Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Jun 23;63(6):e70350. doi: 10.1111/psyp.70350

Does Aerobic Capacity Protect Against Stress? Insights From Two Stress‐Induction Paradigms

Peter Raidl 1,2,✉, Anja Vogl 1, Aljoscha Dreisörner 3, Barbara Wessner 1,3, Urs M Nater 3,4, Robert Csapo 1
PMCID: PMC13291345  PMID: 42338035

ABSTRACT

Prior evidence suggests that individuals with higher fitness exhibit attenuated endocrine and autonomic responses to acute psychological and physical stressors. However, it remains unclear whether this stress‐buffering effect depends on the type of stressor. This study examined whether aerobic capacity modulates psychophysiological responses to two stress‐induction paradigms (Trier Social Stress Tests, TSST; Maastricht Acute Stress Test, MAST). We hypothesized that fitter participants show lower stress reactivity, with stronger attenuation expected for the MAST, which includes a physical stress component. Healthy participants (n = 59; 28 women; age: 27.1 ± 5.5 years) completed the TSST and MAST on separate days under standardized conditions. We assessed self‐reported stress, salivary hormones (cortisol, cortisone, DHEAS), and autonomic cardiac regulation (heart rate and root mean square of successive differences of heart periods). Aerobic fitness was determined via graded exercise testing (peak oxygen uptake; V̇O2peak: 50.4 ± 8.6 mL/kg/min). Multilevel regression models were used to examine the associations between V̇O2peak and stress responses. Higher V̇O2peak was significantly associated with lower heart rate across both stress‐induction paradigms (b = 0.99 bpm, p = 0.002). For other outcomes, no significant association with V̇O2peak was found. The associations did not differ between TSST and MAST. Aerobic fitness was linked to lower heart rate during the experimental sessions, indicating a modest cardiovascular advantage under acute stress. Contrary to our hypothesis, this effect did not differ between the two stress paradigms, suggesting that the influence of fitness was not specific to the type of stressor. Aerobic capacity, therefore, seems to reduce cardiovascular load during stressful situations in a general manner.

Keywords: cardiac capacity, cortisol, HRV, stress

Impact Statement

We demonstrated that higher aerobic capacity is associated with similarly reduced heart rate responses to a psychosocial stress paradigm (TSST) and a psychophysical stress paradigm (MAST), irrespective of sex. This supports the idea that the stress‐buffering effect of physical fitness is generalized across different types of acute stressors and reduces the load to the cardiovascular system.

1. Introduction

The stress response is a crucial physiological function for maintaining and re‐establishing dynamic homeostatic control (Cohen et al. 2016). The hypothalamic–pituitary–adrenal (HPA) axis and the autonomic nervous system (ANS) are systemic regulators, and their effectors are frequently used as measures of the stress response (Cohen et al. 2016).

While activation of the HPA axis and the ANS is necessary for optimal function and survival, chronic or inadequate stress is detrimental to health (Kivimäki et al. 2023; Wirtz and von Känel 2017; BS et al. 2016). Given its psychophysiological nature, the impact of stress on health can be addressed on multiple levels. One potential avenue is to modulate HPA and ANS function through exercise training. It has been suggested that training‐induced improvements in fitness could confer a stress‐buffering effect (Sothmann 2006). Combined evidence from different areas of research is promising: The evidence on exercise programs for stress‐related mental health and well‐being is robust (Gerber and Pühse 2009; Singh et al. 2023; Verhoeven et al. 2023). Also, animal models indicate reduced glucocorticoid response to mild stressors after voluntary exercise programs (Nowacka‐Chmielewska et al. 2022). However, direct effects of fitness on the stress response in humans are inconsistent. Earlier research found some evidence for improved cardiac regulation, but this work is limited by heterogeneous protocols, small samples, and predominantly male participants (Jackson and Dishman 2006; Forcier et al. 2006). While nearly all studies focus on aerobic fitness, only a small proportion use gold‐standard cardiopulmonary exercise tests to measure peak oxygen uptake (Mücke et al. 2018).

More recent studies increasingly used the Trier Social Stress Test (TSST) as a stress‐induction protocol (Mücke et al. 2018), improving standardization but narrowing generalizability to psychosocial stressors. Adaptations relevant to physically aversive demands may therefore have been missed. Thus, it remains unclear whether and when higher fitness buffers human stress responses and whether any buffering is stressor‐specific.

Different stressors may recruit distinct regulatory dynamics (Mason 1971), leading to specific adaptations. For example, a repeated Wingate test leads to greater changes in heart rate variability (HRV) but a smaller subjective stress response compared to the TSST (Hermann et al. 2019). Similarly, athletes and regular exercisers regularly expose themselves to physical stressors, leading to specific adaptations depending on the training regimen: endurance athletes show a higher catecholamine response to the same relative intensity, but a lower response to fixed‐intensity endurance challenge than untrained individuals, indicating adaptive regulation of the adrenal medulla (Zouhal et al. 2008; Deuster et al. 1989). Further adaptations include improved baroreflex sensitivity and cardiac‐autonomic control (Lester et al. 2021; Tulppo et al. 2003), as evidenced by attenuated cardiovascular and autonomic responses to a cold pressor task (Ifuku et al. 2007). Athletes show higher pain tolerance (Tesarz et al. 2012), which may reflect enhanced pain modulation and habituation to interoceptive discomfort and, in turn, reduced nociceptive reflex activation during painful stressors (Huang et al. 2021).

Consistent with habituation, stress responses to repeated TSST exposure attenuate but do not fully transfer to protocol variants (Allen et al. 2014). Taken together, this indicates demand‐specific adaptations of stress regulation. We therefore hypothesized that physical fitness may be more relevant for coping with acute physical stressors compared to psychosocial stressors, such as the TSST.

To test the potential specificity of the stress‐buffering effect of physical fitness, we implemented a randomized crossover design, comparing the stress response to the TSST with that to the Maastricht Acute Stress Test (MAST) (Smeets et al. 2012). In contrast to the TSST, the MAST combines psychosocial stress with the physical challenge of cold exposure, allowing us to test whether physical fitness provides greater protection against stressors that include a physical component. Aerobic fitness was assessed using a graded exercise test (GXT), and self‐reported physical activity (PA) was added as a separate construct (Caspersen et al. 1985; van der Mee et al. 2023).

We hypothesized that (1) higher aerobic capacity (V̇O2peak) and (2) PA would be associated with attenuated psychophysiological stress responses. This would be reflected as (a) reduced glucocorticoid response, (b) lower heart rate (HR) and higher HRV (root mean square of successive differences; RMSSD), indicating more favorable autonomic regulation, and (c) lower subjective stress ratings. Given the differing sensory and psychosocial demands of the two stress paradigms, we further hypothesized that (3) the magnitude of this fitness‐related stress‐buffering would differ between the MAST and TSST, with a stronger protective influence of fitness expected for the MAST. While sex differences in laboratory stress tests have been studied extensively (Liu et al. 2017), to our knowledge, no study to date has examined sex as a factor in relation to the stress‐buffering effect of fitness. In line with previous studies, we hypothesized (4) that women exhibit a lower stress response than men.

Beyond this main effect of sex (H4), we additionally examined sex as a moderator of fitness–stress associations and considered ventilatory thresholds as alternative indicators of aerobic fitness, as they provide physiologically distinct, motivation‐independent markers of cardiorespiratory fitness (exploratory).

2. Materials and Methods

2.1. Study Design

The study was approved by the Ethics Committee of the University of Vienna (Reference Number: 00960). Data processing and all hypotheses were preregistered on OSF (https://osf.io/txsc5). Data collection took place between May 2024 and June 2025 at the Department of Sport and Human Movement Science, University of Vienna. The study was designed as a crossover experiment with a balanced sample of male and female subjects. Each participant completed both a TSST and MAST on two separate days, spaced 1 month apart. Female participants were always scheduled 21 ± 1 days after their first day of the menstrual cycle. The order of the tests was randomly allocated and counterbalanced, stratified by sex. To reduce participant enrollment time while minimizing interference between test days, the assessment of aerobic capacity via GXT was scheduled 2 weeks after the first experimental day.

We used the mlmpower package (v. 1.0.10) to perform a Monte Carlo simulation of the presented multilevel model (see data processing and statistical analysis). We estimated realistic and meaningful effects from earlier studies (Mücke et al. 2018; Smeets et al. 2012) as small to medium for main effects of condition and fitness (R 2 = 0.06–0.12) with high variance explained by cluster structure (ICC = 0.2–0.3). A cross‐level interaction was expected, with a small effect size (R 2 = 0.02). Typical variance for z‐transformed V̇O2peak was based on internal data sets (σ = 8 mL/kg/min). We used 2000 simulations under an accepted 5% type 1 error rate. A sample of n = 60 participants was estimated to achieve a power of 90%, 65%, and 30% for the fixed effects of V̇O2peak, condition, and their interaction, respectively.

