Religiosity and Spirituality as Stressors or Stress Relievers in relation to Sleep Health among African American Women

Abstract

Study Objectives:

To investigate associations between religiosity/spirituality (R/S) and sleep and potential modification by stress among Black/African American (AA) women.

Methods:

Participants from the Study of Environment, Lifestyle, & Fibroids (2010–2018) reported importance of faith and R/S as a source of strength/comfort (ranging from ‘very’ to ‘not at all’) as well as frequency of prayer/meditation (ranging from ‘everyday,’ to ‘<once/month’) at baseline. There were repeated measures of daily stress (‘very high/moderate’ vs. ‘low/none’), short sleep duration (<7-hours), nonrestorative sleep (NRS, waking rested <4-days/week), and insomnia symptoms (difficulty falling/staying asleep). Adjusted prevalence ratios (aPRs) were estimated with Poisson regression with robust variance, and adjusted risk ratios were estimated by applying generalized estimating equations to models. R/S-by-stress terms were tested.

Results:

Among 1,693 Black/AA women (mean age±SD 29.2±3.4 years), 69.3% reported faith as important, 55.6% perceived R/S as a source of strength/comfort, 58.8% prayed/meditated daily, 43.8% reported high/moderate stress, and poor sleep was prevalent (short sleep:58.4%, NRS:62%, and insomnia symptoms:17.9%) at baseline. There were no associations with short sleep and insomnia symptoms. All three R/S measures were each cross-sectionally but not longitudinally associated with lower likelihood of NRS, however, associations differed by stress. R/S as source of strength/comfort and prayer/meditation showed protective associations with NRS prevalence at baseline only among participants with low versus high stress (e.g., prayer everyday vs. <once/month aPR_{high stress}=1.07 [95% CI:0.91–1.26], aPR_{low stress}=0.76 [95% CI:0.63–0.91], p_R/S*stress =0.03).

Conclusions:

Higher R/S was associated with lower NRS prevalence, but protective associations were limited to participants with low/no stress.

Graphical Abstract

INTRODUCTION

Poor sleep is of great public health concern across the globe.¹ In the United States (US), variations in sleep health exist across racial, ethnic, and socioeconomic groups. For example, Black/African American (AA) adults are generally more likely than many other groups to experience short sleep duration (SSD), non-restorative sleep (NRS), and insomnia symptoms (IS).^2,3 These sleep disturbances can influence a broad array of health conditions over the lifespan by contributing to adverse health outcomes such as obesity, type 2 diabetes, cardiovascular disease, and mental health conditions such as depression.^4–6

Multiple social and environmental stressors contribute to these preventable differences in sleep, including economic hardship, neighborhood, and workplace conditions.⁷ Within this broader context, Black/AA women face a distinct and compounded set of challenges due to the intersection of race, sex, and socioeconomic position.⁸ Stressors such as financial instability that contribute to housing and food insecurity, differential treatment in the workplace, caregiving responsibilities, socialization to be strong and invulnerable, and insufficient social support may heighten the social vulnerability of Black women.^9–11 This cumulative and toxic stress, often referred to as “weathering,” exacerbates chronic conditions such as obesity, hypertension, and type 2 diabetes, which are bidirectionally associated with sleep disturbances.^12–15

Religion and spirituality (R/S) may function as a coping mechanism in response to such stressors and is also a well-known independent determinant of health although evidence remains sparse for sleep health.^16–19 According to the Religious Landscape Study conducted by the Pew Research Center, 80% of Black women have described religion as “very important” compared to 53% of the general U.S. population.²⁰ Involving practices such as prayer, meditation, and church attendance, R/S has long been recognized as a culturally salient coping mechanism that fosters psychological resilience and provides social, emotional, informational, and financial support during adverse conditions.^19,21 Beyond its stress-buffering functions, R/S may directly shape health-related behaviors, regulate physiological functioning, and foster identity and community norms that influence both mental and physical health.¹⁹ For Black/AA women, R/S is often deeply intertwined with cultural traditions, gendered expectations, and the historical role of the Black church.^22–25 Religious involvement may encompass active participation in women’s ministries, leadership in faith-based community outreach, and serving as a spiritual anchor for one’s family. The Black church has historically served not only as a religious institution but also as a center for social activism, civic engagement, and health promotion, creating pathways through which R/S influences both individual and community wellbeing.^22–26 These intersecting cultural, social, and gendered roles may enhance the protective effects of R/S through strengthened social support networks, shared identity, and empowerment, but also may introduce stress when religious obligations compete with work, caregiving, and other life demands.²⁷

Perceptions of divine control and religious coping are associated with markers of resilience (i.e., higher adrenal hormone DHEAS and lower cortisol levels) among Black women.²⁸ These physiological responses can directly influence sleep health by reducing hyperarousal and promoting relaxation. Such pathways may operate independently of external stress exposure by fostering routine, meaning-making, social cohesion, and health-promoting behaviors (e.g., avoiding substances, maintaining structure), in addition to reducing psychological distress and lowering allostatic load.^17,19 Moreover, religious involvement and coping have been associated with better mental and cardiovascular health among Black/AA adults.^29,30 A review of seven studies conducted by Hill et al. also found that practices such as prayer, meditation, and participation in faith-based communities have been associated with favorable sleep outcomes, including improved sleep quality, reduced sleep disturbances, and acquiring the recommended sleep duration.¹⁷

Nonetheless, the relationship between R/S and sleep may be complex and not universally positive.¹⁸ While R/S may act as a buffer against stress to promote restorative sleep, it may also introduce additional stress under certain conditions. For example, religious obligations or commitments may conflict with other responsibilities (e.g., work and family life), thereby creating psychological stress and disrupting sleep routines.³¹ Beyond time demands, normative expectations and fear of social sanction within faith communities (i.e., semi-involuntary participation) can create social-evaluative stress that also impairs sleep.³² This may be especially true for women in general and Black women, in particular, due to the historical and ongoing centrality of the church in African American communities, cultural expectations to serve as spiritual anchors for family and congregation, and the intersection of racialized and gendered stressors.^23,33 Individuals experiencing excessive life stress or major life challenges may respond by increasing their engagement in R/S activities, such as attending services more frequently, engaging in extended prayer, or participating in religious retreats, which can provide comfort and community support but may also place additional demands on time and energy that exacerbate stress and interfere with sleep.³⁴ Similarly, individuals experiencing religious struggles/trauma, such as guilt, doubt, or judgement, may find that these experiences exacerbate stress and negatively affect sleep health.^35–38 This bidirectionality highlights the need to consider the protective and potentially adverse effects of R/S on sleep health.

