A meta‐analytic review of the relationship between racial discrimination and alcohol and other drug use outcomes in minoritised racial/ethnic groups
META‐ANALYSIS OF RD AND AOD
Gates et al.
Division of Psychology and Mental Health University of Manchester, Faculty of Biology, Medicine and Health, School of Health Sciences Manchester UK
School of Psychology Liverpool John Moores University Liverpool UK
* CorrespondenceEvie Gates, Division of Psychology and Mental Health, University of Manchester, Faculty of Biology, Medicine and Health, School of Health Sciences, Manchester, UK.
Email: evie.gates@postgrad.manchester.ac.uk
Abstract
Aims
To measure the associations between racial discrimination and distinct alcohol and other drug use outcomes in minoritised racial/ethnic groups and to explore the moderating roles of demographic and methodological characteristics.
Methods
Quantitative studies including racial discrimination as an exposure (both binary and continuous), an alcohol and/or other drug use outcome and a minoritised racial/ethnic sample were identified via database, citation and journal searching. 130 studies contributing 273 effect sizes, across seventeen distinct outcomes, were included in this analysis. Random‐effects meta‐analytic models were implemented. Moderation effects were explored using subgroup analyses.
Results
Racial discrimination was positively associated with sixteen alcohol and other drug use outcomes. The strongest associations were observed for at‐risk/hazardous alcohol use [r = 0.24, 95% confidence interval (CI) = 0.17–0.3, I2 = 94.8%, m = 29, n = 9445], at‐risk/hazardous cannabis use (r = 0.24, 95% CI = 0.18–0.29, I2 = 0%, m = 4, n = 462) and substance use disorder (r = 0.25, 95% CI = 0.14–0.36, I2 = 97.7%, m = 5, n = 21 051). Considerable heterogeneity was observed across fourteen outcomes (I2 = 69.5%–97.7%). Concerning tobacco use, Indigenous North Americans had the largest effect (r = 0.27, 95% CI = 0.2–0.35, I2 = 0%, m = 2, n = 529), followed by Black Americans (r = 0.06, 95% CI = 0.01–0.12, I2 = 81.7%, m = 7, n = 5409). Little evidence for an association was found for Latinxs (r = 0.06, 95% CI = –0.02 to 0.14, I2 = 89.2%, m = 3, n = 5404) or Asian Americans (r = –0.18, 95% CI = –0.8 to 0.43, I2 = 99%, m = 2, n = 572). Regarding composite substance use, Indigenous North Americans had the strongest associations (r = 0.29, 95% CI = 0.23–0.35, I2 = 0%, m = 3, n = 778), followed by Black Americans (r = 0.13, 95% CI = 0.09–0.18, I2 = 62.8%, m = 7, n = 5981) and then Latinxs (r = 0.07, 95% CI = –0.17 to 0.31, I2 = 91.3%, m = 4, n = 1646). Concerning alcohol use problems, younger samples produced stronger associations (r = 0.28, 95% CI = 0.17–0.38, I2 = 38.8%, m = 3, n = 483), while older samples showed larger effects in six other outcomes (rs = 0.13–0.26). Regarding at‐risk/hazardous alcohol use and alcohol use problems/consequences, cross‐sectional studies (rs = 0.23–0.24) produced stronger associations than longitudinal studies (rs = 0.13–0.14). Concerning tobacco and illicit substance use, the strongest associations were identified for lifetime exposure (rs = 0.18–0.32).
Conclusions
Racial discrimination appears to be a consistent correlate of multiple alcohol and other drug use outcomes in minoritised racial/ethnic groups, predominantly based in the United States, yet the magnitude of these associations differs across outcomes. Demographic and methodological characteristics somewhat moderate these associations.
Article notes
Gates E , Cant M , Elliott R , Irizar P , Armitage CJ . A meta‐analytic review of the relationship between racial discrimination and alcohol and other drug use outcomes in minoritised racial/ethnic groups. Addiction. 2025;120(12):2371–2403. 10.1111/add.70131 40667691 PMC12586790
Footnote Group
INTRODUCTION
Race and ethnicity are terms often used interchangeably, yet they are distinct but related constructs. Race is a social construct with no meaningful biological basis that has historically and contemporaneously been used to justify the dominion of the dominant racial group [1, 2, 3, 4]. Despite no universal definition, there is some consensus that race refers to grouping people based on shared ancestry and/or phenotype [5, 6, 7, 8, 9]. Ethnicity is also a multi‐dimensional social construct that pertains to shared cultural, ancestral, linguistic, religious and physical characteristics [10, 11, 12, 13, 14]. The determination of which racial and ethnic groups are minoritised differs widely across countries because of their varied historical, social and political contexts. The present study, however, defines minoritised racial and ethnic groups as those that are either numerically smaller than the rest of the population, hold a non‐dominant political, social or economic position in society or have an ethnicity, religion or language that differs from the majority [15].
Minoritised racial and ethnic groups have been reported to be at increased risk for alcohol and other drug use (AOD) at different stages of the life course [16, 17, 18, 19]. They may also be less likely to ‘age out’ of use [20] and more likely to experience negative consequences of use [21]. Exposure to racial discrimination, which can be defined as unfair treatment attributed to one's ethnicity, race or culture of origin [22, 23, 24], may help explain these increased risks [25, 26, 27]. It has been posited that minoritised racial/ethnic groups may engage in AOD to cope with the stress of discrimination [28, 29]. This is a particularly pertinent theory, because it contextualises racism within pre‐existing models of AOD, such as stress‐coping theory [30], tension‐reduction models [31], motivation models [32] and the self‐medication hypothesis [33]. A 2017 meta‐analysis of six effect sizes, however, failed to show a significant association between racial discrimination and substance use in United States (US)‐based minoritised racial/ethnic groups [34]. Yet later meta‐analyses have reported significant associations (r = 0.16) (r = 0.13) in minoritised racial/ethnic groups residing in the United States and internationally [35, 36], although these effect sizes are smaller than anticipated.
A closer examination of these meta‐analyses reveals important limitations. First, they report the association between racial discrimination and a composite measure of substance use, capturing different AOD outcomes within one generic ‘substance use’ outcome [34, 35, 36]. This method assumes that these outcomes are equivalent. In the US literature, some studies have reported little variation in the association between racial discrimination and distinct AOD outcomes. For example, comparable effect sizes have been reported across tobacco, cannabis, alcohol, prescription opioid and heavy alcohol use [37, 38]. Other research, however, has observed that racial discrimination is differentially associated with distinct types and patterns of AOD, including polysubstance use, dual substance use, binge drinking, alcohol use consequences, cigarette use, alcohol use, cannabis use and prescription substance use [39, 40, 41, 42, 43].
A second potential limitation of previous meta‐analyses is the lack of moderation analyses across distinct AOD outcomes, as there is some evidence from the United States to suggest that the racial discrimination‐AOD relationships vary by race/ethnicity, gender and age. For example, the association between racial discrimination and heavy alcohol use has been documented to be stronger in African Americans and Hispanics, compared to Chinese Americans [44]. Variability by age and gender has not been extensively studied, as they are typically used as covariates, obscuring their potentially moderating effects. However, Assari and colleagues [45] observed that exposure to racial discrimination in adolescence predicts increased cannabis use in adulthood in African American males, but decreased use in females. Gender differences have also been observed in Latinxs, where racial discrimination is more strongly associated with drug and alcohol abuse in males [46]. Moreover, experiences of racial discrimination in African Americans have been demonstrated to have a stronger association with regular smoking in respondents below age 45 [47]. Similar findings are reported for associations between racial discrimination and substance use disorder in African American, Hispanic and Asian participants [48].
Current study
The current study aims to quantitatively synthesise the literature on relationships between racial discrimination and AOD outcomes in minoritised racial/ethnic groups internationally. Considering the findings from previous meta‐analyses, this study intends to (1) determine the strength of associations between racial discrimination and distinct AOD outcomes in minoritised racial/ethnic groups; and (2) assess whether these associations are modified by race/ethnicity, age, gender, exposure timing and study design. To our knowledge, this is the first meta‐analysis to assess the role of racial discrimination across distinct AOD outcomes, within multiple minoritised racial/ethnic groups and across numerous countries. Therefore, it intends to provide critical insights into the role of racial discrimination in AOD, with the aim that this knowledge can guide intervention and prevention strategies, policy and educational practices.
METHODOLOGY
This review was registered on PROSPERO, the systematic review registry (ID: CRD42022381762). This study was also conducted and reported in line with the Preferred Reporting Items for Systematic Reviews and Meta‐analysis (PRISMA) [49].
Search strategy
Combinations of the following subject headings/index terms and free text terms were searched in PubMed, PsychInfo via Ovid, ProQuest for Dissertations and Theses and PsyArXiv: ‘Racism’, ‘racial discrimination’, ‘ethnic discrimination’, ‘racial trauma’, ‘racial abuse’, ‘substance use’, ‘drug use’, ‘addiction’, ‘drug abuse’, ‘alcohol use’, ‘substance abuse’, ‘substance use disorders’, ‘alcoholism’, ‘smoking’, ‘tobacco use’, ‘cannabis use’, ‘marijuana use’, ‘cannabis abuse’, ‘marijuana abuse’, ‘illicit substance use’, ‘opioid use’, ‘amphetamine use’, ‘cocaine use’ were searched. No filters were added to the search, except in ProQuest, to specify that only dissertations and theses should be returned. Searches were performed from January to July 2023 and updated in September 2024. In addition, the Journal of Psychoactive Drugs, Journal of Cultural Diversity and Ethnic Minority Psychology, Addictive Behaviours, Psychology of Addictive Behaviours, Journal of Immigrant and Minority Health, Journal of Ethnicity in Substance Abuse and Substance Use and Misuse were reviewed for additional studies that were not captured by the search strategy. These journals were selected as approximately a quarter of the studies identified via the search strategy were published in these journals. Likewise, citation searching was conducted on previous meta‐analyses and systematic reviews of racial discrimination and health outcomes. See Figure 1.
Eligibility criteria
Inclusion criteria included: use of quantitative methodologies, studies that examined an association between racial discrimination (and known synonyms, i.e. racial harassment, racial bullying) and any type or pattern of AOD (excluding treatment outcomes and measures of craving, relapse or intentions/willingness to use) and samples, which are comprised of people who are members of minoritised racial/ethnic groups.
Measures of racial discrimination included those that capture the frequency, chronicity or count of racially discriminatory events and those that capture racial discrimination appraisal.
Exclusion criteria included: studies that include racial discrimination in a composite measure of general or intersecting discriminations, and/or studies that constitute a previous meta‐analysis/systematic review on the association between racial discrimination and AOD.
In cases where studies have used two‐stage attribution style questionnaires to measure racial discrimination (i.e. the everyday discrimination scale, experiences of discrimination scale), these were only deemed eligible if they had been attributed to race, ethnicity or nationality.
