Discrimination is associated with poor sleep quality in pregnant Black American women

Abstract

Background:

Heightened exposure to racial/ethnic discrimination is associated with poorer sleep health among non-pregnant adults. This relationship has received limited research attention among pregnant women, despite the importance of prenatal sleep quality for optimal maternal and child health outcomes.

Methods:

We utilized perinatal data from a sample of Black American women (n = 600) participating in a cohort study who reported their lifetime experiences of racial/ethnic discrimination and gendered racial stress during early pregnancy and reported on their sleep quality and depressive symptoms during early and mid-pregnancy. Hierarchical multiple linear regression models were fit to examine associations between lifetime experiences of racial/ethnic discrimination or gendered racial stress and sleep quality during early and mid-pregnancy. We also adjusted for women’s concurrent depressive symptoms and tested whether the discrimination/sleep quality association varied by socioeconomic status.

Results:

Greater exposure to racial/ethnic discrimination was associated with poorer sleep quality during early (ΔR² = 0.04, ΔF = 26.08, p < 0.001) and mid-pregnancy (ΔR² = 0.02, ΔF = 9.88, p = 0.002). Similarly, greater gendered racial stress was associated with poorer sleep quality during early (ΔR² = 0.10, ΔF = 65.72, p < 0.001) and mid-pregnancy (ΔR² = 0.06, ΔF = 40.43, p < 0.001. These findings largely held after adjustment for concurrent prenatal depressive symptoms. Socioeconomic status did not modify the observed relationships.

Conclusions:

Efforts to decrease institutional and interpersonal experiences of racial/ethnic discrimination and gendered racism would benefit the sleep quality of pregnant Black American women, particularly during early pregnancy.

1. Introduction

Black American adults experience poorer sleep health than their White counterparts [1]. Unexpected differences in sleep health that systematically and adversely affect historically marginalized individuals are known as sleep health disparities [2]. Sleep health disparities are the likely product of longstanding exposure to psychosocial stressors, such as racial/ethnic discrimination. Indeed, Black American adults are disproportionately exposed to racial/ethnic discrimination [3], and chronic racial/ethnic discrimination is concurrently, and in some cases, prospectively associated with poor sleep health among adults [4]. Heightened exposure to racial/ethnic discrimination may confer risk for poor sleep health via several pathways. Potential mechanisms of transmission include greater likelihood of exposure to environmental living conditions that negatively affect sleep health (e.g., pollution, noise, ambient lighting) among racial/ethnic minorities as a consequence of institutional racism (e.g., housing segregation) [5,6]. In addition, individuals subjected to repeated instances of interpersonal and/or institutional discrimination may develop a sense of “racism-related vigilance” [7] or hyperarousal, to subsequent threat. Such anticipatory anxiety may manifest cognitively as rumination [8], which is positively associated with poorer sleep in both non-pregnant [9] and pregnant individuals [10,11].

To date, empirical evidence for direct associations between racial/ethnic discrimination and sleep health has largely been restricted to the non-pregnant population. However, examining associations between these constructs among Black American adults may be particularly important during pregnancy. First, consistent with an intersectionality framework [12,13], pregnant Black American women hold multiple, overlapping social identities including, but not limited to, their racial/ethnic and gender identities. Associations between discrimination and sleep health within this group of pregnant women may reflect women’s exposure to different, but interrelated interpersonal stressors, including both racial/ethnic discrimination and gendered racial stress. Second, poor sleep quality during pregnancy is associated with systemic inflammation, with pregnant Black women noted to have a greater inflammatory response to sleep disturbance than their White counterparts [14]. Notably, this sleep-induced inflammatory profile is associated with a greater odds of preterm birth [15,16]. Poor sleep health during pregnancy is linked with adverse birth outcomes, particularly among Black American women, who may be unfairly exposed to significant psychosocial stress by nature of their intersectional identities. Fortunately, prenatal sleep health is modifiable and responds well to clinical intervention [17–20]. Examining associations between racial/ethnic discrimination and gendered racial stress and modifiable health behaviors such as sleep quality is critical in addressing health disparities and improving health outcomes for those unjustly exposed to adversity.

Recent calls for research highlight a “critical need … [to understand] the role of racial disparities and systemic racism” on sleep health during pregnancy [21]. While several studies have examined racial/ethnic disparities in prenatal sleep quality [22,23], to our knowledge, only one study has investigated relevant predictors of prenatal sleep quality (e.g., discrimination experiences) within this research framework. Francis and colleagues [24] uncovered positive associations between everyday experiences of racial/ethnic discrimination and poorer prenatal sleep quality in a sample of pregnant Black American women. The authors did not find evidence for associations between these constructs when sleep quality was assessed at a later timepoint during pregnancy; however, these analyses could only be examined in a subset of their sample (n = 133 of N = 640), which may have decreased statistical power to detect effects. As this was the first study of its kind, it remains unclear whether greater experiences of racial/ethnic discrimination are associated with poor sleep quality in early pregnancy, as well as with poor sleep quality later in gestation. Poor sleep quality in early pregnancy is associated with an increased risk of preterm birth [16], and poor sleep quality in late pregnancy is additionally associated with greater risk of preterm birth, longer labor, and a greater likelihood of caesarean delivery [25,26]. The current study design will allow us to examine associations between racial/ethnic discrimination and sleep quality during early and mid-pregnancy in a sample with sufficient statistical power. In addition, because pregnant Black American women are exposed to additional stressors as a function of their gender, their race/ethnicity, and the interaction between these related identities [12,13], we will also examine associations between gendered racial stress and sleep quality during early and mid-pregnancy. This intersectionality framework has been recommended in studies of sleep health [1], but has not yet been applied to studies of sleep health during pregnancy.

