Abstract
Objectives. To examine the accuracy of official estimates of governmental health spending in the United States.
Methods. We coded approximately 2.7 million administrative spending records from 2000 to 2018 for public health activities according to a standardized Uniform Chart of Accounts produced by the Public Health Activities and Services Tracking project. The official US Public Health Activity estimate was recalculated using updated estimates from the data coding.
Results. Although official estimates place governmental public health spending at more than $93 billion (2.5% of total spending on health), detailed examination of spending records from state governments shows that official estimates include substantial spending on individual health care services (e.g., behavioral health) and that actual spending on population-level public health activities is more likely between $35 billion and $64 billion (approximately 1.5% of total health spending).
Conclusions. Clarity in understanding of public health spending is critical for characterizing its value proposition. Official estimates are likely tens of billions of dollars greater than actual spending.
Public Health Implications. Precise and clear spending estimates are material for policymakers to accurately understand the effect of their resource allocation decisions.
The coronavirus disease 2019 (COVID-19) pandemic has illuminated weaknesses in the US public health system.1 Despite spending $3.6 trillion each year on health, the vast majority of this spending goes toward health care, with relatively little toward public health.2,3 Official government estimates indicate that spending on governmental public health accounts for 2.5 cents of every dollar the United States spends on health, which totaled $93.5 billion in 2018.2 However, research indicates that the official government estimate (Public Health Activity estimate [PHAE]) is a substantial overestimation of how much is actually spent on governmental public health in the United States and that we spend far less.4
Research has shown the PHAE conflates governmental public health spending with other related health care spending, such as disability-related clinical care, behavioral health, and publicly supported outpatient clinics; these health services are counted elsewhere in the National Health Expenditure Accounts.5 Whether such health care spending is merely misclassified or double counted is unclear.4 These definitional problems in the PHAE make accurately measuring expenditure alignment between health care, social services, and public health challenging. These problems also may give an incorrect impression about the growth of public health spending over time, relative to other components of national health or social service spending.6 We report the results of an examination of millions of government expenditure records to elucidate how much is actually spent on public health across the United States.
We coded and analyzed state government expenditure data collected by the US Census Bureau; these census data undergird the PHAE as compiled by the Centers for Medicare and Medicaid Services.4 Approximately 2.7 million object-level records for census function code 32 (Health - Other) across 49 states were available from 2000 to 2018. In this iteration of the project, the project team coded data from 2013 to 2018, made available by the Census Bureau.7 Data were coded using the University of Washington Uniform Chart of Accounts crosswalk, which offers a framework for categorizing areas of public health activity.8 Census records previously coded from 2000 to 2012 were updated to the current version of the University of Washington Uniform Chart of Accounts. Records were collapsed from object level (e.g., salaries) to program level (e.g., HIV screening) and coded. Teams of 2 to 3 coders judged each category and resolved conflicts in agreement iteratively. For categories that accounted for less than $5 million in total spending across all available years, 1 coder coded each category. Conflicts were reviewed and resolved manually. Data included records of health-related intergovernmental transfers to local governments. The project team disaggregated these intergovernmental transfers as either for public health or for individual services with the same approach as that described earlier. It was then possible to calculate the share of public health activities financed at the local level by state transfers to local governments.
A revised PHAE was created by (1) aggregating individual, recoded records from states that reported to the census (n = 49 for all years) and (2) adding the publicly available point estimates for 1 state, California, that only reported in aggregate to the census. Total spending on census function code 32 (Health - Other) for states with aggregate-only estimates was reduced proportional to what percentage of public health spending or total function code 32 spending the project team observed nationally after recoding all available data. Local spending on function code 32 estimates was similarly decreased in line with previous approaches.9 The upper end of the revised function code 32 estimate is based on an approach wherein the project team assumed that local spending was wholly public health related (except for state to local intergovernmental transfers that were measurable in the data). On the low end of the revised estimate, the project team assumed that estimates on local spending on function code 32 should be reduced proportional to state and intergovernmental transfer spending.9 In all cases, the default assumption was that if the project team could not clearly identify the spending record, then it would remain categorized as “public health spending.” As such, the revised estimates are potentially quite conservative in reductions from the PHAE overall. Spending estimates were adjusted over time by using the Bureau of Labor Statistics state and local government deflator.
Recoding of state government expenditure data collected by the Census Bureau shows that actual public health spending ranged between $38 billion and $65 billion in 2017 and between $35 billion and $64 billion in 2018. The difference between the nationally reported PHAE and our revised estimate ranges from 36% to 54%, on average, between 2000 and 2018 (Figure 1). The revised PHAE shows more modest growth, compared with the official PHAE, after accounting for inflationary effects. Since 2000, both the official PHAE and the revised estimate show about 2% growth, on average, above inflation. However, the Great Recession blunted the growth of public health spending substantially, and its effects were felt until at least 2014. Thus, the official PHAE continued to rise from 2010 to 2014, but the revised estimate stagnated over this same period. Data from the Census Bureau over the 2000 to 2018 period show that $15 billion to $20 billion in PHAE were transferred per year from state governments to local governments. Analysis of these transfers indicates that 28% in 2018 was for public health activities, and 72% was for health care and related activities, suggesting high levels of transfers to finance direct care or other clinical services.
FIGURE 1— Public Health Spending in the United States: 2000–2018
Source. Centers for Medicare and Medicaid Services, US Census Bureau, and author analysis. Presented in constant 2018 dollars.
Closer scrutiny of the actual spending records shows that the United States spent far less on public health than the official PHAE of $93 billion in 2018. Given available data, actual public health spending is estimated to range between $35 billion and $64 billion (34%–61% of the official estimate). Along with more funding,3 improved accuracy in financial reporting is important to enable proper financial planning to cope with the current pressing US public health crisis. Assessing the value proposition of public health becomes challenging in the absence of accurate spending data, which are increasingly of interest to policymakers, particularly in the wake of the COVID-19 pandemic.3,10,11
Improving definitional clarity and accuracy of spending estimates will also greatly aid efforts to understand the costs and benefits of integration between social services, health care, and public health.12 One approach to this, compatible with the data presented here, is to conduct schema-matching exercises—to crosswalk an organization’s expenditures with a standardized definition set or chart of accounts.8 Limitations to this approach are presented in the Appendix (available as a supplement to the online version of this article at http://www.ajph.org).
Governmental public health is on the front lines in the response to the COVID-19 pandemic. Analysis of government spending data suggests that actual spending on public health is somewhere between one third and two thirds of the official national estimate. Thus, although it may appear somewhat inconsequential whether public health gets 1 cent or 2 cents on the health dollar in the United States, it is extremely consequential, particularly in the context of the COVID-19 pandemic, to have accurate spending data to better understand the value of public health services and ensure necessary resources and capacity to adequately protect the health of the public.
Detailed analysis of government spending data shows that potentially less than half of the official estimate of public health spending goes toward population-focused activities—the remainder goes to health care and related spending on individual services. It is critical for public health agencies to clearly communicate what they do, what it costs to do so, and the benefits of those activities. Without clear estimates of spending, establishing a value proposition in public health remains elusive.
See also Dasgupta, p.
ACKNOWLEDGMENTS
This study was funded by the Robert Wood Johnson Foundation through a Systems for Action grant.
The authors thank the US Census Bureau for provision of data and Rebecca Reif for assistance with data cleaning.
CONFLICTS OF INTEREST
The authors have no conflicts of interest to disclose.
HUMAN PARTICIPANT PROTECTION
Institutional review board approval was not needed because this article uses secondary publicly available data on public spending.