Important gaps persist in the literature. Much of the existing work is cross-sectional, limiting conclusions about directionality or causality, and often relies on narrow or unidimensional measures of religiosity (e.g., attendance only) and single-item or unvalidated sleep assessments.^17,19,28,39 Although some studies provided biological plausibility by linking R/S to stress-related biomarkers,²⁸ and others to mental and self-rated health,^29,30 few explicitly examined sleep health as a primary outcome or consider the interactive effects of R/S and psychosocial stress.^17,19,28 Studies reviewed by Hill et al., had several limitations (e.g., focus on middle-aged or older adults) and none included Black women.¹⁷ Recent systematic reviews further underscore both the relevance and the remaining gaps in this area. A meta-analysis of longitudinal and intervention studies found that spiritual wellbeing was modestly protective against depressive symptoms in youth, whereas negative religious coping (i.e., feeling abandoned by or blaming God) was associated with worse mental health over time, highlighting the multidimensional and context-dependent nature of R/S effects.⁴⁰ Similarly, another systematic review examining R/S and health-related behaviors reported generally positive associations with physical and psychosocial health outcomes but emphasized heterogeneity in measurement, limited longitudinal designs, and the need to clarify underlying mechanisms.⁴¹ Collectively, these gaps underscore the need for rigorous, multidimensional, and population-specific research to better understand how R/S influences sleep health, particularly in the context of chronic stress among Black/AA women, given that few recent reviews have focused specifically on sleep, examined stress as a potential modifier, or included young Black women.

This study builds on existing literature by examining stress as a potential modifier of the relationship between R/S and sleep health outcomes, including SSD, NRS, and IS, among Black/AA women (see Figure 1 for schematic illustration). We hypothesized that 1) higher vs. lower R/S is associated with lower prevalence and risk of SSD, NRS, and IS; and 2) daily life stress modifies the relationship between R/S and sleep, such that higher R/S is associated with lower prevalence and risk of SSD, NRS, and IS among individuals with moderate/high stress, but with weaker or no associations among those with low or no stress.

Figure 1. Study schematic illustrating the hypothesized associations examined between multidimensional religiosity/spirituality exposures and sleep health outcomes with daily life stress as an effect modifier, and potential sociodemographic and health-related confounders.

METHODS

Data source: Study of Environmental, Lifestyle & Fibroids

The Study of Environmental, Lifestyle & Fibroids (SELF) is a prospective cohort study designed to identify risk factors for uterine fibroid incidence and growth among Black/AA women. Enrolled participants were self-identified Black/AA (N = 1,693) women between the ages of 23–35 years who lived in Detroit, Michigan, and had no prior diagnosis of fibroids. Women were excluded from the SELF if they had a prior diagnosis of uterine fibroids, hysterectomy, cancer requiring radiation or chemotherapy, or autoimmune conditions requiring medication. The SELF protocol and a detailed list of exclusion and inclusion criteria have been published previously.⁴² Briefly, at baseline (2010–2012), participants completed a combination of computer-assisted telephone interviews (CATI) and computer-assisted web interviews (CAWI), along with a clinic visit. Measures of sleep and R/S were collected via CAWI which was sent via an online link that could be accessed at any time. The CAWI took approximately 45 minutes to complete. Follow-up assessments occurred at approximately 20-month intervals (range: 18–26 months) and included repeat CATI and CAWI questionnaires using the same standardized procedures as baseline. Study staff attempted to contact participants for approximately 8 months starting at 18 months following the previous visit. Attempts were made via email and phone calls until the visit window closed (after month 25, 45, 65). Participants who missed a given follow-up were eligible to participate in subsequent follow-ups. Retention in the SELF cohort has been high, with approximately 88% of participants completing the first follow-up assessment, 86% completing the second follow-up, and 91% completing the third follow-up. All participants provided written informed consent, and the SELF protocol was approved by the Institutional Review Board of the National Institute of the Environmental Health Sciences and Henry Ford Health System.

Study population

To be eligible for the current study, participants needed to have complete data on R/S (collected at baseline), and at least one sleep dimension (i.e., sleep duration, sleep quality, or insomnia symptoms). For the cross-sectional analyses, which examined baseline R/S in relation to baseline sleep, the analytic sample included all 1,693 Black/AA women with complete data on R/S and baseline sleep measures. For the longitudinal analyses, which examined baseline R/S in relation to repeated measures of sleep across baseline and three follow-up periods, the analytic sample also included all 1,693 Black/AA women; all available person-visit observations were used (6,062 for short sleep duration and 6,128 for non-restorative sleep and insomnia symptoms). Due to the small sample size at baseline (n=19), long sleep duration (>9 hours) was excluded from the analysis where sleep duration was the outcome.

Exposure assessment: Religiosity and spirituality (R/S)

At baseline, participants self-reported importance of faith by responding to the question “How important is your religious faith or spirituality to you?” The responses were categorized as “not at all important,” “somewhat important”, “moderately important”, or “very important,” with “not at all important” serving as the reference category. Participants self-reported R/S as a source of strength and comfort by responding to the following question: “How much is religion or spirituality a source of strength and comfort to you?”. The responses were categorized as “not at all”, “somewhat”, “quite a bit”, or “very much,” with “not at all” serving as the reference category. Prayer frequency was defined by participant’s response to “How often do you pray or meditate?” The responses were categorized as “never/less than once a month” vs. “monthly or few days per month”, “weekly or a few days per week”, “everyday or several times per day,” with “never/less than once a month” serving as the reference category. Internal consistency across the three R/S items was high (Cronbach’s α = 0.83).

