Disparities in Sleep Duration and Quality by Industry of Employment and Occupational Class among Native Hawaiian/Pacific Islanders and Non-Hispanic Whites in the United States

Abstract

Objective:

To investigate disparities in the work-sleep relationship between Native Hawaiian/Pacific Islanders (NHPIs) and Non-Hispanic (NH)-White populations.

Methods:

Using data from a nationally representative sample of US adults (n = 20,828) in the 2014 National Health Interview Survey, we estimated prevalence of short sleep duration (<7 hours) among NHPIs (10%) and NH-Whites for each of 7 employment industry categories and 3 occupational classes. Mean age was 41±0.5 years for NHPIs and 49±0.2 years for NH-Whites. Women comprised 52% of both groups.

Results:

NHPIs were more likely than NH-Whites to report short sleep duration across all industry of employment categories (except for food and accommodation services) and occupational classes. The disparity was widest among NHPI and NH-White workers in the ‘professional/management industry’ category, with NHPIs having higher prevalence of very short (<6 hours; 20% vs. 7%) and short sleep (30% vs. 22%) durations and lower prevalence of recommended sleep duration (45% vs. 68%) and waking up feeling rested (53% vs. 67%). Among the occupational classes, the NHPI-White disparity was widest among participants who held support service occupations. Although professionals had the lowest and laborers had the highest prevalence of short sleep among the three occupational classes in both NHPI and NH-White groups, short sleep duration prevalence was higher among NHPI professionals (35%) than NH-White laborers (33%). NH-White workers across industry and occupational classes had higher sleep medication use prevalence compared to NHPI workers.

Conclusions:

The work environment may contribute to racial/ethnic disparities in short sleep. Further investigations are warranted.

INTRODUCTION

Habitual short sleep duration (<6–7 hours) has been associated with adverse health outcomes including obesity, type 2 diabetes, hypertension, cardiovascular disease, and mortality.^1–7 Sleep problems are common, with approximately one-third of adults in the United States reporting insufficient sleep.^8,9 Studies have increasingly demonstrated racial/ethnic disparities in sleep duration and quality, with non-Hispanic (NH) Blacks, Hispanics/Latinos, and Asians having worse sleep health compared to NH-Whites.^10–13 Of particular concern, the racial/ethnic disparity in sleep health appears to be increasing in recent years.¹⁴ Although the literature is limited, prior studies have demonstrated that racial/ethnic disparities in sleep health among self-identified Native Hawaiian/Pacific Islanders (NHPI) may be similar to disparities observed among NH-Blacks, with up to 40% of both groups reporting short sleep duration compared to 29% of NH-Whites in a nationally representative study.^15,16 Additionally, inadequate sleep duration could potentially exacerbate the disparities observed in chronic diseases, such as type 2 diabetes, hypertension, and obesity observed among NHPIs.^17–19

Myriad causes have been postulated to explain racial/ethnic differences in sleep and its impact on health outcomes. For example, exposure to discriminatory policies and practices influencing the physical and social neighborhood, as well as home sleep environments all potentially play an important role.^20–22 Job-related stressors such as extended work hours, low job control, and workplace discrimination based on factors like race and ethnicity have also been linked to negative sleep outcomes, likely due to their impact on psychological well-being.^23–26 Additionally, poor and insufficient sleep greatly contributes to loss of work productivity,²⁷ potentially widening the racial/ethnic and socioeconomic inequities that contribute to poor health outcomes. Furthermore, occupational factors, including occupation class and industry of employment, have been identified to have particularly important influence.¹² For example, National Health Interview Survey (NHIS) data revealed that those in the management, manufacturing, and transportation/warehousing industries have a higher prevalence of short-sleep duration compared to those in other fields.²⁸ Racial/ethnic disparities in sleep duration have been found between employment industry and occupational classes, with the prevalence of short sleep duration decreasing among NH-Whites but increasing among NH-Blacks and Latinos as professional status increased.^12,29 While similar temporal trends were not observed between Asians and NH-Whites, Asians in professional and support services, but not laborer occupations, were found to have higher prevalence of short sleep compared to NH-Whites.¹¹

Although there is a growing knowledge for how occupational factors may contribute to racial/ethnic disparities in sleep health, their impact on potential disparities between NHPI and NH-Whites has not been investigated. To address this gap, we used two large, nationally representative datasets to examine the association between occupation and industry of employment and the prevalence of insufficient sleep and poor sleep quality among NHPI compared to NH-Whites.

METHODS

National Health Interview Survey

We analyzed data from the NHIS, a series of cross-sectional surveys that use a three-stage stratified cluster probability sampling design to conduct in-person interviews in the households of nationally representative, non-institutionalized U.S. civilians. A detailed description of NHIS procedures has been published elsewhere.²⁹ Briefly, a probability sample of households was identified from which participants were interviewed by trained interviewers from the U.S. Census Bureau National Center for Health Statistics (NCHS) to obtain member household information about health and sociodemographic characteristics on a continual weekly basis. A randomly selected adult and child (not included in this analysis) provided more specific health-related information. Data were collected using computer-assisted personal interviewing. The NCHS received written informed consent from each study participant. Each study participant provided informed consent to the NHIS, and the National Institute of Environmental Health Sciences’ Institutional Review Board waived approval for publicly available, secondary data with no identifiable information.

Study population

In 2014, the NCHS conducted the Native Hawaiian and Pacific Islander (NHPI) NHIS from approximately 3,000 NHPI households located predominantly in Hawaii (38.7%) and California (22%), with 39.3% in all other states.²⁸ The current study focuses on these adults (at least 18 years old) and a separate comparison sample of self-identified NH-White adults in the 2014 NHIS survey. Participants were excluded if they did not have information or were missing data on industry and occupation of their employment, or self-reported sleep characteristics; non-US-born White participants were excluded because evidence suggests that sleep patterns among US immigrants differ from those among individuals born in the United States.²⁹ Non-US-born NHPI participants (18%) were not excluded in order to maximize the limited sample size. Our final sample consisted of 2,134 (82.39%) NHPI [456 (17.61%) NHPI participants excluded] and 18,694 (39.70%) NH-White [28,399 (60.30%) NH-White participants excluded] adults. The excluded NHPI adults were more likely to be younger (approximately 4 years), have lower educational attainment, be single, have household income <$35,000, live in poverty, and be unemployed (Supplemental Table A.1). The excluded NH-White adults were more likely to be younger (approximately 1 year), have lower educational attainment, be single, have household income <$35,000, live in poverty, receive government assistance, be unemployed, and have poorer health (Supplemental Table A.2). The conditional response rate for the sample adult component was 82.8% for the 2014 NHPI-NHIS, and 80.5% for the 2014 NHIS.

Measures

Sleep duration and sleep difficulties

Participants reported their average sleep hours in a 24-hour period. Interviewers rounded the values ≥30 minutes up and <30 minutes down to the nearest hour. Sleep duration was categorized as very short (<6 hours), short (6 hours), recommended (7–8 hours) or long (≥9 hours).³⁰ Seven-to-eight hours of sleep was used as the reference because it is associated with the lowest levels of morbidity and mortality.³¹ Additionally, participants reported how often they had trouble falling and staying asleep, how many days they woke up feeling rested, and how often they used sleep medication to help them fall or stay asleep. Participants were considered to have sleep difficulties if they indicated “trouble falling asleep 3 or more days”, “trouble staying asleep 3 or more days”, “waking up most days (at least 4) feeling rested”, and “took sleep medication 3 or more days” (all of these in the previous week). Three or more days were used to indicate potential insomnia symptoms.³² Participants were considered to have “poor sleep” quality if they indicated unrecommended, short, or long sleep duration, or any of the above sleep difficulties.

Industry of Employment and Occupation

The NHPI dataset included 79 North American Industrial Classification System (NAICS) codes (https://nhis.ipums.org/nhis/14ind.shtml) for currently employed adults or those who had ever worked. These options had already been combined into 14 industry categories in the NHPI-NHIS sample, so we created a final common recode for the two samples with 7 meaningful industry categories: 1) “manufacturing, construction, mining, oil and gas extraction, utilities, agriculture, animal production, forestry, fishing, hunting”; 2) “wholesale/retail”; 3) “professional, management, administration, finance/rental/leasing, information/data services, scientific and technical services, publishing, motion picture and sound, broadcasting, telecommunications”; 4) “education services”; 5) “health care / social assistance services”; 6) “food and accommodation services”; 7) “public administration”. Not included in analyses were the general category of “other” industries, armed forces, performing arts, museums, gambling, or recreation industries.

Occupations were coded using the 2000 Standard Occupation Classification System from the U.S. Census Bureau and the Bureau of Labor Statistics (http://www.bls.gov/soc/). The 23 groupings in the NHIS were combined into 17 categories in the NHPI-NHIS, and a common set of 3 occupational categories was created: 1) “professional”, including management, business, finance, computers, math, engineering, science, architecture; 2) “service” occupations, including community, social, and legal services, education, health care, personal care, sales, office and administrative support; and 3) “labor” occupations, including protective services, food and serving, building, grounds cleaning, farming, fishing, forestry, construction, extraction, installation, repair, maintenance, production, moving, and transportation.

Demographic and Clinical Characteristics

Race/ethnicity. Participants self-identified their race and ethnicity (Hispanic/Latino or not). Participants were categorized into one or more racial/ethnic groups (e.g., American Indian, Alaskan Native, Asian Indian, Black/African American, NH-White, Native Hawaiian, Samoan, Guamanian or Chamorro, Chinese, to name some of the 16 categories). The categories also included ‘other Asian’, ‘other Pacific Islander’, and ‘other race’, and details for each of these were obtained. Participants were coded as NHPI based on their self-identification or mixed-NHPI, if they self-identified as another race/ethnicity, and “other Pacific Islander” with details about the country of origin. Given the small sample size of NHPI and mixed-NHPI, we combined the two categories to make a single NHPI category. The NHPI participants were more likely to be middle-aged, married, living in poverty, and have lower educational attainment but did not differ significantly from the mixed-NHPI participants on sex/gender, household income, government assistance, employment status, health status, or sleep duration (Supplemental Table A.3). NH-White adults represent the dominant culture and the most socially advantaged racial group,³³and had the largest sample size (sample consisted of 63,128 (57.3%) NH-White [US-born and not US-born], 24,455 (22.2%) Hispanic, 14,958 (13.6%) NH-Black, and 7,574 (6.9%) Asian participants) and therefore were chosen as a comparison group for statistical stability.

The following sociodemographic and clinical characteristics of participants were selected to describe the study populations and due to their associations with sleep health based on prior literature.^15,16,34

Socioeconomic and marital status. Indicators of socioeconomic status (SES) included educational attainment and annual household income. Educational attainment was categorized as “<high school”, “high school graduate”, “some college” (including, e.g., vocational training, associate’s degree), and “≥college graduate”. Annual household income was dichotomized as <$35,000 versus ≥$35,000 to represent low income; the income variable in both datasets is categorical rather than continuous, with <$35,000 being the lowest grouping. Marital status was categorized as married/living with a partner, divorced/separated/widowed, or never married. Government assistance was coded ‘yes’ if any family member received welfare, food stamps, SNAP, or WIC benefits. Living in poverty was coded ‘yes’ if the ratio of the family income to the poverty threshold was less than 1.0.

