Abstract
Background Earlier reviews documented the effects of a broad range of climate change outcomes on sleep but have not yet evaluated the effect of ambient temperature. This systematic review aims to identify and summarize the literature on ambient temperature and sleep outcomes in a warming world.
Methods For this systematic review, we searched online databases (PubMed, Scopus, JSTOR, GreenFILE, GeoRef and PsycARTICLES) together with relevant journals for studies published before February 2023. We included articles reporting associations between objective indicators of ambient temperature and valid sleep outcomes measured in real-life environments. We included studies conducted among adults, adolescents, and children. A narrative synthesis of the literature was then performed.
Findings The present systematic review shows that higher outdoor or indoor ambient temperatures, expressed either as daily mean or night-time temperature, are negatively associated with sleep quality and quantity worldwide. The negative effect of higher ambient temperatures on sleep is stronger in the warmest months of the year, among vulnerable populations and in the warmest areas of the world. This result appears consistent across several sleep indicators and measures.
Interpretation Although this work identified several methodological limitations of the extant literature, a strong body of evidence from both this systematic review and previous experimental studies converge on the negative impact of elevated temperatures on sleep quality and quantity. In absence of solid evidence on fast adaptation to the effects of heat on sleep, rising temperatures induced by climate change pose a planetary threat to human sleep and therefore human health, performance and wellbeing.
Introduction
There is accumulating evidence that climate change is increasing health risks and notably heat-related illnesses and mortality.1–5 Beyond mortality, hotter ambient temperatures and heat extremes are associated with increased injuries,6,7 hospitalizations,8 mental health issues,9 health-care costs,10 as well as worsened cognitive performance,11 sentiment,4 labor productivity,12 and activity days.13 One proposed pathway of the association between ambient temperature and health outcomes is disrupted sleep.14–16 Shorter sleep duration, poor sleep quality and sleep disorders (e.g., insomnia, sleep apnea) are prospectively associated with the development of cardiovascular17 and metabolic diseases,18 cancer risks,19 mental health disorders,20 and accidents.21 Although the environmental causes of sleep disruption are multi-factorial (e.g., light and noise pollution),22,23 it is likely that rising ambient temperatures due to ongoing climate change will impair sleep in the hottest seasons of the year at a global scale, barring further adaptation.14,24
Exposure to hot and cold ambient thermal conditions demand the human body to mount a thermoregulatory response to maintain a core body temperature rhythm within the normal range required to support physiological functioning and sound sleep.25–27 Extreme heat can alter human core body temperature outside its normal range when air temperatures exceed that of fully vasodilated skin (35°C), with elevated heat-health risks apparent well-below this threshold.3,28–30 Further, sleep onset is closely coupled with night-time core body temperature decline.25 In hot sleeping environments, heat production can exceed heat loss beyond tolerable levels, increasing core body temperature and disturbing the natural sleep–wake cycle with increased wakefulness.26 In daily life, ambient heat can impact core body temperature (and thus sleep) through at least two plausibly interacting pathways: (i) direct exposure during the day imposing thermal strain, cardiovascular strain and/or dehydration which may carry over into the nocturnal resting period,3 and (ii) exposure at night via a combination of nighttime ambient weather conditions and environmental heat transfer (i.e., the energy accumulated in the built environment, conducted and re-emitted in the bedroom at night) reducing the thermal gradient between the body and ambient environment.26,31,32
The largest investigation of the effect of ambient temperature on sleep thus far – a study based on billions of repeated sleep measurements from sleep-tracking wristbands collected in 68 countries over two years – found that increased nighttime ambient temperature shortens sleep duration, primarily through delayed sleep onset, with stronger negative effects during summer months, in lower-income countries, in warmer climate regions, among older adults, females and after controlling for individual and spatiotemporal confounders.24 These results confirmed those from two previous large-scale national analyses of self-reported sleep outcomes from the United States,33,34 the one including the most representative sample showing that a +1°C increase in monthly nighttime ambient temperatures produces an increase of approximately three nights of subjective insufficient sleep per 100 individuals per month.33 Minor et al. (2022) and Obradovich et al. (2017) also included climate change impact projections, and both estimated that rising ambient temperatures may negatively impact human sleep through the year 2100 under both moderate and high greenhouse gas concentration scenarios, with impacts scaling with the level of emissions barring further adaptation.24,33