Outcome assessment: Self-reported sleep duration, nonrestorative sleep (NRS), and insomnia symptoms (IS)

Participants provided self-reported data on sleep duration, NRS, and IS at baseline and during each follow-up assessment. Each measure was operationalized by categorizing responses in alignment with prior literature.^15,43 Sleep duration was assessed by asking, “How many hours of sleep do you usually get before a workday?” Based on the National Sleep Foundation categories, average sleep duration per day was defined as either short (<7 hours) or recommended (7–9 hours).⁴⁴ NRS was measured by the frequency of waking up feeling rested, based on responses to the question, “How many days per week do you wake up feeling well rested?” Responses were dichotomized (<4 days/week vs. ≥4 days/week).^15,43 IS was assessed by asking, “About how many days per month do you have trouble falling asleep or going back to sleep after waking up?” Responses were dichotomized to indicate the presence of IS (≥10–14 days/month vs. 0–9 days/month).^15,43

Potential confounders

Potential confounding variables included sociodemographic and clinical characteristics that were selected a priori based on the literature.^45,46 Sociodemographic variables were collected at baseline as well as each follow-up and included age, marital status, educational attainment, employment, and annual household income. Clinical characteristics included body mass index (BMI) (assessed with measured height and weight, kg/m²) and self-reported general health status at baseline and the three follow-ups.

Potential modifier

Daily life stress was assessed at baseline and during the three follow-ups as a potential modifier by asking, “How stressful is your day-to-day life?” Responses included “very stressful”, “moderately stressful,” “mildly stressful,” and “not at all stressful”, which were then dichotomized as moderately/very stressful (high) vs. not at all/mildly stressful (low) to distinguish higher versus lower perceived stress and to support stable and interpretable estimation of modification by ensuring adequate cell sizes across stress strata.

Statistical analysis

We computed descriptive statistics to summarize baseline sociodemographic characteristics, as well as health behaviors and characteristics, including sleep dimensions and daily stress. Age (treated as a continuous variable) which was summarized using mean and standard deviation; all other variables were categorical and summarized using frequencies and percentages. We compared baseline characteristics across categories of R/S dimensions, including importance of faith, R/S as a source of comfort/strength, and frequency of prayer/meditation. Statistical significance was assessed using one-way analysis of variance (ANOVA) for continuous variables and chi-squared tests for categorical variables, with p-values <0.05 considered statistically significant. Spearman correlation coefficients were also calculated at baseline to assess the correlations among R/S indicators and among sleep dimensions and to evaluate the extent to which these measures reflected overlapping versus distinct constructs (Supplemental Table S1). We conducted cross-sectional analyses to estimate prevalence ratios (PRs) and 95% confidence intervals (CIs) for the associations between baseline R/S (i.e., importance of faith, R/S as a source of comfort and strength, and prayer/meditation frequency) and baseline sleep dimensions, including sleep duration, NRS, and IS using Poisson regression with robust variance. For longitudinal analyses, we applied generalized estimating equations (GEE) to log-binomial regression models, estimating risk ratios (RRs) and 95% CIs for associations between baseline R/S and repeated measures of sleep duration, NRS, and IS collected at baseline and across follow-up visits (i.e., baseline through the third follow-up). These models leveraged all available person-visit observations and did not specifically examine changes from baseline, but rather estimated average associations between baseline R/S and sleep outcomes across the entire study period. All models were adjusted for baseline age, marital status, educational attainment, employment status, household income, BMI, and overall health status; visit-specific values were used in longitudinal analyses and baseline values in cross-sectional analyses. Statistical significance was defined as a two-sided p-value <0.05. Effect modification by daily stress (high vs. low/no) was tested by including cross-product terms between each R/S measure and the visit-specific stress measure (i.e., stress assessed at the same visit as the sleep outcome) in the longitudinal models and baseline stress in the cross-sectional models. All analyses were conducted using IBM SPSS version 29.

RESULTS

Study population characteristics

Characteristics of the participants at baseline stratified by each R/S domain are presented in Tables 1–3. Among the 1,693 participants (mean age 29.2 ± 3.4 years), 45.7% reported annual household incomes below $20,000, and R/S were commonly reported: 69.3% considered faith very important, 55.6% viewed R/S as a major source of strength/comfort, and 58.8% reported praying/meditating daily. At baseline, 58.4% reported SSD, 62.0% NRS, 17.9% IS, and 43.8% high/moderate daily life stress.

Table 1.

Baseline characteristics of participants by dimensions of importance of faith, Study of Environment, Lifestyle and Fibroids, N = 1693

