Net revenue per resident refers to the estimated annual net revenue across all hospitals. Z scores are calculated by subtracting the mean and dividing by the standard deviation for a given metric. The coronavirus disease 2019 (COVID-19) mean z score is estimated as the mean of z scores for the following outcomes: cumulative COVID-19 deaths per 100 000, cumulative COVID-19 cases per 100 000, non–COVID-19 deaths per 100 000, and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) case-to-test ratio. Higher COVID-19 mean z scores indicate higher burden of COVID-19 (higher COVID-19 cases, higher COVID-19 and non–COVID-19 deaths, higher positive SARS-CoV-2 case-to-test ratio). The mean comorbidity burden z score is the mean of the z scores among the following comorbidities, estimated using population-based data on prevalence: end-stage kidney disease, diabetes, hypertension, obesity, and smoking. Higher mean comorbidity z scores indicate poorer health. The mean hospital finance z score includes the following metrics: mean hospital operating margin and days of cash on hand across Medicare fee-for-service admissions within the county. Higher mean hospital finance z scores indicate better financial health. Best fit lines were fitted linearly for all panels. For each panel, more than 95% of counties are within the bounds of the axes. The remaining counties are top or bottom coded for scatterplot visualization only. The dots indicate counties and the dashed lines indicate the best fit line across all counties.
eTable 1. Sources for Estimates of Health and Financial Need
eTable 2. Method to Estimate Funding Allocation for Medicare-Participating Entities
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Kakani P, Chandra A, Mullainathan S, Obermeyer Z. Allocation of COVID-19 Relief Funding to Disproportionately Black Counties. JAMA. Published online August 07, 2020. doi:10.1001/jama.2020.14978
The Coronavirus Aid, Relief, and Economic Security (CARES) Act and Paycheck Protection Program together designated $175 billion for coronavirus disease 2019 (COVID-19) response efforts and reimbursement to health care entities for expenses or lost revenues.1
The most important factor driving funding allocation is past revenue. However, revenue is an imperfect measure of need because it is also influenced by prices, overuse, payer mix, and market consolidation.2 Moreover, non-White and indigent populations generate lower revenues, due to underinsurance and undertreatment,3,4 and hospitals caring for them may receive less relief despite confronting a greater burden of COVID-19.5
We examined how relief funding relates to the health and financial needs of US counties, focusing on disproportionately Black counties.
Relief funding is allocated according to a complex, nonlinear formula (combining revenue, location, uninsurance, and COVID-19 hospitalizations; see eTables 1 and 2 in the Supplement for details). We measured (1) how this allocation relates to the health and financial needs of counties, and (2) whether this allocation strategy was associated with differences in funding by county racial composition. Ideally, 2 counties receiving equal funding should have equal needs, irrespective of racial composition.
To measure needs, we used county-level data to form 3 categories: COVID-19 burden (cumulative deaths and cases and cumulative severe acute respiratory syndrome coronavirus 2 case-to-test ratio as of June 28, 2020; non–COVID-19 deaths between February 1 and July 1, 2020), comorbidities exacerbating COVID-19 (hypertension, end-stage kidney disease, obesity, smoking, diabetes), and hospital financial health (mean operating margin, cash on hand, as estimated in prior research6). We summarized these categories with mean z scores (eg, for financial health: each hospital’s margin and cash were converted to z scores by subtracting means and dividing by standard deviations [SDs], then averaged across all hospitals in the county).
We then projected anticipated relief funding for Medicare-participating entities based on announced policies as of July 5, 2020 ($120 billion) by county. We modeled allocations to acute care hospitals, the largest category of health spending, using 2015 Cost Reports. We attributed hospital funding to counties by share of 2015 Medicare fee-for-service admissions, and assumed nonhospital entities (for whom revenue data were not available) were proportional.
We regressed measures of need on relief funding and calculated R2 to measure how needs varied with funding allocation. We repeated regressions with an indicator for counties with the largest Black population fraction (top quartile). We used Stata version 16.1 (StataCorp). Two-sided P < .05 defined statistical significance. This study was approved by the National Bureau of Economic Research Institutional Review Board.
The Figure shows relief funding reflected hospital revenues (R2 = 0.37) more than COVID-19 burden (R2 = 0.13), comorbidities (R2 = 0.13), or hospital financial health (R2 = 0.02).
Across 3124 counties, mean (SD) relief funding per resident was $411 ($277). Disproportionately Black counties (29.6% Black, N = 781) received $126 more funding ($506 vs $380, P < .001) than other counties (2.3% Black, N = 2343). However, among counties receiving the same funding, disproportionately Black counties had higher COVID-19 burden (mean z score: +0.50 SD, P < .001), more comorbidities (+0.73 SD, P < .001), and worse hospital finances (–0.12 SD, P < .001) than other counties (Table). Differences in all individual metrics were also large and statistically significant, including more COVID-19 cases and deaths (+404.0 and +17.6 per 100 000, both P < .001), more non–COVID-19 deaths (+33.3 per 100 000, P < .001), lower operating margins (–0.90%, P = .02), and less cash on hand (–15.2 days, P = .001).
Tying relief funding to revenue resulted in allocations largely unrelated to health or financial needs. It also meant disproportionately Black communities received the same level of relief funding as counties with less health and financial need. Although race may co-vary with socioeconomic status or education, it is unique in having special protections under the law. The findings suggest the relief funding allocation may have a “disparate impact” on Black populations, a legal concept referring to policies that negatively affect a protected group, even if they do not explicitly use information about that group.
Study limitations include that it relied primarily on estimated rather than actual disbursement, which may vary with changes in policy or COVID-19 burden, and that nonhospital revenues were not available.
Policy makers should consider aligning funding with measures of need rather than revenue, which would increase both equity and economic efficiency.
Corresponding Author: Ziad Obermeyer, MD, School of Public Health, University of California, Berkeley, 2121 Berkeley Way, Berkeley, CA 94704 (firstname.lastname@example.org).
Accepted for Publication: July 24, 2020.
Published Online: August 7, 2020. doi:10.1001/jama.2020.14978
Author Contributions: Ms Kakani and Dr Chandra had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: Kakani, Chandra, Obermeyer.
Drafting of the manuscript: Kakani, Chandra, Obermeyer.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Kakani, Chandra, Obermeyer.
Administrative, technical, or material support: Chandra.
Supervision: Mullainathan, Obermeyer.
Conflict of Interest Disclosures: Dr Chandra reported serving on the Congressional Budget Office’s Panel of Health Advisors as well as serving as an advisor to SmithRx, Kyruus, and Health Engine and being an academic affiliate of Analysis Group. Dr Obermeyer reported having equity interest in Berkeley Data Ventures, a consultancy that applies machine learning to health care problems, including ways to expand testing for severe acute respiratory syndrome coronavirus 2. No other disclosures were reported.
Funding/Support: This project was supported by the National Institute on Aging grants P01-AG005842 (all authors) and T32-AG000186 (Ms Kakani).
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Create a personal account or sign in to: