Ethnic and racial differences in self-reported symptoms, health status, activity level, and missed work at 3 and 6 months following SARS-CoV-2 infection

Introduction Data on ethnic and racial differences in symptoms and health-related impacts following SARS-CoV-2 infection are limited. We aimed to estimate the ethnic and racial differences in symptoms and health-related impacts 3 and 6 months after the first SARS-CoV-2 infection. Methods Participants included adults with SARS-CoV-2 infection enrolled in a prospective multicenter US study between 12/11/2020 and 7/4/2022 as the primary cohort of interest, as well as a SARS-CoV-2-negative cohort to account for non-SARS-CoV-2-infection impacts, who completed enrollment and 3-month surveys (N = 3,161; 2,402 SARS-CoV-2-positive, 759 SARS-CoV-2-negative). Marginal odds ratios were estimated using GEE logistic regression for individual symptoms, health status, activity level, and missed work 3 and 6 months after COVID-19 illness, comparing each ethnicity or race to the referent group (non-Hispanic or white), adjusting for demographic factors, social determinants of health, substance use, pre-existing health conditions, SARS-CoV-2 infection status, COVID-19 vaccination status, and survey time point, with interactions between ethnicity or race and time point, ethnicity or race and SARS-CoV-2 infection status, and SARS-CoV-2 infection status and time point. Results Following SARS-CoV-2 infection, the majority of symptoms were similar over time between ethnic and racial groups. At 3 months, Hispanic participants were more likely than non-Hispanic participants to report fair/poor health (OR: 1.94; 95%CI: 1.36–2.78) and reduced activity (somewhat less, OR: 1.47; 95%CI: 1.06–2.02; much less, OR: 2.23; 95%CI: 1.38–3.61). At 6 months, differences by ethnicity were not present. At 3 months, Other/Multiple race participants were more likely than white participants to report fair/poor health (OR: 1.90; 95% CI: 1.25–2.88), reduced activity (somewhat less, OR: 1.72; 95%CI: 1.21–2.46; much less, OR: 2.08; 95%CI: 1.18–3.65). At 6 months, Asian participants were more likely than white participants to report fair/poor health (OR: 1.88; 95%CI: 1.13–3.12); Black participants reported more missed work (OR, 2.83; 95%CI: 1.60–5.00); and Other/Multiple race participants reported more fair/poor health (OR: 1.83; 95%CI: 1.10–3.05), reduced activity (somewhat less, OR: 1.60; 95%CI: 1.02–2.51; much less, OR: 2.49; 95%CI: 1.40–4.44), and more missed work (OR: 2.25; 95%CI: 1.27–3.98). Discussion Awareness of ethnic and racial differences in outcomes following SARS-CoV-2 infection may inform clinical and public health efforts to advance health equity in long-term outcomes.

Disparities in recovery after SARS-CoV-2 infection remain largely under-explored (23)(24)(25).Limitations in the few studies that have reported on ethnic and racial differences in recovery from SARS-CoV-2 include heterogeneity in follow-up duration and definition of post-COVID conditions (26)(27)(28), inconsistency of findings (29)(30)(31)(32), limited focus on symptom presence (as opposed to impact) (29)(30)(31), and lack of adjustment for potential confounding by social health determinants (29,30).We sought to evaluate symptoms and health-related impacts following SARS-CoV-2 infection by ethnicity and race to inform effective and equitable health interventions.

Study design and participant recruitment
This is a secondary analysis of data from the Innovative Support for Patients with SARS-CoV-2 Infections Registry (INSPIRE), a multicenter, longitudinal cohort study of the sequelae of SARS-CoV-2 in the United States.Adult participants were followed prospectively with patient-reported outcomes collected every 3 months via survey and linked to digital health data.Additional methods of the parent study were described previously (33).In this study, we focus on reporting results among SARS-CoV-2-positive participants to address our primary objective, namely to assess for differences in symptoms and health-related impacts by ethnicity and race following a first SARS-CoV-2 infection.The SARS-CoV-2negative cohort was included in the analysis to account for non-SARS-CoV-2 impacts.

