Background
The prevalence of extreme temperatures is rising globally, primarily due to climate change [Reference Wang, Jiang and Lang1, Reference Mora, Dousset, Caldwell, Powell, Geronimo and Bielecki2]. The periods of extreme temperatures are characterized by unusually high or low ambient temperatures relative to local norms, and include heat waves (with temperatures above the 90th or 95th percentile) and cold spells (with temperatures below the 5th or 10th percentile), albeit these temperature thresholds vary by region and context [Reference Wang, Jiang and Lang1]. Climate models project that the frequency, intensity, and duration of extreme temperature events will continue to increase, posing growing risks to public health [Reference Mora, Dousset, Caldwell, Powell, Geronimo and Bielecki2, Reference PoshtMashhadi, Maghsoodi and Wood3]. Currently, around 30% of the global population experiences harmful climatic conditions – including periods of elevated temperatures and humidity – for at least 20 days per year [Reference Mora, Dousset, Caldwell, Powell, Geronimo and Bielecki2]. Closely linked is the fact that heatwaves were linked with about 8%–15% increase in emergency department (ED) visits above the baseline, whereas cold waves were associated with a 5%–12% rise in emergence unit visits [Reference PoshtMashhadi, Maghsoodi and Wood3].
Several lines of evidence have highlighted the negative impacts of climate change on mental health, and there is an increasing emphasis on climate change as a multifactorial issue with both direct and indirect effects on mental health through a complex but interconnected pathophysiological mechanism [Reference PoshtMashhadi, Maghsoodi and Wood3]. Notably, pollution – particularly air pollution – plays a significant role in shaping climate change, and its impacts on human health and well-being through detrimental effects on environmental conditions and global warming [Reference PoshtMashhadi, Maghsoodi and Wood3]. While often considered distinct, pollution and climate change may represent two sides of the same coin, as pollutants such as fine particulate matter and greenhouse gases have not only direct impacts on multiple aspects of human health (e.g., respiratory, cardiovascular, and brain) but also indirect impacts through their influence on climate, especially atmospheric temperatures and the occurrence of extreme weather events [Reference PoshtMashhadi, Maghsoodi and Wood3–Reference Polemiti, Hese, Schepanski, Yuan and Schumann5]. In particular, exposure to pollution has been linked to adverse mental health outcomes, including increased risk of depression, anxiety, and cognitive decline, further highlighting its relevance to the focus of this paper. Emerging evidence has implicated an increased exposure to environmental pollutants (such as fine particulate matters [PM2.5], nitrogen oxide [NO₂], and ozone [O₃]) as one of the plausible explanations through which climate change and pollution have independent or joint impacts on brain health [Reference Radua, De Prisco, Oliva, Fico, Vieta and Fusar-Poli4, Reference Polemiti, Hese, Schepanski, Yuan and Schumann5]. A recent umbrella review reported that both climate hazards and air pollution were associated with a range of mental health outcomes, including mood disorders, via identifiable pathophysiological mechanisms that involve neuroinflammation, oxidative stress, and neuroendocrine disruption [Reference Radua, De Prisco, Oliva, Fico, Vieta and Fusar-Poli4]. Another review of epidemiological and neuroimaging data also linked macroenvironmental exposures to structural and functional brain changes affecting emotional regulation and psychiatric vulnerability [Reference Polemiti, Hese, Schepanski, Yuan and Schumann5]. These shared biological pathways – including inflammation, oxidative stress, hypothalamic–pituitary–adrenal (HPA) axis dysregulation, neurotransmitter imbalance, and hippocampal structural changes – may offer mechanistic insight into how environmental pollution and extreme temperatures converge to exacerbate psychiatric symptoms and mental well-being [Reference Radua, De Prisco, Oliva, Fico, Vieta and Fusar-Poli4, Reference Polemiti, Hese, Schepanski, Yuan and Schumann5]. Including this broader context may enrich discussions and provide unique insight into climate-linked mental health risks [Reference PoshtMashhadi, Maghsoodi and Wood3–Reference Polemiti, Hese, Schepanski, Yuan and Schumann5].
Within this context, extreme temperatures are increasingly recognized as both physiological and psychosocial risk factors that may trigger or worsen symptoms and other outcomes in individuals with mood disorders [Reference Thompson, Lawrance, Roberts, Grailey, Ashrafian and Maheswaran6, Reference Rony and Alamgir7]. In support of this risk formulation, heat exposure was associated with cognitive impairment, irritability, and symptom decompensation, whereas cold exposure has been linked to depressive symptoms and heightened stress responses in individuals with mental illness, including those with mood disorders [Reference Thompson, Lawrance, Roberts, Grailey, Ashrafian and Maheswaran6, Reference Rony and Alamgir7]. Beyond these direct effects, extreme temperatures can also indirectly affect mental health and well-being by disrupting sleep, impairing daily functioning, reducing opportunities for social interaction, and increasing economic stress through higher energy demands or lost productivity [Reference Thompson, Lawrance, Roberts, Grailey, Ashrafian and Maheswaran6–Reference Akiskal, Van Valkenburg, Hersen and Turner8].
