The Association of Multidimensional Sleep Health with HbA1c and Depressive Symptoms in African American Adults with Type 2 Diabetes

Abstract

Objective:

Sleep is important for diabetes-related health outcomes. Using a multidimensional sleep health framework, we examined the association of individual sleep health dimensions and a composite sleep health score with hemoglobin A1c (HbA1c) and depressive symptoms among African American adults with type 2 diabetes.

Methods:

Participants (N = 257; mean age 62.5 years) were recruited through local churches. Wrist-worn actigraphy and sleep questionnaire data assessed multidimensional sleep health using the RuSATED framework (regularity, satisfaction, alertness, timing, efficiency, duration). Individual sleep dimensions were dichotomized into poor or good sleep health and summed into a composite score. HbA1c was assessed using the DCA Vantage^™ Analyzer or A1CNow^® Self Check. Depressive symptoms were assessed using the Patient Health Questionnaire (PHQ-9). Regression models examined the association of individual sleep dimensions and composite sleep health with HbA1c and depressive symptoms.

Results:

Higher composite sleep health scores were associated with a lower likelihood of having greater than minimal depressive symptoms (PHQ-9 ≥5) (odds ratio [OR]=0.578, 95% confidence interval [CI]: 0.461–0.725). Several individual sleep dimensions, including irregularity (OR=1.013, CI: 1.005–1.021), poor satisfaction (OR=3.130, CI: 2.095–4.678), and lower alertness (OR=1.866, CI: 1.230–2.833) were associated with a greater likelihood of having depressive symptoms. Neither composite sleep health scores nor individual sleep dimensions were associated with HbA1c.

Conclusions:

Better multidimensional sleep health is associated with lower depressive symptoms among African American adults with type 2 diabetes. Longitudinal research is needed to determine the causal association between multidimensional sleep health and depressive symptoms in this population.

Trial Registry:

ClinicalTrials.gov identifier NCT04282395

INTRODUCTION

Sleep plays a critical role in the health outcomes of individuals with type 2 diabetes (T2D).^1,2 Poor sleep has been linked to heightened insulin resistance, hyperglycemia,³ and comorbidities including hypertension.⁴ Sub-optimal sleep duration and poor sleep quality are associated with a higher prevalence of T2D⁵ and elevated HbA1c,⁶ while daytime sleepiness is associated with poor glycemic control.⁷ Furthermore, sleep irregularity is associated with a higher prevalence and more severe symptoms of depression,⁸ and poor subjective sleep quality is a major predictor of depression onset.⁹

T2D doubles the risk for major depressive disorder,¹⁰ highlighting a need to examine modifiable risk factors for depression in individuals with T2D. Caring for T2D requires continuous self-management, which can cause psychological stress and increase experiences of depressive symptoms.¹¹ T2D is a pro-inflammatory state,¹² and circulating inflammatory markers have been associated with greater depressive symptoms in individuals with T2D.¹³ The relationship between T2D and depression is likely bidirectional, as depressive symptoms may contribute to metabolic disruption via affecting physical activity levels and other relevant health behaviors such as sleep.¹⁴ Furthermore, individuals with co-morbid T2D and depression have more T2D-related health complications.¹¹

Despite racial disparities in the diagnosis and treatment for diabetes,¹⁵ depression,¹⁶ and sleep disorders,¹⁷ most studies examining the interplay between these comorbidities have been conducted in non-Hispanic White adults. In studies where AAs were included, most research analyses combined non-Hispanic White participants into a single group, leaving an important gap in knowledge about the unique sleep health needs, experiences, and differences specific to the AA population. Compared to non-Hispanic Whites, African Americans (AAs) experience disparities in multiple sleep dimensions. AAs exhibit shorter (<6 hours) and longer (>9 hours) sleep duration,¹⁸ both of which are associated with an increased morbidity risk,⁴ and report greater sleep irregularity,⁸ sleep-disordered breathing, and insomnia symptoms.¹⁹ AAs with T2D experience poorer sleep quality⁶ and T2D-related discomforts including hyperglycemic episodes and neuropathy can further impair sleep health.^20,21

