Racial and ethnic disparities in women’s mental health: a narrative synthesis of systematic reviews and meta-analyses of the US-based samples

Abstract

Over 2 decades of research indicate the significance of racial or ethnic disparities in mental illness in the United States. However, minoritized racial or ethnic groups tend to report overall lower prevalence rates of psychiatric disorders than White adults, although this varies depending on gender and race or ethnicity. We conducted a rigorous and systematic narrative synthesis on the differences in the prevalence rates and symptoms that differ across racial or ethnic women in depression, anxiety, eating disorders, and premenstrual syndrome or premenstrual dysphoric disorder. Seven systematic reviews and meta-analyses that examined racial/ethnic differences in depression and eating disorders were included. No review that examined racial/ethnic differences in anxiety or premenstrual syndrome or premenstrual dysphoric disorder met inclusion criteria. Methodological quality of the reviews, which was determined by the Assessment of Multiple Systematic Reviews criteria, revealed that the results of 5 reviews were rated as critically low confidence, one review was rated as low confidence, and one review was rated as high confidence. Findings were inconsistent across systematic reviews and meta-analyses because of the methodological differences in the original studies. Overall, racially or ethnically minoritized women generally report lower prevalence rates in depressive and eating disorders than the White women; however, they exhibit different or greater symptom presentation that could influence prevalence estimates depending on the diagnostic criteria followed. Methodological considerations are provided to strengthen the literature on racial or ethnic mental health disparities in women.

For over 2 decades, disparities in mental health care for minoritized racial or ethnic groups in the United States are well documented, and the interventions aimed at reducing such disparities have been implemented (1, 2). Recent data indicate that minoritized racial/ethnic groups with certain mental health disorders in the United States were less likely to be diagnosed and receive treatment than White adults (3, 4). We generally expect that the prevalence of psychiatric disorders would be higher in minoritized racial or ethnic groups in the United States, given that such groups are more likely to experience social determinants of mental health disorders, such as racism, discrimination, poverty, and childhood adversity (5–7). However, some minoritized racial or ethnic groups actually report low prevalence rates, at least for certain psychiatric disorders (8–11)—a phenomenon referred to as the mental health paradox or the race paradox in mental health (9, 10, 12). Thus, a rigorous and systematic examination of the differences in prevalence rates of specific mental disorders in minoritized racial or ethnic groups is needed to specify which groups exhibit an unexpected advantage or the expected disadvantage.

Sex and gender are also likely to interact with race or ethnicity to impact prevalence and presentation of psychiatric disorders. Although 21% of US adults were affected by mental illness in 2020, the 12-month prevalence of experiencing any mental illness was higher among females (25.8%) than among males (15.8%) (13). Females also had a higher 12-month prevalence of having major depressive disorder (MDD) (10.5% in females vs. 6.2% in males) and any anxiety disorder (23.4% in females vs. 14.3% in males) (13). Furthermore, females are almost twice as likely to experience binge eating disorder (BED), the most commmjpeon eating disorder, over their lifetime (3.5% in females vs. 2.0% in males) and over the past 12 months (1.6% in females vs. 0.8% in males) (13, 14). Finally, certain psychiatric conditions are female-specific and associated with hormonal fluctuations in the reproductive lifecycle, such as premenstrual syndrome (PMS) and premenstrual dysphoric disorder (PMDD). Both PMDD and PMS are characterized by physical and emotional symptoms leading up to the onset of menses, but PMDD is more severe and debilitating (15). The prevalence rates of PMDD in the US and non–US females range from 7% to 54%, depending on the study methods and population sampled (16, 17). Estimates from Germany based on Diagostic and Statistical Manual of Mental Disorders-IV (DSM-IV) criteria indicate that 5.8% of menstruating females experience PMDD in a 12-month period and an additional 18.6% were near-threshold cases, which would qualify as moderate to severe PMS (18).

Given the potential interactive effects of sex and gender as well as race or ethnicity on the aforementioned psychiatric conditions, there is a need to understand racial or ethnic differences in the prevalence rates and symptom variation in adult women in the United States. Specifically, we aim to fill the knowledge gap on the prevalence estimates of these mental health conditions in racially or ethnically minoritized women in the United States and the variation in symptom presentation that could explain these disparities or differences in the prevalence rates. We conducted a high-level narrative synthesis of peer-reviewed systematic reviews and meta-analyses, with the intention to provide a rigorous and comprehensive understanding of the current evidence based on racial and ethnic disparities in mental health conditions that are female-specific or female predominant—namely, depression, anxiety, eating disorders, and PMS/PMDD. In doing so, we aim to provide women’s health decision makers (i.e., researchers, clinicians, and policymakers) with a critical understanding of the summary of effects, heterogeneity, and study quality to inform the next phase of research aimed at reducing disparities in women’s mental health. To our knowledge, this is the first narrative synthesis to examine racial or ethnic differences in female-specific or female-predominant psychiatric conditions.

MATERIALS AND METHODS

We followed standard procedures for conducting a narrative synthesis (19). To be included, systematic reviews or meta-analyses of original research studies needed to compare ≥ 2 female racial and ethnic groups of adults (aged ≥ 18 years) within the psychiatric conditions of interest, be published in or translated to English, and have original research studies conducted on mostly US-based samples, given the regional differences in determinants of psychiatric disorders. There were no limits to publication dates of systematic reviews and meta-analyses. All included reviews were published before January 17, 2023. Systematic reviews and meta-analyses were excluded if original research studies were not based on mostly US-based samples, focused on populations with co-occurring physical and mental health conditions (e.g., depression and diabetes), did not clearly define racial or ethnic groups (e.g., immigrant compared with nonimmigrant populations), or included other age groups in the sample (e.g., pediatric, adolescent populations) without doing a subgroup analysis of adult women (aged ≥18 years).

Our initial search for relevant reviews was conducted in Ovid Medline ALL, APA PsycInfo, and Web of Science on December 5, 2022; please Supplemental Material (available online) for search strategy). A second search was conducted in Cochrane Library on January 14, 2023 and the Cochrane Database of Systematic Reviews on January 17, 2023, but no additional systematic reviews and meta-analyses were found. We used Endnote X9/20, a bibliographic management software to organize and remove duplicates imported from the searches. Article screening occurred in a stepwise manner in Covidence, a systematic review software. First, 2 investigators (P.P.B. and C.E.P.) independently screened and reviewed titles as well as abstracts using the inclusion and exclusion criteria. Disagreements were resolved to determine whether the reference should be considered in full text screening. Articles were then screened independently to determine the final eligibility of each reference. Disagreements were also resolved with consensus of a third investigator (K.A.D.).

Four independent reviewers (P.P.B., C.E.P., D.A.A., H.C.H.) extracted information from each article using an a priori developed data extraction form: first investigator name; year of publication for the systematic review/meta-analysis and the date range of original research studies; country from where the systematic review and meta-analysis was published and from where samples were collected in the original research studies; mental health condition and symptoms (i.e., depression, anxiety, bulimia nervosa (BN)/BED, or PMS/PMDD); recruitment strategy or sampling method; demographic information (e.g., race, ethnicity, sex, age) of the participants represented in the reviews; percentage or number of each race and ethnicity represented; number of databases and date range searched in the reviews; study types included in the reviews (e.g., longitudinal, cross-sectional); assessment methods for mental health conditions of interest; and description of results testing differences in prevalence rates as well as symptoms or correlates of mental health conditions in 2 or more racial or ethnic group.

Systematic reviews and meta-analyses that met inclusion criteria were assessed for methodological quality using the Assessment of Multiple Systematic Reviews (AMSTAR-2) (20). Furthermore, AMSTAR-2 was expanded to include 16 items (e.g., protocol registration before review; adequacy of literature research; risk of bias assessment and interpretation) to guide researchers to conduct rapid and reproducible assessments of the quality of systematic reviews and meta-analyses in randomized and nonrandomized studies. The AMSTAR-2 tool also included an overall rating based on weaknesses in critical domains. Reviews were scored on each of the 16 items as well as assigned an overall confidence rating in the results of the review (high, moderate, low, or critically low) (20).

Relevant data from the included systematic reviews and meta-analyses were summarized. Factors that may explain differences in the direction or size of effects across studies were considered and discussed, as per guidelines for conducting a narrative synthesis (19). Sex and gender and racial or ethnic identities of participants are described as they were in the original review (e.g., Black women will be applied when participants were described as Black women and African American females, when participants were described as African American females). In addition, the term “minoritized” will be used to indicate that systems place individuals into a “minority” status based on certain sociodemographic factors (e.g., race, ethnicity, age, sex, migration background) (21).