The concept of intersectionality posits that the lived experience associated with being a Black woman in America confers both unique strength and vulnerability. In terms of vulnerability, Black American women are at risk of structural and interpersonal exposure to both racial/ethnic discrimination and sexism. Intersectionality theorists assert that these stress exposures occur simultaneously and may be better characterized as an interlocking construct known as gendered racism [27]. Gendered racism is prejudicial treatment and behavior that is tied to women’s racial/ethnic and gender identities [28]. Researchers have recommended that studies examining associations between Black American women’s interpersonal exposure to racial/ethnic discrimination and their health outcomes should also include women’s exposure to gendered racism as an additional predictor variable, given its particular salience for this population [29]. To our knowledge, no study to date has examined associations between pregnant women’s lifetime exposure to gendered racism and their sleep quality during pregnancy or the sleep health of their children. As such, in our review of the extant literature on discrimination and sleep health, we focus on evidence for associations between racial/ethnic discrimination and sleep health, rather than evidence for associations between gendered racism and sleep health. Consistent with recommendations to incorporate an intersectionality conceptualization in studies of discrimination and sleep health [1], in the current study, we broaden the operational definition of discrimination and include measures of both racial/ethnic discrimination and gendered racism or gendered racial stress as our predictor variables. This methodological approach emphasizes intersectionality and allows us to better understand Black American women’s unique lived experiences and exposure to multiple aspects of discriminatory stress. This novel approach will allow us to better understand the ways in which racism and gendered racism are associated with the sleep health of pregnant Black American women and their children.

The majority of studies that examine heightened exposure to racial/ethnic discrimination as a predictor of poor sleep quality include depressive symptoms in predictive models [4]. This is important, as sleep disturbances often predate and/or co-occur with depressive symptomatology [30], so much so that they are listed as a core diagnostic feature of clinical depression [31]. Including depressive symptoms in statistical models allows researchers to determine whether greater lifetime exposure to racial/ethnic discrimination is associated with poor sleep quality beyond concurrent symptoms of depression. This is particularly relevant during pregnancy, when prevalence rates of depression range from 12 to 27% [32]. Francis and colleagues controlled for depressive symptoms in their models, but they reported only unadjusted findings (i.e., without depressive symptoms added to the models) [24], making it difficult to determine how much of the variance in expectant mothers’ sleep quality may have been accounted for by their current depressive symptoms. In the current study, we will adjust for mothers’ current depressive symptoms when examining racial/ethnic discrimination and gendered racial stress as predictors of sleep quality during pregnancy.

Consistent with recommendations to utilize an intersectionality framework in studies of sleep health [1], racial/ethnic discrimination and gendered racial stress may have differential effects on the sleep health of pregnant Black American women at different levels of socioeconomic status (SES). The literature on sleep health among pregnant Black American women is underdeveloped [21]; as such, to date, SES has not been examined as a moderator in the association between racial/ethnic discrimination/gendered racial stress and sleep quality. However, findings from the non-pregnant adult literature suggest that the association between racial/ethnic discrimination and sleep health may be stronger among Black American adults of higher SES.

For instance, sampling a large, population-based cohort, Johnson and colleagues demonstrated that associations between psychological distress and short sleep duration were strongest among Black Americans who received a college education or higher [33]. The “diminishing returns” hypothesis suggests that Black American individuals of higher SES may experience greater discrimination, which may confer greater risk for poorer health outcomes, such as sleep [34]. This effect may occur because high social position predicts recurrent interactions with members of the majority racial/ethnic group (i.e., whites), thus increasing exposure to potential discrimination. An increase in these interactions may concurrently result in more frequent anticipation of being discriminated against. Consistent with prior research [35,36], we hypothesize that the association between racial/ethnic discrimination or gendered racial stress and sleep quality will be stronger among pregnant Black American women of higher SES.

The current study seeks to (1) examine associations between discrimination and sleep health both during early and mid-pregnancy among Black American women, and (2) do so within a framework that acknowledges the importance of intersectional identities in shaping health outcomes. We will also investigate the roles of self-reported depressive symptoms and socioeconomic status in these associations. Our use of an exclusively Black American sample of pregnant women aligns with calls for within-racial/ethnic group studies on sleep health [1]. Such study designs allow us to better understand how the unique lived experiences of a given racial/ethnic group may influence health outcomes.

2. Method

2.1. Participants

The current study constitutes a secondary data analysis of a prospective, longitudinal cohort study (the Prenatal Study; 5R01NR014800–02) that has been described in detail [37]. Recruitment for the Prenatal Study is ongoing and resulted in a sample of n = 600 pregnant women available for current study analyses. All procedures were performed in accordance with the ethical standards of the Institutional Review Board and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Healthy, pregnant women were recruited from prenatal clinics affiliated with two large hospitals – one private, one public – in a Southeastern city of the United States to participate in a study investigating the impacts of prenatal stress and environmental exposures on maternal and child health [37]. Inclusion criteria for enrollment in the Prenatal Study were as follows, Black/African American race/ethnicity and born in the United States (by self-report), singleton pregnancy, fluency in English, maternal age 18–40 years, and the absence of diagnoses of chronic health conditions (e.g., hypertension, diabetes) or chronic prescription medication use.

2.2. Procedure

Data were directly collected from study participants twice: once between eight- and 14-weeks’ gestation (M = 11.16 weeks, SD = 2.25) and once between 24- and 30-weeks’ gestation (M = 26.59 weeks, SD = 2.73). Data analyzed for the current study were collected between February 2014 and February 2021. At the first Prenatal Study visit (early pregnancy), women provided sociodemographic information, retrospectively reported on their lifetime experiences of racial/ethnic discrimination and gendered racial stress and reported on their current sleep quality and current depressive symptoms. At the second Prenatal Study visit (mid-pregnancy), women were asked to repeat the sleep quality and depressive symptom self-report measures. Medical record abstraction to ascertain maternal health and pregnancy characteristics was also performed at the conclusion of the pregnancy.

