Racial Discrimination and Multiple Health Outcomes: An Umbrella Review of Systematic Reviews and Meta-Analyses

Racism is a complex, multifaceted system of oppression and a fundamental cause of health inequalities across populations [1, 2]. Racism encompasses structural racism (e.g., housing policies and neighborhood racial segregation) [3], institutional racism (e.g., policies and practices in organizations) [4], cultural racism (e.g., exalting Western beauty standards) [5], and interpersonal racism, which refers to prejudice and mistreatment experienced on an individual level [6, 7]. Racial discrimination has been shown to impact health in prospective cohort studies [8, 9], resulting in health disparities among different racial groups [3]. In the USA, ethno-racial minorities are disproportionately exposed to stressors and often lack salutary resources, which, in theory, leads to a higher prevalence of multiple health problems [10].

Previous systematic reviews and meta-analyses have demonstrated that racial discrimination is associated with a wide array of adverse health outcomes, including mental health issues such as depression and anxiety, cardiovascular disease, and poorer self-related health [11]. Such reviews have also highlighted the cumulative impact of racial discrimination over the life course and its role in exacerbating existing health disparities [12, 13]. While such literature explored the associations between racial discrimination and related health outcomes, most have focused on specific health outcomes or subsets of individuals experiencing racial discrimination, such as maternal health, perinatal outcomes, cardiovascular outcomes, and the older adult black population in the USA [10, 14‒22]. Including a summary of these findings would help to conceptualize the current studies within the existing body of the literature. However, it is imperative to continuously synthesize emerging literature regarding the impacts of racism, which evolves and shifts over time. In this umbrella review, we focus on racial discrimination with a broad range of health outcomes across multiple nations, focusing on meta-analyses of observational studies. This involved systematically identifying relevant meta-analyses of observational studies, summarizing their outcomes, and evaluating the certainty of evidence; we aimed to offer a comprehensive assessment of the impacts of racial discrimination on health outcomes.

We conducted an umbrella review to summarize and evaluate the associations between racial discrimination and health outcomes. This review followed the PRISMA 2020 guidelines, and its protocol was registered with PROSPERO (Registration No. 42024513366) [23, 24]. Two authors, J.L. and J.K., systematically searched online databases (PubMed/MEDLINE, Embase, Web of Science, and Google Scholar) for meta-analyses of observational studies examining the association between racial discrimination and health outcomes until January 31, 2024. Our search strategy included keywords related to racism and meta-analysis as follows: (“racism” OR “discrimination” OR “racial” OR “colorism” OR “perceived skin tone discrimination”) AND (“meta”) and their variants. We also manually searched references of eligible articles and reviewed titles, abstracts, and full texts (J.L. and J.K.); for an included paper in our study, we contacted authors and retrieved additional data unavailable in the manuscript. Our study included only observational study-based meta-analyses as no randomized controlled trials’ meta-analyses met the criteria to be included in eligible studies for our manuscript. Details on these exclusion criteria are available in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000542988) as an additional resource. The racial discrimination-related factors eligible for our umbrella review included hepatocellular carcinoma (HCC), colon cancer, out-of-hospital cardiac arrest, peripheral artery disease, acute myocardial infarction, nonalcoholic fatty liver disease (NAFLD), and psychosis experiences (PEs). Meta-analyses reporting odds ratio (OR), hazard ratio (HR), and risk ratio (RR) of the association of racial discrimination and its health outcome were included in this review. We recalculated the pool effect size, and HR and RR were converted to equivalent OR (eOR) in reanalysis [23‒25].

The methodological quality of studies included in our review was evaluated using the A Measurement Tool Assessment Systemic Reviews 2 (AMSTAR2) checklist. In instances of disagreement regarding the quality assessment, another researcher (Y.S.) was consulted to facilitate reaching a consensus.