Screening procedure
Screening was performed by two independent reviewers (E.G. and M.C.). First, titles and abstracts were screened against the eligibility criteria in Rayyan Screening Software for Systematic reviews [50]. Second, full‐text versions of studies identified as potentially eligible in the first screening phase were retrieved. Two independent reviewers screened each full‐text article against the eligibility criteria and recorded reasons for ineligibility when applicable. A third independent reviewer also determined eligibility in instances of disagreement (Figure 1).
Coding procedure and data extraction
Two trained and independent coders (E.G. and M.C.) extracted data from full‐text articles onto piloted coding forms. Where data was missing, authors were contacted to obtain this information. Discrepancies between coders were resolved via discussion and reappraisal of the full text articles. Data was extracted into five categories: methodological data (e.g. study design, method of recruitment), participant data (e.g. sample race/ethnicity, gender and age), exposure data (e.g. measurement of exposure, timing of exposure), outcome data (e.g. measure of outcome, operationalisation of outcome) and results data (e.g. statistical models used, effect sizes, 95% CI). The effect sizes extracted included correlation coefficients, OR and β coefficients.
Quality and certainty assessments
Assessment of study quality was performed by E.G. using the National Institute of Health study quality assessment tool for observational cohort and cross‐sectional studies [51]. Studies were assigned a quality rating of poor, fair or good based on assessment tool scores of <50%, 50% to 75% and >75%, respectively. Studies classified as poor were not excluded from analysis, because it has been suggested that there are dangers to blindly excluding poor‐rated studies from systematic reviews/meta‐analysis, as there are no clear‐cut distinctions between high‐ and low‐quality studies. Likewise, study quality assessment tools can only establish whether a study is susceptible to bias, and not whether it is biased [52]. Therefore, alternatively, a sensitivity analysis with the removal of poor rated studies was performed, in line with recommendations for meta‐analysis of observational studies [53].
Certainty assessment was performed by E.G. using the GRADE framework, where the strength of evidence was determined through appraisals of risk of bias, inconsistency, indirectness, imprecision and publication bias [54]. As the evidence included in this review was obtained from observational studies, the certainty of evidence rating for each outcome was initially low [55]. Certainty assessment can be found in Data S1.
DATA SYNTHESIS AND STATISTICAL ANALYSIS
Effect size metric
Correlation coefficients (r) were the effect size metric in this analysis. Where correlation coefficients were not reported, but other effect sizes were, these were converted to correlation coefficients using effect size converters, where possible [56, 57]. Only unadjusted effect sizes were converted because of inconsistencies in the type and number of covariates included in adjusted analyses across studies. Spearman ρ correlations were converted to approximate Pearson correlations using the equation provided by Rupinski and Dunlap [58]. For continuous AOD outcomes, Pearson's, biserial and tetrachoric correlations represented the metrics of interest, because they are statistically comparable metrics that capture underlying continuous constructs [59]. For true dichotomous outcomes, the point‐biserial correlation represented the metric of interest. In outcome domains that contained a combination of biserial, Pearson and tetrachoric coefficients, analysis was performed using raw coefficients, as Fisher's r‐z transformation is inappropriate when combining these types of correlations [59]. However, for outcome domains that only contain Pearson coefficients, the r‐z transformation was performed for analysis and back transformed for interpretation [60, 61].
Management of dependent effect sizes
Samples represent the unit of analysis in this study and are treated as independent. Therefore, only one effect size was included per sample/sub‐sample, per AOD outcome. To ensure this, two sets of prioritisation criteria were developed to determine, which dependent effect sizes were included in the analysis. These criteria addressed dependency within and between studies and can be found in Data S2.
Outcomes of interest
Effect sizes were categorised into discrete outcome domains if ≥4 effect sizes were available for each domain [62]. Therefore, some of the extracted effect sizes could not be included in this meta‐analysis as their corresponding AOD outcome occurred at a frequency of less than 4. Categorisation of outcomes resulted in 17 distinct AOD domains. See Table 1 for operationalisation and measurement of the outcome domains.
| AOD outcome | Operationalisation/definition | Example measurements |
|---|---|---|
| Tobacco use | Frequency of use Quantity of use Frequency × quantity of use | •CDC College Health Risk Behaviour Survey •Monitoring and Future National Survey •Bespoke scales a |
| Smoking status | Outcomes which categorised participants as ‘smokers’ or ‘non‐smokers’ via self‐identification or classification based on responses to smoking‐related questions | •US National Health Interview Survey •Sample Adult Core Questionnaire •Bespoke scales a |
| Presence–absence of tobacco use | Indication of the presence or absence of tobacco use | •National Household Survey on Drug Abuse •Monitoring the Future survey |
| Alcohol use | Frequency of use Quantity of use Frequency × quantity of use | •AUDIT consumption only items •CDC Youth Risk Behaviour Survey •Monitoring and Future National Survey •Daily Drinking Questionnaire •Adolescent Drinking Questionnaire •Drinking Styles Questionnaire •Youth Risk Behaviour Surveillance Scale •Timeline Follow Back •World Health Composite International Diagnostic Interview •Bespoke scales a |
| Alcohol use disorder | Indication of alcohol abuse or dependence as per DSM‐IV or alcohol use disorder as per DSM‐V | •Alcohol Use Disorder and Associated Disabilities Interview Schedule‐5 •World Mental Health Composite International Diagnostic Interview •Alcohol Use Disorder and Associated Disabilities Interview Schedule‐4 •Diagnostic Interview Schedule for Children |
| Binge drinking | Frequency of consuming 4/5 drinks on one occasion | •AUDIT (binge drinking item) •CDC and Prevention's behavioural risk factor surveillance system questionnaire •Timeline follow back •Youth Risk Behaviour Surveillance Scale •Bespoke scales a |
| At‐risk/hazardous alcohol use | Sum of scores or positive indication for AUDIT and CAGE or outcomes, which captured a combination of alcohol consumption, problems and dependency | •AUDIT •CAGE •Bespoke scales a |
| Alcohol problems/consequences | Frequency or number of problems/consequences related to alcohol use | •AUDIT (problem items only)/AUDIT‐P •Drinker Inventory of Consequences •Rutgers Alcohol Problem Index •Addiction severity index •Brief Michigan Alcoholism Screening test •Young adult alcohol consequences questionnaire •Bespoke scales a |
| Presence–absence of alcohol use | Indication of the presence or absence of alcohol use | •National Household Survey on Drug Abuse •Bespoke scales a |
| Cannabis use | Frequency of use Quantity of use Frequency × quantity of use | •Youth Risk Behaviour Surveillance Scale •Monitoring the Future •World Health Composite International Diagnostic Interview •Youth risk behaviour survey •Bespoke scales a |
| Presence–absence of cannabis use | Indication of the presence or absence of cannabis use | •National Household Survey on Drug Abuse •Bespoke scales a |
| Cannabis problems/consequences | Frequency or number of problems/consequences related to cannabis use | •Brief Marijuana Consequences Questionnaire •Bespoke scales a |
| At‐risk/hazardous cannabis use | Sum of scores or positive indication on CUDIT and its revised version | •CUDIT |
| Illicit substance use | Frequency of use Quantity of use Frequency × quantity of use Referring to substances that are prohibited by law and the use of prescription drugs without a prescription from a medical professional. Cannabis was modelled as a separate outcome to illicit substance use because of the variability in its legality status across time, countries and states | •Youth Risk Behaviour Surveillance Scale‐Illicit drug use subscale •Monitoring the Future •World Health Composite International Diagnostic Interview •Youth risk behaviour survey •The Diagnostic Interview Schedule for Children •Addiction severity index •Bespoke scales a |
| Substance use disorder | Indications of substance abuse or dependence as per DSM‐IV or substance use disorder as per DSM‐V, excluding alcohol and tobacco | •World Mental Health Composite International Diagnostic Interview •DSM‐V criteria •National Institute of Alcohol Abuse, Alcoholism, Alcohol use disorder and Associated Disability Interview Schedule •Alcohol Use Disorder and Associated Disabilities Interview Schedule‐5 |
| Substance use problems | Frequency or number of problems/consequences related to substance use, excluding alcohol and tobacco | •Minnesota Survey of Substance Use Problems •DAST •Addiction Severity Index •Bespoke scales a |
Statistical analysis
A series of 17 univariate meta‐analytic models, using inverse‐variance weighting, were implemented.
Random‐effects models were chosen as effect sizes were expected to vary across studies.
Heterogeneity was assessed via Cochran's Q test, where P < 0.05 indicates heterogeneity [63]. The I2 statistic was used to identify the percentage of variation in effect sizes because of heterogeneity, where a value of 75% indicates considerable heterogeneity [63, 64]. The statistical analysis was performed in R Studio using the meta, dmetar and metasens packages.
Calculating a summary effect size across all 17 outcomes was not possible because of violations of the assumption of independence. As some studies reported separate effect sizes for multiple outcomes for the same sample, combining them would result in participant duplication and dependent effects. Moreover, for studies with shared secondary samples, management of dependency was conducted at the outcome level to ensure only one effect size per sample was present in each univariate model; therefore, collating them within one model would result in dependent effects [65, 66].
Subgroup analysis
Where possible, subgroup analysis was performed to explore heterogeneity. The moderators of interest in this analysis were race/ethnicity, age group, gender, study design and timing of exposure. Analysis of subgroups was only possible for subgroups containing ≥2 effect sizes. Our approach for identifying subgroups was data‐driven, whereby the categorisation of subgroups was informed by the studies deemed eligible for inclusion and their associated sample and methodological characteristics.
Publication bias
Publication bias assessment for outcomes with >10 effect sizes was conducted via visual inspection of funnel plots and the Eggers test, where a significant (P < 0.05) result indicates plot asymmetry [67, 68]. However, for outcomes with ≤10 effect sizes, tests of funnel plot asymmetry are underpowered [69, 70]. As such, doiplots were created and inspected for asymmetry, and the Luis Furuya‐Kanamori (LFK) index was used to quantify asymmetry. In which a value that falls between −1 and 1 was deemed to be indicative of plot symmetry [71].
RESULTS
Study selection
A total of 1901 studies were identified via database searching and screened for eligibility at the title and abstract level, where 1494 were excluded for violating the eligibility criteria. A total of 398 were screened at the full‐text level, where 225 were deemed eligible. A further 32 records were identified via citation and journal searching and underwent full‐text screening, where 10 were considered eligible. The results were that 235 studies were eligible, 159 of which provided relevant effect sizes or data to calculate an effect size, 11 studies were removed from the analytical pool during the dependency management process across studies, and a further 18 studies were removed as their AOD outcomes occurred at a frequency of <4. Therefore, 130 studies, with 273 effect sizes, contributed to this analysis (Figure 1).
