Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jan 8.
Published in final edited form as: Environ Int. 2025 Nov 19;206:109942. doi: 10.1016/j.envint.2025.109942

Impact of heat exposure on sleep health and its population vulnerability in the United States

Jiawen Liao a,*, Rima Habre a, Erika Garcia a, Sandrah P Eckel a, Joe Kossowsky b,c,d, Megan M Herting a, Wu Chen a, Chenyu Qiu a, Zhenchun Yang a, Rob McConnell a, Frank Gilliland a, Susan Redline d,e,f,**, Zhanghua Chen a,*
PMCID: PMC12777893  NIHMSID: NIHMS2128376  PMID: 41289815

Abstract

Rising temperatures are associated with impaired sleep quantity and quality. Yet, little is known about population vulnerabilities. We aim to assess the sleep health associations with heat exposure and by sociodemographic, spatial, and health-related factors with projected future changes in the United States. This study linked longitudinal sleep records measured by digital wearables in the All of Us Research Program from 2010 to 2022 with gridded meteorological data. We assessed daily daytime and nighttime temperature anomalies as heat exposures. Sleep duration, continuity, onset, and stages associations with heat were assessed using multi-stage linear mixed effect models. Vulnerability to heat by sociodemographic, spatiotemporal, and health-related factors was estimated. We projected reductions in daily sleep duration from 2020 to 2099, under different future climate scenarios in U.S. by climate zones. This study included 14,232 adults with 12,531,244 nights of sleep duration and onset measurements and 8,134,295 sleep continuity and stage measurements in the continuous U.S. We found that a 10 °C increase of the daytime and nighttime temperature anomalies was associated with a decrease of 2.19 (95 % Confidence Interval (CI): 2.09, 2.29) and 2.63 (95 % CI: 2.51, 2.75) minutes of total sleep time, respectively. The effects of heat on sleep duration were greater (9.2 % – 79.4 %, p-value for interaction < 0.05) among females, Hispanics, and those with lower socioeconomic status or chronic health conditions. We also found substantial spatial differences, with the effect estimates being twice as large in marine climate zones compared with other regions. We project 8.5–24.0 additional hours of sleep reduction could occur per person-year in different climate zones by 2099 compared to the 1995–2014. Higher sleep duration reduction happened in June to September and in Marine climate zones on the west coast. This study highlighted adaptation inequalities across regions, sociodemographic and health-related factors.

Keywords: Sleep Duration, Wearable Sensors, Heat Exposure, All of Us Research Program, Climate and Health

1. Introduction

The world’s average temperature is over 1 °C above the pre-industrial era. (Core Writing Team et al., 2023; Romanello et al., 2024) Exposure to higher ambient temperatures can suppress the normal decrease in core body temperature, which signals sleep onset and entry into the deeper stages. (Murphy and Campbell, 1997) Daytime and nighttime heat exposure can also induce sympathetic activation and dehydration, (Cheng and MacDonald, 1985) alter circadian thermo-regulation; (Chevance et al., 2024) and disturb the natural wake-sleep rhythm; (Chevance et al., 2024; Lan et al., 2017; Gilbert et al., 2004) impairing the multi-dimensional sleep health. The heat-induced sleep health impaired can potentially lead to adverse cardiovascular (Bunker et al., 2016; Liu et al., 2022) and mental health disorders (Sun et al., 2021). Observational and quasi-experimental human studies indicated that higher daytime and nighttime temperatures were associated with decreased total duration of sleep in adults (Cepeda et al., 2018; Ferguson et al., 2023; Obradovich et al., 2017; Minor et al., 2022; Hajdu, 2024; Li et al., 2025) and children. (Quante et al., 2019) These studies provided important evidence supporting that a warming climate could lead to shortened sleep duration. However, growing numbers of studies show that sleep quality (such as sleep stages, sleep continuity and macro-structures) is another important factor that predicts physical, cognitive, and mental health. (Ujma and Bódizs, 2024; Djonlagic et al., 2021; Leary et al., 2020) The similar evidence of how environmental heat affects sleep quality is limited, including in the United States. (Chevance et al., 2024) No studies have comprehensively assessed how outdoor heat exposure impacts sleep quality among a large population with individual-level characteristics. Moreover, little evidence is available for the population vulnerability of heat exposure on sleep health, and no study has incorporated seasonality and regional differences in future projections. In this study, we aim to assess the associations between outdoor heat exposure with multi-dimensional sleep health, including sleep duration, sleep stages, continuity, and macrostructures. Furthermore, we plan to leverage rich demographic, socioeconomic, geographical, and health-related data to examine the population’s vulnerability to heat exposure. Lastly, linking with future climate projection models, we projected the impact of heat exposure on total sleep duration reduction under different climate trajectories.

