Background

Structural racism has been well documented as an important social determinant of health (SODH), the non-medical factor that influences health outcomes [1] and a key driver of health inequities [1,2,3] and a fundamental cause of disease [4] worldwide. Globally, structural racism is considered as a critical determinant of racial inequality in health [5,6,7] even more than 50 years after the ratification of the International Convention on the Elimination of All Forms of Racial Discrimination (ICERD) [8].

Although considered as an important SODH, the term ‘structural racism’ is often conflated and interchangeably used to refer institutional racism or systematic racism in public health literature. But more recently scholars have clarified this confusion asserting that they are different [9,10,11,12].

After analyzing the evolution of different terms and definitions, Bailey et al.’s definition [9] has been considered as the most contemporary definition [13]. We adopt Bailey et al.’s definition which defines.

structural racism as the totality of ways in which multiple macro structural systems (in housing, education, employment, earnings, benefits, credit, media, health care, criminal justice, and so on) and interconnected institutions mutually interact to assert biased and discriminatory policies, practices, beliefs, and distribution of resources for people in a racialized group [9, 11]. Dean and colleagues have argued that such a definition of structural racism establishes a clear distinction from: institutional racism, which associates racism within a particular type of institution, organization or in policy for/against a racial group; systemic racism, that refers to racialized systems of power; racial discrimination, which deals with practices originating from racist beliefs; and cultural racism, which upholds or reflects the values, ideologies, societal norms and practices of a particular racial group [13]. According to Jones, a) internalized racism, which refers to the internalization of oppression by a particular racial group; b) interpersonal bias or racism which refers to ‘personally mediated’ biases or racism, and c) institutional racism are 3 levels of racism [14] which can be operationalized under the larger construct of structural racism. Such mutually reinforced interactions of macro systems and institutions result in adverse health outcomes or racial inequities in health [9] which are determined by social gradient or systematic differences in health for different groups [15].

For example, according to the World Bank data life expectancy at birth in Sierra Leone [9] is 55 years whereas life expectancy at birth in Japan [10] is 84.3 years. There are less resources in Sierra Leone than in Japan leading to speculations about inequity and inaccessibility between countries.

Substantial evidence on vast racial inequities in health in the United States have been well documented [11], although racial health inequities have been a part of government statistics since the founding of colonial America [8, 12]. Any historical account of structural racism goes back to the genocide, enslavement of black people and the indigenous people by the early colonizers [9] followed by the creation of systems of racial oppression by legal initiatives such as the Jim Crow laws [11, 16].

Racial inequalities are not only in health centered organizations, but they also bleed into other organizations such as law enforcement which negatively impacts the safety and wellbeing of people of particular group.

Bor et al. for example, have documented Black Americans are ‘as nearly three times more likely than White Americans’ to be killed by police and ‘five times more likely than White Americans’ to be murdered while unarmed [17]. They are comprised of more than 40% of victims of all police homicides nationwide [17].

There has been a significant increase in the number of studies to examine the health consequences and impact pathways of structural racism in high-income countries. Dennis et al. identified eight mutually reinforcing domains of structural racism in the United States: 1) civil and political rights (including voting rights and citizenship); 2) land/housing (including neighborhoods); 3) education; 4) jobs/benefits/wealth; 5) justice system; 6) health (including health care); 7) migration and movement (including immigration, forced removal, and limited mobility); and 8) racial climate [18]. In terms of pathways between racism and health, Bailey et al. identified 9 empirical pathways in which structural racism determines health outcomes [9].

Similarly, Agénor et al. identified 843 US state laws explicitly or implicitly related to structural racism and found the 10 contemporary mutually reinforcing legal domains (i.e., voting rights laws, stand-your-ground laws, racial profiling laws, mandatory minimum prison sentencing laws, immigrant protections, fair-housing laws, minimum wage laws, predatory lending laws, laws concerning punishment in schools, and stop-and-identify laws) that directly and indirectly impact health in all 50 states and the District of Columbia from 2010 through 2013 [19].

There are numerous empirical studies that document wide range of health impacts for many historically marginalized racial/ethnic groups in the United States (e.g., African American/Black, Latinx, Hispanic, American Indian, Alaska Native, Native Hawaiian or other Pacific Islanders, Asian Americans) and other high income countries [8, 20,21,22].