RESULTS

Study Characteristics

A flowchart of the search and screening process is presented in Figure 1. Of the 490 articles screened at the title and abstract phase, 7 systematic reviews and meta-analyses that examined racial or ethnic differences in depression and eating disorders were ultimately included (a total of 195 original studies) (22–28). Relevant systematic reviews and meta-analyses that examined racial or ethnic differences in anxiety or PMS/PMDD did not meet inclusion criteria for this narrative synthesis. Table 1 displays characteristics and brief description of results of each included review. Across the 7 systematic reviews and meta-analyses, sample sizes ranged from 16 to 16,214 that included college-aged, adult or community-dwelling, or treatment-seeking populations. Of the reviews that reported on the percentage of female participants (22, 25, 26), 9.2% to 100% of females were represented in the original research studies. The 4 remaining reviews that included both males and females conducted subgroup analyses on racial or ethnic differences in women (23, 24, 27, 28). Five reviews reported broad categories of racial or ethnic groups of White/Caucasian, Black/African American, Hispanic/Latina, Asian American, and American Indian or Alaskan Native women (23–27). One systematic review (22) described Mexican American, Chinese American, immigrant Black, or Other racial or ethnic groups, whereas another meta-analysis (28) further specified the description of ethnic groups to include Ghanaian, African Caribbean, Chinese, Japanese, Korean, Indian, Pakistani, Arab, and Russian.

FIGURE 1.

Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flowchart.

TABLE 1.

Study characteristics and summary of racial or ethnic comparisons of depressive and eating disorders

Study (condition)	No. of studies, publication date ranges, and databases	N or percentage (%) of race, ethnicity and femalea	Main results relevant to race or ethnicity differences in women
Lara-Cinisomo et al. (22) Systematic review (depression)	Studies published between 1989 and 2018: 7 Databases searched: 7	Mexican American, other Hispanics, Latina: 1.6%–40% Chinese American, Asian: 25%–42% European American, Caucasian, Whites: 6.5%–58% African American women: 31%–51% Immigrant Black women: 3.1% American Indian or Alaskan Native: 41% Other: 12%	Overall, 5/7 studies found racial or ethnic differences compared with White women in somatization; 3/5 studies found differences between minoritized racial or ethnic groups: Puerto Rican women had significantly the highest prevalence of somatic symptoms whereas non–Hispanic White women had the lowest. African American and Latina women reported significantly more somatic symptoms than White/Caucasian. European American women reported significantly higher somatic symptom than Chinese American women. Hispanic women had significantly lower somatic symptoms than European Whites. African American women were significantly more likely to endorse psychomotor symptoms than White women with similar levels of depression. Latinas reported significantly higher somatic or affective symptoms than Whites and other race or ethnicity, but not higher than those reported for Black women
Mendelson et al. (23) Meta-analysis (depression)	Studies published between 1989 and 2005: 31 Databases searched plus reference lists of included articles: 5	Non–Latino whites: N = 38–24,507 Latinos: N = 22–8308 Percentage of female: 9.2%–100%	No group difference in lifetime prevalence of major depressive disorder Latino adults reported small, yet significantly higher depressive symptoms than White adults Latina women reported significantly higher depressive symptoms than White women (and more pronounced than overall group difference)
Franko et al. (24) Meta-analysis (eating disorders)	Studies published between 2002 and 2010: 11 Clinical trials of BED database searched for peer-reviewed publications	Caucasian: N = 1,159 African American: N = 102 Hispanic/Latino: N = 64 Caucasian female: N = 973 (84%) African American female: N = 92 (90.2%) Hispanic/Latina: 52 (81.3%)	Caucasian, African American, and Hispanic women reported similar episodes of subjective binge eating (16 over the past month) African Americans had significantly higher dietary restraint scores than Caucasians Hispanic women reported significantly higher Global, Eating Concern, and Shape Concern scores on the Eating Disorder Examination than Caucasians and significantly greater Eating Concern scores than African Americans
Goode et al. (25) Systematic review (eating disorders)	Studies published between 1998 – 2019: 38 Databases plus reference sections of included articles: 6	Black women: 7%–40% in n = 18 studies and ≥ 40% in n = 20 studies (sample size of included studies ranged from 16 to 11,222)	Binge eating disorder prevalence in Black women ranged from 0.06 to 2.2% in 30 studies that reported prevalence rates Higher lifetime prevalence of BED found in White women, than Black women reported in 5 studies Prevalence of binge eating between Black and White women: Prevalence of binge eating estimated at 1.5 to 36% in Black women 7/30 studies (23%) reported no significant differences in the prevalence of binge eating 5/30 studies (17%) indicated that White women had significantly higher binge eating 4/30 studies (13%) reported that Black women had a significantly higher prevalence of binge eating Racial differences in eating disorder symptoms: 4/9 studies (44%) showed no racial differences in symptoms of weight, shape, and eating concerns between Black and White women 2/9 studies (22%) found that Black women were less likely to report eating disorder symptoms 2/9 studies (22%) showed few racial differences in the clinical presentation of symptoms, although Black women had higher BMIs than White women.
O’Neill et al. (26) Meta-analysis (eating disorders)	Studies published between 1987 and 2001: 18 Databases plus reference section of included articles and a manual search in the International Journal of Eating Disorders over the past 2 years: 3	Range of Black/African American samples: N = 33–100 Across entire meta-analysis, 30% of sample was Black/African American	No significant differences between African American women and White women in rates of bulimia (Hedges’ d = −.02) or BED (Hedges’ d = .01) Significant overall effect size indicated that African American women reported fewer eating disturbances than White women (Hedges’ d = −.060), despite similar drives for thinness scores between African American and White women. Significant overall effect size indicated that African American women scored significantly lower scores on eating attitudes and concerns than White women (Hedges’ d = −.38)
Roberts et al. (27) Meta-analysis (eating disorders)	Studies published between 1984 and 2004: 55 Databases plus request from professional Listserv of unpublished data: 4	Indicated analyses on Black-White females but did not specifically report N or % across included studies	Significant effect sizes indicated that African American females exhibited greater body satisfaction than did White females (d+ = 0.28) Significant effects were smaller using silhouettes methodology (d+ = 0.14) than weight (d+ = 0.32) or global (d+ = 0.33) questionnaires Age moderated ethnic differences in body satisfaction (differences most pronounced during college years)
Wildes et al. (28) Meta-analysis (eating disorders)	Studies published between 1986 and 1996: 35 Databases plus reference sections of included articles: 3	White or Caucasian: n = 9937 Black or African: n = 5997 African American: n = 5441 Ghanaian: n = 394 African Caribbean: n = 207 Asian: n = 1004 Asian American (Chinese, Japanese, etc.): n = 324 Asian British (Indian, Pakistani, or Bangladeshi origin): n = 680 Other: n = 843 Arab: n = 110 Hispanic: n = 138 Russian: n = 95 “Non–White”: n = 500	Almost 80% of the effect sizes were positive: White samples reported greater eating disturbance and body dissatisfaction than non–White samples in more than three fourths of the recorded outcomes Small to moderate magnitude of effect sizes revealed that White samples scored higher than non–White samples in all studies and outcome measures of eating disturbance and body dissatisfaction (Cohen’s d = 0.29–0.34) Small mean effect sizes found between White and non–White samples differences for bulimia, eating disorder, and weight and dieting concerns Ethnicity determined the magnitude and direction of mean effect sizes: Large effect sizes were significant when studies compared Black and White women White women reported more symptoms of dietary restraint, body dissatisfaction, and smaller ideal body preference Asian women reported significantly greater symptoms of bulimia, concerns for weight and dieting, and body dissatisfaction than White women Asian and White women did not significantly differ in symptoms of eating disturbance/body dissatisfaction, eating disorder, and dietary restraint