Mood disorders – including depression and bipolar disorder – are associated with significant morbidity, functional impairments, and disability [Reference Rony and Alamgir7–Reference Chan, Lam, So, Goggins, Ho and Liu11]. In 2019, mood disorders were one of the leading contributors to the 125.3 million disability-adjusted life years attributed to mental disorders globally [Reference Rony and Alamgir7]. Similarly, the estimates of the prevalence of mood disorders (ranging from 3.3% to 21.4%, depending on the country) are modestly high globally [Reference Kessler, Angermeyer, Anthony, De Graaf, Demyttenaere and Gasquet12]. Considering the recent trends in extreme temperature events, experts have warned that the burden of mood disorders is likely to grow as climate change accelerates [Reference Akiskal, Van Valkenburg, Hersen and Turner8–Reference Chan, Lam, So, Goggins, Ho and Liu11]. However, existing studies vary widely in methodological quality, geographical focus, reported outcomes, and population characteristics [Reference Akiskal, Van Valkenburg, Hersen and Turner8–Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13–Reference Zhou, Ji, Tang, Liu, Yao and Liu31], thereby leading to major gaps in the understanding of the relationship between extreme temperatures and mood disorders. Furthermore, the effects of the interaction between environmental exposures to extreme weather and social determinants of health – such as income, housing, and healthcare access – among individuals with mood disorders remain underexplored.
Given the rising frequency and intensity of extreme temperature events, this review systematically synthesizes existing evidence on their impact on mood disorders, focusing on key outcomes such as symptom severity, hospitalizations, ED visits, outpatient utilization, and adverse events. Ultimately, this study aims to elucidate evidence to inform public health strategies, clinical practice, and future research to mitigate the mental health consequences of climate change on mood disorders, particularly for vulnerable populations.
Methods
This systematic review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [Reference Rethlefsen, Kirtley, Waffenschmidt, Ayala, Moher and Page32] to synthesize findings on the impacts of extreme temperatures on mood disorders across diverse populations and settings. Initial searches were conducted on August 18, 2024, and an updated search and analysis occurred in March 2025. The study protocol was registered on the Open Science Framework (https://doi.org/10.17605/OSF.IO/24EMY). All databases were searched from inception to March 2025 to ensure comprehensive coverage of the literature and inclusion of recent studies.
Definition of extreme temperatures
In this review, extreme temperatures were broadly defined to include exposure to atypical environmental temperature conditions, such as heatwaves, extreme cold, or unusual humidity levels [Reference PoshtMashhadi, Maghsoodi and Wood3]. Recognizing that temperature thresholds vary geographically and temporally, we accepted study-specific definitions of “extreme” that reflected regional and historical baselines, as reported by the original investigators.
Search strategy
The research question was broken down into key concepts to perform a systematic database search and identify a comprehensive list of eligible reports that met the inclusion criteria [Reference Dana, Paynter, Relevo, Hamilton and Nelson33]. MEDLINE/PubMed, PsycINFO, Web of Science, and Scopus were searched. The search strategy (additional details are included in Supplementary Table 1) was developed using strings of search words based on primary concepts [Reference Dana, Paynter, Relevo, Hamilton and Nelson33, Reference Harari, Parola, Hartwell and Riegelman34]. Each database was searched with appropriate controlled vocabulary or Medical Subject Headings, and an explosion was applied only when all subterms were relevant to the research question [Reference Dana, Paynter, Relevo, Hamilton and Nelson33]. Free-text search terms and keywords aligned with the main search concepts [Reference Dana, Paynter, Relevo, Hamilton and Nelson33, Reference Harari, Parola, Hartwell and Riegelman34]. The final search strategy for each was documented following the PRISMA guideline [Reference Rethlefsen, Kirtley, Waffenschmidt, Ayala, Moher and Page32]. Database search results were supplemented by a “snowball” search of relevant articles. Search results were uploaded to Covidence [35] for screening.
Eligibility criteria
This review included studies focused on individuals with mood disorders exposed to extreme temperatures – such as heatwaves, cold, or humidity – using study-specific definitions reflecting regional and historical contexts. Eligible studies reported on outcomes, such as hospitalizations, ED visits, outpatient visits, symptom severity, or adverse events, with full texts available. Studies were excluded if they focused on individuals without mood disorders, non-mood psychiatric conditions, irrelevant exposures or outcomes, or were non-original studies (e.g., reviews, protocols, or editorials).