For example, drawing on ecological studies [16,17,18, 23], Lane et al. identified ecological factors and potential mechanisms that led to disproportionately higher rates of heterosexually transmitted human immunodeficiency virus (HIV) among women of color in Syracuse, New York [24]. They found in Onondaga county, the tenth most populated county in New York State, that the number of African American women and Latinx women diagnosed with acquired immunodeficiency syndrome (AIDS) were nearly at 12.5 times and 9 times higher, respectively, than that of white women [24]. This study also explained how ecological factors (e.g., disproportionate and/or excessive incarceration) can lead to critical changes in population demographics (e.g., low male-to-female sex ratio in Syracuse) resulting in a ‘long term double punch’ effect (imbalanced sex ratio is likely to be associated with females choosing multiple sexual partners) in the increase of HIV transmission among African Americans [24].

Similarly, in Australia the impacts of racism, dispossession, and colonization of aboriginal Australians [25, 26] and statistically significant evidence of racist beliefs, emotions, or practices among healthcare providers in relation to minority groups [27] have been reported. Similar practices have been reported in other high-income countries such as New Zealand [28], Canada [29], Norway [30], France [31].

There are studies that document health impacts of structural racism of all sorts. Groos et al. for example, relates the health outcomes of structural racism to ‘from womb to tomb’ in their systematic review [32]. They found both mental and physical impacts including stress, anxiety, poor psychological well-being, colorectal cancer survival, myocardial infarction, mean arterial blood pressure, episodic memory function, behavioral changes, poor adherence to hypertensive treatment, and delayed HIV testing across the population [32].

It has been strongly argued in previous literature that empirical studies on structural racism and health require scientific theory, hypotheses, data, and research methods [22] in order to systematically capture the broad historical and contemporary impacts of structural racism on health [14].

Critical analysis of evidence on structural racism therefore, indicates an ever increasing scholarship on development of different theoretical frameworks (e.g. ecological, public health critical praxis, critical race theory), approaches, (e.g. sequential approach, mixed method) designs (e.g. more focus on cross-sectional design), measures (largely on common domains such as residential segregation), and indexes or scales of measurement (linear and single dimensional) to explain the etymology, pathways, health impacts and potential solutions to structural racism.

Drawing upon three frameworks: eco-social theory [33], fundamental cause theory [4, 34] and public health critical race praxis [35], Alson et al. critically examined the studies from 2000–2019 to identify studies that used quantitative measures of exposure to systemic racism in population reproductive health studies [36]. Similarly, a wide range of study designs were used in examining the health outcomes of structural racism in the context of the United States. Bor et al. for example, employed quasi-experimental study design to quantify the population-level health impacts of police killings as one of ecological factors of social determinants of health using US behavioral risk factor surveillance system (BRFSS), a nationally representative, telephone-based survey data of non-institutionalized adults aged 18 years and older [17].

Although various frameworks, study designs, measures of exposure, measures of outcome, and indexes have been used by different scholars in studying structural racism, the selection of design, measures and index of measures that responds to the framework of social determinants of health remains unclear and has not been comprehensively studied.

Several challenges have been documented in the current literature. One of the most documented challenges in undertaking empirical studies on structural racism is to develop rigorous methods to study the health impact of structural determinants of racial inequality including laws; institutional policies and practices; national, regional, state, and local economic and political infrastructures; and neighborhood and workplace conditions [24, 25, 37, 38]. For example, the use of cross-sectional design over long-term longitudinal study design on the mental health impact of structural racism has been associated with larger effect size [1]. Paradies et al. found although still statistically significant, the association between racism and health in a long-term longitudinal study with more than one year between exposure and outcome was weaker than cross-sectional or longitudinal study with relatively shorter duration [1]. Some other key methodological challenges include limitations with cross-sectional study design to make conclusion about the temporality and causality in the association between the exposure and outcome [39] and variation of estimates by the geographic unit of analysis [9, 40, 41].

The other key challenges that have been widely documented by the researchers studying structural racism include a large array of measures of exposures and outcomes [26] and measurement scales or indexes [42]. One of the most contested discussions regarding studying structural racism is to deal with the heterogeneity of definitions of measures and indexes of measurement [32]. While some scholars have preferred domain specific measures [43], other scholars have advocated for the use of index measures to study multidimensional impacts of structural racism [44]. While the most common single measure domain is residential segregation [32], the recent development is the introduction of a multidimensional measure of structural racism in Public Use Microdata Areas (PUMA) [45]. Such a heterogeneity of index of measurement indicates the lack of consensus of what index of measurement best fit in investigation of structural racism from SODH context, which calls for the need of comprehensive synthesis of indexes of measurement used in public health research.