2. Methods

2.1. Data collection for heat exposure on sleep health estimates

This study leveraged longitudinal data in the All of Us Research Program (AoU). AoU is a longitudinal national cohort in the United States and has been described previously. (Mayo et al., 2023) Briefly, AoU began in May of 2017 and aims to enroll over 1 million adults in the United States from over 340 recruitment sites through the collection of multi-modal data, including face-to-face survey questionnaires, physical measurements, electronic health records (EHRs), biospecimens, and genomics at baseline. AoU does not focus on any health or disease conditions, and currently, adults (>=18 years) in the United States are being recruited as study participants. Once recruited, participants are invited to complete healthy surveys, physical examinations, bio-specimen collection and join the future ancillary studies that could involve additional follow-up visits. Participants can choose to share their Fitbit data and EHRs with AoU, and this information will be collected longitudinally both retrospectively and prospectively. This study used the controlled tier version 7 (C2022Q4R9) data from 413,457 AoU participants. A subset of over 10,000 participants shared their Fitbit accounts with AoU, including sleep health data measured by Fitbit wearables. In this study, we restrict the data analysis to the subset of AoU participants who shared their Fitbit device data between 2010 and 2022. Fitbit wearables measure daily sleep outcomes, and have been shown to provide accurate estimates for total sleep duration and wake-sleep patterns compared to polysomnography (PSG)–the gold standard sleep measure, and similar performance in sleep staging as actigraphy measures in several validation studies. (Haghayegh et al., 2019; Stucky et al., 2021) We used the daily total sleep duration as the primary outcome. Secondary sleep outcomes included sleep continuity (i.e., sleep efficiency (SE), i.e., the percentage of the asleep time of total time in bed, and wake after sleep onset (WASO), i.e., the duration of wake period after sleep starts), sleep onset timing, and stage-specific sleep durations (light, deep and rapid eye movement (REM) stages). We further excluded participants with less than 3 months’ sleep tracking and less than 18 years of age when sleep tracking begins. Participants’ demographics, including date of birth, sex, race and ethnicity, education levels, household incomes, employment status, homeownership, marital status, smoking, and alcohol use were selected from surveys completed during the enrollment periods, and ZIP Code socio-determinant of health. Existing disease status including cancer, cardiovascular disease, insomnia, sleep apnea, major depressive disorders, diabetes and obesity during follow-up were extracted from the EHR data that are harmonized using the Observational Medical Outcomes Partnership (OMOP) common data model. (Mayo et al., 2023; Zheng et al., 2024).

We used gridded (~4km × 4 km, 1/24th degree) daily gridMET meteorological data, (Abatzoglou, 2013) including maximum temperature, minimum temperature, maximum relative humidity, precipitation accumulation, and wind speed between 1990 and 2023, and averaged to the three-digit ZIP Code that matched with each participants address at AoU enrollment. (ESRI, 2023) Then, we calculate the daytime temperature anomaly (DTA) as the difference between the daily maximum temperature during the sleep tracking day in 2010–2023 and the long-term mean of daily maximum temperature in 1990–2009 for a given ZIP Code. Likewise, the nighttime temperature anomaly (NTA) was calculated as the difference between the daily minimum temperature and the long-term mean of daily minimum temperature. Other environmental variables matched the ZIP Code included climate zones (hot, cold, mixed, and marines) developed by the Department of Energy Building America Program, based on International Energy Conservation Code (IECC) and the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE). (U.S. Energy Information Administration, 2023; Antonopoulos et al., 2021), 2012–2016 five-year averaged air pollution (PM2.5, NO2, and O3) and artificial light at night (ALAN). More details are provided in Supplementary materials for methods.

2.2. Statistical analysis

A multivariate, mixed-effects panel model was used to assess the associations of heat exposures (DTA and NTA) on the daily sleep outcomes, assuming a linear association as in previous studies. (Chevance et al., 2024; Minor et al., 2022) The model adjusted for time-fixed individual covariates, long-term air pollution and ALAN; time-varying covariates including daily precipitation, relative humidity, wind speed and random intercept of participant, Fitbit device types, day of the year, and state-month. The detailed model specifications were provided in the Supplementary materials for methods. An interaction term between heat exposure and selected interaction variable was added to the model to assess effect modifications, and level-specific exposure-outcome associations were also estimated. We assessed the effect modifications of spatiotemporal (month and climate zones), sociodemographic (age, sex, race and ethnicity, income, education, marital, homeownership), and disease status. R kernel for Jupyter Notebook with plm package on the All of Us Research Workbench was used for the statistical analysis.

2.3. Projection of sleep time reduction

We linked the estimated NTA and sleep duration associations by climate zone and month with the projected NTA from Shared Socioeconomic Pathways (SSPs) scenarios to quantify sleep duration reduction in the future. First, we extracted average minimum surface air temperature anomalies laid out by SSPs predicted by Coupled Model Inter-comparison Projects (CMIP6) ensemble models downscaled at 0.25 degree. We used the CMIP6 ensemble median, 10th, and 90th percentile predictions of monthly-averaged anomalies of daily minimal temperature for 2022–2039, 2040–2059, 2060–2079, and 2080–2099 periods at SSP1-2.6, SSP2-4.5, SSP5-7.0, and SSP5-8.5 scenarios compared to 1995–2014 period provided by Climate Change Knowledge Portal, (World Bank, 2024) and calculated climate zones averages. Second, based on the median, 10th, and 90th percentile prediction intervals (PI) of the monthly minimal temperature anomalies, and the mean and the 95th percentile confidence interval of the associations between NTA and total sleep duration by month and climate zones, we assessed the mean and 95th percentile PI of the sleep duration reduction for each month of the year and yearly sums. In the simulation, we assume the CMIP6 ensemble model predicted monthly-averaged minimal temperature anomalies and estimated associations usually are distributed, and we used the Monte Carlo method by generating 1000 samples to quantify the uncertainties of the PI. (Vicedo-Cabrera et al., 2019).

3. Results

3.1. Characteristics of Participants, sleep Outcomes, and heat exposure Summary

Of the 413 360 AoU participants (controlled tier dataset v7 C2022Q4R9), 14,920 participants had Fitbit sleep tracking data shared with AoU, and 14,220 had at least three months of sleep tracking data and were greater than 18 years when sleep tracking began. There was a mean of 4.4 years of follow-up (Fig. S1). Table 1 summarizes the demographic characteristics of the analytical sample (Table S1 by age groups). The included participants had a mean ± SD of age of 50.5 ± 15.6 years when they provided consent to the AoU study; 9,718 (68.3 %) self-reported as female sex at birth, 11,595 (81.5 %) self-reported as White in the race group, and 12,800 (89.9 %) self-reported as non-Hispanic ethnicity. Participants were from throughout the contiguous United States in all 49 States (Fig. S2), and most lived in cold climate zones (Fig. S3).

Table 1.

Characteristics of AoU Participants with Valid Fitbit Sleep Tracking Included in the Analysis.

Characteristics, mean ± SD or N (%) All
N = 14,232
Age at AoU consent 50.5 ± 15.6
Sex
 Male 4,114 (28.9 %)
 Female 9,718 (68.3 %)
 Other or Not Reported 400 (2.8 %)
Race
 White 11,595 (81.5 %)
 Asian 436 (3.1 %)
 Black or African American 711 (5 %)
 Other 464 (3.3 %)
 Not Reported 1,026 (7.2 %)
Ethnicity
 Not Hispanic or Latino 12,800 (89.9 %)
 Hispanic or Latino 873 (6.1 %)
 Not Reported/Other 559 (3.9 %)
Education
 Advanced Degree 5,367 (37.7 %)
 College Graduate 4,663 (32.8 %)
 High School/College not completed 3,779 (26.6 %)
 Not Reported 423 (3 %)
Marital Status
 Married or with partner 9,311 (65.4 %)
 Not Married or Widowed 4,502 (31.6 %)
 Not Reported 419 (2.9 %)
Household Income
 Above 150 k 3,108 (21.8 %)
 100 k – 150 k 2,864 (20.1 %)
 75 k – 100 k 2,062 (14.5 %)
 35 k – 75 k 3,362 (23.6 %)
 Less than 35 k 1,769 (12.4 %)
 Not reported 1,067 (7.5 %)
Employment Status
 Employed 6,252 (65.0 %)
 Retired or Student 3,291 (23.1 %)
 Out of work/Other 1,689 (11.9 %)
Smoking History
 Never Smoked 9,266 (65.1 %)
 Smoked 4,476 (31.5 %)
 Other/Not reported 490 (3.4 %)
Alcohol Use History
 Never Drank 388 (2.7 %)
 Drank Alcohol 13,565 (95.3 %)
 Other/Not reported 279 (2.0 %)
Homeownership
 Own 9,642 (67.7 %)
 Rent 3,366 (23.7 %)
 Other/Not reported 1,224 (8.6 %)
 ZIP Code Median Home Income 67,529 ± 17 598
 ZIP Code Deprivation Index 0.30 ± 0.057
Climate Zone
 Hot 3,540 (24.9 %)
 Mixed 2,371 (16.7 %)
 Cold 7,717 (54.2 %)
 Marine 604 (4.2 %)
Sleep Outcomes
 Sleep tracking days 12,531,244
 Follow-up periods in year, mean ± SD 4.38 ± 2.46
Primary Outcome
 Total Sleep Durationa in minutes, mean ± SD 393.5 ± 109.8
Secondary Outcomes
Sleep Continuity
 Sleep efficiency (SE)2 %, mean ± SD 91.5 ± 3.6
 Wake after sleep onset (WASO)a in minutes, mean ± SD 50.7 ± 19.7
Sleep Onset and Stage-specific Durations
 Sleep Onset Timea (minutes past midnight), mean ± SD −5.7 ± 197.8
 Light Sleep Duration2 in minutes, mean ± SD 258.7 ± 66.6
 Deep Sleep Duration2 in minutes, mean ± SD 60.9 ± 26.5
 REM Sleep Duration2 in minutes, mean ± SD 82.5 ± 35.5
Open in a new tab
a:

N = 12,531,244 observations from n = 14,232 participants 2: N = 8,134,295 observations from n = 12,934 participants.

There were 12,531,244 person-days of sleep tracking (an average of 880 days per person) and nearly 5 billion minutes of tracking data. From these data, the mean ± SD of daily sleep duration was 393.5 ± 109.7 min for the main sleep period. Among 8,134,295 person-days with sleep quality data when photoplethysmography (PPG) was available, the mean ± SD of SE and WASO was 91.5 % ± 3.6 % and 50.7 ± 19.7 min, respectively. The mean ± SD sleep onset timing was −5.7 ± 197.8 min (i.e., 5.7 min before local midnight clock time). The mean ± SD of the daily light, deep, and REM sleep time was 258.7 ± 66.6, 60.9 ± 26.5, and 82.5 ± 35.5 min, respectively (Table 1 and Table S1 by age group). During the sleep tracking days, the mean ± SD of the daytime temperature anomaly (DTA) and nighttime temperature anomaly (NTA) were 0.75 ± 4.86 °C and 0.9 ± 4.24 °C, respectively. Table S2 summarizes the daily relative humidity, precipitation, and wind speed during the sleep tracking period and 5-year averaged long-term air pollution (PM2.5, NO2, and O3) and ALAN. Among the 14,232 participants, 8,680 (61.0 %) shared EHR data. We identified that 3,126 (22.0 %) were - diagnosed with cancer, 2,058 (14.5 %) with cardiovascular diseases (CVD), 1420 (10.0 %) with depressive disorders, 696 (4.9 %) with diabetes, 924 (6.5 %) with insomnia, 1,221 (8.5 %) with sleep apnea, and 1,669 (11.7 %) with obesity before or during sleep tracking follow-ups (Table S3).