Screening
Following the removal of duplicate reports, screening was conducted using Covidence [35], which allowed screeners to decide whether an article was “included” or “excluded.” Articles were screened based on their titles and abstracts, and the reviewers were blinded to the decision of others. At least two independent investigators (N.M., M.K.K., and M.L.) performed title and abstract screening followed by a full-text review. Articles with conflicting decisions were automatically marked for resolution. Conflicts were resolved through discussion among investigators and consultation with the senior author (A.T.O.) when necessary.
Data extraction
Data extraction was conducted in parallel by three team members (N.M., M.K.K., and M.L.), with each article reviewed independently by two extractors and overseen by the senior author (A.T.O.) through regular consultation. Extracted data were compiled in Google Sheets by N.M. Key details collected included publication information, study design, region, population size, exposure definitions, variables analyzed, mood disorder types, outcomes measured, methods of collecting data on extreme temperature, and statistical methods. Primary outcomes – hospital admissions, ED visits, outpatient visits, symptom severity, and adverse events – were defined according to each study’s criteria. The information collected on the methods in individual reports that were included in the final review is presented in Supplementary Table 2.
Quality assessment
To do a quality assessment of the studies included in this review, the Joanna Briggs Institute (JBI) critical appraisal tool was utilized. The JBI checklist includes specific criteria tailored to different study designs, which helps in systematically evaluating the methodological quality of studies [35]. The assessment was completed independently by N.M., M.K.K., and M.L. during data extraction. In cases where discrepancies arose between reviewers, disagreements were first discussed between the two extractors. If consensus could not be reached, the issue was reviewed and resolved through discussion with the senior author (A.T.O.) to ensure consistency and methodological rigor.
Data analysis and synthesis
At least two independent authors (N.M., M.K.K., and M.L.) conducted the synthesis, with guidance and consultation from a senior author (A.T.O.). Findings in the included studies were grouped into five main categories: hospital admissions, ED visits, outpatient visits, symptom severity, and adverse events. Key themes such as mood disorder type, temperature extremes, regional differences, and potential mediating factors were identified.
Results
Selection of eligible studies
The initial database search produced 459 reports, while snowball searching yielded a total of 12 articles, yielding a total of 471 articles. Of these articles, 48 were duplicates, leading to a total of 423 articles being screened by title and abstract. After title and abstract screening, 44 articles were eligible for full-text review. From the full-text review, 22 articles did not meet the eligibility criteria, leading to 22 studies being included in the final review [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9–Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13–Reference Zhou, Ji, Tang, Liu, Yao and Liu31]. See Figure 1 for the article selection process using a PRISMA flow diagram.
PRISMA flow diagram.
Characteristics of the included studies
Table 1 presents findings from the included studies. All studies included in this review were observational. This included 10 time-series studies [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16, Reference Lee, Lee, Myung, Kim and Kim19, Reference McWilliams, Kinsella and O’Callaghan20, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27–Reference Yoo, Eum, Roberts, Gao and Chen29, Reference Zhou, Ji, Tang, Liu, Yao and Liu31], 7 case-crossover studies [Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21–Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23, Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25, Reference Zhang, Yang, Xie, Li, Han and Hou30], 3 cohort studies [Reference Jin, Xu, Cao, Wang, Zeng and Pan17, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24, Reference Sung, Chen and Su26], 1 case-control study [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9], and 1 cross-sectional study [Reference Ding, Berry and Bennett15]. Of the included studies, eight were conducted in Asia [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Jin, Xu, Cao, Wang, Zeng and Pan17, Reference Lee, Lee, Myung, Kim and Kim19, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24, Reference Sung, Chen and Su26, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27, Reference Zhang, Yang, Xie, Li, Han and Hou30, Reference Zhou, Ji, Tang, Liu, Yao and Liu31], seven in North America [Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21–Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23, Reference Wang, Lavigne, Ouellette-kuntz and Chen28, Reference Yoo, Eum, Roberts, Gao and Chen29], five in Europe [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference McWilliams, Kinsella and O’Callaghan20, Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25], and two in Oceania [Reference Ding, Berry and Bennett15, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16]. Looking at when these studies were published, 12 were published between 2020 and 2024 [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Jin, Xu, Cao, Wang, Zeng and Pan17, Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21–Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23, Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25, Reference Yoo, Eum, Roberts, Gao and Chen29–Reference Zhou, Ji, Tang, Liu, Yao and Liu31], 8 between 2010 and 2019 [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9, Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Ding, Berry and Bennett15, Reference Lee, Lee, Myung, Kim and Kim19, Reference McWilliams, Kinsella and O’Callaghan20, Reference Sung, Chen and Su26–Reference Wang, Lavigne, Ouellette-kuntz and Chen28], and 2 between 2000 and 2009 [Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24]. Finally, when analyzing the mood disorder outcomes reported in these studies, nine studies focused on ED visits [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Lee, Lee, Myung, Kim and Kim19, Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23, Reference Wang, Lavigne, Ouellette-kuntz and Chen28, Reference Yoo, Eum, Roberts, Gao and Chen29], eight on hospital admissions [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16, Reference McWilliams, Kinsella and O’Callaghan20, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24, Reference Sung, Chen and Su26, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27], two on outpatient visits [Reference Zhang, Yang, Xie, Li, Han and Hou30, Reference Zhou, Ji, Tang, Liu, Yao and Liu31], two on symptom severity [Reference Ding, Berry and Bennett15, Reference Jin, Xu, Cao, Wang, Zeng and Pan17], and one on adverse events [Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25]. As is common in environmental epidemiology, these designs allow for the investigation of population-level associations but are limited in their ability to infer causality. Supplementary Table 2 provides further details on the methods of each study.
