Abstract
Study Objectives:
Disparities in access to care and healthcare system inequities result in race-based health care outcome differences. Excessive daytime sleepiness (EDS), as a general marker of health, is associated with all-cause mortality. We aimed to determine if the association and its effect size differ among various race and ethnic groups in a health care system with no or minimal race-based access limitation.
Methods:
We extracted Epworth Sleepiness Scale (ESS) of all Veterans with any sleep disorders for the years 1999-2022. ESS scores from 0-10 and 11-24 were categorized as normal and high, respectively. We further extracted the racial/ethnic groups as: White, Black, Hispanic, Native/Hawaiian, and Asian. We estimated unadjusted and adjusted odds ratios (OR and aOR) of all-cause mortality for those with high compared to normal ESS in various race/ethnicity groups. Adjusting factors included age, Body Mass Index (BMI), sex, and the Charlson Comorbidity Index (CCI).
Results:
423,087 Veterans’ records showed at least one ESS value. The sample was overwhelmingly male (95%), with mean age of 54.8 (±14.6), and mean BMI of 32.6 (±6.2). The aOR for all-cause mortality was 14% higher (1.12-1.16) in the high-ESS category in the total sample. This association remained similar for White and Black individuals (aOR: 1.17, 95% CI: 1.15-1.20 for White; aOR: 1.08, 95% CI: 1.03-1.13 for Black population). There were no differences between normal and high ESS for other racial/ethnic groups.
Conclusions:
In this large real-life clinical cohort with a unique eligibility criterion, EDS link with increased all-cause mortality differed among various race/ethnic groups. In White and Blacks, EDS was associated with higher all-cause mortality. In contrast, EDS did not link with higher all-cause mortality in the other groups. Further research is needed to explore the biological reasons underpinning the study findings tying EDS to all-cause mortality.
Keywords: Excessive Daytime Sleepiness, Mortality, Natural Language Processing, Race, Ethnicity
Graphical Abstract
Introduction:
Excessive daytime sleepiness (EDS) refers to difficulty staying awake and alert during the day.(1) EDS can accompany any condition reducing the quantity and/or impairing the quality of sleep. Such conditions include obstructive sleep apnea, narcolepsy, congestive heart failure, neurodegenerative diseases, and psychiatric disorders.(2) Studies link EDS to increased risk of cardiovascular disease, stroke, and mortality.(3, 4) United States (US) military Veterans are particularly susceptible to chronic sleep disruptions due to service-related factors such as habitual irregular sleep-wake cycles and stressful sleeping conditions.(5) Medical and psychiatric conditions that contribute to EDS are also common in this population.(6) In retrospective studies, over half of military personnel referred for sleep medicine evaluations reported EDS.(7-10)
While demographic factors like age and sex have been linked to differences in EDS and its health consequences, the influence of race and ethnicity on this association remains underexplored. Emerging evidence highlights persistent racial and ethnic disparities in sleep health. Black individuals, for example, are more likely to experience short sleep duration, poor sleep quality, and untreated sleep disorders than their White counterparts, even after adjusting for socioeconomic status (11, 12). These disparities stem from longstanding social determinants of health, including residential segregation, exposure to discrimination, unequal access to care, and psychosocial stressors—all of which may contribute to sleep disruption and associated adverse outcomes (11).
Understanding role of race and ethnicity on the relationship between EDS and mortality is critical to identify high-risk groups and to address inequities in clinical care. Prior work suggested that the mortality risk associated with abnormal sleep duration may differ by racial/ethnic group, with White individuals demonstrating stronger associations between short or long sleep and death risk than other groups.(12) However, whether similar racial differences exist in the context of EDS remains unclear.
To further investigate the association between EDS and the mortality risk, we used a validated natural language processing (NLP) approach to extract Epworth Sleepiness Scale (ESS) scores from electronic medical records. ESS, a validated tool, measures subjective EDS. We hypothesized that race and ethnicity may modulate the association between subjective EDS and all-cause mortality.
Methods:
The study was approved by Baylor College of Medicine’s institutional review board (H-35366) and Michael E. DeBakey Veteran Affairs Medical Center research and development committee.
Study design and cohort.
We conducted a retrospective cohort study using electronic medical records (EMRs) from the Veterans’ Health Administration (VHA). The initial cohort included US Veterans referred to the VHA from October 1, 1999, to September 30, 2022, with any International Classification of Diseases (ICD) 9 or 10 diagnostic codes related to sleep and any sleep services use. For this study, we limited the cohort to individuals with medical visits associated with specific sleep study Current Procedural Terminology (CPT) codes, including ‘0203T,’ ‘0204T,’ ‘95782,’ ‘95783,’ ‘95800,’ ‘95801,’ ‘95806,’ ‘95808,’ ‘95810,’ ‘95811,’ ‘95819,’ ‘95821,’ ‘95822,’ ‘95827,’ ‘95828,’ ‘99508,’ ‘G0398,’ ‘G0399,’ and ‘G0400.’ Reports containing keywords such as ‘%Epworth%’ and ‘%ESS%’ in their body of text were selected. For participants with multiple notes, we only used data from the first visit documenting the ESS in the patient’s free-form text considering it as the index date and excluded the other notes. Figure 1 summarizes the procedure used.
Figure 1.
STROBE diagram for the population selection in this study.
Study variables.