Outcomes
The CDC's Person Under Investigation for SARS-CoV-2 symptom list was used to assess 21 COVID-19-like symptoms and/or "other symptoms" at enrollment, 3 months, and 6 months (37).Health status (5-point scale, excellent to poor), activity level compared to before SARS-CoV-2-like symptoms (same, somewhat less, and much less), and missed work due to health reasons in the past 3 months (0-5 workdays, 6-10 workdays, 10-20 workdays, up to 4 weeks, and do not work; overlap of intervals reflects the answer options presented in the survey) were assessed at 3 and 6 months.

Statistical methods
We described socio-demographic and clinical characteristics across ethnicity and race groups and displayed frequency counts by initial SARS-CoV-2 test results.Outcomes were described at enrollment, 3 months, and 6 months after SARS-CoV-2 infection for each ethnic and racial group.
We used generalized estimating equations (GEE) logistic regression to model the association between ethnicity and race (separately) and study outcomes (symptoms, health status, activity level, and missed work) at 3 and 6 months.We leveraged data from the full dataset (including those with and without acute SARS-CoV-2) to fit GEE models with robust standard errors.Independent variables include (1) demographic characteristics (age, gender, education, and family income), social determinants of health, tobacco use, substance use, pre-existing health conditions, and COVID-19 vaccination status (vaccinated or not vaccinated before the index SARS-CoV-2 test); and (2) interactions between ethnicity or race and time point, ethnicity or race and SARS-CoV-2 infection status, and SARS-CoV-2 infection status and time point.Time point was modeled as a categorical variable to account for a non-linear trajectory.Social determinants of health were considered a binary variable coded as "any problem" (vs.no problem) if housing, food security, access to utilities, or access to transportation were unstable (Supplementary Material 1).Substance use was included as present if there was a 'moderate to severe problem' with at least one substance.Individual comorbidities were included as indicator variables.Where appropriate, variable subgroups were collapsed to create larger subgroups.
Using the GEE models, we calculated marginal odds ratios, for brevity referred to as odds ratios hereafter, of individual symptoms (yes, no) by comparing each ethnicity and racial group to the corresponding referent group (non-Hispanic for ethnicity and white for race).For the outcome 'health status' , we estimated the odds ratio of being in 'very good' or 'excellent' health and being in 'fair' or 'poor' health compared to 'good' health.For activity level, the odds ratio of being 'somewhat less' and being 'much less' able to do activities was estimated compared to being the 'same as before' .For missed work, the odds ratio of missing >5 workdays in the prior 3 months due to health reasons was compared to '0 to 5 days' .
We did not adjust for multiple comparisons, given the exploratory nature of this study.All tests were two-sided with an alpha criterion of 0.05.Statistical analyses were performed using the GENMOD procedure of SAS 9.4 (SAS Institute Inc., Cary, NC).Additional information about our GEE methods is available (Supplementary Material 2).

Human subjects approval
This study was approved by the Institutional Review Boards of all eight study sites (33).

Results
The participant flow diagram is shown in Figure 1.Three-month data were available for all participants (N = 3,161), and 6-month data were available for 1,771 participants.Of the 3,161 total participants, ethnicity was reported by 3,155 participants, and race was reported by 3,133 participants.There were differences by ethnicity and race in the proportion of participants completing the follow-up surveys, with lower completion among Hispanic compared to non-Hispanic participants and among Black participants compared to other race groups (Supplementary Material 3).

