Racial and Ethnic Disparities in COVID-19 Infection and Hospitalization in the Active Component US Military

Abstract

Objectives. To assess COVID-19 disparities in the active component US military with an emphasis on race and ethnicity.

Methods. In this retrospective cohort study, we calculated the incidence of COVID-19 testing, infection, and hospitalization in the active component US military in calendar year 2020.

Results. Overall, 61.3 per 100 population per year were tested for COVID-19, 10.4% of tests were positive, and 1.1% of infected individuals were hospitalized. Non-Hispanic Blacks and Hispanics had a rate of testing for COVID-19 similar to that of Whites but had a higher risk of infection (adjusted risk ratio [ARR] = 1.25 and 1.26, respectively) and hospitalization (ARR = 1.28 and 1.21, respectively).

Conclusions. Although of lower magnitude than seen in civilian populations, racial and ethnic disparities in COVID-19 infection and hospitalizations exist in the US military despite universal eligibility for health care, similar rate of testing, and adjustment for comorbidities and other factors. Simply making health care coverage available may be insufficient to ensure health equity. Interventions to mitigate disparities in the US military should target the patient, provider, health care system, and society at large. (Am J Public Health. 2021;111(12):2194–2201. https://doi.org/10.2105/AJPH.2021.306527)

Soon after the COVID-19 outbreak began in the United States, it became apparent that the pandemic was having a disproportionate effect on persons from Black, Indigenous, and other racial and ethnic groups.^1,2 Early data from the US Centers for Disease Control and Prevention (CDC) indicated that Black and Hispanic/Latino populations had at least 2.6 times the rate of COVID-19 cases and 4.6 times the rate of COVID-19 hospitalizations of non-Hispanic White nonpopulations.³ A growing number of health care systems have reported racial and ethnic disparities in COVID-19 cases and related deaths, further demonstrating that the virus disproportionately affects persons from minority groups.^4–7

The causes of health disparities related to COVID-19 and other conditions are complex, interrelated, and difficult to quantify. Contributing factors arise from all social–ecological levels of society, including individual, interpersonal, organizational, and community levels and include age, comorbidities, education, socioeconomic status, housing, household structure, geography, language proficiency, cultural background, bias, and stereotyping.⁸ Access to care, complex in and of itself, also is a significant driver of health disparities. It includes eligibility for insurance and availability of care (the supply side of access) and acceptance of health recommendations and services (the demand side).⁹ Trust plays an important role in the acceptance of health recommendations on individual and population levels, and research shows that Black persons report lower levels of trust of health care providers than White persons.¹⁰ Previous studies have demonstrated lower uptake of influenza and COVID-19 vaccines among Black persons.^11,12

Health care benefits are available to all military members, their families, and military retirees through the Military Health System (MHS). These benefits provide care to approximately 1.4 million active component (AC) service members and 9.5 million total beneficiaries.¹³ Given the large racially and ethnically diverse population covered, the analysis of health outcomes in the MHS presents a unique opportunity to explore health equality. This is especially true because universal eligibility for health care in the MHS has been purported to eliminate health disparities in some areas, including outcomes following coronary artery bypass grafting and other surgical procedures,¹⁴ cancer,¹⁵ and initiation of treatment of mental health conditions.¹⁶

Our study objective was to assess factors associated with COVID-19 testing, infection, and hospitalization among AC service members, with special emphasis on assessing whether disparities exist by race and ethnicity. We hypothesized that COVID-19 outcomes in the US military would be similar among persons belonging to all racial and ethnic groups because of the provision of universal health care coverage.

METHODS

The Uniformed Services University institutional review board approved this study. The review board waived informed consent because this was a retrospective cohort study and we de-identified the patient data. We followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cohort studies.¹⁷

Population

This was a population-based, retrospective cohort study of the incidence of COVID-19 testing, infection, and hospitalization among all AC service members in service during calendar year 2020. The Armed Forces Health Surveillance Division (AFHSD) maintains the Defense Medical Surveillance System (DMSS), a continuously expanding relational database of military personnel and medical data.¹⁸ For this analysis, we used DMSS to identify the study population, COVID-19 outcomes, and information on demographics and other variables.