Identification of subgroups
Subgroups were categorised into the following: age groups—youth and/or adolescents, young adults and/or adults and mixed (i.e. multiple age groups); gender—male, female and mixed (i.e. both male and female participants); study design—longitudinal and cross‐sectional; exposure timing—lifetime, past year and less than past year; and minoritised racial/ethnic groups—Black American (African and/or Afro‐Caribbean heritage and residing in the United States), Latinx (Central and Southern American heritage and residing in the United States), Asian American (Asian heritage and residing in the United States), Indigenous North American (Indigenous peoples of North America), Black Canadians (African and/or Afro‐Caribbean heritage and residing in Canada), Aboriginal Australian and Torres Strait Islander (Indigenous Australian or Torres Strait Islands heritage and residing in Australia), South Asian (South Asian heritage and residing in Hong Kong), multi‐racial/ethnic (heritage from multiple racial/ethnic groups, irrespective of country), and diverse (multiple different racial/ethnic groups not analysed separately, irrespective of country).
The diverse race/ethnicity group, the mixed gender group and the mixed age group were not considered meaningful categories for comparison and were not included in the subgroup analyses.
Study characteristics
Characteristics of each study are displayed in Table 2. Studies were conducted/published between 1997 and 2024. Sample sizes ranged from 55 to 17 115 and 75% of studies used cross‐sectional designs, and the length of follow‐up for longitudinal studies ranged from up to 3 weeks to 13 years. A total of 109 of the studies were published in academic journals, 20 were dissertations/theses and one was a pre‐print. The majority of studies were conducted in the United States (94%), and the remainder were conducted in Canada, Hong Kong and Australia. In the US‐based studies, 56 had Black American‐only samples, 24 had Latinx‐only samples, nine had Asian American‐only samples, six had Indigenous North American‐only samples and one had a multi‐racial/ethnic only sample. In the Canada‐based studies, one had a Black Canadian‐only sample and two had Indigenous North American‐only samples. The Australia‐based study had an Aboriginal and Torres Strait Islander only sample, and the Hong Kong‐based study had a South Asian only sample. The remaining studies had a diverse sample of minoritised racial/ethnic groups that were not analysed separately or had multiple different minoritised racial/ethnic groups that were stratified for analytical purposes. A total of 70% of studies used a secondary data source, and 73% of studies had predominantly female samples and mean ages ranged from 9.5 to 49. With regards to methodological quality, 50 studies were rated as poor, 77 were rated as fair and three were rated as good. The most common methodological quality concerns among ‘poor’ rated studies were: a lack of clarity regarding the study eligibility criteria, inability to determine if the exposure preceded the outcome and insufficient information or no information regarding the reliability and validity of the outcome measurement.
| Study author and date | n | Design | Type | Country | Outcomes | Racial discrimination measurement | Secondary source | % female | Mean age | Race/ethnicity | Effect size | Study quality |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Khazvand et al. 2022 [72] | 501 | Cross‐sectional | Journal | United States | Composite substance use | Index of Race–Related Stress‐Brief | Unknown | 59.50 | 23.75 | Diverse | 0.17 | Poor |
| Layland 2020 [73] | 217 | Longitudinal (2 y) | Dissertation | United States | Alcohol use problems; alcohol use; at‐risk alcohol use; binge drinking; cannabis use; tobacco use; substance use disorder | Bespoke scale a | Healthy Young Mens Cohort | NA | 22.3 | Diverse | 0.21; 0.13; 0.16; 0.13; 0.16; 0.23; 0.22 | Fair |
| Lee et al. 2018 [74] | 465 | Longitudinal (4 y) | Journal | United States | Alcohol use problems; alcohol use | Daily Life Experiences scale | Flint Adolescent study | Unknown | Unknown | Black American | 0.16; 0.25 | Fair |
| Gerrard et al. 2017 [75] | 508 | Longitudinal (9 y) | Journal | United States | Alcohol use problems; alcohol use | Schedule of Racist Events (modified) | FACHS | 94 | Unknown | Black American | 0.07; −0.01 | Fair |
| Tran 2016 [76] | 131 | Cross‐sectional | Dissertation | United States | At‐risk alcohol use | Asian American Racism‐Related Stress Inventory | NA | 50.40 | 32.49 | Asian American | 0.09 | Fair |
| Drazdowski et al. 2016 [77] | 200 | Cross‐sectional | Journal | United States | Cannabis use; illicit substance use | Racism and Life Experiences scale (daily life experiences subscale) | Unknown | 53 | Unknown | Diverse | 0.00; 0.17 | Fair |
| Lorenzo‐Blanco et al. 2015 [78] | 1919 | Longitudinal (2 y] | Journal | United States | Tobacco use | Unknown | Project RED | 52 | 14.1 | Latinx | 0.07 | Fair |
| Sanders‐Phillips et al. 2014 [79] | 567 | Cross‐sectional | Journal | United States | Alcohol use; cannabis use | Bespoke scale a | Unknown | 61 | 15.6 | Black American | 0.06; 0.00 | Poor |
| Ornelas et al. 2011 [80] | 275 | Cross‐sectional | Journal | United States | Binge drinking | Bespoke scale a | Men as Navigators for Health and Hombres Manteniendo Bienestar y Relaciones Saludables | 0 | 28.4 | Latinx | 0.09 | Fair |
| Copeland‐Linder et al. 2010 [81] | 232‐268 | Longitudinal (2 y) | Journal | United States | Alcohol use; cannabis use; tobacco use | Racism and Life Experiences Scale | Unknown | Unknown | 46.40 | Black American | 0.01; 0.06; −0.07; 0.07; 0.01; 0.02 | Fair |
| Flores et al. 2010 [82] | 110 | Longitudinal (6 months) | Journal | United States | Alcohol use; illicit substance use | Discrimination Stress Scale | NA | 46 | 18.8 | Latinx | 0.35; 0.22; 0.16 | Fair |
| Stone et al. 2017 [83] | 4249 | Cross‐sectional | Journal | United States | Alcohol use; cannabis use; tobacco use | Bespoke scale a | Health Behaviours in School Aged Children | 51.7; 54.1; 49.7; 50.6; 50.8 | Unknown | Black American; Latinx; Asian American; other ethnic minority, multi‐racial/ethnic | −0.13; 0.05; 0.02; 0.05; 0.14; −0.097; 0.03; 0.17; −0.06; −0.03; −0.097; ‐0.03; −0.5; −0.03; 0.05 | Poor |
| Berkel et al. 2022 [84] | 571 | Longitudinal (1 y) | Journal | United States | Substance use problems | Racism and Life Experiences Scale | NA | 54 | Unknown | Black American | 0.29 | Fair |
| Mata‐Greve et al. 2018 [85] | 233 | Cross‐sectional | Journal | United States | At‐risk alcohol use | The Brief Perceived Ethnic Discrimination Questionnaire | NA | 73 | 36.32 | Latinx | 0.23 | Poor |
| Liu et al. 2022 [86] | 289 | Cross‐sectional | Journal | United States | At‐risk alcohol use | AAPI Hate Reporting Centre's Incident Report Questionnaire | NA | 43 | 33.1 | Asian American | −0.09 | Fair |
| Song et al. 2022 [87] | 602 | Longitudinal (5 y) | Journal | United States | Alcohol use | Bespoke scale a | Unknown | 54 | 12.92 | Latinx | 0.03 | Fair |
| Su et al. 2022 [88] | 383 | Cross‐sectional | Journal | United States | Alcohol use problems | Schedule of Racist Events | Cultural Experiences and Alcohol Use study | 81 | 20.65 | Black American | 0.13 | Fair |
| Iwamoto et al. 2022 [89] | 1432 | Cross‐sectional | Journal | United States | Alcohol use problems | The Everyday Racial Discrimination Scale | Unknown | 73.20 | 19.81 | Asian American | 0.22 | Fair |
| Heads et al. 2020 [90] | 266 | Cross‐sectional | Journal | United States | At‐risk alcohol use; illicit substance use | The Scale of Ethnic Experience ‐ Perceived Discrimination subscale | Multi‐site university study of identity and culture | 71.80 | 19.8 | Black American | 0.14; 0.1 | Fair |
| Zimmerman et al. 2022 [91] | 1333 | Cross‐sectional | Journal | United States | Presence–absence of alcohol use; Presence–absence of cannabis use, Presence–absence of tobacco use | Unknown | Project on Human Development in Chicago Neighbourhoods | 52.44 | 17.6 | Diverse | 0.23; 0.2; 0.17 | Poor |
| Steele et al. 2022 [92] | 291 | Longitudinal (unknown length) | Journal | United States | Binge drinking | Schedule of Racist Events | FACHS | 100 | Unknown | Black American | 0.07 | Fair |
| Schick et al. 2021 [93] | 106 | Cross‐sectional | Journal | Canada | Alcohol use; alcohol use problems | Bespoke scale a | Unknown | 50 | 14.6 | Indigenous North American | 0.3; 0.29 | Fair |
| Crichlow et al. 2022 [94] | 1514 | Cross‐sectional | Journal | United States | Composite substance use | Bespoke scale a | Unknown | 56.90 | 13.56 | Black American | 0.12 | Poor |
| Su et al. 2021 [95] | 165 | Cross‐sectional | Journal | United States | Alcohol use | Daily Life Experiences scale | Unknown | 75 | 21.56 | Black American | 0.09 | Poor |
| Keum et al. 2021 [96] | 387 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Perceived Online Racism Scale | NA | 57 | Unknown | Diverse | 0.47; 0.37 | Poor |
| Zapolski et al. 2021 [97] | 399 | Cross‐sectional | Journal | United States | At‐risk alcohol use; at‐risk cannabis use | The Racial and Ethnic Microaggression scale | NA | 61.40 | 20.7 | Black American | 0.23; 0.24 | Fair |
| Bakhtiari et al. 2020 [98] | 121 | Cross‐sectional | Journal | United States | Cannabis use | Adolescent Discrimination Distress index | Schools, Peers, and Adolescent Development Project | 54 | 15.56 | Latinx | 0.02 | Poor |