Although there have been several studies on the health implications of structural racism, the lack of study that comprehensively maps study methods including study design, measures, and indexes in the context of SODH framework seems to exist in the literature.

Therefore, this review aimed to systematically map and summarize the global evidence on the use of different research approaches and methods and identify additional research on measures that are needed in studying structural racism as social determinant of health.

This review contributes to knowledge by providing researchers and organizations interested in combatting structural racism a comprehensive synthesis of methods, measures of exposure and health outcome and the scales that have been used in studying structural racism. This review is needed to improve the current state of research surrounding structural racism within the public health domain. This paper will guide other researchers interested in racial equity in choosing appropriate and contextual study design, measures, and scales of measurement for a research design.

Methods

This scoping review aimed (1) to describe the literature on structural racism as it has used different study designs in studying structural racism; (2) to identify the measures of exposures and outcomes of structural racism for SODH framework; (3) to identify different measurement scales or indexes used by different by scholars, and (4) to describe different methodological challenges in studying structural racism from the framework of SODH. In undertaking this scoping review, we followed the framework of Levac et al. [46] which was based on the framework given by Arksey and O’Malley [28]. As consultation with key stakeholders, the 6th step in the methodology, is optional, consultation with key stakeholders was not conducted under this current review.

Step 1—Identify the research question

The broader research question for this review was:

What public health research methods have been used in studying structural racism as social determinant of health?

The specific research questions developed for this review were:

  1. 1.

    What study designs have been used in studying structural racism as social determinants of health?

  2. 2.

    What measures of exposure and outcome have been reported in structural racism studies?

  3. 3.

    What measurement scales or indexes have been used to explain the health impacts of structural racism?

  4. 4.

    What methodological challenges in studying structural racism as social determinants of health have been reported in the studies?

Step 2—Identify the relevant literature

The identification of relevant literature involved a systematic search in five primary databases: PubMed, Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycInfo, Web of Science, and ProQuest. In addition, google search was used for grey literature. The development of systematic search strategy included the breaking down the broader research question into key concepts and finding key search terms for the respective concepts. Each domain of SODH and related terms were used as key word in the search strategy. To develop a comprehensive list of key words, relevant studies were reviewed. Key words were used in database to develop subject headings. A final search string including key words and subject heading terms for each concept was used. Finally, a combined search syntax with search strings of all the concepts was run in the respective database. Modifications to relevant field tagging were also done. Several rounds of consultations with the librarian throughout the development and database search took place. The systematic search across databases was conducted on 30th January 2022 and was repeated on 30th April 2022. The peer-reviewed English articles were considered for the review. Table 1 provides the complete search syntax for PubMed. Syntax for other databases can be available on request.

Table 1 PubMed Search Syntax
Full size table

Step 3—Select the literature

Bibliographic citations which were retrieved through the electronic databases were imported into Rayyan [29], a systematic review management software. Bibliographic databases were reviewed and duplicates were removed. Two stage relevance screening was conducted using the inclusion criteria (Table 2).

Table 2 Inclusion and exclusion criteria used to identify the articles for the review
Full size table

At first, titles and abstracts were screened. Secondly, full texts of potentially relevant articles were screened. Articles were independently screened by two reviewers (MKA and DS) at both stages of screening. Articles were relevant to this review if they considered social determinants of health (SODH) framework or any other theoretical framework relevant to SODH and studied structural racism as a determinant of health.

The articles that were considered ineligible by one reviewer were cross examined by the 2nd reviewer. At the second level the reviewers discussed the papers one by one for the ones they disagreed with for exclusion. Article was included when two reviewers reached to an agreement.

Step 4—Chart the data

A priori structured data charting matrix was developed by MKA, reviewed by DS and supervised by SDL. The development of data charting matrix was done in accordance with broader and specific research questions. The charting matrix captured information from studies on year of publication, country of study, study objective, theoretical concepts or frameworks, study design, measures of exposure and outcome, measurement scale or indexes, data analysis plan, key findings and methodological challenges reported in the study. After charting data from all studies, the second reviewer screened the matrix for data accuracy.