Critically, climate change and other anthropogenic environmental changes related to increased heat exposure, such as urban heat island, are altering outdoor ambient temperatures where populations reside. Although sound and sufficient slumber underpins human functioning, the current prevalence of insufficient or poor sleep is already elevated in high-income countries (e.g., above 50% in the US, 31% in Europe and 23% in Japan).35 Although insufficient sleep prevalence estimates are relatively lower for low (7%)- and middle (8.2%)-income countries,36 a recent study found that habitants from those countries disproportionally suffered greater sleep loss due to heat,24 suggesting heightened vulnerability to elevated temperatures (i.e., reduced or missing access to personal or collective cooling strategies).37 In parallel, the 1.5°C global temperature threshold above the pre-industrial climate is expected to be exceeded by 2040 under most scenarios of the Intergovernmental Panel for Climate Change, including the increasingly plausible “Shared Socioeconomic Pathway” SSP2-4.5.38 Although there is some evidence of adaptation to increasing temperatures in high-income countries for heat-attributed mortality,39 temperature projections for the coming decades remain concerning for both current and future generations.40–44 In this context, having a precise and comprehensive understanding of the influence of ambient temperature and extreme heat on sleep is crucial.
Earlier reviews documented the effects of a broad range of climate change outcomes on sleep (see14 which included six studies about the specific role of ambient temperature and heat), the influence of the bed micro-environment on sleep physiology,26,45,46 as well as the specific role of humidity in sleep regulation measured mostly in laboratory settings.47 However, no previous systematic review has described the state of the literature on the effect of ambient temperature on sleep in humans under real-life conditions (i.e., observational studies conducted in real-life environments), in contrast with laboratory studies that experimentally manipulate both behavior and temperature in the micro-environment.48–50 This systematic review aims to identify and summarize the literature on ambient temperature, notably heat, and sleep outcomes in a warming world. Specifically, we aim to synthesize the available evidence and research gaps on this topic to inform researchers seeking to explore new facets of the temperature-sleep association, as well as decision makers and interventionists trying to promote climate adaptation.
Methods
Methods for collecting and summarizing data met the standards of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.51 The study protocol was registered in PROSPERO (CRD42021284139).
Inclusion and Exclusion Criteria
Studies were included if they: (i) reported associations between objective indicators of ambient temperature and valid sleep outcomes measured in real-life, environments; (ii) included adults, adolescents, or children; and (iii) were peer-reviewed. Eligible measures of sleep included: self-reported (subjective) sleep questionnaires; accelerometer-based actigraphy and commercial-grade activity monitors (e.g., fitness bands and sleep-tracking wearable devices); sleep sensors and polysomnography. This selection criteria thus excluded sleep-adjacent articles such as those interested in the association between ambient temperature and the prescription of hypnotics, which are only indirectly related to sleep issues.52 Eligible measures for temperature were narrowed to objective records via weather stations, climate reanalysis data or indoor temperature sensors; this criterion excluded articles interested in the associations between sleep outcomes and seasons without measuring ambient temperature,53 subjective perceptions of temperature54 or using climate classifications (i.e., Koppen’s weather climate classification).55 We excluded articles with valid measures of both sleep and temperature when the association between the two outcomes was insufficiently reported (e.g., association plotted with insufficient details to be numerically interpreted and reported),56 or simply not tested (i.e., some articles include measures of temperature and sleep but focus on other outcomes).57,58 Because we focused on ambient temperature and sleep outcomes measured in real-life contexts, experimental studies manipulating in-laboratory temperatures were also excluded.48–50 Beyond heat manipulation, those experimental studies often include invasive skin and rectal temperature measures as well as behavioral constraints and thus can neither be considered ecologically valid in the context of daily life nor the present review.
Data Sources and Searches
Studies were identified by searching PubMed, Scopus, JSTOR, GreenFILE, GeoRef and PsycARTICLES between March and April 2022, and then updated in February 2023. Search strategies and algorithms are available in the supplemental material. Relevant reviews and articles cited in the introduction were also scanned. Specific journals were also inspected, in the sleep literature (e.g., Sleep; Sleep Medicine, Journal of Sleep Research) and the environmental health domain (e.g., Environment International, Environmental Research Letters, Environmental Health Perspectives, The Lancet Planetary Health). After duplicates were removed, titles and abstracts of all studies identified were examined independently by three authors (GC, KM, PB) to determine those meeting the selection criteria.
Data Extraction and Synthesis
An a priori data extraction form was developed and tested with 5 articles. Data were coded from each paper by three coders (KM, EC, PB) and double checked by a single coder (GC). The following information was extracted: first author’s name, sample region and period, study design, main research question, estimation strategy, temperature and sleep assessment methods and outcomes, the inclusion of relevant control variables, main results, conclusion, and relevant additional information. A narrative synthesis of the literature was then performed. We did not perform meta-analyses given the diversity of study designs, temperature and sleep outcomes and the statistical strategies used. A preliminary synthesis was conducted by the first authors (GC, KM) and then discussed and revised by all authors.
Quality rating and risk of bias
We used the items from the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies59 as a basis to develop a custom list of 14 quality criteria relevant to the question of ambient temperature and sleep. For each criterion, we indicated whether or not the study met the requirement (i.e., binary outcome, yes/no). Criteria are displayed in Table 1 below.
Role of the funding sources
The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
Results
Descriptive findings
As depicted in the study flowchart (Figure 1), a first iteration resulted in a total of 1735 independent records. After screening title and abstract, 58 articles were inspected in closer detail and 27 articles were included in the present review.
Included articles were published in journals from diverse academic disciplines, including public health, sleep, physiology, engineering, economics, environmental science and medicine. The oldest article retrieved was published in 1992 and the most recent one 30 years later, in 2022. Figure 2 illustrates the countries in which the studies were performed; the US is the most represented country (42%). In terms of participants’ characteristics, studies included samples of various ages from ∼13 years old to nearly 80 years old. As for identified sex, most studies reported at least ∼30% female sample composition, while females were not included in two studies.60,61 All studies were correlational but six articles used intensive repeated measurements and appropriate statistical methods to estimate covariation between changes in temperatures and sleep outcomes within-participants, and thus are labelled as quasi-experimental studies here24,60,62–65.
Table 2 and Table 3 provide, respectively, a brief and detailed description (constrained by the scope of the current review) of each individual study included. Outdoor ambient temperature was measured via weather stations in 17 articles and indoor temperature was measured using local sensors installed in the home environment (e.g., HOBO temperature data logger, Thermochrons iButtons logger) in 11 articles. One study measured and reported results for both indoor (via local sensor) and outdoor (via weather station) temperature outcomes (see Table 2).66 Sleep measures showed a greater diversity of assessment methods with ten articles using self-reported questionnaires or diaries,33,34,64–71 eight studies using commercial activity monitors,24,62,63,71–75 five studies using polysomnography61,69,76–78, four studies using research-grade accelerometers64,79–81 and three studies using specific sleep sensors60,82,83. Beyond assessment methods, 15 articles used daily (24-hour) aggregated measures of temperature, the remaining 12 articles focused on average nighttime temperature (see Table 3, column “sample period”). For sleep, outcomes ranged from subjective sleep duration34, perception of insufficient sleep33 or sleep quality70 to accelerometer-derived signals providing information about sleep efficiency or wakefulness after sleep onset64 as well as sleep-related outcomes measured via polysomnography such as sleep stages78 and sleep disorders (e.g., episodes of sleep apnea).76