3.2. Associations of temperature anomaly with sleep duration and secondary outcomes of sleep Continuity, onset and stage-specific durations

Table 2 shows the associations between DTA and NTA with the primary outcome of total sleep duration (all stages in the main sleep period) and secondary outcomes of stage-specific durations, sleep continuity, and sleep onset latency, respectively. We found that a 10 °C increase of the DTA and NTA was associated with a decrease of 2.19 (95 % Confidence Interval (CI): 2.09, 2.29) and 2.63 (95 CI: 2.51, 2.75) minutes of total sleep time, respectively. In the analysis of the secondary outcomes, we found that a 10 °C increase of NTA was associated with 0.03 % (95 % CI: 0.03 %, 0.04 %) decrease in sleep efficiency (SE), 0.05 (95 % CI: 0.02, 0.08) minutes increase in wake after sleep onset (WASO), and 1.66 (95 % CI: 1.53, 1.8) minutes increase of sleep onset time. We also found reductions in light, REM, and deep stage sleep by 1.58 (95 % CI: 1.49, 1.67), 0.19 (95 % CI: 0.15, 0.23) and 0.93 (95 % CI: 0.88, 0.98) minutes, respectively. Similar associations were observed for DTA except for WASO and deep stage durations, where non-significant associations were found (Table 2). Table S4 listed linear mixed model results of all covariates.

Table 2.

Estimated Associations of Daytime Temperature Anomaly (DTA) and Nighttime Temperature Anomaly (NTA) with Sleep Duration and Sleep Quality Measures.

Sleep Outcome Nighttime Temperature
Anomaly (NTA)		 Daytime Temperature
Anomaly (DTA)
Point Estimate
(95 % CI)		 P
value		 Point Estimate
(95 % CI)		 P
value
Primary Outcome
Total Sleep Timea (minutes) −2.63 (−2.75, −2.51) <0.001 −2.19 (−2.29, −2.09) <0.001
Secondary Outcomes
Sleep Continuity
Sleep Efficiencyb (SE) % −0.03 (−0.04, −0.03) <0.001 −0.02 (−0.03, −0.02) <0.001
WASOb (minutes) 0.05 (0.02, 0.08) <0.001 0.01 (−0.02, 0.03) 0.504
Sleep Onset and Stage-specific Durations
Sleep Onset Timea (minutes past midnight) 1.66 (1.53, 1.8) <0.001 0.8 (0.69, 0.91) <0.001
Light Sleep Timeb (minutes) −1.58 (−1.67, −1.49) <0.001 −1.52 (−1.6, −1.45) <0.001
Deep Sleep Timeb (minutes) −0.19 (−0.23, −0.15) <0.001 −0.02 (−0.05, 0.01) 0.241
REM Sleep Timeb (minutes) −0.93 (−0.98, −0.88) <0.001 −0.64 (−0.68, −0.59) <0.001
Open in a new tab

Associations were shown as per the 10 °C increase of DTA and NTA. Models adjusted for age, sex, race, ethnicity, education, employment, smoking, alcohol use, homeownership, marital status, year, air pollution, artificial light at night, humidity, precipitation, wind speed, year, day of the week, and random effects of participants, date and state-month.

a:

N = 12,531,244 observations from n = 14,232 participants.

b:

N = 8,134,295 observations from n = 12,934 participants.

3.3. Effect Modification by season and climate zone

Analyses by month of the year and climate zones showed that the nighttime heat exposures had significantly different associations with sleep duration reduction by month of the year and across climate zones in the contiguous United States (Fig. 1A). For month-stratified associations between NTA and total sleep duration, the strongest associations were observed for months in the summer, especially late spring or early summer (e.g., May, June) and late summer or early fall (e.g., September, October) (Fig. 2A). Analysis by climate zones (hot, mixed, cold, and marine climate zones, Fig. S3) (U.S. Energy Information Administration, 2023; Antonopoulos et al., 2021) shows the strongest associations with sleep duration reduction in the marine climate zone. Notably, the estimated effects of nighttime heat exposure on sleep duration in marine climate zones were more than twice those associations for other climate zones. Per 10 °C increase of NTA, the total sleep time reduction was estimated to be 8.56 (95 % CI: 7.63, 9.48) minutes. The reductions of secondary outcomes of light, deep, and REM sleep duration associated with NTA were also greater in late spring and early fall months and marine climate zone (Table S5). Like NTA, daytime heat exposure as DTA was also associated with reduced total sleep duration (Fig. S4A). The sensitivity analysis by both month and climate zones indicated that heat exposure during summer in the marine climate zone had the highest effects on sleep duration compared to the effects during other climate zones and other months of the year (Table S6). We found that in June and October and in the marine climate zone, a 10 °C increase in NTA was associated with 12.21 (95 % CI: 8.78, 15.65) and 9.66 (95 % CI: 6.81, 12.51) minutes of total sleep reduction.

Fig. 1.

Fig. 1.

Open in a new tab

Subgroup Analysis of NTA with Sleep Time by Spatiotemporal (A), Sociodemographic (B) and Disease Status (C). Total sleep duration associations with 10 °C increase of NTA *: interaction p-value (Int. p) < 0.05 comparing stratified results with reference levels in the interaction analysis.

Fig. 2.

Fig. 2.

Open in a new tab

Additional Total Sleep Time Reduction by Climate Zone in the Contiguous United States at Projected Climate Trajectories by Shared Socioeconomic Pathways (SSPs) from 2020 to 2099 Compared to 1995–2014 Periods for Yearly Data (A) and Monthly Data (B).