Findings from the studies included in this review
Abbreviations: AF, attributable fraction; AN, attributable number; ED, emergency department; ER, excess risk; HR, hazard ratio; IRR, incidence rate ratio; ME, marginal effect; OR, odds ratio; RR, relative risk.
a Results of some studies are partially reported due to the extensive amount of data points reported, and some values have been independently calculated based on data available.
Extreme temperatures and hospital admissions
Studies from Asia [Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Lee, Lee, Myung, Kim and Kim19, Reference Sung, Chen and Su26, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27], Oceania [Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16], and North America [Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21] reported a significant link between high temperatures and hospitalizations for mood disorders in general. Chan et al. [Reference Chan, Lam, So, Goggins, Ho and Liu11] note hospital admissions specifically for episodic mood disorders (RR: 1.3; 95% CI: 1.05–1.71), but not for depressive disorders, for which no significant association was observed (RR: 0.95; 95% CI: 0.72–1.24). In this study, episodic mood disorders and depressive disorders were treated as two distinct diagnostic categories, underscoring that the observed associations applied only to the former. Niu et al. [Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21] reported increased hospital admissions during heat events. While the association with depression was not significant across age groups, there was a significant increase in bipolar disorder admissions among 12–17-year-olds (odds ratio [OR]: 1.39; 95% CI: 1.06–1.83), but not in the 18–25 age group (OR: 1.37; 95% CI: 0.85–2.22) [Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21]. Demographic factors such as age and sex appear to have some influence on these associations. Notably, Hasen et al. [Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16] noted that females aged 15–64 had an incidence rate ratio (IRR) of 1.118, which is notably higher compared with males aged 15–64 with an IRR of 1.057. However, when comparing the sexes aged 75 and older, males had an IRR of 1.271, and females had an IRR of 0.928 [Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16]. In contrast, Trang et al. [Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27] found no statistically significant association between extreme heat (above the 95th percentile, >36°C) and daily hospital admissions (RR = 1.07; 95% CI: 0.82–1.40).
Extreme temperatures and ED visits
The studies included in this review highlight an association between heat waves and increased ED visits, although the strength of these associations varies, with weaker trends observed at extreme temperature percentiles [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9, Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Lee, Lee, Myung, Kim and Kim19]. Deng et al. [Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14] noted that across mood disorders, temperature exhibited a short-term effect on ED visits compared to the heat index. Similarly, Runkle et al. [Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23] note that pregnant individuals with short-term exposure to high ambient temperatures during the warm season were linked to an increased risk of ED visits for depression and bipolar disorder. The risk of depression-related visits was highest in suburban areas (IRR = 1.25 on the same day, increasing to 1.24 over 6 days) and lowest in rural regions, where the association became apparent only after prolonged exposure (IRR = 1.46 at 6 days) [Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23]. Similarly, bipolar disorder visits showed the strongest association in urban areas (IRR = 1.49 on the same day, rising to 1.81 over 6 days), whereas no significant risk was observed in rural settings [Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23].
Lavigne et al. [Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18] examined ED visits and reported an OR of 0.985 (95% CI: 0.959–1.011) for extreme cold exposure, suggesting no statistically significant association between extreme cold and ED visits. Wang et al. [Reference Wang, Lavigne, Ouellette-kuntz and Chen28] also reported no notable associations between cold temperatures and ED visits, aligning with findings from the other studies.