Outcome variable
The primary outcome of interest was all-cause mortality. All-cause mortality data were gathered from the corporate data warehouse VHA Vital Status table (sensitivity 98.3% and specificity 99.8% relative to the National Death Index)(13, 14). Mortality data were recorded until August 23, 2023.
Exposure variable
We used ESS as the exposure variable. ESS is a well-validated, easily administered, 8-item self-reported sleepiness questionnaire.(2, 15) Its routine inclusion with sleep medicine assessment makes it a useful clinical tool for identifying sleepy individuals at risk for adverse outcomes. For this analysis, as per published literature and usual convention, EDS was defined as ESS score > 10.(15) Therefore, we created two categories: (1) High ESS or h-ESS (ESS > 10) and normal ESS or n-ESS (ESS ≤ 10). An already developed and validated NLP pipeline was used to extract ESS scores from EMR.(4) We considered the date of ESS progress note as the index date.
Other variables
Other variables included patient demographics (age, body mass index {BMI} sex, race {White, Black, and others}, ethnicity), and Charlson Comorbidity Index (CCI) scores. Individuals with BMI > 30 were classified as obese. (16) Further, we extracted Each individual’s comorbid conditions and sedative/hypnotic/antipsychotic medication regimen at the index date. For the CCI, we calculated inpatient and outpatient medical conditions for one year before the index date.(17) A CCI≥2 reflects the presence of multiple comorbid conditions.(18) To further assess whether the association of EDS and mortality differed according to race and ethnicity groups, we divided the cohort into five already previously defined groups of race and ethnicity (i.e., White, Black, Hispanic, Native/Hawaiian, and Asian. (19)
Statistical analysis
For categorical variables, we used the Chi-square test, and for continuous variables we used the student t-test. We reported the odds ratio (OR) and 95% confidence intervals (95%CI) of participants using binary logistic regression model. The reference group was patients with n-ESS (ESS < 10). We adjusted the OR for age, sex, BMI, and CCI and reported the raw OR95%CI and the adjusted OR95%CI. A significance level of <0.05 was used. Data management and analysis were performed with SPSS version 27 (IBM SPSS Statistics, Armonk, NY).
Results:
Demographic analysis:
After applying NLP to the electronic medical records of 495,514 individuals, the total cohort comprised of 423,087 participants with an ESS reported in their notes. Table 1 presents the demographic characteristics of the entire study cohort and race and ethnicity sub-cohorts. Among them, 90.5% were male, 60.7% were White, with the mean age of 54.8±14.6 years. The mean BMI was 32.6±6.2kg/m2, and the prevalence of obesity (BMI ≥ 30 kg/m2) was 63.6%. The mean CCI was 1.2 ±1.8, and 27.2% of the cohort had a CCI ≥ 2. n-ESS group (N=217,938; 51.5%), compared to h-ESS (N= 205,149; 48.5%), were older (mean age: 56.5±14.7 vs. 52.9±14.2), with higher prevalence of White individuals (66.8% vs. 57.7%), lower prevalence of obesity (62.1% vs. 65.2%) and higher prevalence of comorbidities (29.2% vs. 25.1%). Similar trends were detected in various races and ethnicity sub-cohorts comparing ESS categories.
Table 1.
Summary of demographic characteristics of the entire cohort and different race and ethnicity sub cohorts.
| Race | All | White | Black | Hispanic | Native/Hawaiian | Asian | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Number (%) | 423,087 | 263,850(60.7) | 87,614(20.7) | 35,463(8.3) | 3,562(0.8) | 5,475(1.3) | ||||||
| ESS Category | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 |
| Number | 217,938 | 205,149 | 145,537 | 118,313 | 38,954 | 48,660 | 15,512 | 19,951 | 1,612 | 1,950 | 2,304 | 3,171 |
| ESS, M(SD) | 5.6 (3.1) | 15.4 (3.4) | 5.6 (3.0) | 15.0 (3.2) | 5.5 (3.2) | 16.0 (3.6) | 5.7 (3.0) | 15.8 (3.6) | 5.7 (3.0) | 15.8 (3.4) | 5.9 (3.1) | 15.8 (3.5) |
| Age, M(SD) | 56.5 (14.7) | 52.9 (14.2) * | 58.0 (14.7) | 54.6 (14.5) * | 54.2 (13.2) | 51.4 (12.5) * | 49.9 (15.3) | 47.7 (14.4) * | 53.9 (14.9) | 50.4 (13.9) * | 47.9 (15.2) | 46.5 (14.0) * |
| Age <50, N (%) | 69,027 (31.7) | 84,292 (41.1) * | 22,632 (15.5) | 23,058 (19.4) * | 6,402 (16.4) | 9,711 (19.9) * | 5,101 (32.8) | 7,181 (35.9) * | 360 (22.3) | 541 (27.7) * | 897 (38.9) | 1,237 (39.0) |
| 50≤Age <65, N (%) | 76,998 (35.3) | 76,189 (37.1) * | 67,830 (46.6) | 63,571 (53.7) * | 24,083 (61.8) | 32,110 (65.9) * | 7,369 (47.5) | 10,025 (50.2) * | 814 (50.5) | 1,098 (56.3) * | 1,023 (44.4) | 1,566 (49.3) * |
| Age≥ 65, N (%) | 71,913 (33.0) | 44,668 (21.8) * | 55,075 (37.8) | 31,684 (26.7) * | 8,469 (21.7) | 6,839 (14.0) * | 3,042 (19.6) | 2,745 (13.7) * | 438 (27.1) | 311 (15.9) * | 384 (16.6) | 368 (11.6) * |
| Sex, Male, N (%) | 198,693 (91.2) | 184,144 (89.8) * | 135,425 (93.1) | 109,018 (92.1) * | 32,978 (84.7) | 40,949 (84.2) * | 14,109 (91.0) | 18,046 (90.5) | 1,442 (89.5) | 1,741 (89.3) | 2,067 (89.7) | 2,829 (89.2) |
| BMI, M(SD) | 32.4 (6.1) | 32.9 (6.2) * | 32.3 (6.1) | 32.9 (6.2) * | 32.5 (6.2) | 33.0 (6.2) * | 32.2 (5.8) | 32.4 (5.8) * | 32.3 (6.3) | 32.8 (6.3) * | 29.57 (5.1) | 29.80 (5.1) |
| BMI ≥ 30, N (%) | 135,265 (62.1) | 133,692 (65.2) * | 90,287 (62.0) | 77,734 (65.7) * | 24,690 (63.3) | 32,272 (66.3) * | 9,642 (62.1) | 12,473 (62.5) | 986 (61.1) | 1,250 (64.1) | 926 (40.1) | 1,346 (42.4) |
| CCI≥ 2, N (%) | 63,617 (29.2) | 51,534 (25.1) * | 43,902 (30.2) | 31,908 (27.0) * | 11,674 (30.0) | 12,119 (24.9) * | 3,259 (21.0) | 3,573 (17.9) * | 432 (26.8) | 391 (20.1) * | 351 (15.2) | 366 (11.5) * |
BMI = body mass index, CCI = Charlson Comorbidity Index, ESS = Epworth Sleepiness Scale, h-ESS = high Epworth Sleepiness Scale, n-ESS = normal Epworth Sleepiness Scale, SD = standard deviation.
P-value≤0.05 for comparison between ESS 0–10 and ESS 11–24 within each racial group
Table 2 demonstrates the prevalence of comorbidities and sedative/hypnotic/antipsychotic medications prescribed for the different race and ethnicity sub cohorts and the ESS categories. The prevalence of various comorbidities exceeded in n-ESS compared to h-ESS across all race and ethnicity groups while prevalence of psychiatric disorders surpassed in h-ESS compared to n-ESS. Comparing the different comorbidities among different races and ethnicities showed that generally the rate of chronic comorbidities is highest in White and Black individuals. For instance, the White individuals have the highest rates of metabolic (67.1% in h-ESS and 72.1% in n-ESS), pulmonary (65.0% in h-ESS and 68.0% in n-ESS) disorders and cancer (51.1% in h-ESS and 56.9% in n-ESS). The cardiovascular disorders were most common in Black individuals (69% in h-ESS and 73.7% in n-ESS) followed by White individuals (67.1% in h-ESS and 72.1% in n-ESS). Similarly, the renal disorders showed highest prevalence in Black individuals (20.2% in h-ESS and 24.1% in n-ESS) followed by White individuals (18.9% in h-ESS and 21.4% in n-ESS). Psychiatric disorders, insomnia, and sleep apnea on the other hand were most common in Hispanic individuals (Psychiatry disorders: 66.6% in h-ESS and 64.7% in n-ESS; Insomnia: 21% in h-ESS and 23% in n-ESS; Sleep apnea: 43.9% in h-ESS and 37.5% in n-ESS).
Table2.
Summary of comorbidities and medications prescribed for the different race and ethnicity sub cohorts.
| Comorbidities | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Race | White | Black | Hispanic | Native/Hawaiian | Asian | |||||
| ESS category | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 | ESS 0-10 | ESS 11-24 |
| Cardiovascular1 | 104,941 (72.1) | 79,446 (67.1) | 28,701 (73.7) | 33,555 (69.0) | 8,674 (55.9) | 10,039 (50.3) | 1,089 (67.6) | 1,206 (61.8) | 1,179 (51.2) | 1,426 (45.0) |
| Metabolic2 | 103,610 (71.2) | 79,447 (67.1) | 25,010 (64.2) | 29,121 (59.8) | 9,524 (61.4) | 11,687 (58.6) | 1,079 (66.9) | 1,178 (60.4) | 1,308 (56.8) | 1,703 (53.7) |
| Neurology3 | 33,104 (22.7) | 25,573 (21.6) | 9,228 (23.7) | 9,979 (20.5) | 3,250 (21.0) | 3,868 (19.4) | 376 (23.3) | 339 (17.4) | 363 (15.8) | 445 (14.0) |
| Psychiatry4 | 79,743 (54.8) | 67,798 (57.3) | 23,480 (60.3) | 30,385 (62.4) | 10,031 (64.7) | 13,287 (66.6) | 930 (57.7) | 1,139 (58.4) | 1,170 (50.8) | 1,659 (52.3) |
| Renal5 | 31,205 (21.4) | 22,393 (18.9) | 9,384 (24.1) | 9,819 (20.2) | 2,630 (17.0) | 3,012 (15.1) | 340 (21.1) | 321 (16.5) | 270 (11.7) | 294 (9.3) |
| Pulmonary6 | 98,978 (68.0) | 76,951 (65.0) | 23,858 (61.2) | 29,218 (60.0) | 9,412 (60.7) | 12,623 (63.3) | 1,052 (65.3) | 1,173 (60.2) | 1,238 (53.7) | 1,625 (51.2) |
