Racial/ethnic minorities have greater declines in sleep duration with higher risk of cardiometabolic disease: An analysis of the U.S. National Health Interview Survey

Abstract

Objective:

Habitual insufficient sleep may contribute to cardiometabolic disease in the United States, particularly among racial and ethnic minorities. However, there is mixed evidence on secular trends in U.S. sleep duration.

Study design:

Cross-sectional data from 413,417 individuals were acquired from the National Health Interview Survey from 2005–2018. Variables included self-reported sleep duration as well as lifetime diagnosis of hypertension, coronary heart disease, diabetes, and pre-diabetes, and obesity. Population-weighted robust Poisson models estimated annual trends in sleep duration and the prevalence risk ratios (PRR) of cardiometabolic disease as a function of sleep duration.

Setting:

Population-based survey.

Results:

Daily sleep duration decreased −0.62 min ([−0.71, 0.54], p < 0.01) annually from 2005–2018. However, this decline began only after 2010, when sleep duration fell by 1.04 min ([−1.21, −0.86], p < 0.01) each year. This trend varied by race (two-way ANOVA, p = 0.02), such that Mexican Hispanic individuals saw a greater decline (−1.83 [−2.37, −1.30] min per year, p < 0.01) than whites (−0.83 [−1.02, −0.64] min per year, p < 0.01). Additionally, a 1-h loss in daily sleep duration was linked to 4% greater prevalence of hypertension (PRR: 1.04, [1.04, 1.05]), 3% greater prevalence of diabetes (PRR: 1.03 [1.01, 1.05]), and 8% greater prevalence of obesity (PRR: 1.08 [1.07, 1.09]) after adjusting for age, sex, employment, marital status, and survey year.

Conclusions:

There is a secular decline in U.S. daily sleep duration that is greater among Mexican Hispanic individuals. Moreover, reduced sleep duration is associated with more prevalent cardiometabolic disease.

1. Introduction

There is no consensus on whether adult sleep duration is decreasing in the United States. In their systematic review, Bin and colleagues [1] reported that the proportion of short sleepers increased from 1985 to 2006, but that average sleep duration increased during the same period. Ford and colleagues [2] analyzed the National Health Interview Survey (NHIS) and reported that sleep duration decreased between 1985 and 2004, but that no changes occurred between 2004 and 2012. Youngstedt and colleagues [3] reviewed 168 studies of objective sleep duration and reported that polysomnographic and actigraphic measures of sleep duration showed no change in total sleep time between 1960 and 2013, a finding which “challenged the notion of a modern epidemic of insufficient sleep.” Indeed, Basner and Dinges [4] analyzed the American Time Use Survey (ATUS) and reported that self-reported sleep duration had increased between 2003 and 2016 among students, employed Americans, and retirees. By contrast, Hisler and colleagues [5] found that average sleep duration decreased (and sleep difficulties increased) in the NHIS from 2013 to 2017, while Sheehan et al. [6] reported that, the prevalence of short sleep was stable from 2004 to 2012 but increased from 2013 to 2017, especially among Hispanics and Black Americans. Ultimately, Hoyos and colleagues [7] conducted a systematic review and found “little evidence” for significant declines in sleep duration prior to 2012.

Several factors may clarify the confusion on this issue. First, data from 2004 to 2012 generally show no change [2, 3, 7] or a small increase [1, 4] in sleep duration, while data after 2013 suggest a decrease [5, 6]. Thus, annual trends in sleep duration may have changed around 2012. Second, categorical analyses of short sleep [1,6] (defined as less than 7 h) typically use logistic regression models which indirectly and inaccurately estimate relative risk using odds ratios [8]. Third, claims about U.S. trends in sleep duration are often derived from the NHIS [2,5,6,9], the ATUS [4], or similar epidemiological surveys which have unique limitations. For the NHIS, sleep duration is derived from a single question asking, “on average, how many hours of sleep do you get in a 24-h period?”, which Ogilvie and Patel [10] point out has never been validated. By contrast, the ATUS uses a daily diary that starts and ends at 4 AM. While sleep diaries tend to be more robust measurement tools, the ATUS daily diary interrupts most individuals’ typical sleep periods (leading to measurement error) and is more likely to capture time in bed than time asleep [10]. Finally, the specter of reporting and recall bias are always present with self-report data, even in large samples and despite the known correlations between self-report and objective measures of sleep duration [11–13].

