Social determinants of health exacerbate disparities in COVID-19 illness severity and lasting symptom complaints

BACKGROUND Increasing reports of long-term symptoms following COVID-19 infection, even among mild cases, necessitates systematic investigation into the prevalence and type of lasting illness. Notably, there is limited data regarding the influence of social determinants of health, like perceived discrimination and economic stress, which may exacerbate COVID-19 health risks. The primary goals of this study are to test the bearing of subjective experiences of discrimination, financial security, and quality of care on illness severity and lasting symptom complaints. METHODS 1,584 recovered COVID-19 patients that experienced mild to severe forms of the disease provided information about their illness, medical history, lasting symptoms, and psychosocial information. Prevalence data isolated differences in patients infected early versus late in the pandemic. Path analyses examined hypothesized associations between discrimination, illness severity, and lasting symptoms. Post hoc logistic regressions tested social determinants hypothesized to predict neurological, cognitive, or mood symptoms. RESULTS 70.6% of patients reported presence of one or more lasting symptoms after recovery. Neural systems were especially impacted, and 19.4% and 25.1% of patients reported mood or cognitive/memory complaints, respectively. Path models demonstrated that frequency and stress about experiences of discrimination predicted increased illness severity and increased lasting symptom count, even when adjusting for sociodemographic factors and mental/physical health comorbidities. Notably, this effect was specific to stress related to discrimination, and did not extend to general stress levels. Further, perceived but not objective socioeconomic status (SES) was associated with increased lasting symptom complaints after recovery. Finally, associations between discrimination and illness differed with individual perceptions about quality of medical care. CONCLUSIONS Lasting symptoms after recovery from COVID-19 are highly prevalent and neural systems are significantly impacted. Importantly, psychosocial factors (perceived discrimination and perceived SES) can exacerbate individual health risk. This study provides actionable directions for improved health outcomes by establishing that sociodemographic risk and medical care influence near and long-ranging health outcomes.


Title: Social determinants of health exacerbate disparities in COVID-19 illness severity and lasting symptom complaints
Authors: Moriah E. Thomason Figure S1) VI. Path model results (Table S4)

Data validation and exclusion criteria
Data assurance and quality checking were performed using R version 4.0.2. Quality of patient responses were evaluated by isolating implausible and/or inconsistent responses. For example, a height feet value greater than 7, or a height inches value greater than 12 were considered implausible, and conflicting self-reported 'date of birth' (DOB) and 'current age' were considered inconsistent. Respondents with implausible or inconsistent responses were excluded (N=72). Exclusions were also made on the basis of self-reporting not being ill with COVID-19 (N=65), incomplete survey data (N=483), and date of illness provided falling outside of February 2020 to March 2021. All data presented herein are based on the resulting sample of 1,584 recovered patients described in Table S1, below. Variation in illness severity in the sample is depicted in Figure 1, main text.

Assessment of lasting symptoms, life disruption and illness anxiety
The NCIPR survey includes questions about a broad array of lasting symptoms (n = 28), covering multiple functional systems, and also questions specific to cognitive/memory and mood complaints following infection. Magnitude of lasting symptoms was determined by summing the number of endorsed lasting complaints, and for post hoc analyses, lasting complaints were subdivided in to general complaints and central nervous system (CNS) specific complaints. The latter included fatigue, cognitive symptoms, mood symptoms, change in the way things taste, change in the way things smell, and sleep disturbance. In addition, individual NCIPR questions addressed COVID-illness life disruption and COVID-illness anxiety using a 6-point Likert scale ranging from none (0) to extreme (5).

Discrimination
Subjective experiences of discrimination were assessed using questions from the Perceived Discrimination Scale by Williams and colleagues. [2] Participants answered the following multiple choice items: (1) In your day-to-day life, have you experienced discrimination?; (2) What do you think was the reason(s) for this/these experience(s)? Pick as many as apply (check all that apply); and (3) Over your entire lifetime, how stressful have experiences of unfair treatment or discrimination usually been for you? Heading and item response choices are provided in Table S2. Table S2. Questions assessing perceived discrimination frequency, type, and associated stress

Instructions
These are some questions about discrimination that you may or may not experience in your day-to-day life. By discrimination, we mean being treated unfairly because of your race, ethnicity, income level, social class, sex, gender, age, sexual orientation, physical appearance, or religion.