| Cancer | 82,868 (56.9) | 60,505 (51.1) | 19,829 (50.9) | 22,190 (45.6) | 6,154 (39.7) | 7,072 (35.4) | 778 (48.3) | 754 (38.7) | 663 (28.8) | 736 (23.2) |
| Insomnia | 29,208 (20.1) | 21,943 (18.5) | 8,845 (22.7) | 10,290 (21.1) | 3,566 (23) | 4,187 (21) | 315 (19.5) | 349 (17.9) | 426 (18.5) | 591 (18.6) |
| Sleep Apnea | 51,836 (35.6) | 40,162 (33.9) | 12,565 (32.3) | 16,041 (33.0) | 5,812 (37.5) | 8,750 (43.9) | 600 (37.2) | 691 (35.4) | 805 (34.9) | 1,106 (34.9) |
| Sedative/Hypnotic/Antipsychotic Medications | ||||||||||
| Any Medications | 74,134 (50.9) | 62,377 (52.7) | 20,148 (51.7) | 25,274 (51.9) | 8,057 (51.9) | 10,743 (53.8) | 819 (50.8) | 955 (48.9) | 892 (38.7) | 1,234 (38.9) |
| Medications≥ 2 | 38,964 (26.7) | 32,068 (27.1) | 10,249 (26.3) | 12,144 (24.9) | 4,107 (26.4) | 5,512 (27.6) | 433 (26.8) | 431 (22.1) | 393 (17.0) | 503 (15.8) |
| SSRI7 | 48,926 (33.6) | 43,596 (36.8) | 13,655 (35.0) | 18,173 (37.3) | 5,538 (35.7) | 7,729 (38.7) | 562 (34.8) | 668 (34.2) | 593 (25.7) | 914 (28.8) |
| Benzodiazepine | 33,024 (22.7) | 25,741 (21.7) | 6,425 (16.5) | 7,286 (14.9) | 3,175 (20.4) | 4,230 (21.2) | 344 (21.3) | 338 (17.3) | 279 (12.1) | 328 (10.3) |
| SNRI8 | 16,399 (11.2) | 12,426 (10.5) | 4,808 (12.3) | 5,211 (10.7) | 1,914 (12.3) | 2,892 (14.4) | 178 (11.0) | 185 (9.4) | 173 (7.5) | 198 (6.2) |
| Tricyclic | 19,989 (13.7) | 15,463 (13.0) | 5,780 (14.8) | 6,667 (13.7) | 1,958 (12.6) | 2,235 (11.2) | 225 (13.9) | 216 (11.0) | 222 (9.6) | 274 (8.6) |
| Antipsychotic | 16,533 (11.3) | 13,374 (11.3) | 5,455 (14.0) | 6,305 (12.9) | 1,990 (12.8) | 2,414 (12.0) | 193 (11.9) | 197 (10.1) | 215 (9.3) | 224 (7.0) |
| Opioid | 5,124 (3.5) | 4,102 (3.5) | 832 (2.1) | 793 (1.6) | 281 (1.8) | 304 (1.5) | 55 (3.4) | 48 (2.5) | 23 (1.0) | 30 (0.9) |
Any of the following: Hypertension, Heart Disease, Other and Ill-Defined Heart Disease, Cardiovascular Disease including Hypertension, Coronary Atherosclerosis and other Heart Disease, Heart Failure, Peripheral Artery Disease, Congestive Heart Failure, Arrhythmia, Cardiomyopathy, Acute Cardiac Injury, Acute Myocardial Infarction
Any of the following: Hyperlipidemia, Diabetes, Diabetes Type Two, Diabetes Without/with Complication
Any of the following: Cerebrovascular Disease, Chronic Neurological Disease, Ischemic Stroke, Headache, Myalgia, Epilepsy, Parkinson’s, Alzheimer
Any of the following: Major Depressive Disorder, Anxiety, PTSD, Alcohol Dependence, Drug Dependence, Bipolar, Schizophrenia
Any of the following: kidney disease, chronic kidney disease, Nephrosis, Urinary Stone, Acute Kidney Failure, Chronic Kidney Failure, Dialysis
Any of the following: Obstructive Sleep Apnea, Chronic Lung Disease, Chronic Obstructive Pulmonary Disease, Smoking, Dyspnea, Cough, Bronchitis, Asthma, Acute Respiratory Failure, Pulmonary Heart Disease, Lower Respiratory Infection, Venous Thromboembolism, acute respiratory distress syndrome, Emphysema, Pulmonary Embolism, Pulmonary
Selective Serotonin Reuptake Inhibitors
Serotonin and Norepinephrine Reuptake Inhibitors
BMI = body mass index, CCI = Charlson Comorbidity Index, ESS = Epworth Sleepiness Scale, PTSD = post-traumatic stress disorder, SD = standard deviation, SNRI = serotonin and norepinephrine reuptake inhibitors, SSRI = selective serotonin reuptake inhibitors.
The proportions of those with any sedative/hypnotic/antipsychotic prescriptions were similar between the ESS categories in Black individuals (51.9% in h-ESS vs 51.7% in n-ESS) and Asian individuals (38.9% in h-ESS vs 38.7% in n-ESS). In contrast, proportions of those who were prescribed the medications slightly differed in the White individuals (52.7% in h-ESS vs. 50.9% in n-ESS), the Hispanic individuals (53.8% in h-ESS vs. 51.9% in n-ESS), and the Native/Hawaiian individuals (48.9% in h-ESS vs. 50.8% in n-ESS). The main driver of this difference in White and Hispanic individuals was higher prescription for selective serotonin reuptake inhibitors (36.8% in h-ESS vs. 33.6% in n-ESS in White individuals; 38.7% in h-ESS vs. 35.7% in n-ESS in Hispanic individuals) while among Native/Hawaiian individuals it appears to be other medications like benzodiazepines and tricyclic antidepressants. Among all, the highest prescription of medications was for Hispanic individuals in h-ESS.