Most sleep research examines either a single sleep dimension or multiple sleep dimensions independently, which fails to acknowledge the interconnected nature of multiple sleep dimensions and the combined influence of multiple sleep dimensions on health outcomes. The RuSATED (regularity, satisfaction, alertness, timing, efficiency, duration) multidimensional sleep health framework^22,23 emphasizes the importance of multiple sleep dimensions being associated with optimal health and psychological well-being. This framework broadens the definition of sleep health beyond the absence of sleep disorders and facilitates early prevention and intervention efforts to ameliorate the harmful effects of sub-optimal sleep on health.²² A lower composite sleep health score, constructed by aggregating the six individual RuSATED dimensions into a single measure,^24–26 has been associated with a higher prevalence of diabetes, hypertension,²⁷ depressive symptoms,²⁸ and cardiovascular disease.²⁹

Given that AAs are disproportionately affected by T2D and report poorer sleep quality than non-Hispanic Whites,^6,30 it is crucial to gain a deeper understanding of the association between multidimensional sleep health and T2D-related health outcomes in this population. Accordingly, the overall aims of this study were to: 1) characterize multidimensional sleep health in AAs with T2D; 2) determine the association between multidimensional sleep health and HbA1c in AAs with T2D; and 3) determine the association between multidimensional sleep health and depressive symptoms in AAs with T2D. It is hypothesized that better multidimensional sleep health is associated with lower HbA1c and lower depressive symptoms. To comprehensively address these aims, an established self-reported sleep questionnaire and objectively measured sleep data were used to examine the association between individual sleep dimensions and a composite sleep health score with T2D-related health outcomes.

METHODS

Participants

The current study used baseline data (N = 284) from TX STRIDE (Texas Strength Through Resilience in Diabetes Education), an ongoing clinical trial examining the effectiveness of a resilience-based diabetes self-management education and support program on diabetes-related health outcomes.³¹ Participants were recruited through local churches in Austin, TX, and the surrounding areas. Inclusion criteria were being AA, 18 years of age or older, and being diagnosed with T2D. Individuals were excluded if they were pregnant/lactating or had medical conditions for which changes in diet and/or physical activity would be contraindicated.

Study procedures

Data were collected from August 2020 to April 2023. Due to the COVID-19 pandemic, remote baseline testing was conducted from August 2020 to July 2021 (n = 82). In-person baseline testing was conducted from July 2021 to April 2023 on-site (n = 202). During data collection, participants completed a self-reported sleep questionnaire and depressive symptoms questionnaire. Finger-prick blood samples were collected for HbA1c measurement. Participants wore actigraphy on their non-dominant wrist for 24 hours a day for 7 consecutive days for objective sleep measurements. We excluded participants with less than four out of seven nights of actigraphy data based on evidence that four nights provides similar estimates of sleep duration as nine nights.³² We also excluded participants who were missing sleep questionnaire or sleep medication data. Therefore, the final analytic sample size was N = 257 (Figure 1). Self-report questionnaires and HbA1c measurement were completed before actigraphy. The study protocol was approved by the Institutional Review Board at The University of Texas at Austin and all participants provided written informed consent.

Figure 1.

Participant selection for final analytic sample size

Measurements

HbA1c.

The measurement of HbA1c was performed using the DCA Vantage^™ Analyzer (DCA, Siemens Medical Solutions Diagnostics, Malvern, PA) or the A1CNow^® Self Check (A1CNow, PTS Diagnostics, Whitestown, IN), an at-home self-testing kit. Before the emergence of COVID-19, HbA1c was planned to be measured in person using the DCA. However, when in-person research activities were paused due to the pandemic, a modification was implemented, and participants were provided an A1CNow kit to measure their HbA1c at home with remote assistance of research staff. When in-person research activities resumed, HbA1c was measured in a subsample of participants using both the A1CNow and DCA to assess the accuracy of the A1CNow compared to the DCA as a reference standard. In the present sample, A1CNow measurements were significantly lower than DCA measurements (t[32] = −5.89, p < .001). Therefore, we applied a regression equation (y = 0.665 + 1.003x) to adjust HbA1c assessed using the A1CNow based on findings from our recent validation study.³³ DCA measurements were used in the analysis whenever possible (n = 182) and adjusted A1CNow measurements were used for participants without DCA measurements (n = 75).