Two researchers (J.K. and J.L.) independently screened titles, abstracts, and selected articles for a comprehensive, full-text review. These two researchers have extracted the following key data from the selected articles: publication year, number of primary studies included, outcomes, country of study, number of cases and participants, study design, effect estimation model (random or fixed effects), heterogeneity, and maximally adjusted effect size with 95% confidence interval (CI). Each meta-analysis was re-analyzed by the Der Simonian and Laird random fixed-effects model [25‒27]. Dose-response meta-analyses were not re-analyzed if they lacked sufficient data. Additional analyses were performed to evaluate specific aspects. I² statistics was performed to evaluate heterogeneity, and an I² value exceeding 50% indicates significant heterogeneity. P-curve analysis was used to detect p-hacking [25‒27]. The 95% prediction interval (PI) was examined to assess the uncertainty of the observed estimates and predict the value of new future observations based on Bayesian statistics. To mitigate type 1 errors, the Knapp-Sidik-Jonkman random-effects model was utilized [25‒27]. Publication bias was assessed with Egger’s test, indicating bias at a p value <0.1. eOR for various metrics, including the RR, were calculated following the latest guidelines [28]. These analyses were conducted using the “meta” package of R software (version 4.2.2; R Foundation, Vienna, Austria), with significance determined at a two-sided p value of less than 0.05 [29].

In this review, we assessed the class and quality of evidence (CE) for each outcome, using criteria from previous umbrella reviews [25‒27]. Observational study associations were categorized into five levels based on the strength of evidence for potential factors (class I, convincing; class II, highly suggestive; class III, suggestive; class IV, weak; NS, not significant). The credibility of evidence from observational studies was rated considering various factors, including the number of events related to the outcome of interest, the p value of the association, the presence of small study effects, excess of significance bias, PIs, statistical significance in the largest study, and heterogeneity.

According to the criteria of observational study, the credibility of evidence was graded. Class I involved exceeding 1,000 cases (or over 20,000 participants); significant summary associations as per random-effects calculations, with p < 10⁻⁶; the absence of indications pointing toward small-study effects; no observed evidence suggesting an excess of significance bias; not including the null value in PIs; the largest study is nominally significant, p < 0.05; a low degree of heterogeneity, specifically, I² value below 50%, Class II involved exceeding 1,000 cases (or over 20,000 participants); significant summary associations as per random-effects calculations, with p < 10⁻⁶; the largest study is nominally significant, p < 0.05. Class III involved exceeding 1,000 cases (or over 20,000 participants); significant summary associations as per random-effects calculations, with p < 10⁻³. Class IV involved any other associations that present a p value lower than 0.05; no significant evidence was defined when p > 0.05.

Of 1,081 studies screened, 286 full-text studies were assessed after duplicate removal, resulting in 8 meta-analyses that examined 15 unique associations between racial discrimination and overall health outcomes; data from one study were directly obtained from the author (Fig. 1 and Table 1). These meta-analyses spanned from 2007 to 2023, covering 113 articles from the USA and the UK (Table 1). In addition, we present the characteristics of systematic reviews measuring racial discrimination among multiple health outcomes (Table 1) [30‒37].

Utilizing the AMSTAR2 evaluation method, our study determined the quality of the original meta-analysis; those articles were high quality in two meta-analyses, moderate quality in three meta-analyses, and low quality in two meta-analyses (Table 2 and online suppl. Table 2). In the analysis, three meta-analytical associations (20.00%) showed suggestive evidence, five meta-analytical associations (33.33%) met the criteria for weak evidence, and seven meta-analytical associations (46.67%) were found to have nonsignificant evidence (online suppl. Table 3). Except for the mortality of NAFLD of Hispanics, the shape of the p-curve was highly right-skewed for the binomial metrics (p < 0.25), indicating no evidence of p-hacking. In our re-analysis of the thirteen associations through random effects analyses, 40.00% (6/15) demonstrated significant heterogeneity with an I² value greater than 75 (I² > 75). Using Egger’s regression test, we observed statistical evidence of publication bias in none of the studies included. The forest plot, funnel plot, and p curve for each association are shown in online supplementary Figures 1–15.