2.3. Measures

2.3.1. Lifetime experiences of racial/ethnic discrimination and gendered racial stress

At the first Prenatal Study visit, women retrospectively reported on their lifetime exposure to racial/ethnic discrimination using the Experiences of Discrimination (EOD) measure [38]. Women also reported on their experiences of gendered racial stress using the Jackson, Hogue, Phillips Contextualized Stress (JHP) measure [39].

EOD.

The EOD asks participants to respond (yes/no) to the following question:

“Have you ever experienced discrimination, been prevented from doing something, or been hassled or made to feel inferior in any of the following [nine] situations (i.e., at school, getting hired/getting a job, at work, getting housing, getting medical care, getting service at store/restaurant, getting credit/bank loans/mortgage, in public, with police/in courts) because of your race, ethnicity, or color?”

Women’s responses on the EOD are limited to what they are willing to disclose, and do not necessarily represent the full range of discriminatory experiences. No items on the EOD reference sleep quality. Responses are summed to result in a total score (range = 0–9). This score is considered reliable and valid: it has good internal consistency, good test-retest reliability over a month timespan, and good convergent validity with other self-report measures of discrimination [38]. This measure of racial/ethnic discrimination has previously been used in work investigating pregnant Black American women’s exposure to discrimination as a predictor of psychological outcomes [40–42]. Q-Q plots indicated that total scores in the current study sample were normally distributed (skewness = 0.91, SE = 0.10, kurtosis = −0.19, SE = 0.20). Internal consistency in the current study sample was good: Cronbach’s α = 0.83.

JHP.

The lived experience of Black American women is shaped by both racial/ethnic identity and gender identity [12,43]. As such, Black American women may be exposed to both racial/ethnic discrimination and gender-based discrimination, and these constructs are not wholly separable. The JHP is a 39-item self-report measure that assesses Black American women’s gendered racial stress. Participants indicate whether statements broadly describe their lived experience (0 = Strongly Agree, 1. = Agree, 2 = Unsure, 3 = Disagree, 4 = Strongly Disagree). Participants are not provided with a time anchor. Statements include, “Everyone expects me to be strong for them,” “As an African American woman, I can withstand great pressure,” “Racism is a problem in my life,” and “I have to work harder than White women to earn recognition.” The JHP is comprised of five subscales: burden, coping, racism, personal history, and work. Subscale scores are summed to result in a total summary score (range = 43–159). We tested current study hypotheses using summary scores from the JHP as a predictor variable, as we did not have a priori predictions about any particular subscale and wanted to tap the full construct of gendered racial stress. The JHP was developed via qualitative interviews with n >/= 400 women in metropolitan Atlanta, and has good convergent validity with self-report measures of anger, depressive symptoms, and anxiety [39]. In the current study sample, internal consistency for the JHP total summary score was good: Cronbach’s α = 0.84.

2.3.2. Prenatal sleep quality

At both Prenatal Study visits, women reported on their sleep quality in the past week using the Patient-Reported Outcomes Measurement Information System (PROMIS) Sleep Disturbance Short Form. [44] The PROMIS is a validated measure comprised of eight items that assess sleep quality (e.g., “My sleep was restless.“). No items directly reference mood symptoms or stress. Participants are asked to decide whether statements describe their sleep quality (1 = not at all, 2 = a little bit, 3 = somewhat, 4 = quite a bit, 5 = very much). Responses are summed to compute a total continuous raw score (range = 8–40); higher scores indicate worse sleep quality. These continuous scores were used as our outcome measure in current study analyses. The PROMIS has high internal consistency and good convergent validity with other measures of sleep quality [44,45] and has previously been used in samples of perinatal women [17,46]. In the current study sample, internal consistency for the PROMIS was good to excellent (early pregnancy Cronbach’s α = 0.90; mid-pregnancy Cronbach’s α = 0.89).

2.3.3. Prenatal depressive symptoms

Also at both Prenatal Study visits, women’s current depressive symptoms were assessed using summary scores from the Edinburgh Postnatal Depression Scale (EPDS) [47]. Participants indicated whether ten statements referencing low mood, suicidality, anhedonia, and stress described them within the past week. One item on the EPDS directly references sleep and fatigue (“I have been so unhappy that I have had difficulty sleeping.“) and was removed from all analyses to avoid methodological overlap, resulting in a total of nine EPDS items. Scores on the EPDS range from zero to 30. In the current study sample, internal consistency for the 9-item scale was good: Cronbach’s α = 0.83 (early pregnancy), 0.86 (mid-pregnancy).

2.3.4. Socioeconomic status

At the first Prenatal Study visit, women provided their highest education level and their income. Because socioeconomic status (SES) is best captured by multiple factors rather than a single variable [48] we operationalized SES as the average of standardized measures of education and income. This methodological approach is consistent with previous discrimination and sleep health research [49–51].

2.3.5. Covariates

We selected specific covariates for inclusion in modeling because of their known associations with prenatal sleep quality, including age, BMI, gestational age, and parity; higher levels of each are associated with poorer sleep quality [22,23,52]. Adjusting for these variables allows us to better isolate the role of racial/ethnic discrimination and gendered racial stress in predicting sleep quality in pregnancy. Each of these covariates was ascertained from the maternal medical record using a standardized abstraction. Maternal age in years was based on age at enrollment (first study visit). Prenatal body mass index (BMI) was calculated from measured height and weight, and gestational weeks of pregnancy was determined from the clinical record (based on last menstrual period in combination with obstetrical ultrasound according to accepted clinical standards [53].