Depressive Symptoms.

We measured depressive symptoms using the Patient Health Questionnaire-9 (PHQ-9) to assess the frequency and severity of depressive symptoms over the past two weeks.³⁴ Each item was rated on a 4-point Likert scale (0 = not at all; 1 = several days; 2 = more than half the days; 3 = nearly every day) with higher scores indicating more severe depressive symptoms. The reliability of the PHQ-9 in the current study was moderately strong (α = .85).

Sleep Assessment.

Participants were asked to wear an actigraphy monitor (model wGT3X-BT, ActiGraph Corp, Pensacola, FL) on their non-dominant wrist for 7 consecutive days. All devices were initialized using ActiLife software version 6.13.4. Devices used a sampling frequency of 30 Hz and began recording at 7:00 am the morning of data collection (Saturday). After the 7-day wear time, participants were instructed to mail the actigraphy using a postage-paid return envelope. Sleep data were downloaded and analyzed in 1 min epochs using the Cole-Kripke and ActiLife algorithms (ActiGraph Corp) for sleep periods and sleep duration detection. All sleep health actigraphy data (i.e., sleep period graph, bedtime, wake-up time, efficiency, total sleep time) were checked for accuracy by three trained research staff members. Two staff members independently reviewed actigraphy records and identified ambiguous periods of sleep onset/offset; decisions were finalized by the co-first author (HML) and the PI. Self-reported sleep was measured via the Pittsburgh Sleep Quality Index (PSQI),³⁵ which assesses sleep quality over the past month.

Multidimensional Sleep Health.

Multidimensional sleep health was operationalized using the RuSATED framework.²² Individual sleep health dimensions were obtained using actigraphy data (regularity, timing, efficiency, and duration) or PSQI (satisfaction and alertness). Sleep regularity was defined as the extent to which individuals had consistent sleep and wake-up time schedules across days. Sleep timing was defined as the midpoint from sleep onset to wake up. Sleep efficiency was operationalized as the ratio of total sleep time to total time in bed. Sleep duration was operationalized as the total amount of sleep obtained from sleep onset to wake-up. Sleep satisfaction was defined as a subjective assessment of individuals’ sleep quality. Sleep alertness was defined as the ability to maintain wakefulness without getting sleepy during the day. Table 2 shows the sources of measurement for each individual sleep dimension and descriptive information.

Table 2.

Descriptive information of individual sleep health dimensions (N = 257)

Sleep dimensions	Sources	Means (SD)	Cutoff values	Good (%)
Regularity	Actigraphy derived SD of midpoint of sleep (hh:mm)	1:06 (0:55)	≤1 hour = good47 >1 hour = poor	64%
Satisfaction	“How would you rate your sleep quality overall?” on PSQI: 0 (very good) to 3 (very bad)	1.25 (.87)	0,1 = good47 2,3 = poor	70%
Alertness	“How often have you had trouble staying awake while driving, eating meals, or engaging in social activity?” on PSQI: 0 (not during the past month) to 3 (three or more times a week)	0.30 (.74)	0 = good 1–3 = poor	82%
Timing	Actigraphy-derived midpoint from sleep onset to wake up (hh:mm)	3:24 (1:20)	2:00 am - 4:00 am = good48 <2:00 am or >4 :00 am = poor	60%
Efficiency	Actigraphy-derived ratio of total sleep time to time in bed (%)	87.6 (5.9)	≥85% = good49 <85 % = poor	75%
Duration	Actigraphy-derived total amount of sleep obtained (hh:mm)	6:37 (1:09)	6–8 hours = good22 <6 hours or >8 hours = poor	62%

Note: SD = standard deviation; PSQI = Pittsburgh Sleep Quality Index. Good (%) = percentage of participants who had good sleep health based on the cutoff values.