Clinical characteristics. Self-reported height and weight were used to calculate body mass index (BMI), which was recoded into “overweight/obese” (BMI>=25 kg/m²) versus “underweight/normal” (BMI<25 kg/m²).³⁵ Self-reported general health status was categorized as “excellent/very good,” “good,” or “fair/poor.” Smoking status was characterized as “never”, “former”, and “current,” and alcohol consumption as “current” versus “never/former.” Current leisure time physical activity had three levels based on 2008 federal guidelines: “no activity”, “insufficiently active” (<150 minutes/week), and “sufficiently active” (150 minutes or more/week). Participants also reported whether they were ever told by a doctor/health professional that they had hypertension, diabetes, stroke, or heart disease. Serious psychological distress was defined as a score ≥13 on the 6-item Kessler distress scale (range 0–24).³⁶

Statistical analysis

We conducted separate analyses for the NHPI and NH-White participants. Since the two surveys were conducted separately and each had its own sampling protocol, each dataset had its own stratification and sampling weight variables that accounted for the unequal probabilities of selection resulting from the sample design, nonresponse, and oversampling of certain subgroups (i.e., minoritized races/ethnicities; older adults). Standard errors or variance estimates were calculated by using Taylor series linearization.³⁷ Stata, version 15,³⁸ software was used for all analyses.

Categorical variables are presented using raw (unweighted) frequencies accompanied by the weighted percentages; additionally, all percentages are standardized to the age structure of the 2010 Census (with 2 age groups: 18–49, 50+) to account for potential age-related variations in sleep. Although the two sample groups could not be analyzed together, it is possible to compare percentage values between the two groups using z-score testing (https://www.cdc.gov/nchs/nhis/nhpi/faqs.htm).^39,40

We examined the age-standardized prevalence of short sleep duration (<7 hours), stratified by various demographic and clinical characteristics in NH-White and NHPI participants, and tested whether these two groups significantly differed in each prevalence (e.g., prevalence of short sleep among former smokers, in NH-White versus NHPI participants). We also examined age-standardized prevalence of sleep duration categories and sleep quality indicators in NH-White and NHPI participants, which were stratified by industry groups (Table 2) and occupation groups (Table 3) and again tested for significant differences between NPHI versus NH-White participants. Further, we repeated the analyses described above by excluding unemployed participants and found that inferences of the results did not change, thus, these results are presented in the supplemental tables (see supplemental Tables A.4-A.6). P-values <0.05 were considered significant.

Table 2.

Age-Standardized Sleep Duration and Quality Indicators Prevalence among US White Participants (N=18,694) and Native Hawaiian/Pacific Islanders (N=2,134) by Industry Groups, National Health Interview Survey, 2014^a

