Changes in Outpatient Health Care Use After COVID-19 Infection Among Veterans

Abstract

Importance: The association of COVID-19 infection with outpatient care utilization is unclear. Many studies reported population surveillance studies rather than comparing outpatient health care use between COVID-19-infected and uninfected cohorts.

Objective: To compare outpatient health care use across 6 categories of care (primary care, specialty care, surgery care, mental health, emergency care, and diagnostic and/or other care) between veterans with or without COVID-19 infection.

Design, setting, and participants: In a retrospective cohort study of Veterans Affairs primary care patients, veterans with COVID-19 infection were matched to a cohort of uninfected veterans. Data were obtained from the Veterans Affairs Corporate Data Warehouse and the Centers for Medicare & Medicaid Services Fee-for-Service Carrier/Physician Supplier file from January 2019 through December 2022. Data analysis was performed from September 2022 to April 2023.

Exposure: COVID-19 infection.

Main outcomes and measures: The primary outcome was the count of outpatient visits after COVID-19 infection. Negative binomial regression models compared outpatient use over a 1-year preinfection period, and peri-infection (0-30 days), intermediate (31-183 days), and long-term (184-365 days) postinfection periods.

Results: The infected (202 803 veterans; mean [SD] age, 60.5 [16.2] years; 178 624 men [88.1%]) and uninfected (202 803 veterans; mean [SD] age, 60.4 [16.5] years; 178 624 men [88.1%]) cohorts were well matched across all covariates. Outpatient use in all categories (except surgical care) was significantly elevated during the peri-infection period for veterans with COVID-19 infection compared with the uninfected cohort, with an increase in all visits of 5.12 visits per 30 days (95% CI, 5.09-5.16 visits per 30 days), predominantly owing to primary care visits (increase of 1.86 visits per 30 days; 95% CI, 1.85-1.87 visits per 30 days). Differences in outpatient use attenuated over time but remained statistically significantly higher at 184 to 365 days after infection (increase of 0.25 visit per 30 days; 95% CI, 0.23-0.27 visit per 30 days). One-half of the increased outpatient visits were delivered via telehealth. The utilization increase was greatest for veterans aged 85 years and older (6.1 visits, 95% CI, 5.9-6.3 visits) vs those aged 20 to 44 years (4.8 visits, 95% CI, 4.7-4.8 visits) and unvaccinated veterans (4.5 visits, 95% CI, 4.3-4.6 visits) vs vaccinated veterans (3.2 visits; 95% CI, 3.4-4.8 visits).

Conclusions and relevance: This study found that outpatient use increased significantly in the month after infection, then attenuated but remained greater than the uninfected cohorts' use through 12 months, which suggests that there are sustained impacts of COVID-19 infection.

Conflict of interest statement

Conflict of Interest Disclosures: Dr Hynes reported receiving a Patient-Centered Outcomes Research Institute subcontract from the University of North Carolina at Chapel Hill; grants from Pacific Source Community Services, Inc, and the David and Lucille Packard Foundation; and consulting fees from Quality Insights outside the submitted work. Dr Maciejewski reported owning Amgen stock through his spouse’s employment. No other disclosures were reported.

Figures

Figure 1.. Derivation of COVID-19–Infected and Matched Uninfected Cohorts

aThere were 1 222 272 comparators matched to more than 1 case. In the numbers presented in this figure, comparators who were matched to more than 1 infected case have been counted as many times as they appeared in the data set. The number of unique comparators at each step in the flow diagram are presented in the footnotes below. bThose excluded in the prior matching included those with no Care Assessment Need score; no primary care practitioner; missing height, weight, or implausible value; age missing or implausible value; zip code missing or not in Washington, DC, or 50 states; had a Medicare COVID-19 diagnosis; and no suitable match. Those who became a case in the same month were also excluded among the comparators. cThe unique number of uninfected comparators matched to an infected case was 3 014 091. dThe unique number of matched comparators with unique identifiers was 3 012 243. eThe unique number of alive matched comparators at the index date was 3 006 856. fThe unique number of matched comparators who were dropped was 2 798 500. gThe unique number of matched comparators in initial best match was 201 070. Of the comparators in the initial best match, 6091 were matched to more than 1 infected case. hThere were 5052 cases (and their matched comparators) excluded from analysis because they had an updated infection date in a prior infection month. An additional 124 comparators (and their matched case) were excluded from analysis because they had an updated infection date in a prior month or the same month as their matched case.

Figure 2.. Unadjusted Weekly Outpatient Visit Use for COVID-19–Infected and Uninfected Veterans, by Category of Outpatient Visit, for the 52 Weeks Before and 52 Weeks After Infection

Figure 3.. Difference-in-Difference Estimates of the Association of COVID-19 Infection With Outpatient Visits Over 30 Days Across 3 Postinfection Periods, by Category and Mode (Telecare vs In-Person) of Outpatient Visit

Figure 4.. Difference-in-Difference Estimates of the Association of COVID-19 Infection With Total Outpatient Visits Over 30 Days Across 3 Postinfection Periods, by Patient Subgroups

The estimates are the difference between infected patients and uninfected patients in the difference from baseline to each of the defined follow-up periods, for each of the defined subgroups. Models included fixed effects for Medicare eligibility, which was imbalanced between infected and uninfected cohorts within many subgroups. Models for comorbidity subgroups additionally controlled for smoking status, and models for vaccination subgroups additionally controlled for age and risk of hospitalization or death as measured by the Care Assessment Need score because of standardized mean differences >0.1 in these variables for these subgroups.