We categorized the outcomes into four domains: oncology, cardiology, gastroenterology, and mental health, within identifying six significant associations with racial discrimination (Table 3 and online suppl. Table 4). The results are consolidated into evidence maps in our umbrella review, providing a structured overview of how racial discrimination impacts diverse health outcomes (Table 4). Eight meta-analysis articles have covered over 1 million individuals with 15 unique health outcomes related to racial discrimination. These outcomes included mortality of HCC over multiple races (black, Hispanic, and Asian) compared to white patients, early-stage diagnosis of HCC over multiple races (black, Hispanic, and Asian) who experience disparities on the cancer care continuum compared to white patients, mortality of out-of-hospital cardiac arrest who had treatment within 30 days, mortality of peripheral artery diseases, mortality of acute myocardial infarction, NAFLD over multiple races (black and Hispanic), psychotic experiences, suicidal ideation (SI), and suicidal attempt. Our reanalyses identified eight statistically significant associations, highlighting the impact of racial discrimination on health outcomes (Table 4).

Racial discrimination was significantly associated with three health outcomes, which are mortality of HCC among black patients compared to white (reference) patients (eOR, 1.10 [95% CI, 1.06 to 1.13], CE = weak), mortality of HCC among Asian patients compared to white patients (eOR, 0.83 [95% CI, 0.72 to 0.96], CE = suggestive), and early stage of HCC diagnosed among black patients with disparities on the cancer care continuum, compared to white patients (eOR, 0.77 [95% CI, 0.64 to 0.92], CE = suggestive).

Among three racial discriminations associated with cardiological health outcomes, all had nonsignificant CE. No health outcomes showed statistically significant associations with racial discrimination.

Among two racial discrimination-associated gastroenterological health outcomes, both had weak CE. Racial discrimination was significantly and negatively associated with both clinical outcomes, which are among Hispanic patients with NAFLD (eOR, 1.47 [95% CI, 1.35 to 1.61], CE = weak) and black patients (eOR, 0.74 [95% CI, 0.69 to 0.80], CE = weak) with socioeconomic status/differential access to health care, compared to white patients.

For mental health outcomes, racial discriminations were significantly associated with increased odds of psychotic experiences (2.40 [1.87 to 3.07], CE = weak), SI (1.80 [1.63 to 2.00], CE = suggestive), and suicidal attempt (1.61 [1.25 to 2.07], CE = weak). Numerous significant associations were from weak to suggestive evidence levels, indicating variability in the evidence.

To our knowledge, our study is the first umbrella review on the relationship between racial discrimination and health outcomes. While previous reviews have focused on specific domains, such as sole examination of mental health outcomes and biomarkers or analysis only among the USA, and are outdated, our review provides a comprehensive highlight across multiple health disparities impacting ethnic minority groups throughout the globe [12, 38‒40]. Eight meta-analyses covering over 1 million participants identified 15 potential associations between racial discrimination and health outcomes. Of 15 previously reported associations, racial discrimination was associated with eight health outcomes. We found five notable health outcomes that were significantly associated with racial discrimination, which are the mortality rate, NAFLD, PEs, SI, and suicide attempts. Importantly, we also found three health outcomes: HCC-related death among Asian patients compared to white patients, early-stage diagnosis of HCC among black patients compared to white patients, and prevalence of NAFLD among black patients with socioeconomic status/differential access to health care, showed significantly lower risk with suggestive or weak CEs, respectively.

Numerous studies have explored the link between racial discrimination and health outcomes, showing that racial discrimination is associated with poorer health. Numerous models have been proposed to help conceptualize the pathways between racial discrimination and health, which include race-based trauma, minority stress theory, and Hatzenbuehler’s integrative mediation framework [41], among other models. However, the pathways are complex and still emerging [42].