3.4. Effect Modification by Demographics, socioeconomic Factors, and chronic disease conditions

Results of the analyses by demographic and socioeconomic factors showed that the associations of NTA with total sleep duration differed by participants’ age, sex, ethnicity, household income, education, marital status, homeownership, and menopause indicators (Fig. 1B). Population aged 40—–50 years showed decreased total sleep duration by 2.76 (95 % CI: 2.47, 3.04) minutes per 10 °C NTA, 19.6 % higher than population less than 40 years (p for interaction = 0.016). The older age population (>= 65 years) did not show a significant difference compared to the youngest age group (<40 years). This may be due to the residual impact of differential sociodemographic characteristics among different age groups (Table S1). Female participants also showed decreased total sleep duration by 2.65 (95 % CI: 2.51, 2.79) minutes per 10 °C NTA, 22.9 % higher than males (p for interaction = 0.001). Hispanic ethnicity was associated with a 30.6 % larger reduction in sleep duration per unit increase of NTA compared to the non-Hispanic population (p for interaction = 0.01). Lower socioeconomic status, measured by income, education, marital status, and homeownership, was also associated with a larger reduction in total sleep duration. Compared to participants in the highest income category (>$150 000 annual household income), participants with annual household income less than $35,000 had a 36.6 % higher total sleep time reduction per unit increase of NTA (p for interactions < 0.001). Participants with the lowest education status (high school) showed a 30.5 % higher total sleep duration reduction compared to participants with advanced degrees, per unit increase of NTA (p for interactions < 0.001). Living conditions also appeared to influence an individual’s vulnerability to nighttime heat exposure for sleep duration. Participants who were not married or widowed had a reduction in total sleep duration by 2.98 (95 % CI: 2.76, 3.19) minutes per 10 °C increase in NTA, 30.5 % higher than participants who are married or living with partners (p for interaction < 0.001). Participants who are living in rental residences showed 2.91 (95 % CI: 2.64, 3.17) minutes reductions of sleep duration, 18.8 % higher than those who lived in the residences they own (p for interaction = 0.002). Numeric results are provided in the Table S7, and similar results have been found for DTA (Fig. S4B).

Lastly, the analysis by chronic physical and mental conditions among a subset (N = 8,680, 61.0 %) of participants with EHR found that participants diagnosed with CVD, obesity, and depression (major depressive disorder) appeared to be more susceptible to heat exposure (Fig. 1C). We found that a 10 °C increase in NTA in participants with CVD and obesity was estimated to be associated with a reduction by 2.70 (95 % CI: 2.42, 2.92) and 2.75 (95 % CI: 2.44, 3.07) minutes of total sleep respectively, 24.3 % and 24.0 % higher than participants without (p for interaction = 0.002 and 0.03, respectively), respectively. Participants with depression also showed a reduction in total sleep time by 23.3 % more than participants without this condition.

3.5. Projection of the additional sleep time reduction under future climate scenarios

The projected median, 10th, and 90th percentiles of the monthly mean of the daily minimal temperature anomalies predicted by CMIP6 ensemble models at different climate projections by shared socioeconomic pathways (SSP) scenarios at each climate zone are shown in Table S8. The projected sleep duration reduction is shown in Fig. 2A (Table S9). Under the SSP5-8.5 (high emission and high economic growth) scenario, compared to 1995–2014 period, in 2080–2099 period the population living in cold, hot, marine and mixed climate zones is projected to have an additional reduction in total sleep duration by 8.5 (95 % prediction interval (PI): 2.3, 15.6), 11.8 (95 % PI: 4.0, 21.4), 24.0 (95 % PI: 8.9, 44.9), and 8.5 (95 % PI: 0.9, 17.2) hours per person-year, most of which can be attributed to period between May–October. The national average is estimated as 10.0 (95 % PI: 3.4, 17.0) hours per person-year. The predicted results show significant differences across climate zones, with the population in the marine climate zones showing the highest and more than twice the sleep time reduction than those in the hot, cold, and mixed climate zones. At the monthly level, between May and October under the SSP5-8.5 scenario, the population in marine zones was estimated to have a reduction in total sleep duration of more than 2 h per month, with the highest reduction estimated for August [3.4 (95 % PI: 1.7, 5.6) hours per month, which equals to 7 min of sleep reduction every day] (Fig. 2B, Table S10).

4. Discussion

To our knowledge, this is the first and largest study examining effect modifications of spatiotemporal, sociodemographic, and health-related factors on temperature anomaly exposures and sleep health. Combined with the projected climate trajectories under different SSPs, by the end of the century, we estimated that 8.5–24.0 h of sleep are expected to decrease yearly for different climate zones compared to 1995–2014. The study provides important scientific and public health implications for understanding the potential effects of heat exposure on sleep duration, continuity, and latency, as well as for identifying vulnerable populations. The differences in associations across subgroups highlight in-equities in heat adaptations by climate zones, sociodemographic and health-related factors, including lower socioeconomic status, rental and non-marital living conditions, Hispanic background, and with chronic physical and mental conditions.

A growing number of studies indicate that higher temperature affects sleep in population-based observations. (Obradovich et al., 2017; Minor et al., 2022; Li et al., 2025) Most studies have identified that high temperature and heat exposure reduce sleep duration and quality. (Chevance et al., 2024) The recent 2024 Lancet Countdown on Health and Climate Change (Romanello et al., 2024) newly included an additional sleep health indicator, primarily based on previous population-based studies from 68 countries based on 7 million sleep records, (Minor et al., 2022) and estimated a 6 % reduction in sleep duration occurred globally in 2023 compared to 1986–2005. Based on a national cohort from AoU with longer follow-up and 12.5 million sleep-tracking records, the current study provided similar overall marginal estimates of sleep duration reduction among all participants, (Minor et al., 2022) but also found that associations varied with sociodemographic and health-related factors as well as by climate zones. Specifically, higher sleep loss was observed in the hot and the marine climate zones. We projected that people living in marine and hot climate zones may have much more sleep loss than the previously estimated 11.3 h of sleep loss from the national estimates, (Minor et al., 2022) but lower compared to the a new study estimating more than 30 h of sleep loss in China. (Li et al., 2025) More importantly, we found that the projected sleep reduction is mainly during the summer months, which could result in an average of more than 10 min per day of sleep reduction in August for the population living in the Marine climate zone, potentially reaching clinically relevant effects. The overall associations with shorter sleep duration, decreased REM duration, decreased sleep efficiency, and delayed sleep onset are relevant to health, given the abundant data linking these measures with higher mortality and higher risk of physical and mental diseases. (Ujma and Bódizs, 2024; Djonlagic et al., 2021; Leary et al., 2020).

Currently, there is a scarcity of research that has evaluated variations in the effects of heat on sleep in potentially vulnerable groups. (Obradovich et al., 2017; Minor et al., 2022) Leveraging the rich individual-level characteristics and EHRs and using objective sleep assessments, our analysis suggests that certain demographic groups, including the older population (aged between 40–55 years), females, and Hispanics, are more affected by heat exposure. Notably, our study provided empirical evidence showing lower socioeconomic status (lower income, lower education level, and home rental), single or widow marital status, and existing chronic physical and mental conditions (including CVD, depression and obesity) were associated with an increased vulnerability to heat-related adverse sleep outcomes. Vulnerability among these groups may be explained by personal-level differences in heat exposure not captured by area-level measurements related to individual residence characteristics, such as fewer green spaces and less access to air conditioning (AC) and other cooling strategies. (Mitchell and Chakraborty, 2015; Gronlund, 2014) The differences may also reflect increased susceptibilities and sensitivity to environmental heat and altered thermoregulation from existing chronic diseases. (Liu et al., 2015) These findings align with the increasing literature reporting the individual vulnerabilities of mortality and morbidity of heat exposure, including racial, ethnic, and socioeconomic factors, (Manware et al., 2022) and existing chronic conditions and medications, (Xi et al., 2024) providing crucial public health implications for prioritizing these vulnerable subgroups for reducing adverse sleep outcomes from heat exposure. Our estimated effects of temperature anomalies on sleep duration revealed overall effect sizes (averaged across season, location, and subgroups) that are similar to several other studies of outdoor temperature, which are in the range of 0.2–0.4 min of total sleep reduction per 1 °C increase. (Chevance et al., 2024; Minor et al., 2022) However, our stratified analysis revealed that in the early summer, sleep duration reductions could reach 0.49 min per 1 °C for all participants and 1.33 min per 1 °C for participants in marine climate zones.

Our findings should be interpreted considering limitations and the possibility that random error in the outcome and exposure assessment may underestimate the association and the projection. First, outdoor temperature was determined using the first three digits of the ZIP Code of the participants’ address. While this approach provided much greater spatiotemporal resolution than other population-based studies, (Cepeda et al., 2018; Obradovich et al., 2017; Minor et al., 2022) there still may be variations in outdoor heat exposure within the 3-digit ZIP Code area. Indoor and outdoor temperatures vary greatly due to AC use, building insulation, and cooling roofs, (Jay et al., 2021) and this study does not have information on these heat adaptation strategies. The significantly higher estimated effects of heat exposure on sleep duration among participants in the marine zones in summer may be partially explained by lower residential AC usage than in other areas, as 51 % of households in the 2020 Residential Energy Consumption Survey reported using AC compared to 89 % nationwide. (Residential Energy Consumption Survey (RECS), 2020) Second, the address information was collected during study enrollment; any subsequent change in address during sleep tracking periods could have introduced random errors. Third, Fitbit sleep tracking devices have random errors in estimating several sleep metrics. (Haghayegh et al., 2019) Participants might not always wear Fitbit devices, even though we have conducted data quality checks to remove participants who do not wear Fitbit devices. Fourth, the study population is predominantly female and of the white race and ethnicity who shared Fitbit data with AoU. More studies should be conducted on other populations. Furthermore, the projections of future impact do not consider the changing demographics such as age, sex, racial structures, ALAN, and air pollution or heat adaptations. Many important heat adaptation strategies, such as cooling roofs, AC use, and better building insulation, which can reduce indoor heat exposure and alleviate the adverse effects, have not been considered in the future projections. Lastly, our study projected limited sleep loss with future climate scenarios, it might underestimate the overall impact of future climate trajectories on sleep loss, since we did not include impact of many other extreme weather events (Medicine NA of S Engineering, 2016; Stott, 2016) such as hurricanes, wildfire event and flooding could also increase frequency with changing climate and disrupt sleep (Rifkin et al., 2018).

The present study observed the adverse effects of heat exposures on multiple objectively assessed aspects of sleep health in a large cohort from All of Us Research Program with over 12.5 million nights. Our study’s projection of future sleep loss from future climate senarios highlighted the potential vulnerability of individuals living in marine and hot climate zones. We identified sociodemographic factors such as lower socioeconomic status, non-marital status, living in rental residences, and physical and mental chronic disease might increase vulnerabilities. These findings could inform targeted interventions for improving heat adaptations and resilience.

Supplementary Material

1
2
3

Acknowledgments

This work is supported by the National Institutes of Health (P30ES007048, R01ES033707). JL is supported by the GeoCAFE Scholar Program with grant 2427815 from the US National Science Foundation and CAFE Research Coordinating Center of the NIH Climate Change and Health Initiative (National Institutes of Health U24ES035309).

