Perceived discrimination as a modifier of health, disease, and medicine: empirical data from the COVID-19 pandemic

The study addresses the growing concern of long-term symptoms following recovery from COVID-19 (Long COVID/PASC), which include not only fatigue and anosmia but also cognitive and psychiatric complaints. Prior work (e.g., large EMR analyses and UK Biobank longitudinal imaging) indicates persistent physical and neurological effects, potential CNS involvement via multiple biological pathways, and neuroinflammation. The pandemic has disproportionately affected racial/ethnic minorities, with social determinants of health (including discrimination and socioeconomic stress) implicated as key drivers once biomedical and access factors are considered. Chronic perceived discrimination may increase disease risk through barriers to care and elevated allostatic load. The current study aims to quantify prevalence and types of lasting symptoms and test whether perceived discrimination (frequency and associated stress) and socioeconomic factors relate to illness severity and lingering symptoms, while examining moderating effects of perceived care quality and timing of infection (early vs late in the pandemic).

Evidence reviewed includes: persistent multi-system symptoms after COVID-19, including neurological and psychiatric sequelae; EMR studies showing higher rates of anosmia, dysgeusia, chest pain, chronic fatigue, dyspnea, and diabetes among those with prior COVID-19; UK Biobank findings of reduced global brain volume and neurocognitive decline post-infection; mechanistic hypotheses for CNS involvement via respiratory, hematogenous, and olfactory routes with associated inflammation, endothelial dysfunction, and coagulopathy. Disparities literature highlights disproportionate impacts on racial/ethnic minority groups, with social determinants and comorbidities explaining much variance. Perceived discrimination is linked to underutilization of care, mistrust, delayed treatment, and chronic stress/allostatic load, potentially exacerbating post-COVID outcomes. Gaps identified include limited rigorous evaluation of prior health, perceived equity, and stress in relation to Long COVID/PASC.

Design: Cross-sectional survey of adults with prior COVID-19 infection identified from NYU Langone Health records. Records queried in Feb 2021 for ICD-10 B97.29. Eligible individuals (alive, with email, and permitted for research contact) were invited; surveys completed between Feb 23, 2021 and Apr 4, 2021. Participants received a $25 gift card. IRB-approved with informed consent and data sharing provisions. Participants: 2,212 respondents; 1,584 retained post-inclusion/quality criteria. Ages 18–96. Illness timing spanned NYC’s two incidence peaks (April 8, 2020; January 7, 2021), enabling early vs late comparisons. Measures: Self-reported medical and psychosocial data, including illness severity (composite from multiple indicators such as fever severity, illness duration, complications, hospitalization, and perceived severity), lasting symptoms (counts and domains: neurological, cognitive/memory, mood), discrimination frequency and stress from discrimination, current stress/anxiety, COVID-illness life disruption, prior mental and physical health conditions. Socioeconomic status: Objective SES composite (income-to-needs, education, housing stability, public assistance) and perceived SES composite (financial satisfaction, financial worries, perceived stability, MacArthur ladder). Confirmatory factor analyses indicated excellent fit (objective SES: RMSEA=0.074, CFI=0.958, χ2=1773.75, p<0.001; perceived SES: RMSEA=0.012, CFI=0.990, χ2=623.03, p<0.001). Statistical analysis: Group comparisons for early vs late infection using chi-squared tests and two-sample t-tests with 5000 bootstrap samples. Path analyses tested direct effects of discrimination frequency on illness severity and lasting symptom count, indirect effects via illness severity, and moderation by stress from discrimination. Secondary analyses substituted current stress level as moderator (with discrimination stress as covariate) to assess specificity. Additional path models stratified by perceived quality of medical care (excellent vs less-than-excellent). Exploratory post hoc logistic regressions examined predictors of domain-specific lasting symptoms (neurological, cognitive, mood), including discrimination frequency, perceived SES, COVID-life disruption, and illness activity indicators. All models controlled for race (white vs non-white), objective SES risk, perceived SES, prior mood/anxiety disorder, diabetes/heart disease, COVID-illness life distress, COVID-illness anxiety, and early vs late illness onset. FIML used to handle missingness; mediation tested with 5000 bootstrap samples; analyses conducted in Mplus v8.

