Structural Stigma, Racism, and Sexism Studies on Substance Use and Mental Health: A Review of Measures and Designs

Key Takeaways

• Studies on substance use and mental health issues related to structural stigma, racism, and sexism recently have increased sharply in number.
• The structural determinant measures used in the studies were diverse and often innovative.
• Most of the studies used passive-observational (correlational) designs, which limited the ability to assess whether the structural determinants examined were causally related to substance use and mental health. 
• The studies typically did not discuss the practical significance of the statistically significant associations found between structural determinants and substance use and mental health.
• Collectively, the studies showed that a wide range of structural determinants correlate with the substance use and mental health of many groups within and across nations.

Introduction

Factors such as government policies, institutional practices, economic systems, social norms, and the physical environment can impact health. Called structural determinants of health in medical parlance,¹ several of these factors known to be associated with substance use and mental health have long been studied. Examples include alcohol taxation (associated with alcohol use and mortality),^2-4 alcohol outlet density (associated with alcohol-related health),⁵ tobacco taxes (associated with tobacco use),⁶ tobacco control policies (associated with smokeless tobacco use),⁷ essential/precursor chemical regulations (associated with illicit drug use and morbidity),^8‑11 neighborhood/city characteristics (associated with mental health),^12,13 and immigration policies (associated with mental health).^14,15

However, other structural determinants of substance use and mental health exist, many of which are being identified in the emerging fields of structural stigma, structural racism, and structural sexism. Although the need to consider measures from these fields has been discussed,^16-21 a survey of the measures and related research elements used in substance use and mental health studies across the three fields has not been conducted to date. This narrative review helps address this gap. 

Definitions of Structural Stigma, Structural Racism, and Structural Sexism

As background, it will be helpful first to outline what structural stigma, structural racism, and structural sexism are. Hatzenbuehler and Link defined structural stigma as the “societal-level conditions, cultural norms, and institutional policies that constrain the opportunities, resources, and well-being of the stigmatized.”^22(p2) Examples of stigmatized groups are persons living with human immunodeficiency virus (HIV), persons living with mental illness, persons who misuse substances, sexual and gender minorities, and marginalized racial and ethnic groups.¹⁹

According to Bailey et al., structural racism involves interconnected institutions, whose linkages are historically rooted and culturally reinforced: “It refers to the totality of ways in which societies foster racial discrimination, through mutually reinforcing inequitable systems (in housing, education, employment, earnings, benefits, credit, media, health care, criminal justice, and so on) that in turn reinforce discriminatory beliefs, values, and distribution of resources, which together affect the risk of adverse health outcomes.”^23(p1454) Academics often use the terms “structural racism” and “institutional racism” interchangeably.^17,24 Institutional racism, as defined by Willams et al., “refers to the processes of racism that are embedded in laws (local, state, and federal), policies, and practices of society and its institutions that provide advantages to racial groups deemed as superior, while differentially oppressing, disadvantaging, or otherwise neglecting racial groups viewed as inferior.”^24(p107)

Javidan defined structural sexism as “discriminatory beliefs or practices on the basis of sex and gender that are entrenched in societal frameworks and which result in fairly predictable disparities in social outcomes related to power, resources, and opportunities. Sexism is structural when it is integral to how a society is organized, which can be observed in its institutions, systems, and/or patterns of social relations. Structural sexism also functions to normalize and legitimize such beliefs, practices, and inequalities of conditions and outcomes.”^25(p1)

Approach of the Review

This narrative review surveys the objective measures of structural determinants (e.g., government policies, institutional actions, community opinions/norms), the substance use and mental health outcome measures, and the research designs used in studies of structural stigma, structural racism, and structural sexism. Objective measures of structural determinants are considered here for multiple reasons. Compared with self-reports of structural determinants from individuals (e.g., self-reports of discrimination at a workplace), objective structural measures avoid issues such as respondents’ bias or lack of awareness of structural effects.¹⁶ Objective structural measures also avoid same-source bias, which can introduce spurious results when the independent and dependent variables are measured with the same method.^19,26 Finally, as noted by Groos et al., objective quantitative measurement is important to the identification of reliable and causal associations between structural determinants and health.¹⁸