Outcomes

AFHSD also maintains a master list of MHS beneficiaries, including AC service members, with reverse transcription-polymerase chain reaction (RT-PCR) or antigen test laboratory–confirmed COVID-19 infection. We updated this list daily, and it is composed of cases identified using RT-PCR and antigen tests for COVID-19 in Composite Health Care System Health Level 7–formatted and MHS Genesis laboratory data extracted by the Navy and Marine Corps Public Health Center, as well as medical event reports of RT-PCR laboratory–confirmed COVID-19 infection from the Disease Reporting System Internet. We derived testing and infection outcomes from this COVID-19 master list.

We considered patients to have been hospitalized if hospitalization was documented in the Disease Reporting System Internet (assessed via chart review for approximately 50% of COVID-19 cases tracked in the Disease Reporting System Internet) or if inpatient encounter data extracted from DMSS indicated a diagnosis of COVID-19–like illness in diagnosis code position 1 or 2 within 30 days of a COVID-19 infection. We empirically generated the 30-day window as the point past which very few additional hospitalizations occurred. Given the typical clinical course of COVID-19, hospitalizations past this window were likely attributable to other causes.

Exposure

We obtained self-reported race and ethnicity data from the DMSS records collected at the time of entry into military service and categorized this as non-Hispanic White, non-Hispanic Black, Hispanic, other, and unknown or missing.

Other Variables

We used the DMSS to obtain each service member’s age (younger than 20 years, 5-year age groups 20–44 years, and 45 years or older), sex (male or female), service branch (Army, Navy, Air Force, Marines, Coast Guard), rank (enlisted or officer), education level (high school or less, some college, bachelor’s or advanced degree, and other or unknown), marital status (single and never married, married, or other or unknown), military occupation (combat specific, motor transport, pilot or air crew, repair or engineering, communications or intelligence, health care, and other or unknown), and geographic region of assignment (Northeast, Midwest, South, West, overseas, and unknown or missing). We also used the DMSS to identify comorbidities from administrative records of inpatient and outpatient medical encounters, which include encounters from fixed military treatment facilities as well as outsourced care reimbursed by TRICARE (the health care program of the US Department of Defense [DoD] MHS). We considered an individual to have a comorbidity if they had an inpatient or outpatient encounter containing a diagnosis with an International Classification of Diseases, 10th Revision (Geneva, Switzerland: World Health Organization; 1992 [ICD-10]) code for that comorbidity in any diagnostic position between January 1, 2019 and December 31, 2020. The list of ICD-10 codes, selected based on a review of the existing literature, can be found in Table A (available as a supplement to the online version of this article at http://www.ajph.org).

Statistical Analysis

We calculated adjusted rate ratios for COVID-19 testing using a negative binomial regression model, which we selected over a Poisson or zero-inflated model because it has the lowest Akaike’s information criteria value. We censored person-time at risk at the date of the first COVID-19 test. We used an offset of the log of the follow-up time for each individual to account for different population sizes. For analyses of the risk of testing positive among those tested and the risk of being hospitalized among those infected, we calculated adjusted risk ratios (ARRs) using Poisson regression with robust variance estimation to avoid exaggeration of effect estimates because of violation of the rare disease assumption.¹⁹ We adjusted all models for age, sex, race, service branch, rank, education level, marital status, military occupation, geographic region, and presence of any comorbidity using SAS version 9.4 (SAS Institute, Cary, NC). We selected these covariates a priori on the basis of being potential confounders.²⁰ We did not perform any analysis of effect modification, as we had no prespecified hypothesis for this.

RESULTS

The US active component military population used in this study is shown in Table 1 and is similar to previously published estimates.²¹ There were 694 878 AC service members tested for COVID-19 in 2020, a rate of 61.3 per 100 person-years, of which 10.4% (n = 72 152) tested positive, and 1.1% (n = 846) were hospitalized (Table 1).