Participant characteristics
Among 2,354 SARS-CoV-2-positive participants with ethnicity data, 330 (14.0%) were Hispanic and 2,024 (86.0%) were non-Hispanic.Among 801 SARS-CoV-2-negative participants with ethnicity data, 132 (16.5%) were Hispanic and 669 (83.5%) were non-Hispanic.Assessing both SARS-CoV-2-positive and SARS-CoV-2-negative participants, compared with non-Hispanic participants, Hispanic participants were younger, less educated, more likely never married, and had lower family income (Table 1).Hispanic participants were more likely to lack health insurance, experience housing and/or food insecurity, have limited access to utilities and transportation, and be non-health essential workers working outside the home throughout the pandemic.Hispanic participants were more likely to report prescription abuse and pre-existing obesity and less likely to report hypertension and having no pre-existing health conditions.The proportion of SARS-CoV-2-positive participants was similar among Hispanic and non-Hispanic participants (75.2% vs. 71.4%,respectively, value of p = 0.089).Hispanic participants were less likely to be vaccinated against COVID-19, and the two groups utilized different COVID-19 testing sites (Table 1).
Among 2,341 SARS-CoV-2-positive participants with race data, 258 (11.0%) were Asian, 186 (7.9%) were Black, 232 (9.9%) were Other/ Multiple races, and 1,665 (71.1%) were white (Table 2).Among 792 SARS-CoV-2-negative participants with race data, 117 (14.8%) were Asian, 104 (13.1%) were Black, 64 (8.1%) were Other/Multiple races, and 507 (64.0%) were white.Assessing both SARS-CoV-2-positive and SARS-CoV-2-negative participants, Black participants had the highest prevalence of low family income, lack of health insurance, housing insecurity, and food insecurity, limited utility access, and limited transportation access, followed by Other/Multiple race participants in most instances.Black participants also had the highest prevalence of pre-existing health conditions.Employment in essential services was highest among Black and Other/Multiple race participants, and healthcare setting employment was highest among Asian and Other/ Multiple race participants.Black participants were most likely to be tested for COVID-19 in a hospital setting, and COVID-19 vaccination was lowest among Black and Other/Multiple race participants.

Symptoms, health status, activity level, and work among SARS-CoV-2-positive participants
At each time point, the reported prevalence of symptoms, health status, activity level, and missed work among SARS-CoV-2-positive participants varied by ethnicity and race (Table 3; results for the SARS-CoV-2-negative cohort included in Supplementary Material 4).

Association between ethnicity and race and study outcomes among SARS-CoV-2-positive participants
Figure 2 presents adjusted odds ratios for having 21 COVID-like symptoms or "other symptoms" (22 symptoms queried total) comparing Hispanic to non-Hispanic participants and comparing different races to white participants who tested positive for SARS-CoV-2 at 3 and 6 months (entire cohort results are included in Supplementary Materials 5.1-5.3GEE adjusted odds ratio output).
Evaluating health status, activity level, and missed work at 3 and 6 months after SARS-CoV-2 infection, some minoritized groups had worse health status, less physical activity, and more missed days of work.At 3 months, Hispanic compared to non-Hispanic participants were more likely to report fair/poor health (OR: 1.94, 95%CI: 1.36-2.78)and less activity (somewhat less, OR: 1.47, 95%CI: 1.06-2.02;much less, OR: 2.23, 95%CI: 1.38-3.61)(Figure 2, estimates reported in Supplementary Material 5.1).These differences in health and activity level by ethnicity were not found at 6 months (Figure 2, estimates reported in Supplementary Material 5.1).There were no significant differences in missed work at 3 or 6 months by ethnicity.

The relative importance of ethnicity and race in driving study outcomes compared to other covariates
The adjusted GEE models (fit using the complete dataset of SARS-CoV-2-positive and negative participants) were examined to explore associations between included covariates and study outcomes to gauge the relative importance of ethnicity and race, respectively, in driving study outcomes (Supplementary Materials 6.1-6.3 summary plots of GEE model parameter estimates).Adjusted GEE parameter estimates demonstrate a broad positive association between SARS-CoV-2positive status, older age, female gender, any problem in social determinants of health, asthma, and lack of COVID vaccination and the 22 symptoms assessed (Supplementary Materials 6.1, 6.2 summary plots of GEE model parameter estimates).By contrast, infrequent associations between ethnicity or race and symptoms were found when adjusting for other covariates.
Examining associations of covariates that were adjusted for in the model with health status, activity level, and missed work, prominent associations were found between identifying as female or transgender, any problem in social determinants of health, asthma, and obesity and worse study outcomes (Supplementary Material 6.3 summary plot of GEE model parameter estimates).