We gratefully acknowledge All of Us participants for their contributions, without whom this research would not have been possible. We also thank the National Institutes of Health’s All of Us Research Program for making available the participant data examined in this study. The All of Us Research Program is supported by the National Institutes of Health, Office of the Director: Regional Medical Centers: 1 OT2 OD026549; 1 OT2 OD026554; 1 OT2 OD026557; 1OT2 OD026556; 1 OT2 OD026550; 1 OT2 OD 026552; 1 OT2 OD026553; 1 OT2 OD026548; 1OT2 OD026551; 1 OT2 OD026555; IAA #: AOD 16037; Federally Qualified Health Centers: HHSN 263201600085U; Data and Research Center: 5 U2C OD023196; Biobank: 1 U24 OD023121; The Participant Center: U24 OD023176; Participant Technology Systems Center: 1U24 OD023163; Communications and Engagement: 3 OT2 OD023205; 3 OT2 OD023206; and Community Partners: 1 OT2 OD025277; 3 OT2 OD025315; 1 OT2 OD025337; 1 OT2OD025276.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envint.2025.109942.

Footnotes

CRediT authorship contribution statement

Jiawen Liao: Writing – review & editing, Writing – original draft, Visualization, Resources, Methodology, Investigation, Funding acquisition, Conceptualization. Rima Habre: Writing – review & editing, Methodology, Investigation. Erika Garcia: Writing – review & editing, Investigation. Sandrah P. Eckel: Writing – review & editing, Methodology. Joe Kossowsky: Writing – review & editing, Methodology. Megan M. Herting: Writing – review & editing, Resources. Wu Chen: Writing – review & editing, Methodology. Chenyu Qiu: Writing – review & editing, Resources. Zhenchun Yang: Writing – review & editing. Rob McConnell: Writing – review & editing, Funding acquisition. Frank Gilliland: Writing – review & editing, Funding acquisition. Susan Redline: Writing – review & editing, Supervision, Methodology. Zhanghua Chen: Writing – review & editing, Supervision, Resources, Project administration, Conceptualization.

Declaration of competing interest

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.

Data availability

Data and code are published on All of US Research Workbench (Identifier: aou-rw-05580e60), which is available to all researchers with approval to the All of Us Workbench.