The study was announced on the institute's website. After filling out a contact form, all interested volunteers were contacted directly. Eligibility was screened via phone or email, after which participants were invited to the laboratory to provide informed consent. No financial compensation was offered. However, participants received their individual GXT results, which were discussed with an experienced sport scientist. We did not inform participants about the specifics of the experiments. The aim of the study was described as understanding the hormonal response to “challenges” (instead of stress). Participants were therefore partially blinded to the study's aim and structure. It was not possible to blind the experimenters.

2.2. Participants

We aimed to recruit a balanced sample of male and female participants aged 18 to 40 years. Exclusion criteria were acute injuries or health concerns, use of medications or performance‐enhancing drugs regulated by the World Anti‐Doping Agency, regular intake of ashwagandha or rhodiola rosea within the last month (Lopresti et al. 2022), diagnosed chronic diseases, and conditions affecting hormonal balance, including the use of hormonal contraception, pregnancy, lactation, or irregular menstrual cycles (less than 9 consecutive periods in the last year or cycle length < 21 or > 35 days (Elliott‐Sale et al. 2021)). Participants were also required to maintain a regular sleep schedule for the time of the experiment, avoid traveling across multiple time zones in the weeks of testing, and refrain from intense exercise, supplement intake, or alcohol and other drugs for at least 24 h before each experimental day. In addition, they were asked to abstain from caffeine and nicotine for 18 h, and from eating, drinking caloric beverages, chewing gum, or brushing their teeth for 1 h prior to testing. Participants were asked for compliance at the beginning of each experimental day. If they had not complied, the test was rescheduled.

2.3. Study Protocol

An overview of the experimental design is shown in Figure 1. Appointments were kept consistent within each participant and scheduled between 2 pm and 4 pm to reduce bias related to the circadian variability of hormonal markers (Goodman et al. 2017). It has previously been demonstrated that the stress response is influenced by the menstrual cycle (Kajantie and Phillips 2006). Therefore, female participants attended the stress tests on day 21 (±1) of their self‐reported menstrual cycle.

FIGURE 1.

FIGURE 1

Open in a new tab

Study design. Two experimental days were separated by a one‐month washout period (or the length of one menstrual cycle for female participants). The order of the experiments was counterbalanced within each sex. After a baseline assessment, participants completed either the Trier Social Stress Test (TSST) or the Maastricht Acute Stress Test (MAST), followed by a 30‐min recovery period. Vertical arrows indicate measurement time points for State Anxiety Inventory (STAI), the visual analog scale for perceived stress (VAS), and saliva sampling. Heart rate (HR) was recorded continuously throughout the experiment, with explicit time windows for standardized analyses.

Written informed consent was obtained after participants received detailed information at the beginning of the first visit. They were then equipped with a chest strap heart rate monitor and seated in a quiet waiting room. To minimize personal interactions, the investigator leading the experiment left the room between measurements. The waiting room contained only neutral reading material, a pen, and paper. To control for individual variations in blood glucose, which are directly linked to cortisol regulation, participants consumed 30 g of dextrose dissolved in 200 mL of water (Labuschagne et al. 2019) and subsequently rinsed their mouth with 100 mL of water to prevent contamination of saliva samples. Fifteen minutes after ingesting the glucose drink, participants were guided into the test room and given instructions for the upcoming stress test. After the stress test, they were returned to the waiting room for follow‐up measurements of the stress response.

2.4. Stress‐Induction Protocols

2.4.1. TSST

The TSST consists of a five‐minute mock job interview, followed by a five‐minute mental arithmetic task performed in front of a committee. The present study followed the guidelines provided by Labuschagne et al. (Labuschagne et al. 2019). The committee comprised one male and one female interviewer, both dressed formally and instructed to maintain neutral facial expressions and body language to increase uncertainty. Interviews with female participants were conducted by a male interviewer, and those with male participants by a female interviewer. After receiving task instructions, participants were returned to the waiting room for a five‐minute preparation phase. A saliva sample was collected at the end of this phase to capture anticipatory changes in stress (see Physiological measurements).

2.4.2. Mast

The MAST combines elements of the TSST with repeated cold exposure of the dominant hand as a physical stressor. The MAST was performed under the same conditions as the TSST, following the protocol described by Smeets et al. (Smeets et al. 2012). Briefly, each test was administered by a single tester. To ensure consistency between the protocols, the sex of the tester was kept constant within each participant. Male testers administered the test to women, and vice versa. Participants received task instructions while seated at a laptop computer. They were then repeatedly asked to immerse their dominant hand in ice water. The cold exposure lasted between 60 and 90 s and alternated with arithmetic tasks similar to those in the TSST.

2.5. Measures

2.5.1. Physiological Measures

Saliva samples were collected at six time points: at baseline, during stress anticipation, and after each stress test protocol (see Figure 1). Synthetic swabs (Salivettes, Sarstedt, Germany) were placed in the mouth for 2 min and stored at 5°C until the end of the experiment. Subsequently, they were centrifuged at 2000 RCF (4°C) and frozen at −20°C until analysis. Analyses were performed by an external laboratory (BIOLYZ Inc., Tulln, Austria) using targeted liquid chromatography‐mass spectroscopy (LC–MS) with a ZenoTOF 7600 mass spectrometer (SCIEX, USA) and an ACQUITY M‐CLASS LC system (Waters, USA). Concentrations of free cortisol, cortisone, and DHEAS were quantified against heavy isotope‐labeled internal standards.

Heart periods were continuously recorded at 1000 Hz using a chest‐strap ECG sensor (Polar H10, Finland) on each experimental day. Three resting time frames were analyzed: after glucose administration, and after collection of the third and fifth saliva samples (see Figure 1). During these periods, participants were instructed to sit still, rest their backs against the backrest of a chair, and breathe normally. Raw data were filtered for artifacts and ectopic beats using an adaptive threshold filter with a 50‐beat sliding window and thresholds between 1333 to 300 ms, corresponding to heart rates of 45 to 200 bpm. HRV metrics were calculated in 5‐minute segments across the entire experiment using 1‐minute moving windows. Mean HR was chosen as an overall metric of cardiac autonomic load. The root mean square of successive differences (RMSSD) served as the primary outcome for vagally‐mediated cardiac control (Quigley et al. 2024).

2.5.2. Self‐Reported Measures

Participants completed the Depression Anxiety Stress Scale (DASS‐21) at the beginning of each experimental day. The DASS‐21 assesses symptoms experienced during the preceding week across three subscales—depression, anxiety, and stress—using a 4‐point response format (0 = “did not apply to me at all” to 3 = “applied to me very much or most of the time”). Scale scores were computed according to the official scoring guidelines, with higher scores indicating greater symptom severity.

Participants further completed the 10‐item State Anxiety Inventory (STAI10) (Grimm 2009) before and after each stress test (see Figure 1). This questionnaire is widely used to assess transient anxiety in laboratory stress paradigms. Previous meta‐analytic evidence indicates that administering the STAI questionnaire before or after stress induction does not affect the stress response (Goodman et al. 2017). The German version (Grimm 2009) applied in this study employs an 8‐point Likert scale (1 = “not at all” to 8 = “completely”) with four items reverse‐scored.

A 100‐mm visual analog scale (VAS100) was administered simultaneously with saliva sampling. The horizontal scale, which was labeled at the extremes with “relaxed” (minimum) and “stressed” (maximum) in German but did not contain intermediate ticks, was digitized after the session.

Participants were given a printed two‐week PA diary to fill out during their enrollment. The diary differentiated between activities of daily living, commuting, and deliberate exercise or sports training, and included time of day, duration, and subjective intensity on a 10‐point Likert scale. In addition, participants could note if they did the activity alone, with a partner/friend, or in a group. Only 36 participants handed in a filled activity diary. Therefore, we used recall data for PA only and report this analysis. The interpretation of this deviation from our planned analysis should be read with some caution. The long version of the International Physical Activity Questionnaire (IPAQ) (Craig 2009) was also administered on the day of the exercise test. It assesses PA over the preceding 7 days, from which we calculated energy expenditure as the sum of metabolic equivalent of task (MET‐minutes) and the total time spent in vigorous activity, according to the official scoring guidelines.

2.5.3. Fitness

A GXT was conducted on a separate occasion using a stationary cycle ergometer (Ergoline GmbH, Ergoselect 200, Germany) in a climatized laboratory. Respiratory parameters were measured using a metabolic gas analysis system (Cortex, METALYZER 3B, Germany). Heart rate was continuously recorded using a chest strap ECG (Polar H10, Polar, Finland). Participants were instructed to arrive well‐rested and to refrain from intense PA for 24 h preceding the test.

The test protocol consisted of a 2‐minute rest period seated on the ergometer, followed by a 2‐minute warm‐up phase with unloaded pedaling. During the main phase, resistance was increased every 2 min until exhaustion. The increase per step was determined individually based on sex, age, body mass, and height, following the recommendations by Primus et al. (2022). A 2‐minute recovery phase concluded the session.

The test was terminated at volitional exhaustion or if participants could not maintain a minimal pedaling cadence of 60 RPM. Participants were instructed to remain seated and to refrain from speaking during the test protocol.