| Brown et al. 2021 [99] | 399 | Cross‐sectional | Journal | United States | Illicit substance use | Everyday Discrimination scale | NA | 26.80 | 34 | Diverse | 0.4 | Poor |
| Nalven et al. 2021 [100] | 598 | Cross‐sectional | Journal | United States | Binge drinking | Unknown | National Epidemiologic Survey on Alcohol and Related Conditions‐III | 54.40 | 40.28 | Diverse | 0.095 | Poor |
| Clifton et al. 2021 [101] | 147 | Cross‐sectional | Journal | United States | Composite substance use | Daily life Experiences Scale | NA | 81.70 | 23.16 | Black American | 0.22 | Poor |
| Marks et al. 2021 [102] | 196 | Cross‐sectional | Journal | United States | Alcohol use; alcohol use problems | Inventory of Microaggressions Against Black Individuals | NA | Unknown | Unknown | Black American | 0.2; 0.01 | Fair |
| Motley et al. 2021 [103] | 300 | Cross‐sectional | Dissertation | United States | Alcohol use problems; alcohol use; illicit substance use; substance use problems | Classes of Racism Frequency of Racial Experiences scale | NA | 50.70 | 19 | Black American | 0.014; 0.256; 0.15; 0.131 | Fair |
| Waldron et al. 2021 [104] | 245 | Longitudinal (1 y) | Journal | United States | Alcohol use; alcohol use problems | Schedule of Racist Events (modified) | NA | 73.10 | Unknown | Latinx | 0.23; 0.09 | Fair |
| Kogan et al. 2020 [105] | 505 | Longitudinal (3 y) | Journal | United States | Binge drinking; cannabis use | Schedule of Racist Events | African American Mens Project | 0 | 20.26 | Black American | 0.1; 0.14 | Fair |
| Glass et al. 2020 [106] | 17 115 | Cross‐sectional | Journal | United States | Alcohol use disorder | Experiences of Discrimination scale (modified) | National Epidemiologic Survey On Alcohol and Related Conditions III | 52.60 | Unknown | Diverse | 0.11 | Fair |
| Lui 2020 [107] | 988 | Cross‐sectional | Journal | United States | Alcohol use problems; alcohol use | Everyday Discrimination scale | NA | 42; 60.4; 64.7 | 22.42; 22.69; 23.02 | Black American; Asian American; Latinx | −0.16; 0.23; 0.08; 0.05; 0.08; −0.06 | Poor |
| Meca et al. 2020 [108] | 1101 | Longitudinal (3 y) | Journal | United States | Alcohol use | Unknown | Project RED | 51.20 | 13.99 | Latinx | 0.14 | Fair |
| Jelsma et al. 2019 [109] | 610 | Longitudinal (3 y) | Journal | United States | Alcohol use; cannabis use | Bespoke scale a | Maryland Adolescent Development in Context Study | 49 | Unknown | Black American | 0.17; 0.16 | Fair |
| Su et al. 2020 [110] | 383 | Cross‐sectional | Journal | United States | Alcohol use | Schedule of Racist Events | Cultural Experiences and Alcohol Use study | 81 | 20.65 | Black American | 0.07 | Fair |
| Hicks et al. 2018 [111] | 505 | Longitudinal (1.5 y) | Journal | United States | Tobacco use | Schedule for Racist Events (modified) | African American Men's Health Project | 0 | 20.26 | Black American | 0.19 | Fair |
| Nieri et al. 2022 [112] | 259 | Longitudinal (unknown length) | Journal | United States | Alcohol use; binge drinking; cannabis use | Unknown | NA | 61 | 15 | Diverse | −0.088; −0.033; 0.004 | Poor |
| Piña‐Watson et al. 2019 [113] | 796 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Perceived racism scale for Latinxs ‐ Frequency of exposure to racism subscale | NA | 66.30 | 19.45 | Latinx | 0 | Fair |
| Desalu et al. 2017 [42] | 251 | Cross‐sectional | Journal | United States | Alcohol use problems; binge drinking | The Perceived Ethnic Discrimination Questionnaire | NA | 66 | 20 | Black American | 0.3; −0.01 | Fair |
| Le et al. 2019 [114] | 311 | Longitudinal (1 y) | Journal | United States | Alcohol use; alcohol use problems | Everyday Discrimination scale | Unknown | 55 | 18.1 | Asian American | 0.15; 0.01 | Fair |
| Dickerson et al. 2019 [43] | 182 | Cross‐sectional | Journal | United States | Alcohol use problems; alcohol use; binge drinking; cannabis use; tobacco use; cannabis use problems | Microaggressions Distress Scale | NA | 50 | 15.6 | Indigenous North American | 0.35; 0.13 | Fair |
| Franco et al. 2019 [115] | 466 | Cross‐sectional | Journal | United States | Composite substance use | Multiracial Challenge and Resilience scale (General discrimination subscale) | NA | 66.20 | 29.7 | Multi‐racial/ethnic | 0.1 | Poor |
| Zapolski et al. 2019 [116] | 612 | Cross‐sectional | Journal | United States | Composite substance use | Bespoke scale a | Unknown | 58.40 | Unknown | Black American | 0.15 | Poor |
| Zapolski et al. 2018 [117] | 388 | Cross‐sectional | Journal | United States | Alcohol use | Schedule of Racist Events | NA | 62.40 | 20.6 | Black American | 0.56 | Fair |
| Gibbons et al. 2018 [118] | 889 | Longitudinal (13 y) | Journal | United States | Smoking status; tobacco use | Schedule of Racist Events | FACHS | 54 | 15.5 | Black American | 0.1; 0.12 | Fair |
| Lee et al. 2018 [119] | 681 | Longitudinal (3 y) | Journal | United States | At‐risk alcohol use | Daily Life Experiences scale | Unknown | 51 | Unknown | Black American | 0.1 | Fair |
| Metzger et al. 2018 [120] | 235 | Cross‐sectional | Journal | United States | Alcohol use; binge drinking | Daily Life Experiences scale | Activties and Behaviours in College study | 73.60 | 20.56 | Black American | 0.16; 0.21 | Poor |
| Pittman et al. 2018 [121] | 469 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Index of Race‐Related Stress (brief version) | Unknown | 100 | 20.24 | Black American | 0.18 | Poor |
| Ornelas et al. 2015 [122] | 5313 | Cross‐sectional | Journal | United States | Binge drinking | Brief Perceived Ethnic Discrimination Questionnaire (community version) | Hispanic Community Health study/Study of Latinos ‐ Sociocultural ancillary study | 55 | 42 | Latinx | 0.1 | Fair |
| Demianczyk 2015 [123] | 418 | Cross‐sectional | Dissertation | United States | Alcohol use problems | The Racial and Ethnic Microaggression scale | NA | Unknown | Unknown | Asian American; Black American; Latinx; multi‐racial/ethnic | 0.44; 0.35; 0.38; 0.3 | Poor |
| Zapolski et al. 2016 [124] | 1521 | Cross‐sectional | Journal | United States | Composite substance use | Bespoke scale a | Unknown | 56.30 | Unknown | Black American | 0.07 | Poor |
| Thompson et al. 2016 [125] | 144 | Cross‐sectional | Journal | United States | Alcohol use; at‐risk alcohol use, smoking status | Experiences of Discrimination scale | The Black LIFE | 52 | 44.6 | Black American | 0.0325; 0.229; 0.58 | Poor |
| Greenfield 2015 [126] | 347 | Cross‐sectional | Dissertation | United States | Alcohol use; binge drinking; illicit substance use | The Microaggressions scale | NA | 61.50 | 28.45 | Indigenous North American | 0.022; 0.014; 0.134 | Fair |
| Armenta et al. 2015 [127] | 674 | Longitudinal (7 y) | Journal | United States; Canada | Alcohol use disorder; presence–absence of alcohol use | Schedule of Racist Events | Unknown | Unknown | Unknown | Indigenous North American | 0.15; 0.07 | Good |
| Tse et al. 2014 [128] | 202 | Cross‐sectional | Journal | Hong Kong | At‐risk alcohol use | Unknown | NA | 0 | Unknown | South Asian | 0.03 | Poor |
| Rodriguez‐Seijas et al. 2015 [129] | 5191 | Cross‐sectional | Journal | United States | Alcohol use disorder; substance use disorder | Bespoke scale a | National Survey of American Life | Unknown | Unknown | Black American | 0.23; 0.22 | Poor |
| Brondolo et al. 2015 [130] | 518 | Longitudinal (up to 3 weeks) | Journal | United States | Smoking status | Perceived Ethnic Discrimination Questionnaire (community version) | Unknown | Unknown | Unknown | Black American | 0.19; 0.03 | Fair |
| Cano et al. 2015 [131] | 129 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Environmental scale from the Social attitudes, Familial, and Environmental Acculturation scale | NA | 70 | 19.41 | Latinx | 0.23 | Poor |
| Cheng et al. 2015 [132] | 203 | Longitudinal (1 y) | Journal | United States | At‐risk alcohol use | General Ethnic Discrimination scale | NA | 59.10 | 24.06 | Latinx | 0.18 | Fair |
| Spence et al. 2014 [133] | 340 | Cross‐sectional | Journal | Canada | Presence–absence of cannabis use | Measure of Indigenous Racism Experience Interpersonal racism scale | Researching Health in Ontario Communities | 54.90 | 41 | Indigenous North American | 0 | Fair |
| Kapadia 2013 [134] | 13 914 | Cross‐sectional | Dissertation | United States | Alcohol use disorder | Experiences of Discrimination scale | National Epidemiologic survey on Alcohol and related conditions II | 61 | Unknown | Diverse | 0.16; 0.14 | Fair |
| Hurd et al. 2014 [26] | 681 | Longitudinal (3 y) | Journal | United States | Alcohol use; tobacco use | Unknown | Unknown | Unknown | Unknown | Black American | 0.21; 0.02 | Fair |
| Otiniano Verissimo et al. 2014 [135] | 2312 | Cross‐sectional | Journal | United States | Alcohol use; alcohol use disorder; illicit substance use; substance use disorder | Everyday Discrimination scale | NLAAS | 55 | Unknown | Latinx | 0.11; 0.13; 0.4; 0.23; 0.24; 0.22; 0.5; 0.25 | Fair |
| Latzman et al. 2013 [136] | 336 | Cross‐sectional | Journal | United States | Alcohol use problems | Unknown | NA | 70.50 | 20.4 | Diverse | 0.56 | Poor |
| Borrell et al. 2012 [137] | 1169 | Longitudinal (13 y) | Journal | United States | Presence–absence of alcohol use | Bespoke scale a | CARDIA | Unknown | Unknown | Black American | 0.21 | Fair |