Sociodemographic characteristics	Importance of Faith
	Not at all N = 58 (3.4%)	Somewhat important N = 181 (10.7%)	Moderately important N = 280 (16.5%)	Very important n = 1174 (69.3%)	Total n = 1693 (100.0%)	p-value
Age, years (mean ± SD)	28.4 ± 3.5	29.1 ± 3.4	28.3 ± 3.4	29.5 ± 3.4	29.2 ± 3.4	<.001
Educational attainment, n (%)
≤High school	23 (39.7)	62 (34.3)	73 (26.1)	212 (18.1)	370 (21.9)	<0.001
Some college or Associate’s/Technical degree	22 (37.9)	88 (48.6)	138 (49.3)	600 (51.1)	848 (50.1)
≥Bachelor’s degree	13 (22.4)	31 (17.1)	69 (24.6)	362 (30.8)	475 (28.1)
Employment status, n (%)
Not employed	28 (48.3)	88 (48.6)	107 (38.2)	417 (35.6)	640 (37.9)	0.025
Employed part-time	6 (10.3)	18 (9.9)	32 (11.4)	154 (13.2)	210 (12.4)
Employed full-time	24 (41.4)	75 (41.4)	141 (50.4)	600 (51.5)	840 (49.6)
Annual household income, n (%)
<$20,000	36 (62.1)	102 (56.4)	130 (46.4)	.505 (43.0)	773 (45.7)	0.002
$20,000 to $50,000	15 (25.9)	52 (28.7)	112 (40.0)	451 (38.4)	630 (37.2)
>$50,000	7 (12.1)	27 (14.9)	38 (13.6)	218 (18.6)	290 (17.1)
Marital status, n (%)
Never married	48 (82.8)	114 (63.0)	169 (60.4)	661 (56.3)	992 (58.6)	<0.001
Previously married or living as married	6 (10.3)	14 (7.7)	35 (12.5)	181 (15.4)	236 (13.9)
Currently married or living as married	4 (6.9)	53 (29.3)	76 (27.1)	332 (28.3)	465 (27.5)
Health behaviors and characteristics
Smoking status, n (%)
Never	38 (65.5)	114 (63.0)	198 (70.7)	895 (76.2)	1245 (73.5)	<0.001
Former	1 (1.7)	19 (10.5)	16 (5.7)	89 (7.6)	125 (7.4)
Current	19 (32.8)	48 (26.5)	66 (23.6)	190 (16.2)	323 (19.1)
Alcohol use, n (%)
None	17 (29.3)	37 (20.4)	48 (17.1)	399 (34.0)	501 (29.6)	<0.001
Moderate	29 (50.0)	94 (51.9)	158 (56.4)	580 (49.4)	861 (50.9)
Heavy	12 (20.7)	50 (27.6)	74 (26.4)	195 (16.6)	331 (19.6)
BMI (kg/m2), n (%)
Underweight/normal (<25)	13 (22.4)	32 (17.7)	46 (16.4)	244 (20.8)	335 (19.8)	0.463
Overweight (25 - <30)	8 (13.8)	36 (19.9)	58 (20.7)	248 (21.1)	350 (20.7)
Obese (≥30)	37 (63.8)	113 (62.4)	176 (62.9)	682 (58.1)	1008 (59.5)
Sleep duration, n (%)
Short (<7 hours)	33 (56.9)	106 (58.6)	168 (60.0)	682 (58.1)	989 (58.4)	0.752
Recommended (7–9 hours)	25 (43.1)	71 (39.2)	110 (39.3)	479 (40.8)	685 (40.5)
Long (>9 hours)	0 (0.0)	4 (2.2)	2 (0.7)	13 (1.1)	19 (1.1)
Insomnia symptoms, n (%)
Yes	7 (12.1)	34 (18.8)	60 (21.4)	202 (17.2)	303 (17.9)	0.241
No	51 (87.9)	147 (81.2)	220 (78.6)	972 (82.8)	1390 (82.1)
Non-restorative sleep, n (%)
Yes	38 (65.5)	122 (67.4)	188 (67.1)	702 (59.8)	1050 (62.0)	0.044
No	20 (34.5)	59 (32.6)	92 (32.9)	472 (40.2)	643 (38.0)
General health, n (%)
Poor/Fair	13 (22.4)	32 (17.7)	58 (20.7)	178 (15.2)	281 (16.6)	0.083
Good	18 (31.0)	78 (43.1)	106 (37.9)	440 (37.5)	642 (37.9)
Very good/Excellent	27 (46.6)	71 (39.2)	116 (41.4)	556 (47.4)	770 (45.5)
Daily life stress, n (%)
High	25 (43.1)	87 (48.1)	127 (45.4)	502 (42.8)	741 (43.8)	0.544
Low/No	33 (56.9)	94 (51.9)	153 (54.6)	672 (57.2)	952 (56.2)

Note. Values are mean ± SD or n (%). p-values were obtained using one-way ANOVA for Age and Pearson’s χ² tests for all other (categorical) variables; tests were two-sided with α=0.05. Missing values were excluded pairwise. Moderate alcohol use = 1–5 drinks on days when having alcohol; Heavy alcohol use = 6+ drinks on days when having alcohol

Missing data for employment: n = 3

Table 3.

Baseline characteristics of participants by dimensions of prayer/meditation frequency, Study of Environment, Lifestyle and Fibroids, N = 1693

Sociodemographic characteristics	Frequency of prayer/meditation
	Never/<once a month n = 156 (9.2%)	≤Once per month n = 187 (11.0%)	Every week n = 355 (21.0%)	Everyday n = 995 (58.8%)	Total n = 1693 (100%)	p-value
Age, years (mean ± SD)	28.79 ± 3.58	28.77 ± 3.42	29.08 ± 3.29	29.42 ± 3.47	29.22 ± 3.44	0.021
Educational attainment, n (%)
≤High school	64 (41.0)	44 (23.5)	57 (16.1)	205 (20.6)	370 (21.9)	<0.001
Some college or Associate’s/Technical degree	62 (39.7)	97 (51.9)	186 (52.4)	503 (50.6)	848 (50.1)
≥Bachelor’s degree	30 (19.2)	46 (24.6)	112 (31.5)	287 (28.8)	475 (28.1)
Employment status, n (%)
Not employed	78 (50.0)	76 (40.9)	121 (34.1)	365 (36.8)	640 (37.9)	0.012
Employed part-time	15 (9.6)	15 (8.1)	47 (13.2)	133 (13.4)	210 (12.4)
Employed full-time	63 (40.4)	95 (51.1)	187 (52.7)	495 (49.8)	840 (49.7)
Annual household income, n (%)
<$20,000	91 (58.3)	88 (47.1)	144 (40.6)	450 (45.2)	773 (45.7)	0.002
$20,000 to $50,000	39 (25.0)	78 (41.7)	147 (41.4)	366 (36.8)	630 (37.2)
>$50,000	26 (16.7)	21 (11.2)	64 (18.0)	179 (18.0)	290 (17.1)
Marital status, n (%)
Never married	116 (74.4)	118 (63.1)	208 (58.6)	550 (55.3)	992 (58.6)	<0.001
Previously married	14 (9.0)	22 (11.8)	45 (12.7)	155 (15.6)	236 (13.9)
Currently married	26 (16.7)	47 (25.1)	102 (28.7)	290 (29.1)	465 (27.5)
Health behaviors and characteristics
Smoking status, n (%)
Never	104 (66.7)	138 (73.8)	262 (73.8)	741 (74.5)	1245 (73.5)	0.424
Former	11 (7.1)	15 (8.0)	26 (7.3)	73 (7.3)	125 (7.4)
Current	41 (26.3)	34 (18.2)	67 (18.9)	181 (18.2)	323 (19.1)
Alcohol use, n (%)
None	41 (26.3)	42 (22.5)	76 (21.4)	342 (34.4)	501 (29.6)	<0.001
Moderate	73 (46.8)	99 (52.9)	202 (56.9)	487 (48.9)	861 (50.9)
Heavy	42 (26.9)	46 (24.6)	77 (21.7)	166 (16.7)	331 (19.6)
BMI (kg/m2), n (%)
Underweight/Normal(<25)	35 (22.4)	38 (20.3)	60 (16.9)	202 (20.3)	335 (19.8)	0.158
Overweight (25 - <30)	21 (13.5)	45 (24.1)	72 (20.3)	212 (21.3)	350 (20.7)
Obese (≥30)	100 (64.1)	104 (55.6)	223 (62.8)	581 (58.4)	1008 (59.5)
Sleep duration, n (%)
Short (<7 hours)	87 (55.8%)	105 (56.1%)	214 (60.3%)	583 (58.6%)	989 (58.4%)	0.739
Recommended (7–9 hours)	68 (43.6%)	78 (41.7%)	138 (38.9%)	401 (40.3%)	685 (40.5%)
Long (>9 hours)	1 (0.6%)	4 (2.1%)	3 (0.8%)	11 (1.1%)	19 (1.1%)
Insomnia symptoms, n (%)
Yes	19 (12.2%)	35 (18.7%)	58 (16.3%)	191 (19.2%)	303 (17.9%)	0.152
No	137 (87.8%)	152 (81.3%)	297 (83.7%)	804 (80.8%)	1390 (82.1%)
Non-restorative sleep, n (%)
Yes	105 (67.3)	114 (61.0)	233 (65.6)	598 (60.1)	1050 (62.0)	0.141
No	51 (32.7)	73 (39.0)	122 (34.4)	397 (39.9)	643 (38.0)
General health, n (%)
Poor/Fair	34 (21.8)	29 (15.5)	57 (16.1)	161 (16.2)	281 (16.6)	0.240
Good	62 (39.7)	77 (41.2) ^	143 (40.3)	360 (36.2)	642 (37.9)
Very good/Excellent	60 (38.5)	80 (43.3)	155 (43.7)	474 (47.6)	770 (45.5)
Daily life stress, n (%)
High	68 (43.6)	78 (41.7)	166 (46.8)	429 (43.1)	741 (43.8)	0.618
Low/No	88 (56.4)	109 (58.3)	189 (53.2)	566 (56.9)	952 (56.2)