Figure 3 illustrates the range of observed ambient temperatures and latitudes for each of the studies in this review. Aggregating these ranges suggests a greater observational density for hotter compared to colder temperatures (bottom heat bar, Figure 3A) and for northern latitudes compared to both the equatorial region and the southern hemisphere (right heat bar, Figure 3B). Notably, just four (15%) of the studies investigated temperature-related sleep responses in the Tropics, even though the region is home to approximately 40% of the global human population (blue histogram, Figure 3B).24,61,65,69
Narrative synthesis for the association between ambient temperature and sleep
Overall, most studies (i.e., 22 articles, 80%) concluded that higher temperatures were associated with poorer sleep, with two articles explicitly characterizing the association as “modest” in terms of effect size.64,79 Among the five remaining studies: (i) two observed a positive effect of temperature on sleep, with one study performed in China reporting a positive association between temperature and self-reported sleep duration,68 another, performed in Brazil, showing that apnea-hypopnea index was inversely correlated with ambient temperature;77 (ii) two other studies were not explicit in their conclusion about this specific temperature-sleep association, with one not observing a significant association between temperature and sleep duration (but delayed sleep timing),62 and another observing a significant association between temperature and the number of awakenings but focusing on sleep adaptation in expatriates (not the impact of heat);61 and (iii) one study showing a non-significant association (neither positive nor negative) between indoor temperature and sleep duration.75 In regards to sleep apnea specifically, the abovementioned result77 contradicts two other studies included in the present review showing a significant and positive association between temperature and apnea-hypopnea indexes;76,78 while a fourth study did not observe significant relationships between temperature (indoor) and apnea-hypopnea indexes.69
Results are less contrasted with other sleep outcomes and assessment methods. For example, the four studies using research-grade accelerometers64,79–81 all found that higher temperature was associated with poorer sleep (i.e., notably reduced sleep duration and sleep efficiency, as well as increased WASO). In the same vein, four of the six studies using commercial-grade activity monitors found negative associations between ambient temperature and sleep duration,24,63,71 efficiency71,73 (but see74,75 for a non-significant association), WASO and REM sleep,71 nighttime body-movements72 and slow wave sleep.74 Eight studies using questionnaires reported that higher temperature was associated with poorer subjective sleep.33,34,65–67,69–71 One study did not observe a significant association between temperature and subjective sleep in adults with episodic migraine,64 and another study reported a positive association between temperature and sleep duration in a large cohort of students.68 Among the five higher quality studies that fulfilled at least half of the quality criteria,24,33,62,64,65 all concluded that higher temperature was associated with poorer sleep outcomes.
For the six studies that identified significant negative effects of temperature on sleep duration, estimated effect sizes appeared to vary in magnitude from modest to large, with effects scaling across the temperature distribution.24,34,63,67,71,79,80 For instance, a multi-country study found that at the colder end of local temperature distributions, a +1°C increase reduced sleep by just 0.20 minutes during winter months but sleep was reduced by 0.97 minutes per +1°C during the last month of summer,24 whereas a US-based national time use diary study conducted across all seasons estimated a linear sleep reduction of 0.45 minutes per +1°C.34 By comparison, a Boston-based study conducted during a summer heatwave estimated a larger sleep reduction of 2.7 minutes per +1°C increase in bedroom temperature for a sample of young adults,63 while a Shanghai-based study estimated a large magnitude reduction of 5.10 minutes in total sleep time per +1°C increase for a sample of elderly participants.71 Taken together, these results suggest that ambient temperatures may exact both cumulatively and progressively larger sleep impacts at higher ambient temperatures, with one study24 estimating that nights exceeding 25°C push 4,600 additional people to experience a short (<7 hour) night of sleep per 100,000 adults compared to the cold optimum identified.
Other relevant results
Other notable results that further contextualize the temperature-sleep association include (i) the role of seasons, (ii) the impact of other weather and environmental variables, (iii) the presence of non-linear associations between temperature and sleep, (iv) the potential mechanisms explaining how temperature impacts sleep, (v) the specific role of indoor temperature beyond outdoor temperature, and (vi) the role of adaptation measures.