Step 5 – Collate, summarize, and report results

All 54 articles were charted by the year of publication, study objective, theoretical concepts or frameworks, study design, measures of exposure and outcome, measurement scale or indexes, data analysis plan, key findings and methodological challenges reported in the study. For the geographical location the study setting was considered. For review article the country of the first author was considered. Descriptive statistics using frequency counts and percentages was used to depict an overview of the characteristics of the studies. For the measures of exposure and outcome, cluster plotting with the help of Microsoft Excel was done to illustrate the dominant health implications of structural racism. Key findings related to the specific research questions were summarized narratively.

Results

Study characteristics

The search strategy yielded 1,793 articles. After the removal of the duplicates, 1,542 articles remained for the title and abstract screening. Reviewers screened the titles and abstracts of 1,542 articles and dropped 1,480 articles. 62 articles were taken for full text review. Individual hand search of all studies which led to an additional 4 articles was also done. During the full text review 12 articles were excluded. This left 54 articles for final review.

Figure 1 shows the flow diagram of the process of inclusion and exclusion of the studies for this review. Articles included in this review reported or discussed a wide range of public health exposures and outcomes of structural racism, which was captured employing different study designs, theoretical frameworks, scales of measurements. The review found that majority of the included articles were primary research (n = 43/54 articles; 79.6%), conducted in the United States (n = 46/54 articles; 85.2.6%), and predominantly used quantitative research method (n = 32/54 articles; 59.3%, Fig. 2).

Fig. 1
figure 1

Flow diagram of the article screening process

Full size image
Fig. 2
figure 2

Countries of publication of included studies (n = 54) from 2000–2022

Full size image

The majority of the studies used theoretical frameworks which conceived the SODH approach (n = 36/54 articles; 66.7%; Table 3). The review also found that the number of studies has increased substantially since 2015 (Fig. 3).

Table 3 Characteristics of included studies (n = 54)
Full size table
Fig. 3
figure 3

Year of publication of included studies (n = 54) from 2000–2022

Full size image

In terms of study designs the most preferred study design in studying structural racism was cross-sectional (n = 14/54 articles; 31.5%). Black women by ethnicity and gender (n = 7/54 articles; 13.0%) and people of color adults in general (n = 7/54 articles; 13.0%) were the most studied study population. It was also found that majority of the studies (n = 39/54 articles; 72.2%) reported at least 1–4 measures of exposure of structural racism and almost all the studies (n = 49/54 articles; 90.7%) reported at least 1 outcome measure of structural racism as social determinant of health. Similarly, majority of the studies (n = 37/54 articles; 68.5%) used measurement index in capturing the association between structural racism and health.

What study designs have been used in studying structural racism as social determinants of health?

The review found 19 types of study design that have been used in studying structural racism as social determinant of health in the developed country context. It is clearly noticeable that the cross-sectional study design was the most commonly used design (n = 17/54 articles; 31.5%) [22, 26,27,28,29,30,31,32, 39, 47,48,49,50,51,52,53,54] followed by systematic review (n = 7/54 articles; 13.0%) [23, 40, 41, 43, 44, 55] and narrative review (n = 6/54 articles; 11.1%) [18, 45, 46, 56,57,58] (Fig. 3). The cross-sectional study design was used to examine a wide range of exposure clusters of structural racism including access to healthcare [22, 26, 53], civil and legal system discrimination [29, 53], educational attainment [26,27,28,29, 31, 32, 51, 53], employment and income [30, 39, 41, 43, 47, 48, 50, 53, 54], health coverage [51], health status [26, 49], housing and residential segregation [22, 26, 30, 31, 51, 53], incarceration [22, 29], structural violence [22], sociodemographic characteristics [26, 28, 39, 47, 50, 52], socioeconomic status [26, 29, 31], institutional and personal discrimination [54], racial discrimination [48, 50] and immigration related discrimination [28]. Similarly, the clusters of health outcome measures that were examined by cross-sectional design included infant health outcomes [26, 29, 30, 50], chronic conditions [27, 47, 48, 53], access and quality of healthcare [49, 52], quality of life [30, 47, 48], communicable diseases [22, 32], and mental health [28, 31].