Methods

Search Strategy

The PubMed, PsycINFO, and Scopus databases were searched for articles on May 11, 2023, with the help of a health sciences librarian and using a focused search approach. Searches included thesaurus terms, such as Medical Subject Headings (MeSH), when available, as well as text words to search in the title or abstract (tiab), or author-supplied keyword fields of the article records in the databases. The terms “structural,” “institutional,” “institutionalized,” or “systemic” were paired with the terms “stigma,” “racism,” “barriers,” and “sexism”; all of these were combined with terms for substance use or mental health.

The following search strategy for PubMed demonstrates the search terms used and the actual search strategy that was employed and then translated to the other databases: ((“Systemic Racism” [MeSH] OR “systemic racism” [tiab: ~3] OR “systemic stigma” [tiab: ~3] OR “systemic barriers” [tiab: ~3] OR “systemic sexism” [tiab: ~3] OR “structural racism” [tiab: ~3] OR “structural stigma” [tiab: ~3] OR “structural barriers” [tiab: ~3] OR “structural sexism” [tiab: ~3] OR “stigma system” [tiab: ~3] OR “racism system” [tiab: ~3] OR “institutional racism” [tiab: ~3] OR “institutional stigma” [tiab: ~3] OR “institutional barriers” [tiab: ~3] OR “institutional sexism” [tiab: ~3] OR “institutionalized racism” [tiab: ~3] OR “institutionalized stigma” [tiab: ~3] OR “institutionalized barriers” [tiab: ~3] OR “institutionalized sexism” [tiab: ~3]) AND (“Substance-Related Disorders” [MeSH] OR “Drug Misuse” [MeSH] OR Smoking [MeSH] OR “Substance Abuse Treatment Centers” [MeSH] OR “Mental Disorders” [MeSH] OR “Mental Health” [MeSH] OR ““Mental Health Services” [MeSH] OR “Community Mental Health Centers” [MeSH] OR “substance use” [tiab] OR “substance abuse” [tiab] OR “substance dependence” [tiab] OR “substance misuse” [tiab] OR “drug use” [tiab] OR “drug abuse” [tiab] OR “drug misuse” [tiab] OR “use disorder*” [tiab] OR “addiction” [tiab] OR “mental health” [tiab] OR “mental disorder*” [tiab] OR “mental illness” [tiab])).

The MeSH terms listed above often had descendent MeSH terms that were automatically searched. For example, the MeSH term “Mental Disorders” had descendent MeSH terms such as “Anxiety Disorders,” “Mood Disorders,” and “Schizophrenia Spectrum and Other Psychotic Disorders”; many of these had their own descendent MeSH terms, all of which were automatically searched.

Eligibility Criteria

Article inclusion criteria were as follows: (1) original research, (2) peer-reviewed, (3) written in English, and (4) one or more objective structural determinant measures used to help explain behavioral health outcomes (e.g., self-reports from individuals regarding alcohol use or depression). Other types of studies on structural determinants (e.g., commentaries, review papers, qualitative analyses, studies with no behavioral health outcome measures) were excluded.

Literature Search Results

The PubMed, PsycINFO, and Scopus searches identified 2,020, 1,639, and 827 records, respectively. Duplicates (n = 1,950) were eliminated, leaving a total of 2,536 records to review. The titles and abstracts were reviewed for relevance to the topic, and the full text of the resulting selected (potentially relevant) studies were reviewed for further analysis. This resulted in the exclusion of 2,487 studies. Forty-nine studies (30 related to structural stigma,^27-56 16 related to structural racism,^57-72 and three related to structural sexism^73‑75) met the inclusion criteria and are reviewed here. 

Results

The number of structural stigma, racism, and sexism studies examining substance use and mental health has recently accelerated. The first identified study was published in 2004.³⁰ Eight were published during 2004 through 2017, eight were published during 2018 through 2020, and 33 were published during 2021 through May 11, 2023; this last period included 11 studies on substance use and 22 studies on mental health.