Mortality prediction based on ESS:
Table 3 presents mortality data grouped by race and ethnicity and two exposure levels for ESS (n-ESS and h-ESS). For the Black, Hispanic, Native/Hawaiian, and Asian individuals, mortality rates were significantly higher in the n-ESS category compared to the h-ESS category (p < 0.001), while for the White group, there was no significant difference in mortality rates between the h-ESS and n-ESS categories (both categories had 19.3% rate of mortality). Figure 2 illustrates the all-cause mortality rates across various race and ethnicity groups, supporting the findings presented in Table 3. The adjusted odds ratios (aORs) of all-cause mortality significantly exceeded, h-ESS compared to n-ESS, in Whites (1.17; 95% CI: 1.15, 1.20) and Blacks (1.08; 95% CI: 1.03, 1.13). The aORs of all-cause mortality, comparing h-ESS to n-ESS, did not differ significantly in Asians (1.11; 95% CI: 0.85, 1.45), Hispanics (0.97; 95% CI: 0.89, 1.06) and Native/Hawaiian (0.87; 95% CI: 0.70, 1.07).
Table 3.
Race-modulated mortality rate, unadjusted and adjusted odds ratio for two categories of ESS
| Mortality Rate, N (%) | Unadjusted OR (95%CI) | Adj OR (95%CI) † | ||
|---|---|---|---|---|
| All races | ESS 0-10 | 36,906 (16.9) | Reference | Reference |
| ESS 11-24 | 32,167(15.7) | 0.91 (0.90,0.93) | 1.14 (1.12,1.16) | |
| White | ESS 0-10 | 28,186(19.3) | Reference | Reference |
| ESS 11-24 | 22,900(19.3) | 0.99 (0.98,1.01) | 1.17 (1.15,1.20) | |
| Black | ESS 0-10 | 4,799(12.3) | Reference | Reference |
| ESS 11-24 | 5,502(11.3) | 0.91 (0.87,0.95) | 1.08 (1.03,1.13) | |
| Hispanic | ESS 0-10 | 1,324(8.5) | Reference | Reference |
| ESS 11-24 | 1,461(7.3) | 0.85 (0.78,0.91) | 0.97 (0.89,1.06) | |
| Native/ Hawaiian | ESS 0-10 | 254(15.7) | Reference | Reference |
| ESS 11-24 | 227(11.6) | 0.7 (0.58,0.85) | 0.87 (0.70,1.07) | |
| Asian | ESS 0-10 | 124(5.3) | Reference | Reference |
| ESS 11-24 | 154(4.8) | 0.90 (0.70,1.14) | 1.11 (0.85,1.45) |
ESS = Epworth sleepiness score, OR (95%CI) = Odds ratio (95% confidence interval),
Adjusted by age, sex, BMI, and Charlson comorbidity index
Figure 2—
Mortality rates in different race and ethnicity groups, stratified by daily sleepiness.
Discussion:
In this large, nationally representative cohort of U.S. Veterans receiving care within an integrated healthcare system with minimal race-based access barriers, we investigated the association between subjective excessive daytime sleepiness (EDS), as measured by the Epworth Sleepiness Scale (ESS), and all-cause mortality across different racial and ethnic groups. After adjusting for age, sex, BMI, and comorbidities, EDS associated with increased odds of mortality among White and Black but not among Hispanic, Asian, or Native Hawaiian groups. These findings suggest that the relationship between EDS and mortality is not uniform across all racial and ethnic groups.
Prior research showed disparities between White and Black populations regarding health outcomes, often showing worse outcomes among Black individuals (20-22). In contrast, we demonstrated that the magnitude and direction of the association between EDS and mortality are similar between these two groups. This observation suggests that shared biological or aging-related vulnerabilities may play a more critical role than previously assumed structural or systemic disparities in shaping this particular outcome within the relatively equitable framework of the VA system (23, 24).
The link between EDS and increased odds of mortality in White and Black populations became evident only after adjusting for various confounding factors, highlighting the importance of accounting for variables like age, sex, BMI, and comorbidity burden (CCI). One possible explanation is the role of age in this relationship, as the mean age was higher among individuals with normal ESS scores compared to those with high ESS scores. Similarly, the prevalence of medical comorbidities was greater in the normal ESS group than in the high ESS group for both White and Black individuals. In contrast, the proportion of individuals with obesity (BMI ≥30) was higher in the high ESS group. Since older age, a higher burden of comorbid conditions, and lower BMI are independently associated with increased mortality risk(25), adjusting for these variables helps isolate the independent effect of EDS on mortality. Moreover, the overall profile of comorbidities, BMI distribution, and age patterns was generally similar between White and Black Veterans, suggesting that the observed association between EDS and mortality in these two groups may reflect shared underlying risk mechanisms.
