns, developing methods to infer MA plans’ networks is crucial, as little is known about how network extent relates to plan cost and quality. Therefore, our work has both immediate policy relevance and is foundational for future study of the economics of MA plans.