Thirty-six of the 49 studies identified were conducted in the United States; this included 18 of 30 structural stigma studies, 15 of 16 structural racism studies, and all three of the structural sexism studies. Six of the 12 structural stigma studies conducted outside the United States were multinational in that they examined samples drawn from a range of four countries to more than 100 countries;^{30,41,43,45,51,54} the remaining structural stigma studies were from Australia,⁴⁷ Brazil,⁴⁶ Italy,⁵⁰ Sweden,^36,55 and the United Kingdom.³⁴ The one structural racism study conducted outside of the United States was from England.⁵⁷

Outcome Measures

Fifteen studies examined the association between structural determinants and substance use-related outcome measures. Approximately twice that number (31 studies) examined the association between structural determinants and mental health-related outcome measures. Another three studies examined how structural determinant measures were associated with both substance use- and mental health-related outcomes. All studies are summarized in Appendix 1. 

Measures related to alcohol use constituted the substance use outcome category most often examined (Table 1). However, multiple studies also examined tobacco use, other drug use, and/or substance use-related services. When different studies considered the same type of substance use outcome (e.g., binge drinking), there was often variation in the specific measures used. For example, Greene et al.³⁵ defined binge drinking as four or more drinks for women and five or more drinks for men on a single occasion in the past 30 days; Drabble et al. defined it as eight or more drinks in a single day during the past year;³¹ Everett et al. defined it as six or more drinks in a day in the past 12 months;³² and McKetta et al. (in two studies) defined it as five or more drinks in a row during the past 2 weeks.^74,75 Among the four studies that had smoking outcome measures, one considered cigarette use in the past year,³⁷ one examined current use of cigarettes,⁵³ one examined 7-day abstinence from cigarettes at week 26 of a clinical trial,⁶⁸ and one examined current smoking at the census tract level without specifying whether it was limited to cigarettes.⁶⁶

Table 1. Substance Use Outcome Categories: Number of Structural Stigma, Structural Racism, and Structural Sexism Studies

Outcome	Total Studies	Structural Stigma Studies	Structural Racism Studies	Structural Sexism Studies
Alcohol*	10	6 (31,32,35,40,44,51)	2 (66,71)	2 (74,75)
Tobacco Use	5	3 (37,44,53)	2 (66,68)	0
Marijuana Use	2	2 (31,38)	0	0
Illicit Drug Use	3	2 (38,51)	1 (71)	0
Substance Use-Related Services	3	3 (33,40,52)	0	0
Drug-Related Deaths	1	0	1 (70)	0

Note: Reference numbers are in parentheses. Studies that examined multiple outcome categories are counted in each of those categories.
*Any alcohol-related outcome (does not include substance use-related services)

The mental health outcome most often examined was depression (Table 2). Multiple studies also considered anxiety, eating disorders, general mental health, and mental health-related services. Several studies used the same mental health outcome measures. For example, six studies assessed depression using versions of the Center for Epidemiologic Studies-Depression (CES-D) scale,^{32,39,41,48,50,72} and five assessed anxiety using the General Anxiety Disorder-7 (GAD-7) scale.^{34,39,45,56,67}

Table 2. Mental Health Outcome Categories: Number of Structural Stigma, Structural Racism, and Structural Sexism Studies

Outcome	Total Studies	Structural Stigma Studies	Structural Racism Studies	Structural Sexism Studies
Depression	13	7 (32,39,41,43,48,50,54)	6 (58,61,65,67,69,72)	0
General Mental Health	8	5 (28,29,42,47,55)	3 (60,64,66)	0
Anxiety	7	5 (34,39,45,50,56)	2 (61,67)	0
Stress/Worry	4	2 (35,56)	2 (59,63)	0
Eating Disorders	2	0	1 (62)	1 (73)
Mental Health-Related Services	6	4 (27,30,46,49)	2 (57,58)	0

Note: Reference numbers are in parentheses. Studies that examined multiple outcome categories are counted in each of those categories.