The other study designs that were found in the studies include case (n = 1/54 articles; 1.9%) [59], cohort (n = 2/54 articles; 3.7%) [60, 61], ecological (n = 2/54 articles; 3.7%) [62, 63], longitudinal cohort (n = 2/54 articles; 3.7%) [64, 65], longitudinal randomized controlled trial (n = 1/54 articles; 1.9%) [66], non-experimental survey (n = 3/54 articles; 5.7%) [67,68,69], prospective study (n = 1/54 articles; 1.9%) [70], quasi-experimental (n = 1/54 articles; 1.9%) [11], retrospective cohort (n = 1/54 articles; 1.9%) [71], sequential quantitative and qualitative (n = 1/54 articles; 1.9%) [72], exploratory qualitative (n = 3/54 articles; 5.7%) [73,74,75], grounded theory (n = 1/54 articles; 1.9%) [76], integrative review (n = 1/54 articles; 1.9%) [77], meta-analysis (n = 2/54 articles; 3.7%) [78, 79], and policy surveillance (n = 1/54 articles; 1.9%) [13] (Fig. 4).

Fig. 4
figure 4

Distribution of study designs of included studies (n = 54) from 2000–2022

Full size image

What measures of exposure and outcome have been reported in structural racism studies?

It was found that a total of 87 measures of exposure have been reported in all 54 studies. The following paragraph addresses racial bias and/or discrimination as they relate to a wide number of variables. Please see the additional data file for a full list of exposure and outcome (Fig. 5).

Fig. 5
figure 5

Cluster plotting of measures of exposure and measures of outcome in the included study (n = 54) from 2000–2022

Full size image

Exposure

This list of measures of exposure was  grouped in 20 clusters: access to healthcare (n = 7), civil and legal system discrimination (n = 5), cultural, language and values (n = 4), educational attainment (n = 13), employment and Income (n = 12), everyday discrimination (4), geographical segregation (n = 2), health and wellbeing (n = 2), health coverage (n = 2), health status (n = 3), healthcare discrimination (n = 6), housing and residential segregation (n = 13), immigration related discrimination (n = 3), incarceration (n = 5), Institutional and personal discrimination (n = 1), political context (n = 2), racial discrimination (n = 6), religious discrimination (n = 1), sociodemographic characteristics (n = 13), socioeconomic status (n = 8) and structural violence (n = 6).

Access to healthcare cluster includes specific measure of exposure such as asthma rate [62], access to healthcare during pregnancy [26], maternity care system [77], health facility based segregation [46], health care [53], accessibility barriers [72], constraints on access to sexually transmitted diseases (STD) services [22].

The list of reported exposures related to civil and legal system discrimination include legal regulation [54), voting rights [13], stand your-ground laws [13], racial profiling laws [13], mandatory minimum prison sentencing laws [13], immigrant protections [13], fair-housing laws [13], minimum-wage laws [13], predatory lending laws [13], laws concerning punishment in schools [13], stop-and identify laws [13], criminal justice [23, 53], home mortgage discrimination [23], juvenile custody rate [29], sentencing rates [29] and capital punishment [29].

The educational attainment cluster remains to be one of the largest clusters reported by the highest number of studies (n = 13). Reported as one of the most important social determinants of health, educational attainment cluster includes individual education level and attainment [26, 28, 29, 51, 53, 63, 68, 70], high school education, math and English test score [62], parental education [80], education inequality [31] and school stability rate [62].

Similarly, employment-income cluster was reported as another important  determinant of health for the people of color in general (n = 12) through the exposure of having/not having employment [18, 26, 27, 29, 51, 53], employment inequity [31], kind/status of employment [68] household income [27, 29, 68], individual income [26, 50, 63], income inequity [31], income to needs ratio [3], poverty [18, 47, 68].

Housing and residential segregation cluster was also reported by highest number of studies through number of exposures of structural racism including government help for rent [26], household characteristics and conditions [51, 53, 59], length of residence in the neighborhood [26], neighborhood safety [26], residence in public housing [26], residential housing pattern [23], racial housing segregation [22, 30, 31, 57, 61, 69, 79], residential vacancy rate [62].

Sociodemographic characteristics including age [39, 56, 68, 70], gender [28, 45, 67, 70], marital status [26, 28], race/ethnicity [39, 45, 50, 52, 65, 68, 70, 76], race related stress [67] have also been widely reported cluster of exposure in all 54 studies. The other widely reported exposure cluster was socioeconomic status which included car ownership (65), class consciousness (77), homeownership [26, 31], occupational status [29, 63], and wealth [18, 23, 67].