| Thoma et al. 2013 [138] | 276 | Cross‐sectional | Journal | United States | Alcohol use; binge drinking; cannabis use; tobacco use | Schedule of Racist Events | Diverse Adolescents Sexual Health study | 33 | 17.45 | Black American | −0.02; 0.15; 0.22; 0.1 | Poor |
| Horton et al. 2011 [139] | 573 | Cross‐sectional | Journal | United States | Tobacco use | General Ethnic Discrimination scale | Unknown | 47 | Unknown | Black American; Latinx | 0.17; 0.03 | Poor |
| Gibbons et al. 2013 [140] | 889 | Longitudinal (8 y) | Journal | United States | Composite substance use | Schedule of Racist Events | FACHS | 54 | 18.5 | Black American | 0.1 | Good |
| Park 2010 [141] | 1628 | Cross‐sectional | Dissertation | United States | Alcohol use | Bespoke scale a | NLAAS | 52.50 | 41.6; 41.9; 43 | Asian American | 0.05; −0.03; 0.23 | Poor |
| Respress 2010 [142] | 514 | Cross‐sectional | Dissertation | United States | Alcohol use; binge drinking; cannabis use; presence–absence of cannabis use | Bespoke scale a | The National Longitudinal Study of Adolescent Health | 52.50 | Unknown | Black American | 0.042; 0.082; 0.196; 0.16 | Poor |
| Krieger et al. 2011 [143] | 504 | Cross‐sectional | Journal | United States | Smoking status | Everyday Discrimination scale | My body, My story | 69.20 | 48.6 | Black American | 0.01 | Fair |
| Kam et al. 2011 [144] | 728 | Longitudinal (2 y) | Journal | United States | Composite substance use | Bespoke scale a | Unknown | 54 | 12.3 | Latinx | 0.11 | Fair |
| Galliher et al. 2010 [145] | 133 | Cross‐sectional | Journal | United States | Composite substance use; substance use problems | Unknown | NA | Unknown | Unknown | Indigenous North American | 0.371; −0.229 | Fair |
| Borrell et al. 2010 [44] | 2266 | Cross‐sectional | Journal | United States | Smoking status | Detroit Area Study Discrimination Questionnaire | Multi‐ethnic study of Atherosclerosis | 51.4; 51.9 | Unknown | Asian American; Latinx | 0.1; 0.14 | Poor |
| Kwate et al. 2009 [146] | 139 | Cross‐sectional | Journal | United States | Binge drinking; at‐risk alcohol use | Daily Life Experiences scale (race and bother scales) | Unknown | 100 | 34 | Black American | 0.071; −0.097 | Fair |
| Yoo et al. 2009 [147] | 271 | Cross‐sectional | Journal | United States | Presence–absence of alcohol use | Unknown | Asian Pacific Arizona Initiative survey | 52.20 | 43.6 | Asian American | 0.05 | Poor |
| Semino 2008 [148] | 110 | Cross‐sectional | Dissertation | United States | At‐risk alcohol use | Schedule of Racist Events | NA | 100 | 32.6 | Black American | 0.13 | Fair |
| Kulis et al. 2009 [149] | 1374 | Cross‐sectional | Dissertation | United States | Alcohol use; cannabis use; illicit substance use; tobacco use | Bespoke scale a | Unknown | 51.20 | 10.36 | Latinx | 0.14; 0.165; 0.119; 0.154 | Poor |
| Gibbons et al. 2007 [150] | 606 | Longitudinal (5 y) | Journal | United States | Cannabis use; illicit substance use | Schedule of Racist Events | FACHS | Unknown | 10.5 | Black American | 0.15; 0.08 | Fair |
| Kwate et al. 2003 [151] | 33–34 | Longitudinal (unknown length) | Journal | United States | Alcohol use; tobacco use | Schedule of Racist Events | Unknown | 100 | 44.4 | Black American | 0.23; 0.13 | Poor |
| Guthrie et al. 2002 [152] | 178 | Cross‐sectional | Journal | United States | Presence–absence of tobacco use | Everyday Discrimination scale | Female Adolescent Substance Experience study | 100 | 15.45 | Black American | 0.35 | Poor |
| Lui et al. 2020 [153] | 740 | Cross‐sectional | Pre‐print | United States | At‐risk alcohol use | Everyday Discrimination scale | NA | 53 | Unknown | Asian American; Black American | 0.1; 0.24; 0.18; 0.12 | Poor |
| Keum et al. 2023 [154] | 407 | Cross‐sectional | Journal | United States | At‐risk alcohol use | The Perceived Online Racism scale | NA | 57 | 34.12 | Diverse | 0.45 | Fair |
| Cave et al. 2019 [155] | 424–443 | Longitudinal (5 y) | Journal | Australia | Presence–absence of alcohol use; Presence–absence of tobacco use | Bespoke scale a | Footprints in time: The Longitudinal study of Indigenous Children | 49.60 | Unknown | Aboriginal Australian and Torres Strait Islanders | 0.088; 0.38 | Fair |
| Yen et al. 1999 [156] | 716 | Cross‐sectional | Journal | United States | Alcohol use problems; at‐risk alcohol use | Bespoke scale a | 1993–1995 San Francisco Muni Health and Safety study | Unknown | Unknown | Diverse | 0.07; 0.19 | Poor |
| Greenfield et al. 2021 [157] | 347 | Cross‐sectional | Journal | United States | Tobacco use | Experiences of Discrimination scale | NA | 65.60 | 28.45 | Indigenous North American | 0.26 | Poor |
| Assari et al. 2019 [158] | 595 | Longitudinal (13 y) | Journal | United States | Cannabis use | Daily Life Experiences scale | Flint Adolescent Study | 53 | 14.8; 14.9 | Black American | −0.04; 0.18 | Fair |
| Tao et al. 2022 [159] | 450 | Cross‐sectional | Journal | United States | Composite substance use | Ethnic Racial Discrimination index | Unknown | 52.40 | 20.84 | Diverse | 0.26 | Fair |
| Walsh et al. 2021 [160] | 139 | Cross‐sectional | Journal | United States | At‐risk alcohol use; substance use problems | Experiences of Discrimination scale | Unknown | 100 | 26.78 | Diverse | 0.03; 0.1 | Fair |
| Currie et al. 2021 [161] | 210 | Cross‐sectional | Journal | Canada | Composite substance use | Everyday Discrimination scale | Unknown | 47.10 | Unknown | Diverse | 0.34 | Poor |
| Chae et al. 2008 [162] | 2073 | Cross‐sectional | Journal | United States | Alcohol use disorder | Unknown | NLAAS | 52.40 | 41 | Asian American | 0.08 | Fair |
| Keum et al. 2022 [163] | 322 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Perceived Online Racism Scale | NA | 57 | 23.3 | Diverse | 0.44 | Fair |
| Otiniano Verissimo et al. 2014 [46] | 6294 | Cross‐sectional | Journal | United States | Substance use disorder | Experiences of Discrimination scale | National Epidemiological Survey on Alcohol and Related conditions II | 49.18 | 43.82 | Latinx | 0.16 | Fair |
| Squires et al. 2017 [164] | 203 | Cross‐sectional | Journal | United States | Alcohol use; smoking status | Unknown | Unknown | 32 | 44.1 | Black American | 0.098; 0.15 | Fair |
| Martin et al. 2019 [165] | 674 | Longitudinal (5 y) | Journal | United States | Composite substance use | Unknown | NA | 50 | Unknown | Latinx | 0.13 | Good |
| Buckner et al. 2022 [166] | 347 | Cross‐sectional | Journal | United States | Alcohol use; alcohol use problems | Perceived Ethnic Discrimination questionnaire | Unknown | Unknown | 21.4 | Latinx | 0.15; −0.04 | Fair |
| Buckner et al. 2021 [167] | 160 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Perceived Ethnic Discrimination questionnaire | Unknown | 82.90 | 21.7 | Black American | 0.27 | Fair |
| Zaso et al. 2022 [168] | 241 | Cross‐sectional | Journal | United States | Alcohol use | Perceived Ethnic Discrimination questionnaire | Unknown | 63; 72 | 20.03; 20.06 | Black American | 0.2; −0.01 | Fair |
| Call 2001 [169] | 97 | Cross‐sectional | Dissertation | United States | Alcohol use; alcohol use problems; illicit substance use; substance use problems | Index of Race‐Related Stress | NA | 100 | 33.9 | Black American | 0.174; 0.073; 0.216; 0.084 | Fair |
| Woodson 2021 [170] | 1146 | Cross‐sectional | Dissertation | United States | Composite substance use; presence–absence of alcohol use; presence–absence of cannabis use; presence–absence of tobacco use | Everyday Discrimination scale | National Survey of American Life Adolescent supplement | 50 | 15 | Black American | 0.21; 0.19; 0.14; 0.16 | Poor |
| Breeden 2004 [171] | 599 | Cross‐sectional | Dissertation | United States | Cannabis use; illicit substance use | Bespoke scale a | Woodlawn project | 51.80 | Unknown | Black American | 0.34; 0.37 | Poor |
| Swann et al. 2020 [172] | 352 | Cross‐sectional | Journal | United States | At‐risk alcohol use; at‐risk cannabis use | Brief Perceived Ethnic Discrimination questionnaire (community version) | Female‐assigned at birth sexual and gender minorities | 100 | 20.27 | Diverse | 0.17; 0.17; 0.26 | Fair |
| Lee et al. 2014 [173] | 136 | Cross‐sectional | Journal | United States | Composite substance use | Bespoke scale a | Korean Adoption Project | 54 | 15.16 | Asian American | 0.19 | Poor |
| Greene 1997 [174] | 189 | Cross‐sectional | Dissertation | United States | Composite substance use | Bespoke scale a | NA | 49.70 | Unknown | Latinx | −0.27 | Poor |
| Pittman et al. 2017 [175] | 649 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Index of Race‐Related Stress (brief version) | Unknown | 74.70 | 20.42 | Black American | 0.2 | Fair |
| Whitbeck et al. 2002 [176] | 195 | Cross‐sectional | Journal | United States | Alcohol use problems; composite substance use | Unknown | NA | 46 | 12.1 | Indigenous North American | 0.18; 0.29 | Poor |
| Pro et al. 2017 [177] | 322 | Cross‐sectional | Journal | United States | Cannabis use | Racial and Ethnic Microaggression scale | NA | 66.40 | 22.6 | Diverse | 0.08 | Poor |
| Somerville 2024 [178] | 199 | Cross‐sectional | Dissertation | United States | Binge drinking; cannabis use; illicit substance use | Index of Race‐Related Stress | Unknown | 100 | Unknown | Black American | 0.14; 0.2; 0.11 | Fair |