BED = Binge eating disorder

^aRace, ethnicity, and gender inclusion described as it is in the systematic review or meta-analysis

Quality Assessment

A summary of AMSTAR-2 ratings is in Table 2. One review was rated as high confidence in the results of the review (25), 1 review was rated as low (22), and 5 reviews were rated as critically low (23, 24, 26–28). Low or critically low confidence ratings were because of limited information on reasons for study inclusion or no reference list of excluded studies, lack of an assessment of the risk of bias in the original studies and its impact on results of the review, or lack of publication and/or study registration of an a priori written protocol.

TABLE 2.

Summary of AMSTAR-2 Ratings

1 = PICO provided; 2 = “a priori” design provided; 3 = explanation study design selection; 4=comprehensive literature search strategy; 5=study selection in duplicate; 6=data extraction in duplicate; 7=list and justification of excluded studies provided; 8=description of included studies available; 9 = use of satisfactory technique: assessment RoB of individual studies; 10 = sources of funding reported; 11 = use of appropriate methods for statistical combination of results – RCT; 12 = assessment of potential impact of RoB in results of individual studies on the results; 14 = explanation for and discussion of heterogeneity in results; 15 = if quantitative synthesis: adequate investigation of RoB and discussion on the results conducted; 17 = conflict of interest statement PICO = population/ intervention/ comparison/ outcome; AMSTAR = assessment of multiple systematic reviews; RoB = risk of bias

