A few degrees, a lot more stress: Associations between room temperature and reactivity to psychosocial stress

Abstract

Increases in ambient temperatures pose a potential risk to mental and physical health. Heat has already been associated with alterations in bodily stress systems, and stress in turn is another major risk factor for various diseases. Yet, the precise impact of moderate temperature rises on stress reactivity remains unclear. This study investigated associations between moderate room temperature variations (17°C–28 °C) and physiological (cortisol, alpha amylase, heart rate, and mean arterial pressure) and subjective stress reactivity in 73 healthy adult participants exposed to a psychosocial stressor, the Trier Social Stress Test. Higher room temperatures were associated with increased reactivity in both the hypothalamic-pituitary-adrenal axis (cortisol concentrations) and the autonomic nervous system (alpha amylase activity, heart rate, mean arterial pressure), but not subjective stress. These findings highlight a potential pathway through which rising temperatures could exacerbate stress-related health vulnerabilities, even in non-extreme conditions.

Highlights

• •Room temperature (17–28 °C) is associated with psychosocial stress reactivity.
• •Effects found for cortisol, alpha amylase, heart rate, and mean arterial pressure.
• •Subjective stress reactivity was not associated with room temperature.
• •Findings were robust to outliers, covariates, and weather station controls.
• •Even mild temperature increases may amplify reactivity to psychosocial stressors.

1. Introduction

The rate of global warming is steadily increasing [1], driving a rise in ambient temperatures across the globe. While first studies explored health effects of increased temperature and heat stress, the precise impact on many physical and psychosocial outcomes is still unclear. Alarmingly, recent reports and meta-analyses linked elevated temperatures to higher physical and mental health-related morbidity and mortality [2,3]. However, most studies so far focused on high to extreme temperatures, making it challenging to assess effects at moderate temperature rises, which will occur even more frequently. In a pioneering review on endocrine effects of heat exposure, the authors concluded that two major bodily stress systems, the hypothalamic-pituitary-adrenal (HPA) axis and the autonomic nervous system (ANS), are strongly affected by heat. Heat results in increased hormone concentrations (e.g., cortisol and noradrenaline) to facilitate heat adaptation, for instance, by promoting heat loss through vasodilation [4]. Considering that stress and subsequent stress system activation affects onset and aggravation of physical (e.g., inflammatory or cardiovascular) and mental disorders (e.g. Ref. [5]), heat-induced alterations in stress system activation may partially explain negative health effects of heat. However, little is known regarding the impact of mild temperature variations closer to the thermoneutral zone on stress reactivity. Understanding these subtler effects is crucial given the frequency of moderate temperature elevations and considering that psychosocial stress is a common burden in many societies.

This report assessed how room temperature differences affect healthy participants’ physiological and subjective reactivity to a psychosocial stressor, the Trier Social Stress Test (TSST), a gold standard test to induce stress in humans without direct physical strain [6]. We quantified HPA axis, ANS, and subjective stress reactivity and examined their associations with variations in room temperature. Importantly, room temperature variation was moderate (17°C–28 °C), allowing for the assessment of variability in stress circuitry reactivity close to the thermoneutral zone, i.e., without extreme heat.

2. Materials and Methods

2.1. Participants and procedure

We analyzed a dataset including 73 healthy participants (38 men and 35 women, age: 19–40 years, M = 26.63, SD = 6.00; see Supplementary Information (SI) 1 for in- and exclusion criteria) as part of a larger study regarding stress reactivity and habituation, performed in Hamburg, Germany, a city with oceanic climate. A screening interview confirmed participants’ health and excluded conditions affecting physiological stress systems. After observing temperature fluctuations during prior testing sessions, we systemically recorded temperature during the TSST for all subsequent participants, resulting in the sample size of n = 73. The study was approved by the local ethics committee (MSH-2021/164, January 18, 2022), performed in accordance with relevant laws and institutional guidelines, and participants received monetary compensation or partial course credit.