Denney et al. (12) examined the interaction between race and ethnicity, different sleep durations, and all-cause mortality. They found that the risk of mortality differs among various race and ethnicities as in black individuals sleeping more than 9 hours, the adjusted hazard ratio was significant and 26% less than white population sleeping >9hours serving as the reference group. It is noteworthy that the results were adjusted for demographics, socioeconomic factors and severe mental illnesses but not for medical comorbidities. Further, the Southern Community Cohort Study found a U-shaped association between sleep duration and all-cause mortality in the White population. However, in the Black population, this association was weaker, even after adjusting for socioeconomic and sociodemographic factors such as education, household income, marital status, and employment status (26). In line with our results, a study by Petrov et al. assessed the association between sleep duration and risk of incident stroke in black and white races and found that in both races, sleep duration of >9 hours compared to normal sleep duration did not significantly associate with risk of stroke and the pattern of findings did not change event after adjusting for various sociodemographic and clinical variates. In contrast, stratifying race groups based on sex showed that white males sleeping >9 hours had significantly higher risk of stroke compared to those sleeping 7-9 hours, while white female, black male, and black female populations still did not show significant differences (27).
The absence of a significant association between EDS and mortality in Hispanic, Asian, and Native Hawaiian Veterans may generally reflect the small sample size of these groups in our cohort warranting further research. It may also reflect cultural differences in symptom perception, varying norms regarding fatigue, or healthcare engagement patterns (24, 28). Prior studies suggested that self-reported symptoms such as sleepiness can be underrecognized or underreported in certain cultural groups due to stigma, differences in health communication, or varying thresholds for concern(29). Additionally, measurement limitations may have influenced our findings. The ESS was developed primarily in White populations (30), and its cross-cultural validity may be limited in other racial or ethnic groups (31, 32). Future research incorporating culturally sensitive instruments or objective sleep measures could help clarify these findings. For Asian and Native Hawaiian Veterans in particular, the combination of limited statistical power and possible measurement bias warrants caution in interpretation.
Our findings are also consistent with broader mortality trends. According to CDC data, Black, White, and Native Hawaiian individuals have the highest overall mortality rates (33), while Asian and Hispanic populations have the lowest (34). Cause-specific mortality also varies: Non-Hispanic White individuals show higher rates of respiratory disease and Alzheimer’s disease, while non-Hispanic Black individuals have higher rates of heart disease, cancer, cerebrovascular diseases, influenza/pneumonia, diabetes, kidney disease, and septicemia. (35) These variations underscore the importance of considering both shared and distinct mechanisms underlying mortality risk in diverse populations.
EDS is often dismissed as a vague complaint, yet it may serve as an early marker of declining health, especially in aging populations. Our findings highlight the need for targeted screening and intervention strategies for EDS among White and Black Veterans. Given the known association of sleep health with cardiovascular, metabolic, and cognitive outcomes (36, 37), integrating EDS assessment into routine care could improve early detection of at-risk individuals.
From a health equity perspective, our study emphasizes the lack of disparities in the clinical outcomes among White and Black veterans referred to sleep clinics. This observation aligns with prior research on racial and ethnic disparities within the Veterans Health Administration (VHA), which has shown a lower prevalence of such disparities compared to those seen in the private healthcare sector (28, 38, 39). The VHA, through its statutory mandates and policies, ensures equal access to care and minimizes out-of-pocket costs for all eligible Veterans. As a system designed to serve all qualified Veterans, the VHA may help buffer against the types of bias that often drive racial disparities in other healthcare environments (19). These findings underscore the added value of stratifying by race/ethnicity, which uncovered differential associations masked in aggregate analyses. This is conceptually and methodologically distinct from our previous work (4) and addresses an important gap in the literature on sleep health disparities.
Limitations
Our study is subject to several limitations. First, the findings demonstrate associations, not causality. Second, although the ESS is a widely used clinical tool, it does not always align with objective sleep metrics, such as the Multiple Sleep Latency Test (MSLT). For example, a study from the Wisconsin Sleep Cohort found differing associations between self-reported sleepiness and objective hypersomnolence with depression: self-reported sleepiness was linked to increased odds of depression, while objective sleep propensity measured by the MSLT was associated with decreased odds of depression (40). Further, although our study reported the rate of insomnia and sleep apnea in different groups, other disorders such as restless legs syndrome and hypersomnolence may also influence sleepiness and warrant investigation in future studies. Moreover, over 90% of our sample was male, limiting the generalizability of findings to female populations. The lack of cause-specific mortality data and psychiatric diagnoses (e.g., depression or anxiety) further limits interpretability. In addition, psychiatric comorbidities and psychotropic medications were described but not included in our adjusted models; sedative, hypnotic, and antipsychotic medications were analyzed collectively, and future studies should examine them separately to clarify their distinct effects on daytime sleepiness. Also, we did not apply formal statistical corrections for multiple testing across racial/ethnic subgroups; therefore, some findings may be subject to type I error and should be interpreted with caution. Finally, the unequal representation of racial/ethnic groups—particularly the underrepresentation of Hispanic, Asian, and Native Hawaiian populations—limits the generalizability of our findings. Future studies should consider applying statistical weighting or oversampling to ensure adequate power to detect group-specific effects.
Conclusion
Our study provides novel insights into the role of race and ethnicity in the association between excessive daytime sleepiness and all-cause mortality. Data suggested that EDS associated with increased odds of mortality only among White and Black but not among Hispanic, Asian, or Native Hawaiian groups. These findings suggest that the relationship between EDS and mortality is not uniform across all racial and ethnic groups. Future research should aim to uncover the social and biological mechanisms that underlying these patterns and to refine sleep-related screening tools for diverse populations. Understanding how EDS functions as a mortality risk factor across racial and ethnic groups is essential to designing equitable and effective interventions in both clinical and public health.
