Standardization of government spending on health care | Health matters

It is well known that health care costs per person in the United States are higher, and in some cases much higher, than in many similar countries. While this is true for the US as a whole, it is not true everywhere in the country. In 2019, total health care spending per person was twice as much in Washington, DC than in Utah. What’s more, states like Utah at the lower end of the health care spending spectrum have spending patterns closer to the averages of other high-income countries than to the average US health care spending levels. Just as very few patients spend the average amount on health—most spend significantly more or less—many states spend significantly more or less than the U.S. average.

The large differences between countries can be explained, at least in part, by some known factors. Differences reflect underlying demographic and economic characteristics, as well as health administration and financing mechanisms unique to each country. To investigate them, we published a study in Health matters in August that projected health care spending by state through 2019 using the Centers for Medicare and Medicaid Services (CMS) State Health Expenditure Accounts (SHEA). At the time, these estimates only extended back to 2014.

As part of this analysis, we quantified the economic and demographic factors associated with cost growth and variation across states. Ten days after our study was published, CMS (much to our surprise) released the official SHEA update with data for 2020. Although these new data obviate the need for cost projections like our own, our investigation of cost trends remains informative for look beyond raw spending totals to compare spending across US states and understand the drivers of variation. To this end, we repeated our analysis from August 2022 using the new SHEA data to confirm our initial findings with official data and to assess the strength of evidence related to health care spending trends. We chose to leave 2020 out of our analysis because of the additional complexity introduced in disentangling pandemic effects, but a similar analysis of pandemic and post-pandemic data would also be informative.

When evaluating differences in state spending, it is important to compare apples to apples and to consider state characteristics that may influence variation but are generally considered beyond the control of the health care system. For example, the two-fold difference in per capita spending between Washington, DC, and Utah at least partially reflects differences in age structure, underlying population health, and cost of living, in addition to critical differences in the health care systems of each location. Understanding these fundamental aspects of variation and accounting for factors beyond the reach of most policy interventions are essential to identifying effective cost control measures.

To standardize health care spending levels for an apples-to-apples comparison, we replicated the methods used in our August publication. We first standardized health care costs per person by age and sex (using indirect age-standardization methods) and adjusted them for inflation and regional price parity (using data from the Bureau of Economic Analysis). Second, we used regression analysis to control for per capita income, population density, behavioral health risk, and time (the tendency for health spending to increase each year in all states). With this approach, almost half of the variation in state health care spending is explained by income and regional prices. These two factors together with time, health risks (physical activity and smoking prevalence), population density, age and gender explain more than 75 percent of the variation (Appendix 1).

Figure 1: Sources of variation among states in per capita state health care expenditures, 1990–2019.

Source: Authors’ estimates obtained from government health care spending data from the Centers for Medicare and Medicaid Services.

The remaining, unexplained variation reflects heterogeneity caused by other factors, some of which may respond to cost containment measures. This analysis and our August publication reveal the extent to which variables outside the health system influence spending outcomes. It also highlights the significant remaining variation that provides an opportunity for cross-country comparisons and improvements.

After SHEA standardized state spending for the factors external to the health care system described above, the distribution of high- and low-spending states shifted, but substantial differences between states persisted (Appendix 2). For countries with significantly different populations (in terms of age and health risk) and different levels of economic development, standardized costs are more similar across countries than unstandardized costs. States like Alaska and West Virginia have remained at the extremes of spending and are indicative of places where other factors — some of which are likely to be key health policies — affect spending amounts.

Appendix 2: Unstandardized Health Care Expenditure per Person and Standardized Health Care Expenditure per Person by Country, 2019

Source: Authors’ estimates obtained from government health care spending data from the Centers for Medicare and Medicaid Services. Notes: Each row represents the unstandardized and standardized expenditures for one state in 2019. States with interesting or noteworthy trends are highlighted in color and the state abbreviation is shown.

After controlling for external influences on government spending growth, our August publication examines the relationships between standardized spending and Medicaid expansion, both in the aggregate and by type of insurance. Using recently updated SHEA data, Figure 3 shows that standardized spending increased at very similar rates between 2013 and 2019 for states that expanded Medicaid before 2016 and those that did not. Importantly, while Medicaid expansion was clearly associated with increases in Medicaid spending growth, expansion states had lower average growth in private insurance spending than their peers.

Example 3: Annual growth in standardized health care expenditures per person by state from 2013 to 2019, by Medicaid expansion status

Source: Authors’ estimates obtained from government health care spending data from the Centers for Medicare and Medicaid Services.

In our regression analysis, we estimated the significance of expanded Medicaid access after controlling for the standardization factors mentioned above; we saw that raising the income eligibility thresholds for children and adults was significantly associated with higher total health care costs, but that increasing the eligibility thresholds for pregnant women was associated with lower total costs (Exhibit 4). Medicaid expansion in each state resulted in 2 percent higher total spending in subsequent years than states without expansion.

Figure 4: Factors associated with changes in total state-level health care spending per person for the years 2000–19 after controlling for age, cost, income, population density, and behavioral health risk; percentage changes

Source: Authors’ estimates obtained from government health care spending data from the Centers for Medicare and Medicaid Services. ** p < 0.05; *** p < 0.0001

Official health care expenditure reports are an invaluable tool for health researchers and policymakers, and the release of SHEA by CMS by 2020 opens the door to many informative research projects. The links between Medicaid expansion and increased health care spending discussed in our article provide insight into ways to potentially slow spending growth, which could have an impact regardless of state profiles. Future researchers should keep in mind that there are many potential explanations for why health care spending varies so much across countries, and that it is critical to determine how much of this variation is within policy makers’ ability to influence. Methods such as our standardization analysis are a productive way to identify and account for drivers of health care costs that we cannot easily control, and to ensure that state comparisons are made on a level playing field.

Authors’ note

Research funding for this project came from Gates Ventures and the Peterson Center on Healthcare. Dr. Joseph L. Dilleman currently has funding from the Bill & Melinda Gates Foundation.

Leave a Comment

Your email address will not be published. Required fields are marked *