Discussion
In this prospective longitudinal cohort of individuals with acute SARS-CoV-2-like symptoms, there were few differences in adjusted odds of symptoms by ethnicity or race at 3 and 6 months among SARS-CoV-2-positive participants.However, there were differences in health status, activity level, and missed work.By ethnicity, at 3 months, Hispanic compared to non-Hispanic participants had worse health and lower activity levels; these differences were not present at 6 months.By race, at 3 months, Other/Multiple races compared to white participants had worse health and lower activity levels; at 6 months, these differences persisted.Additionally, at 6 months, higher odds of worse health were found for Asian participants and of missed work for Black and Other/Multiple race participants compared to white participants.The definition currently used to identify post-COVID conditions (i.e., Long COVID) is limited in scope to continuing or developing "signs, symptoms, and conditions" following acute SARS-CoV-2 infection.Notably, the differential impacts on participants' lives by ethnicity and race identified in this study would not be captured within the current definition of post-COVID conditions.A broadening of our understanding of post-COVID conditions may be necessary to fully capture the health-related consequences of SARS-CoV-2.The strength of the association between health status, activity level, and missed work following acute SARS-CoV-2 illness and ethnicity or race is eclipsed by the strength of the association between these impacts and other determinants that we adjusted for in our model, including pre-existing health conditions and social determinants of health.
Few prior studies have evaluated longer-term sequelae of COVID-19 through the lens of ethnicity and race.Several studies used a threshold of ≥28 days to define Long COVID symptoms (28)(29)(30).We assessed the presence of SARS-CoV-2-like symptoms at least 3 months after initial infection in accordance with the current World Health Organization definition of Long COVID (38).We accounted for known ethnic and racial disparities in social determinants of health, adjusted for demographic characteristics and pre-existing health conditions, controlled for non-SARS-CoV-2 impacts through the inclusion of participants testing negative for SARS-CoV-2, and considered not only differences in persisting symptoms but other overall health measures as well.Associations between SARS-CoV-2like symptoms and SARS-CoV-2 infection status, time point from the onset of acute symptoms, pre-existing health conditions, and lack of vaccination found in our GEE models are consistent with what is known in the literature, supporting the validity of our results.
Others have reported inconsistent associations between ethnicity and race and Long COVID symptoms.The Arizona CoVHORT study and a study of American SARS-CoV-2-positive adults who tested positive during the Omicron surge reported no significant

