Racial and ethnic disparities in sleep are increasingly recognized as an important public health issue for their potential contribution to disparities in overall health, health outcomes, and mortality (Jackson et al., 2015; Billings et al., 2021; Khot et al., 2023; Denney et al., 2023). This concern was best conveyed by Grandner et al. (2016) who said “sleep disparities represent a pathway by which larger disparities emerge.” Sleep disparities are significant health issues by themselves but they are an even more critical problem seen as a driver of potentially much larger downstream health disparities.

Previous studies amply demonstrated that racial and ethnic minority people have shorter sleep time and poorer quality compared to White people (Ruiter et al., 2011; Grandner et al., 2013; Ahn et al., 2021). A recent study showed that the percentage of short sleepers (< 7 h of sleep) among African American or Black individuals was persistently higher than that among Caucasian or White individuals in a 15-year period between 2004 and 2018, with the gap between the two groups growing larger over time (Caraballo et al., 2022). However, much of the evidence on racial and ethnic disparities in sleep is based on subjective sleep and the lack of objective sleep data are often recognized as a limitation in this available research (Piccolo et al., 2013).

In our scoping review conducted in 2021 (Ahn et al., 2021), we found that the most consistent and significant finding in the sleep disparities literature was the disparities in subjective sleep duration experienced by Black people compared to White people. Data for Hispanic people or other racial and ethnic groups were much more limited and their results were mixed or insufficient. Importantly, the evidence on the racial/ethnic disparities in objectively measured sleep duration was mixed. In addition to sleep duration, other objective measures of sleep also were largely mixed or insufficient. To fill this gap in the literature, our aim is to examine sleep disparities between races and ethnicities using objectively-measured sleep data from a large academic medical center in the US.

Our sample included 5,331 PSG studies conducted on the same number of patients who met our inclusion criteria. Patient characteristics by their race and ethnicity are shown in Table 1. Of all patients, 63% were White, 21% Black, 4% Hispanic, and 12% other or unknown race/ethnicity. The total patient population was 51 ± 14 years at the time of the PSG study, with Black and Hispanic patients about 5 and 8 years younger than White patients (p < 0.001). While the entire sample was comprised of 41% male, only 30% and 32% were male among Black and Hispanic patients (p < 0.001). While over 51% of White patients were married or living with a significant other, only 28% and 44% were among Black and Hispanic patients (p < 0.001).

Mean BMI for the sample was 34.7 ± 8.9 kg/m² with Black patients having the highest BMI (37.7 ± 9.3 kg/m²) of all groups (p < 0.001). While 80% of Black patients were obese (BMI ≥ 30 kg/m²), only 62% and 71% were among White and Hispanic patients. Black patients also had the highest rates of Class III obesity (BMI ≥ 40 kg/m²) at 33% compared to ~20% for both White and Hispanic patients.

More than 84% of all patients had one or more co-existing conditions with over 40% having 3 or more. Other than obesity, the most prevalent comorbidity was hypertension (59%), followed by diabetes (35%), depression (25%), and chronic lung disease (23%). Of all racial/ethnic groups, Black patients had the highest prevalence of hypertension (74%), diabetes (69%), and chronic lung disease (25%), while White patients had the highest prevalence of depression (26%).

Our results show that Black and Hispanic patients were not significantly different from White patients in most objective measures of sleep, including TST, AHI, or sleep efficiency. However, we found some interesting differences in sleep stage outcomes. In adjusted models, Black patients spent almost 10 min shorter in N3 (Slow Wave) sleep and spent 2 min longer in REM sleep than White patients. It is notable that both Black and Hispanic patients did not show differences in any of the measures of sleep disordered breathing, including AHI, OSA (AHI ≥ 5), severe OSA (AHI ≥ 30), POSA, ePOSA-5, -10, and -15. To the best of our knowledge, this study is the first that examined racial and ethnic disparities in positional OSA.

Overall, on most other measures that were significantly different in adjusted analyses, Black patients tended to have more favorable sleep compared to White patients, including shorter sleep latency, higher mean oxyhemoglobin saturation, shorter time spent in oxygen saturation < 89%, fewer PLMs, and awakenings. The only unfavorable sleep feature identified in Black patients was total stage changes compared to White patients. Hispanic patients show differences that are similar in both direction and magnitude of those in Black patients compared to White patients. However, their differences were mostly insignificant, perhaps due to their smaller sample size.

In the sleep disparities literature, the largest number of studies were conducted for racial disparities in subjective sleep duration, followed by subjective sleep quality/insomnia, and sleep disordered breathing (Ahn et al., 2022). Evidence so far overwhelmingly indicates that Black patients have shorter subjective sleep duration than White patients (Caraballo et al., 2022; Jackson et al., 2014; Cunningham et al., 2016; Seixas et al., 2019), but disparities in objective sleep duration is not so clear. A meta-analysis published in 2011 reported that, in eight studies of objective sleep duration, Black patients had 28 min (p < 0.05) shorter TST than White patients among normal sleepers (Ruiter et al., 2011). However, individual studies based on PSG reported mostly non-significant and inconsistent results (Redline et al., 1997; Rao et al., 1999; Irwin et al., 2000; Profant et al., 2002; Stepnowsky et al., 2003; Walsleben et al., 2004; Thomas et al., 2006; Mills et al., 2007; Mezick et al., 2008). It appears that the one actigraphy-based study had contributed a lion's share to the combined effect due to its large effect size as well as sample size (Lauderdale et al., 2006). Since then, a few additional studies have examined disparities in objectively measured sleep. Studies in which sleep duration was assessed using actigraphy showed relatively consistent findings favoring White patients (Mezick et al., 2008; Matthews et al., 2019; Song et al., 2011; Chen et al., 2015), while those assessed with PSG still showed mixed results (Hall et al., 2009; Fiorentino et al., 2006). It is our conjecture that the difference between actigraphy-based and PSG-based sleep time may derive from the sleeping environments at the time of measurement. For example, actigraphy studies are typically conducted at home and study subjects sleep in their usual sleeping environment, while in-lab PSG study subjects sleep in a highly uniform sleeping environment. Thus, actigraphy-measured sleep time is likely to also reflect differences in sleeping environments in terms of crowding, air quality, light, and noise at night (Jackson, 2017; Billings et al., 2019; Fang et al., 2015; Patel, 2019; Rudolph et al., 2019; Hammer et al., 2014).