Laboratory visits were scheduled to start between 11:30 a.m. and 5:40 p.m., where diurnal fluctuations in HPA axis activity are moderate. Participants were instructed to arrive well-rested for the study and to abstain from consuming psychotropic substances within 24 h prior to the appointment (no alcohol, other drugs, and/or smoking five or more cigarettes). After arrival, participants provided written informed consent and answered several questionnaires not related to the current analysis (for an experimental timeline, see Fig. 1). To assess stress reactivity, we recorded mean arterial pressure (MAP) and heart rate. To this end, diastolic and systolic blood pressure and heart rate were measured with the RS2 (OMRON, the Netherlands), an automated wrist monitor. Assessments were each performed twice with a pause of 30 s, and mean values were used for analyses to ensure more reliable measurements. To assess mean arterial pressure for each time point as a comprehensive measure under autonomic control [7], we used the following formula: diastolic blood pressure + (systolic blood pressure – diastolic blood pressure)/3. Afterwards, we collected saliva samples for salivary cortisol concentration as HPA axis marker and salivary alpha amylase (sAA) activity for noradrenergic and thus ANS activity [8]. Saliva was collected via Salivettes® (code blue, Sarstedt, Germany) with synthetic swabs (see SI 2 for salivary measurement details). Measurements were taken two times before the TSST (T1,T2) and 4 times thereafter (T3-T6), at 10-min intervals. Participants also rated current subjective stress levels (feeling anxious/nervous/stressed, 1–9 scale) at each sampling and once during the TSST. Natural variations in room temperature (17°C–28 °C) were recorded at TSST onset.

Fig. 1.

Experimental timeline including details relevant to this publication. Participants arrived in the lab approximately 55 min prior to the stress induction and answered a battery of questionnaires not reported here. The TSST took place in a separate laboratory room and room temperature was assessed at the beginning of the TSST. CSS = current subjective stress measurement; HR/BP = heart rate and blood pressure; S = salivary measurement; TSST = Trier Social Stress Test.

The TSST [6] is a standardized procedure for psychosocial stress induction in humans, which simulates a job interview. Participants are asked to first prepare a speech to convince a selection committee (one man, one woman; introduced as experts in analyzing facial expressions and body language) that they are the perfect candidate (5 min) and then to hold that speech in front of the committee, which keeps a neutral and detached demeanor (5 min). Afterwards, participants are asked to perform a challenging mental arithmetic task (serial subtraction) for an additional 5 min (see SI 3 for TSST procedure details).

2.2. Statistical analysis

To validate TSST-induced stress responses, one-way ANOVAs with the within-subject factor time were conducted. We then constructed stress reactivity indices: for ANS markers (MAP, heart rate, sAA), T3 minus Min (pre-TSST) as baseline-to-peak measure; for cortisol, global Max minus global Min as recommended by Miller et al. [9]; and for subjective stress (averaged across the three items per time point due to high internal consistency; Cronbach's α = .75-.88), a similar Max-Min score. Room temperature was Pearson-correlated with each stress reactivity index to assess associations between temperature and stress reactivity. To follow up, robustness checks were performed including outlier correction (removal of values exceeding 3 SDs of the mean: cortisol: n = 3; sAA: n = 2; other reactivity measures were free from outliers) and non-parametric Spearman correlations due to non-normally distributed data. Significant correlations were followed by further sensitivity checks: Potential covariates were tested for associations with reactivity indices (T1 stress marker levels, body-mass-index, gender, smoking (yes/no), hormonal contraception (yes/no)) or room temperature (T1 stress marker levels) using correlations or Welch two-sample t-tests. Significant covariates were included in multiple regressions to examine whether room temperature effects persisted after accounting for the covariate and its interaction with room temperature. Following reviewer suggestion, we additionally carried out explorative regressions including gender, room temperature, and gender x room temperature for all stress markers as men and women may differ in both, thermoregulation [10] and stress reactivity [11]. Finally, to rule out that findings were driven by general temperature fluctuations due to, e.g., seasonal influences rather than room temperature per se, analyses were repeated using nearest weather station temperatures at TSST onset and correlations are reported, if they exceed room temperature findings. Significance was set at p < .05 (two-tailed) and 95 % confidence intervals are reported in brackets. Analyses were conducted in R (v4.4.0 [12]), with full data and code available at osf.io/yfpwd.