Socioeconomic status
A measure of cumulative objective socioeconomic status (SES) was generated by standardizing and summing the following demographic variables: household income-to-needs ratio (i.e., income relative to household size), education level, stability of housing, and receipt of public assistance. Next, a measure of perceived SES was computed by standardizing and summing financial satisfaction, financial worries, perceived financial stability, and the MacArthur ladder of perceived social standing. Confirmatory factor analyses were used to verify fit of SES composite variables, which indicated excellent fit for both variables (objective SES: RMSEA = 0.029, CFI = .985, χ2 = 177.53, p < 0.001; perceived SES: RMSEA = 0.0, χ2 = 632.03, p < 0.001).
Demographic questions and response options used to derive these measures are available at: https://osf.io/82rkj, via the NCIPR Demographic Survey. .00008). Direction of these effects was such that patients in the early infection group reported more severe illness and more long-COVID symptoms. Early versus late groups were also significantly different in anxiety about illness (p = 9.2E-8) and ratings of COVID illness-related life disruption (p = .001), where the early infection group reported increased concerns anxiety and life disruption. Analyses controlled for multiple comparisons (see Table S3). Select outcomes are plotted for early and late infection groups in Figure S1, below.

Path model results
Full results from the observed path model are reported in Table S4.  Figure S2. Analytical control model testing moderation by general stress levels. Current stress similarly moderates the impact of discrimination frequency on lasting symptoms, but does not moderate the direct effect of discrimination frequency on illness severity, nor the indirect association between discrimination frequency, illness severity, and lasting symptoms. All coefficients are standardized. On all pathways, we controlled for discrimination stress, non-White race, objective SES score, perceived SES score, history of mood/anxiety disorder, history of diabetes/heart disease, COVID-illness life disruption, COVID-illness anxiety, and early versus late illness onset (i.e., peak 1 versus peak 2). * p < .05, *** p < .001.

Quality of clinical care during COVID-19 illness as a protective factor
Path analyses examined whether quality of medical care during COVID-19 illness moderated conditional associations between discrimination, illness severity, and lasting symptoms, again controlling for non-White race, objective SES score, perceived SES score, history of mood/anxiety disorder, history of diabetes or heart disease, COVID-illness life disruption, COVID-illness anxiety, and early versus late illness onset. For individuals who reported that their quality of care was not excellent (n = 740; Figure S3), there was a significant direct association between illness severity and lasting symptoms (β = .38, p < .001), and a trending association between discrimination frequency and lasting symptoms (β = .069, p = .06). Although the direct unconditional effect of discrimination frequency on illness severity was not significant (B = -.02), the direct effect conditional on discrimination stress was significant (β = .08, p = .04). Analysis of simple slopes indicated that the effect of discrimination frequency on illness severity was more positive for individuals other significant direct or indirect associations (all βs < .07, all ps > .13). These findings suggest that high quality of care may buffer associations between discrimination, illness severity, and lasting symptoms. Figure S3. In individuals reporting non-excellent care (A), there were significant direct effects of discrimination frequency on lasting symptoms, and on illness severity, which was conditional on discrimination stress. There was also a significant conditional indirect association between discrimination frequency, illness severity, and lasting symptoms. In individuals reporting excellent care (B), there was only a direct effect of illness severity on lasting symptoms, suggesting that high quality of care may buffer associations between discrimination, illness severity, and prolonged symptoms. All coefficients are standardized. On all pathways, we controlled for discrimination stress, BIPOC race, cumulative SES risk score, perceived SES score, history of mood/anxiety disorder, history of diabetes/heart disease, COVID-illness life disruption, COVID-illness anxiety, and early versus late illness onset (i.e., peak 1 versus peak 2). ⏊ p < .10, * p < .05, *** p < .001. Figure S4. Patient report of COVID-19 symptoms and related conditions. The most prevalent concerns were chest pressure/shortness of breath, loss of taste or smell and fatigue. Pneumonia was the most frequent secondary complication of infection.

Frequency of symptoms experienced during COVID-19 illness
Patients were asked "What symptoms did you experience while you were ill with COVID? (check all)," "What was the most concerning COVID symptom or medical complication that you experienced? (check all)", and "What was the most concerning COVID symptom or medical complication that you experienced? (check one)". The most frequently endorsed concerning symptoms were difficulty breathing and loss of taste or smell.