Following the test, breath‐by‐breath data were exported and linearly interpolated to 1‐s intervals. A 30‐s running mean was applied for smoothing. V̇O2peak was defined as the highest value of the smoothed oxygen uptake. The thresholds (VT1 and VT2) were determined visually by two raters (PR and AV) using breakpoints in the parameters V̇E, V̇E/V̇CO2, V̇E/V̇O2, PetCO2, and PetO2. After independent analysis, consensus was reached on the five cases with the highest between‐rater difference. A detailed analysis of rater agreement is available in the supplemental files. To control for differences in body dimensions, V̇O2peak and oxygen uptake at VT1 and VT2 were adjusted for body mass.

2.6. Data Processing and Statistical Analyses

Overall hormonal responses during each experiment were quantified by calculating the area under the curve with respect to baseline (AUCi) (Pruessner et al. 2003). Similarly, overall reported subjective stress was quantified as the area under the curve with respect to zero (AUCg), which is more appropriate given its fixed lower limit.

For hormonal and VAS100 data, single missing time points were linearly interpolated; datasets with more than one missing value were excluded. Both hormonal and HRV metrics were natural log‐transformed to approximate normal distributions. Except for RMSSD, the presented estimators were transformed to improve interpretability. Data points exceeding three times the interquartile range were classified as extreme outliers.

Statistical analyses were performed in R (v.4.5.1). The significance level was set at p < 0.05. Hypothesis testing employed multilevel linear models with random intercepts, specifying a repeated‐measures structure for the stress test condition (level 1 variable: MAST vs. TSST) nested within participants (level 2 variables: sex and fitness). Models were fitted using the lme4 package (v.1.1.37). Metric predictors for fitness were grand‐mean centered. Model comparisons were based on changes in Akaike Information Criterion (AIC) and likelihood ratio (χ 2) tests under maximum likelihood estimation. Parameter estimates were obtained using restricted maximum likelihood, and effects were considered statistically significant when the bootstrapped 95% confidence intervals (95% CIs) did not include zero.

The time kinetics of the various stress markers were examined in a similar framework, with time points treated as the lowest level. Estimated marginal means were used for contrast testing following significant fixed effects. For multiple pairwise comparisons, p‐values were Bonferroni‐adjusted and reported alongside 95% CIs.

3. Results

3.1. Sample Characteristics

Of the 63 participants initially deemed eligible, four discontinued participation due to acute illness (n = 2), pregnancy (n = 1), or excessive discomfort during the protocol (n = 1). The final sample comprised 31 men and 28 women, with V̇O2peak values of 55.1 ± 8.1 mL/kg/min and 44.2 ± 5.7 mL/kg/min, respectively. Overall, 58% of participants self‐identified as competitive athletes—defined as those who participated in at least one official competition within the current year—and 37% trained or competed in a team setting. Sample characteristics are presented in Table 1.

TABLE 1.

Participant characteristics.

Women (n = 28) Men (n = 31) Total (n = 59)
Age (years) 26.3 ± 5.7 27.7 ± 5.3 27.1 ± 5.5
Body mass (kg) 61.8 ± 6.9 78.6 ± 7.8 70.65 ± 11.2
Height (cm) 166.9 ± 5.50 180.6 ± 5.9 174.1 ± 8.9
BMI (kg/m2) 22.1 ± 1.9 24.1 ± 2.1 23.2 ± 2.2
Physical activity (MET‐min/week) 3993 ± 2171 4343 ± 2665 4177 ± 2429
V̇O2peak (ml/kg/min) 44.2 ± 5.7 55.1 ± 8.1 50.4 ± 8.6
Team athletes (n) 10 (36%) 12 (39%) 22 (37%)
Paid work, ≥ 10 h/week (n) 22 (70%) 17 (60%) 39 (66%)
Student (n) 18 (58%) 21 (75%) 39 (66%)
Competitive athletes (n) 13 (46%) 21 (68%) 34 (58%)
DASS21 depression 1 [0, 4] 1 [0, 4] 1 [0, 4]
DASS21 anxiety 1 [0, 8] 1 [0, 10] 1 [0, 10]
DASS21 stress 4 [0, 15] 1 [0, 7] 2 [0, 15]
Open in a new tab

Note: Values are presented as mean ± standard deviation, median [minimum, maximum], or number (%).

3.2. Data Quality and Measurement Reliability

Twelve of the 720 saliva samples contained insufficient volume for analysis. Missing data points were interpolated in six cases, leaving six missing data points coming from three different cases. The quality of the heart rate period data was acceptable, with a mean rate of outlier intervals of 0.4% ± 0.9%. Three datasets with > 5% outlier heart periods due to a defective heart rate monitor were excluded. One GXT had to be repeated because the participant was insufficiently accustomed to the cycle ergometer during the initial trial, which led to implausible results.

To evaluate the sensitivity of the salivary hormone measures, signal‐to‐noise ratios (SNRs) were calculated separately for each analyte's concentration‐time curve. The SNR was defined as the difference between the mean maximal and baseline concentrations divided by the standard error of the mean, providing an index of how strongly the observed hormonal responses exceeded assay variability. The highest signal‐to‐noise ratio was observed for cortisol (SNR = 25.2), followed by VAS100 (SNR = 22.2), total glucocorticoids (SNR = 21.7), and cortisone (SNR = 19.3). Lower SNRs were found for the cortisone‐to‐cortisol ratio (cn2c; SNR = 11.2), and DHEAS (SNR = 5.2).

Internal consistency of the psychometric self‐report measures was evaluated using Cronbach's α. For the DASS‐21, internal consistency was good for the depression (α = 0.80, 95% CI [0.62, 0.86]) and stress subscales (α = 0.87, 95% CI [0.77, 0.92]), but low for anxiety (α = 0.39, 95% CI [0.10, 0.56]). Internal consistency of the STAI10 was acceptable (α = 0.76, 95% CI [0.65, 0.84]).

Two independent raters (PR and AV) determined the ventilatory thresholds. If the difference in relative oxygen uptake exceeded 5 mL/kg, raters discussed the case to reach a consensus. This was necessary in six cases. After this procedure intraclass correlation coefficients of ICC (2, 3) = 0.988 (95% CI [0.98, 0.993]) and ICC (2, 3) = 0.97 (95% CI [0.927, 0.985]) with standard errors of measurement of SEM = 0.26 mL/kg/min (95% CI [0.19, 0.41]) and SEM = 0.16 mL/kg/min (95% CI [0.13, 0.21]) were reached for VT1 and VT2, respectively. The mean of the two ratings was used as a predictor for the following analyses.

3.3. Verification of Stress‐Induction

Before testing the hypothesized influence of aerobic fitness and stressor type on stress response, we first verified that both stress paradigms elicited robust endocrine, autonomic, and subjective responses. Because of multicollinearity, the anticipation phase (time point 2) was excluded from all glucocorticoid models. The final models for time kinetics included cross‐level interactions and revealed statistically significant main effects of time and time × condition interaction for all key stress markers—cortisol, cortisone, total glucocorticoids, DHEAS, VAS100, STAI10, HR, and lnRMSSD—indicating effective stress‐induction across both tasks.

Contrary to our hypothesis, sex did not improve model fit for any outcomes, except for the cortisone‐to‐cortisol ratio (Cn:C) and HR. In the case of HR, the main effect of sex reached statistical significance, indicating overall higher HR in women than in men (ΔHR = 6.6 bpm, 95% CI [0.5, 12.8]). Figure 2 shows statistically significant effects between baseline and all other timepoints, as well as between the stress test conditions or sexes at each timepoint. Full parameter estimates for all time‐course models are presented in the Supporting Information.

FIGURE 2.

FIGURE 2

Open in a new tab

Time course of salivary hormones, heart rate, HRV, and subjective stress responses. Error bars indicate 95% confidence intervals (CIs). The Maastricht Acute Stress Test (MAST) and the Trier Social Stress Test (TSST) are presented separately in blue dots and red triangles, respectively. *: Statistically significant differences compared to baseline; #: Statistically significant differences between MAST and TSST; $: Differences between men and women; Cn:C: Cortisone‐to‐cortisol ratio; VAS100: 100 mm horizontal visual analogue scale for subjective stress; lnRMSSD: Natural log‐transformed root mean square of successive differences; The x‐axes (abscissa) indicate the time in minutes, with the gray area corresponding to the stress‐induction phase.

3.4. Influence of Aerobic Fitness and Stressor Type

Mean cortisol response (AUCi) was 76.1 ± 83.7 ng/mL·min. In the intercept‐only model, 23.5% of the variance in cortisol AUCi was attributable to person‐level variability. The model fit improved with the addition of the fixed‐effect predictor condition (ΔAIC = −3.16, χ 2(1) = 5.2, p = 0.023). No further improvements were observed when aerobic fitness indicators (V̇O 2 peak: ΔAIC = 3.5, χ 2(2) = 0.501, p = 0.778; VT1: ΔAIC = 3.8, χ 2(2) = 0.2, p = 0.905; VT2: ΔAIC = 2.81, χ 2(2) = 1.188, p = 0.552) were included. The final model revealed a variance attribution of ICC = 0.27 and a model fit of AIC = −251.7. Predicted geometric mean cortisol AUCi were 57.8 ng/mL·min (95% CI [25.1, 91.7]) and 87.8 ng/mL·min (95% CI [54.3, 122.0]) for MAST and TSST, respectively.