Note. Values are mean ± SD or n (%). p-values were obtained using one-way ANOVA for Age and Pearson’s χ² tests for all other (categorical) variables; tests were two-sided with α=0.05. Missing values were excluded pairwise. Moderate alcohol use = 1–5 drinks on days when having alcohol; Heavy alcohol use = 6+ drinks on days on having alcohol

Missing data for employment: n = 3

Compared with women reporting lower R/S, those reporting higher R/S were generally older, had higher educational attainment and income, and were more likely to be employed and married (Tables 1–3). At baseline, the prevalence of NRS differed across categories of importance of faith (59.8%–67%, p = 0.04) and R/S as a source of strength/comfort (58.4%–72%, p < 0.001), whereas SSD and IS did not differ across R/S dimensions. The prevalence of high/moderate daily life stress at baseline ranged from 42% to 48% across R/S categories. Across all study visits, participants with high/moderate stress were more likely to report SSD, NRS, and IS compared to those with lower stress (Supplemental Table S2).

Spearman correlations among baseline R/S and sleep dimensions are presented in Supplemental Table S1; R/S dimensions were moderately to strongly correlated with one another (r_range: 0.49–0.73). Correlations between sleep dimensions were weak to moderate (r_range: 0.17–0.32). Correlations between R/S and sleep dimensions were weak (r_range: −0.09 to 0.04), with statistically significant inverse associations observed for NRS with importance of faith (r: −.07) and R/S as a source of strength/comfort (r: −0.09). To characterize the stability of sleep measures over time, intraclass correlation coefficients (ICCs) were calculated across repeated assessments. ICCs indicated low within-person stability, with values of 0.30 (95% CI: 0.27–0.32) for sleep duration, 0.32 (95% CI: 0.29–0.35) for NRS, and 0.39 (95% CI: 0.36–0.41) for insomnia symptoms. The ICC for daily stress was 0.35 (95% CI: 0.32–0.38), similarly indicating low within-person stability over time.

Cross-sectional associations between R/S and sleep

Figure 3 (Supplemental Table S3) presents the adjusted cross-sectional associations between baseline R/S and baseline sleep dimensions. None of the R/S indicators were significantly associated with SSD or Insomnia. Multiple R/S dimensions were significantly associated with lower prevalence of non-restorative sleep. A lower prevalence of NRS was observed for those reporting faith was “very important” compared to “not important” (adjusted prevalence ratios [aPR] = 0.77, 95% CI: 0.59–0.99, p = .04), those reporting that R/S was “very much” a source of strength/comfort compared to “not at all” (aPR = 0.75, 95% CI: 0.59–0.97, p = .03), and those who reported everyday prayer/meditation compared to those who never engaged in prayer/meditation (aPR = 0.77, 95% CI: 0.65–0.93, p = .01).

Figure 3. Adjusted prevalence ratios and 95% confidence intervals for cross-sectional associations between religiosity/spirituality and sleep at baseline.

Note. High, medium, and low exposure levels are defined separately for each religiosity/spirituality measure. For importance of faith, high = very important; medium = moderately important, low = somewhat important. For religion/spirituality is a source of strength/comfort, high = very much, medium = quite a bit, low = somewhat. For frequency of prayer/meditation, high = everyday, medium = every week, low = ≥once a month.

Longitudinal associations between R/S and sleep

Figure 4 (Supplemental Table S4) presents the adjusted longitudinal associations between baseline R/S and repeated measures of sleep across baseline and three follow-up assessments. None of the R/S indicators were statistically significantly associated with SSD, NRS, or IS over time.

Figure 4. Adjusted risk ratios and 95% confidence intervals for longitudinal associations between baseline religiosity/spirituality and repeated measures of sleep.

Note. High, medium, and low exposure levels are defined separately for each religiosity/spirituality measure. For importance of faith, high = very important; medium = moderately important, low = somewhat important. For religion/spirituality is a source of strength/comfort, high = very much, medium = quite a bit, low = somewhat. For frequency of prayer/meditation, high = everyday, medium = every week, low = ≥once a month.

Modification of the R/S-Sleep relationship by stress

Figure 5 (Supplemental Table S5) presents the results from cross-sectional models examining whether daily life stress modifies the associations between R/S and sleep dimensions at baseline. No statistically significant effect modification was observed by daily life stress on R/S in relation to SSD or IS. There was evidence of interaction between R/S as a source of comfort (p-value for interaction term = 0.02) and frequency of prayer/meditation (p-value for interaction term = 0.03). For all dimensions of R/S those with low stress and higher R/S had protective associations with NRS, while those with high stress had null or positive associations. For instance, among women reporting low/no stress, those reporting R/S as “very much” a source of strength/comfort had lower prevalence of NRS compared to those reporting “not at all” (aPR = 0.74, 95% CI: 0.58–0.95, p = 0.02), while among women reporting high stress the same contrast suggested an increased prevalence of NRS (aPR = 1.37, 95% CI: 0.95–1.98, p = 0.10). Similarly, among women reporting low/no stress, those praying/meditating everyday had lower prevalence of NRS compared to those who reported never praying (aPR = 0.76, 95% CI: 0.63–0.91, p < 0.01), while everyday prayer/meditation was not associated with NRS among women reporting high/moderate stress (aPR = 1.07, 95% CI: 0.91–1.26, p = 0.39).