- Prevalence: 70.6% (1,118/1,584) reported at least one lasting symptom post-recovery; mean lingering symptoms 3.06 (SD=3.73). Mood complaints in 19.4% and cognitive/memory complaints in 25.1%. - Early vs late infection: Early cases showed higher composite illness severity and more lasting symptoms, greater life disruption, slightly higher discrimination frequency/stress, and lower satisfaction with care; non-White representation was higher in early cases (p=0.03). Several differences remained significant after multiple-comparisons correction (e.g., illness severity, lasting mood complaints, life disruption, care satisfaction). - Discrimination effects: Higher discrimination frequency predicted greater illness severity and higher lasting symptom counts, controlling for sociodemographic and health covariates. Stress from discrimination moderated these relationships: effects of discrimination frequency on illness severity and lasting symptoms were stronger at higher discrimination-related stress levels; an indirect pathway from discrimination frequency to lasting symptoms via illness severity was observed under higher discrimination stress. - Specificity to discrimination-related stress: Replacing the moderator with current general stress showed moderation for lasting symptoms only; current stress did not moderate the discrimination–illness severity path nor the indirect path via severity, indicating a unique role for discrimination-related stress. - Perceived care quality: Path model effects largely held in the low perceived care quality group but attenuated or became non-significant in the high-quality group, suggesting perceived excellent care may buffer adverse associations. - Domain-specific outcomes: Logistic regressions showed greater discrimination frequency associated with higher likelihood/number of lasting neurological symptoms (p=0.02). Perceived (but not objective) SES predicted lasting cognitive complaints. There was a trend toward association between discrimination frequency and lasting mood symptoms (p=0.07).

Findings demonstrate that lasting post-COVID symptoms are common and include substantial neurological and cognitive components. The study provides evidence that perceived discrimination—particularly when experienced as stressful—relates to greater acute illness severity and more numerous lasting symptoms, advancing understanding of social determinants in Long COVID/PASC. The moderation by discrimination-related stress and the attenuation of effects among those reporting excellent care suggest that both chronic social stressors and the healthcare experience shape outcomes. Early-pandemic infections were linked to worse severity and more persistent complaints, aligning with contextual differences in healthcare burden and treatment availability early on. The distinct predictive value of perceived SES over objective SES underscores the importance of subjective socioeconomic standing in health trajectories. Collectively, these results support integrating psychosocial assessments and improving perceived care quality to mitigate long-term sequelae.

This study shows that 70.6% of recovered COVID-19 patients report persistent symptoms, with notable impacts on neurological and cognitive domains. Perceived discrimination (frequency and stress) predicts greater illness severity and more lasting symptoms, effects that are partly buffered by perceived high-quality medical care and not fully explained by general stress. Timing of infection (early pandemic) is associated with worse outcomes. Perceived SES is a stronger predictor of certain long-term complaints than objective SES. These findings emphasize the need for public health and clinical strategies that address psychosocial determinants, reduce discriminatory stressors, and enhance patient-centered care. Future research should longitudinally track symptom trajectories, incorporate objective neurobiological measures alongside patient-reported outcomes, and examine structural drivers of perceived care quality and discrimination to inform targeted interventions.

The study relies on self-reported data for symptoms, discrimination experiences, and perceived care quality, which may introduce reporting and recall biases. The cross-sectional design precludes causal inference about the relationships among discrimination, illness severity, and lasting symptoms. The authors note they cannot directly address how perceived discrimination maps onto structural inequalities. The sample was drawn from a single health system and region during specific pandemic waves, which may limit generalizability. Some measures (e.g., quality of care) are subjective, and early vs late comparisons may reflect unmeasured contextual changes over time.