| Arreola 2024 [179] | 115 | Cross‐sectional | Dissertation | United States | At‐risk alcohol use | Brief Perceived Ethnic Discrimination questionnaire (community version) | NA | 81.70 | Unknown | Latinx | 0.28 | Fair |
| Jones 2024 [180] | 108 | Cross‐sectional | Dissertation | United States | Alcohol use problems; alcohol use; alcohol use disorder; at‐risk alcohol use | Perceived Online Racism scale | NA | 52.80 | 21.7 | Black American | 0.279; 0.036; 0.257; 0.231 | Fair |
| McDowell 2023 [181] | 126 | Cross‐sectional | Dissertation | United States | At‐risk alcohol use; at‐risk cannabis use | Schedule of Racist Events | NA | 39 | 23.6 | Black American | 0.43; 0.16 | Fair |
| Dean 2019 [182] | 152 | Longitudinal (up to 2 months) | Dissertation | United States | Composite substance use | Schedule of Racist Events | NA | 84.20 | 19.4 | Black American | 0.11 | Fair |
| Centeno et al. 2023 [183] | 703 | Cross‐sectional | Journal | United States | Illicit substance use | Adult and Peer Discrimination scale | Unknown | 47.60 | 16 | Latinx | 0.12 | Poor |
| Dyar et al. 2023 [184] | 304 | Longitudinal (up to 44 days) | Journal | United States | Alcohol use problems; cannabis use problems; alcohol use; cannabis use | Experiences of Discrimination scale | Unknown | Unknown | Unknown | Black American; Latinx; other ethnic minority | 0.11; −0.01; 0.07; 0.26; −0.08; 0.03; 0.07; 0.03; 0.13; −0.02; 0.09; 0.05 | Fair |
| Mora 2023 [185] | 55 | Cross‐sectional | Dissertation | United States | Composite substance use | General Ethnic Discrimination scale | NA | 57 | Unknown | Latinx | 0.35 | Poor |
| Moreno et al. 2024 [186] | 164 | Cross‐sectional | Journal | United States | Tobacco use | Experiences of Discrimination scale (modified) | Spit for Science study | 81.10 | 19.9 | Diverse | 0.27 | Poor |
| Assari 2023 [187] | 2514 | Longitudinal (Up to 36 months) | Journal | United States | Presence–absence of cannabis use; Presence–absence of tobacco use | Bespoke scale a | Adolescent Brain Cognitive Development Study | 49.50 | 9.5 | Black American | 0.083; 0.028 | Fair |
| Buckner et al. 2024 [188] | 164 | Cross‐sectional | Journal | United States | Alcohol use problems; alcohol use | Perceived Ethnic Discrimination questionnaire | Unknown | 82.90 | 21.7 | Black American | 0.27; 0.17 | Fair |
| Zapolski and Depperman 2023 [189] | 390 | Cross‐sectional | Journal | United States | At‐risk alcohol use; at‐risk cannabis use | Schedule of Racist Events | NA | 62 | 20.6 | Black American | 0.67; 0.24 | Fair |
| Cénat et al. 2023 [190] | 860 | Cross‐sectional | Journal | Canada | Composite substance use | Everyday Discrimination scale | Black Community Mental Health Project | 75.10 | 25 | Black Canadian | 0.348 | Fair |
| Schick et al. 2023 [191] | 52; 1743 | Cross‐sectional | Journal | United States; Canada | Alcohol use | Perceived Discrimination scale | Our Youth, Our Future | 50; 45.2 | 15.4 | Indigenous North American | 0.31; 0.14 | Fair |
| Espinosa et al. 2023 [192] | 7037 | Cross‐sectional | Journal | United States | Substance use disorder | Experiences of Discrimination scale | National Epidemiological survey on alcohol and related conditions wave 3 | 56.10 | 39.9 | Latinx | 0.13 | Fair |
| Macias Burgos et al. 2024 [193] | 426 | Cross‐sectional | Journal | United States | At‐risk alcohol use | Perceived Discrimination scale | Community Health Research and Implementation Science Data | 65.50 | 40 | Latinx | 0.41 | Fair |
| Nie 2024 [194] | 356 | Cross‐sectional | Journal | United States | Tobacco use | Bespoke scale a | NA | 49.90 | 48.3 | Asian American | 0.13 | Poor |
| Barry et al. 2023 [195] | 510 | Cross‐sectional | Journal | United States | Alcohol use; cannabis use; illicit substance use | Bespoke scale a | NA | 52 | Unknown | Indigenous North American | 0.03; 0.07; 0.0 | Poor |
| Morris et al. 2023 [196] | 152 | Cross‐sectional | Journal | United States | Alcohol use; tobacco use; cannabis use | Daily Life Experiences and Racism scale | Unknown | 74.50 | 21.5 | Black American | 0.018; −0.045; 0.024 | Poor |
Summary effect sizes
The summary effect sizes, 95% CI, P‐values, heterogeneity indices and prediction intervals for each AOD outcome are presented in Table 3. The data supporting these results can be found in Data S3.
| AOD outcome domain | n | m | k | r | 95% CI LL | 95% CI UL | P | Q | PI LL | PI UL | I2 (%) |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tobacco use | 13 256 | 17 | 23 | 0.07 | 0.002 | 0.14 | 0.04 | 250.34 | −0.26 | 0.4 | 91 |
| Alcohol use | 23 623 | 41 | 55 | 0.09 | 0.06 | 0.13 | <0.001 | 393.82 | −0.13 | 0.32 | 86 |
| Cannabis use | 12 630 | 20 | 28 | 0.08 | 0.04 | 0.12 | <0.001 | 170.46 | −0.12 | 0.29 | 84.20 |
| Illicit substance use | 8022 | 14 | 15 | 0.18 | 0.12 | 0.24 | <0.001 | 97.42 | −0.05 | 0.41 | 85.60 |
| Binge drinking | 9601 | 15 | 15 | 0.09 | 0.07 | 0.11 | <0.001 | 20.12 | 0.07 | 0.11 | 30.40 |
| At‐risk/hazardous alcohol use | 9445 | 29 | 33 | 0.24 | 0.17 | 0.3 | <0.001 | 614.2 | −0.12 | 0.59 | 94.80 |
| Alcohol use problems/consequences | 8127 | 23 | 30 | 0.2 | 0.15 | 0.25 | <0.001 | 184.56 | −0.06 | 0.47 | 84.30 |
| Substance use problems/consequences | 1240 | 5 | 5 | 0.08 | −0.09 | 0.25 | 0.33 | 32.25 | −0.5 | 0.62 | 87.60 |
| Alcohol use disorder | 41 387 | 7 | 9 | 0.19 | 0.13 | 0.26 | <0.001 | 177.67 | −0.05 | 0.43 | 96 |
| Substance use disorder | 21 051 | 5 | 6 | 0.25 | 0.14 | 0.36 | <0.001 | 218.45 | −0.15 | 0.64 | 97.70 |
| Composite substance use | 10 578 | 19 | 19 | 0.17 | 0.11 | 0.23 | <0.001 | 134.6 | −0.1 | 0.45 | 86.60 |
| Smoking status | 4795 | 7 | 9 | 0.15 | 0.04 | 0.26 | 0.009 | 80.39 | −0.26 | 0.56 | 90.00 |
| Presence–absence of alcohol use | 5017 | 6 | 6 | 0.15 | 0.09 | 0.21 | <0.0001 | 21.61 | −0.06 | 0.36 | 76.90 |
| Presence–absence of tobacco use | 5614 | 5 | 5 | 0.21 | 0.08 | 0.34 | 0.001 | 78.6 | −0.28 | 0.7 | 94.90 |
| Presence–absence of cannabis use | 6174 | 7 | 7 | 0.13 | 0.08 | 0.18 | <0.001 | 19.68 | −0.01 | 0.27 | 69.50 |
| At‐risk/hazardous cannabis use | 462 | 4 | 4 | 0.24 | 0.18 | 0.29 | <0.001 | 1.01 | 0.12 | 0.35 | 0 |
| Cannabis use problems/consequences | 1267 | 2 | 4 | 0.09 | 0.001 | 0.18 | 0.05 | 2.1 | −0.11 | 0.29 | 0 |
Positive associations between racial discrimination and 16 AOD outcomes were identified. The median effect size across outcomes was 0.15. The strongest associations were observed for at‐risk/hazardous alcohol use (r = 0.24, 95% CI = 0.17–0.3, I2 = 94.8%, m = 29, n = 9445), at‐risk/hazardous cannabis use (r = 0.24, 95% CI = 0.18–0.29, I2 = 0%, m = 4, n = 462) and substance use disorder (r = 0.25, 95% CI = 0.14–0.36, I2 = 97.7%, m = 5, n = 21 051). The weakest association was observed for tobacco use (r = 0.07, 95% CI = 0.002–0.14, I2 = 91%, m = 17, n = 13 256). See Data S4 for forest plots. Results of the sensitivity analysis with only fair/good‐rated studies can be found in Data S5.
Differences across AOD outcome domains
Based on non‐overlapping 95% CI, racial discrimination has a stronger association with at‐risk/hazardous drinking, alcohol use problems/consequences and alcohol use disorder (AUD) than with alcohol use. Similarly, at‐risk/hazardous cannabis use and substance use disorder had stronger associations with racial discrimination than with cannabis use. However, illicit substance use has a stronger association with racial discrimination than with cannabis use. AOD outcomes with overlapping 95% CI, however, do not necessarily indicate any differences. Although for AOD outcomes whose 95% CI contain the mean correlation coefficient for another AOD outcome, equivalent effects can be inferred [197]. As such, no differences were observed between tobacco use and alcohol use, cannabis use and smoking status. No differences were found between alcohol use, cannabis use and binge drinking. Furthermore, no differences were identified between cannabis use and cannabis problems, or substance use disorder and AUD.
Publication bias
Across tobacco use, alcohol use, cannabis use, illicit substance use and alcohol use problems/consequences outcomes, funnel plots did not indicate asymmetry, which was supported by non‐significant Eggers tests (P = 0.06–0.96). Correspondingly, doiplots and the LFK index across these domains did not suggest plot asymmetry (LFKs = −0.77 to 0.6). However, for the binge drinking domain, minor asymmetry was suspected in the funnel plot, yet the Eggers test was non‐significant (t = −0.88, P = 0.4) and doiplots and LFK index indicated minor plot asymmetry (LFK = −1.36). Similarly, a non‐significant Eggers test was observed (t = 0.54, P = 0.59) for the composite substance use domain with minor plot asymmetry in doiplots and LFK index (1.53). Contrastingly, for at‐risk/hazardous alcohol use, the Eggers test indicated plot asymmetry (t = −2.92, P = 0.007), while doiplot and LFK index did not. Moreover, for cannabis use problems/consequences and at‐risk/hazardous cannabis use, doiplots and LFK index indicated minor plot asymmetry (LFK = −1.35, 1.92). For the substance use problems/consequences, AUD, substance use disorder, presence–absence of alcohol use, presence–absence of tobacco use domains, major plot asymmetry was detected via dioplot and LFK index (LFKs = −4.16, 3.24). Doiplots and LFK index for smoking status and presence–absence of cannabis use did not indicate plot asymmetry (LFK = 0.84, −0.35).