High ESS is associated with higher mortality in White and Black Veterans
Results suggest the need for race-specific EDS screening in Veterans
Adjusting for age, sex, BMI, and comorbidities revealed the risk pattern
Further studies should explore social/biological causes and improve tools
No link found between EDS and mortality in Hispanic, Asian, or Native Hawaiian Veterans
Disclosure Statement:
This work was done in its entirety at Baylor College of medicine and Michael E. DeBakey VA Medical Center. All the authors have reviewed and approved the manuscript. This study was financially supported by National Institutes of Health, National Heart, Lung, and Blood Institute (NHLBI) K25 funding (1K25HL152006-01 [J.R.]); and the use of facilities and resources at the Center for Innovations in Quality, Effectiveness, and Safety (CIN 13-413), Michael E. DeBakey VA Medical Center, Houston, TX. The opinions expressed are those of the authors and not necessarily those of the Department of Veterans Affairs, the U.S. government, or Baylor College of Medicine. The authors do not report any off-label or investigational use, and no relevant conflict of interest related to the topic of this study. All the authors have reviewed and approved the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References:
- 1.Littner MR, Kushida C, Wise M, G. Davila D, Morgenthaler T, Lee-Chiong T, et al. Practice parameters for clinical use of the multiple sleep latency test and the maintenance of wakefulness test. Sleep. 2005;28(1):113–21. [DOI] [PubMed] [Google Scholar]
- 2.Slater G, Steier J. Excessive daytime sleepiness in sleep disorders. Journal of thoracic disease. 2012;4(6):608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wang L, Liu Q, Heizhati M, Yao X, Luo Q, Li N. Association between excessive daytime sleepiness and risk of cardiovascular disease and all-cause mortality: a systematic review and meta-analysis of longitudinal cohort studies. Journal of the American Medical Directors Association. 2020;21(12):1979–85. [DOI] [PubMed] [Google Scholar]
- 4.Maghsoudi A, Azarian M, Sharafkhaneh A, Jones MB, Nozari H, Kryger M, et al. Age modulates the predictive value of self-reported sleepiness for all-cause mortality risk: insights from a comprehensive national database of veterans. Journal of Clinical Sleep Medicine. 2024;20(11):1785–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Good CH, Brager AJ, Capaldi VF, Mysliwiec V. Sleep in the United States military. Neuropsychopharmacology. 2020;45(1):176–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Alexander M, Ray MA, Hébert JR, Youngstedt SD, Zhang H, Steck SE, et al. The national veteran sleep disorder study: descriptive epidemiology and secular trends, 2000–2010. Sleep. 2016;39(7):1399–410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Vinnikov D, Blanc PD, Alilin A, Zutler M, Holty J-EC. Fatigue and sleepiness determine respiratory quality of life among veterans evaluated for sleep apnea. Health and quality of life outcomes. 2017;15(1):1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mysliwiec V, Brock MS, Pruiksma KE, Straud CL, Taylor DJ, Hansen S, et al. A comprehensive evaluation of insomnia, obstructive sleep apnea and comorbid insomnia and obstructive sleep apnea in US military personnel. Sleep. 2022;45(12):zsac203. [DOI] [PubMed] [Google Scholar]
- 9.Foster SN, Hansen SL, Capener DC, Matsangas P, Mysliwiec V. Gender differences in sleep disorders in the US military. Sleep health. 2017;3(5):336–41. [DOI] [PubMed] [Google Scholar]
- 10.Capener DC, Brock MS, Hansen SL, Matsangas P, Mysliwiec V. An initial report of sleep disorders in women in the US Military. Military medicine. 2018;183(9-10):e266–e71. [DOI] [PubMed] [Google Scholar]
- 11.Carnethon MR, De Chavez PJ, Zee PC, Kim K-YA, Liu K, Goldberger JJ, et al. Disparities in sleep characteristics by race/ethnicity in a population-based sample: Chicago Area Sleep Study. Sleep medicine. 2016;18:50–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Denney JT, Zamora-Kapoor A, Hansen DA, Whitney P. Race/ethnicity, sleep duration, and mortality risk in the United States. SSM-Population Health. 2023;21:101350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Sohn M-W, Arnold N, Maynard C, Hynes DM. Accuracy and completeness of mortality data in the Department of Veterans Affairs. Population health metrics. 2006;4(1):1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ramanathan D. VIReC Factbook: corporate data Warehouse (CDW) vital sign 1.1 domain. Hines, IL. 2018. [Google Scholar]
- 15.Johns MW. Reliability and factor analysis of the Epworth Sleepiness Scale. Sleep. 1992;15(4):376–81. [DOI] [PubMed] [Google Scholar]
- 16.Finucane MM, Stevens GA, Cowan MJ, Danaei G, Lin JK, Paciorek CJ, et al. National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9· 1 million participants. The lancet. 2011;377(9765):557–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. Journal of clinical epidemiology. 1992;45(6):613–9. [DOI] [PubMed] [Google Scholar]
- 18.Huang Y-q, Gou R, Diao Y-s, Yin Q-h, Fan W-x, Liang Y-p, et al. Charlson comorbidity index helps predict the risk of mortality for patients with type 2 diabetic nephropathy. Journal of Zhejiang University Science B. 2014;15:58–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Razjouyan J, Helmer DA, Li A, Naik AD, Amos CI, Bandi V, et al. Differences in COVID-19-related testing and healthcare utilization by race and ethnicity in the veterans health administration. Journal of racial and ethnic health disparities. 2022:1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bailey ZD, Krieger N, Agénor M, Graves J, Linos N, Bassett MT. Structural racism and health inequities in the USA: evidence and interventions. The lancet. 2017;389(10077):1453–63. [DOI] [PubMed] [Google Scholar]
