Assessment of Changes in US Veterans Health Administration Care Delivery Methods During the COVID-19 Pandemic

This cross-sectional study assesses changes in health care delivery methods, including all forms of care either purchased or provided by the Veterans Health Administration (VHA), for VHA enrollees in response to the COVID-19 pandemic.


Introduction
The Veterans Health Administration (VHA) manages an integrated health care system that has expenditures of nearly $100 billion per year and serves more than 9 million enrollees. 1 Like other health care systems, the VHA has faced unprecedented challenges in responding to the COVID-19 pandemic. Although its large size, diverse operating environments, and geographically dispersed patient population make it difficult for the VHA to pivot nimbly and ensure access to care, this health system was able to leverage its existing infrastructure and prior planning to rapidly scale virtual care services (ie, telephone and video) for enrollees in 2020. 2,3 In this study, we took a broad look at how VHA care patterns, including all forms of care either purchased (known as community care) or provided by the VHA, have shifted in association with the COVID-19 pandemic.

Methods
For this cross-sectional study, we extracted records from the US Department of Veterans Affairs Corporate Data Warehouse on all health care encounters purchased or provided by the VHA from January 1, 2019 to March 28, 2021. Encounters were then classified into mutually exclusive categories by location (VHA and community care) and type of care delivered (inpatient, emergency department, urgent care, and outpatient). Patient age, sex, and race and ethnicity demographics were collected from the records to assess whether they reflected the general population of VHA enrollees. Further details on data aggregation are provided in the eMethods in the Supplement.
Encounter data were aggregated by epidemiologic weeks. To estimate the total number of missing encounters in 2020, we performed local polynomial regression of total weekly encounters on prepandemic 2019 numeric weeks and applied the smoothed values (ie, expected total encounters per week if prior levels had persisted) to 2020 numeric weeks. We subtracted the smoothed values from actual 2020 totals and summed the differences across all weeks.
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The Stanford University Institutional Review Board approved the study. This was an observational study without any direct patient contact and was considered to be of minimal risk; therefore, a waiver of informed consent was obtained.

Results
The VHA provided or paid for 179.5 million encounters for 6 737 274 unique patients between    The

Discussion
In this cross-sectional study of health care use patterns of VHA enrollees, we observed substantial and persistent changes in the number of virtual and community care encounters over the year since the start of the COVID-19 pandemic in 2020. Limitations of this study include a likely lag in adjudication of more recent community care claims, which indicates that current estimates of community care encounters may be underestimated.
Like other systems, the VHA experienced large reductions in care early in the pandemic and was well positioned to quickly transition to providing virtual care. 2,3 However, the number of encounters for VHA in-person care declined substantially more than that for community care encounters and has not yet recovered. Our results indicate that the VHA has likely adopted a more conservative reopening strategy compared with community care providers. These providers tend to have different financial incentives to resume in-person care, and many returned to close to prepandemic inpatient and outpatient levels by September 2020. 4,5 In the wake of concerns about health care access, the VHA has steadily increased spending on community care, and our results indicate that existing trends pushing the VHA toward being a mixed payer and provider may have accelerated.

ARTICLE INFORMATION
Accepted for Publication: August 10, 2021.