Nearly all of the substance use-related studies (17 of 18) were conducted in the United States (see Appendix 1). In contrast, 12 of the 34 studies examining mental health-related outcome measures were conducted outside of the United States. (Note that the three studies^27,32,66 that examined both substance use- and mental health-related outcomes were included in the above count of 18 substance use studies and again in the count of 34 mental health studies.)

Structural Determinant Measures

Government legislation/policy was the subject area most often addressed by structural determinant measures (it was used in 23 studies); examples include legislative protections for sexual and gender minorities and for reproductive rights, legislative bills with stigmatizing language, and the gender and political parties of legislators (see Appendix 1). However, government legislation/policy was not used as a structural determinant in any of the structural racism studies. The other three top structural determinants measured were community opinions/norms (e.g., neighborhood racism, community opinions regarding sexual minorities), which were used in 15 studies; area characteristics (e.g., areas with low socioeconomic status, housing remediation, segregation), which were used in 13 studies; and institutional policy/action (e.g., financial loan discrimination, college policies to promote inclusion of sexual and gender minorities), which were used in eight studies. Various other determinants (e.g., regional development, law enforcement, historical trauma) were examined as well.

No two different lead authors used the exact same measures for structural determinants. That said, several studies constructed structural determinant measures using data from the same sources. For example, nine studies used U.S. Census data;^{41,59,62,64-68,70} three used data from the Movement Advancement Project;^27,31,42 and three used data (grades) from the Home Owners’ Loan Corporation.^64,66,67

The studies’ structural determinant measures can be classified into four types: single indicators, multiple single indicators, indexes, and composite indexes. An example of a single indicator would be a community opinion/norm or the enactment of new legislation. An example of multiple single indicators would be the use of a community opinion/norm and new legislation as individual structural determinant measures in the same study. Indexes consist of multiple related indicators (e.g., two different financial lending measures) combined into one measure. A composite index is a single measure constructed by combining an indicator/index for one topic with an indicator/index for at least one other topic. An example would be the combining of an index for a state’s policies with an indicator of poverty in that same state. 

The most frequently used type of structural determinant measure was the index (used in 17 studies). This was followed by composite indexes (13 studies), single indicators (13 studies), and multiple single indicators (six studies). All four types were used in structural stigma studies (seven used single indicators,^{32,34,39,40,46,47,50} four used multiple indicators,^29,33,41,56 nine used indexes,^{27,28,31,35,42,48,49,51,52} and 10 used composite indexes^{30,36-38,43-45,53-55}). None of the structural racism studies used composite indexes; eight used indexes,^{59,61,62,64,65,67,68,70} six used single indicators,^{57,58,63,66,71,72} and two used multiple single indicators.^60,69 All three of the structural sexism studies used composite indexes.^73-75

Compared to composite indexes, the use of indexes might be on the rise. Sixteen of the 17 studies using indexes were published in 2021 or later, whereas only eight of the 13 studies using composite indexes were published in 2021 or later.

Types of Research Designs Used

None of the 49 studies reviewed used experimental designs. Seven used quasi-experimental designs/methods: Specifically, three used pretest-posttest comparisons (two with pretest groups that differed from their posttest groups and none with a comparison group),^32,39,50 two used propensity score matching,^42,73 one used a difference-in-differences design,⁷¹ and one used an interrupted time series design.⁴⁰ These latter two designs were among the strongest regarding causal inference, but the studies using them did not find statistically significant associations between their structural determinant measures and outcome measures. Forty-two of the 49 studies used passive-observational designs (i.e., correlational designs including, for example, regressions with control variables, path analyses, and structural equation models). Five of these examined ecological correlations (i.e., correlations in which the units studied were populations or groups rather than individuals).^{33,52,58,64,66} Whether studies examined structural stigma, structural racism, or structural sexism, passive-observational designs predominated.

Types of Associations

Most of the studies (39 of 49) focused exclusively on direct associations between their structural determinant and outcome measures. Five studies (one on substance use⁷⁴ and four on mental health^36,43,62,65) considered both direct and indirect associations. Five studies (all mental health studies) reported indirect or moderator associations but not direct associations between their structural and outcome measures.^34,54-56,61

The majority of the studies reported that statistically significant associations were found between their structural measures and outcome measures. Whether the associations found had practical significance was generally not discussed.