Extreme temperatures and outpatient visits
Looking at the impact of extreme temperatures on outpatient visits, short-term exposure to extreme temperatures, both hot and cold, has been linked to an increased risk of outpatient visits [Reference Zhang, Yang, Xie, Li, Han and Hou30, Reference Zhou, Ji, Tang, Liu, Yao and Liu31]. Zhang et al. [Reference Zhang, Yang, Xie, Li, Han and Hou30] conducted a case-crossover study examining daily hospital outpatient visits for depressive and affective disorders. The study found that exposure to both extreme cold and extreme heat increased the risk of outpatient visits, with cold temperatures having a more pronounced effect [Reference Zhang, Yang, Xie, Li, Han and Hou30]. In Shenzhen, extreme cold (2.5th percentile) over a lag of 0–9 days was associated with a significant increase in outpatient visits for depressive disorders (OR = 1.26; 95% CI: 1.18–1.35) [Reference Zhang, Yang, Xie, Li, Han and Hou30]. Similar associations were observed in Zhaoqing (OR = 1.22; 95% CI: 1.00–1.48) and Huizhou (OR = 1.32; 95% CI: 1.08–1.62) [Reference Zhang, Yang, Xie, Li, Han and Hou30]. Zhou et al. [Reference Zhou, Ji, Tang, Liu, Yao and Liu31] supported these findings, as extremely high humidex values were significantly associated with increased depression visits (OR = 1.179; 95% CI: 1.081–1.286), with an estimated attributable number (AN) of 1,709 visits (95% CI: 819–2,577) and an attributable fraction (AF) of 1.10% (95% CI: 0.50%–1.61%). The effect was more pronounced among females and older adults, highlighting specific vulnerable populations [Reference Zhou, Ji, Tang, Liu, Yao and Liu31].
Extreme temperatures and symptom severity
Extreme temperatures significantly impact the severity of symptoms in mood disorders, particularly depressive disorders. Jin et al. [Reference Jin, Xu, Cao, Wang, Zeng and Pan17] conducted a longitudinal cohort study among 5,600 older adults and found that prolonged residence in regions with extreme temperatures – both cold and hot – was associated with an increased risk of developing depressive symptoms. Specifically, high temperatures were associated with a hazard ratio (HR) of 1.027 (95% CI: 1.013–1.041), and low temperatures with an HR of 1.023 (95% CI: 1.011–1.035) [Reference Jin, Xu, Cao, Wang, Zeng and Pan17]. Similarly, Ding et al. [Reference Ding, Berry and Bennett15], using K10 scores to assess psychological distress, found that a one-unit increase in temperature was linked to a 0.2% increase in the probability of high or very high distress (p < 0.001; 99% CI: 0.1–0.3%). This effect more than doubled to 0.5% at the 99th percentile of humidity (p < 0.001; 99% CI: 0.2–0.7%), suggesting that humidity may amplify the mental health effects of heat [Reference Ding, Berry and Bennett15].
Extreme temperatures and adverse events
When it comes to adverse events, Stivanello et al. [Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25] conducted a case-crossover study in Italy involving adults with pre-existing mental health conditions to examine the impact of extreme heat on adverse outcomes, including all-cause mortality. Among mental health service users, those with depression were significantly more vulnerable to heat-related all-cause mortality, with an OR of 1.083 (95% CI: 1.030–1.138) [Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25]. In contrast, no statistically significant association was observed for individuals with bipolar disorder (OR = 1.027; 95% CI: 0.881–1.197), including those experiencing mania [Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25].
Quality assessment
The quality assessment of the studies included in this systematic review, based on the JBI critical appraisal tool [35], reveals a generally inherent moderate standard of methodological rigor, as seen in Supplementary Table 3. The majority of the studies were time series (n = 10) [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10–Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16, Reference Lee, Lee, Myung, Kim and Kim19, Reference McWilliams, Kinsella and O’Callaghan20, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27, Reference Wang, Lavigne, Ouellette-kuntz and Chen28, Reference Zhou, Ji, Tang, Liu, Yao and Liu31] or case-crossover (n = 7) [Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21–Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23, Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25, Reference Zhang, Yang, Xie, Li, Han and Hou30] designs, with some cohort (n = 3) [Reference Jin, Xu, Cao, Wang, Zeng and Pan17, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24, Reference Sung, Chen and Su26], case-control (n = 1) [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9], and cross-sectional (n = 1) [Reference Ding, Berry and Bennett15] studies. Sample sizes ranged widely, reflecting substantial variability in population coverage. The risk of bias was predominantly low across most studies, with only five studies rated as having moderate risk, mainly due to limited confounder adjustment or incomplete reporting of methodological details. Most studies used appropriate methods for exposure and outcome measurement, although these were not standardized across studies – temperature metrics and mental health outcomes varied, which may limit comparability. Statistical analyses were robust and appropriate in nearly all studies, enhancing the reliability of the reported effect sizes. Overall, the quality of the included studies was rated as high (n = 14) [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16–Reference Lee, Lee, Myung, Kim and Kim19, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21, Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27–Reference Zhou, Ji, Tang, Liu, Yao and Liu31], good (n = 3) [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9, Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23], and moderate (n = 5) [Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Ding, Berry and Bennett15, Reference McWilliams, Kinsella and O’Callaghan20, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24, Reference Sung, Chen and Su26], consistently presented from highest to lowest throughout the manuscript.