	Manufacturing/ etc.		Wholesale/ Retail		Professional / etc.		Education		Health		Food / accommodation		Public administration
	%b	SE	%	SE	%	SE	%	SE	%	SE	%	SE	%	SE
	95% CI		95% CI		95% CI		95% CI		95% CI		95% CI		95% CI
NHPI
	N = 539		N = 284		N = 414		N = 208		N = 254		N = 207		N = 228
Sleep duration
<6 hours	13.91	3.40	16.51 d	6.46	20.04 **	4.80	15.59 *	5.28	8.78	2.12	8.05	2.62	10.29	2.02
	(6.9, 20.9)		(3.3, 29.8)		(10.2, 29.9)		(4.8, 26.4)		(4.4, 13.1)		(2.7, 13.4)		(6.1, 14.4)
6-<7 hours	34.07 **	3.90	26.13	3.32	30.13 *	3.21	25.04	4.35	38.29 **	4.90	25.30	5.38	32.28 *	4.12
	(26.1, 42.1)		(19.3, 32.9)		(23.6, 36.7)		(16.1, 34.0)		(28.2, 48.4)		(14.3, 36.3)		23.8, 40.7)
7–9 hours	47.54 ***	3.72	54.71	6.18	45.40 ***	3.92	56.39 ***	4.25	51.05 **	4.71	58.67	5.47	52.65 *	5.47
	(39.9, 55.2)		(42.0, 67.4)		(37.4, 53.4)		(47.7, 65.1)		(41.4, 60.7)		(47.5, 69.9)		(41.4, 63.9)
>9 hours	4.48	1.64	2.64 *	0.83	4.43	1.37	2.99	1.59	1.87	0.96	7.98	2.84	4.78	2.57
	(1.1, 7.8)		(1.0, 4.3)		(1.6, 7.2)		(0.0, 6.2)		(0.0, 3.8)		(2.1, 13.8)		(0.0, 10.1)
Trouble falling asleep (≥3 days/week)	21.96	3.48	13.03 **	2.66	24.94	2.82	33.81 **	5.73	23.07	3.33	13.55 ***	2.71	19.90	2.80
	(14.8, 29.1)		(7.6, 18.5)		(19.2, 30.7)		(22.1, 45.6)		(16.2, 29.9)		(8.0, 19.1)		(14.2, 25.6)
Trouble staying asleep (≥3 days/week)	31.25	3.25	18.71 ***	3.46	29.11	4.43	32.90	5.20	30.81	5.40	20.59 ***	4.44	26.55	4.19
	(24.6, 37.9)		(11.6, 25.8)		(20.0, 38.2)		(22.2, 43.6)		(19.7, 41.9)		(11.5, 29.7)		(17.9, 35.2)
Woke rested most days (≥4 days/w eek)	63.00	3.77	57.39	5.97	53.41 ***	3.39	61.57	4.56	56.09	4.47	61.35	6.64	64.24	5.4 7
	(55.3, 70.7)		(45.1, 69.6)		(46.5, 60.4)		(52.2, 70.9)		(46.9, 65.3)		(47.7, 75.0)		(53.0, 75.5)
Sleep medication (≥3 days/week)	7.72	2.69	2.34 ***	0.99	9.84	2.57	5.39 ***	1.28	6.03 ***	1.61	4.91 ***	2.25	4.99	2.36
	(2.2, 13.2)		(0.3, 4.4)		(4.6, 15.1)		(2.8, 8.0)		(2.7, 9.3)		(0.3, 9.5)		(0.2, 9.8)
Poor sleep c	70.79 ***	2.80	67.27	4.65	73.80 ***	3.44	67.48 *	4.26	70.15	3.75	64.37	4.85	67.93 *	4.47
	(65.0, 76.5)		(57.7, 76.8)		(66.7, 80.9)		(58.7, 76.2)		(62.5, 77.8)		(54.4, 74.3)		(58.8, 77.1)
Non-Hispanic White
	n = 5,120		n = 2,634		n = 3,740		n = 2,275		n = 2,576		n = 1,157		n = 1,192
Sleep duration
<6 hours	9.11	0.54	8.19	0.77	7.26	0.72	4.77	0.61	6.73	0.62	12.77	1.69	7.23	1.03
	(8.0, 10.2)		(6.7, 9.7)		(5.8, 8.7)		(3.6, 6.0)		(5.5, 8.0)		(9.4, 16.1)		(5.2, 9.3)
6-<7 hours	21.52	0.89	20.91	1.07	21.80	1.01	18.51	1.15	22.73	1.28	23.84	2.22	22.77	1.86
	(19.8, 23.3)		(18.8, 23.0)		(19.8, 23.8)		(16.3, 20.8)		(20.2, 25.3)		(19.5, 28.2)		(19.1, 26.4)
7–9 hours	65.23	1.06	65.70	1.29	68.41	1.14	73.00	1.30	67.07	1.43	55.33	2.71	67.03	1.98
	(63.1, 67.3)		(63.2, 68.2)		(66.2, 70.7)		(70.5, 75.6)		(64.3, 69.9)		(50.0, 60.7)		(63.1, 70.9)
>9 hours	4.15	0.32	5.20	0.65	2.52	0.36	3.71	0.54	3.47	0.55	8.06	1.83	2.97	0.59
	(3.5, 4.8)		(3.9, 6.5)		(1.8, 3.2)		(2.6, 4.8)		(2.4, 4.6)		(4.5, 11.7)		(1.8, 4.1)
Trouble falling asleep (≥3 days/week)	18.05	0.76	22.28	1.11	19.24	1.03	16.09	0.99	22.52	1.11	30.17	2.34	17.40	1.51
	(16.6, 19.5)		(20.1, 24.5)		(17.2, 21.3)		(14.2, 18.0)		(20.3, 24.7)		(25.6, 34.8)		(14.4, 20.4)
Trouble staying asleep (≥3 days/week)	29.07	0.95	30.78	1.14	28.82	0.97	24.01	1.12	30.71	1.26	37.88	2.41	25.79	1.81
	(27.2, 30.9)		(28.5, 33.0)		(26.9, 30.7)		(21.8, 26.2)		(28.2, 33.2)		(33.1, 42.6)		(22.2, 29.4)
Woke rested most days (≥4 days/week)	64.41	1.01	62.92	1.30	67.00	1.12	67.60	1.23	61.91	1.54	57.65	2.50	66.83	1.75
	(62.4, 66.4)		(60.4, 65.5)		(64.8, 69.2)		(65.2, 80.0)		(58.9, 64.9)		(52.7, 62.6)		(63.4, 70.3)
Sleep medication (≥3 days/week)	9.70	0.62	10.34	0.76	9.97	0.77	10.54	0.87	13.05	0.95	14.50	1.76	9.94	1.07
	(8.5, 10.9)		(8.9, 11.8)		(8.4, 11.5)		(8.8, 12.3)		(11.2, 14.9)		(11.0, 18.0)		(7.8, 12.1)
Poor sleep c	60.76	0.92	62.87	1.23	58.14	1.18	57.98	1.40	64.15	1.62	70.52	2.23	57.12	1.92
	(58.9, 62.6)		(60.4, 65.3)		(55.8, 60.5)		(55.2, 60.7)		(61.0, 67.3)		(66.1, 74.9)		(53.3, 60.9)

CI, confidence interval; NHPI, Native Hawaiian Pacific Islander; SE, standard error;

^aWhite participants are from National Health Interview Survey (NHIS) 2014 survey, and Native Hawaiian/Pacific Island (NHPI) participants are from special NHIS-NHPI survey in 2014 as well.

^bBolded percentages are significantly different between NHPI and White samples.

^cPoor sleep is defined by non-optimal duration (<7 hours or >9 hours) or one or more sleep quality problems (trouble falling asleep, staying asleep, not waking rested, using sleep medications, 3 or more days)

^dItalicized percentages have a Relative Standard Error >30%

Comparing NHPI versus Non-Hispanic White:

^*p<0.05

^**p<0.01

^***p<0.001

Table 3.

