Association for Public Policy Analysis & Management

This study examines market and procedural characteristics that drive consumer shopping behavior in healthcare markets. In September 2017, FAIR Health released its New York Healthcare Online Shopping Tool (NY HOST), an independent, publicly-accessible state-wide consumer shopping tool. NY HOST displays benchmarks for insurers’ allowed amounts and providers’ billed charges for each 3-digit geozip (areas designated by the first three digits of a zipcode) by procedure for a range of values including benchmarks for the average and the 50th to 95th percentiles’ billed charges. The site also provides educational resources. The rollout was accompanied by an extensive, multi-pronged marketing effort. We embedded a randomized experiment in NY HOST – individual provider billed charges (list prices) were randomized across New York State’s 3-digit geozips for a set of common procedures.

We utilized a unique, comprehensive dataset of claims drawn from the FAIR Health database, which includes about 75% of all private insurance claims across the country, to analyze the effects of this randomized experiment on billed charges. The claims dataset encompassed all claims for 104 procedures which were selected from the database based on high frequency of utilization across a segment of specialties in New York State from 2016 through the second quarter of 2019. The dataset encompassed approximately 115 million claims, over 200,000 unique National Provider Identifiers (NPI), and multiple private insurers. We also utilized data provided by FAIR Health on utilization of the tool, including the date of search and the procedure codes that were searched.

Of the searches, the majority of searches were for medical and surgical services, and dental services accounted for nearly a quarter of all searches. Procedures searched included orthopedic procedures, gastrointestinal procedures (colonoscopy and endoscopy), dermatology procedures, and cardiovascular procedures. Of the codes searched on the NYHOST website, only about a quarter of searches were included in a published classification of “shoppable services,” while the majority of searches were for more emergent services not typically classified as shoppable in the literature.

We found that consumers appeared to use the tool strategically and market characteristics affected shopping behavior. Consumers searched more in procedure markets with provider-specific price information availability, more frequent out-of-network utilization, higher charges, significant charge dispersion, and substantial provider competition. Demographic characteristics of service users were also associated website use; there were more searches in a given geozip if users of that procedure in that geozip were disproportionately women and young adults. We also found substantial evidence that provider-level price information is valuable; areas and procedures randomized for release of physician level charges were more likely to exhibit utilization of the tool.

In conclusion, consumers use the tool for shoppable services, which may help contain provider pricing, but even more to assess prices for emergent services, perhaps as an ex-post negotiating tool in response to surprise bills. In addition, shopping behavior appears to respond to the level and variance of costs of services in the market and the availability of local competitors offering the same service.