2.4. Data analytic plan

2.4.1. Preliminary analyses

All analyses were performed using IBM SPSS version 27.0 [54]. Statistical significance was set at a two-sided p-value of <0.05. A priori power analyses using G*Power [55] revealed that a sample size of n = 395 was needed to detect small effects (Cohen’s f [2] = 0.02; α = 0.05; 1 - β = 0.80; number of tested predictors = 1; total number of predictors = 5). All participants (N = 600) in the current study sample completed the measure of experiences of racial/ethnic discrimination. Of these women, n = 597 completed the measure of gendered racial stress. Most women (n = 595) completed the sleep quality measure at the first Prenatal Study visit, and n = 406 women completed the sleep quality measure at the second Prenatal Study visit.

We examined our data for missingness in compliance with the Journal Article Reporting Standards (JARS) for Research in Psychology [56]. To determine whether our data were Missing Completely at Random (MCAR), Missing at Random (MAR), or Missing Not at Random (MNAR) we dummy-coded variables to signify whether participants were missing data on study variables at the first and/or second Prenatal Study visits. Next, we conducted Independent Samples T-tests and Chi-Squared tests to determine whether missingness was associated with predictor, outcome, and/or sociodemographic variables. Finally, missing data were multiply imputed using the multiple imputation package in SPSS. All results reported in the text, Tables, and Figures are based on the imputed data.

2.4.2. Primary analyses

We plotted standardized residuals against standardized predicted values to inspect for linearity, homoscedasticity, and normality and examined our data for influential outliers. We used the PROCESS macro in SPSS [57] to test our moderation hypothesis.

To test our primary hypothesis, the experiences of racial/ethnic discrimination score was evaluated as a predictor of women’s sleep quality at eight – 14 weeks gestation and 24–30 weeks gestation, in two separate regression models. Next, the gendered racial stress score was evaluated as a predictor of women’s maternal sleep quality at eight – 14 weeks gestation and 24–30 weeks gestation, also in two additional separate regression models. We first performed simple linear regressions to test all four associations, and then adjusted for covariates (i.e., women’s age, prenatal body mass index (BMI), gestational weeks of pregnancy, prenatal depressive symptoms) in subsequent hierarchical linear regression models.

We used the PROCESS macro in SPSS [57] to test our third study hypothesis. As before, we tested racial/ethnic discrimination and gendered racial stress as individual predictor variables of (1) early pregnancy sleep quality and (2) mid-pregnancy in sleep quality in separate models. SES was entered as the moderator variable in all models. This resulted in a total of four separate PROCESS models. Women’s age, prenatal BMI, gestational weeks of pregnancy (in either early or mid-pregnancy), and depressive symptoms (in either early or mid-pregnancy) were included as covariates in all four models.

2.4.3. Sensitivity analyses

A portion of data collection occurred during the COVID-19 pandemic. Given recent research suggesting poorer sleep quality among pregnant women during the early phases (i.e., March–April 2020) of the pandemic [58], we repeated all analyses with all participants who completed the first and/or second Prenatal Study visit after March 1, 2020 removed. This resulted in a total sample of N = 567 women available for sensitivity analyses: n = 13 women completed both Prenatal Study visits after March 1, 2020, and n = 20 women completed only the second Prenatal Study visit after March 1, 2020.

3. Results

3.1. Participants

There were no significant differences in predictor (i.e., racial/ethnic discrimination and gendered racial stress), outcome (i.e., sleep quality), or sociodemographic variables (all p’s > 0.05) between participants who did and did not have available racial/ethnic discrimination and gendered racial stress or sleep quality data at Prenatal Visit 1. However, participants who did not have available sleep quality data at Prenatal Visit 2 participated in Prenatal Visit 1 earlier during pregnancy (M gestational weeks = 10.77, SD = 2.16) than participants who had available sleep quality data at Prenatal Visit 2 (M gestational weeks = 11.35, SD = 2.26, t = −2.97, p = 0.003). There were no other significant differences between these groups (all p’s > 0.05). Given the significant association between gestational weeks of pregnancy at Prenatal Visit 1 and missingness at Prenatal Visit 2, we concluded that our data met Missing At Random (MAR) specifications [59].

Table 1 displays descriptive statistics for variables in the current study sample. On average, study participants were 25.36 years old (SD = 5.08 years) and for 46.3% of participants, this pregnancy was their first. The median highest level of education achieved was a high school diploma or General Educational Development (GED). Participants’ median income was less than 100% of the Federal Poverty Level. Table 2 displays bivariate correlations between variables in the current study sample. Correlations between racial/ethnic discrimination/gendered racial stress and sleep quality were small to moderate (r’s = 0.13–0.32, all p’s < 0.001). On average, participants’ sleep was not considered poor during early (m = 21.47, sd = 7.86) or mid-pregnancy (m = 22.01, sd = 7.44). However, 16% of women (n = 96) reported poor sleep during early pregnancy and mid-pregnancy (n = 98).

Table 1.

Descriptive statistics.