A composite sleep health score was calculated by dichotomizing individual sleep health dimensions using cut-off values based on prior research: Poor sleep health was coded as 0 and good sleep health was coded as 1. Scores were summed across dimensions to create a composite sleep health score ranging from 0 to 6 with higher scores indicating better sleep health. The process of creating a composite sleep health score was consistent with other studies^25,26 to facilitate comparisons across studies and enhanced translation of study findings into clinical settings.

Statistical Analysis

The Shapiro-Wilks normality test and visual inspection of histograms assessed normality of study variables. Based on these methods, depressive symptoms exhibited positive skew (W = .86, p = <.001), so two modeling approaches were used, described below. Descriptive statistics were used to examine participant characteristics, sleep health, HbA1c, and depressive symptoms. Correlations among study variables were assessed using Pearson (continuous variables) and point-biserial (continuous and categorial variables) correlation coefficients. Bar graphs were plotted to examine the association of HbA1c and depressive symptoms with composite sleep health scores. Multiple linear regression analysis was performed to examine the association of individual sleep health dimensions (regularity, satisfaction, alertness, timing, efficiency, duration) and a composite sleep health score with HbA1c. Associations between sleep health and depressive symptoms were determined using two approaches. First, depressive symptoms were dichotomized as having no or minimal depressive symptoms (PHQ ≤ 4; coded 0) or having greater than minimal depressive symptoms (PHQ ≥ 5; coded 1), as in other studies.^36,37 Multiple logistic regression analysis was performed to examine the association of individual sleep health dimensions and a composite sleep health score with the presence of depressive symptoms. Second, the PHQ-9 scores reflected a negative binomial distribution, exhibiting significant positive skew and variance substantially larger than the mean. Rather than transform the data, which would be problematic given the distribution, we left the data untransformed and performed a negative binomial regression (Supplemental Digital Content analysis). Each model exhibited good fit (Pearson χ² range: 0.65–0.74).

Sociodemographic covariates were chosen a priori based on established associations with HbA1c and depressive symptoms. These included age^38,39, sex,^40,41 education,⁴² employment status,^43,44 and marital status.^45,46 Other covariates were sleep medication use and diabetes medication use (for models predicting HbA1c). The proportion of actigraphy days that were weekend days relative to total actigraphy days was associated with later sleep timing (r = .13, p = .042), so the proportion of weekend actigraphy nights was included as a covariate for models in which sleep timing was the predictor of interest. Two participants were missing PHQ-9 scores, resulting in a sample size of 255 for models predicting depressive symptoms. Participants had complete data on all other study variables. All data analyses were performed using SPSS version 28 (IBM, Chicago, Illinois). A post-hoc power analysis (G*Power 3.1.9.7) with power = .80, N = 257, and α = .05, showed that linear regression models including 7–8 predictors (one focal predictor and 6–7 covariates per model) could detect small effect sizes (f² = .03) of the focal predictors.

RESULTS

Participants had an average age of 62.5 ± 10.6 years, and a majority were female (72%) and had some college experience (48%) or a college degree (30%). Selected characteristics of study participants are shown in Table 1.

Table 1.

Participant characteristics (N = 257)

Characteristics	Mean (SD) or n (%)	Range
Age (years)	62.5 (10.6)	25–86
Male/female (% female)	73/184 (72%)
Education
High school or lower	56 (22%)
Some college/technical school	124 (48%)
Undergraduate degree	44 (17%)
Graduate degree	33 (13%)
Employment status (currently employed)	127 (49%)
Marital status (currently married)	123 (48%)
Body mass index (kg/m2)	36.3 (8.4)	19.5–66.0
Diabetes diagnosis length (years)	11.4 (8.7)	.1–50.0
Diabetes medication use
Oral medications/non-insulin injectable only	154 (60%)
Insulin only	22 (9%)
Both	63 (25%)
No medication	18 (7%)
Sleep medication use	64 (25%)
Composite sleep health score (0–6)	4.1 (1.3)	0–6
HbA1c (mg/dL)	8.0 (1.8)	5.4–14.0
Depressive symptoms (PHQ-9 sum; range 0–27)a	5.2 (5.1)	0–26
Greater than minimal depressive symptoms (PHQ-9 ≥ 5)a	111 (44%)

Note:

^aN = 255

At least 60% of participants meet the threshold for “good” sleep health in each sleep dimension (Table 2). In bivariate analyses, higher composite sleep health scores were negatively associated with depressive symptoms (r = −.42, p < .001). Additionally, regularity (r = .26, p < .001), satisfaction (r = .55, p < .001), and alertness (r = .36, p < .001) were associated with less depressive symptoms. However, neither the composite sleep health score nor individual sleep health dimensions were associated with HbA1c (Table 3).

Table 3.

Associations among continuous individual sleep health dimensions, the composite sleep health score, HbA1c, and depressive symptoms (N = 257)

	1	2	3	4	5	6	7	8
1. Regularity	—
2. Satisfaction	.14*	—
3. Alertness	.02	.17**	—
4. Timing	.03	.10	.09	—
5. Efficiency	−.19**	−.11	−.06	.02	—
6. Duration	−.41***	−.06	−.05	−.07	.45***	—
7. Composite sleep health	−.40***	−.42***	−.45***	−.23***	.44***	.41***	—
8. HbA1c	.08	.09	.09	.03	.01	−.02	−.08	—
9. Depressive symptomsa	.26***	.55***	.36***	.12	−.01	−.05	−.42***	.15*

Note:

^aN = 255;

^*p < .05;

^**p < .01;

^***p < .001.

Bar graphs illustrating the distribution of HbA1c and depressive symptoms over the range of sleep health composite scores are shown in Figure 2. HbA1c was not linearly associated with composite sleep health scores (F[5, 251] = 1.43, p = .21), although the highest HbA1c was observed at the lowest composite sleep health score. Conversely, participants experiencing poor sleep indicated by lower composite sleep health scores exhibited higher depressive symptom scores (F[5, 249] = 11.67, p < .001).

Figure 2.

Bar graphs with standard errors showing HbA1c levels (Panel A) and depressive symptom scores (Panel B) at each composite sleep health score.

Table 4 shows the multiple linear regression results of individual sleep health dimensions and the composite sleep health score with HbA1c, adjusted for covariates. Older age (b range = −.154 to −.130, p range = .041 to .092) and female sex (b range = −.239 to −.228, all ps < .001) were significantly associated with lower HbA1c levels. However, HbA1c was not significantly associated with the composite sleep health score or any individual sleep health dimensions (b range = −.032 to .078, p range = .228 to .828. In sex-stratified analyses, the pattern of findings in both women and men was similar to those in the full sample (Supplemental Digital Content, Tables S1 and S2).

Table 4.

Multiple linear regression of individual sleep health dimensions and the composite sleep health score with HbA1c (N = 257). Each row represents an independent regression model.

	HbA1c				Model R2
	B	SE	b	p
Regularity (SD sleep midpoint; minutes)	.001	.002	.045	.476	.076
Satisfaction (0–3; lower scores = better)	.160	.132	.078	.228	.079
Alertness (0–3; lower scores = better)	.131	.151	.054	.387	.076
Timing (sleep midpoint; minutes)a	.000	.001	.014	.828	.074
Efficiency (%)	.009	.019	.029	.641	.074
Duration (total sleep time; hours)	.001	.002	.030	.633	.075
Sleep health (0–6; higher scores = better)	−.042	.084	−.032	.619	.075

Note: B = unstandardized coefficients, SE = standard errors, b = standardized coefficients. Models adjusted for age, sex, education, employment status, marital status, diabetes medication use, and sleep medication use.

^aAdjusted for proportion of weekend actigraphy nights.