References

  1. Core Writing Team, Lee H, Romero J, 2023. Summary for policymakers climate change 2023: synthesis report. contribution of working groups i, ii and iii to the sixth assessment report of the intergovernmental panel on climate change. IPCC, Geneva, Switzerland, 10.59327/IPCC/AR6-9789291691647.001 (accessed Sept 8). [DOI] [Google Scholar]
  2. Romanello M, Walawender M, Hsu S-C, et al. , 2024. The 2024 report of the lancet countdown on health and climate change: facing record-breaking threats from delayed action. Lancet 404, 1847–1896. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Murphy PJ, Campbell SS, 1997. Nighttime drop in body temperature: a physiological trigger for sleep onset? Sleep 20, 505–511. [DOI] [PubMed] [Google Scholar]
  4. Cheng JL, MacDonald MJ, 1985. Effect of heat stress on vascular outcomes in humans. J. Appl. Physiol 2019 (126), 771–781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chevance G, Minor K, Vielma C, et al. , 2024. A systematic review of ambient heat and sleep in a warming climate. Sleep Med. Rev, 101915. [DOI] [PubMed] [Google Scholar]
  6. Lan L, Tsuzuki K, Liu YF, Lian ZW, 2017. Thermal environment and sleep quality: a review. Energ. Buildings 149, 101–113. [Google Scholar]
  7. Gilbert SS, van den Heuvel CJ, Ferguson SA, Dawson D, 2004. Thermoregulation as a sleep signalling system. Sleep Med. Rev 8, 81–93. [DOI] [PubMed] [Google Scholar]
  8. Bunker A, Wildenhain J, Vandenbergh A, et al. , 2016. Effects of air temperature on climate-sensitive mortality and morbidity outcomes in the elderly; a systematic review and meta-analysis of epidemiological evidence. EBioMedicine 6, 258–268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Liu J, Varghese BM, Hansen A, et al. , 2022. Heat exposure and cardiovascular health outcomes: a systematic review and meta-analysis. The Lancet Planetary Health 6, e484–e495. [DOI] [PubMed] [Google Scholar]
  10. Sun S, Weinberger KR, Nori-Sarma A, et al. , 2021. Ambient heat and risks of emergency department visits among adults in the United States: time stratified case crossover study. BMJ 375, e065653. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cepeda M, Koolhaas CM, van Rooij FJA, et al. , 2018. Seasonality of physical activity, sedentary behavior, and sleep in a middle-aged and elderly population: the Rotterdam study. Maturitas 110, 41–50. [DOI] [PubMed] [Google Scholar]
  12. Ferguson T, Curtis R, Fraysse F, et al. , 2023. Weather associations with physical activity, sedentary behaviour and sleep patterns of Australian adults: a longitudinal study with implications for climate change. Int. J. Behav. Nutr. Phys. Act 20, 30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Obradovich N, Migliorini R, Mednick SC, Fowler JH, 2017. Nighttime temperature and human sleep loss in a changing climate. Sci. Adv 3, e1601555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Minor K, Bjerre-Nielsen A, Jonasdottir SS, Lehmann S, Obradovich N, 2022. Rising temperatures erode human sleep globally. One Earth 5, 534–549. [Google Scholar]
  15. Hajdu T, 2024. Temperature exposure and sleep duration: evidence from time use surveys. Economics & Human Biology 54, 101401. [DOI] [PubMed] [Google Scholar]
  16. Li A, Luo H, Zhu Y, et al. , 2025. Climate warming may undermine sleep duration and quality in repeated-measure study of 23 million records. Nat. Commun 16, 2609. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Quante M, Wang R, Weng J, et al. , 2019. Seasonal and weather variation of sleep and physical activity in 12–14-year-old children. Behav. Sleep Med 17, 398–410. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ujma PP, Bódizs R, 2024. Sleep alterations as a function of 88 health indicators. BMC Med. 22, 134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Djonlagic I, Mariani S, Fitzpatrick AL, et al. , 2021. Macro and micro sleep architecture and cognitive performance in older adults. Nat. Hum. Behav 5, 123–145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Leary EB, Watson KT, Ancoli-Israel S, et al. , 2020. Association of rapid eye movement sleep with mortality in middle-aged and older adults. JAMA Neurol. 77, 1241–1251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Mayo KR, Basford MA, Carroll RJ, et al. , 2023. The all of us data and research center: creating a secure, scalable, and sustainable ecosystem for biomedical research. Annu. Rev. Biomed. Data Sci 6, 443–464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Haghayegh S, Khoshnevis S, Smolensky MH, Diller KR, Castriotta RJ, 2019. Accuracy of wristband fitbit models in assessing sleep: systematic review and meta-analysis. J. Med. Internet Res 21, e16273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Stucky B, Clark I, Azza Y, et al. , 2021. Validation of fitbit charge 2 sleep and heart rate estimates against polysomnographic measures in shift workers: naturalistic study. J. Med. Internet Res 23, e26476. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Zheng NS, Annis J, Master H, et al. , 2024. Sleep patterns and risk of chronic disease as measured by long-term monitoring with commercial wearable devices in the all of Us Research Program. Nat. Med 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Abatzoglou JT, 2013. Development of gridded surface meteorological data for ecological applications and modelling. Int. J. Climatol 33, 121–131. [Google Scholar]
  26. ESRI. USA ZIP Code Three-Digit Areas. 2023; published online Dec 20. https://www.arcgis.com/home/item.html?id=2690036a601b4e9a937466884a594938 (accessed Dec 7, 2024).
  27. U.S. Energy Information Administration. Climate Zones - DOE Building America Program, U.S. Department of Energy Building America Program. 2023; published online Nov 21. https://codes.iccsafe.org/content/IECC2012P5 (accessed Nov 17, 2024). [Google Scholar]
  28. Antonopoulos CA, Gilbride TL, Margiotta ER, Kaltreider CE, 2022. Guide to determining climate zone by county: building america and IECC 2021 updates. Pacific Northwest National Laboratory (PNNL), Richland, WA (United States), 10.2172/1893981. [DOI] [Google Scholar]
  29. World Bank. Climate Change Knowledge Portal. 2024. https://climateknowledgeportal.worldbank.org/ (accessed Oct 1, 2024).
  30. Vicedo-Cabrera AM, Sera F, Gasparrini A, 2019. Hands-on tutorial on a modeling framework for projections of climate change impacts on health. Epidemiology 30, 321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mitchell BC, Chakraborty J, 2015. Landscapes of thermal inequity: disproportionate exposure to urban heat in the three largest US cities. Environ. Res. Lett 10, 115005. [Google Scholar]
  32. Gronlund CJ, 2014. Racial and socioeconomic disparities in heat-related health effects and their mechanisms: a review. Curr. Epidemiol. Rep 1, 165–173. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Liu C, Yavar Z, Sun Q, 2015. Cardiovascular response to thermoregulatory challenges. American Journal of Physiology-Heart and Circulatory Physiology 309, H1793–H1812. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Manware M, Dubrow R, Carrión D, Ma Y, Chen K, 2022. Residential and race/ethnicity disparities in heat vulnerability in the United States. GeoHealth 6 e2022GH000695. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Xi D, Liu L, Zhang M, et al. , 2024. Risk factors associated with heatwave mortality in chinese adults over 65 years. Nat. Med 1–10. [DOI] [PubMed] [Google Scholar]
  36. Jay O, Capon A, Berry P, et al. , 2021. Reducing the health effects of hot weather and heat extremes: from personal cooling strategies to green cities. Lancet 398, 709–724. [DOI] [PubMed] [Google Scholar]
  37. Residential Energy Consumption Survey (RECS) 2020. Washington, DC: U.S. Energy Information Administration (EIA), 2023 https://www.eia.gov/consumption/residential/data/2020/ (accessed Nov 11, 2024). [Google Scholar]
  38. Medicine NA of S Engineering, and, Studies D on E and L, Climate B on AS and, Attribution C on EWE and CC. Attribution of Extreme Weather Events in the Context of Climate Change. National Academies Press, 2016. [Google Scholar]
  39. Stott P, 2016. How climate change affects extreme weather events. Science 352, 1517–1518. [DOI] [PubMed] [Google Scholar]
  40. Rifkin DI, Long MW, Perry MJ, 2018. Climate change and sleep: a systematic review of the literature and conceptual framework. Sleep Med. Rev 42, 3–9. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2128376-supplement-1.docx (35.4KB, docx)
2
NIHMS2128376-supplement-2.docx (1.2MB, docx)
3
NIHMS2128376-supplement-3.xlsx (275.6KB, xlsx)

Data Availability Statement

Data and code are published on All of US Research Workbench (Identifier: aou-rw-05580e60), which is available to all researchers with approval to the All of Us Workbench.

RESOURCES