These arguments about measurement bias and statistical methods may appear trivial, but adequate sleep is critical for human health. As such, claims about systematic sleep deficits should not be made lightly. The official position of the American Academy of Sleep Medicine and the Sleep Research Society is that the average adult needs 7 to 8 h of sleep [14], in part because insufficient sleep is linked to significant morbidity and mortality. An overview of systematic reviews [15] compiling data from over 4 million participants across 30 countries found that short sleep increased all-cause mortality by 6 to 12%, risk of cardiovascular disease by 7%, risk of type 2 diabetes by 9%, and incident obesity by up to 38%. Moreover, clinical and experimental evidence support a mechanistic relationship between short sleep and hypertension [16], type 2 diabetes [17], and other cardiometabolic diseases [18].

Unfortunately, the prevalence and impact of insufficient sleep does not fall equally on all Americans. In 2009, 34% of Black Americans reported sleeping 5–6 h compared to 25.4% of White Americans, while nearly twice as many Black Americans reported sleeping less than 5 h as White Americans [9]. Black Americans are more likely to report insufficient sleep than White Americans [6,8,19], as are Hispanic and Asian Americans [19,20], and the disparity between Black and White Americans’ sleep may develop as early as age 24 [21]. Moreover, the observed deficits between Black or Hispanic Americans’ sleep and White Americans’ sleep appear to be growing [22], and these racial disparities in sleep duration may increase the risk of hypertension, diabetes, and obesity [23–27]. Given the apparent consequences of this sleep health disparity, accurately quantifying secular trends in sleep duration among racial/ethnic groups is critical to establishing the true magnitude of this problem and what, if any, interventions should be leveraged to address it.

Therefore, the present study examined data collected by the NHIS from 2005 to 2018 to evaluate secular trends in self-reported sleep duration and cardiometabolic disease, both within the general population and by race/ethnicity. The primary hypothesis was that daily sleep duration would decrease between 2005 and 2018 with greater declines in racial/ethnic minorities. A secondary hypothesis was that the annual change in sleep duration would change sometime around 2012, reflecting prior literature showing an inflection in most annual sleep trends. Finally, it was hypothesized that reduced daily sleep duration would be associated with an increased prevalence of cardiometabolic diseases.

2. Materials and methods

2.1. Data source

Cross-sectional data from 413,417 respondents were collected for the years 2005 to 2018 from the National Health Interview Survey (NHIS, https://www.cdc.gov/nchs/nhis/1997-2018.htm), which is conducted annually by the Centers for Disease Control and Prevention in all 50 states and the District of Columbia. The NHIS divides each state into individual counties, groups of contiguous counties, or metropolitan areas, and then into address clusters based on the most recent census housing data. A subset of address clusters is randomly selected and NHIS interviewers go to these clusters and interview the occupants. Responses are then weighted based on census data so that the NHIS is representative of the civilian, noninstitutionalized population of the United States.

2.2. Variables

The primary outcome was self-reported sleep duration, which was assessed using a single question that asked, “on average, how many hours of sleep do you get in a 24-h period?”. Responses were integers ranging from 1 to 24 h, although responses greater than 12 were top coded to 12. Sleep duration was converted to minutes and secondarily categorized as short (< 7 h), recommended (reference, 7–8 h), and long (≥ 9 h). The secondary outcomes were obesity and morbid obesity determined by body mass index (obesity: reference, BMI<30 vs BMI 30–39; morbid obesity: reference, BMI< 40 vs. BMI≥ 40), as well as hypertension (no/yes), coronary heart disease (no/yes), diabetes (no/yes), and prediabetes (no/yes). BMI was determined from height and weight measurements and the other conditions were assessed by single items which asked, “have you EVER been told by a doctor or other health professional that you had… [short description of disease]”. The primary predictors were survey year (between 2005 and 2018) and race/ethnicity (White, Black/African American, Mexican Hispanic, Other Hispanic, Asian). Covariates included age (reference, 18–34; 35–64; and 65 +), sex (reference, Male; Female), BMI (< 18.5; reference, 18.5–25; 25–30; 30–40; 40 +), employment (reference, employed; unemployed; homemaker; student; retired), and marital status (reference, never married; married; widowed; divorced/separated; living with partner).