Variable	Imputed Data (N = 600)
	M (SD) or n (%)
Age (years)	25.36 (5.08)
Primiparous Women	278 (46.3%)
In a Relationship and Cohabitating	292 (48.7%)
Education	–
8th Grade or Less	1 (0.2%)
Some High School	94 (15.7%)
Graduated High School or GED	241 (40.2%)
Some College or Technical School	166 (27.7%)
Graduated College	66(11.0%)
Some Graduate Work or Degree	32 (5.3%)
Income	–
<100% of the Federal Poverty Level	265 (44.2%)
100–132% of the Federal Poverty Level	92 (15.3%)
133–149% of the Federal Poverty Level	54 (9.0%)
150–199% of the Federal Poverty Level	85 (14.2%)
200–299% of the Federal Poverty Level	41 (6.8%)
300–399% of the Federal Poverty Level	25 (4.2%)
>/= 400% of the Federal Poverty Level	38 (6.3%)
Gestational Weeks	–
Prenatal Study visit 1	11.16(2.25)
Prenatal Study visit 2	26.59 (2.73)
Body Mass Index (BMI)a	–
Prenatal Study visit 1	28.87 (7.83)
Racial/Ethnic Discrimination (EOD)b	–
Prenatal Study visit 1	2.13 (2.37)
Gendered Racial Stress (JHP)c	–
Prenatal Study visit 1	96.36 (20.91)
Depressive Symptoms (EPDS)d	–
Prenatal Study visit 1	7.18 (5.48)
Prenatal Study visit 2	7.03 (5.53)
Sleep Quality (PROMIS)e	–
Prenatal Study visit 1	21.47 (7.86)
Prenatal Study visit 2	22.01 (7.44)

Note.

^aBMI was calculated based on women’s height and weight and was only collected at the first Prenatal Study visit.

^bScores on the EOD range from 0 to 9.

^cScores on the JHP range from 43 to 159.

^dEPDS scores range from 0 to 30; in racial/ ethnic minority women, scores greater than or equal to 10 suggest clinically significant depressive symptomatology. EPDS means reported here include the sleep item, but when using EPDS summary scores in all study analyses, the sleep item was removed. Bivariate correlations between the 9 and 10-item EPDS scores were high (r = 0.99, p < 0.001 at Prenatal Study visit 1 and 2).

^ePROMIS scores range from 8 to 40; scores greater than or equal to 25 suggest mild sleep disturbance.

Table 2.

Bivariate Correlations Between Study Variables (N = 600 for all cells).

	1	2	3	4	5	6	7	8	9	10	11	12
1) Maternal age	-
2) Parity	0.33**	–
3) Maternal Body Mass Index	0.13**	0.09*	–
4) Gestational Weeks, Early Pregnancy	0.03	0.04	−0.01	–
5) Gestational Weeks, Mid-Pregnancy	−0.07	0.06	−0.10*	0.09*	–
6) Maternal SES	0.42**	−0.09*	−0.01	0.09*	−0.08	–
7) Racial/Ethnic Discrimination	0.14**	0.02	0.001	−0.05	−0.05	0.17**	–
8) Gendered Racial Stress	0.15**	0.08	0.05	−0.07	0.02	−0.01	0.32**	–
9) Sleep Quality, Early Pregnancy	0.09*	−0.07	0.02	0.03	0.04	0.10*	0.20**	0.32**	–
10) Sleep Quality, Mid-Pregnancy	0.12**	0.06	0.06	−0.004	0.13**	0.02	0.13**	0.25**	0.51**	–
11) Depressive Symptoms, Early Pregnancy	−0.02	−0.02	−0.02	−0.08	0.02	−0.07	0.26**	0.53**	0.38**	0.32**	–
12) Depressive Symptoms, Mid-Pregnancy	−0.02	0.08	−0.04	−0.02	0.01	−0.06	0.19**	0.32**	0.23**	0.31**	0.54**	–

Note.

^**p < 0.01

^*p < 0.05.

3.2. Main effects of discrimination on sleep quality during pregnancy

The results of simple linear regression analyses (Models 1a and 1b in Tables 3 and 4) reflect the bivariate correlations displayed in Table 2. Women’s reports of greater lifetime experiences of racial/ethnic discrimination measured in early pregnancy were associated with poorer sleep quality in early pregnancy (ΔR [2] = 0.04, ΔF = 26.08, p < 0.001; Model 1a, Table 3) and associated with poorer sleep quality in mid-pregnancy (ΔR [2] = 0.02, ΔF = 9.88, p < 0.002; Model 1a, Table 4). Similarly, women’s reports of gendered racial stress measured in early pregnancy were associated with poorer sleep quality in early pregnancy (ΔR [2] = 0.10, ΔF = 65.72, p < 0.001; Model 1b, Table 3) and associated with poorer sleep quality in mid-pregnancy (ΔR [2] = 0.06, ΔF = 40.43, p < 0.001; Model 1b, Table 4).

Table 3.

Early pregnancy sleep quality regressed on Women’s exposure to racial/ethnic discrimination or gendered racial stress (N = 600).

Predictor Variable – Racial/Ethnic Discrimination	B (SE)	95% CI (B)	β	t	p
Model 1a. Unadjusted	0.68 (0.13)	0.42–0.94	0.20	5.11	<0.001
Model 2a. Model 1 + adjustment for maternal age, gestational weeks at Prenatal Visit 1, prenatal BMI, SES, parity	0.64 (0.14)	0.38–0.91	0.19	4.74	<0.001
Model 3a. Model 2 + adjustment for depressive symptoms at Prenatal visit 1	0.30 (0.13)	0.04–0.56	0.09	2.26	0.02
Predictor Variable – Gendered Racial Stress	B (SE)	95% a (B)	β	t	p
Model 1b. Unadjusted	0.12 (0.02)	0.09–0.15	0.32	8.11	<0.001
Model 2b. Model 1 + adjustment for maternal age, gestational weeks at Prenatal Visit 1, prenatal BMI, SES, parity	0.12 (0.02)	0.09–0.15	0.32	8.19	<0.001
Model 3b. Model 2 + adjustment for depressive symptoms at Prenatal visit 1	0.06 (0.02)	0.02–0.09	0.14	3.20	0.001

Note.

^*BMI was only measured at Prenatal Study Visit 1 (early pregnancy).

Table 4.