Moderation analysis
Table 4 provides effect sizes, 95% CI, P‐values and I2 values for subgroup analyses.
| AOD domain | Moderator | k | r | 95% CI LL | 95% CI UL | I2 (%) | P (subgroup differences) |
|---|---|---|---|---|---|---|---|
| Tobacco use | Study design | 0.45 | |||||
| Longitudinal | 9 | 0.1 | 0.05 | 0.14 | 55.60 | ||
| Cross‐sectional | 14 | 0.05 | −0.05 | 0.16 | 94 | ||
| Race/ethnicity | <0.001 | ||||||
| Black American | 11 | 0.06 | 0.01 | 0.12 | 81.70 | ||
| Latinx | 4 | 0.06 | −0.023 | 0.14 | 89.20 | ||
| Indigenous North American | 2 | 0.27 | 0.2 | 0.35 | 0 | ||
| Asian American | 2 | −0.18 | −0.8 | 0.43 | 99 | ||
| Gender | 0.22 | ||||||
| Male | 3 | 0.14 | 0.02 | 0.27 | 72.50 | ||
| Female | 2 | 0.03 | −0.09 | 0.15 | 0 | ||
| Age group | 0.1 | ||||||
| Youth and adolescents | 11 | 0.06 | −0.11 | 0.12 | 94.60 | ||
| Young adult and adults | 5 | 0.13 | 0.04 | 0.22 | 76.20 | ||
| Exposure timing | 0.05 | ||||||
| Lifetime | 3 | 0.18 | 0.06 | 0.29 | 62.80 | ||
| Past year | 4 | 0.09 | −0.05 | 0.23 | 80.60 | ||
| Less than past year | 7 | −0.06 | −0.21 | 0.1 | 94.1 | ||
| Alcohol use | Study design | 0.77 | |||||
| Longitudinal | 17 | 0.1 | 0.04 | 0.16 | 75.50 | ||
| Cross‐sectional | 38 | 0.09 | 0.05 | 0.13 | 88.70 | ||
| Race/ethnicity | 0.72 | ||||||
| Black American | 27 | 0.11 | 0.05 | 0.17 | 91.40 | ||
| Latinx | 11 | 0.08 | 0.03 | 0.14 | 73 | ||
| Asian American | 6 | 0.06 | −0.01 | 0.14 | 77.10 | ||
| Indigenous North American | 6 | 0.12 | 0.03 | 0.21 | 71.50 | ||
| Gender | 0.5 | ||||||
| Male | 3 | 0.09 | 0.04 | 0.15 | 18.10 | ||
| Female | 4 | 0.12 | 0.07 | 0.17 | 0 | ||
| Age group | 0.25 | ||||||
| Youth and adolescents | 20 | 0.08 | 0.04 | 0.13 | 83.60 | ||
| Young adult and adults | 13 | 0.15 | 0.05 | 0.13 | 92.50 | ||
| Exposure timing | 0.09 | ||||||
| Lifetime | 6 | 0.06 | −0.01 | 0.12 | 32 | ||
| Past year | 12 | 0.16 | 0.07 | 0.25 | 93.10 | ||
| Less than past year | 14 | 0.04 | −0.02 | 0.09 | 75.6 | ||
| Cannabis use | Study design | 0.95 | |||||
| Longitudinal | 11 | 0.08 | 0.03 | 0.14 | 58.10 | ||
| Cross‐sectional | 17 | 0.08 | 0.02 | 0.14 | 89.00 | ||
| Race/ethnicity | 0.78 | ||||||
| Black American | 15 | 0.1 | 0.04 | 0.17 | 89.50 | ||
| Latinx | 4 | 0.08 | −0.01 | 0.16 | 80.80 | ||
| Indigenous North American | 2 | 0.07 | −0.001 | 0.15 | 0.00 | ||
| Gender | 0.9 | ||||||
| Male | 3 | 0.09 | −0.06 | 0.23 | 81 | ||
| Female | 3 | 0.07 | −0.06 | 0.21 | 73.00 | ||
| Age group | 0.1 | ||||||
| Youth and adolescents | 16 | 0.05 | 0.006 | 0.1 | 83.60 | ||
| Young adult and adults | 5 | 0.16 | 0.04 | 0.28 | 81 | ||
| Exposure timing | 0.52 | ||||||
| Lifetime | 2 | 0.17 | −0.16 | 0.5 | 97.30 | ||
| Past year | 6 | 0.08 | −0.01 | 0.17 | 67.30 | ||
| Less than past year | 11 | 0.03 | −0.02 | 0.09 | 72.3 | ||
| Illicit substance use | Study design | 0.44 | |||||
| Longitudinal | 2 | 0.13 | −0.003 | 0.26 | 49.30 | ||
| Cross‐sectional | 13 | 0.18 | 0.12 | 0.25 | 86.60 | ||
| Race/ethnicity | 0.28 | ||||||
| Black American | 6 | 0.17 | 0.07 | 0.27 | 86.50 | ||
| Latinx | 5 | 0.18 | 0.12 | 0.24 | 72.40 | ||
| Indigenous North American | 2 | 0.06 | −0.07 | 0.2 | 73.50 | ||
| Age group | 0.13 | ||||||
| Young adult and adults | 2 | 0.26 | 0.05 | 0.48 | 90.80 | ||
| Youth and adolescents | 5 | 0.10 | 0.05 | 0.14 | 49.10 | ||
| Exposure timing | 0.03 | ||||||
| Lifetime | 2 | 0.32 | 0.18 | 0.46 | 54.80 | ||
| Past year | 3 | 0.15 | 0.08 | 0.21 | 0 | ||
| Binge drinking | Study design | 0.46 | |||||
| Longitudinal | 4 | 0.07 | 0.009 | 0.13 | 25.10 | ||
| Cross‐sectional | 11 | 0.1 | 0.07 | 0.12 | 35.50 | ||
| Race/ethnicity | 0.84 | ||||||
| Black American | 8 | 0.09 | 0.04 | 0.14 | 45.60 | ||
| Latinx | 2 | 0.1 | 0.07 | 0.13 | 0 | ||
| Indigenous North American | 2 | 0.07 | −0.05 | 0.19 | 48.40 | ||
| Gender | 0.42 | ||||||
| Male | 3 | 0.1 | 0.04 | 0.17 | 0 | ||
| Female | 3 | 0.05 | −0.08 | 0.17 | 58.40 | ||
| Age group | 0.23 | ||||||
| Young adult and adults | 2 | 0.15 | 0.04 | 0.25 | 51.70 | ||
| Youth and adolescents | 3 | 0.06 | −0.03 | 0.15 | 45.80 | ||
| At‐risk/hazardous alcohol use | Study design | 0.01 | |||||
| Longitudinal | 3 | 0.13 | 0.07 | 0.19 | 0 | ||
| Cross‐sectional | 30 | 0.24 | 0.18 | 0.31 | 95.00 | ||
| Race/ethnicity | 0.06 | ||||||
| Black American | 14 | 0.27 | 0.16 | 0.37 | 96.20 | ||
| Latinx | 6 | 0.22 | 0.1 | 0.34 | 91.70 | ||
| Asian American | 4 | 0.07 | −0.05 | 0.2 | 74.50 | ||
| Gender | 0.1 | ||||||
| Male | 5 | 0.19 | 0.04 | 0.35 | 84.70 | ||
| Female | 7 | 0.19 | 0.12 | 0.26 | 57.10 | ||
| Exposure timing | 0.94 | ||||||
| Lifetime | 6 | 0.23 | 0.12 | 0.34 | 86.10 | ||
| Past year | 3 | 0.26 | −0.15 | 0.67 | 99 | ||
| Less than past year | 6 | 0.26 | 0.09 | 0.44 | 95.80 | ||
| Alcohol use problems/consequences | Study design | 0.03 | |||||
| Longitudinal | 8 | 0.14 | 0.08 | 0.19 | 24.90 | ||
| Cross‐sectional | 22 | 0.23 | 0.16 | 0.29 | 86.90 | ||
| Race/ethnicity | 0.23 | ||||||
| Black American | 12 | 0.16 | 0.08 | 0.23 | 73 | ||
| Latinx | 5 | 0.16 | 0.05 | 0.27 | 61.30 | ||
| Asian American | 4 | 0.25 | 0.14 | 0.36 | 73.20 | ||
| Indigenous North American | 3 | 0.28 | 0.17 | 0.38 | 38.80 | ||
| Age group | 0.04 | ||||||
| Young adult and adults | 11 | 0.15 | 0.1 | 0.2 | 49.70 | ||
| Youth and adolescents | 3 | 0.28 | 0.17 | 0.38 | 38.80 | ||
| Exposure timing | 0.13 | ||||||
| Lifetime | 3 | 0.08 | 0.02 | 0.13 | 0 | ||
| Past year | 4 | 0.16 | 0.03 | 0.29 | 80.20 | ||
| Less than past year | 11 | 0.19 | 0.08 | 0.30 | 78.1 | ||
| Alcohol use disorder | Race/ethnicity | 0.33 | |||||
| Black American | 2 | 0.23 | 0.2 | 0.26 | 0 | ||
| Latinx | 2 | 0.32 | 0.15 | 0.48 | 95.20 | ||
| Gender | 0.45 | ||||||
| Male | 2 | 0.28 | 0.04 | 0.51 | 98.50 | ||
| Female | 2 | 0.18 | 0.09 | 0.27 | 89.90 | ||
| Composite substance use | Study design | 0.64 | |||||
| Longitudinal | 5 | 0.15 | 0.07 | 0.23 | 65 | ||
| Cross‐sectional | 14 | 0.18 | 0.1 | 0.26 | 89 | ||
| Race/ethnicity | 0.003 | ||||||
| Black American | 7 | 0.13 | 0.09 | 0.18 | 62.80 | ||
| Latinx | 4 | 0.07 | −0.17 | 0.31 | 91.30 | ||
| Indigenous North American | 3 | 0.29 | 0.23 | 0.35 | 0 | ||
| Age group | 0.01 | ||||||
| Young adult and adults | 2 | 0.25 | 0.18 | 0.33 | 0 | ||
| Youth and adolescents | 10 | 0.11 | 0.03 | 0.19 | 83.90 | ||
| Smoking status | Study design | 0.67 | |||||
| Longitudinal | 2 | 0.11 | −0.04 | 0.27 | 70.70 | ||
| Cross‐sectional | 7 | 0.16 | 0.02 | 0.3 | 92.20 | ||
| Race/ethnicity | 0.5 | ||||||
| Black American | 5 | 0.2 | 0.01 | 0.4 | 94.40 | ||
| Latinx | 2 | 0.1 | −0.002 | 0.2 | 60.90 | ||
| Asian American | 2 | 0.09 | 0.03 | 0.14 | 0 | ||
| Exposure timing | 0.36 | ||||||
| Lifetime | 3 | 0.27 | −0.03 | 0.57 | 96.60 | ||
| Past week | 2 | 0.11 | −0.04 | 0.27 | 70.70 | ||
| Presence–absence of alcohol use | Study design | 0.55 | |||||
| Longitudinal | 3 | 0.13 | 0.04 | 0.22 | 81.10 | ||
| Cross‐sectional | 3 | 0.17 | 0.07 | 0.27 | 74.10 | ||
| Presence–absence of cannabis use | Study design | 0.36 | |||||
| Longitudinal | 3 | 0.09 | 0.06 | 0.13 | 0 | ||
| Cross‐sectional | 4 | 0.13 | 0.05 | 0.21 | 74.20 | ||
| Presence–absence tobacco use | Study design | 0.94 | |||||
| Longitudinal | 2 | 0.02 | −0.14 | 0.55 | 98.30 | ||
| Cross‐sectional | 3 | 0.22 | 0.1 | 0.33 | 72.00 |
The results of the subgroup analyses can be found in Table 4. The majority of these analyses did not provide evidence for moderation effects. Yet across racial/ethnic groups, differences were found for tobacco use, with Black American and Latinx subgroups having comparable positive correlation coefficients, however, Indigenous North Americans had the strongest positive associations, and a negative association was observed for Asian Americans. Likewise, differences were also identified in the composite substance use outcome, with Indigenous North Americans again having the strongest associations, compared to Black Americans and Latinxs. Within the at‐risk/hazardous alcohol use domain, however, Black Americans had the strongest association, followed by Latinx and then Asian Americans.