2.3. Statistical analysis

Survey-weighted linear regression models estimated how sleep duration (in minutes) changed across survey years. Analyses were first unadjusted and then adjusted for age, sex, race/ethnicity, BMI, employment, and marital status. Stratified analyses examined significant race/ethnicity by year interactions after adjustment for age, sex, and other covariates. As a secondary analysis, the prevalence risk ratios of categorical short sleep by survey year were estimated using quasi-Poisson models (which relaxes the Poisson model constraint requiring the mean equal the variance by introducing an overdispersion term). Models were first unadjusted and then adjusted for age, sex, race/ethnicity, BMI, employment, and marital status. Stratified analyses then evaluated significant year by race/ethnicity interactions in prevalence risks after adjustment for age, sex, and other covariates. Finally, survey-weighted quasi-Poisson regression models estimated how a one-hour reduction in average daily sleep duration affected the prevalence risk ratios of hypertension, coronary heart disease, diabetes, pre-diabetes, and obesity. All models were adjusted for age, sex, race/ethnicity, employment, marital status, and survey year. Analyses were conducted using R (v. 4.1.0, R Foundation for Statistical Computing, Vienna, Austria), and the code used for this analysis is available at https://github.com/atubbs-sleep. Outcomes are reported as unstandardized regression coefficients (B) or prevalence risk ratios (PRR) and 95% confidence intervals.

3. Results

3.1. Trends in nightly sleep duration

Between 2005 and 2018, average daily sleep duration decreased from 429.2 min (7.15 h) to 422.8 min (7.05 h) (Fig. 1 A). In unadjusted analyses, daily sleep duration decreased significantly (−0.62 min per year [−0.71, −0.53], p < 0.01), and adjusting for age, sex, race/ethnicity, BMI, employment status, and marital status did not eliminate this result (−0.65 min per year [−0.74, −0.57], p < 0.01). This equates to roughly 4 h of sleep lost each year over 13 years. A visual review of the data showed that sleep duration appeared constant from 2005 until 2010, but then fell from 2010 to 2018. Therefore, the data were split at 2010 and reanalyzed. From 2005 to 2010 there was no change in sleep duration in unadjusted or adjusted models (all p > 0.05). However, from 2010 to 2018 sleep duration decreased by 1.04 [−1.21, −0.86] min per year in the unadjusted model (p < 0.01), and −0.99 [−1.16, −0.83] min per year when adjusted for age, sex, race/ethnicity, BMI, employment, and marital status (p < 0.01).

Fig. 1. Changes in self-reported sleep duration by sex and race/ethnicity from 2005 to 2018 in the U.S.

(A) Average self-reported sleep duration from 2005 to 2010. While sleep duration remained stable from 2005 to 2010, it then declined after 2010. Shaded band represents the 95% confidence interval. (B) Average self-reported sleep duration by race/ethnicity. While sleep duration declined significantly among whites, the declines were greater among Black Americans and Mexican Hispanics. (C) Annual trends in sleep duration by race for men and women based on linear modeling.

3.2. Trends in nightly sleep duration by race/ethnicity

For 2005 to 2018, there was no difference in trends by sex (two-way ANOVA, p = 0.05), but there was a difference in trends by race/ethnicity (two-way ANOVA p < 0.01). In post-hoc testing, greater annual declines were observed among Black Americans (−1.03 [−1.30, −0.76] min per year, p < 0.01) and Mexican Hispanics (−1.44 [−1.72, −1.17] min per year, p < 0.01) compared to White Americans (−0.47 [−0.57, −0.37] min per year, p < 0.01). For 2005 to 2010, there were no annual trends in sleep duration (one-way ANOVA, p = 0.86), nor differences in trends by sex (two-way ANOVA p = 0.09) or race (p = 0.13). However, from 2010 to 2018, there was variation in annual trends by race/ethnicity (two-way ANOVA, p = 0.02). In post-hoc testing, declines were evident among White Americans (−0.83 [−1.02, −0.64] min per year, p < 0.01) that were greater still among Mexican Hispanics (−1.83 [−2.37, −1.30] min per year, p < 0.01). The annual trends in sleep duration are reported by sex and race/ethnicity in Table 1 and presented graphically in Fig. 1 B,C.

Table 1.

Annual trends in sleep duration by race/ethnicity from 2005 to 2018.