Discussion
This systematic review reports current findings from extant literature, highlighting the association between extreme temperatures and several aspects of mood disorders, including increased hospital admissions, ED visits, outpatient visits, symptom severity, and adverse events [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9–Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13–Reference Zhou, Ji, Tang, Liu, Yao and Liu31]. These results are broadly consistent with earlier reviews demonstrating temperature-related changes in mental health outcomes, particularly during heatwaves [Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Ding, Berry and Bennett15, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21, Reference Zhou, Ji, Tang, Liu, Yao and Liu31]. However, this review contributes novel insights by disaggregating outcomes by setting (inpatient, ED, and outpatient) and incorporating symptom-level data, which have been underexplored in prior literature [Reference Thompson, Lawrance, Roberts, Grailey, Ashrafian and Maheswaran6, Reference Rony and Alamgir7].
A key pattern emerging from the findings is the differential effect of extreme heat and cold on mood disorder outcomes. While extreme heat showed a clear and consistent impact, cold exposure presented mixed results and was reported in fewer studies [Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Yoo, Eum, Roberts, Gao and Chen29]. This asymmetry may reflect differences in the acute versus chronic physiological stress associated with thermal extremes or the potential disparities in the sensitivity of surveillance for heat-related versus cold-related outcomes [Reference Li, Zhang, Li, Zhang, Lu and Brown36]. For instance, hospitalizations for mood disorders were consistently elevated during heat events across multiple regions, particularly for episodic and bipolar disorders [Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Sung, Chen and Su26]. In contrast, admissions for depressive disorders showed more variability across studies, which may be due to differences in diagnostic classification, cultural help-seeking behaviors, and hospital coding systems [Reference Li, Zhang, Li, Zhang, Lu and Brown36, Reference Liu, Varghese, Hansen, Xiang, Zhang and Dear37].
The acute mental health burden of extreme heat was reflected in increased ED utilization, particularly for depression and bipolar disorder. These spikes were more pronounced in urban and suburban areas and occurred more rapidly than in rural settings, likely due to greater heat exposure, urban infrastructure, or healthcare-seeking behaviors [Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Runkle, Sugg, Berry, Reed, Cowan and Wertis23, Reference Bell, Gasparrini and Benjamin38]. Some studies noted that risk estimates plateaued or weakened at the highest temperature percentiles, which may be indicative of behavioral adaptation, statistical ceiling effects, or exposure misclassification [Reference Aguglia, Serafini, Escelsior, Canepa, Amore and Maina9, Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Lee, Lee, Myung, Kim and Kim19]. Conversely, the relationship between extreme cold and ED visits remained inconsistent and largely non-significant, suggesting that cold may exert a more diffuse or delayed influence on psychiatric crises that would warrant ED visits, or that current methodologies inadequately capture these effects of cold temperatures [Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Wang, Lavigne, Ouellette-kuntz and Chen28, Reference Yoo, Eum, Roberts, Gao and Chen29].
Outpatient data provided further evidence of the nuanced effects of both heat and cold on mental health care-seeking behaviors. In particular, cold temperatures were more consistently associated with increased outpatient visits for depressive symptoms than with hospital or ED visits, possibly reflecting the more chronic and subacute burden of cold stress [Reference Zhang, Yang, Xie, Li, Han and Hou30]. In contrast, heat – especially when measured using composite indices like humidex – also correlated with increased outpatient presentations, particularly among older adults and women [Reference Zhang, Yang, Xie, Li, Han and Hou30, Reference Zhou, Ji, Tang, Liu, Yao and Liu31]. This finding is biologically plausible, given known age- and sex-related differences in thermoregulation and vulnerability to dehydration, sleep disturbance, and medication side effects during heat events [Reference Zhang, Yang, Xie, Li, Han and Hou30, Reference Zhou, Ji, Tang, Liu, Yao and Liu31, Reference Liu, Varghese, Hansen, Xiang, Zhang and Dear37].
Overall, age and sex emerged as important modifiers of risk to extreme temperatures. Younger adults and women aged 15–64 appeared more vulnerable to heat-related hospitalizations, potentially due to hormonal or occupational factors, whereas adults aged 75 and older faced increased vulnerability linked to comorbidities, impaired thermoregulation, and reduced adaptive capacity [Reference PoshtMashhadi, Maghsoodi and Wood3, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16, Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Liu, Varghese, Hansen, Xiang, Zhang and Dear37]. This pattern was also seen in outpatient and ED visits, emphasizing the importance of intersectional factors – such as age, sex, housing, and healthcare access – in shaping vulnerability to extreme temperature events [Reference PoshtMashhadi, Maghsoodi and Wood3, Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Liu, Varghese, Hansen, Xiang, Zhang and Dear37]. Moreover, socioeconomic factors such as isolation and inadequate access to cooling resources appeared to further intensify the mental health risk to extreme temperature, especially among individuals living in urban heat islands or disadvantaged settings [Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Liu, Varghese, Hansen, Xiang, Zhang and Dear37].