Moderation by gender did not reveal any differences between male and female‐only samples across all AOD outcome domains.
In the alcohol use problems/consequences domain, adolescents and/or youths had stronger associations than young adults and/or adults. However, across the tobacco use, alcohol use, cannabis use, illicit substance use and binge drinking domains, there was a general trend of larger effect sizes in the young adult and/or adult subgroups.
Concerning moderation by study design, in the at‐risk/hazardous alcohol use and alcohol use problems/consequences domains, stronger associations were observed for cross‐sectional studies, compared to longitudinal studies.
Subgroup analysis by racial discrimination exposure timing demonstrated that, in the illicit substance use domain, lifetime exposure had a stronger association than past‐year exposure. Likewise, in the tobacco use domain, lifetime exposure also had the strongest positive association, followed by past year exposure and a negative association was observed for less than past year exposure.
DISCUSSION
We used meta‐analytic methods to understand the association between racial and AOD outcomes. The findings suggest that there is considerable evidence to suggest that racial discrimination is a consistent correlate of distinct AOD outcomes in minoritised racial/ethnic groups, predominantly based in the United States. Across AOD outcomes, all the associations were in the small range [198], but varied in magnitude. This study also found preliminary evidence that race/ethnicity, age group, exposure timing and study design moderate these associations to some degree.
Given that most of the participants included within this review resided in the United States, it is likely that our findings mainly reflect the nature of racial discrimination and AOD in the United States context. Our findings, therefore, demonstrate that the United States’ history of racism continues to harm racially and ethnically minoritised groups residing in the country. However, considering all of the studies included in this review were conducted in post‐colonial and imperialist states/regions, the findings also point to enduring negative effects of these systems.
Previous research
The findings of this study align with previous meta‐analyses, which report significant, positive associations between racial discrimination and AOD [35, 36, 199]. The magnitude of the associations between racial discrimination and alcohol use, binge drinking and alcohol use problems/consequences in the current study were similar to those reported by Desalu et al. [200], but the current study reports larger effect sizes for at‐risk/hazardous alcohol use and AUD. This may arise from the current study using the full Alcohol Use Disorder Identification Test (AUDIT) scale to measure at‐risk/hazardous alcohol use, whereas Desalu et al. [200] use the Alcohol Use Disorders Identification Test‐Consumption (AUDIT‐C). The full AUDIT captures health‐harming aspects of alcohol use, such as dependency and problems related to use, which have been previously demonstrated to have stronger associations with racial discrimination than use alone [95, 201, 202]. Moreover, despite both studies using the same operationalisation for AUD, differences in the racial discrimination‐AUD relationship across race/ethnicity might explain this discrepancy, as this study included a diverse minoritised racial/ethnic sample, while Desalu et al.'s [200] sample was Black American only.
Stronger associations between racial discrimination and at‐risk/hazardous alcohol use, alcohol problems and AUD compared to alcohol use have been reported in previous work [46, 93, 110, 202]. This suggests that more harmful alcohol outcomes are more closely related to racial discrimination, rather than just use. Substance use motives may explain this, as motives to avoid negative internal states are more related to maladaptive alcohol use, whereas social and enhancement motives are stronger predictors of use 32, 203, 204]. Unexpectedly, comparable associations between alcohol use and binge drinking were found, despite the unhealthy nature of binge drinking. Yet some evidence suggests that social motives are the strongest predictors of binge drinking [205, 206].
Our finding that racial discrimination had a weaker association with cannabis use, compared to other illicit substance use, is important. It demonstrates that incorporating cannabis into measures of illicit substance use in the context of racial discrimination is inappropriate and suggests that minoritised racial/ethnic groups exposed to racial discrimination may be at specific risk for illicit substance use. This finding corresponds to that reported by Garrett et al. [39] in their analysis of Cherokee Nation adolescents, which also observed that effect sizes for prescription and other illicit drug use were larger than for cannabis use. A recent analysis of diverse racial/ethnic groups (including White respondents) also reported larger effect sizes for the association between discrimination and illicit substances such as methamphetamines, than for cannabis, tobacco or alcohol, but smaller or comparable effect sizes for other illicit substances, including cocaine [207]. As the current study used a composite outcome for illicit substance use, the larger effect size may be driven by racial discrimination's stronger association with specific illicit substances. Yet it should be noted that, as the studies that contributed to both the illicit substance and cannabis use outcomes all used US‐based samples, these findings are likely only applicable to minoritised racial/ethnic groups based in the United States.
The finding that Indigenous North Americans appear to be at higher risk for tobacco use when exposed to racial discrimination compared to Black Americans, Latinxs and Asian Americans, may be because of the accessibility of tobacco, as Indigenous North Americans living on reservations report paying less for tobacco than Black Americans, Latinxs or Asian Americans [208] because of the exploitative marketing practices of tobacco companies [209, 210]. Therefore, the higher price point for tobacco products for other minoritised racial/ethnic groups may make the use of tobacco to cope with racial discrimination stress unviable. The similar finding observed for the composite substance use outcome suggests that Indigenous North Americans may be particularly vulnerable to the pernicious effects of racial discrimination. However, the current study is the first to document these findings, as studies that have assessed the moderating role of race/ethnicity in the racial discrimination‐AOD associations have consistently not included Indigenous North American participants.
The smaller effect sizes observed for Asian Americans, compared to Black Americans and Latinxs in the at‐risk/hazardous drinking domain, could be accounted for by the model minority myth (MMM). As there is some evidence to suggest that internalisation of the MMM impedes the ability to accurately perceive discrimination [211, 212], which may affect the degree to which discrimination is appraised as stressful. Accordingly, Asian Americans may be less likely to engage in maladaptive coping strategies, such as AOD, to cope with the stress of racial discrimination.
The larger effect sizes for young adults and/or adults across six of the AOD outcome domains, compared to adolescents and/or youths, may be accounted for by the simultaneous depletion of familial protective influences and increased exposure to racial discrimination, which have been documented to occur with aging into adulthood [213, 214, 215]. However, the opposite trend observed in the alcohol problems/consequences domain was unexpected. Notably, all the participants in the youth and/or adolescent subgroup for this outcome were Indigenous North American, who may be at particular risk for AOD when exposed to racial discrimination as discussed above. Therefore, this effect may be more driven by race/ethnicity than age.
The subgroup analyses identified that cross‐sectional studies produced larger effect sizes than longitudinal studies in the at‐risk/hazardous alcohol use and alcohol use problems domains. This finding aligns with Paradies et al. [216] meta‐analyses, where they noted that racism produced stronger effects on negative mental health in cross‐sectional studies, which has also been reported for minoritised ethnic groups living in the United Kingdom [217]. These findings suggest that the effects of racial discrimination may attenuate over time, potentially because of the progressive development of resiliency or the recruitment of protective resources, post‐exposure to racial discrimination. Alternatively, there is the possibility that the larger effect sizes observed for cross‐sectional studies reflect common‐method bias, which can inflate effect size estimates, as cross‐sectional studies are particularly vulnerable to this form of bias [218, 219, 220].
Our findings that lifetime exposure to racial discrimination had a stronger association with illicit substance and tobacco use, compared to recent exposure, are consistent with Carter et al.'s [36] results. Together, these findings support the weathering hypothesis [221], whereby as risk accumulates across the life course, so does the likelihood of negative health outcomes. Yet it is unclear why exposure timing is a significant moderator for some AOD outcomes and not others.
Limitations
The present study should be interpreted in the context of some limitations. First, our subgroup analysis by race/ethnicity could only be conducted across broad racial categories. It was uncommon in the primary studies to capture data on distinct ethnic identities, and therefore, grouped participants were grouped primarily by the continent of heritage. This overlooks the substantial heterogeneity within these broad racial/ethnic categorisations. Therefore, future research should aim to collect detailed data on the ethnic identities of participants to facilitate more thorough investigations of how the racial discrimination‐AOD associations operate across distinct ethnic identities. Second, for some of the AOD outcomes, including at‐risk/hazardous cannabis use and cannabis use problems/consequences, only a small number of studies contributed to effect sizes and therefore should be interpreted with caution. Moreover, the low number of effect sizes available within subgroups, notably in the male and female‐only samples, likely limited our statistical power to detect moderation effects. Third, a small minority of studies included in this analysis were rated as good methodology quality, therefore, the inclusion of predominantly low to moderate‐quality studies can threaten the reliability and validity of these findings.
Future research
Our findings have identified key areas for future research. First, the racial discrimination‐AOD relationships warrant further research in Indigenous North American populations, as this study provides preliminary evidence that they may be especially vulnerable to the impacts of racial discrimination. Second, as the vast majority of studies included in this review were based on US samples, research on this association in minoritised racial/ethnic groups outside of the United States is required. Third, this study suggests that future research should give considerably more attention to how the timing of racial discrimination impacts different AOD outcomes.
CONCLUSIONS
In summary, the present study provides consistent evidence that racial discrimination is associated with AOD outcomes in minoritised racial/ethnic groups, yet this association varies to some degree across distinct AOD outcomes. We, therefore, provide evidence against the use of composite AOD measurements in this field. Moreover, the present study also suggests that these associations are somewhat modified by race/ethnicity, age, exposure timing and study design. The findings of this study could be informative for the development of prevention and intervention practices to mitigate the harmful effects of racism within minoritised racial/ethnic groups.
DECLARATION OF INTERESTS
None.
CLINICAL TRIAL REGISTRATION
PROSPERO registration ID CRD42022381762 (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=381762)
Supporting information
ACKNOWLEDGEMENTS
We acknowledge and thank the Economic and Social Research Council and the Biotechnology and Biological Sciences Research Council for funding this project. We also thank Martyna Kosciuszko for taking time out of her PhD to contribute to this project. C.J.A. is supported by National Institute for Health and Care Research (NIHR) Manchester Biomedical Research Centre and NIHR Greater Manchester Patient Safety Research Collaboration. Views of the authors do not necessarily represent those of the NIHR or Department of Health and Social Care. R.E. is supported by NIHR Manchester Biomedical Research Centre. Views of the authors do not necessarily represent those of the NIHR.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available in supplementary materials 5.