Age-Standardized Sleep Duration and Quality Indicators Prevalence among US White Participants (N=18,694) and Native Hawaiian/Pacific Islanders (N=2,134) by Occupation Groups, National Health Interview Survey, 2014^a

	Professional / management, etc.			Support Service			Labor
	%b	SE	95% CI	%	SE	95% CI	%	SE	95% CI
	NHPI
	N= 359			N= 945			N= 830
Sleep duration
<6 hours	10.81 *	1.75	(7.2, 14.4)	14.30 **	2.34	(9.5, 19.1)	14.58 **	1.37	(11.8, 17.4)
6-<7 hours	23.94	2.76	(18.3, 29.6)	32.94 ***	2.41	(28.0, 37.9)	31.08 ***	2.43	(26.1, 36.1)
7–9 hours	63.07 *	3.17	(56.6, 69.6)	49.62 ***	2.41	(44.7, 54.6)	48.54 ***	2.66	(43.1, 54.0)
>9 hours	2.17 d	0.97	(0.2, 4.2)	3.13	0.61	(1.9, 4.4)	5.80	1.54	(2.6, 9.0)
Trouble falling asleep (≥3 days/week)	18.77	2.50	(13.6, 23.9)	21.12	1.64	(17.8, 24.5)	23.19	2.44	(18.2, 28.2)
Trouble staying asleep (≥3 days/week)	21.06	3.63	(13.6, 28.5)	28.61	2.10	(24.3, 32.9)	29.63	2.11	(25.3, 34.0)
Woke rested most days (≥4 days/week)	61.57 *	2.82	(55.8, 67.4)	58.30 **	2.15	(53.9, 62.7)	60.10	2.68	(54.6, 65.6)
Sleep medication (≥3 days/week)	5.38 *	1.01	(3.3, 7.5)	4.98 ***	0.98	(3.0, 7.0)	8.41	2.25	(3.8, 13.0)
Poor sleep c	61.61	3.51	(54.4, 68.8)	69.98 ***	2.05	(65.8, 74.2)	70.72 **	2.29	(66.0, 75.4)
			Non-Hispanic White
	N= 4,126			N= 8,927			N= 5,641
Sleep duration
<6 hours	6.40	0.54	(5.3, 7.5)	6.69	0.37	(6.0, 7.4)	10.51	0.56	(9.4, 11.6)
6-<7 hours	21.15	0.87	(19.4, 22.9)	20.54	0.58	(19.4, 21.7)	22.62	0.83	(21.0, 24.2)
7–9 hours	70.00	0.98	(68.1, 71.9)	68.67	0.71	(67.3, 70.1)	61.52	1.05	(59.5, 63.6)
>9 hours	2.46	0.28	(1.9, 3.0)	4.09	0.31	(3.5, 4.7)	5.35	0.62	(4.1, 6.6)
Trouble falling asleep (≥3 days/week)	15.90	0.83	(14.3, 17.5)	20.56	0.61	(19.4, 21.8)	21.80	0.85	(20.1, 23.5)
Trouble staying asleep (≥3 days/week)	27.01	0.99	(25.1, 29.0)	28.86	0.65	(27.6, 30.1)	30.84	1.07	(28.7, 32.9)
Woke rested most days (≥4 days/week)	68.69	0.97	(66.8, 70.6)	64.92	0.72	(63.5, 66.3)	61.36	1.02	(59.3, 63.4)
Sleep medication (≥3 days/week)	7.98	0.49	(7.0, 8.9)	11.32	0.51	(10.3, 12.3)	11.19	0.66	(10.0, 12.5)
Poor sleep c	56.41	1.09	(54.3, 58.5)	60.83	0.74	(59.4, 62.3)	63.90	0.99	(62.0, 65.8)

SE, standard error; CI, confidence interval; NHPI, Native Hawaiian Pacific Islander

^aWhite participants are from National Health Interview Survey (NHIS) 2014 survey, and Native Hawaiian/Pacific Island (NHPI) participants are from special NHIS-NHPI survey in 2014 as well.

^bBolded percentages are significantly different between NHPI and White samples.

^cPoor sleep is defined by non-optimal duration (<7 hours or >9 hours) or one or more sleep quality problems (trouble falling asleep, staying asleep, not waking rested, using sleep medications, 3 or more days)

^dItalicized percentages have a Relative Standard Error >30%

Comparing NHPI versus Non-Hispanic White:

^*p<0.05

^**p<0.01

^***p<0.001

RESULTS

Study Population Characteristics

The NHPI participants (N=2,134) included 48% NHPI-only and 52% NHPI-mixed race. NHPI were approximately 8 years younger on average (41 years vs. 49 years) than NH-Whites (N=18,694). The age-standardized prevalence of short sleep (<7 hours) among NHPI participants compared to NH-Whites was consistently higher across a wide range of demographic and clinical strata (see Table 1). The few nonsignificant exceptions included having received government assistance, and among participants who suffered a stroke. Generally, across cardiometabolic health outcomes (other than stroke), and for general health status categories, leisure-time activity categories, and the psychological distress strata, short sleep was more prevalent for the deleterious or negative outcome within each group. For example, NH-Whites with hypertension had a higher prevalence of short sleep (33%) than NH-Whites without hypertension (28%). Similarly, NHPIs with hypertension had a higher prevalence of short sleep (50%) than NHPIs without hypertension (42%).

Table 1.

Age-Standardized Prevalence of Short Sleep (<7 hours) across Sociodemographic, Health Behavior, and Clinical Characteristics among US White Participants (N=18,694) and Native Hawaiian/Pacific Islanders (N=2,134), National Health Interview Survey, 2014^a