Groups Sampled

Sexual minority groups were sampled more often than any other group (Table 3). Racial and ethnic minority groups, primarily persons who were Black, were sampled second most often. Other groups sampled included, for example, migrants, youth, women, psychiatric patients subject to involuntary hospitalization, and patients who experienced overdoses from substance use. Sometimes the samples consisted of items rather than people. For instance, Conley²⁸ and Conley and Baum²⁹ analyzed samples of mental health-related bills from all 50 states.

The three fields tended to differ regarding groups examined. For example, almost all of the studies that considered sexual and gender minority groups were structural stigma studies. Most of the studies that focused on persons who were Black were structural racism studies. All three of the structural sexism studies used surveys that primarily sampled cisgender persons.

Table 3. Groups Sampled: Number of Structural Stigma, Structural Racism, and Structural Sexism Studies

Groups Sampled*	Total Studies	Structural Stigma Studies	Structural Racism Studies	Structural Sexism Studies
Sexual and Gender Groups
Sexual Minority†	17	16 (31,32,35-38,41-45,47,49,50,53,55)	1 (60)	0
Gender Minority	1	1 (40)	0	0
Sexual/Gender Minorities	4	4 (27,39,54,56)	0	0
Females	2	0	0	2 (74,75)
Racial and Ethnic Groups
Black Individuals	10	0	10 (58,59,61,62,65,67-69,71,72)	0
Latino Individuals	1	0	1 (63)	0
Black and Latino Individuals	1	0	1 (70)	0
Other Groups of Persons
General Population	3	1 (51)	2 (64,66)	0
People who Inject Drugs	1	1 (33)	0	0
Migrants	1	1 (34)	0	0
Patients Experiencing Overdoses	1	1 (52)	0	0
Patients in Mental Health Hospitals	1	0	1 (57)	0
Persons With Behavioral Disorders	1	1 (30)	0	0
Youth	3	2 (46,48)	0	1 (73)
Items/Entities
Mental Health-Related Legislative Bills	2	2 (28,29)	0	0

Note: Reference numbers are in parentheses.
*Groups were categorized according to the major focus of a study. For instance, in the Miedema et al. study of sexual minority men in four Asia-Pacific region countries,⁴¹ the focus was primarily on sexual minority status. The study consequently was listed in the sexual minority category.
^†Includes one study that sampled sexual minority youth of color.

Discussion

This narrative review found a substantial, recent rise in substance use studies and mental health studies within the collective fields of structural stigma, structural racism, and structural sexism. Although increases in both types of studies contributed to the overall rise, those involving mental health exceeded those involving substance use by about two to one.

The outcome measures used in the studies (e.g., measures of alcohol use and depression) were generally the same or similar to those commonly used in substance use and mental health epidemiology analyses. Notwithstanding this, the structural determinant measures used were diverse and often innovative, at least in comparison to those used in the long-standing lines of structural determinant research (e.g., alcohol taxation, essential/precursor chemical control, immigration policy) noted earlier. The structural determinant measures of the reviewed studies addressed, for example, community opinions about social issues, the gender of legislators, financial loan discrimination, economic vulnerability, college policies, law enforcement practices, historical trauma, and legislative protections for reproductive rights and for sexual and gender minorities.

The structural measures were also relatively novel in that more than half were indexes or composite indexes. This differs from the structural determinant measures used in alcohol taxation and other long-standing lines of structural determinant research, which typically have used single indicators. Some of the indexes and composite indexes in the reviewed studies were complex. For example, Beccia et al. used a composite index that assessed political participation, state policies, socioeconomic measures, institutional support, and access to reproductive care.⁷³

In the studies reviewed, the use of indexes for measuring structural determinants increased over time. Authors have debated the use of single indicators versus indexes and composite indexes.^76,77 Single indicators may provide clearer guidance for policy in the sense that each indicator’s contribution to an outcome can be described. On the other hand, an index or composite index provides a single score representing a set of structural determinants, and policymakers and the public may be more comfortable considering that single score as opposed to pondering how each of several single indicators contributes to an outcome. If the goal of a study is to guide policy directly, single indicators may be preferable. If the goal is to increase awareness of structural determinants, then indexes or composite indexes may be more helpful.