Seasonality
In regards to seasonality, several studies showed that the negative impact of ambient temperature on sleep was more pronounced in the hottest months of the year as discussed earlier in this review,24,60,81,82 although one study failed to replicate this effect.76 A study conducted in the Netherlandshighlighted that the seasonality of sleep duration – with people sleeping less during summer and more during winter – appeared to be mainly driven by ambient temperature (i.e., the percentage of variance explained by seasons decreased significantly when temperature was controlled for).80 However, another study conducted across the US, found that seasons and day length were the only significant variables associated with sleep duration over 3 principal components mainly representing temperature, wind, humidity and cloud coverage (all of these derived components were not significantly associated with sleep duration).62 However, all subjects in this last study were information technology workers working in thermally controlled indoor offices, potentially buffering against outdoor thermal exposures.
Effect modification with other weather variables
Other weather outcomes such as humidity, precipitation, cloud cover or wind speed were included as control variables in several studies.24,34,62,68 In the largest device-based study included in this review,24 authors showed that (i) the impact of temperature on sleep duration was independent from other weather variables and that (ii) higher levels of precipitation, wind speed and cloud coverage marginally increase sleep duration while both low and high levels of humidity significantly reduce sleep duration. Additionally, high diurnal temperature range (the difference between daily maximum and minimum temperature) further reduced sleep duration, albeit to a lesser degree than high night-time temperature.24 Concerning humidity specifically, several experimental studies previously showed that the combination of high levels of indoor humidity and heat create the worst conditions for sleep.32,84 At high ambient temperature, high levels of humidity compromise the body’s evaporative cooling thermoregulatory response and thus increase the risk of heat stress and hyperthermia, possibly challenging the nocturnal core body temperature decline.47 Although evidence remains sparse, one article included in our review investigated the combined effect of heat and humidity on sleep, finding that higher heat index values progressively reduced sleep duration.24
Functional form of the temperature-sleep association
Regarding the functional forms that the temperature-sleep association may take, most studies assumed a linear response but three studies found non-linear associations.24,33,60 For example, one global-scale study observed a monotonic decline in sleep duration as night-time temperature increased, but uncovered an inflection point (i.e., a steeper decline) at 10°C, with progressively larger effects at higher temperatures.24 Interestingly, the same study found that the marginal effect of ambient temperature on sleep loss was over twice as large in the warmest climate regions compared to the coldest areas, consistent with the kinked functional form identified. A second study found a U-shape association between nighttime temperature and wake after sleep onset, as well as an inverse U-shape functional form for sleep efficiency, with the lowest level of wake after sleep onset and higher sleep efficiency in the range 10-15°C.60
Potential mechanisms
Several researchers have discussed putative causal mechanisms linking temperature and sleep or the role of sleep as a mediator of the association between heat and other health and behavioral outcomes,14,15,85,86 although evidence remains limited in this regard. Two studies inferred that body skin temperature may be one of the mechanisms linking temperature and sleep outcomes in the elderly, with higher skin temperature associated with poorer sleep in studies conducted in Japan and China.71,84 A second study investigated the interplay between temperature, sleep and mental health.34 This last study proposed that sleep loss may be a plausible mechanism explaining the effect of higher temperature on emergency department visits for mental disorders and suicide attempts – both of which exhibited consistent functional forms – but the authors did not perform a proper mediation analysis to test this hypothesis. Another study showed that shorter sleep duration induced by higher temperature may be a mediating factor of the negative association between temperature and cognitive functions.63 However, the mediation analysis was only significant for one of five cognitive outcomes assessed. The authors concluded that a small sample size precluded their analysis from yielding conclusive results about the mediating role of sleep in cognitive effects.