	Non-Hispanic White (N=18,694)				Native Hawaiian/Pacific Islander (N=2,134)
	n	%	SE	95% CI	n	%	SE	95% CI	P<
Overall short sleep duration	5,491	29.2	0.46	(28.3, 30.1)	925	41.7	1.48	(38.7, 44.8)	0.00001
Overall Mean age	-	49.3	0.24			41.1	0.45
Age, years
18–49	2,588	29.9	0.67	(28.6, 31.3)	480	38.8	1.84	(35.1, 42.6)	0.00001
50–64	1,770	33.3	0.92	(31.5, 35.1)	288	50.8	3.53	(43.6, 58.0)	0.00001
65+	1,133	22.2	0.72	(20.7, 23.6)	157	43.2	3.84	(35.3, 51.1)	0.00001
Sex/Gender
Male	2,525	29.8	0.70	(27.6, 30.3)	436	45.2	2.45	(40.2, 50.2)	0.00001
Female	2,966	29.0	0.67	(28.4, 31.2)	489	43.5	2.23	(38.9, 48.1)	0.00001
Annual household income *, **
<$35,000/year	1,919	34.1	0.90	(32.4, 35.9)	207	47.4	2.89	(41.5, 53.4)	0.0001
≥$35,000/year	3,205	27.9	0.61	(26.7, 29.2)	640	43.7	2.16	(39.3, 48.2)	0.00001
Living in poverty *, **
Yes	708	35.6	1.43	(32.8, 38.4)	99	51.4 e	6.23	(38.6, 64.2)	0.05
No	4,550	29.0	0.53	(28.0, 30.0)	784	43.8	1.90	(39.9, 47.7)	0.00001
Government assistance
Yes	812	42.6	2.09	(38.5, 46.7)	187	47.0	4.18	(38.4, 55.6)	NS
No	4,663	28.1	0.49	(27.1, 29.0)	738	43.2	1.82	(39.4, 46.9)	0.00001
Educational attainment
Less than high school	399	36.7	2.29	(32.1, 41.2)	84	49.5	6.63	(35.9, 63.1)	NS
High school graduate	1,613	32	0.95	(30.2, 33.9)	295	43.6	2.14	(39.2, 48.0)	0.00001
Some college	1,968	31.5	0.85	(29.8, 33.2)	360	45.6	3.12	(39.2, 52.0)	0.0001
College graduate or more	1,499	24	0.70	(22.7, 25.4)	185	41.6	2.22	(37.1, 46.2)	0.00001
Marital status
Married	2,450	28.4	0.65	(27.1, 29.7)	518	42.3	2.45	(37.2, 47.3)	0.00001
Divorced / separated / widowed	1,909	38.2	1.10	(36.1, 40.4)	184	59	4.88	(49.0, 69.0)	0.0001
Never married	1,121	27.1	1.30	(24.6, 29.7)	222	48.1	3.48	(41.0, 55.3)	0.00001
Employment status b
Unemployed/Not in labor force	2,007	30.0	0.91	(28.2, 31.7)	276	39.9	2.53	(34.7, 45.1)	.001
Employed	3,483	29.7	0.67	(28.3, 31.0)	649	49.3	3.00	(43.1, 55.4)	0.00001
Smoking status
Never	2,641	26.0	0.65	(24.7, 27.3)	489	41.6	1.86	(37.8, 45.4)	0.00001
Former	1,503	32.1	1.03	(30.1, 34.2)	252	46.0	4.22	(37.3, 54.6)	0.01
Current	1,341	36.6	1.26	(34.1, 39.0)	183	49.9	2.68	(44.4, 55.4)	0.00001
Alcohol consumption
Never/former	1,615	29.7	1.01	(27.7, 31.7)	326	45.7	3.51	(38.5, 52.9)	0.0001
Current	3,847	29.1	0.55	(28.1, 30.2)	594	44.3	1.99	(40.2, 48.3)	0.00001
Leisure-time physical activity
Never/unable	1,800	32.8	0.84	(31.1, 34.4)	237	48.6	5.18	(38.0, 59.3)	0.01
Insufficiently active (<150 min/week)	1,043	29.8	1.11	(27.6, 31.9)	166	44.7	5.33	(33.8, 55.7)	0.01
Sufficiently active (150+ min/week)	2,630	27.5	0.66	(26.2, 28.8)	507	43.0	2.09	(38.8, 47.3)	0.00001
Cardiometabolic health outcomes
Overweight/Obese c **
Yes	3,662	31.6	0.59	(30.4, 32.7)	623	44.2	2.50	(39.0, 49.3)	0.00001
No	1,678	25.6	0.76	(24.1, 27.1)	198	41.4	3.07	(35.1, 47.7)	0.00001
Hypertension
Yes	1,989	32.6	0.89	(30.8, 34.3)	378	50.0	3.26	(43.3, 56.6)	0.00001
No	3,500	28.4	0.60	(27.2, 29.6)	545	41.8	2.76	(36.2, 47.5)	0.00001
Diabetes
Yes	546	33.1	2.04	(29.1, 37.1)	150	52.4	5.29	(41.5, 63.2)	0.001
No	4,847	29.2	0.52	(28.1, 30.2)	775	42.9	1.71	(39.4, 46.4)	0.00001
Heart disease
Yes	786	36.1	1.59	(33.0, 39.3)	102	48.2	5.29	(37.3, 59.0)	0.05
No	4,698	28.6	0.51	(27.6, 29.6)	819	44.3	2.05	(40.1, 48.5)	0.00001
Stroke
Yes	190	35.0	3.69	(27.7, 42.2)	35	44.2	7.00	(29.8, 58.6)	NS
No	5,299	29.3	0.47	(28.4, 30.2)	890	44.2	1.78	(40.6, 47.8)	0.00001
Health status
Excellent/very good	2,959	25.8	0.61	(24.6, 27.0)	449	36.5	2.38	(31.6, 41.4)	0.0001
Good	1,543	31.9	1.03	(29.9, 33.9)	293	48.9	3.28	(42.2, 55.6)	0.00001
Fair/poor	985	43.7	1.59	(40.6, 46.8)	182	57.7	3.35	(50.8, 64.5)	0.001
Serious mental illness d
Yes	309	51.2	2.70	(45.9, 56.5)	59	70.8	6.00	(58.5, 83.1)	0.01
No	5,137	28.7	0.47	(27.8, 29.6)	860	43.4	1.87	(39.5, 47.2)	0.00001

CI, confidence interval; SE, standard error

^*Non-Hispanic White: 7.4% missing income, 4.7% missing poverty status; all other variables have <3% missing.

^**Native Hawaiian/Pacific Islander (NHPI): 8.3% missing income, 4.7% missing poverty status, 10.9% missing body mass index (BMI); all other variables have <3% missing.