When examining multiple structural determinants, future studies could consider performing one analysis with single indicators and another using indexes or composite indexes. The former could help provide direct guidance regarding individual determinants, whereas the latter could provide a simpler—and perhaps easier to remember—representation of the structural determinants’ overall impact. 

Little information on whether structural determinants and behavioral health outcomes were causally related was provided, as most of the reviewed studies used passive-observational (correlational) research designs. Seven used quasi-experimental designs. Of these, the two studies that arguably had the strongest designs—a difference-in-differences design and an interrupted time series design—reported null effects.^40,71 Relatively strong quasi-experimental designs are usually needed to help evaluate whether structural determinants are responsible for changes in behavioral health outcomes.^78-80 Such designs have commonly been used in some of the long-standing lines of structural determinant-behavioral health research; this includes studies on alcohol taxation,^2-4 tobacco taxation,⁶ and essential/precursor chemical regulation.^8-11

Although most of the reviewed studies reported statistically significant results, they generally did not discuss practical significance.⁸¹ These two types of significance can differ as illustrated in the following example. In a trial of a vaccine for the common cold, people who received the vaccine had an average of 1.6 colds per year, whereas those who did not receive the vaccine averaged 2.1 colds. The difference in the number of colds between the two groups had high statistical significance (p < .000001). Nevertheless, the difference was thought to lack practical significance because an average of 0.5 fewer colds per year was considered too small to justify the vaccine’s expense and possible side effects.^82,83 Similarly, when examining structural determinants, it is usually important to consider not only the statistical significance of an effect (association) but also its practical significance. For example, is the effect substantial enough to warrant real-world actions such as policy changes, efforts to shift social norms, or the introduction of new legislation?

The studies reviewed drew samples from a varied array of groups (see Appendix 1 and Table 3). However, within each of the three fields, some groups tended to be sampled with particular frequency. Structural stigma studies, for example, often considered sexual and gender minority persons, structural racism studies often considered persons who were Black, and structural sexism studies analyzed survey samples predominantly composed of cisgender individuals. Numerous groups remain to be examined. For instance, although 18 studies considered outcomes related to substance use, none were specific to persons who used methamphetamine, cocaine, or psychedelics. None of the studies focused on persons who were homeless or who had physical, intellectual, or psychiatric disabilities.⁸⁴ Also, none of the studies analyzed population samples specific to American Indians, Alaska Natives, Hawaiian Natives, or other Indigenous people in colonized countries.

Interest in the study of structural stigma, structural racism, and structural sexism appeared to differ geographically. Structural stigma studies were conducted in a wide range of nations, while all of the structural sexism studies and all but one of the structural racism studies were conducted in the United States.

Limitations

This review was exclusive to structural stigma, structural racism, and structural sexism studies that used objective structural determinant measures to help explain outcomes related to substance use and mental health. Studies from these fields that used qualitative analyses and self-reports to assess structural determinants related to substance use and mental health were not considered and remain to be reviewed. Bias may have occurred due to not including manuscripts from the gray literature. This review focused on measures and designs, not on study findings. Nations can differ regarding policies, norms, economic systems, and so on. Such differences could limit the extent to which associations between structural determinants and behavioral health outcomes can be generalized from one nation to another.

Conclusions

Structural stigma, structural racism, and structural sexism studies using objective structural determinant measures to help explain behavioral health issues have recently accelerated in number. These studies employed a diverse array of structural determinant measures, many of which were innovative. However, most of the studies used passive-observational (correlational) research designs and thus provided little evidence as to whether causal associations existed between the structural determinants and the substance use and mental health outcomes examined. Moving forward, relatively strong quasi-experimental designs could be used to help address this limitation. In the meantime, public health proponents can use the studies reviewed to help foster awareness that a wide range of structural determinant measures correlate with the substance use and mental health of many groups within and across nations.