	2005 to 2018		2005 to 2010		2010 to 2018
Unadjusted	B	95% CI	B	95% CI	B	95% CI
White	−0.44	[−0.54, −0.33]	0.35	[−0.02, 0.72]	−0.86	[−1.06, −0.67]
Black	−0.95	[−1.21, −0.68]	−0.20	[−1.09, 0.69]	−1.25	[−1.77, −0.72]
Mexican Hispanic	−1.34	[−1.64, −1.04]	−0.39	[−1.15, 0.36]	−1.85	[−2.44, −1.27]
Other Hispanic	−0.47	[−0.87, −0.08]	0.99	[−0.20, 2.17]	−1.11	[−1.87, −0.34]
Asian	−0.69	[−1.03, −0.34]	−0.53	[−1.88, 0.82]	−1.03	[−1.62, −0.44]
Adjusted	B	95% CI	B	95% CI	B	95% CI
White	−0.47	[−0.57, −0.37]	0.23	[−0.13, 0.59]	−0.83	[−1.02, −0.64]
Black	−1.03	[−1.30, −0.76]	−0.51	[−1.41, 0.39]	−1.28	[−1.81, −0.76]
Mexican Hispanic	−1.44	[−1.72, −1.17]	−0.64	[−1.38, 0.09]	−1.83	[−2.37, −1.30]
Other Hispanic	−0.49	[−0.89, −0.10]	0.56	[−0.60, 1.72]	−1.01	[−1.76, −0.25]
Asian	−0.67	[−1.03, −0.31]	−0.47	[−1.84, 0.90]	−1.03	[−1.62, −0.43]

^*Adjusted for age, sex, BMI, employment, and marital status.

3.3. Prevalence risk of short sleep across years by race and ethnicity

In unadjusted models, the prevalence risk of short sleep increased 1.3% each year (PRR 1.013 [1.011, 1.015], p < 0.01), and this annual increase remained significant after adjusting for covariates (PRR 1.01 [1.010, 1.013], p < 0.01). There was variation in prevalence risk over time by race/ethnicity (two-way ANOVA, p < 0.01). Post-hoc testing showed the risk of short sleep rose more among Mexican Hispanics (PRR 1.03 [1.02, 1.04], p < 0.01) than White Americans (PRR: 1.01 [1.01, 1.01], p < 0.01). These data are presented in Fig. 2.

Fig. 2. Annual trends from 2005 to 2018 in the prevalence of long sleep, recommended sleep, and short sleep by race/ethnicity.

Shaded bands represent 95% confidence intervals.

3.4. Sleep duration and cardiometabolic disease by race/ethnicity

The prevalence risk ratios associated with a 1-h shorter daily sleep duration are presented in Table 2. In adjusted models, losing 1 h of daily sleep was associated with a 4% greater prevalence of hypertension, 3% greater prevalence of coronary artery disease and diabetes, a 13% greater prevalence of pre-diabetes, a 10% greater prevalence of obesity, and a 19% greater chance of morbid obesity. Risks were notably higher for hypertension among Black Americans (PRR 1.06 [1.04, 1.07]), Mexican Hispanics (PRR 1.12 [1.09, 1.15]), Other Hispanics (PRR 1.10 [1.06, 1.15]), and Asians (PRR 1.08 [1.04, 1.11]) than among White Americans (PRR 1.03 [1.02, 1.04]); higher for diabetes among Other Hispanics (PRR 1.11 [1.03, 1.19]) than White Americans (PRR 1.03 [1.01, 1.05]); and higher for pre-diabetes among Mexican Hispanics (PRR 1.20 [1.11, 1.29]), Other Hispanics (PRR 1.33 [1.18, 1.45]), and Asian Americans (PRR 1.26 [1.14, 1.38]) than White Americans (PRR 1.10 [1.07, 1.13]).

Table 2.

The prevalence risk of cardiometabolic disease due to a 1-h decrease in daily sleep by race/ethnicity.