Outcome

In terms of outcome 58 measures of health outcomes in which exposures to racism had measurable impacts on health were reported in the included studies. These 58 measures were categorized into 10 clusters of outcome measures: access and quality of healthcare (n = 7) [40, 49, 52, 55, 72, 76], chronic absenteeism [62], chronic conditions (n = 9) [27, 47, 48, 53, 56, 61, 64, 69, 71], communicable diseases (n = 4) [19, 55, 63, 79], general health outcomes (n = 5) [15, 42, 58, 59, 61], infant health outcome (n = 10) [18, 26, 29, 30, 43, 50, 60, 63, 79, 81], mental health (n = 9) [11, 18, 28, 31, 43, 58, 65, 66, 78], maternal health (n = 1) [77], socioemotional health (n = 1) (74), quality of life (n = 10) [1, 39, 41, 44, 51, 54, 67, 68, 70, 73].

Among all the clusters infant health outcome and quality of life were mostly affected by structural racism. The infant health outcome included preterm birth [43, 50, 60], low birth weight [26], preterm birth and low birth weight [50, 79], neonatal outcome and home environments [81], infant mortality [18, 30, 43], and cortisol reactivity [43]. The quality of life was reported through the measures of health outcomes including dementia [39], disability pattern [51], everyday experiences of discrimination [70], green space in the neighborhood [41], health and wellbeing [54, 73], increased self-awareness [44], mental health [39, 40], wellness score [67], years of life loss [68].

Chronic conditions such as Framingham risk score (FRS) for cardiovascular disease [71], acute respiratory syndrome [27], allostatic load [47], body mass index [53], DNAm (methylation) [64], late-stage diagnosis of cancer [69], number of chronic conditions [61], psychological (e.g., anger, fear) stress responses [56] and waist circumference [48] were reported to be associated with structural racism.

What measurement scales or indexes have been used to explain the health impacts of structural racism?

The review found that majority of the studies used measurement scales or indexes of structural racism (n = 37/54 articles, 68.5%) while 17 (31.5%) studies used no scale of measurement. The total number of indexes that were reported in 37 studies was 73. Concentration of Extremes [29, 51, 63] Dissimilarity Index [23, 31, 41], Everyday Discrimination Scale (EDS) [1, 18, 43, 52], Experience of Discrimination Scale (EDS) [1, 18, 43, 50, 70], Five Segregation Scale [46, 79], Index of Race Related Stress (IRRS) [23, 56, 67, 78], Isolation Index [23, 30, 69] and Perceived Racism Scale (PRS) [1, 78] were the most used scales of measurement in the studies (Table 4).

Table 4 Full list of Indexes reported in the studies
Full size table

What methodological challenges in studying structural racism as social determinants of health have been reported in the studies?

It was found that 44 (n = 44/54 articles; 81.4%), studies discussed methodological challenges related to studying structural racism as social determinant of health. The most widely reported methodological challenges were found to relate to study design (n = 9/54 articles; 16.7%) [11, 26, 27, 43, 48, 51, 53, 55, 62], scales of measurement (n = 9/54 articles;16.7%) [18, 26, 39, 52, 53, 56, 63, 70, 81], measures of exposure (n = 8/54 articles; 14.81%) [40, 46, 49, 53, 56, 61, 67, 79], and data analysis approach (n = 5/54 articles; 9.2%) [1, 30, 31, 50, 78].

The other methodological challenges were related to sample size and sampling method (n = 3/54 articles; 5.5%) [54, 64, 66, 68], study population (n = 4/54 articles; 7.4%) [40, 65, 74, 76], study duration (n = 1/54 articles; 1.9%) [13], study approach (n = 3/54 articles; 3.5%) [45, 72, 73], lack of availability of data on structural level (n = 2/54 articles; 3.7%) [57, 68], weighted SODH score (n = 1/54 articles; 1.9%) [71], use of secondary data (n = 1/54 articles; 1.9%) [28], bias, confounding and misclassification (n = 2/54 articles; 3.7%)[29, 32].

Although cross-sectional was commonly used study design, it was associated with several methodological challenges and limitations in terms of temporal ordering of variables, biases towards type II errors for physical outcomes. Therefore, it has been argued that causality cannot be understood from cross-sectional studies [48]. Hall et al. found cross-sectional study design with limited ability to explain predictive relationships for chronic conditions between a risk factor (e.g., exposure to a biased health care provider) and an outcome (e.g., a patient’s psychological distress) related to structural racism [55]. Dougherty et al. has documented similar observations on the use of cross-sectional study design [53]. Clay et al., on the other hand, found cross-sectional design particularly suitable for ‘fragile or risk population’ such as non-Hispanic White and Black unmarried women with lower educational attainment where women were found to have  low-birth weight infants. Retrospective design was also associated with limitation [27]. However, ecological design was considered suitable and effective to examine association and correlation for macro level factors [22, 62].

