Variation in Health Care Prices and Service Use Across U.S. Health Care Markets

Tuesday, June 25, 2019
Exhibit Hall C (Marriott Wardman Park Hotel)

Presenter: William Johnson

Co-Author: Kevin Kennedy

We studied how health care spending in the commercially insured population varied across geographic markets and the role of differences in the prices and use of health care services in that spending variation. Compared to public programs, there has been less research about these questions within the commercially insured population – in large part due to the lack of widely available private health care claims data. We leveraged data on the allowed amounts paid for 1.8 billion commercial claims from the Health Care Cost Institute (HCCI) database spanning 112 Core Based Statistical Areas (CBSAs) across 43 states from 2012 to 2016.

We measured the total spending on, use of, and prices paid for a basket of common inpatient, outpatient, and professional health care services. This allowed us to create indices of prices and use and then to document the variation and co-variation of these measures across CBSAs. Overall, we found wide variation in spending across CBSAs. By 2016, spending in the CBSA at the 75th percentile was 16% above the median and the CBSA at the 25th percentile 12% below. In our sample, while variation in prices and use both played large roles in determining variation in spending, variation in use was responsible for the majority of spending variation. Even so, in our sample, variation in use accounted for a substantially smaller share of spending variation than previous research has found among the Medicare population (where price variation is constrained).

We categorized whether the level of spending in a particular CBSA was primarily attributable to price or use levels by decomposing our spending index as the product of our price and use indices. In our sample, only 43% of CBSAs in the top half of the spending distribution (“high spending” CBSAs) were in the top half of both the price and use distributions. Another 34% of high spending CBSAs were only in the top half of the use distribution and 23% were only in the top half of the price distribution. We also found a symmetric trend for CBSAs in the bottom half of the spending distribution. Our findings suggest that within high and low spending CBSAs there are distinct groups of CBSAs with different factors driving their high or low level of spending.

The differences in price and use levels reflect different potential causes and potentially different policy solutions. Differences in prices partly reflect differences in the cost of living across markets but may also result from differences in provider or insurer market power. Differences in use rates may stem from differences in providers’ patterns of practice but also reflect disparities in health across markets. Although a full analysis of those factors is beyond the scope of this project, our findings are a starting point for future research to understand variation in levels of health care spending in different markets across the country.