Racial and Ethnic Comparisons in Depression

One meta-analysis (23) and one systematic review (22) examined racial or ethnic differences of MDD and symptom presentation. Mendelson et al. (23) examined the prevalence of MDD in Latino and non–Latino White adults only to determine whether ethnicity is a risk factor for depression in the general population. They excluded original studies that considered other specialized populations (e.g., individuals with home-lessness, individuals with severe traumatic experiences) as depression may be different in these populations compared with the general population. Lara-Cinisomo et al. (22) reviewed differences in somatic symptoms of depression across women of diverse racial or ethnic groups (European American/Caucasian White, Hispanic/Latinas, Chinese American/Asian, African American, immigrant Black, American Indian or Alaskan Native, and other). The prevalence and symptoms of MDD were determined using structured clinical interviews (e.g., Composite International Diagnostic Interview [CIDI] (29); Diagnostic Interview Schedule [DIS] [30]) and validated measures for depression and depression subscales (e.g., Center for Epidemiologic Studies Depression Scale [CES-D] (31) ; Beck Depression Inventory [BDI] (32); Hamilton Rating Scale [HRS] [33]). Somatic symptoms of depression were assessed using validated global measures (e.g., BDI; HRS; Somatic Symptoms Index [34]) or measures with somatic subscales (somatic domain of the CES-D; the somatic-affective symptoms from the BDI).

Only one meta-analysis focused on lifetime prevalence of MDD and current depressive symptoms in 2 ethnic groups: Latinos and non–Latino Whites, encompassing males and females. They also conducted subgroup analyses in Latinos of Mexican descent and in Latino and non–Latino White females. The lifetime prevalence of MDD in Latino and non–Latino Whites was determined as odds ratios (ORs) and differences in current symptoms of depression were determined as standardized mean differences (Hedge’s adjusted g). The summary OR revealed that although Latinos had slightly lower odds of lifetime prevalence of MDD than non–Latino Whites, lifetime prevalence did not significantly differ between the 2 groups (OR, 0.89; 95% confidence interval [CI], 0.72–1.10) (23). Summary depressive symptom scores were small, but significantly higher in Latinos by 0.19 standard deviations (SDs) (95% CI, 0.12–0.25) than non–Latino whites. Summary effect estimates for the subgroups revealed that symptoms of depression were similar in Mexican Americans (males and females, 0.21; 95% CI, 0.15–0.27) and somewhat more extreme in Latino and non–Latino White women (0.38, 95% CI, 0.18–0.57) compared with the overall summary effect estimate in Latinos and non–Latino Whites. Overall, the lifetime prevalence rates of MDD appears to be relatively similar between Latinos and non–Latino Whites, although current depressive symptoms were more pronounced in Latinos, especially in Latina women (23).

Significant ethnic differences also emerged in the somatic presentation of depression in a systematic review of 7 original research studies (22). Specifically, 43% of the original research studies found that African American women with depression had higher somatic symptoms than White (P<.05) and Latina women (P<.001), and Hispanic/Latina women with depression had a higher prevalence of somatic symptoms than White/Caucasian women (.001<P<.05). Other studies found that European American women reported higher somatic symptoms compared with Chinese American women (P<.05) or found no differences in somatic symptoms between Asians, non–Hispanic Whites, American Indians, and Alaskan Natives. Although racial or ethnic differences in the somatic symptoms of depression was not consistently observed, Lara-Cinisomo et al. (22) discussed that African American and Hispanic/Latina women may be more likely to endorse somatic symptoms because of the cultural stigma of mental illness and potentially seek treatment to target the physical symptoms of depression (22).

Racial and Ethnic Comparisons in Eating Disorders

Five meta-analyses and one systematic review examined racial or ethnic differences in the prevalence of BN or BED (25, 26) or eating disorder symptoms (shape, weight, or eating concerns, binge eating, dietary restraint, or body dissatisfaction) (24–28). The prevalence rates of BN or BED and associated symptoms were assessed by validated self-report questionnaires (e.g., Eating Disorder Examination Questionnaire (35); Binge Eating Scale [36]) or structured clinical interviews [e.g., Eating Disorder Examination (37), Structured Clinical Interview for Axis I DSM-Ⅳ Disorders (38)]. Four reviews synthesized differences in prevalence rates and eating disorder symptoms among Black/African American women compared with White women (24–26, 28); 2 reviews compared disordered eating symptoms in Hispanic/Latina (24) and Asian American (28) with White women.

Only 2 reviews examined differences in the prevalence rates of BN or BED in Black women compared with White women (25, 26). No other review examined the prevalence rates of BN or BED in any other racial/ethnic group. One meta-analysis found that African American women did not significantly differ from White women in prevalence rates of BN (small effect size Hedges’ d, −0.02; 95% CI, −0.07 to +0.04; P=.56) or BED (small effect size Hedges’ d, 0.02; 95% CI, −0.04 to 0.05; P<.86) (26). Similarly, 17% of the original studies reported in the systematic review by Goode et al. (25) found that White women had a significantly higher lifetime prevalence rate of BED than Black women, although the exact prevalence rate was not reported. No original studies in this systematic review indicated that Black women had a higher lifetime prevalence rate of BED than White women (25).

In terms of racial or ethnic differences in specific disordered eating symptoms, one meta-analysis broadly defined groups as White and non–White (i.e., Black or African, Asian or Asian American, Arab, Hispanic, or Russian). These investigators found small to moderate effect sizes across all studies and measures of eating disturbances and body dissatisfaction (Cohen’s d, 0.29–0.34; SD, 0.41) with White samples scoring approximately one quarter of an SD above non–White samples (28). However, mean effect sizes differences for bulimia (mean [M] = 0.19, SD = 0.32), eating disorder (M 0.15, SD = 0.35), and weight and dieting concerns (M = 0.16, SD = 0.33) between White and non–White samples were small (28), suggesting that disordered eating symptoms between these 2 groups are relatively similar.

However, evidence of racial or ethnic differences in certain disordered eating symptoms emerged when specific racial or ethnic groups were compared, for example, Black compared with White women. Most of the reviews (4/6) focused on differences between Black/African American and White/Caucasian women. Overall evidence found no significant difference in binge eating frequency between Black/African American, White/Caucasian, and Hispanic/Latina women (24, 25). Compared with White/Caucasian women only, overall evidence also found that Black/African American women reported lower rates or fewer symptoms of self-induced vomiting and laxative/diuretic misuse (24, 25), eating disturbances (anorexia, bulimia, BED; small effect size, Hedges’ d, −0.06; 95% CI, −0.09 to −0.03; P<.001) (26), and concerns for disordered eating (small effect size, Hedges’ d, −0.38; 95% CI, −0.55 to −0.22: P<.001) (26). In 2 of the 3 meta-analyses, Black/African American women had lower rates of dietary restraint (small effect size, Hedges’ d, −0.17; 95% CI, −0.22 to −0.11; P<.001) (26), whereas White women reported more symptoms of dietary restraint (M, 0.93; SD, 0.34; P<.001) (28). In contrast, African American women with a high body mass index (BMI) reported significantly higher dietary restraint scores than White/Caucasian women with similar BMIs (P<.0.01) (24). Thus, differences in dietary restraint scores between Black/African American and White/Caucasian women may be due to differences in BMI, rather than ethnic differences.

We found 2 meta-analyses that compared differences in body dissatisfaction (or satisfaction) between Black/African American and White/Caucasian women. Overall evidence indicated that White women reported more symptoms of body dissatisfaction (M, 0.46; SD, 0.34) and smaller ideal body size (M, 0.81; SD, 0.16; P<.001) than Black/African American women (28). Alternatively, Black women exhibited significantly greater body satisfaction than White women (d₊ = 0.28; 95% CI, 0.25–0.32) (27). However, the magnitude of the effect was dependent on the assessment for body satisfaction, with original studies in this meta-analysis using the silhouettes methodology produced significantly smaller effects (d₊ = 0.14) than original studies using either the weight-related (d₊ = 0.32) or global (d₊=0.33) questionnaire approaches (between group heterogeneity Q_B(2) = 21.04, P<.001). Age also significantly moderated Black–White differences in body satisfaction, with the greatest ethnic differences occurred among women in their early 20s and then rapidly narrowed by age 40 on global and weight questionnaire methods (β = −0.46; z score = −6.48; P<.0001) (27).

The evidence of racial or ethnic differences in disordered eating symptoms in Hispanic/Latina and Asian American women was limited, with only 2 meta-analyses estimating differences in these groups. Treatment-seeking Hispanic women reported significantly higher Global, Eating Concern, and Shape Concern scores on the Eating Disorder Examination than Caucasian women (P<.05) as well as significantly higher Eating Concern scores compared with African American women (P < .05) (24). A similar effect was found for Asian American women, who reported significantly greater symptoms of bulimia (P=.01), concerns for weight and dieting (P=.05), and body dissatisfaction (P=.05) than White women (28).

DISCUSSION

This narrative synthesis aimed to investigate the racial/ethnic mental health paradox and where prevalence rates of depression, anxiety, eating disorders, and PMS or PMDD differed among US women. Although we conducted a comprehensive and systematic review of the available literature across 5 biomedical databases, 7 systematic reviews and meta-analyses that totaled 195 original studies met final inclusion criteria for this narrative synthesis. Relevant evidence was only available for depression and eating disorders.

Our narrative synthesis found evidence that Latino and non–Latino White adults (males and females) had similar odds of the lifetime prevalence of MDD (23). However, we found significant effects in the prevalence of depressive symptoms across racial and ethnic groups (22, 23). Specifically, Latino adults reported higher depressive symptom scores than White adults (in an analysis combining males and females). This difference was more pronounced in a subgroup analysis of women, with Latinas reporting high depressive symptoms (23). In addition, Puerto Rican, African American, and Hispanic/Latina women with depression reported significantly higher somatic symptoms than White/Caucasian women with depression (22). However, this finding was not consistent across all studies in the systematic review, some of which found that Chinese American and Hispanic women reported significantly lower somatic symptoms than European American women (22).

A similar pattern emerged across the 5 reviews that considered racial/ethnic differences in the prevalence rates of eating disorders and symptoms. We identified 2 reviews that investigated differences in the prevalence rates of BN or BED in Black/African American and White/Caucasian women. Black/African American and White women did not significantly differ in the prevalence rates of BN/BED (26) or White women had higher lifetime prevalence of BED than Black/African American women (25). Other than binge eating frequency, where Black/African American, White/Caucasian, and Hispanic/Latina women showed similar findings (24, 25), most of the reviews found small to moderate effect sizes indicating that Black/African American women generally reported fewer disordered eating symptoms, concerns for disordered eating, and increased body satisfaction (25–27). Black/African American women also reported significantly lower dietary restraint scores than White women (26), with the exception among Black/African American women with high BMI who reported higher dietary restraint scores than White women with similar BMI (24). For Hispanic/Latina and Asian American women, the overall evidence was limited to 2 meta-analyses that found that Asian American women reported greater symptoms of bulimia and weight and dieting concerns and higher overall body dissatisfaction than White women (28); Hispanic/Latina women reported higher eating, shape, and weight concerns than those reported by both White and Black/African American women (24).

It is interesting that no systematic review or meta-analysis that considered racial or ethnic differences in anxiety met inclusion criteria for this narrative synthesis. Thus, this is a clear gap in the current literature. Systematic reviews and meta-analyses that considered anxiety as an outcome were excluded because these reviews either did not consider or examine prevalence rates and symptoms across the racial and ethnic groups, women, or both; reported prevalence rates of anxiety comorbid with physical health conditions; or considered prevalence rates and symptoms in an adolescent and/or non–US-based sample. In 2020, 12-month prevalence rates in anxiety disorders across racial and ethnic adults ranged from 16.1% in Black or African American adults to 29.9% White adults (13). However, the significant cultural variation in the expression of anxiety across the racial and ethnic groups in the United States likely impacts the reporting of the prevalence rates of anxiety disorders and symptoms when defined by the DSM-5 (39, 40).

We also did not find any published systematic reviews or meta-analyses that examined the racial or ethnic differences in PMS or PMDD among US women. At the time of our search, only one systematic review on the prevalence rates and symptoms of PMS/PMDD had been published, but it was excluded from this narrative synthesis because it focused on a sample of women in India (41). A meta-analysis also examined racial or ethnic differences in treatment responses for PMDD but did not consider such differences in prevalence rates or symptom presentation of PMDD (42). A study protocol for a meta-analysis was published recently that will estimate the global and regional prevalence and risk factors of PMS/PMDD (43). With PMDD recognized as a formal psychiatric disorder in 2013, research on racial and ethnic differences in the prevalence and presentation of PMS/PMDD is still emerging. However, one cross-sectional study found higher lifetime prevalence rates of PMDD and premenstrual symptoms in White women (4.4% and 56.0%, respectively) than those among Black women (2.9% and 43.0%, respectively) (44). However, PMS/PMDD were measured using the premenstrual symptoms module of the World Mental Health Composite International Diagnostic Interview (45) based on DSM-IV criteria (46). Such nationally representative epidemiology studies using DSM-5 criteria are needed to understand differences in symptoms and prevalence rate across the racial and ethnic groups of US women.

Overall, the findings described from these systematic reviews and meta-analyses should be interpreted with caution given the results of our AMSTAR-2 assessment and methodological challenges described in the reviews. Our assessment on the quality of these reviews using AMSTAR-2 identified only one review with a high confidence rating (25). The remaining reviews were rated as low (22) to critically low (23, 24, 26–28), primarily owing to lack of description of the risk of bias assessment and impact on results of the original studies and limited justification and description of excluded studies. In addition, methodological considerations likely explain variation in the prevalence rates and symptoms of disorders reported in this narrative synthesis. First, there was significant heterogeneity in the assessment tools and approaches used to measure depression and eating disorders. For example, one systematic review noted that 11 measures were used to determine binge eating outcomes (25). Although meta-analyses statistically accounted for the heterogeneity across assessment tools and sample size as well as type, the levels of heterogeneity could limit the reliability of prevalence comparisons significantly across studies. Second, validated assessments are typically normed on most of the population [i.e., White/Caucasian populations (47)] and may not adequately assess diagnosis or symptoms in racially/ethnically minoritized women if symptom presentation and reporting practices differ. Third, the reviews were limited by the relatively small sample size of racial or ethnic women represented in the original research studies (22–24, 26), thus limiting the statistical power to examine interaction effects between race/ethnicity and sex/gender.

Our findings should be considered in the context of several limitations. First, because of our inclusion and exclusion criteria, we only included evidence from 7 systematic reviews and meta-analyses. Excluded reviews that defined other population types [e.g., Indigenous vs. nonindigenous (48)] could provide additional explanation for differences in prevalence rates or symptoms of psychiatric disorders across minoritized groups. Systematic reviews and meta-analyses were also excluded if they considered racial and ethnic differences of the psychiatric conditions comorbid with physical health conditions (e.g., obesity, diabetes) or during reproductive stages (e.g., the perinatal period, the menopause transition). Given that certain psychiatric disorders are highly comorbid with physical health conditions (49), future research should examine whether the racial or ethnic variation in psychiatric disorders extend to populations with physical health and/or reproductive-related conditions. Second, we focused on mental health disparities in racially or ethnically minoritized women only and excluded reviews that considered additional minoritized social identities (e.g., sexual orientation). Significant research in this area is needed to understand how women with multiple minoritized social identities may be differentially impacted by psychiatric disorders compared with women with fewer minoritized identities (50).

Despite these limitations, we conducted the first narrative synthesis to provide a rigorous and systematic analysis of prevalence rates and clinical symptoms of female-specific and female-predominant psychiatric disorders across racially and ethnically minoritized US women. We also conducted a quality assessment of the included reviews based on the revised AMSTAR-2 criteria (20). It is evident that there are multiple and complex contributors to variation in prevalence rates and symptoms in racially or ethnically minoritized women. We offer methodological recommendations to address gaps in this literature (e.g., coordinating assessment methods that are validated in diverse populations and culturally sensitive recruitment of racially or ethnically minoritized women to examine both between- and within-group differences). We also provide a direction to determine the prevalence rates of these disorders, the presentation of clinical symptoms, and the unique sociocultural experiences (e.g., racial discrimination, income inequality) that may uniquely impact the racially or ethnically minoritized women(51, 52). Researchers aimed to better understand the differential presentation and prevalence of psychiatric disorders across racial or ethnic groups could consider using the Research Domain Criteria from the National Institute of Mental Health (53). This is a framework aimed to shift focus from diagnostic categories to symptom presentation to better understand how different racial and ethnic groups may present different patterns of psychiatric symptoms.

CONCLUSION

We identified that although, racially or ethnically minoritized women generally report lower prevalence rates in depressive and eating disorders, they exhibit a different, and at times greater, symptom presentation than White women. Nationally representative epidemiological and community-based studies are needed to determine the racial or ethnic differences in presentations of psychiatric disorders and how these could impact the prevalence rate estimates depending on the diagnostic criteria used. Nonetheless, recent sociocultural events over the past 2 years—the effects of the coronavirus disease 2019 pandemic disproportionately burdening female care-givers (54) and the sharp rise in racism and discrimination toward Asian American women (55)—suggest that racially or ethnically minoritized women may be more vulnerable to psychiatric disorders than before. It is imperative that the women’s health professionals and decision makers recognize and consider the variations that exist in the mental health experience of all women, but more so in racially and ethnically minoritized women whose needs are often not given equal weight as that given to the needs of most population.