Mid-pregnancy sleep quality regressed on Women’s exposure to racial/ethnic discrimination or gendered racial stress (N = 600).

Predictor Variable – Racial/Ethnic Discrimination	B (SE)	95% CI (B)	β	t	p
Model 1a. Unadjusted	0.40 (0.13)	0.15–0.65	0.13	3.14	0.002
Model 2a. Model 1 + adjustment for maternal age, gestational weeks at Prenatal Visit 2, prenatal BMI, SES, parity	0.39 (0.13)	0.14–0.64	0.13	3.07	0.002
Model 3a. Model 2 + adjustment for depressive symptoms at Prenatal Visit 2	0.20 (0.13)	−0.05–0.45	0.06	1.61	0.11
Predictor Variable – Gendered Racial Stress	B (SE)	95% a (B)	β	t	p
Model 1b. Unadjusted	0.09 (0.01)	0.06–0.12	0.25	6.36	<0.001
Model 2b. Model 1 + adjustment for maternal age, gestational weeks at Prenatal Visit 1, prenatal BMI, SES, parity	0.08 (0.01)	0.06–0.11	0.23	5.87	<0.001
Model 3b. Model 2 + adjustment for depressive symptoms at Prenatal Visit 2	0.05 (0.02)	0.02–0.08	0.15	3.64	<0.001

Note.

^*BMI was only measured at Prenatal Study Visit 1 (early pregnancy).

After adjustment for conceptually relevant covariates (Models 2a and 2b in Tables 3 and 4), effects were attenuated slightly but the overall pattern of findings remained the same. Women’s reports of greater lifetime experiences of racial/ethnic discrimination measured in early pregnancy were associated with poorer sleep quality in early pregnancy (ΔR [2] = 0.04, ΔF = 22.50, p < 0.001; Model 2a, Table 3) and passociated with poorer sleep quality in mid-pregnancy (ΔR [2] = 0.02, ΔF = 9.44, p = 0.002; Model 2a, Table 4). Similarly, women’s reports of gendered racial stress measured in early pregnancy were associated with poorer sleep quality in early pregnancy (ΔR [2] = 0.10, ΔF = 67.10, p < 0.001; Model 2b, Table 3) and associated with poorer sleep quality in mid-pregnancy (ΔR [2] = 0.05, ΔF = 34.41, p < 0.001; Model 2b, Table 4).

Associations adjusted for women’s concurrent prenatal depressive symptoms are shown in Models 3a and 3b in Tables 3 and 4 In early pregnancy, associations persisted, although effect sizes were attenuated. Women’s reports of greater lifetime experiences of racial/ethnic discrimination (ΔR [2] = 0.01, ΔF = 5.12, p = 0.02; Model 3a, Table 3) and gendered racial stress (ΔR [2] = 0.01, ΔF = 10.22, p = 0.001; Model 3b, Table 3) accounted for only a small percentage of the variance in early pregnancy sleep quality. In mid-pregnancy, after adjustment for concurrent prenatal depressive symptomatology, associations between racial/ethnic discrimination and sleep quality were no longer significant (ΔR [2] = 0.004, ΔF = 2.59, p = 0.11; Model 3a, Table 4). However, gendered racial stress continued to explain a statistically significant percentage of the variance in mid-pregnancy sleep quality, although this effect size was small (ΔR [2] = 0.02, ΔF = 13.22, p < 0.001; Model 3b, Table 4).

3.3. Interaction effects of discrimination and socioeconomic status on sleep quality during pregnancy

PROCESS analyses revealed no significant interaction between SES and racial/ethnic discrimination in the prediction of sleep quality in early (t = −0.54, p = 0.59) or mid-pregnancy (t = −0.99, p = 0.32). Similarly, there was no significant interaction between SES and gendered racial stress in the prediction of sleep quality in early (t = 0.59, p = 0.76) or mid-pregnancy (t = 0.75, p = 0.45).

3.4. Sensitivity analyses

Results of sensitivity analyses, in which we limited analyses to data points collected prior to the COVID-19 pandemic, indicated the same pattern of findings for all three study hypotheses. This suggests that the inclusion of data collected during the early COVID-19 pandemic did not influence our findings (data not shown).

4. Discussion

The current study adds to the growing literature on racial/ethnic discrimination and women’s sleep health during pregnancy. We found positive associations between racial/ethnic discrimination and poorer sleep quality during early pregnancy, replicating the work of Francis and colleagues [24]. In addition, we broadened our definition of exposure to discrimination to include a measure of women’s experiences of gendered racial stress, and found positive associations between gendered racial stress and poorer sleep quality during early and mid-pregnancy, respectively. Each of these associations held after accounting for sociodemographic factors such as SES, health-related factors such as women’s age, BMI and week of pregnancy, and women’s concurrent prenatal depressive symptoms. In contrast, we did not find evidence for associations between women’s exposure to racial/ethnic discrimination and their sleep quality during mid-pregnancy, after adjustment for their prenatal depressive symptoms.

In keeping with prior research methodology [4], we examined associations between women’s experiences of racial/ethnic discrimination or gendered racial stress and their prenatal sleep quality with adjustment for women’s concurrent prenatal depressive symptomatology. While associations largely held – apart from associations between women’s exposure to racial/ethnic discrimination and their mid-pregnancy sleep quality– effect sizes diminished in all models. This may suggest that pregnant women’s concurrent depressive symptoms are a better predictor of their sleep quality than their previous exposure to racial/ethnic discrimination or gendered racial stress. This finding is not entirely surprising: poor sleep quality and greater depressive symptomatology frequently co-occur during pregnancy, and likely have a bidirectional relationship, such that pregnant women who experience poor sleep quality may be more vulnerable to low mood and vice versa [60]. This bidirectional relationship between depressive symptomatology and sleep health is reflected in both our assessment of perinatal depressive symptoms [47], and in our diagnostic classification system [31]. While we aimed to eliminate statistical overlap by removing an item from our assessment of prenatal depressive symptomatology that taps prenatal sleep quality, these theoretical constructs may not be entirely separable.

