Panel Paper: Association Between Insurance Network Breadth and Hospital Characteristics and Quality

Friday, November 3, 2017
Toronto (Hyatt Regency Chicago)

*Names in bold indicate Presenter

John Graves1, David Howard2, Sydney Broadhead1 and Ashish Jha3, (1)Vanderbilt University, (2)Emory University, (3)Harvard University


Research Objective: The Affordable Care Act’s health insurance Marketplaces have been revolutionary in reforming the standards to which insurers are held and how health benefits are structured. Given the restrictions placed upon insurers, in the form of regulations such as guaranteed issue and essential health benefits, insurers have increasingly turned to restricting provider networks to limit costs. Key unanswered questions include whether these networks may limit access to certain types of hospitals (e.g., teaching hospitals), or may differ in terms of the quality of in-network hospitals. Our study’s primary objective was to investigate the quality and characteristics of hospitals in private plans sold on the ACA’s health insurance marketplaces in 2016.

Study Design: We extracted the full provider networks for 96% of Silver marketplace plans sold in the health insurance marketplaces for 38 states. We classified plans as basic (i.e., narrow), standard or broad based on the hospital participation rate, or the percentage of local hospitals considered “in-network” for a given plan in its rating area. In line with CMS definitions, basic plans were those with a hospital participation rate one standard deviation or more below the mean, while broad plans were those one standard deviation or more above the mean. We coupled these data with additional hospital-level measures from the American Hospital Association Annual and Supplemental Surveys, as well as quality data from the CMS Hospital Compare website. Our primary statistical analyses relied on fixed effects models that effectively compared the characteristics of in-network hosptials among plans within the same local market.

Population Studied: 1,402 Silver Marketplace Plans in 38 states.

Principal Findings: Overall, 20.8% of plans were characterized as “basic” or narrow-network, though the fraction of narrow network plans ranged from 7.7% in markets with 1-2 issuers; 16.7% in markets with 3-5 issuers; and 33.9% in markets with 5 or more issuers. Compared to standard plans within the same rating area, basic network plans were substantially less likely to include a teaching hospital (-25.2 percentage points, 95% CI -26.9 to -23.5), though the in-network hospitals in basic networks were slightly larger, on average (+48.0 beds, 95% CI 42.5 to 53.4). In addition, basic network plans were more slightly more likely to include hospitals that were part of integrated delivery systems (+3.3%, 95% CI 1.3 to 5.3). There were few differences in quality in basic network plans as compared to other plans in the same rating area, with hospital performance on composite process quality, patient satisfaction, and 30-day mortality and readmission scores statistically indistinguishable across network sizes.

Conclusions: In-network hospitals among plans listed in the ACA’s health insurance marketplace plans differed substantially in terms of structural characteristics, but did not differ along any dimension of observable quality.

Implications for Policy or Practice:  While hospital networks among narrow-network plans did not display any differences in terms of quality, patients with specialty medical needs (e.g., cancer patients enrolled in clinical trials) may find that such plans limit access to certain types of hospitals.