For subjective stress ratings (VAS100 AUCg), the intercept‐only model showed moderate between‐person variability (ICC = 0.275). Adding condition improved the model fit (ΔAIC = −19.25, χ 2(1) = 21.253, p < 0.001), with higher perceived stress during the TSST (predicted mean difference of AUCg: 278.2 mm·min; 95% CI [278.2, 734.9]). Aerobic fitness did not significantly account for additional variance (V̇O 2 peak: ΔAIC = 0.94, χ 2(2) = 3.065, p = 0.216; VT1: ΔAIC = 0.79, χ 2(2) = 3.208, p = 0.201; VT2: ΔAIC = 1.78, χ 2(2) = 2.222, p = 0.329). In the final model (AIC = 219.6), 43.1% of the variation was explained by the clustering structure of the data.

In the HR model, between‐subject variability was relatively largest (ICC = 0.357). Including condition improved model fit (ΔAIC = −1060.33, χ 2 (1) = 1062.328, p < 0.001), and further improvements occurred with V̇O 2 peak and the V̇O 2 peak × condition interaction (ΔAIC = −8.19, χ 2(2) = 12.189, p = 0.002). The mediating role of sex did not further improve the model (ΔAIC = 2.94, χ 2(4) = 5.061, p = 0.281). In the final model (AIC = −75, R 2 = 0.71, ICC = 0.588), the predicted geometric mean HR at the 25th percentile of V̇O2peak (42.6 mL/kg/min) was 78.7 bpm (95% CI [74.6, 83.1]) during the MAST and 91.8 bpm (95% CI [87.0, 96.9]) during the TSST. At the 75th percentile (55.1 mL/kg/min), the corresponding values were 71.6 bpm (95% CI [68.2, 75.1]) and 84.1 bpm (95% CI [80.1, 88.3]), respectively. This corresponded to a reduced HR of 0.99 bpm (95% CI [0.98, 1.00], p = 0.002) for each mL/kg/min change in V̇O2peak.

When HR reactivity was modeled as the change from baseline to the stress phase, only condition improved the fit (ΔAIC = −57.96, χ 2 (1) = 59.962, p < 0.001), whereas aerobic fitness did not (V̇O2peak: ΔAIC = 3.58, χ 2 (2) = 0.419, p = 0.811).

Similarly, the fit for lnRMSSD (ICC = 0.294) was improved by the inclusion of condition (ΔAIC = −22.22, χ 2(1) = 24.2, p < 0.001), whereas V̇O 2 peak did not add further explanatory power (ΔAIC = 3.93, χ 2(2) = 0.1, p = 0.967).

Figure 3 illustrates the observed data for cortisol, VAS100, absolute HR, and lnRMSSD, together with the predicted model including interaction between stressor condition and V̇O2peak.

FIGURE 3.

FIGURE 3

Open in a new tab

Predicted area under the curve for cortisol, visual analogue scale (VAS100), heart rate (HR) during stressor condition, and natural log‐transformed root mean square of successive difference (lnRMSSD) during stressor condition, by peak oxygen uptake relative to body mass (V̇O2peak). Error bars represent 95% confidence intervals (CI) of the fixed effect estimator. The blue dashed lines represent the models for the Maastricht Acute Stress Test (MAST), while the blue dots indicate the sample data. Conversely, the models for the Trier Social Stress Test (TSST) are depicted as red solid lines, with the sample data shown as red triangles.

3.5. Influence of Physical Activity and Stressor Type

Only 36 of 59 participants returned a PA diary for at least 7 days. Therefore, PA was estimated using the IPAQ. Total PA was expressed as MET‐minutes per week and calculated according to the official scoring manual (Craig et al. 2003). To reduce the absolute difference in predictor magnitudes, the mean‐centered values were divided by 1000 for model building only. Total PA did not increase model fit for cortisol (ΔAIC = −20.94, χ 2(2) = 1.4, p = 0.491), VAS (ΔAIC = 13.25, χ 2(2) = 0.9, p = 0.638), HR (ΔAIC = 18.1, χ 2(2) = 1.4, p = 0.4993), or RMSSD (ΔAIC = 13.0, χ 2(2) = 1.4, p = 0.498), compared to the basic model.

When only weekly hours of vigorous PA were considered, model fit improved significantly for cortisol (ΔAIC = 14.93, χ 2(2) = 8.6, p = 0.0134) and VAS (ΔAIC = 8.21, χ 2(2) = 7.1, p = 0.0287). Model improvements did not reach statistical significance for HR (ΔAIC = 18.6, χ 2(2) = 1.4, p = 0.4993) and RMSSD (ΔAIC = 15.2, χ 2(2) = 0.13, p = 0.939). Although adding vigorous PA improved overall model fit for cortisol, neither the main effect of PA (β = −0.004, 95% CI [−0.010, 0.003]) nor the condition × PA interaction (β = −0.006, 95% CI [−0.013, 0.002]) reached significance. Only VAS showed a significant condition × PA interaction (β = 0.071, 95% CI [0.018, 0.123]), indicating a positive association between vigorous PA and VAS responses during TSST compared to MAST. For the MAST condition, estimated marginal means AUCg VAS indicated higher values at the 25th percentile (1102, 95% CI [909, 1335]) than at the 75th percentile of vigorous PA (955, 95% CI [808, 1128]). During TSST, AUCg VAS was lower at the 25th percentile (1470, 95% CI [1213, 1781]) compared to the 75th percentile (1575, 95% CI [1336, 1857]). Both estimates were significantly different from MAST (p = 0.005 and p < 0.001, respectively). Observed datapoints and model predictions are presented in Figure 4. Models including sex as a covariate yielded substantively identical conclusions for the condition × PA interactions. The full analysis is available in the supplements.

FIGURE 4.

FIGURE 4

Open in a new tab

Predicted area under the curve for cortisol, visual analogue scale (VAS100), heart rate (HR) during stressor condition, and natural log‐transformed root mean square of successive difference (lnRMSSD) during stressor condition, by hours of vigorous physical activity per week, estimated from the International Physical Activity Questionnaire (IPAQ). Error bars represent 95% confidence intervals (CI) of the fixed effect estimator. The blue dashed lines represent the models for the Maastricht Acute Stress Test (MAST), while the blue dots indicate the sample data. Conversely, the models for the Trier Social Stress Test (TSST) are depicted as red solid lines, with the sample data shown as red triangles.

3.6. Submaximal Measures of Physical Fitness

To evaluate whether ventilatory thresholds provided a better fitness proxy for HR during stress, we replaced V̇O2peak with VT1 or VT2 in otherwise identical models. These substitutions did not improve explanatory power: marginal R2 values were 0.718 for the VT1 model and 0.710 for the VT2 model, compared with 0.710 for the V̇O2peak model. Accordingly, the model fits were not significantly different (VT1: ΔAIC = 5.65, χ 2(0) = 0, p > 0.999; VT2: ΔAIC = 8.85, χ 2(0) = 0, p > 0.999).

4. Discussion

4.1. Overall Stress Response

This study examined the physiological and subjective stress responses to two established laboratory‐based stress‐induction protocols, the TSST and MAST, and evaluated the modulating effect of aerobic fitness and PA in a mixed‐sex sample. Across the glucocorticoids, autonomic‐cardiac, and subjective domains, both tests elicited pronounced stress reactions, confirming the validity of the paradigms. The TSST induced stronger activation of the hypothalamic–pituitary–adrenal (HPA) axis and autonomic nervous system, reflected by higher glucocorticoid and DHEAS concentrations, elevated heart rate, reduced vagal cardiac modulation, and greater subjective stress compared with the MAST.

Differences in task posture and engagement likely contributed to these effects. The TSST requires participants to stand while they actively engage in public speaking. This may introduce mild orthostatic and motor demands, which can amplify autonomic‐associated cardiac activity (Nater et al. 2013). The MAST, in contrast, is performed while seated and involves cold exposure, which may provoke localized vasoconstriction and transient autonomic adjustments. However, evidence is inconclusive as to whether partial body cold exposure has a measurable effect on HRV (Jdidi et al. 2024). Taken together, these results reaffirm the TSST as a strong, reliable inducer of psychophysiological stress, while the MAST represents a less socially loaded, more physically loaded stressor.

It is noteworthy that our finding of divergent responses to the two stress paradigms updates results reported by Smeets et al. (Smeets et al. 2012), who compared the TSST and MAST in a sample of 20 male undergraduate students in a crossover design and found no differences in salivary cortisol, α‐amylase (often used as a marker of sympathetic activity (Nater and Rohleder 2009)), or subjective stress. Several methodological and analytical factors likely explain this divergence. First, constructs and instruments differed: Smeets et al. used the Positive and Negative Affect Schedule (PANAS) at two time points, whereas we assessed state anxiety (STAI10) and perceived stress (VAS100). The PANAS (Watson et al. 1988) measures affect on two orthogonal dimensions. In contrast, the STAI10 captures anxiety, which is neurophysiologically and phenomenologically highly related to stress (Daviu et al. 2019). The VAS100 was used to report the level of overall stress. Compared to affective measures, stress and anxiety may be regarded as higher‐order constructs that integrate affect with other components like arousal (Barrett and Russell 1999). Second, the higher temporal resolution of our sampling (six vs. two time points) enabled time‐course modeling and AUC‐based summaries, increasing sensitivity to between‐task differences.

Further, the autonomic indices comprised different physiological mechanisms. Smeets et al. measured blood pressure as an indicator of sympathetic activation (Smeets et al. 2012). In contrast, we assessed HR as a measure of overall cardiac autonomic load and RMSSD, which is interpreted as a measure of cardiac vagal modulation (Quigley et al. 2024). The stronger effects of the TSST in our study may therefore reflect the different autonomic pathways probed. Finally, sample size and composition may have played a role. Our sample was larger (n = 59 vs. n = 20) and included both sexes. The greater statistical power and biological heterogeneity may have revealed effects that were undetectable in the smaller, all‐male cohort of Smeets et al. (Smeets et al. 2012).

The inclusion of male and female participants represents a key strength of the present study. Research on PA and fitness on stress has traditionally relied predominantly on male samples (Sothmann 2006; Forcier et al. 2006), despite accumulating evidence that biological sex and sex hormones modulate HPA‐axis and autonomic regulation (Jackson and Dishman 2006; Mücke et al. 2018). Meta‐analytic evidence indicates that men typically show higher cortisol responses than women (Liu et al. 2017; Gu et al. 2022). However, during the luteal phase—marked by elevated levels of estrogen and progesterone—women's cortisol responses to the TSST have been shown to approximate those of men (Liu et al. 2017). In our analysis, sex‐related differences were generally small. Across both stress‐induction protocols, the only significant difference in endocrine responses emerged for Cn:C shortly after the stressor.

Besides the direct effects of estrogen on the HPA axis, it also increases corticoid‐binding globulin, which binds free, bioactive cortisol, measured in saliva (Handa and Weiser 2014). The faster time kinetics, combined with the lower overall cortisol response in men (even if not significant in our study), may have led to a brief drop in Cn:C before excess cortisol would be enzymatically converted to cortisone. This conversion could be further delayed by the higher saliva flow in men (Rutherfurd‐Markwick et al. 2017), reducing the contact time with 11β‐HSD, the main enzyme in this process. Regarding autonomic regulation, women exhibited higher mean heart rates than men, independent of time point or stressor. This finding aligns with well‐established sex differences in cardiac structure and function, including smaller heart size and stroke volume in women (Prabhavathi et al. 2014). No further sex‐related differences were observed for endocrine, autonomic, or subjective outcomes, supporting the view that, under controlled hormonal conditions, acute stress responses are broadly comparable between sexes.

4.2. Stress‐Buffering Effects

With a mean V̇O2peak of 50.4 ± 8.6 mL/kg/min, our participants demonstrated above‐average aerobic fitness. Across all measured domains, stress responses to both tasks were largely independent of aerobic capacity. Cortisol, RMSSD, and subjective stress (VAS100) did not vary as a function of fitness. The only consistent association emerged for absolute HR during stress‐induction, which was inversely related to V̇O2peak. This pattern aligns with established cardiovascular adaptations to endurance training—such as increased cardiac dimensions and stroke volume (Hellsten and Nyberg 2015)—that enable trained individuals to maintain lower heart rates at a given submaximal load. HR reactivity—the change of HR from baseline to during the stressor—did not vary significantly with fitness. Thus, while fitter participants exhibited a comparable relative stress response, their absolute cardiac load may be lower.

The expected interaction between fitness and stressor type—a stronger stress‐buffering effect during the MAST—did not reach statistical significance, although descriptive trends in HR and VAS100 were in the hypothesized direction (Figure 3). It is conceivable that the physiological specificity of training adaptations limits their transferability to psychosocial stressors such as the TSST, whereas stressors with a physical component, like the MAST, may share greater overlap with the stimuli encountered during exercise training. Overall, however, the small and non‐significant effects suggest that, under acute laboratory conditions, aerobic fitness may exert only a modest influence on HPA axis or autonomic activation.

Due to low adherence to the diary, PA was estimated solely from recall. Therefore, the analysis was interpreted cautiously. While overall weekly MET‐minutes did not improve model fit, vigorous PA did. However, the estimated effects on hormonal and autonomic outcomes were either very small or did not reach statistical significance, and the overall explained variance of the models was lower compared to objective measures of fitness. Our data suggest higher subjective stress during the TSST among people engaging in more vigorous physical activity than during the MAST. This was surprising and warrants further investigation. A negative trend between vigorous PA and subjective stress during MAST could be explained by athletes' higher pain tolerance (Tesarz et al. 2012). The opposite effect during the TSST, however, is contrary to our hypothesis and warrants further investigation. As the TSST overall induced greater psychophysiological stress, one explanation could be greater interoceptive awareness among people regularly engaging in high‐intensity training.

A possible effect of higher‐intensity PA relative to lower‐intensity PA would be consistent with the underlying assumption of the cross‐stressor adaptation hypothesis, which states that PA and exercise may train the stress‐moderating systems through repeated exposure to physical demands. However, only higher intensity activities challenge certain physiological systems and lead to increases in cortisol and catecholamines (Athanasiou et al. 2023; Hackney 2006). Another study by Gerber et al. (Gerber et al. 2017) also found evidence for a stress‐buffering effect of higher‐intensity activity over the recommended amount of moderate activity, using accelerometer‐based measures of PA. Also, the systematic review by Mücke et al. (Mücke et al. 2018) indicated a moderating effect of intensity.

In an exploratory extension, we examined ventilatory thresholds (VT1 and VT2) as alternative fitness indices. These did not improve model fit for any stress parameter. Variance explained by the included fixed effects was similar to the models including V̇O2peak. Therefore, the use of submaximal thresholds may be sufficient in future work to determine fitness without requiring maximal effort in less‐trained or clinical populations.

In summary, aerobic capacity was only associated with lower absolute HR during stress exposure, but did not attenuate endocrine, autonomic, or subjective stress responses. This finding aligns with previous experiments. Rimmele et al. observed that while elite athletes showed lower cortisol and HR responses, amateurs showed only blunted HR responses compared to a third, non‐exerciser group. This was interpreted as a stronger stress‐buffering effect of the ANS‐related compared to the HPA‐related outcomes (Rimmele et al. 2009). However, the reported effects of fitness on HR are inconsistent (Mücke et al. 2018). A more pronounced stress‐buffering effect for ANS‐related metrics would, however, align with literature on the acute hormonal effects of exercise. Trained individuals show a more pronounced elevation of adrenaline and alpha‐amylase compared to non‐trained individuals, whereas the same effect is either smaller or absent for cortisol (Zouhal et al. 2008; Athanasiou et al. 2023; Kunz et al. 2015). These findings suggest that under controlled laboratory conditions, the stress‐buffering effect of fitness is modest. Endurance training elicits the strongest adaptive effects on cardiovascular function. Our data suggest that this system‐specific training effect also shows the most promising protective effect against psychosocial demands. Under this assumption, future studies may investigate the effect of other physical capacities and physiological adaptations to exercise training on psychological stress.

4.3. Strengths and Limitations

This study was the first to examine the relationship between fitness and stress responses across two different stressor paradigms based on a within‐subject design. The strength of our experiment lies in the collection and analysis of multiple markers of stress and the application of state‐of‐the‐art methods for hormone quantification and cardiac autonomic analysis. Both stress‐induction protocols were standardized and implemented according to best‐practice guidelines, and physical fitness was determined by an incremental GXT, the current gold standard.

Furthermore, both biological sexes were included, and hormonal influences were partially controlled by scheduling women during the luteal phase and excluding those using hormonal contraception. Contrary to our hypothesis, we did not find extended sex effects, nor a sex‐dependent stress‐buffering effect of fitness, which is probably due to low statistical power for these effects. Previous research has shown the influence of intraindividual sex hormone fluctuations on the acute stress response, with estrogen stimulating cortisol‐binding globulin and simultaneously increasing adrenal sensitivity to ACTH, both of which contribute to the accumulation of saliva cortisol under stress in opposite directions (Kajantie and Phillips 2006). Combined with the extensive expression of estrogen receptors (types α and β) in stress‐related regions of the brain (Kajantie and Phillips 2006), this complex interplay typically results in lower cortisol response in women than in men and higher free cortisol levels during the luteal phase than during the follicular phase in eumenorrheic women (Liu et al. 2017; Kudielka and Kirschbaum 2005). However, the menstrual phase was self‐reported only. To better account for intraindividual hormone variation, future studies should incorporate repeated ovulation testing (Schmalenberger et al. 2021; Sims and Heather 2018) and statistically control for menstrual phase and contraceptive use—an approach that enhances ecological validity but requires larger samples.

We found lower absolute HR during stress in fitter participants, suggesting lower cardiac load with aerobic training under other types of demand. However, it remains unclear whether this finding extends to the protective effect of exercise on the cardiovascular system during stress. To further understand cardiovascular stress, future studies should monitor stroke volume. Laboratory stress‐induction protocols would allow the implementation of transthoracic bioimpedance (Harford et al. 2019) or modeling the arterial pulse waveform (Bogert and van Lieshout 2005) for an accurate estimate. To our understanding, this has not been investigated, but it could improve our understanding of cardiac demand during non‐exercise stress.

With a V̇O2peak of 50.4 ± 8.6 mL/kg/min, the fitness level of our sample can be considered higher than average. Although the study aimed to sample a broader population, volunteers were fitter and more active than average (Holler et al. 2026). This may have limited our ability to find effects. Other research identified evidence for the stress‐buffering effect of exercise only when contrasting groups with low and high levels of aerobic fitness (Rimmele et al. 2009, 2007). Therefore, our overall null findings may be partly explained by the relatively homogeneous nature of our sample with respect to aerobic capacity.

To reduce the complexity of the statistical models and control for variation in time‐kinetics, we used the area under the curve with respect to the baseline (AUCi) for glucocorticoids. Alternative approaches—such as the AUC with respect to zero (AUCg) or change scores—capture slightly different constructs. While AUCi is more related to changes from baseline and the sensitivity of the system, AUCg represents total hormone levels (Pruessner et al. 2003).

Baseline cortisol appears unrelated to exercise volume (Cevada et al. 2014), although regular endurance training is related to higher chronic cortisol exposure (Skoluda et al. 2012), complicating the interpretation of absolute concentrations. For these reasons, we decided to report AUCi but acknowledge the value of AUCg for estimating total allostatic load and mitigating baseline measurement error (Pruessner et al. 2003).

Unfortunately, we had low compliance with the PA diary, so we reported only an analysis based on IPAQ data. To further differentiate physical fitness and activity, the implementation of extended accelerometer‐based measures is highly recommended (Sylvia et al. 2014). Future studies could explore additional aspects of activity and exercise (e.g., enjoyment, mood, environment, or social engagement), as the potential buffering effect of exercise may partly stem from psychological and social factors.

Finally, the follow‐up period after stress cessation was limited to 30 min, and full hormonal recovery was not reached within this timeframe. Extending the recovery window, as recommended for the TSST (Labuschagne et al. 2019), would allow a more precise characterization of post‐stress regulation. It remains plausible that higher aerobic fitness accelerates HPA axis feedback and recovery kinetics even in the absence of lower peak responses (Mücke et al. 2018).

5. Perspective

The positive effect of exercise on stress‐related outcomes surpasses the acute response investigated in this study. The potentially negative health consequences of chronic stress include tissue catabolism, for example, in the hippocampus and other brain areas with a high density of glucocorticoid receptors (BS et al. 2016), and skeletal muscle (Kraemer et al. 2020). Also, stress modulates inflammatory processes and reduces cardiac vagal regulation, contributing to cardiovascular health (Wirtz and von Känel 2017). Indirectly, these effects can be exacerbated by behavioral influences of stress, such as changes in diet and increased food intake (Hill et al. 2022). PA and exercise, on the other hand, have repeatedly been shown to positively influence all these outcomes via pathways that are unrelated to stress (Martín‐Rodríguez et al. 2024; Liu et al. 2024).

Taken together, this study demonstrated a robust stress response to the MAST and TSST spanning multiple interconnected psychophysiological systems. Our data indicate that higher levels of vigorous PA are associated with stressor‐dependent differences in subjective stress experience. We did not find evidence for stress‐buffering effects of aerobic fitness on measures of HPA axis activity, cardiac vagal control, and subjective stress. Nevertheless, as with physical demands, people with higher aerobic capacity require a lower heart rate to adequately respond to the same stressor. This sparing effect on the cardiovascular system under stress may be a key protective factor against everyday life stress.

Author Contributions

Barbara Wessner: methodology, conceptualization, writing – review and editing, supervision, resources. Peter Raidl: conceptualization, investigation, writing – original draft, methodology, visualization, writing – review and editing, formal analysis, data curation, project administration, validation, software. Robert Csapo: conceptualization, methodology, resources, supervision, writing – review and editing. Anja Vogl: investigation, writing – review and editing, validation. Urs M. Nater: conceptualization, writing – review and editing. Aljoscha Dreisörner: conceptualization, writing – review and editing, methodology.

Funding

This study was supported by the SOLE Research platform of the University of Vienna.

Ethics Statement

The Ethics Committee of the University of Vienna approved the study. Reference Number: 00960.

Consent

All participants provided informed consent in accordance with the terms of the ethics application.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: psyp70350‐sup‐0001‐Supinfo.pdf.

PSYP-63-e70350-s001.pdf (907.7KB, pdf)

Acknowledgments

We thank Tom Smeets for the original material and protocol of the MAST. We further thank Jonas Gary and Nadine Kunt for their assistance during data collection. Open Access funding provided by Universitat Wien.

Data Availability Statement

All data and analysis code are made available on OSF: https://osf.io/ex2t4/overview?view_only=6a67433550424110b82d24dc0613b5d9.

References

  1. Allen, A. P. , Kennedy P. J., Cryan J. F., Dinan T. G., and Clarke G.. 2014. “Biological and Psychological Markers of Stress in Humans: Focus on the Trier Social Stress Test.” Neuroscience and Biobehavioral Reviews 38: 94–124. 10.1016/j.neubiorev.2013.11.005. [DOI] [PubMed] [Google Scholar]
  2. Athanasiou, N. , Bogdanis G. C., and Mastorakos G.. 2023. “Endocrine Responses of the Stress System to Different Types of Exercise.” Reviews in Endocrine & Metabolic Disorders 24, no. 2: 251–266.  10.1007/s11154-022-09758-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barrett, L. F. , and Russell J. A.. 1999. “The Structure of Current Affect: Controversies and Emerging Consensus.” Current Directions in Psychological Science 8, no. 1: 10–14. 10.1111/1467-8721.00003. [DOI] [Google Scholar]
  4. Bogert, L. W. J. , and van Lieshout J. J.. 2005. “Non‐Invasive Pulsatile Arterial Pressure and Stroke Volume Changes From the Human Finger.” Experimental Physiology 90, no. 4: 437–446. 10.1113/expphysiol.2005.030262. [DOI] [PubMed] [Google Scholar]
  5. BS, M. E. , Nasca C., and Gray J. D.. 2016. “Stress Effects on Neuronal Structure: Hippocampus, Amygdala, and Prefrontal Cortex.” Neuropsychopharmacology 41, no. 1: 3–23. 10.1038/npp.2015.171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Caspersen, C. J. , Powell K. E., and Christenson G. M.. 1985. “Physical Activity, Exercise, and Physical Fitness: Definitions and Distinctions for Health‐Related Research.” Public Health Reports 100, no. 2: 126–131. [PMC free article] [PubMed] [Google Scholar]
  7. Cevada, T. , Vasques P. E., Moraes H., and Deslandes A.. 2014. “Salivary Cortisol Levels in Athletes and Nonathletes: A Systematic Review.” Hormone and Metabolic Research 46, no. 13: 905–910. 10.1055/s-0034-1387797. [DOI] [PubMed] [Google Scholar]
  8. Cohen, S. , Gianaros P. J., and Manuck S. B.. 2016. “A Stage Model of Stress and Disease.” Perspectives on Psychological Science 11, no. 4: 456–463. 10.1177/1745691616646305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Craig, A. B. 2009. “How Do You Feel—Now? The Anterior Insula and Human Awareness.” Nature Reviews Neuroscience 10, no. 1: 59–70. 10.1038/nrn2555. [DOI] [PubMed] [Google Scholar]
  10. Craig, C. L. , Marshall A. L., Sjöström M., et al. 2003. “International Physical Activity Questionnaire: 12‐Country Reliability and Validity.” Medicine and Science in Sports and Exercise 35, no. 8: 1381–1395. 10.1249/01.MSS.0000078924.61453.FB. [DOI] [PubMed] [Google Scholar]
  11. Daviu, N. , Bruchas M. R., Moghaddam B., Sandi C., and Beyeler A.. 2019. “Neurobiological Links Between Stress and Anxiety.” Neurobiology of Stress 11: 100191. 10.1016/j.ynstr.2019.100191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Deuster, P. A. , Chrousos G. P., Luger A., et al. 1989. “Hormonal and Metabolic Responses of Untrained, Moderately Trained, and Highly Trained Men to Three Exercise Intensities.” Metabolism 38, no. 2: 141–148. 10.1016/0026-0495(89)90253-9. [DOI] [PubMed] [Google Scholar]
  13. Elliott‐Sale, K. J. , Minahan C. L., de Jonge X. A. K. J., et al. 2021. “Methodological Considerations for Studies in Sport and Exercise Science With Women as Participants: A Working Guide for Standards of Practice for Research on Women.” Sports Medicine 51, no. 5: 843–861. 10.1007/s40279-021-01435-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Forcier, K. , Stroud L. R., Papandonatos G. D., et al. 2006. “Links Between Physical Fitness and Cardiovascular Reactivity and Recovery to Psychological Stressors: A Meta‐Analysis.” Health Psychology 25: 723–739. 10.1037/0278-6133.25.6.723. [DOI] [PubMed] [Google Scholar]
  15. Gerber, M. , Ludyga S., Mücke M., Colledge F., Brand S., and Pühse U.. 2017. “Low Vigorous Physical Activity Is Associated With Increased Adrenocortical Reactivity to Psychosocial Stress in Students With High Stress Perceptions.” Psychoneuroendocrinology 80: 104–113. 10.1016/j.psyneuen.2017.03.004. [DOI] [PubMed] [Google Scholar]
  16. Gerber, M. , and Pühse U.. 2009. “Review Article: Do Exercise and Fitness Protect Against Stress‐Induced Health Complaints? A Review of the Literature.” Scandinavian Journal of Public Health 37, no. 8: 801–819. 10.1177/1403494809350522. [DOI] [PubMed] [Google Scholar]
  17. Goodman, W. K. , Janson J., and Wolf J. M.. 2017. “Meta‐Analytical Assessment of the Effects of Protocol Variations on Cortisol Responses to the Trier Social Stress Test.” Psychoneuroendocrinology 80: 26–35. 10.1016/j.psyneuen.2017.02.030. [DOI] [PubMed] [Google Scholar]
  18. Grimm, J. 2009. “STAI‐Test: State‐Trait‐Anxiety Inventory (Deutsche Version).” https://empcom.univie.ac.at/fileadmin/user_upload/p_empcom/pdfs/Grimm2009_StateTraitAngst_MFWorkPaper2009‐02.pdf.
  19. Gu, H. , Ma X., Zhao J., and Liu C.. 2022. “A Meta‐Analysis of Salivary Cortisol Responses in the Trier Social Stress Test to Evaluate the Effects of Speech Topics, Sex, and Sample Size.” Comprehensive Psychoneuroendocrinology 10: 100125. 10.1016/j.cpnec.2022.100125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hackney, A. C. 2006. “Stress and the Neuroendocrine System: The Role of Exercise as a Stressor and Modifier of Stress.” Expert Review of Endocrinology and Metabolism 1, no. 6: 783–792. 10.1586/17446651.1.6.783. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Handa, R. J. , and Weiser M. J.. 2014. “Gonadal Steroid Hormones and the Hypothalamo‐Pituitary‐Adrenal Axis.” Frontiers in Neuroendocrinology 35, no. 2: 197–220. 10.1016/j.yfrne.2013.11.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Harford, M. , Clark S. H., Smythe J. F., et al. 2019. “Non‐Invasive Stroke Volume Estimation by Transthoracic Electrical Bioimpedance Versus Doppler Echocardiography in Healthy Volunteers.” Journal of Medical Engineering & Technology 43, no. 1: 33–37. 10.1080/03091902.2019.1599074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hellsten, Y. , and Nyberg M.. 2015. “Cardiovascular Adaptations to Exercise Training.” Comprehensive Physiology 6, no. 1: 1–32. 10.1002/cphy.c140080. [DOI] [PubMed] [Google Scholar]
  24. Hermann, R. , Biallas B., Predel H. G., and Petrowski K.. 2019. “Physical Versus Psychosocial Stress: Effects on Hormonal, Autonomic, and Psychological Parameters in Healthy Young Men.” Stress 22, no. 1: 103–112. 10.1080/10253890.2018.1514384. [DOI] [PubMed] [Google Scholar]
  25. Hill, D. , Conner M., Clancy F., et al. 2022. “Stress and Eating Behaviours in Healthy Adults: A Systematic Review and Meta‐Analysis.” Health Psychology Review 16, no. 2: 280–304. 10.1080/17437199.2021.1923406. [DOI] [PubMed] [Google Scholar]
  26. Holler, P. , Dohr S., Tuttner S., Amort F. M., van Poppel M., and Jaunig J.. 2026. “Associations Between Perceived Physical Literacy and Physical Fitness Tests Among the Adult Population.” Frontiers in Sports and Active Living 8: 1783118. 10.3389/fspor.2026.1783118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Huang, M. , Yoo J. K., Stickford A. S. L., et al. 2021. “Early Sympathetic Neural Responses During a Cold Pressor Test Linked to Pain Perception.” Clinical Autonomic Research 31, no. 2: 215–224. 10.1007/s10286-019-00635-7. [DOI] [PubMed] [Google Scholar]
  28. Ifuku, H. , Moriyama K., Arai K., and Shiraishi‐Hichiwa Y.. 2007. “Regulation of Cardiac Function During a Cold Pressor Test in Athletes and Untrained Subjects.” European Journal of Applied Physiology 101, no. 1: 75–79. 10.1007/s00421-007-0475-y. [DOI] [PubMed] [Google Scholar]
  29. Jackson, E. M. , and Dishman R. K.. 2006. “Cardiorespiratory Fitness and Laboratory Stress: A Meta‐Regression Analysis.” Psychophysiology 43, no. 1: 57–72. 10.1111/j.1469-8986.2006.00373.x. [DOI] [PubMed] [Google Scholar]
  30. Jdidi, H. , Dugué B., de Bisschop C., Dupuy O., and Douzi W.. 2024. “The Effects of Cold Exposure (Cold Water Immersion, Whole‐ and Partial‐ Body Cryostimulation) on Cardiovascular and Cardiac Autonomic Control Responses in Healthy Individuals: A Systematic Review, Meta‐Analysis and Meta‐Regression.” Journal of Thermal Biology 121: 103857. 10.1016/j.jtherbio.2024.103857. [DOI] [PubMed] [Google Scholar]
  31. Kajantie, E. , and Phillips D. I. W.. 2006. “The Effects of Sex and Hormonal Status on the Physiological Response to Acute Psychosocial Stress.” Psychoneuroendocrinology 31, no. 2: 151–178. 10.1016/j.psyneuen.2005.07.002. [DOI] [PubMed] [Google Scholar]
  32. Kivimäki, M. , Bartolomucci A., and Kawachi I.. 2023. “The Multiple Roles of Life Stress in Metabolic Disorders.” Nature Reviews. Endocrinology 19, no. 1: 10–27. 10.1038/s41574-022-00746-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kraemer, W. J. , Ratamess N. A., Hymer W. C., Nindl B. C., and Fragala M. S.. 2020. “Growth Hormone(s), Testosterone, Insulin‐Like Growth Factors, and Cortisol: Roles and Integration for Cellular Development and Growth With Exercise.” Frontiers in Endocrinology 11: 33. 10.3389/fendo.2020.00033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kudielka, B. M. , and Kirschbaum C.. 2005. “Sex Differences in HPA Axis Responses to Stress: A Review.” Biological Psychology 69, no. 1: 113–132. 10.1016/j.biopsycho.2004.11.009. [DOI] [PubMed] [Google Scholar]
  35. Kunz, H. , Bishop N. C., Spielmann G., et al. 2015. “Fitness Level Impacts Salivary Antimicrobial Protein Responses to a Single Bout of Cycling Exercise.” European Journal of Applied Physiology 115, no. 5: 1015–1027. 10.1007/s00421-014-3082-8. [DOI] [PubMed] [Google Scholar]