TABLE 1—

Characteristics of Active Component US Military Service Members: 2020

	No. (%)
Total	1 361 399 (100.0)
Sex
Male	1 128 236 (82.9)
Female	233 163 (17.1)
Age, y
< 20	73 164 (5.4)
20–24	428 450 (31.5)
25–29	322 394 (23.7)
30–34	222 445 (16.3)
35–39	170 204 (12.5)
40–44	89 981 (6.6)
≥ 45	54 761 (4.0)
Race/ethnicity
Non-Hispanic White	755 302 (55.5)
Non-Hispanic Black	215 414 (15.8)
Hispanic	230 987 (17.0)
Other	131 325 (9.6)
Unknown/missing	28 371 (2.1)
Service
Army	473 296 (34.8)
Navy	335 391 (24.6)
Air Force	329 241 (24.2)
Marines	182 728 (13.4)
Coast Guard	40 743 (3.0)
Rank
Enlisted	1 116 284 (82.0)
Officer	245 115 (18.0)
Education level
High school or less	864 258 (63.5)
Some college	166 168 (12.2)
Bachelor’s or advanced degree	306 116 (22.5)
Other/unknown	24 857 (1.8)
Marital status
Single, never married	600 135 (44.1)
Married	693 161 (50.9)
Other/unknown	68 103 (5.0)
Military occupation
Combat specific	180 668 (13.3)
Motor transport	43 879 (3.2)
Pilot/air crew	46 491 (3.4)
Repair/engineering	410 123 (30.1)
Communications/intelligence	292 532 (21.5)
Health care	115 471 (8.5)
Other/unknown	272 235 (20.0)
Geographic region
Northeast	43 358 (3.2)
Midwest	87 443 (6.4)
South	627 277 (46.1)
West	356 503 (26.2)
Overseas	148 890 (10.9)
Unknown/missing	97 928 (7.2)
Comorbidities
Yes	458 138 (33.7)
No	903 261 (66.3)
COVID-19 outcomes
Tested	694 878 (61.3)a
Infected	72 152 (10.4)b
Hospitalized	846 (1.1)c

^aPer 100 person-years. ^bProportion of those tested. ^cProportion of those infected.

Those who self-reported as non-Hispanic Black or Hispanic had a similar but small and marginally statistically significant increase in testing compared with Whites (for Blacks, ARR = 1.01; 95% confidence interval [CI] = 1.00, 1.02; for Hispanics, ARR = 1.06; 95% CI = 1.04, 1.07), as seen in Table 2 and Table B (available as a supplement to the online version of this article at http://www.ajph.org). Among the population of those who were tested, non-Hispanic Blacks had 1.25 (95% CI = 1.22, 1.27) times the risk of testing positive for COVID-19 and Hispanics had 1.26 (95% CI = 1.24, 1.28) times the risk compared with non-Hispanic Whites after adjusting for covariates (Table 2; Table C, available as a supplement to the online version of this article at http://www.ajph.org). Similarly, all racial/ethnic groups demonstrated a higher risk of hospitalization for COVID-19 than did non-Hispanic Whites, with the highest risk found among those who reported “other” race/ethnicity (ARR = 1.39; 95% CI = 1.10, 1.75), followed by non-Hispanic Blacks (ARR =  1.28; 95% CI = 1.08, 1.53) and Hispanics (ARR = 1.21; 95% CI = 1.01, 1.45; Table 2; Table D, available as a supplement to the online version of this article at http://www.ajph.org).

TABLE 2—

Adjusted Associations of Race and Ethnicity With Testing, Infection, and Hospitalization Among Active Component US Military Service Members: 2020

Race/Ethnicity	Testing, ARRa (95% CI)	Infection, ARRa (95% CI)	Hospitalization, ARRa (95% CI)
Non-Hispanic White	1 (Ref)	1 (Ref)	1 (Ref)
Non-Hispanic Black	1.01 (1.00, 1.02)	1.25 (1.22, 1.27)	1.28 (1.08, 1.53)
Hispanic	1.06 (1.04, 1.07)	1.26 (1.24, 1.28)	1.21 (1.01, 1.45)
Other	1.03 (1.01, 1.04)	1.00 (0.97, 1.02)	1.39 (1.10, 1.75)
Unknown	0.96 (0.94, 0.99)	1.04 (0.99, 1.11)	0.85 (0.47, 1.54)

Note. ARR = adjusted risk ratio; CI = confidence interval.

^aAdjusted for the other variables in the table as well as age, sex, rank, comorbidities, service branch, geographic region, occupation, and marital status.

We also found other health disparities by sex and rank (Tables B–D). Female service members had a modestly higher adjusted rate of testing than did males (ARR = 1.14; 95% CI = 1.13, 1.15), lower risk of infection (ARR = 0.94; 95% CI = 0.93, 0.96), and similar risk of hospitalization. Officers had a lower rate of testing than did enlisted ranks (ARR = 0.89; 95% CI = 0.88, 0.91), a similar risk of infection, and a lower risk of hospitalization (ARR = 0.69; 95% CI = 0.52, 0.91). Finally, although we did not see an association between the presence of any of the listed comorbidities and testing for or infection with COVID-19, the ARR for hospitalization among those with any comorbidity was elevated at 4.67 (95% CI = 3.99, 5.45).

DISCUSSION

By the end of calendar year 2020, 61.3 per 100 person-years of AC military service members had been tested for COVID-19, 10.4% of those tested positive, and 1.1% of those infected were hospitalized. Non-Hispanic Blacks and Hispanics had a similar rate of testing for COVID-19 as Whites, but they had a higher risk of infection (ARRs = 1.25 and 1.26, respectively) and hospitalization (ARRs = 1.28 and 1.21, respectively). Officer rank, a military correlate of higher socioeconomic status, was not associated with infection but was associated with a 31% lower adjusted risk of hospitalization. These associations persisted despite equal eligibility for health care; despite similar rate of testing; and after adjusting for comorbidities, occupation, and other factors associated with COVID-19. The presence of comorbidities was associated with a large increase in the risk of hospitalization.

This study builds on the emerging literature demonstrating the disproportionate impact of COVID-19 on persons having certain racial and ethnic characteristics. Race and ethnicity have been associated with infection and hospitalizations for COVID-19 in numerous populations, including US military veterans.^4–6,22,23 As of July 2021, the CDC has reported a 2.8 times higher unadjusted rate of hospitalization for both Black persons and Hispanic persons, as compared with non-Hispanic White persons, but only a 1.1 and 1.9 times higher rate of infection, respectively.²⁴ However, associations between race/ethnicity and mechanical ventilation and deaths from COVID-19 have been inconsistent,^22,23 with 1 study actually showing better survival among Black and Hispanic populations.²⁵ We found a higher risk of both infection and hospitalization among Blacks and Hispanics, although these were of lesser magnitude than seen previously in civilian populations. This attenuation is likely to be attributable in part to the military’s provision of universal eligibility for health care, resulting in increased access to care. The regional variation of infection and hospitalization seen in this study is generally consistent with that seen in the United States, although there has been heterogeneity among and in regions over time.²⁶ Racial and ethnic disparities also persisted over time, although they varied by region and became less pronounced over time in each region.

Associations of sex and age with COVID-19 in civilian populations were similar to those identified in this study.⁶ The association of COVID-19 with chronic medical conditions was also similar to that found in previous studies,⁴ although we found a significant association only with hospitalization and not with infection or testing.

Strengths and Limitations

A major strength of this study is the use of a large, well-characterized, and enumerated population with equal eligibility for health care and the ability to identify relevant health events, notably including testing for COVID-19.

The most important limitation is the possible misclassification of outcomes. There may have been cases of testing, infection, and hospitalization that we did not identify. In particular, asymptomatic or mildly symptomatic infections may have gone undiagnosed because of individuals not seeking medical care and because we initially limited testing to symptomatic cases. Because of universal eligibility for care in the MHS, we expected misclassification of outcomes to be nondifferential, and thus the associations seen in this study may be underestimates. Because of differences in the classification of ethnicity by service branch, we could not further differentiate the “other” race/ethnicity category into subcategories (e.g., Native Hawaiian/Pacific Islander or American Indian/Alaska Native).

Further, because operational deployment data were not available for the complete surveillance period at the time of the analysis, we may have misclassified geographic region, particularly for the overseas region, as that did not include deployments to Iraq or Afghanistan. The selection of individuals for testing may also have been biased by the DoD’s tiered testing strategy, which prioritized testing of critical national capabilities, engaged field forces, and forward deployed forces.²⁷ Differential access to surveillance testing likely resulted in some of the associations seen with testing, such as among health care personnel and pilots.

Because the military has different population characteristics, better preexisting health status, and universal health care coverage, the findings from this study may not be generalizable to civilian populations. Finally, the associations seen in this study may have changed over time because of the changing nature of the pandemic. However, including a time component in the models did not significantly change the associations observed (data not shown).

Public Health Implications

Health disparities related to race and ethnicity in the general US population have been attributed to differential access to health care, the presence of comorbidities, current work and living circumstances, and systemic racism and inequities in the underlying society.^7,8,28 This study shows that although they are attenuated compared with civilian populations, significant disparities in COVID-19 health events remain despite the same eligibility for care available in the DoD, similar work and living circumstances, and after adjusting for other important confounding variables such as age and comorbidities. Racial and ethnic health disparities have been identified for many conditions in the underlying US population, so it is not surprising to find that disparities exist in the US military. These disparities are objectionable in and of themselves, and they further pose a serious threat to the ability of the MHS to maintain a medically ready force, which is critical to national security.¹³ The persistence of COVID-19–related health disparities among AC service members calls for deeper exploration and action aimed at improving care among non-Hispanic Black and Hispanic service members.

The military places special emphasis on equality and makes health care universally available through equal eligibility for care. Consistent with the notion that equal access to care ameliorates health disparities, this study demonstrated similar rates of testing among all racial and ethnic groups and reduced disparities in infection and hospitalization compared with civilian populations. The persistence of disparities in infection and hospitalization, however, suggests that the provision of universal health care does not necessarily result in equal access to care or the elimination of health disparities. There may be racially based differences in the perceived threat posed by the pandemic, although available evidence suggests that COVID-19 prevention behaviors do not differ among persons with different racial and ethnic characteristics and that thus individual behaviors do not explain COVID-19 disparities.²⁹ Persistent disparities are likely the result of subtle and complex social and societal mechanisms, such as distrust in the health care system, delays in care, and culturally inappropriate care.^7,8 Although these factors are believed to be largely mitigated in the DoD, this study suggests that the impact of these societal forces persists to some degree even in the military.

The CDC’s COVID-19 Response Health Equity Strategy includes immediately available actions aimed at responding to COVID-19 disparities. The CDC’s priority strategy 1 is to expand the evidence base by collecting and analyzing data relevant to health disparities, which is often lacking or of poor quality in both military and civilian populations.³⁰ Although the military collects race and ethnicity data on service members, these data are often unavailable for other beneficiaries, such as family members and retirees. Availability of these data would allow more robust identification of racial and ethnic factors related to COVID-19 and other conditions and assist with the development of interventions to promote health equity. The Uniformed Services Health Equity Collaboratory at the Uniformed Services University has recently been established to increase the evidence base, improve collaboration, and promote equity in the MHS.³¹

The US military can further optimize the health of service members and other MHS beneficiaries by ensuring equity in all aspects of COVID-19 care. In addition, the military can strive for early, frequent, transparent, and culturally and linguistically appropriate communication related to the pandemic. This has the potential to promote health literacy, engender trust, and improve adherence to and participation in recommended interventions, such as testing, nonpharmaceutical interventions, therapeutics, and vaccination.^7,8 It may also help identify and reduce delays in care, ensure culturally appropriate care, and increase trust in the MHS for all persons, but particularly those groups at greatest risk for health disparities. For the military to completely eliminate health disparities, the underlying societal causes of disparities must be addressed in both the military and the underlying US population from which the US military is drawn. Responses include structurally competent reforms and interventions aimed at eliminating the structural racism that drives many existing health inequities.²⁸

The DoD’s systematic efforts at promoting diversity, equity, and inclusion have historically been focused on employment and leadership opportunities but recently began to take a broader approach that includes the range of attributes, such as enhancing military performance.³² The Uniformed Services Health Equity Collaboratory will continue to engage with civilian, military, and veteran communities to assess and address the role of structural racism and other factors that contribute to health disparities.

Conclusions

The COVID-19 pandemic disproportionately affects Black and Hispanic AC service members despite the same eligibility for health care as White service members. Addressing health disparities is important for designing effective interventions that control the pandemic, ensure military readiness, and achieve health equity. Additional interventions should be targeted at the patient, provider, health care system, and societal levels, including timely, accurate, and consistent public health education aimed at building trust, culturally appropriate care, community-based nonpharmaceutical interventions, and vaccination advocacy efforts.