Outcome	Unadjusted		Adjusted†
Hypertension	PRR	95% CI	PRR	95% CI
Overall	1.01	[1 0.00, 1.02]	1.04	[1.04, 1.05]
White	0.98	[0.97, 0.99]	1.03	[1.02, 1.04]
Black	1.02*	[1.01, 1.03]	1.06*	[1.04, 1.07]
Mexican Hispanic	1.13*	[1.10, 1.17]	1.12*	[1.09, 1.15]
Other Hispanic	1.13*	[1.08, 1.18]	1.10*	[1.06, 1.15]
Asian	1.10*	[1.06, 1.15]	1.08*	[1.04, 1.11]
Coronary Artery Disease	PRR	95% CI	PRR	95% CI
Overall	0.91	[0.89, 0.93]	1.03	[1.01, 1.05]
White	0.87	[0.85, 0.90]	1.02	[0.99, 1.04]
Black	0.99	[0.94, 1.04]	1.07	[1.02, 1.13]
Mexican Hispanic	1.05	[0.95, 1.16]	1.10	[1.00, 1.20]
Other Hispanic	1.11	[0.94, 1.29]	1.11	[0.95, 1.26]
Asian	1.13	[0.97, 1.30]	1.13	[1.01, 1.26]
Diabetes	PRR	95% CI	PRR	95% CI
Overall	0.98	[0.97, 1.00]	1.03	[1.02, 1.05]
White	0.95	[0.93, 0.97]	1.03	[1.01, 1.05]
Black	1.00	[0.97, 1.03]	1.05	[1.02, 1.09]
Mexican Hispanic	1.04	[0.99, 1.09]	1.04	[1.00, 1.09]
Other Hispanic	1.12*	[1.04, 1.21]	1.11*	[1.03, 1.19]
Asian	1.02	[0.93, 1.10]	0.99	[0.93, 1.06]
Pre-Diabetes	PRR	95% CI	PRR	95% CI
Overall	1.12	[1.10, 1.15]	1.13	[1.11, 1.16]
White	1.08	[1.05, 1.11]	1.10	[1.07, 1.13]
Black	1.13	[1.08, 1.19]	1.15	[1.09, 1.21]
Mexican Hispanic	1.26*	[1.17, 1.35]	1.20*	[1.11, 1.29]
Other Hispanic	1.37*	[1.22, 1.52]	1.33*	[1.18, 1.47]
Asian	1.33*	[1.20, 1.46]	1.26*	[1.14, 1.38]
Obesity	PRR	95% CI	PRR	95% CI
Overall	1.10	[1.09, 1.11]	1.09	[1.08, 1.10]
White	1.10	[1.09, 1.11]	1.10	[1.09, 1.11]
Black	1.07	[1.05, 1.08]	1.07	[1.06, 1.09]
Mexican Hispanic	1.11	[1.08, 1.13]	1.09	[1.07, 1.12]
Other Hispanic	1.13	[1.08, 1.17]	1.12	[1.07, 1.16]
Asian	1.18	[1.08, 1.27]	1.18	[1.08, 1.27]
Morbid Obesity	PRR	95% CI	PRR	95% CI
Overall	1.19	[1.16, 1.21]	1.18	[1.15, 1.20]
White	1.18	[1.15, 1.22]	1.19	[1.15, 1.23]
Black	1.12	[1.08, 1.16]	1.13	[1.09, 1.18]
Mexican Hispanic	1.22	[1.14, 1.30]	1.21	[1.12, 1.30]
Other Hispanic	1.23	[1.05, 1.41]	1.22	[1.03, 1.42]
Asian	1.23	[0.75, 1.71]	1.30	[0.78, 1.82]

^†Adjusted for age, sex, employment, marital status, and survey year.

^*Significant difference (p < 0.05) from whites.

4. Discussion

Self-reported daily sleep duration decreased by 7 min from 2005 to 2018, which equates to roughly 4 h per year over the past 13 years. This decline appeared to start in 2010, averaging approximately 1 min per year. Sleep loss disproportionately impacted Black Americans and Mexican Hispanics, who saw 2-fold and 3-fold greater annual declines, respectively. The possible consequences of these changes are not trivial, as a 1-h loss of daily sleep was estimated to increase the prevalence risk of cardiometabolic disease by 3–13%, with racial/ethnic minorities experiencing greater risks for hypertension, diabetes, pre-diabetes, and obesity.

This evidence of declining U.S. daily sleep duration stands in apparent contrast to prior reports [2–4,7]. However, there are key factors that might explain this discrepancy. The data reviewed by Bin et al. [1] and Hoyos et al. [7] were from, at the latest, 2012, as were the NHIS data analyzed by Ford et al. [2]. While the present study found a decreasing trend in sleep duration from 2005 to 2018, data from 2005–2010 showed no significant changes, thus confirming these prior studies. This perspective similarly applies to the trends in objective sleep duration reported by Youngstedt et al. [3], where 2013 was the last year examined. The time window difference, however, does not explain why Basner and Dinges [4] found evidence of increasing sleep duration from 2003 to 2016. It is possible that this disconnect arises from inherent differences between measuring sleep duration in the NHIS (an unvalidated single-item measure) versus the ATUS (an unvalidated activity diary ranging from 4AM to 4AM the next day). Both assessment strategies have flaws as outlined above and by Ogilvie and Patel [10], thus more work is necessary to clarify the issue.