Greater depressive symptomatology during pregnancy may reflect greater exposure to discrimination throughout one’s lifetime. In the current study, bivariate correlations between women’s exposure to racial/ethnic discrimination and their experiences of gendered racial stress and sleep quality ranged from small to moderate. Previous research indicates that Black American women who have experienced greater racial/ethnic discrimination report poorer psychological functioning during pregnancy [61]. Similarly, using a multi-ethnic sample of women (38% identified as Black American), Earnshaw and colleagues demonstrated that women’s reports of greater lifetime exposure to racial/ethnic discrimination were associated with greater prenatal depressive symptoms [62]. Prenatal depressive symptomatology does not occur within a vacuum: it may be associated with comorbid psychopathology, such as posttraumatic stress, and it may reflect the broader social context within which women exist. With respect to posttraumatic stress, we were unable to assess for the contribution of women’s posttraumatic stress symptomatology, despite the hypothesis that higher levels of discrimination are associated with poorer sleep quality via cognitive and/or physiological hyperarousal [7]. While women’s concurrent depressive symptoms accounted for a greater proportion of the variance in prenatal sleep quality in the current study sample, it is noteworthy that women’s lifetime experiences of racial/ethnic discrimination and gendered racial stress – potentially experienced prior to the conception of their children – largely remained small, but statistically significant predictors of sleep quality during pregnancy. While our questionnaire-based methods did not collect information on the timing of exposure, racial/ethnic and gendered racial/ethnic discrimination are particularly insidious because of their continuous and repetitive nature [63]. Our findings highlight the importance of deliberately attending to the lived experiences of racial/ethnic minority women: even in the presence of prenatal depressive symptoms, the stress associated with the lived experience of being a Black woman in America may confer additional risk for poor prenatal sleep quality, particularly during early pregnancy.

Associations between racial/ethnic discrimination/gendered racial stress and sleep quality were stronger during early versus mid-pregnancy. The attenuation of effect observed in mid-pregnancy likely stems from both conceptual and methodological factors. Conceptually, sleep quality worsens during the transition from the second to third trimester [52,64]. Mindell and colleagues collected data on women’s sleep quality across all nine months of gestation, and found that women increasingly attributed their sleep difficulties to pregnancy-specific symptoms (e.g., frequency of nocturnal urination, hip/pelvic pain, fetal movement, contractions, uncomfortable sleep position) as pregnancy progressed [65]. Changes in body morphology may contribute to the increased wakefulness after sleep onset and lighter, more fragmented sleep observed during the second and third trimesters [66]. In essence, normative physiological factors may explain more of the variance in poor sleep quality as gestational age increases, decreasing the likelihood that distally measured psychological stressors (e.g., exposure to discrimination) will explain a significant proportion of the variance in women’s sleep quality. In addition, proximal psychological stressors (e.g., everyday versus lifetime racial/ethnic discrimination/gendered racial stress, such as exposure to microaggressions) may be better predictors of both early and mid-pregnancy sleep quality. We did not collect information on either pregnancy-specific contributors to sleep or everyday discrimination in the current study and cannot test these hypotheses empirically. Methodologically, participants retrospectively reported on their lifetime exposure to racial/ethnic discrimination and gendered racial stress only once, during early pregnancy. These measures were not repeated in mid-pregnancy, whereas sleep quality information was collected at both study timepoints. Greater effect sizes in early versus mid-pregnancy may also reflect shared method variance. We encourage future researchers to collect information on discrimination and related stress at multiple timepoints throughout pregnancy, and to use measures that tap both lifetime and current or everyday exposure to discrimination.

In the current study, almost 40% of women reported never being exposed to racial/ethnic discrimination on the EOD, a pattern of findings that mirrors prior work. Slaughter-Acey and colleagues examined pregnant Black American women’s impressions of personal and group-directed racism. In their study, 42.3% of women denied ever being personally exposed to racism whereas only 7.5% of women indicated that racism did not negatively affect the lives of other Black Americans [67]. Our measures of racial/ethnic discrimination and gendered racial stress reflect women’s personal exposure and may have resulted in underestimates of study participants’ true exposure to racial/ethnic discrimination and related stress. At the same time, underreport of exposure to these stressors may be informative in and of itself. There is evidence suggesting that Black American women who underreport or deny exposure to racial/ethnic discrimination and gendered racial stress are at greater risk of poor health outcomes than those who report greater levels of exposure to these stressors [68]. As sleep health is a precursor to cardiometabolic health outcomes such as blood pressure [69], future studies of discrimination experiences and structural racism impacts on sleep in pregnancy are needed to understand the full extent of these associations.

We did not find evidence to support our proposed interaction model but encourage future researchers to consider different moderators in the association between racial/ethnic discrimination/gendered racial stress and prenatal sleep quality, such as neighborhood diversity and quality. When the ethnic density (i.e., proportion of individuals with racial/ethnic minority status) in an area increases, mental health problems may decrease as a function of community support [70]. Greater social support has been found to facilitate stress recovery and prevent or diminish development of mental health problems as a response to life stress [71]. However, increase in racial/ethnic minority volume is also associated with lower economic resources, so communities high in ethnic density may also lack the capital to protect members against negative mental health outcomes [71]. This may contribute to why SES did not have a moderating effect on sleep quality in our sample. These competing forces may have statistically cancelled each other out, thus nullifying any potential interaction effects. In the current study sample, women who reported fewer experiences of discrimination on the EOD had lower SES (Table 2). This finding potentially supports the idea of social support as a buffering effect for women in our sample who may be more likely to work and/or live in ethnically dense environments [72]. Context matters, and while we did not find evidence of an interaction effect, SES-related factors may still differentially affect the association between racial/ethnic discrimination/gendered racial stress and sleep quality, as Black American adults at different levels of SES are differentially exposed to environmental risk and protective factors. Future studies that can better parse exposures related to SES are needed to inform sleep health interventions.