Indoor and outdoor temperatures
Only two studies included in the systematic review combined measures and simultaneous analyses of indoor ambient temperature in parallel with outdoor temperature.66,70 A first study, performed in The Netherlands, showed that outdoor temperature was no longer associated with self-reported sleep disturbances when also including indoor (bedroom) temperature in their model specification, with the latter being significantly associated with sleep disturbances.66 Interestingly, the second study, conducted in New York, showed that indoor ambient temperature was systematically higher than outdoor temperature, even in summer and with 92 % of the sample reporting air conditioning ownership.70 However, this study focused on the effect of indoor temperature and did not statistically control for outdoor temperature when testing the association with sleep, rendering it impossible to disentangle the effect of indoor versus outdoor temperature on sleep. A third study measured both indoor and outdoor temperature for elderly residents in Shanghai but only included indoor temperatures in the final analysis.71 The range of reported indoor temperatures closely approximated the outdoor range. A separate study found greater sleep loss on days with larger diurnal temperature ranges and cumulatively larger lagged negative effects of outdoor ambient temperature on sleep loss, suggesting that interior environments may trap ambient heat and prolong temperature-related sleep loss.24
Adaptation measures
Finally, only a minority of studies investigated behavioral or technological adaptations that might protect sleep from heat. One study, conducted in Sydney, Australia, showed that air conditioning was rarely operated compared to open windows and the use of fans, and that air conditioning did not interact with the study results showing that higher bedroom temperature was associated with poorer sleep.73 Similarly, a separate study found that only a third of elderly participants activated their AC units to cool their rooms at night.71 Despite uniform AC and fan access, higher bedroom temperatures were associated with large reductions in sleep quantity and quality. Another study showed that older adults did not report higher levels of hydration (i.e., drinking episodes) when temperature increased, suggesting that participants lacked adaptive behavioral strategies.72 One study also investigated whether people adapt to night-time sleep impacts with compensatory sleep during the day (napping), week (catch up sleep) or across summer months (intra-annual acclimatization), but did not find any evidence of sleep adaptation.24 This same study also found that residents already living in warmer climate regions were more affected per degree of temperature increase than those living in colder areas, suggestive of limited long-run adaptation. This may indicate an upper threshold for human physiology and appears similar to the pattern observed for the temperature-mortality relationship in Europe.43
Narrative synthesis for available climate change projections
Two studies investigated whether warming nighttime temperatures due to climate change would increase the incidence of insufficient sleep in the future.24,33 Obradovich et al. 2017 calculated nighttime temperature anomalies for 2050 and 2099 for the Representative Concentration Pathways “high greenhouse gas (GHG) concentration” scenario (RCP8.5; IPCC) and the United States based on a large empirical self-reported sleep dataset.33 Assuming no further adaptation and that the same functional sleep response persists in the future climate, these authors inferred that climate change may cause between 6 to 14 additional nights of insufficient sleep per 100 individuals by 2050 and 2099, with the greatest increase in climate change-induced nights of insufficient sleep evident in areas of the western and northern United States. Assuming that future adaptation responses do not exceed those observed across the diverse global climate regions examined in the recent historical record, Minor et al. 2022 projected the impact of climate change on sleep for two scenarios: the end- of the century GHG stabilization scenario (RCP4.5) and the increasing GHG concentration scenario (RCP8.5), using empirical sleep data from 68 countries.24 Their globally averaged, population-weighted projections indicate that by 2099, sleep loss might vary between approximately 50 hours per year in a stabilized GHG scenario (RCP4.5) to 58 hours under a less plausible increasing GHG scenario (RCP8.5). By the end of the century, the authors separately estimate that individuals might experience 13 (RCP4.5) to 15 (RCP8.5) excess short (<7 hours) nights of sleep per person per year. These last simulations also indicate that global inequalities in the effect of climate change on sleep loss may scale with future greenhouse gas concentrations, with the warmest regions of the world disproportionately impacted. The authors reported that future sleep loss may also be larger for certain demographics that were less represented in their sample composition, referring to their subgroup analyses that found larger marginal effects for residents from lower-income countries (by a factor of approximately 3), older adults (by a factor of 2) and women (∼25% higher).24
Discussion
Projection studies estimate that, with ongoing climate change, the number of nights with insufficient sleep may significantly increase by the end of the century.24,33 Since the global prevalence of poor sleep is already high, it is crucial to develop a detailed and comprehensive understanding of the effect of temperature on sleep. The present systematic review, which includes 27 original articles, shows that higher outdoor or indoor ambient temperatures, expressed either as daily mean or nighttime temperature, are negatively associated with various sleep outcomes worldwide. This negative effect of higher ambient temperatures on sleep is stronger in the warmest months of the year, among vulnerable populations, notably in the elderly, and in the warmest areas of the world. This result seems consistent across various sleep indicators including sleep quantity, timing or quality and measured via various means including questionnaires, polysomnography, research-grade or commercial activity monitors (see Tables 2 and 3 for a summary of the results). Although the heat-related results are in accordance with those from previous reviews focused on experimental studies manipulating indoor temperatures,26 studies investigating both cold and hot outdoor ambient temperatures have found elevated sleep duration during colder temperatures, suggesting that people may be better at adapting to low ambient temperature than to ambient heat.