3. Results

As expected, we observed a pronounced activation of both HPA axis and ANS, as well as increases of subjective stress after the TSST as confirmed by repeated measures ANOVAs and the respective post hoc tests (see Fig. 2 as well as Supplemental Table S1 and S2). Importantly, though, room temperature correlated positively with all physiological stress reactivity markers, demonstrating increased physiological stress reactivity with higher temperatures (see Fig. 3): salivary cortisol (r₇₀ = .313 [.087, .508], p = .008), sAA (r₇₀ = .438 [.229, .608], p < .001), heart rate (r₇₁ = .276 [.049, .476], p = .018), and MAP (r₇₁ = .279 [.052, .478], p = .017). Notably, sensitivity analyses excluding potential outliers resulted in similar findings (cortisol: r₆₇ = .390 [.169, .574], p < .001, sAA: r₆₈ = .275 [.042, .479], p = .021). Results were also robust when using Spearman correlations (see Supplemental Table S3). Likewise, including potential confounders (see Materials and Methods) in regression analyses (baseline sAA, which was associated with sAA reactivity, r₇₀ = .299 [.072, .496], p = .011; body-mass-index associated with MAP reactivity, r₇₁ = .281 [.054, .480], p = .016), did not alter our findings (effect of room temperature on sAA reactivity: β = 44.418 [13.854, 74.983], p = .005; R_adj² = .278, F_3,68 = 10.110, p < .001, MAP: β = 6.753 [.993, 12.513], p = .022; R_adj² = .148, F_3,69 = 5.183, p = .003) except for baseline MAP (negatively associated with room temperature, r₇₁ = −.294 [-.491, −.069], p = .011) which diminished the effect for MAP (β = −4.703 [−10.789, 1.382], p = .128; R_adj² = .082, F_3,69 = 3.149, p = .030). Regarding regressions including gender x room temperature interactions, gender and its interaction with room temperature were only significant for subjective stress reactivity (β = 7.433 [.850, 14.017], p = .028 and β = −.324 [−.639, −.009], p = .044; R_adj² = .092, F_3,69 = 3.439, p = .021) suggesting that women had larger increases in subjective stress then men, but that the difference in reactivity became smaller as temperatures increased. Moreover, our findings were specific to effects of room temperature and not driven by general ambient (outside) temperature variations. In contrast, subjective stress reactivity was not affected by variations in room temperature, but negatively correlated with general ambient temperature measured at the nearest weather station (r₇₁ = −.240 [-.446, −.011], p = .041).

Fig. 2.

Measurements of A) salivary cortisol concentration, B) salivary alpha amylase activity, C) subjective stress levels (arbitrary units, a. u.), D) heart rate, and E) mean arterial pressure over time. First two measurement points were baseline measures. Grey bar indicates the Trier Social Stress Test (TSST). Data show means ± 95 % bootstrapped CI. Statistics of One-Way-ANOVAs over time are reported above (for ηp² with 95 % CI see Supplemental Table S1; for post hoc tests see Supplemental Table S2). ∗p < .001.

Fig. 3.

Pearson correlations of room temperature with stress reactivity markers, i.e., difference scores of (A) salivary cortisol concentration, B) salivary alpha amylase activity, C) subjective stress, D) heart rate, and E) mean arterial pressure. Line indicates linear regression over all data points with 95 % CI. r with 95 % CI is also reported above each graph.