Findings on mood symptom severity showed a worsened trend during thermal extremes, highlighting the direct physiological and psychological impacts of environmental stress [Reference PoshtMashhadi, Maghsoodi and Wood3–Reference Polemiti, Hese, Schepanski, Yuan and Schumann5]. Higher temperatures and humidity levels were associated with increased psychological distress, particularly among older individuals [Reference Ding, Berry and Bennett15, Reference Jin, Xu, Cao, Wang, Zeng and Pan17]. The synergistic effect of heat and humidity appears to magnify distress – possibly by impairing sleep, elevating inflammation, or disrupting circadian regulation [Reference Ding, Berry and Bennett15, Reference Taliercio39, Reference Fischer, Naegeli, Cardone, Filippini, Merla and Hanusch40]. While mechanistic studies remain limited, there is a growing evidence base that suggests that both acute and cumulative exposure to temperature extremes can exacerbate mood disorder symptoms through complex biological pathways, involving inflammation, oxidative stress, HPA axis dysregulation, neurotransmitter imbalance, and hippocampal structural changes [Reference PoshtMashhadi, Maghsoodi and Wood3–Reference Polemiti, Hese, Schepanski, Yuan and Schumann5, Reference Ding, Berry and Bennett15, Reference Taliercio39].
In terms of adverse events, depression was significantly associated with increased mortality during extreme heat, whereas bipolar disorder showed weaker associations. This divergence may relate to differences in medication regimens, social support, or behavioral coping during thermal stress [Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25]. Individuals with depression may be at greater risk of isolation, poor judgment, lack of motivation to access care for both physical and mental health needs, or diminished self-care, thereby increasing susceptibility during heatwaves [Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25, Reference Li, Zhang, Li, Zhang, Lu and Brown36]. This diagnostic heterogeneity underscores the need for targeted risk stratification and diagnosis-specific adaptations in clinical and public health planning.
Despite the overall moderate-to-high quality of the included studies, several methodological considerations must be addressed when interpreting the evidence. Many studies employed robust longitudinal designs with large samples, enhancing causal inference [Reference Bundo, de Schrijver, Federspiel, Luterbacher, Franco, Müller and Vicedo-Cabrera10, Reference Deng, Brotzge, Tracy, Chang, Romeiko and Zhang14, Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker16–Reference Lee, Lee, Myung, Kim and Kim19, Reference Niu, Girma, Liu, Schinasi, Clougherty and Sheffield21, Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Stivanello, Chierzi, Marzaroli, Zanella, Miglio and Biavati25, Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27–Reference Zhou, Ji, Tang, Liu, Yao and Liu31]. However, inconsistencies in exposure definitions (e.g., percentiles vs. absolute thresholds, and humidex vs. raw temperature), variable outcome measures (e.g., administrative data vs. self-report), and inadequate confounder adjustment introduce heterogeneity and limit cross-study comparisons [Reference Chan, Lam, So, Goggins, Ho and Liu11, Reference Christodoulou, Laaidi, Fifre, Lejoyeux, Akaoui and Geoffroy13, Reference Ding, Berry and Bennett15, Reference McWilliams, Kinsella and O’Callaghan20, Reference Shapira, Shiloh, Potchter, Hermesh, Popper and Weizman24–Reference Trang, Rocklöv, Giang, Kullgren and Nilsson27]. These issues highlight the need for standardized exposure metrics and psychiatric outcome definitions to facilitate future comparative and meta-analytic studies to synthesize findings across settings. Moreover, the reliance on ecological and hospital-based data may miss subclinical symptoms or vulnerable groups outside formal care systems. Future research should also prioritize longitudinal designs and stronger confounder control to clarify causal pathways and inform climate-sensitive mental health interventions.
Study Implications
Despite the abovementioned shortcomings, the evidence from this study indicates that extreme temperatures – particularly heat – pose a clinically relevant risk for individuals with mood disorders, albeit this risk can vary geographically and be impacted by environmental and social determinants [Reference Higgins, Thomas, Chandler, Cumpston, Li and Page41–Reference Brandt, Adorjan, Catthoor, Chkonia, Falkai and Fiorillo47]. The public health implications of these study findings are wide-ranging. First, there is a need to be cognizant of the significant mental health impacts of extreme temperatures and the factors that can intensify exposure and reduce adaptive capacity to extreme heat, including living in socioeconomically disadvantaged areas (such as urban heat islands), limited access to cooling, and baseline socioeconomic deprivation [Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Higgins, Thomas, Chandler, Cumpston, Li and Page41, Reference Dang, Wang, Ma, Cai, Guo and Liu43–Reference Brandt, Adorjan, Catthoor, Chkonia, Falkai and Fiorillo47]. Furthermore, vulnerable groups – including older adults, adolescents, pregnant people, and those with severe mental illness – require special attention due to heightened physiological and behavioral risks [Reference Almendra, Loureiro, Silva, Vasconcelos and Santana45–Reference Brandt, Adorjan, Catthoor, Chkonia, Falkai and Fiorillo47]. Such extra attention can include early-warning systems, targeted outreach, and the provision of support tailored to age and cognitive needs [Reference Lavigne, Maltby, Côté, Weinberger, Hebbern and Vicedo-Cabrera18, Reference Higgins, Thomas, Chandler, Cumpston, Li and Page41].