Table 5 shows the multiple logistic regression results for individual sleep health dimensions and the composite sleep health score with depressive symptoms, adjusted for covariates. The composite sleep health score was significantly associated with greater than minimal depressive symptoms (odds ratio [OR] = 0.578, 95% confidence interval [CI]: 0.461–0.725), such that participants with higher composite sleep health scores had a lower likelihood of having greater than minimal depressive symptoms. Additionally, participants with more irregular sleep timing (OR = 1.013, CI: 1.005–1.021), poorer sleep satisfaction (OR = 3.130, CI: 2.095–4.678), and lower alertness (OR = 1.866, CI: 1.230–2.833) had a higher likelihood of having greater than minimal depressive symptoms. Sleep timing, efficiency, and duration were not significantly associated with likelihood of endorsing greater than minimal depressive symptoms. The negative binomial regression determined that better sleep health was associated with lower depressive symptoms (B = −.233, p < .001), while greater satisfaction (B = .510, p < .001) and alertness (B = .294, p = .001) were associated with greater depressive symptoms (Supplemental Digital Content, Table S3).

Table 5.

Multiple binary logistic regression of individual sleep health dimensions and the composite sleep health score with greater than minimal depressive symptoms (N = 255). Each row represents an independent regression model.

	Greater than minimal depressive symptoms (PHQ-9 ≥ 5)
	OR (95% CI)	p
Regularity (SD sleep midpoint; minutes)	1.013 (1.005 – 1.021)	.002
Satisfaction (0–3; lower scores = better)	3.130 (2.095 – 4.678)	<.001
Alertness (0–3; lower scores = better)	1.866 (1.230 – 2.833)	.003
Timing (sleep midpoint; minutes)a	1.003 (1.000 – 1.007)	.088
Efficiency (%)	0.985 (0.943 – 1.029)	.511
Duration (total sleep time; hours)	0.997 (0.993 – 1.001)	.147
Sleep health (0–6; higher scores = better)	0.578 (0.461 – 0.725)	<.001

Note: OR = odds ratio, CI = confidence interval. Models adjusted for age, sex, education, employment status, marital status, and sleep medication use.

^aAdjusted for proportion of weekend actigraphy nights.

DISCUSSION

The current study examined the association of multidimensional sleep health with HbA1c and depressive symptoms among AAs with T2D. Higher composite sleep health scores were significantly associated with a lower likelihood of having depressive symptoms. Additionally, several individual sleep health dimensions, including irregular sleep, poor sleep satisfaction, and lower alertness, were significantly associated with a higher likelihood of having depressive symptoms. HbA1c was not significantly associated with the composite sleep health score or individual sleep dimensions. These results support an important link between multidimensional sleep health and depressive symptoms in AAs with T2D.

Composite sleep health scores showed a negative linear association with depressive symptoms in the present study. Additionally, a 1-unit increase in sleep health was associated with 42% lower odds of endorsing greater than minimal depressive symptoms. These findings align with previous research demonstrating the significant association between multidimensional sleep health and depressive symptoms.⁵⁰ For example, in a large sample of female nurses, poor composite sleep health scores at baseline were associated with greater odds of having depressive symptoms in a gradient fashion.⁵¹ However, a majority of available studies assessed multidimensional sleep health using self-reported questionnaires alone, and participants were predominantly non-Hispanic Whites or Asians. In addition to self-report questionnaires, the present study utilized objective measures of sleep, which provide a more accurate and robust assessment of sleep health in association with depressive symptoms. The present study not only reinforces the association between sleep health and depressive symptoms but also extends this finding to AA adults with T2D who may be at elevated risk for depressive symptoms. Increased awareness of sleep health may be important for AAs given the strong stigma associated with depression and the underutilization of mental health services in this population.^52,53

In addition to the composite sleep health score, several individual sleep dimensions, including more irregular sleep timing, poorer sleep satisfaction, and lower alertness were also associated with higher depressive symptoms. These results are aligned with previous studies reporting that poor sleep satisfaction was associated with 2.1-fold higher odds of having depressive symptoms⁵¹ and that daytime sleepiness was associated with increased prevalence and incidence of depressive symptoms.^50,54 In the present study, self-reported sleep health measures showed stronger associations with depressive symptoms than objective measures of sleep. More specifically, participants with poor satisfaction had a 3.1-fold increased likelihood of experiencing greater than minimal depressive symptoms, while those with lower alertness had a 1.9-fold increased likelihood of experiencing greater than minimal depressive symptoms. Previous studies reported that poor subjective sleep satisfaction was associated with postpartum depression in mothers, but objectively measured sleep data, including sleep duration, efficiency, and disturbances, were not associated with depressive symptoms.^55,56 Taken together, these findings suggest that individuals’ perception of sleep may be more salient for depressive symptoms than objective measures of sleep.