4. Discussion

Our results demonstrate that higher room temperatures are associated with increased HPA axis and ANS reactivity to acute stress. These findings are particularly noteworthy considering that temperature variations were rather mild compared to previous research on ambient temperature and health which investigated markedly higher temperatures or exertion heat (e.g., Ref. [2,4]). In contrast, the temperature range here is close to the so-called thermoneutral zone, but still, temperature had marked effects. Moreover, our participants were healthy, stressor intensity was moderate, and physical strain minimal, as participants stood calmly during the TSST, suggesting that our findings may be a conservative estimate of temperature effects in real-life. Interestingly, temperature did not affect baseline physiological measures (except for a negative association with MAP), indicating that our findings were not merely driven by temperature adaptation processes occurring before stressor onset, which contrasts studies utilizing higher temperatures, where baseline effects of thermoregulation were present (e.g. Ref. [4]). Finally, our results were robust to several sensitivity analyses, further supporting the relationship between higher temperatures and physiological stress reactivity.

Alternative explanations for higher stress marker levels without actual stress reactivity increases, such as temperature-induced changes in salivary cortisol and sAA availability (e.g., through alterations in saliva flow rate or protein binding), seem unlikely to be fully explanatory for our results. First, salivary cortisol appears to be stable under different temperature conditions and while sAA output increases in extreme heat, it appears to have an inverse relationship with temperature under more moderate conditions [8,13]. Second, these alternative explanations would be specific to single stress markers and thus not explain that the effect of temperature pertained to all physiological markers assessed. However, the precise mechanism underlying the observed association between temperature and stress reactivity is still unclear, likely to be highly complex and may depend on several other factors, e.g., humidity [4], participant age [14], and participant's habituation to different climatic zones.

In comparison to physiological markers, subjective stress reactivity was not related to room temperature (only in interaction with gender). However, it was negatively associated with ambient temperature recorded at the nearest weather station, which might reflect seasonal effects on mood regulation, i.e., better mood in warmer seasons [15]. While dissociations between subjective and physiological stress markers are often reported [16], the underlying mechanisms are unclear. However, they suggest that investigating effects of temperature increases by subjective self-reports may miss important physiological alterations.

Importantly, these findings are based on secondary analyses of data collected from a convenience sample, leveraging naturally occurring variation in room temperature during the course of data collection. While this approach enhances ecological validity and reflects realistic laboratory conditions, it inherently limits experimental control and thus precludes causal interpretations. Future studies should employ randomized experimental designs in which room temperature is systematically manipulated, ideally independent of season and external environmental conditions. Additionally, due to the nature of the sample as well as its size, covariates could only be investigated in broad terms, for example, self-reported smoking status was analyzed as binary yes or no instead of more nuanced measures. While gender was included in our statistical analyses, women who were not on hormonal contraception were tested during the luteal phase. Whereas this enabled us to experimentally control effects of the menstrual cycle on thermoregulation [10] and stress reactivity [11], it precludes us from assessing how temperature-related changes in stress reactivity may vary across cycle phases. Lastly, while we show that room temperature variation is correlated with physiological stress reactivity, we cannot exclude that effects are due to normal thermoregulation processes adapting to the given temperatures. However, since we investigated reactivities, i.e., increases in stress marker values instead of absolute values, interactive effects of room temperature and psychosocial stressor seem likely. Ultimately, experimental studies manipulating room temperature in a randomized fashion will be needed to validate and better understand the presented findings.

To summarize, even mild variations in temperature, which are typically not considered heat, affect physiological stress reactivity. This observation has important methodical implications for stress researchers as laboratory temperature and its possible effects on stress reactivity should be taken into consideration in future studies.

CRediT authorship contribution statement

Kim Fricke: Writing – original draft, Visualization, Formal analysis. Marie-Christin Barthel: Writing – review & editing, Methodology, Investigation, Conceptualization. Nina Alexander: Writing – review & editing, Methodology, Conceptualization. Ulf Baumgärtner: Writing – review & editing. Markus Muehlhan: Writing – review & editing, Methodology, Conceptualization. Susanne Vogel: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization.

Funding source

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The publication was funded by the Open Access budget of MSH Medical School Hamburg as part of Project DEAL.

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.

Appendix A. Supplementary data

The following is the Supplementary data to this article: