The effect of the coronavirus disease 2019 (COVID-19) pandemic on health care spending in the US has important implications for payers, clinicians, hospitals, health care systems, and patients, and has been the subject of much debate. Understanding how early spending changes varied by disease incidence and implementation of policies to limit transmission can inform expectations about health care spending as the pandemic evolves.
For the first 14 weeks of 2020, we analyzed multipayer deidentified claims from FAIR Health spanning roughly 75% of the commercially insured and 50% of the Medicare Advantage population. Our analysis included fee-for-service (not capitated) claims submitted to FAIR Health by June 25, 2020, and used 2019 data to correct for unreported claims (eAppendix in the Supplement). The study was determined not to be human subjects research and thus not reviewed by the Harvard Medical School Committee on Human Studies.
We calculated weekly aggregate medical spending in total and categorized by type of care and 4 prespecified groups of states defined by cumulative incidence of confirmed cases by April 7 (COVID-19 activity) and timing of social distancing policies1 (1) New York; (2) high activity (excluding New York); (3) low activity, early social distancing; and (4) low activity, late social distancing (eAppendix in the Supplement). We also examined spending for typically elective procedures and COVID-19 inpatient spending (where identifiable by an emergently introduced diagnosis code).2
We focused on changes in spending from week 9 (ending March 3, 2020) to week 14 (ending April 7, 2020). We quantified the extremeness of these changes using a randomization inference approach (eAppendix in the Supplement).
From week 9 to 14, aggregate medical spending decreased by $2.7 billion per week, or 46.0% (Figure 1). Spending decreased across all categories of services except telehealth, ranging from −26.9% for inpatient to −86.2% for ambulatory surgical center care (Figure 2).
Spending reductions were greater in New York (−48.8%) and other high-activity states (−49.9%) than low-activity states, where the reduction was greater in those implementing social distancing policies earlier (−44.2%) rather than later (−39.7%). Identifiable inpatient spending for COVID-19 increased more in high-activity than low-activity states (reaching 25.2% vs 3.5% of week 9 total inpatient spending by week 14), but total inpatient spending fell similarly in high-activity (−25.1%) and low-activity (−26.9%) states. In contrast, spending on elective procedures declined more in high-activity (−85.5%) than low-activity (−76.1%) states. Within each state category, spending reductions for elderly and nonelderly enrollees were largely similar.
All reductions were more extreme than all values produced by permutations of week pairs in 2019; differential percentage reductions in total spending associated with COVID-19 activity and social distancing were more negative than 99.7% and 95.5% of estimates produced by these permutations, respectively.
The early phase of the COVID-19 pandemic was associated with a massive decrease in medical spending for the privately insured. Areas with higher COVID-19 activity had larger spending reductions, as increases in COVID-19 spending were more than offset by decreases in spending on non–COVID-19 care. Spending reductions in high-activity states were similarly substantial for seniors, who are at greater risk of hospitalization for COVID-19. Findings for low-activity states suggest a relationship between elevated concern for viral transmission (manifest as earlier distancing policies) and lower use of medical care. As cases surge in new areas and concerns about transmission prompt states to reimpose restrictions, these findings suggest that health care spending will likely rise and fall inversely with the severity of the pandemic and remain below prepandemic levels for its duration. Resurgent or persistent concerns may limit spending via demand-side or supply-side mechanisms, including patient fear of exposure and effects of infection control measures on faciliy capacity and nonessential care.3
The findings of this study do not reflect intensified COVID-19 testing or pent-up demand from deferred care. As incidence fell after our study period, outpatient utilization rebounded but remained below prepandemic levels through mid-June 2020,4 even for older adults for whom pent-up demand should be greater. Our findings suggest that subsequent COVID-19 surges that started in late June should be associated with corresponding spending reductions.
For commercial insurers, the prospect of net gains in 2020 raises questions about optimal use of unspent premium dollars5; rebates mandated by medical loss ratio regulations are partial and delayed by design. For Medicare Advantage plans, our findings suggest potentially substantial cumulative savings. Medicare could remand these savings to preserve the Trust Fund as tax revenue falls.6
Accepted for Publication: August 14, 2020.
Corresponding Author: J. Michael McWilliams, MD, PhD, Department of Health Care Policy, Harvard Medical School, 180 Longwood Ave., Boston, MA 02115 (mcwilliams@hcp.med.harvard.edu).
Published Online: November 9, 2020. doi:10.1001/jamainternmed.2020.5333
Correction: This article was corrected on December 7, 2020, to fix an error in the author order of the byline.
Author Contributions: Dr McWilliams and Ms Russo 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: All authors.
Drafting of the manuscript: McWilliams, Russo.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: McWilliams, Russo.
Administrative, technical, or material support: Mehrotra.
Supervision: McWilliams.
Conflict of Interest Disclosures: Dr McWilliams reports serving as a paid member of the Academic Advisory Board for FAIR Health. No other conflicts were reported.
Disclaimer: The conclusions reached reflect the interpretation of the authors alone and do not necessarily reflect the views of FAIR Health, Inc.
Additional Contributions: We thank David Cheng, MS, Josh Smolinsky, MS, and Eric Okurowski, MBA, of FAIR Health, Inc, for conducting the data analysis, providing claims expertise, and constructing data visualizations. They were not compensated.