Findings from the current study suggest that efforts to decrease institutional and interpersonal racial/ethnic discrimination would benefit the sleep quality of pregnant Black American women, particularly during early pregnancy. Racism is a complex system that is difficult to dismantle [63], so clinicians should engage in multiple antiracist actions at multiple levels of influence [73]. In addition to completing trainings on unconscious biases, clinicians may consider screening for women’s experiences of racial/ethnic discrimination and gendered racial stress during prenatal clinic visits. These screening efforts may identify pregnant women who are most at-risk for poor prenatal sleep quality and potential consequences (e.g., the development of a sleep disorder, preterm birth, etc.). More broadly, clinicians and sleep researchers can actively support programs and policies, through both education and advocacy, to address the determinants of sleep health disparities and mitigate these effects, especially among disadvantaged populations [73]. Relatedly, clinicians should intentionally name the detrimental effects of racism as opposed to race when informing pregnant Black American women of risks to their health during pregnancy [74]. Clinicians may also directly intervene on pregnant women’s sleep quality, with particular attention to Black American women who report heightened comorbid depressive symptomatology.

The current study has several limitations. First, the measures we selected asked women to consider their lifetime exposure to racial/ethnic discrimination and gendered racial stress. We were unable to obtain information on the timing of exposure, so we cannot state whether and how often these stressors occurred pre-conception or during women’s pregnancies. Given evidence that daytime incidences of racial/ethnic discrimination are associated with poorer sleep health that same evening [75], future research on discrimination and sleep health in pregnant women may consider taking a more fine-grained approach to better understand how daily stressors (e.g., microaggressions) are associated with sleep. Similarly, we were only able to consider Black American women’s race/ethnicity and gender, and how stress associated with these social identifiers may bear on prenatal sleep quality. We encourage future researchers to consider other aspects of women’s identity, including, but not limited to, their sexual identity and their nativity status (of note, our cohort is comprised of American-born women). Second, while the PROMIS has been used in samples of pregnant and postpartum women [17,46], it has not been validated in this population. However, in the current study sample, there was adequate variability in the distribution of sleep quality scores in both early and mid-pregnancy suggesting that poor sleep quality was not merely associated with physiological hallmarks of pregnancy that often prohibit restorative sleep. An additional limitation is that the current study sample was comprised of women who accessed prenatal medical care during early and mid-pregnancy at urban medical centers. Importantly, pregnant Black American women who report greater lifetime exposure to racial/ethnic discrimination are more likely to delay prenatal care to the second or third trimester of pregnancy [76]; eligibility criteria for the current study would not have allowed us to understand the experiences of these women. Finally, data on our predictor and outcome variables were obtained using self-report, rather than using objective measures of exposure to discrimination or sleep quality. With respect to discrimination, we were interested in women’s perceptions of their exposure to interpersonally inflicted racism and gendered racism and its associations with their prenatal sleep quality. We aimed to add to the extant literature by applying a framework that has been used in previous studies [24,62,77] that examine Black American women’s exposure to discrimination and its effects on prenatal health outcomes. Regarding our outcome variable, we examined women’s subjective sleep quality, consistent with the only known study to date that we sought to replicate [24]. The data from the current study constitute a secondary analysis of a larger study on prenatal health, and the focus of the larger study did not warrant inclusion of objective measures of prenatal sleep health, assessment of sleep disorders, or of women’s sleep health prior to pregnancy. Sleep health is a multidimensional construct, and also includes sleep duration and sleep architecture, both of which may be better measured via objective measures such as actigraphy and polysomnography [69]. However, sleep health also includes individuals’ satisfaction with their sleep, and this can only be measured subjectively. Measures of sleep quality may better reflect pregnant women’s clinical concerns [78], and subjective reports of sleep are associated with important health outcomes [79]. Nonetheless, future studies on discrimination and prenatal sleep health should still make an effort to include both objective and subjective measures of sleep health. While data on sleep disorders were not collected, they may be relevant to this analysis and future studies examining discrimination and sleep during pregnancy may consider including sleep disorder data when possible.

Despite these limitations, the current study has several strengths. We replicated previous work demonstrating direct effects between racial/ethnic discrimination and sleep quality during early pregnancy [24], and extended these findings to mid-pregnancy. In addition to assessing women’s lifetime exposure to racial/ethnic discrimination, we also included a measure of gendered racial stress as a predictor variable, to better capture the contextual stress associated with being both Black and female in America. Finally, we had limited missing data, but nevertheless used multiple imputation to maximize sample size in analyses. In sum, among pregnant Black American women in the current study sample, greater lifetime exposure to racial/ethnic discrimination and gendered racial stress was associated with poorer sleep quality during early, and in some cases, mid-pregnancy. Given the importance of optimal sleep quality during pregnancy for healthy birth outcomes [15] and maternal psychological wellbeing [80], future clinical research should be devoted to mitigating sleep health disparities among pregnant Black American women [50]. Efforts to dismantle racial/ethnic and gender-based discrimination may decrease both sleep health disparities during pregnancy and adverse maternal child health outcomes that disproportionately affect Black American women and their infants.

Data availability statement

Participants of the current study did not consent for their data to be shared with journal outlets, so supporting data is not available.