^aNon-Hispanic White participants are from National Health Interview Survey (NHIS) 2014 survey, and NHPI participants are from special NHIS-NHPI survey in 2014 as well.

^bUnemployed N includes participants no longer in the labor force.

^cOverweight/obese (BMI 25+) versus normal or underweight (BMI <25).

^dKessler 6-item distress scale: possible scores from 0 to 24: 13 or more indicates serious psychological distress.

^eItalicized percentages have a Relative Standard Error >30%

NOTE: Bolded percentages are significantly different between NHPI and White samples.

Sleep Characteristics by Industry of Employment among NHPIs and NH-Whites

Table 2 provides the age-standardized prevalence of various sleep duration and quality indicators, stratified by industry of occupation groups, separately for NHPI vs. NH-White participants. NHPI participants had consistently higher prevalence of very short sleep (< 6 hours) in all industry sectors except food/accommodation, with significant differences compared to NH-White participants in the professional/management/financial sector and in education. NHPI also had higher prevalence of short sleep (<7 hours) across all sectors, with significant differences from NH-Whites again in the professional/management/financial sector, as well as manufacturing/construction/agriculture, health, and public administration. Furthermore, NHPIs had lower prevalence of having the recommended 7–9 hours of sleep in all industry sectors except food/accommodation, with significant differences compared to their NH-White counterparts in the manufacturing/construction/agriculture, professional/management/financial, education, health, and public administration. The two groups did not differ significantly in long sleep (>9 hours); the exception found was wholesale/retail, with NH-White participants having higher prevalence of longer sleep duration.

There were no significant differences in difficulty falling asleep between NHPI and NH-White participants in the manufacturing/construction, professional/management/financial, health, and public administration sectors. NH-White participants had significantly higher prevalence of trouble falling asleep in wholesale/retail and in food/accommodation sectors, while NHPI participants had significantly higher prevalence in the education sector. There were also no significant differences in difficulty staying asleep across most sectors, except for wholesale/retail and food/accommodation with NH-Whites having significantly higher prevalence of trouble staying asleep. NH-Whites were somewhat more likely to wake up feeling rested in all sectors, but only significantly so in the professional/management sector. Likewise, NH-Whites were more likely to rely on sleep medication across sectors, significantly in wholesale/retail, education, health, and food/accommodation. On the global measure of poor sleep, NHPI participants fared worse in all sectors except food/accommodation, with significantly higher prevalence of poor sleep in manufacturing/construction, professional/management/financial, education, and public administration sectors.

Sleep Characteristics by Occupational Class among NHPIs and NH-Whites

Table 3 provides the age-standardized prevalence of various sleep duration and quality indicators, stratified by occupational class, separately for NHPI vs. NH-White participants. NHPIs had significantly higher prevalence of very short sleep and significantly lower prevalence of recommended sleep (7–9 hours) across all three occupation classes. NHPIs did not differ significantly from NH-Whites in trouble falling asleep or staying asleep, but had significantly lower prevalence of waking up feeling rested than NH-Whites in both the professional and service occupations. In these two occupational classes, NHPIs also had significantly lower prevalence of taking sleep medications. NHPIs had significantly higher prevalence of poor sleep in both the service and labor occupations.

DISCUSSION

In this descriptive study comparing two surveys (i.e., NHPI NHIS and NHIS) conducted separately and uncombined among a nationally representative sample of Native Hawaiian/Pacific Islanders and NH-White working adults, we observed NHPI workers across industry and occupational class reported significantly higher prevalence of short sleep (<7 hours) (Figures 1 and 2) and a significantly lower prevalence of recommended sleep duration compared to their NH-White counterparts. Importantly, we found that NHPI-White disparities in sleep health outcomes varied by industry sectors and occupational classes. Specifically, the prevalence of short sleep duration and poor sleep was significantly higher among NHPI workers in the manufacturing, professional, education, health, and public administration sectors when compared to NH-White workers in the same industries, with the greatest disparity in the professional industry. Further, NHPI adults were more likely to have very short sleep and less likely to get recommended amount of sleep across the 3 occupational classes, but the disparity was widest in support service class. Interestingly, although professionals had the lowest and laborers had the highest prevalence of short sleep among the three occupational classes in both NHPI and NH-White groups, the prevalence of short sleep duration was higher among NHPI professionals (35%) than NH-White laborers (33%) (Table 3). Additionally, NH-White workers across industry sectors and occupational class had higher prevalence of sleep medication use compared to NHPI workers, which could lead to improvements in sleep duration and quality, thus potentially contributing to lower prevalence of short sleep duration observed among NH-White workers.

Figure 1. Age-Standardized Prevalence of Short Sleep (<7 hours) by Industry of Employment and Race/Ethnicity. NHPI, Native Hawaiian Pacific Islander; NH-White, Non-Hispanic White.

Figure 2. Age-Standardized Prevalence of Short Sleep (<7 hours) by Occupation Class and Race/Ethnicity. NHPI, Native Hawaiian Pacific Islander; NH-White, Non-Hispanic White.

Prior research suggest that work environments influence sleep duration through various potential pathways such as long works hours, job strain, and discrimination,^23–26 and that sleep is not equitably distributed among different racial/ethnic groups.^41–43 Previous studies showed that short sleep duration is most prevalent in professional/management, manufacturing, food, and health industries.^41,42 There is limited research on the sleep patterns of NHPIs, and one prior NHIS study showed a higher prevalence of sleep duration <7 hours and poor sleep quality, and lower sleep medication use among NHPIs compared to NH-Whites.¹⁵ However, this study did not investigate the work-sleep relationship.