- 21.Williams DR, Mohammed SA. Discrimination and racial disparities in health: evidence and needed research. Journal of behavioral medicine. 2009;32:20–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nelson A. Unequal treatment: confronting racial and ethnic disparities in health care. Journal of the national medical association. 2002;94(8):666. [PMC free article] [PubMed] [Google Scholar]
- 23.Trivedi AN, Grebla RC, Wright SM, Washington DL. Despite improved quality of care in the Veterans Affairs health system, racial disparity persists for important clinical outcomes. Health affairs. 2011;30(4):707–15. [DOI] [PubMed] [Google Scholar]
- 24.Saha S, Freeman M, Toure J, Tippens KM, Weeks C, Ibrahim S. Racial and ethnic disparities in the VA health care system: a systematic review. Journal of general internal medicine. 2008;23:654–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sakr Y, Elia C, Mascia L, Barberis B, Cardellino S, Livigni S, et al. Being overweight or obese is associated with decreased mortality in critically ill patients: a retrospective analysis of a large regional Italian multicenter cohort. Journal of critical care. 2012;27(6):714–21. [DOI] [PubMed] [Google Scholar]
- 26.Xiao Q, Blot WJ, Matthews CE. Weekday and weekend sleep duration and mortality among middle-to-older aged White and Black adults in a low-income southern US cohort. Sleep Health. 2019;5(5):521–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Petrov ME, Howard G, Grandner MA, Kleindorfer D, Molano JR, Howard VJ. Sleep duration and risk of incident stroke by age, sex, and race: the REGARDS study. Neurology. 2018;91(18):e1702–e9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Peterson K, Anderson J, Boundy E, Ferguson L, McCleery E, Waldrip K. Mortality disparities in racial/ethnic minority groups in the Veterans Health Administration: an evidence review and map. American Journal of Public Health. 2018;108(3):e1–e11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Jean-Louis G, Magai CM, Cohen CI, Zizi F, von Gizycki H, DiPalma J, et al. Ethnic differences in self-reported sleep problems in older adults. Sleep. 2001;24(8):926–33. [DOI] [PubMed] [Google Scholar]
- 30.Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep. 1991;14(6):540–5. [DOI] [PubMed] [Google Scholar]
- 31.Bindi P, Bayles A, Acquadro C, Johns M. Challenges in translating the Epworth Sleepiness Scale into 66 languages. Value in Health. 2016;19(3):A93. [Google Scholar]
- 32.Gelaye B, Lohsoonthorn V, Lertmeharit S, Pensuksan WC, Sanchez SE, Lemma S, et al. Construct validity and factor structure of the pittsburgh sleep quality index and epworth sleepiness scale in a multi-national study of African, South East Asian and South American college students. PloS one. 2014;9(12):e116383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Dwyer-Lindgren L, Kendrick P, Kelly YO, Baumann MM, Compton K, Blacker BF, et al. Cause-specific mortality by county, race, and ethnicity in the USA, 2000–19: a systematic analysis of health disparities. The Lancet. 2023;402(10407):1065–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hummer RA, Benjamins MR, Rogers RG. Racial and ethnic disparities in health and mortality among the US elderly population. Critical perspectives on racial and ethnic differences in health in late life. 2004:53–94. [Google Scholar]
- 35.Anderson NB, Bulatao RA, Cohen B, on Race P, Council NR. Racial and ethnic disparities in health and mortality among the US elderly population. Critical perspectives on racial and ethnic differences in health in late life: National Academies Press (US); 2004. [PubMed] [Google Scholar]
- 36.Mehrdad M, Azarian M, Sharafkhaneh A, Alavi A, Zare R, Rad AH, et al. Association between poor sleep quality and glycemic control in adult patients with diabetes referred to Endocrinology Clinic of Guilan: a cross-sectional study. International Journal of Endocrinology and Metabolism. 2021;20(1):e118077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Grandner MA, Jackson NJ, Pak VM, Gehrman PR. Sleep disturbance is associated with cardiovascular and metabolic disorders. Journal of sleep research. 2012;21(4):427–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Burgess DJ, Van Ryn M, Grill J, Noorbaloochi S, Griffin JM, Ricards J, et al. Presence and correlates of racial disparities in adherence to colorectal cancer screening guidelines. Journal of general internal medicine. 2011;26:251–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Selim AJ, Fincke G, Berlowitz DR, Cong Z, Miller DR, Ren XS, et al. No racial differences in mortality found among Veterans Health Administration out-patients. Journal of clinical epidemiology. 2004;57(5):539–42. [DOI] [PubMed] [Google Scholar]
- 40.Plante DT, Finn LA, Hagen EW, Mignot E, Peppard PE. Subjective and objective measures of hypersomnolence demonstrate divergent associations with depression among participants in the Wisconsin Sleep Cohort Study. Journal of Clinical Sleep Medicine. 2016;12(4):571–8. [DOI] [PMC free article] [PubMed] [Google Scholar]