Nonetheless, the methodological quality of most studies included in the present systematic review is low (see Table 4). The current literature is notably limited by a relatively poor consideration of key potential individual, spatiotemporal and social confounders. For instance, only 41% (11/27) of the studies statistically adjusted for location-specific seasonality (see Table 4), even though seasonality is associated with changes in daylight, environmental characteristics and behaviors that may also influence or otherwise spuriously associate with sleep. Although the impact of temperature on sleep appears robust – even when controlling for other weather variables (i.e., precipitation, cloud cover, humidity, wind speed, diurnal temperature range), only few studies properly handle these covariates.24,34,62 The negative association between ambient temperature and sleep also remains significant when controlling for adaptation measures such as the utilization of air conditioning, but this should also be further explored.34,65,70,71 Additionally, only 18% (5/27) of the studies assessed the plausible lagged effects of ambient temperature conditions on sleep in addition to the contemporaneous effect (see Table 4).24,61,64,82
Moreover, the relative importance of indoor and outdoor ambient temperatures remains remarkably unclear and virtually unassessed. According to the only study that accounted for both measures, indoor temperature (i.e., measured in the bedroom), more than outdoor temperature, appeared to drive the relationship between ambient heat and sleep.66 It is worth highlighting that climate change is shifting the underlying distribution of local outdoor temperatures, yet adaptation will continue to transpire both outdoors and indoors. Thus, both ambient temperature measures likely impact human sleep through potentially distinct and/or overlapping pathways that should be investigated in future research, both independently – and where possible – in combination. Similarly, it’s unclear how daytime and nighttime temperatures interact to impact sleep. To our knowledge, only one study tested this effect and found that a higher diurnal temperature range was independently associated with decreased sleep duration.24
These limitations and the quality assessment performed for this review help to draw a set of recommendations for future studies. First, researchers and funding agencies should pursue large-scale cooperative projects leveraging repeated person-level sleep measures (including, but not limited to personal sensing technologies) and longitudinal study designs across larger, and more globally diverse populations, and for longer periods of unobtrusive observation.87 The geographic distribution of studies conducted so far does not cover the global distribution of the human population (Figures 2, 3B). Drawing on these timeseries data, researchers should explicitly control for time-invariant between-individual differences to identify within-person temperature-sleep responses.
Second, key spatiotemporally-varying factors should be more consistently controlled in future studies, including daily precipitation, percentage of cloud cover, relative humidity, average wind speed, local climatological conditions for these meteorological variables, and potentially, other relevant ambient environmental factors (e.g., air pollution, day length, etc.).24 Further, for quasi-experimental study designs that seek to identify plausibly causal effects from as good as random variation in ambient temperature fluctuations in real-life environments, researchers should control for location-specific seasonality as well as socio-temporal trends by accounting for day of study-specific shocks due to calendar-induced behavioral changes and macro events that might spuriously associate with both temperature and sleep outcomes.