Furthermore, the risk of adverse cardiometabolic health outcomes was found to be greater among NHPIs with suboptimal sleep compared to their NH-White counterparts.¹⁵ Another study utilizing the 2014 NHIS and 2014 NHPI-NHIS data found that NHPI and Black adults were two times more likely to have very short sleep, and NHPI, Hispanic and Black adults were more likely to have short sleep compared to NH-Whites.¹⁶ Similar to our study, they also found that NHPIs, Blacks, and “Others” were less likely to experience the recommended amount of sleep compared to NH-Whites.¹⁶

Studies examining the work-sleep relationship in Asian, Black, and Hispanic/Latino populations showed similar patterns to our study.^11,12,29 Using NHIS data, one study found that Asians had a higher prevalence of short sleep than NH-Whites, especially among those in professional and support service occupations, not among laborers.¹¹ Another NHIS study revealed that Black adults had a higher prevalence of short sleep compared to NH-Whites, with the widest disparity observed in professional occupations.¹² Moreover, immigrant status exacerbated racial/ethnic disparities in sleep health, with the highest prevalence of short sleep found among non-US-born Black adults, followed by US-born Black adults in all occupations.²⁹ The authors found that short sleep duration was most prevalent among Black adults who were not US-born, followed by US-born Black adults in all occupations.²⁹ Additionally, short sleep duration increased with increasing professional roles for Black and Latino adults, while it decreased for NH-Whites.²⁹ Lastly, US-born Latinos in professional and service occupations had a higher prevalence of short sleep compared to NH-Whites, but not those in labor occupations.²⁹

Research demonstrates that lower socioeconomic status is generally associated with higher rates of sleep disturbance.⁴⁴ Sociocultural factors, such as educational attainment, influences occupational status and income creating differential access to and utilization of health-related information and services. The NHPI population faces barriers to education that are not experienced by NH-White adults. Specifically, family obligations, traditional or religious restrictions, being geographically isolated without access to higher education, and the financial burden of having to relocate to acquire education have been identified as common barriers among NHPIs,⁴⁵ all of which influence the type of job one is able to secure. Greater family obligations among NHPIs may also create greater financial burden even among those with higher income or occupational status, leading to greater stress and sleep difficulties. Depending on geographic location and relative cost of living (e.g., Hawaii), there might be multiple generations or extended family members in one household, potentially increasing stress level of household income earners who may work multiple jobs and have less opportunity for the recommended hours of sleep within a 24-hour period. Households with large number of family members also might experience differences in member sleep quality when compared to households with fewer individuals, especially as sleeping arrangements (i.e., shared bedrooms) can alter noise and comfort levels. Furthermore, it is important to note that minoritized racial/ethnic groups experience disproportionate socioeconomic disadvantage, likely associated with racism, and NHPI no less experience racial discrimination in their daily lives. For example, 1 in 4 Micronesians in Hawaii have experienced workplace discrimination because of their ethnicity.⁴⁶ Discrimination based on race in the workplace is linked to shorter sleep durations and the emergence of insomnia symptoms in marginalized racial/ethnic groups.²³ Further research is needed to examine its potential impact on the NHPI community. Additional research is also warranted to investigate the influence of acculturation and the associated acculturative stress on work-sleep relationship among NHPI adults.

In considering the clinical and policy implications and avenues for intervention, it is essential to recognize the broader structural and public health dimensions influencing sleep health disparities among NHPI populations. Beyond the scope of our study, there is a need to explore and implement targeted interventions addressing the socioeconomic determinants of poor sleep. Community-based sleep health programs tailored to NHPI cultures could enhance awareness and education within these communities. Additionally, workplace interventions, such as advocating for policies promoting work-life balance and stress management, may mitigate occupational factors contributing to sleep disparities. Policymakers could play a pivotal role in advocating for affordable housing initiatives and neighborhood improvements, recognizing the impact of living conditions on sleep health. Ensuring equitable access to healthcare services, coupled with cultural competency training for healthcare professionals, is crucial to addressing the unique healthcare needs of NHPI populations. Moreover, social support programs and advocacy for educational opportunities are potential strategies to address broader socio-economic disparities.

Our study has several limitations. First, the cross-sectional design prevented investigating prospective associations between occupation and sleep duration/quality. Second, using self-reported sleep data could result in bias. However, a validation study examining the concordance between self-reported and objectively measured sleep found that self-reported sleep duration was overestimated compared to objective sleep duration by all racial/ethnic groups, which suggests that our finding is not simply due to measurement bias.⁴⁷ We also lacked data on sleep disorders (e.g., sleep apnea), which may affect sleep duration/quality. We also could not account for the household size or the number of children in the household, which likely influences sleep and differs by race. Given the strong family ties among NHPIs, it is possible that this group is more likely than NH-White adults to live in multigenerational households. We also did not have data on shiftwork, which is associated with shorter sleep duration and sleepiness.⁴⁸ Moreover, statistical power was insufficient to test for differences between Native Hawaiians and different Pacific Islander subgroups. Indeed, a recent NIH workshop has revealed that the aggregation of data among Asian American and NHPI groups have resulted in masking of critical health disparities.¹⁹ Future research on disaggregation of NHPI groups is warranted. Our study has important strengths despite the limitations. For instance, we used a large nationally representative sample of understudied Native Hawaiians and Pacific Islanders. With the large sample of NHPIs, we were able to stratify based on industry groups and occupation classes and provide robust estimates extending our understanding of sleep disparities among NHPI adults.

In conclusion, NHPI-White disparities in sleep varied by industry of employment and occupation class, suggesting that the work environment may serve as an important setting that contributes to racial/ethnic disparities in sleep health as well as an important setting to intervene and mitigate observed disparities.⁴⁹ Further investigation is needed to identify determinants of disparities in the workplace and associated factors, and subsequently develop tailored interventions to address disparities in hopes of improving sleep health among US workers.