  36. Labuschagne, I. , Grace C., Rendell P., Terrett G., and Heinrichs M.. 2019. “An Introductory Guide to Conducting the Trier Social Stress Test.” Neuroscience and Biobehavioral Reviews 107: 686–695. 10.1016/j.neubiorev.2019.09.032. [DOI] [PubMed] [Google Scholar]
  37. Lester, G. R. , Abiusi F. S., Bodner M. E., Mittermaier P. M., and Cote A. T.. 2021. “The Impact of Fitness Status on Vascular and Baroreceptor Function in Healthy Women and Men.” Journal of Vascular Research 59, no. 1: 16–23. 10.1159/000518985. [DOI] [PubMed] [Google Scholar]
  38. Liu, J. J. W. , Ein N., Peck K., Huang V., Pruessner J. C., and Vickers K.. 2017. “Sex Differences in Salivary Cortisol Reactivity to the Trier Social Stress Test (TSST): A Meta‐Analysis.” Psychoneuroendocrinology 82: 26–37. 10.1016/j.psyneuen.2017.04.007. [DOI] [PubMed] [Google Scholar]
  39. Liu, M. , Hui M. D., Lu Z. Y., Kang N., Min W. D., and Chen G.. 2024. “Meta‐Analysis of Exercise Intervention on Health Behaviors in Middle‐Aged and Older Adults.” Frontiers in Psychology 14: 1308602. 10.3389/fpsyg.2023.1308602. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Lopresti, A. L. , Smith S. J., and Drummond P. D.. 2022. “Modulation of the Hypothalamic‐Pituitary‐Adrenal (HPA) Axis by Plants and Phytonutrients: A Systematic Review of Human Trials.” Nutritional Neuroscience 25, no. 8: 1704–1730. 10.1080/1028415X.2021.1892253. [DOI] [PubMed] [Google Scholar]
  41. Martín‐Rodríguez, A. , Gostian‐Ropotin L. A., Beltrán‐Velasco A. I., et al. 2024. “Sporting Mind: The Interplay of Physical Activity and Psychological Health.” Sports 12, no. 1: 37. 10.3390/sports12010037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Mason, J. W. 1971. “A Re‐Evaluation of the Concept of “Non‐Specificity” in Stress Theory.” Journal of Psychiatric Research 8, no. 3–4: 323–333. 10.1016/0022-3956(71)90028-8. [DOI] [PubMed] [Google Scholar]
  43. Mücke, M. , Ludyga S., Colledge F., and Gerber M.. 2018. “Influence of Regular Physical Activity and Fitness on Stress Reactivity as Measured With the Trier Social Stress Test Protocol: A Systematic Review.” Sports Medicine 48, no. 11: 2607–2622. 10.1007/s40279-018-0979-0. [DOI] [PubMed] [Google Scholar]
  44. Nater, U. M. , Ditzen B., Strahler J., and Ehlert U.. 2013. “Effects of Orthostasis on Endocrine Responses to Psychosocial Stress.” International Journal of Psychophysiology 90, no. 3: 341–346. 10.1016/j.ijpsycho.2013.10.010. [DOI] [PubMed] [Google Scholar]
  45. Nater, U. M. , and Rohleder N.. 2009. “Salivary Alpha‐Amylase as a Non‐Invasive Biomarker for the Sympathetic Nervous System: Current State of Research.” Psychoneuroendocrinology 34, no. 4: 486–496. 10.1016/j.psyneuen.2009.01.014. [DOI] [PubMed] [Google Scholar]
  46. Nowacka‐Chmielewska, M. , Grabowska K., Grabowski M., Meybohm P., Burek M., and Małecki A.. 2022. “Running From Stress: Neurobiological Mechanisms of Exercise‐Induced Stress Resilience.” International Journal of Molecular Sciences 23, no. 21: 13348. 10.3390/ijms232113348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Prabhavathi, K. , Selvi K. T., Poornima K. N., and Sarvanan A.. 2014. “Role of Biological Sex in Normal Cardiac Function and in Its Disease Outcome—A Review.” Journal of Clinical and Diagnostic Research 8, no. 8: BE01–BE04. 10.7860/JCDR/2014/9635.4771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Primus, C. , Wonisch M., Berent R., and Auer J.. 2022. “Praxisleitlinien Ergometrie Und Spiroergometrie//Practice Guidelines for Exercise Testing.” Journal Für Kardiologie 29, no. 1–2: 17–26. [Google Scholar]
  49. Pruessner, J. C. , Kirschbaum C., Meinlschmid G., and Hellhammer D. H.. 2003. “Two Formulas for Computation of the Area Under the Curve Represent Measures of Total Hormone Concentration Versus Time‐Dependent Change.” Psychoneuroendocrinology 28, no. 7: 916–931. 10.1016/S0306-4530(02)00108-7. [DOI] [PubMed] [Google Scholar]
  50. Quigley, K. , Gianaros P., Norman G., Jennings J., and Geus E.. 2024. “Publication Guidelines for Human Heart Rate and Heart Rate Variability Studies in Psychophysiology‐Part 1: Physiological Underpinnings and Foundations of Measurement.” Psychophysiology 61, no. 9: e14604. 10.1111/psyp.14604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Rimmele, U. , Seiler R., Marti B., Wirtz P. H., Ehlert U., and Heinrichs M.. 2009. “The Level of Physical Activity Affects Adrenal and Cardiovascular Reactivity to Psychosocial Stress.” Psychoneuroendocrinology 34, no. 2: 190–198. 10.1016/j.psyneuen.2008.08.023. [DOI] [PubMed] [Google Scholar]
  52. Rimmele, U. , Zellweger B. C., Marti B., et al. 2007. “Trained Men Show Lower Cortisol, Heart Rate and Psychological Responses to Psychosocial Stress Compared With Untrained Men.” Psychoneuroendocrinology 32, no. 6: 627–635. 10.1016/j.psyneuen.2007.04.005. [DOI] [PubMed] [Google Scholar]
  53. Rutherfurd‐Markwick, K. , Starck C., Dulson D. K., and Ali A.. 2017. “Salivary Diagnostic Markers in Males and Females During Rest and Exercise.” Journal of the International Society of Sports Nutrition 14, no. 1: 27. 10.1186/s12970-017-0185-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Schmalenberger, K. M. , Tauseef H. A., Barone J. C., et al. 2021. “How to Study the Menstrual Cycle: Practical Tools and Recommendations.” Psychoneuroendocrinology 123: 104895. 10.1016/j.psyneuen.2020.104895. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Sims, S. T. , and Heather A. K.. 2018. “Myths and Methodologies: Reducing Scientific Design Ambiguity in Studies Comparing Sexes and/or Menstrual Cycle Phases.” Experimental Physiology 103, no. 10: 1309–1317. 10.1113/EP086797. [DOI] [PubMed] [Google Scholar]
  56. Singh, B. , Olds T., Curtis R., et al. 2023. “Effectiveness of Physical Activity Interventions for Improving Depression, Anxiety and Distress: An Overview of Systematic Reviews.” British Journal of Sports Medicine 57, no. 18: 1203–1209. 10.1136/bjsports-2022-106195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Skoluda, N. , Dettenborn L., Stalder T., and Kirschbaum C.. 2012. “Elevated Hair Cortisol Concentrations in Endurance Athletes.” Psychoneuroendocrinology 37, no. 5: 611–617. 10.1016/j.psyneuen.2011.09.001. [DOI] [PubMed] [Google Scholar]
  58. Smeets, T. , Cornelisse S., Quaedflieg C. W. E. M., Meyer T., Jelicic M., and Merckelbach H.. 2012. “Introducing the Maastricht Acute Stress Test (MAST): A Quick and Non‐Invasive Approach to Elicit Robust Autonomic and Glucocorticoid Stress Responses.” Psychoneuroendocrinology 37, no. 12: 1998–2008. 10.1016/j.psyneuen.2012.04.012. [DOI] [PubMed] [Google Scholar]
  59. Sothmann, M. S. 2006. “The Cross‐Stressor Adaptation Hypothesis and Exercise Training.” In Psychobiology of Physical Activity, 149–160. Human Kinetics. [Google Scholar]
  60. Sylvia, L. G. , Bernstein E. E., Hubbard J. L., Keating L., and Anderson E. J.. 2014. “A Practical Guide to Measuring Physical Activity.” Journal of the Academy of Nutrition and Dietetics 114, no. 2: 199–208. 10.1016/j.jand.2013.09.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Tesarz, J. , Schuster A. K., Hartmann M., Gerhardt A., and Eich W.. 2012. “Pain Perception in Athletes Compared to Normally Active Controls: A Systematic Review With Meta‐Analysis.” Pain 153, no. 6: 1253–1262. 10.1016/j.pain.2012.03.005. [DOI] [PubMed] [Google Scholar]
  62. Tulppo, M. P. , Hautala A. J., Mäkikallio T. H., et al. 2003. “Effects of Aerobic Training on Heart Rate Dynamics in Sedentary Subjects.” Journal of Applied Physiology 95, no. 1: 364–372. 10.1152/japplphysiol.00751.2002. [DOI] [PubMed] [Google Scholar]
  63. van der Mee, D. J. , Gevonden M. J., Westerink J. H. D. M., and de Geus E. J. C.. 2023. “Cardiorespiratory Fitness, Regular Physical Activity, and Autonomic Nervous System Reactivity to Laboratory and Daily Life Stress.” Psychophysiology 60, no. 4: e14212. 10.1111/psyp.14212. [DOI] [PubMed] [Google Scholar]
  64. Verhoeven, J. E. , Han L. K., Lever‐van Milligen B. A., et al. 2023. “Antidepressants or Running Therapy: Comparing Effects on Mental and Physical Health in Patients With Depression and Anxiety Disorders.” Journal of Affective Disorders 329: 19–29. 10.1016/j.jad.2023.02.064. [DOI] [PubMed] [Google Scholar]
  65. Watson, D. , Clark L. A., and Tellegen A.. 1988. “Development and Validation of Brief Measures of Positive and Negative Affect: The PANAS Scales.” Journal of Personality and Social Psychology 54, no. 6: 1063–1070. 10.1037/0022-3514.54.6.1063. [DOI] [PubMed] [Google Scholar]
  66. Wirtz, P. H. , and von Känel R.. 2017. “Psychological Stress, Inflammation, and Coronary Heart Disease.” Current Cardiology Reports 19, no. 11: 111. 10.1007/s11886-017-0919-x. [DOI] [PubMed] [Google Scholar]
  67. Zouhal, H. , Jacob C., Delamarche P., and Gratas‐Delamarche A.. 2008. “Catecholamines and the Effects of Exercise, Training and Gender.” Sports Medicine 38, no. 5: 401–423. 10.2165/00007256-200838050-00004. [DOI] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Data S1: psyp70350‐sup‐0001‐Supinfo.pdf.

PSYP-63-e70350-s001.pdf (907.7KB, pdf)

Data Availability Statement

All data and analysis code are made available on OSF: https://osf.io/ex2t4/overview?view_only=6a67433550424110b82d24dc0613b5d9.

Articles from Psychophysiology are provided here courtesy of Wiley

RESOURCES