Menu

How Broad is Narrow? Measuring Network Breadth for Physician Services in New York State Health Plans

Tuesday, June 25, 2019: 11:00 AM
Wilson B - Mezzanine Level (Marriott Wardman Park Hotel)

Presenter: Thomas Koch

Co-Author: Marshall Thomas

Discussant: Amanda Starc


Narrow network design in health insurance plans reflects tradeoffs faced by plan designers and beneficiaries: narrow networks arise when health plans follow through on threats to exclude providers. These plans negotiate lower prices paid to providers and limit options to beneficiaries, which reduces the value to those seeking care. There is a nascent theoretical literature on why negotiated networks might be narrow, with scant empirical evidence about how many of them actually are narrow.

Recent research on narrow networks focuses on rarely utilized and lumpily measured inpatient care. We consider network design in outpatient physician services markets. Physician networks provide a natural setting to measure the breadth of all health insurance networks and to understand why networks might be uniformly broad, narrow, or vary substantially across plans marketed to beneficiaries in the same area. Unlike hospital networks, which bundle their in-network services together across service lines, health insurance plans can construct physician networks across specialties (e.g., a network’s General Practitioner coverage could differ from its General Surgery coverage). Plans constructing networks might face different market structures across specialties in the same geographic area.

We use network participation data from New York State for health insurance plans in 2017. New York State collects the in-network status of plans sold and marketed in New York. This data indicates an affirmative declaration by the insurer that the provider is in network. With this data, we measure the number of providers that might be in-network for any insurer in any particular county against those that are actually in-network for any particular plan. This approach is superior to inferences based on in-network status from insurer claims databases, since it does not rely upon utilization to capture in-network status.

We document the substantial geographic variation in employer-sponsored health plan data.

  • We examine seven different specialty groupings, including General Practitioners, General Surgeons, OBGYNs, Pediatricians, Hospitalists, Orthopedic Surgeons, and Emergency Medicine specialists.
  • Plans often have the same physician and specialty across multiple counties. Physicians could belong to more than one plan, and have more than one specialty across counties.
  • The median network is narrower in rural areas across all specialties.
  • More populated areas, such as New York City, Syracuse, and Buffalo, have broader median networks.

We also evaluate some potential sources of both the levels of network breadth and the variations across plans within a geographic area. We calculate the concentration levels of the physician providers as well as local hospitals to determine which better describe patterns of network breadth.