Our study shows a significantly higher risk of mortality from HCC among black patients compared to white patients, part of which is associated with racial discrimination. While multiple factors might affect the prevalence of HCC, it can be specifically attributed to the prevalence of hepatitis B and C. HCC, the third most common cause of cancer-related deaths worldwide, is significantly influenced by the presence of chronic hepatitis B and C [43, 44]. Research indicates that Africans, African Americans, and Sub-Saharan Africans have a high prevalence of hepatitis B virus [45]. Similarly, hepatitis C virus prevalence is notably higher among African Americans, highlighting the association between these viral infections and HCC risk within these populations. This trend can be related to socioeconomic disparities and structural racism, leading to limited health care access, missed opportunities for screening and vaccination, and higher engagement in high-risk behaviors. Furthermore, the increased engagement in high-risk behaviors, such as injection drug use, is associated with a risk of hepatitis B virus and hepatitis C virus infections [46]. Additionally, the chronic stress from perceived racial discrimination, which affects black individuals of all ages, has been linked to higher alcohol consumption, contributing to the observed associations between racial discrimination and health disparities in these populations [47]. The higher prevalence of HCC among black individuals, driven by factors like health disparities and socioeconomic conditions, correlates with increased mortality rates, linking it closely to racial discrimination. Interestingly, the trend inversely appears in early-stage HCC diagnoses among black populations, indicating a complex relationship between racial discrimination and health outcomes in HCC [48, 49]. Among various factors that might contribute to such outcomes, systematic racism can be taken as an example: black patients with hepatitis C virus being diagnosed with HCC at earlier liver disease stages yet facing delayed detection due to less rigorous screening not tailored to their risk profile. For instance, the common Fibrosis Index Based on 4 Factors threshold excludes nearly one-third of black patients from HCC screening, highlighting a health care system gap in adequately addressing early-stage HCC diagnosis disparities among black individuals [50].

NAFLD in the Hispanic population shows a higher risk when individuals experience racial discrimination. This link may be due to factors rooted in institutional and structural racism, including access to healthy foods, health care access, and the allostatic load borne by those in impoverished conditions [51, 52]. While both Hispanic and black populations are minorities, a lower risk exists between NAFLD and the racial discrimination-experienced black population is observed, contrasting with Hispanics. In terms of the risk of NAFLD among the black population compared to the white population, while it is common knowledge that NAFLD disease burden is related to metabolic syndrome components such as obesity, diabetes, and dyslipidemia, a notable finding is that some studies suggest black American populations (who have higher rates of diabetes and vs. white populations) may have significantly lower risk of NAFLD. This suggests an explanation through racial discrimination mixed of structural, cultural, and interpersonal racism; for instance, institutional racism can produce unresponsive services and generate mistrust of the black population [53‒56]. The expansion of managed care and for-profit health care, along with historical unethical medical research [57, 58] may have contributed to the observed negative association between NAFLD and racial discrimination, resulting in less detection of NAFLD as more racial discrimination is presented.

In addition, five ethnic groups (Bangladeshi, black Caribbean, Indian, Irish, and Pakistani) experiencing racial discrimination showed a higher vulnerability to PEs, SI, and suicide attempts, suggesting discrimination is associated with increased PE risk through limiting access to health care and resources, exposure to risk factors, and direct stress effects [38]. It is possible to claim that discrimination is associated with limited access to health care, life-promoting resources, and education, resulting in increased exposure to other risk factors during the early stage of developing PEs [57, 59‒61]. Also, social stressors and perceived ethnic discrimination might cause direct, significant emotional stress, leading individuals experiencing racial discrimination to have negative mental health outcomes; such outcomes could also be associated with functional and structural changes in the central nervous system [62].