Third, future studies should strive to investigate the effects of indoor versus outdoor temperatures and diurnal versus nighttime temperatures on sleep.66 Fourth, studies should consider the lagged and cumulative effects of temperature and other meteorological variables on sleep outcomes. Fifth, behavioral and technological adaptation measures should be more consistently measured and included in analyses; this includes hydration, behaviors related to sleep hygiene and the utilization of fans or air conditioners.31,63,72 Sixth, non-linearity in the association between temperature and sleep, as well as other meteorological controls, should be systematically inspected and reported.24 Seventh, a priority should be given to vulnerable populations who received scant attention so far, including habitants of low-income countries, individuals with low financial resources within high-income countries, women in peripartum period,88 developing infants and children,79,89 residents living in the tropics (Figure 3B), residents living in extremely cold and hot environments (Figure 3A), incarcerated populations with limited environmental controls, individuals with mental health disorders, and those with sleep disorders such as insomnia and restless legs syndrome.54 Eighth, and as argued before,15,85 more mechanistic studies are still very much needed to both better understand (i) the potential mediators of the temperature-sleep association beyond physiological parameters (e.g., mental health)9, and (ii), although not the main focus of this systematic review, the contribution of sleep issues in the pathway between ambient temperatures and health outcomes (e.g., mortality).29,90 Given the congruence between the recently identified temperature-sleep functional response and temperature-mental health functional forms,11,24,33,34,91 carefully designed field experiments that enable rigorous assessments of mediation while also experimentally shutting down other temperature-sensitive pathways are needed to inform well-targeted policy responses that bolster heat and sleep resilience. Finally, only 30% (8/27) of the studies in this review investigated temperature-sleep relationships across multiple cities or geographic regions,24,33,34,62,65,66,76,82 and only one featured multiple countries.24 Since spatial autocorrelation is likely high within single county and city studies – likely introducing bias to results and interpretation92 – researchers should strive to carry out research investigations and multi-country collaborations across diverse geographic regions while statistically accounting for spatially correlated errors.
Fostering adaptation
Beyond observational studies, there is an urgent need for interventional studies aiming to foster heat adaptation at different levels, from interventions focused on individuals to environmental and structural modifications.93 At the individual level, evidence-based sleep hygiene measures should be tested to see whether such behavioral measures can foster adaptation to ambient heat. This includes general sleep health measures, such as the avoidance of caffeine, nicotine, alcohol and daytime naps, stress management, sleep timing regularity, management of bedroom noise and artificial light.94 Additionally, heat-specific behavioral adaptations should also be assessed, including cool showers before bedtime, the use of fans (when relative humidity <30%),95,96 water sprays, daytime hydration, reduced bedding, and light cotton clothing.88 Traditional lifestyle interventions, such as the promotion of regular physical activity, are also crucial given the role of physical fitness towards heat adaptation.97,98 These interventions could be implemented through traditional randomized controlled trials or using innovative designs such as just-in-time interventions using weather forecasts that might be particularly relevant for heatwaves.99,100 At the societal level, equitable adaptation should be promoted101. These efforts should ideally be combined with environmental measures such as urban greening,102 urban water features, passive cooling and the improvement of buildings’ insulation and ventilation systems.31,103
Conclusion
The present systematic review shows that higher temperatures are generally associated with poorer sleep outcomes worldwide. Given the absence of solid evidence on fast sleep adaptation to heat, rising average temperatures induced by climate change pose a serious threat to human sleep and therefore human health, performance, and wellbeing. Although this work identified several methodological limitations of the extant literature, a strong body of evidence from both this systematic review and previous experimental studies converge on the negative impact of elevated temperatures on sleep quality and quantity. Pertinent to policymakers, planners and sleep researchers, the intensity of night-time warming is projected to continue to exceed daytime warming in most populated areas,29,104–106 while urbanization will likely further exacerbate night-time ambient heat exposure for most of humanity.107 Even if these relationships and their associated pathways can be refined further through future well-designed observational studies as we advise here, we argue that interventional studies are now urgently needed to foster adaptation and safeguard the essential restorative role of sleep in a hotter world.
Data Availability
Not appropriate
Acknowledgements
The authors want to thanks Pr Manolis Kogevinas and Dr Michael P Mead for their feedbacks on the manuscript.
Footnotes
Funding: We acknowledge support from the Spanish Ministry of Science and Innovation and State Research Agency through the “Centro de Excelencia Severo Ochoa 2019-2023” Program (CEX2018-000806-S), and support from the Generalitat de Catalunya through the CERCA Program. GC has also been awarded with the grant RYC2021-033537-I, supported by MCIN/AEI/10.13039/501100011033 and by the European Union “NextGenerationEU”/PRTR”.
Issue with Table 3. Some studies were missing.
References
- 1.↵
- 2.
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.
- 42.
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.
- 106.↵
- 107.↵