Figure 5. Adjusted prevalence ratios and 95% confidence intervals for cross-sectional associations between religiosity/spirituality and sleep at baseline stratified by daily stress.

Note. P-values shown in each panel correspond to cross-product terms between each religiosity/spirituality measure and stress in the models.

No clear evidence of effect modification by daily life stress was observed in the longitudinal models Figure 6 (Supplemental Table S6).

Figure 6. Adjusted risk ratios and 95% confidence intervals for longitudinal associations between baseline religiosity/spirituality and repeated measures of sleep stratified by daily stress.

Note. P-values shown in each panel correspond to cross-product terms between each religiosity/spirituality measure and stress in the models.

DISCUSSION

This study investigated cross-sectional and longitudinal associations between multiple dimensions of R/S and sleep health among young Black/AA women, and potential modification of these associations by daily stress. Most women in the SELF study reported faith was very important to them, frequently engaged in prayer or meditation, and often perceived R/S as a source of strength and comfort. SSD, NRS, IS, and daily life stress were also common among women in the study. Our key findings indicate that the relationship between R/S and sleep was context-dependent and varied by both the dimension of R/S examined and the level of perceived stress. All R/S dimensions, including importance of faith, R/S as a source of strength/comfort, and frequency of prayer/meditation, were associated with lower prevalence of NRS overall. However, these associations differed by stress. Among women reporting low/no stress, women perceiving R/S as “very much” a source of strength/comfort and engaging in everyday prayer/meditation had approximately 24% and 26% lower prevalence of NRS (defined as waking up feeling rested <4 days/week, a frequency-based threshold consistent with clinically relevant sleep dissatisfaction) compared with women who did not consider R/S a source of strength or pray. However, among women reporting high stress, estimates for R/S as a source of strength/comfort and prayer/meditation were generally in the direction of higher prevalence of NRS, although most associations were not statistically significant. These findings were partially consistent with our hypotheses: while daily stress modified the relationship between R/S and sleep, the direction of association differed from expectations, with protective associations observed among women reporting low/no stress rather than among those with high/moderate stress. No significant associations were observed between R/S and sleep health over time in the longitudinal analysis.

The high prevalence of R/S observed in this cohort is generally consistent with prior national data. In the present study, 69.3% of participants reported that faith was very important in their lives, which is comparable, though somewhat lower, than estimates from national surveys such as the Pew Research Center, which report that approximately 80% of Black women describe religion as very important.²⁰ This difference likely reflects cohort effects, as the present study focused on young Black/AA women (ages 23–35 years) in one geographical area, whereas national estimates span the full adult age range. Prior research consistently shows that religiosity and institutional religious involvement increase with age, suggesting that the prevalence observed in this cohort is expected rather than indicative of substantively lower religious engagement given the younger age distribution.^47,48

The findings partially align with prior research suggesting that R/S can serve as a protective factor for sleep health, although not uniformly across stress levels. For example, studies have shown that religious coping is associated with lower psychological distress, improved cardiovascular profiles, and reduced allostatic load, which may support sleep regulation through stress-buffering mechanisms.^17,28–30 Faith-based practices and beliefs may also offer meaning, belonging, and tools for coping with adversity. Among Black/AA women, R/S is often deeply integrated into cultural and community life, and has been shown to promote psychological resilience and emotional regulation.^19,20,29 However, rather than uniformly buffering the effects of stress, as proposed by the stress-buffering hypothesis,⁴⁹ our findings suggest that the relationship between R/S and sleep may depend on the broader stress context. Specifically, the potentially buffering role of two dimensions of R/S on NRS was observed only among women reporting low/no stress, and these protective associations were not observed among women with high/moderate stress. Therefore, the current results do not reflect the classic stress-buffering pattern of R/S suggested in prior literature.⁵⁰

Collectively, these findings highlight that different dimensions of R/S may relate to sleep health through distinct pathways that vary by stress context. Notably, associations were observed only for non-restorative sleep and not for sleep duration or insomnia symptoms, suggesting that R/S may operate through psychosocial and emotional processes that influence perceived sleep restoration specifically, rather than broader sleep patterns or symptom profiles. Furthermore, perceiving R/S as a source of strength or comfort likely reflects an internalized psychosocial resource that supports emotional regulation and restorative sleep,⁵¹ particularly under conditions of lower stress, where these associations were observed. Similarly, engaging in everyday prayer or meditation may reflect routine, non-distress-driven R/S practices that are associated with better perceived sleep restoration under lower stress conditions,⁵² whereas under higher stress, competing demands or stress-related processes may attenuate these associations. Alternatively, private or internalized dimensions of R/S may be less effective under high stress, whereas social dimensions (e.g., religious support or cohesion) may be more relevant, warranting further investigation.

Contrary to our hypothesis, we did not observe significant longitudinal associations between R/S and any of the sleep dimensions. Several explanations may account for this null finding. First, R/S was measured only at baseline, and thus our analysis did not capture religious history or changes in religious or spiritual engagement and experience (e.g., struggle/trauma) over time. R/S is not a static construct and may fluctuate in response to life events, changing stress levels, or personal growth.⁵³ Second, it is possible that R/S exerts short-term or immediate effects on sleep that do not persist or accumulate over longer periods. Third, participants may have experienced shifts in their stress exposure or sleep health over time, introducing additional variability that our baseline R/S indicators could not capture. Fourth, independent of changes in stress exposure, variability in sleep measures over time may also contribute to the lack of observed longitudinal associations. Intraclass correlation coefficients indicated low within-person stability for sleep duration, NRS, and IS across follow-up, suggesting meaningful within-person variability over time. Such variability may be developmentally expected during early to mid-adulthood, when women experience substantial life transitions (e.g., pregnancy, caregiving, work demands) that influence sleep, likely reflecting the dynamic and context-sensitive nature of sleep during this life stage. Accordingly, differences between longitudinal and cross-sectional findings may reflect this variability, which can introduce misclassification in repeated assessments and attenuate longitudinal associations.

It is also important to interpret these findings in the context of the substantial economic disadvantage experienced by many women in this cohort, with nearly half reporting annual household incomes below $20,000. Economic hardship and financial strain are chronic stressors that can adversely affect sleep through pathways such as housing instability, neighborhood stressors, caregiving burden, and constrained access to health-promoting resources.^54,55 In such contexts, R/S may serve as a critical psychosocial resource by providing meaning, hope, and social support, potentially influencing how individuals cope with stress and manage sleep. At the same time, persistent material hardship may limit the extent to which psychosocial resources alone can offset the physiological and environmental conditions that undermine sleep health.

Our findings should be further interpreted in the context of several limitations. First, although we used validated self-reported sleep measures, objective sleep data (e.g., actigraphy) were not available, which may have introduced measurement error or recall bias.⁵⁶ In addition, standardized clinical instruments such as the Insomnia Severity Index⁵⁷ were not collected in the SELF cohort. As a result, insomnia symptoms were assessed using symptom-based items rather than validated clinical scales with established diagnostic thresholds, which may limit direct comparability with studies using standardized instruments. Similarly, R/S was assessed only at baseline using limited items that did not distinguish between organizational, non-organizational, and intrinsic religiosity, nor did they capture religious coping styles (e.g., positive vs. negative coping) or experiences of religious struggle. Religious service attendance patterns were also not assessed, so we could not examine attendance-specific associations or compare our findings with the limited existing evidence on sleep health.⁵⁸ These nuances may be important in shaping how R/S influences sleep health. Future studies should include multidimensional and longitudinal assessments of R/S to better understand its dynamic role in sleep health over time and to address the potential for reverse causation, such as the possibility that sleep problems may influence engagement in religious or spiritual practices rather than result from them. The distinctions between religion and spirituality should also be examined as they could have differential impacts on sleep health. Future research incorporating validated clinical sleep instruments alongside symptom-based measures would help extend and replicate these findings in both epidemiologic and clinical contexts. Additionally, R/S indicators were modeled as categorical variables without assuming a linear trend, which limited our ability to assess potential dose-response or non-linear relationships across levels of R/S. As a result, observed differences should be interpreted as category-specific associations rather than evidence of a continuous functional form. Future studies that assess R/S using continuous or more finely graded measures may be better positioned to evaluate potential non-linear relationships. Second, our measure of stress was a single-item assessment of perceived day-to-day stress, which, while meaningful, may not fully capture the complexity or chronicity of stress experiences among Black/AA women. Inclusion of validated multidimensional stress scales or biomarkers of stress (e.g., cortisol, allostatic load) in future studies could help clarify the biopsychosocial mechanisms linking stress, R/S, and sleep. We also did not capture detailed childcare and other caregiving responsibilities (e.g., number/ages of children, nighttime caregiving, partner support), which may confound the observed R/S–sleep associations despite adjustment for marital status and employment. Third, given the evaluation of multiple dimensions of R/S across several sleep outcomes and stress strata, the potential for inflated Type I error due to multiple statistical tests should be considered. However, all analyses were specified a priori based on conceptual hypotheses rather than conducted post hoc, and formal multiplicity corrections were not applied, consistent with epidemiologic guidance suggesting that routine adjustment for multiple comparisons may not be warranted in hypothesis-driven research.⁵⁹ Finally, the sample included young Black/AA women from Detroit, Michigan, which may limit generalizability to other populations or age groups. However, this demographic is often underrepresented in sleep and health disparities research, and our study helps fill a critical gap by focusing on early adulthood, a life stage when chronic sleep problems often emerge. Despite these limitations, this study has notable strengths. We leveraged data from a large, well-characterized cohort of Black/AA women with repeated assessments over time. The inclusion of multiple sleep dimensions and the consideration of both main along with modifying effects of R/S and stress offers a nuanced understanding of the role of R/S in sleep health.

Our findings underscore the importance of considering culturally salient coping mechanisms, such as R/S, when studying sleep health disparities. While public health and clinical sleep interventions often focus on behavioral and cognitive strategies, integrating the cultural and spiritual values of target populations may improve acceptability and effectiveness, particularly for Black/AA women. For example, faith-based sleep health promotion efforts or spiritually integrated cognitive behavioral therapy for insomnia ⁶⁰ may resonate more deeply within communities for whom R/S plays a central role.⁶¹ More broadly, evidence from intervention research supports the clinical relevance of integrating R/S-informed approaches into health promotion efforts. A recent systematic review and meta-analysis of randomized controlled trials found that religious and spiritual interventions were associated with significant improvements in mental health symptoms, well-being, and spiritual outcomes.⁶² The present findings provide practical insight by suggesting that clinicians should not assume that all forms of R/S engagement confer sleep benefits; rather, the perceived meaning and function of R/S may determine whether it operates as a resilience factor or varies in its association with sleep depending on context. In practice, clinicians and community-based programs might first assess whether R/S functions as a source of comfort and resilience for a given individual versus reflecting different patterns of engagement that may not confer the same sleep-related benefits, and tailor support accordingly, acknowledging heterogeneity in how religious or spiritual engagement may impact sleep health. Future research should investigate the role of religious coping styles, religious struggle/trauma, and spiritual support networks in shaping sleep health over time. Longitudinal and qualitative studies may help clarify whether certain forms of R/S serve as proactive coping mechanisms or reactive responses to distress, and how this distinction influences sleep outcomes. Moreover, interdisciplinary approaches that bridge epidemiology, psychology, theology, and community engagement are warranted to develop holistic, equity-informed interventions.

In conclusion, this study highlights the complex and context-dependent relationship between R/S and subjective sleep health among young Black/AA women. All three dimensions of R/S, including importance of faith, R/S as a source of strength/comfort, and prayer/meditation, were associated with lower prevalence of non-restorative sleep overall; however, specific associations varied by stress. Protective associations with considering R/S as a source of strength and comfort as well as with daily prayer or meditation were observed among women reporting low/no stress, not among women with high stress. No longitudinal associations were found. Combined, these results underscore the need for future studies with repeated measures of R/S, more granular assessments of both stress and coping, and objectively measured sleep. These findings ultimately emphasize the importance of incorporating salient, culturally relevant psychosocial factors into sleep health research to better inform interventions aimed at reducing disparities and improving well-being in Black/AA communities.