Another important element was the confirmation and improved quantification of how changes in sleep duration are disproportionately impacting racial and ethnic minorities. In their analyses of the NHIS, Sheehan et al. [6] reported an increasing incidence of short sleep starting in 2013 that was greater among non-Hispanic Black and Hispanic respondents. Similarly, the present study found a secular decline in sleep duration that was more pronounced among Black and Mexican Hispanic individuals. However, this trend appeared to start after 2010, and when examining 2010 to 2018 more specifically, only Mexican Hispanic individuals showed a greater decline in sleep duration than White Americans (−1.85 v. −0.86 min) or a greater increase in the prevalence of short sleep (3% vs. 1% annually). This is likely because Mexican Americans started with greater sleep duration and less prevalent short sleep in 2005 and have subsequently declined to match the levels reported by White Americans. This decline may be due to acculturation to an American lifestyle, which affects population and individual patterns of health, disease risk, and unhealthy behaviors [28–30]. Greater acculturation among Hispanics is linked to increased cardiometabolic risk [31], particularly among Mexican Americans [32, 33]. Mexican immigrants are 40% more likely than US-born Mexican Americans to be short sleepers, which is largely due to negative acculturation [34]. Mexican Americans are also more likely to report difficulties in maintaining sleep, early morning awakenings, nonrestorative sleep, and daytime sleepiness [35]. In a sample of Hispanic/Latina women, Kachikis and Breitkopf [36] found greater acculturation was associated with shorter sleep duration and poorer sleep quality. Similar findings were noted among individuals of Mexican descent at the US-Mexico border, where greater US acculturation was associated with less sleep duration, worse insomnia symptoms, poorer sleep quality, and increased sleep apnea [37]. Lastly, first-generation immigrants who were less acculturated maintained better sleep, where greater acculturation was associated with worse sleep [38]. These findings indicate that cultural behaviors and norms may play an important role in maintaining good sleep in the Hispanic population.

The null finding for Black Americans, however, does not establish their equivalence with White Americans; in 2010, Black Americans reported less sleep duration and more prevalent short sleep at baseline compared to White Americans, even if the subsequent annual trends were indistinguishable. Thus, the divergent findings from Sheehan et al. [6] may be the result of differing timeframes or the use of quasi-Poisson models to directly estimate prevalence risks rather than logistic models to estimate prevalence odds ratios.

Finally, these results illustrate how declining sleep duration relates to cardiometabolic disease. The loss of a single hour of daily sleep significantly increased the prevalence risk of hypertension, coronary heart disease, diabetes, pre-diabetes, and obesity and, in some cases, more so among racial/ethnic minorities. The most dramatic differences were noted for pre-diabetes, where loss of one hour of sleep increased the prevalence risk by 20–33% for Asian and Hispanic Americans over White Americans. Given the extensive morbidity, mortality, and healthcare burden [39,40] associated with cardiometabolic diseases, increasing sleep duration may be a reasonable and cost-effective strategy to reduce this burden, particularly in minority communities.

The strengths of this study include the large sample (over 400,000 respondents), the sample period (2005 to 2018), and the robust analytic strategy (direct estimation of prevalence risks using quasi-Poisson models). As with all epidemiological studies of sleep duration, however, this study is limited by the cross-sectional study design which precludes causal claims and the lack of objective measures of sleep and cardiometabolic disease. Although sleep duration was measured using a single-item self-report measure of daily sleep duration which is vulnerable to recall and reporting bias, the simplicity of the question likely provides a reasonable, aggregate estimate of population sleep duration. Future studies with more robust methodologies are needed to clarify these findings, particularly among racial/ethnic minorities. The use of longitudinal/repeated measures study designs, subjective and objective measures of sleep, objective measures of cardiometabolic disease, and sleep-extension interventions would all contribute to a greater understanding of how sleep influences cardiometabolic disease at the population level.

5. Conclusions

Annual self-reported daily sleep duration has fallen by 7 min in the United States, with steeper declines among Black/African American and Mexican Hispanic individuals. The consequences of these declines include rising insufficient sleep and greater prevalence of obesity, diabetes, and hypertension. Thus, sleep may play a pivotal role in the development, and/or prevention, of cardiometabolic diseases in the United States.