Several methodological challenges with different scales of measurement were reported in several studies. The most widely documented challenge was the reliability issue of the use of self-reporting data with Experiences of Discrimination (EOD) and EDS scale of measurement [18]. The other methodological issues related to the different scales of measurement include lack of sensitivity to non-uniform difference score and inflated Cronbach's Alpha for internal consistency reliability in studying internalized discrimination [70], the use of secondary data based on pre-designed questionnaire in studying incarceration [81], lack of construct validity in measuring the latent variable in the confirmatory factor modeling [53], narrow focus of the scales of measurement on the individual along with lack of validated measures of institutional, cultural and structural racism [56]. The lack of longitudinal studies to examine the multiple pathways and dimensional aspects of structural racism and its health outcomes has also been reported [56].

In terms of measures of exposure, residential segregation has been associated with difficulties and potential measurement error [79] as it has not been easy to identify the right kind of measures in examining segregation and its health outcomes. White et al. found it unclear whether to select direct measures versus proxy measures in understanding segregation and health outcomes [46]. Direct or explicit measures with one item have been associated with social desirability bias [32]. The inclusion of one category of study participant such as aboriginal women in New Zealand [81], physicians [44], non-government organization staffs [78] and one ethnic category [54] in the United States context have been associated with non-conclusive and non-generalizable findings.

Other reported methodological issues include lack of structural level data [61], bias, confounding and misclassification [39] due to unavailability of study participants’ in-depth information such as individual income information, low standardized entropy for our latent class model and lack of control potential confounders in analysis [41], unvalidated and weighted SODH score for cardiovascular events [75], and the inability of correlational data to explain causal relationship between exposure and outcome [82].

See full description of the included studies in Table 5 (additional file).

Table 5 Full description of the included studies
Full size table

Discussion

Our review synthesized the growing and recent body of literature on structural racism and highlighted the current methods of structural racism research. Our review particularly offers a comprehensive synthesis of methodological current practices and issues in terms of study design, inventory of measures of exposure of structural racism and health outcomes, inventory of comprehensive list of scales/indexes of measurement and current methodological challenges. This review highlights several important findings:

A striking finding was that the current structural racism research is heavily based on quantitative research approach followed by mixed method and qualitative research. Several studies used the community based participatory research approach which was found to be effective. Similar suggestions on using mixed method research approach have also been given in recent other studies [20, 80]. It has been argued that the use of mixed method (supplementing quantitative data with qualitative data in sequential design or triangulation approach) can facilitate a greater understanding of the social determinant of health by describing and analyzing multidimensional and multiple impact pathways of health outcomes of structural racism [20]. This calls for the need of more research on the use of qualitative approach in studying structural racism.

Secondly, this review shows that the majority of the current racial research employs cross-sectional study design suggesting the dearth of longitudinal studies that consider multiple impact pathways and dimensions. It also indicates the paucity of longitudinal studies in the current research trend and practices by racism scholars. Furthermore, this review also highlights the limitations associated with the cross-sectional design. Several longitudinal studies found that cross-sectional designs were associated with type II error and biases in relation to physical outcomes of structural discrimination [91]. In another study in the United States cross-sectional design was found with the limitation to make conclusion about the temporality in the association between the exposure of discrimination and anxiety scores [97]. Similar limitations on the examination of temporality and causal assumptions have also been reported in study with indigenous South Australians [39]. Our review highlights the several gaps regarding the appropriate study designs: 1) there is no agreed upon best practices in selecting study design while studying structural racism in a given country context, 2) there is critical need to conduct more research with particular focus on the suitability of different study design and develop a gold standard for structural racism research, and finally 3) there is need for methodological innovations for better understanding which can inform the design of future programs, policy or practices regarding structural racism.

Some other key methodological challenge are the unavailability of data on structural levels [10] and variation of estimates by the geographic unit of analysis which has been documented in several other studies [9, 40, 41]. When residential segregation is considered as exposure, the association between segregation and health outcome tend to vary as the most reliable estimates are found for smaller unit of analysis. This analysis is not consistent for all health outcomes. It often becomes impossible to differentiate the potential mediating and moderating effects in the association. The current research practice adopted by some scholars to overcome this challenge is to control variables related to socio-economic condition which are the pathways of how racism affects health outcomes [40, 41]. This sheds light on future research needs to identify the proximal mechanisms, interaction pattern between the exposure and health outcomes by using longitudinal data and advanced statistical method to develop concrete understanding around temporality and causality.