Figure 2. Study design and assessment timeline in the SELF cohort.

Table 2.

Baseline characteristics of participants by dimensions of amount religion/spirituality is a source of strength, Study of Environment, Lifestyle and Fibroids, N = 1693

Sociodemographic characteristics	Amount religion/spirituality is a source of strength/comfort
	Not at all n = 63 (3.7%)	Somewhat n = 293 (17.3%)	Quite a bit n = 395 (23.3%)	Very much n = 942 (55.6%)	Total n = 1693 (100%)	p-value
Age, years (mean ± SD)	28.22 ± 3.51	29.22 ± 3.54	28.96 ± 3.34	29.39 ± 3.43	29.22 ± 3.44	0.019
Educational attainment, n (%)
≤High school	26 (41.3)	91 (31.1)	94 (23.8)	159 (16.9)	370 (21.9)	<0.001
Some college or Associate’s/Technical degree	24 (38.1)	141 (48.1)	203 (51.4)	480 (51.0)	848 (50.1)
≥Bachelor’s degree	13 (20.6)	61 (20.8)	98 (24.8)	303 (32.2)	475 (28.1)
Employment status, n (%)
Not employed	33 (52.4)	129 (44.0)	148 (37.6)	330 (35.1)	640 (37.9)	0.014
Employed part-time	6 (9.5)	32 (10.9)	41 (10.4)	131 (13.9)	210 (12.4)
Employed full-time	24 (38.1)	132 (45.1)	205 (52.0)	479 (51.0)	840 (49.7)
Annual household income, n (%)
<$20,000	43 (68.3)	155 (52.9)	167 (42.3)	408 (43.3)	773 (45.7)	<0.001
$20,000 to $50,000	13 (20.6)	98 (33.4)	165 (41.8)	354 (37.6)	630 (37.2)
>$50,000	7 (11.1)	40 (13.7)	63 (15.9)	180 (19.1)	290 (17.1)
Marital status, n (%)
Never married	51 (81.0)	174 (59.4)	225 (57.0)	542 (57.5)	992 (58.6)	0.002
Previously married	7 (11.1)	30 (10.2)	54 (13.7)	145 (15.4)	236 (13.9)
Currently married	5 (7.9)	89 (30.4)	116 (29.4)	255 (27.1)	465 (27.5)
Health behaviors and characteristics
Smoking status, n (%)
Never	43 (68.3)	192 (65.5)	272 (68.9)	738 (78.3)	1245 (73.5)	<0.001
Former	1 (1.6)	26 (8.9)	34 (8.6)	64 (6.8)	125 (7.4)
Current	19 (30.2)	75 (25.6)	89 (22.5)	140 (14.9)	323 (19.1)
Alcohol use, n (%)
None	20 (31.7)	57 (19.5)	86 (21.8)	338 (35.9)	501 (29.6)	<0.001
Moderate	32 (50.8)	159 (54.3)	215 (54.4)	455 (48.3)	861 (50.9)
Heavy	11 (17.5)	77 (26.3)	94 (23.8)	149 (15.8)	331 (19.6)
Body Mass Index (kg/m2), n (%)
Underweight/normal (<25)	16 (25.4)	54 (18.4)	69 (17.5)	196 (20.8)	335 (19.8)	0.122
Overweight (25 - <30)	6 (9.5)	62 (21.2)	96 (24.3)	186 (19.7)	350 (20.7)
Obese (≥30)	41 (65.1)	177 (60.4)	230 (58.2)	560 (59.4)	1008 (59.5)
Sleep duration, n (%)
Short (<7 hours)	33 (52.4)	180 (61.4)	228 (57.7)	548 (58.2)	989 (58.4)	0.788
Recommended (7–9 hours)	30 (47.6)	110 (37.5)	162 (41.0)	383 (40.7)	685 (40.5)
Long (>9 hours)	0 (0.0)	3 (1.0)	5 (1.3)	11 (1.2)	(1.1)
Insomnia symptoms, n (%)
Yes	8 (12.7)	67 (22.9)	69 (17.5)	159 (16.9)	303 (17.9)	0.079
No	55 (87.3)	226 (77.1)	326 (82.5)	783 (83.1)	1390 (82.1)
Non-restorative sleep, n (%)
Yes	38 (60.3)	211 (72.0)	251 (63.5)	550 (58.4)	1050 (62.0)	<0.001
No	25 (39.7)	82 (28.0)	144 (36.5)	392 (41.6)	643 (38.0)
General health, n (%)
Poor/Fair	12 (19.0)	56 (19.1)	83 (21.0)	130 (13.8)	281 (16.6)	<0.001
Good	21 (33.3)	128 (43.7)	148 (37.5)	345 (36.6)	642 (37.9)
Very good/Excellent	30 (47.6)	109 (37.2)	164 (41.5)	467 (49.6)	770 (45.5)
Daily life stress, n (%)
High	24 (38.1)	135 (46.1)	190 (48.1)	392 (41.6)	741 (43.8)	0.100
Low/No	39 (61.9)	158 (53.9)	205 (51.9)	550 (58.4)	952 (56.2)

Note. Values are mean ± SD or n (%). p-values were obtained using one-way ANOVA for Age and Pearson’s χ² tests for all other (categorical) variables; tests were two-sided with α=0.05. Missing values were excluded pairwise. Moderate alcohol use = 1–5 drinks on days when having alcohol; Heavy alcohol use = 6+ drinks on days on having alcohol

Missing data for employment: n = 3

Statement of significance:

Most Americans identify as religious or spiritual, yet religiosity/spirituality (R/S) remains an understudied determinant of sleep health despite its impact on health and well-being. Using multidimensional measures of R/S, we found that valuing faith, finding strength/comfort in R/S, and prayer/meditation were each associated with restorative sleep, primarily among those reporting low/no daily stress. These findings underscore the importance of incorporating R/S into culturally relevant sleep health research and interventions.

Data Availability Statement:

Participant data that underlie the primary results reported in this article can be requested for purposes of replication or meta-analysis by accessing the Study Request Form at https://www.niehs.nih.gov/research/atniehs/labs/epi/resources/data-sharing. All data releases will require a data use proposal and a data use agreement and will comply with the IRB and consent of the SELF study, which may require omission of some data elements. The NIH IRB may be asked to review the request and study consent forms to approve a data transfer.