Further research is needed to consider various confounding factors and establish robust causation between health outcomes and racial discrimination. Additionally, the ethical implications of our findings highlight the urgent need for health care policies to address and mitigate the impacts of racial discrimination. Given the observed associations found, governments should address these health disparities with targeted interventions to eliminate racism in all its forms and facets, as well as efforts to provide tailored health services in communities. Possible policies could include implementing mandatory cultural competency training among health care providers, establishing community health worker programs, and developing standardized protocols for identifying and addressing discrimination in health care settings. Such findings would help inform the development of comprehensive anti-racist interventions and health care policies that prioritize equity and inclusivity.

Our umbrella review, a first of its kind on racial discrimination factors, provides evidence of the association of health outcomes with racial discrimination. However, as an umbrella review, it inherently relies on existing meta-analyses, which may introduce limitations related to the scope and quality of those underlying studies. Yet, our study is limited by its keyword scope, focusing only on “racism” and “racial discrimination,” potentially overlooking variations like “perceived skin tone disparities/discrimination” [63, 64], “perceived ethnic discrimination” [32], and “perceived colorism” [65], limiting our search to specific terms related to racism might have narrowed the scope of our findings. However, our focus on structural, institutional, cultural, and interpersonal racism required a precise approach, excluding broader or related terms to maintain a consistent definition throughout our research. Additionally, the manuscript predominantly includes studies conducted in the USA and the UK, which limits the generalizability of the findings to a global context. As it is well known that structural and cultural dimensions vary among regions and areas, it might diminish the relevance of the findings for non-Western populations; such limitation calls for the need for more nuanced contextualization, which we address in the discussion section by exploring regional variation and their implication. Moreover, the inclusion of papers under “racial discrimination” that discuss both racism-related and genetic-related discrimination led to some ambiguity in the analysis of the meta-analyses, potentially blurring the distinctions between these types of discrimination. Not directly assessing the original quality of texts within included meta-analyses might affect the precision of identified issues – also, methodological variances in evaluating each sample size of the meta-analyses, heterogeneity, and statistical significance. If the mathematical method used in this umbrella review changes, efficacy certification could alter the efficacy conclusions. Although our umbrella review highlights the association between health outcomes and racial discrimination, causation remains to be further investigated, as discussed frequently above. The regional focus on North America and Europe also restricts a comprehensive global understanding of the impact of racial discrimination on health.

Our umbrella review identified patterns between racial discrimination and health outcomes, revealing weak and nonsignificant associations based on various methodological approaches. We also observed eight notable associations between racial discrimination and health outcomes. These findings underscore the complexity of these relationships and highlight the necessity for further research to understand such associations better while acknowledging the limitations of current observational evidence and the need for rigorous study designs to establish potential causal pathways.

This systematic review article does not require Institutional Review Board approval. Our systematic review and meta-analysis protocol was registered with PROSPERO (Registration No. CRD42024513366).

We declare no competing interests.

This research was supported by grants from the National Research Foundation of Korea (NRF) funded by the Korea government (MSIT; RS–2023–00248157) and the Ministry of Science and ICT (MSIT), Korea, under the Information Technology Research Center (ITRC) support program (IITP-2024-RS-2024–00438239) supervised by the Institute for Information & Communications Technology Planning & Evaluation (IITP). The funders had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Dr. Dong Keon Yon had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. All the authors approved the final version before submission. Study concept and design; acquisition, analysis, or interpretation of data; drafting of the manuscript; and statistical analysis: Jun Hyuk Lee, Hyeri Lee, Yejun Son, Jiseung Kang, and Dong Keon Yon; critical revision of the manuscript for important intellectual content: all the authors; study supervision: Dong Keon Yon and Jiseung Kang supervised the study and those are the guarantors for this study. Jun Hyuk Lee, Hyeri Lee, and Yejun Son contributed as co-first authors. Dong Keon Yon and Jiseung Kang contributed equally as the corresponding authors. The corresponding authors attest that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Jun Hyuk Lee, Hyeri Lee, and Yejun Son are joint first authors.

The data used in this review were derived from publicly available systematic reviews and meta-analyses.