Drug use
Daily or near daily 0 (0.0) 0 (0.0) 1 (0.4) 7 (0.4) 8 (0.3) 1 (0.9) 2 (1.9) 1 (1.6) 2 (0.4) 6 (0.8) Weekly 0 (0.0) 1 (0.5) 1 (0.4) 4 (0.2) 6 (0.3) 0 (0.0) 0 (0.0) 0 (0.0) 1 (0.2) 1 (0.1) Monthly 0 (0.0) 1 (0.5) 5 (2.2) 5 (0.3) 11 (0.5) 0 (0.0) 0 (0.0) Conversely, a U.S. Veterans Affairs EHR-based cohort study found differences between Black and white participants at 6 months, though not to the detriment of one particular group (31).The COVID States Project, a 6-weekly internet survey conducted in all 50 states and the District of Columbia between February 2021 and July 2022 (N > 16,000), found that Hispanic, Other, and white participants testing positive for SARS-CoV-2 were more likely than Asian participants to report Long COVID symptoms (32).Multiple prior studies found ethnic and racial minoritized populations to have a higher risk of Long COVID (28,(39)(40)(41).Some investigations have indicated that differences by ethnicity and race might be partially accounted for by other factors.In the RECOVER Program, a retrospective study using EHR data of participants with and without COVID-19, Black and Hispanic participants experienced higher symptom burden and a different distribution of symptoms/conditions 31-180 days after testing positive for SARS-CoV-2 than white participants (27); however, adjusting for neighborhood-level socioeconomic status attenuated several differences.In the University of California Los Angeles COVID Ambulatory Monitoring Program, a prospective cohort study of adults with SARS-CoV-2, no significant difference in symptoms 60 days after acute illness was found by ethnicity or race after adjusting for other factors, including demographic/clinical characteristics, insurance type, social vulnerability index, and baseline function (26).Differences that we observed among ethnic and racial minoritized populations in health-related outcomes 3 and 6 months following acute SARS-CoV-2 illness might be explained by additional factors, beyond those adjusted for in this study.Several potential factors are described in the literature on health disparities.Socioeconomic deprivation has independently been associated with a higher risk of Long COVID and may mediate disparities in SARS-CoV-2 impact for ethnic and racial minoritized populations (42).Higher loss of work days may be driven by the overrepresentation of minoritized populations in physically demanding frontline industries without the option to work from home (6,43,44).Poor health outcomes may result from barriers to care, including inadequate health insurance and medical mistrust (6,(43)(44)(45)(46).Mistrust and fear have been shown to deter ethnic and racial minoritized individuals with persistent SARS-CoV-2 symptoms from seeking care, compounding structural and systemic barriers (47).Ethnic and racial inequities in access to care have been illustrated in an administrative claims study, which showed that for Asian, Black, and Hispanic patients, a significantly longer time elapsed between initial infection and Long COVID diagnosis than for non-Hispanic white patients (48).Activity levels among ethnic and racial minoritized groups have been associated with differences in the built environments where they live (49)(50)(51)(52)(53)(54)(55).Finally, sequelae of COVID-19 among ethnic and racial minoritized populations may be driven by institutional, cultural, and structural racism.Experiences of discrimination adversely impact mental health and physical health through inflammation, telomere shortening, cortisol dysregulation, and increased allostatic load (56)(57)(58).Experiences of discrimination have been shown to negatively affect the quality of care and form a barrier to seeking help (59,60).Further research is needed to understand the remaining variation in SARS-CoV-2's impact on health status, activity level, and missed work by ethnicity and race.
Our study has several limitations.First, various ethnic and racial subgroups had small sample sizes, which reduced precision in identifying differences within these subgroups.Second, sparse data precluded adjustment for insurance and frontline worker status in GEE analysis.Third, individuals who agreed to participate in this study may not have been representative of their larger ethnic and racial subgroups.Fourth, representativeness across survey time points may have been further impacted by non-response bias.Given the variation in response rates by ethnicity and race, these limitations likely differentially impact the conclusions for specific ethnic and racial groups.Fifth, we did not evaluate important neurological and mental health sequelae of SARS-CoV-2, including cognitive impairment, difficulty concentrating, and anxiety (26,31).Sixth, participants were recruited at different stages of the pandemic, and      the ethnic and racial composition of newly recruited participants fluctuated over time (61).Heterogeneity in symptom profile and SARS-CoV-2 impact by ethnicity and race may be influenced by differences in the dominant SARS-CoV-2 variant at the time of enrollment.Finally, we did not adjust for multiple comparisons, and the generated hypotheses should be tested in confirmatory studies.

Conclusion
Despite similar symptom prevalence, ethnic minoritized populations compared to non-Hispanic populations and racial minoritized populations compared to white populations experience more negative impacts following SARS-CoV-2 infection in terms of health status, activity level, and missed work.Increased focus on understanding drivers of ethnic and racial differences in health impacts may inform approaches to advance health equity after SARS-CoV-2 infection.The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

TABLE 1 (
Continued) Do you work in a healthcare setting such as a hospital, clinic, or nursing/rehabilitation care facility?Are you a non-health essential worker that was asked to work outside the home throughout the pandemic?

TABLE 2
Characteristics by race of adult INSPIRE participants stratified by SARS-CoV-2 status (N = 3,133).