Thirdly, our review, to our best knowledge is one of the first reviews to systematically synthesize all available and reported measures of exposure and health outcomes in relation with structural racism as social determinants of health. Our review has documented a total of 87 measures of exposure and 56 measures of health outcomes. It also highlights the most common clusters of exposures which include educational attainment, employment and income, sociodemographic characteristics. The most common clusters of health outcomes found in our review were infant health outcome, chronic condition, mental health, and quality of life. Similar observations have consistently been documented in several studies [8, 95]. Such as a wide range of clusters of exposures and outcomes pinpoints the magnitude of public health implications in the United States and other high-income countries. This also suggests that there is critical need to develop a comprehensive and integrated framework for measures of exposure and outcomes related to the study of structural racism.

Fourthly, our review documents all available scales or indexes of measurement related to structural racism available in the literature. It lists 73 scales of measurement and discusses methodological challenges related to widely used scales of measurement such as individual focus, self-reporting and personally mediated, non-linear nature which are associated with bias and confounding. This have been documented in other studies [85].

The other limitations with the current indexes of measurement are more focus on liner domain-based measurements, failure to study multidimensional and multilevel impacts of structural racism, and applicability to limited number of ethnic groups. These findings are consistent with recently published literature [13]. Our review demonstrates that there is no scientific consensus on the use of index of measurements that help us further understand and explain the dynamics and pathways of multilevel interactions of mutually reinforced systems and institutions.

Finally, our review clearly highlights the gaps in the current research on structural racism. Some of the widely documented gaps are lack of systematic, longitudinal studies that examine multiple pathways and ecological factors by which racism can affect health over the life course [60], the use of single dimensions of structural racism (e.g., housing, education, employment, incarceration, etc.) [20], exclusion of the appropriate respondents in the appropriate settings (e.g. exclusion of prisoners in the most of the studies) [33], methodological challenges involving individuals, levels and spatiotemporal scale [33]. These findings are consistent with the findings of other studies [91, 95]. Last not but the least, our review shows the lack of best practice regarding the selection of measures of exposure and outcome, study design and index of measurement and therefore, calls for more research initiatives and develop a standard guideline for the researchers interested in structural racism. Such a call has also been made in other study [93]. The findings from our review can guide researchers, academicians, and other relevant stakeholders in designing future research and programs on structural racism.

Limitations

This scoping review anticipates several limitations and biases. One bias could be the selection bias as we included studies which were published in English. Secondly, although comprehensive searches across databases were done, it is possible to miss relevant sources which could have been eligible for this review. Thirdly, the review did not conduct quality assessment of the included studies. Finally, we found that more studies on structural racism were done in the United States than other countries. We therefore recommend the exercise of caution in the use of findings in other context. Despite these limitations our review has documented the current research trend, practices, challenges, and future research needs on structural racism.

Conclusion

Our scoping review found that despite repeated calls from racism scholars for more comprehensive approaches, traditional research methods are being followed by most of the scholars. We found that there is a severe lack of longitudinal studies and availability of structural or ecological data. There is growing understanding among the racism scholars that it is imperative to understand the ways in which the social or ecological context including all the structure, institutions, laws, policies, and practices affect the health of a racial group. It has also been recognized that a detailed and comprehensive characterization of the exposures in their social context is required [98, 99]. Our review sheds lights on several key gaps and research priorities on structural racism. First of all, there is a need to develop agreed upon measures with concrete indicators that can comprehensively assess multidimensional and multilevel health outcomes of structural racism. Secondly, the current methodological gaps such as lack of consensus or framework on appropriate study design that can capture the complex interactions of systems and interconnected institutions need to be addressed through further research on structural racism. Thirdly, there is also a need to develop a framework on choosing the index of measurement which needs to be selected prior to the design of research. We, therefore, recommend the development and use of new structural racism measures which could be a good fit at different levels and geographical location for consistent and reliable estimates. We also recommend the use index measures based on a set of concrete indicators which capture complex interactions of exposure and outcomes and undertaking of longitudinal studies using a life-course approach to measurement.

Finally, we acknowledge that there is also growing recognition among the racism scholars that studying structural racism requires more scientific, rigorous and context specific study methods of structural racism or discrimination.