Scalable, interdisciplinary, and multiprong interventions are urgently needed. Such intervention requires an integrated approach that combines clinical, environmental, and public health strategies. These include psychotropic medication reviews (especially those that impair thermoregulation) during heatwaves [Reference Taliercio48, Reference Wong, Iwagami, Taniguchi, Kawamura, Suzuki and Douglas49], hydration support, environmental modifications (e.g., central cooling/heating), and increased provider awareness of climate-sensitive exacerbations [Reference Almendra, Loureiro, Silva, Vasconcelos and Santana45, Reference Taliercio48–Reference Errett, Hartwell, Randazza, Nori-Sarma, Weinberger and Spangler52]. Urban planning must prioritize cooling infrastructure and green space expansion in high-risk areas [Reference Brennan, O’Shea and Mulkerrin53, Reference Jay, Capon, Berry, Broderick, de Dear and Havenith54]. Integrating mental health into climate adaptation strategies – such as Heat Action Plans and real-time psychiatric-weather surveillance systems – is essential [Reference Mahmoudi, Kazmi, Vatandoust, Athari, Sadigh-Eteghad and Morsali44, Reference Klingelhöfer, Braun, Brüggmann and Groneberg46, Reference Taliercio48, Reference Hess, Errett, McGregor, Busch Isaksen, Wettstein and Wheat55–Reference Vaidyanathan57].
Public education and community-engaged, multisectoral approaches are critical for equitable and locally appropriate solutions. This includes culturally tailored outreach, targeted emergency response plans, and accessible materials for populations with cognitive, linguistic, or mobility barriers [Reference Mahmoudi, Kazmi, Vatandoust, Athari, Sadigh-Eteghad and Morsali44, Reference Klingelhöfer, Braun, Brüggmann and Groneberg46, Reference Vaidyanathan57, Reference Hasan, Marsia, Patel, Agrawal and Razzak58]. Recognizing regional and demographic differences in vulnerability is key to mitigating immediate harms and promoting long-term climate resilience and justice [Reference Brandt, Adorjan, Catthoor, Chkonia, Falkai and Fiorillo47].
The study findings also underscore the advantages of innovative ideas and technology in clinical practice. For instance, the use of telemedicine may play a vital role in mitigating the impacts of extreme temperatures on at-risk individuals, particularly elderly patients or those with mobility issues [Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Bell, Gasparrini and Benjamin38, Reference Brennan, O’Shea and Mulkerrin53, Reference Sorensen and Hess59]. By allowing for continued access to care without requiring travel during heatwaves or cold spells, virtual consultations could reduce exposure risks while maintaining continuity of mental health support [Reference Sorensen and Hess59]. Clinicians might also consider incorporating environmental risk assessments into routine care planning, particularly during seasonal transitions or in regions prone to thermal extremes [Reference Sorensen and Hess59]. This may include proactive outreach, medication adjustments, or connecting patients with community-based cooling or heating resources as part of personalized care [Reference Nori-Sarma, Sun, Sun, Spangler, Oblath and Galea22, Reference Bell, Gasparrini and Benjamin38, Reference Brennan, O’Shea and Mulkerrin53, Reference Sorensen and Hess59]. In addition, raising awareness among patients about the potential mental health impacts of extreme temperatures – such as heightened risk of mood disturbances, anxiety, or sleep disruption – and equipping them with coping strategies may further strengthen resilience and support overall well-being [Reference Li, Zhang, Li, Zhang, Lu and Brown36]. Lastly, future research is needed to guide evidence-based clinical practice and guidelines to mitigate the mental health impacts of extreme temperatures and weather events.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2025.10110.
Data availability statement
The search strategy is provided in the Supplementary Material. Full search results and data entry forms are available from the authors upon request.
Acknowledgements
The authors would like to acknowledge Ms. Kaitryn Campbell, health research and information specialist at St. Joseph’s Healthcare Hamilton, for her support. They are also grateful for the feedback that was provided during the presentation of the abstract at the 33rd European Congress of Psychiatry in Madrid, Spain in April 2025.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
The authors declare no competing interests.
Open access
Comments
No Comments have been published for this article.