Although previous studies utilizing the composite sleep health score found significant associations with various health outcomes including chronic conditions and physical dysfunction,^24–26 neither the composite sleep health score nor individual sleep dimensions were associated with HbA1c in the current study. It is possible that the cross-sectional design of the study may have weakened the ability to establish a significant association between HbA1c and sleep health. As HbA1c reflects average blood glucose levels over the past 3 months, sleep health at baseline may not be linked to HbA1c levels at baseline. Lack of significant findings may have also been due to the widespread use of T2D-related medications in our sample (93%), which could confound associations between sleep health and glucose control. Although we statistically adjusted for T2D medication use, medication use has a complex and likely nonlinear association with glucose control. Medications may substantially lower HbA1c for some individuals, while having minimal impact for others. Those with poorly controlled T2D are more likely to use medications than those with well-controlled T2D, but effective medications can mask underlying physiological glucose dysregulation. Thus, there is a need to study African Americans with T2D who are not taking medications to better determine how sleep health is related to glucose control in this population. Future research should also employ longitudinal designs with repeated measures of sleep over time among AAs to provide a more comprehensive understanding of the influence of the combined effects of multiple sleep dimensions on HbA1c in this population. However, the highest HbA1c levels and highest depressive symptom scores were observed at the lowest composite sleep health score based on descriptive plots which may suggest that AAs with the poorest sleep health could benefit from public health interventions aimed at enhancing sleep health to improve health outcomes.

As in any other studies, the findings of this study should be considered in light of several limitations. First, the current study is cross-sectional in design, thus the causality between sleep health and depressive symptoms cannot be determined. It is possible that the presence of depressive symptoms may have affected sleep health. Second, the current study was conducted in a specific population, AAs with T2D living in the greater Austin area. Thus, the generalizability of the study findings to other population groups is limited. Third, depressive symptoms, daytime alertness, and sleep satisfaction were assessed using retrospective self-reported questionnaires, which may not accurately reflect the presence of clinically significant depressive symptoms and the actual sleep health status of participants. Lastly, this study was conducted in a primarily female sample, which may limit the extent to which results are relevant for African American males with diabetes. However, we conducted sex-stratified analyses and did not find evidence of sex differences in our sample.

CONCLUSIONS

The current study demonstrated that multidimensional sleep health was associated with depressive symptoms but not with HbA1c among AAs with T2D. The association between sleep health and depressive symptoms was observed and confirmed in different analyses. First, the bar graph illustrating the distribution of depressive symptoms over the range of sleep health composite scores showed that participants with lower composite sleep health scores had higher depressive symptom scores. Second, higher composite sleep health scores were associated with a lower likelihood of having depressive symptoms. Third, individual sleep dimensions, including more irregular sleep, poorer sleep satisfaction, and lower alertness, were associated with a higher likelihood of having greater than minimal depressive symptoms. Importantly, these results were obtained in AAs who are significantly underrepresented in sleep research. Longitudinal research is needed to determine the causal association between multidimensional sleep health and depressive symptoms in this population.

Abbreviations used in the text:

HbA1c

hemoglobin A1c

RuSATED

regularity, satisfaction, alertness, timing, efficiency, duration

PHQ-9

Patient Health Questionnaire

T2D

type 2 diabetes

AA

African American

TX STRIDE

Texas Strength Through Resilience in Diabetes Education

PSQI

Pittsburgh Sleep Quality Index

Data Availability:

Data described in the manuscript and analytic code may be made available upon request to the corresponding author pending application and approval.