Our results utilizing objective sleep duration are not consistent with most of the existing literature on disparities in subjective sleep duration. There are many reasons why objective and subjective sleep durations may be different, including the response bias inherent in self-reported values. Several studies (Takahashi et al., 2022; O'Donnell et al., 2009; Benz et al., 2023; Utsumi et al., 2022; Lauderdale et al., 2008) reported substantial differences between subjective and objective sleep durations which may be as large as 20% of TST which was driven by over-reporting among shorter sleepers (Lauderdale et al., 2008). One study even suggested that subjective and objective sleep durations may capture two different aspects of sleep (Benz et al., 2023). Another reason may be that our sample consisted of persons who had sleep complaints and were sleeping outside the usual sleeping environment with electroencephalogram (EEG) electrodes placed on the body. In addition, these are patients who had financial access to polysomnographic studies, either through their health insurance or self-payment (Stanchina, 2022). As such, our sample may represent a selective subset that may be different from those found in the communities. These are some of the explanations why racial/ethnic disparities based on subjective and objective sleep durations might diverge.

Another noteworthy finding in our study is the lack of disparities in most measures of sleep disordered breathing. Historically, evidence on racial and ethnic disparities in sleep disordered breathing in adults is largely mixed (Ahn et al., 2021; Dudley and Patel, 2016). Two recent studies illustrate the current state of evidence in this area. The first study based on data from an academic medical center in the US (Pranathiageswaran et al., 2013) showed a significantly higher AHI for the African American patients compared to White patients (median 32.7 vs 22.4; p = 0.01). The second study based on in-home PSG data for a subset of the Multi-Ethnic Study of Atherosclerosis (MESA) cohort (Chen et al., 2015) showed that Black participants were not significantly different from White participants in SDB severity, while Hispanic patients were. Our study based on in-lab PSGs showed no difference between groups in terms of AHI counts, prevalence of OSA or severe OSA (AHI ≥ 30), and measures of positional OSA between Black and White patients. On the other hand, Hispanic patients were doing significantly better than White patients in terms of all measures of positional OSA. While the first study counted hypopneas using the 3% desaturation rule, the MESA study and ours used the 4% desaturation rule. This might be one reason that can potentially explain the different results. Given that the first study and ours were based on in-lab PSGs, while the MESA study was based on in-home measures, the location does not seem to affect AHI measurement. More studies are needed if these two factors (hypopnea counting rule and measurement location) underlie the lack of clear evidence on racial and ethnic differences in sleep ordered breathing between racial and ethnic groups.

While the literature seems to offer insufficient or mixed results on racial and ethnic disparities in restorative sleep such as N3 and REM sleep (Ahn et al., 2021), our data show that Black patients had shorter N3 sleep but longer REM sleep, while Hispanic patients had longer N3 and REM sleep, compared to their White counterparts. Because these stages of sleep are linked to various health outcomes such as important for immune and autonomous nervous system and cardiovascular and brain health (N3) and emotional regulation, mortality and depression (REM) (Gottesman et al., 2024), there is a possibility that these differences can account for some part of the racial and ethnic disparities in these outcomes.

Our study has implications for clinical practice. First, sleep disparities are increasingly thought of as an important contributor to disparities in job performance, productivity, wages, and health. Recently, several investigators argued that sleep disparities may be the fundamental contributor to disparities in health outcomes, including cardiovascular disease, diabetes, and mortality (Jackson et al., 2015; Grandner et al., 2016; Kingsbury et al., 2013). And recommendations have been made to work on sleep as an intervention target for improved cardiovascular health (Jackson et al., 2015; Grandner et al., 2016; Kingsbury et al., 2013), obesity (Grandner, 2017), and disparities in overall health (Buxton and Marcelli, 2010). This is an important and even overdue development. However, from our vantage point after going through a large database of objective sleep measures, we feel that more studies are needed to confirm and understand disparities using objective sleep measures between races and ethnicities, especially because we did not find any significant disparities in sleep time and disordered breathing, two factors that are of paramount importance to cardiovascular disease (Jackson et al., 2015; Linz et al., 2015).

Second, there is a large number of sleep studies that are being done every year at many medical centers across the US. These studies are being recorded in raw data and sleep measures are being scored and pulled into either textual reports or computerized data fields in the electronic health records (EHR) systems. Combined with other data available from EHR systems such as diagnoses, procedures, lab results, vitals, and medication prescriptions, these real-life, objective data can be a large and valuable source of data that should be utilized more frequently in sleep research.