Immediately after perceiving a stressor threatening an organism’s homeostasis, the SNS is activated. Activation of the SNS leads to the release of both adrenaline and noradrenaline (7). Adrenaline and noradrenaline release provides the body with additional energy to increase alertness and attention helping to prepare for the so-called fight-or-flight reaction (8). In addition, the activity of the parasympathetic branch of the autonomous system (i.e., the rest-and-digest reaction) is decreased. Characteristically, activation of the SNS results in an elevated heart rate and blood pressure, accelerated respiration, and constricted digestion. Activation of the SNS occurs rapidly and returns to normal as soon as the source of stress is resolved (9).

In contrast, activation of the HPA axis is slower. It involves the release of a cascade of hormones, resulting in a release of cortisol into the bloodstream. Unbound cortisol (~5-10%) has various effects on its target systems throughout the body, aiming to supply additional energy to the organism (4). Notably, peak cortisol secretion is observable between fifteen and thirty minutes after the onset of the stress reaction and about fourteen minutes after the peak activation of the SNS (10). Moreover, the HPA axis can be active in anticipation of a stressor as well as chronically (11).

Both the SNS and the HPA stress reactions can be measured by saliva samples, namely salivary alpha-amylase (sAA; 12, 13) and salivary cortisol (sC; 14). Alpha-amylase is a salivary enzyme that hydrolyses starch to glucose and maltose in the oral cavity. In recent years, sAA has been investigated as a non-invasive biomarker to assess activation of the autonomic nervous system, more specifically, of the SNS. Research has shown alpha-amylase to be highly sensitive to stress-related changes (13).

For the organism to adaptively cope with acute stress, it can be assumed that the SNS and the HPA axis interact and coordinate. Anatomically, both systems intersect at the level of the central nervous system (9). However, the details of their interplay are not yet fully understood.

Due to its complexity, the interaction between SNS and HPA axis is most often studied in acute stress conditions using laboratory stress tasks such as the Trier Social Stress Test (TSST; 15). Intuitively, a coherent response of both systems would seem optimal, such that the perception of a stressor activates both the SNS and the HPA axis. Accordingly, research has shown that levels of sAA and sC throughout a stressful situation in the laboratory were reliably correlated at different time lags (10). However, studies using pharmacological suppression of either system also show an active mechanism of mutual compensation. Partial suppression of the HPA-axis stress response in an acute stress paradigm increased SNS activity compared to a placebo group (16). Conversely, suppressing SNS activity was associated with increased sC levels after stress exposure (16). Thus, the active system seems to compensate for the suppressed system (9).

An asymmetry between the SNS and HPA axis indicating dysregulation can be observed without any pharmacological suppression, i.e., with chronic stress experiences. Studies assessing stress reactivity have shown that a stressful childhood may be associated with an asymmetry in the reaction of SNS and the HPA axis. An exposure to traumatic stressors in childhood can also affect the brain significantly (6, 17). Gordis et al. (18) found an asymmetric relationship between sAA and sC in maltreated youths compared to a control group, such that SNS and HPA axis measures in reaction to the TSST were positively correlated in a control group but not among maltreated youth.

To further examine the interplay between the SNS and the HPA axis, Ali and Pruessner (19) calculated the ratio of sAA over sC as an index of asymmetry. This ratio serves as a measure that includes markers for both SNS and HPA axis reactivity as well as their balance. The authors found that adults with a history of early life adversity (ELA) showed greater asymmetry (i.e., a higher ratio of sAA/sC) in reaction to the TSST compared to adults without ELA. Moreover, the sAA/sC ratio was more strongly related to ELA as well as chronic stress and depressive symptoms compared to either sAA or sC alone. Thus, stressful events in childhood seem to be associated with a dysregulated physiological stress response, and this asymmetry can be captured in a ratio. However, studies assessing SNS and HPA axis activity in natural settings instead of stress reactivity in laboratory settings are rare. Karhula et al. (20) studied differences in sC and sAA levels in healthcare professionals with either high or low job strain with a laboratory stress task (i.e., the TSST) as well as in the field (i.e., shift work). They found that individuals with high levels of job-strain had a higher sC reactivity in the TSST compared to the group with low job-strain. Moreover, subjects working in the morning shift also had higher sAA levels 30 minutes after awakening compared to the low job strain group. However, the two groups did not differ in the sAA/sC ratio or the sC/sAA ratio. Nislin et al. (21) investigated early childhood professionals in Finland. They assessed five saliva samples per working day and one weekend day, as well as burnout symptoms and self-reported work engagement. Exploring the sAA/sC ratio and the sC/sAA ratio, they did not find significant correlations between either ratio values and self-reported work engagement nor between ratio values and burnout. However, they noted that their specific sample scored lower on burnout compared to teachers in general education and other professions. Thus, the relationship between the physiological stress reaction and ongoing work stress is less clear.

Individuals vastly differ in stress reactivity levels, and stress reactivity is often speculated to be related to susceptibility to different diseases. Individual susceptibility to cardiovascular disease is linked to cardiovascular stress responses, such that greater stress reactivity and poorer recovery from acute mental stressors predict future cardiovascular disease (22). A recent review by Turner et al. (23) showed that this reactivity hypothesis also seems true regarding other biomarkers (e.g., blunted salivary cortisol reactivity), and stress consequences (e.g., regulatory T cell percentage). Moreover, they assume an invented-U model, in which both an exaggerated and a blunted stress response are related to adverse longitudinal physiological health consequences. However, in this overview, only two studies included assessed markers of both physiological systems and longitudinal physiological health outcomes, with inconclusive results.

Psychological stress consequences and their association with the physiological stress response are less clear. A coherent response of all involved stress systems (i.e., the psychological system or either of the two physiological stress systems) is often considered adaptive, meaning that a stressor concurrently activates both, the psychological processing of the stressful stimulus and the physiological stress response. Accordingly, once the stress is averted, all systems deactivate and recover. However, research shows that activation of the psychological and the physiological system is not strongly associated (24, 25). Even if the physiological stress reaction is pharmacologically suppressed, the emotional stress reaction still occurs (26). Thus, the psychological and physiological stress systems appear to function almost independently of each other. Accordingly, people often do not consciously perceive their own physiological stress reactions. If the physiological arousal is perceived, it is often misattributed to a different source and not to the acute stressor. Experiencing psychological stress therefore appears to depend on other factors and not necessarily on physiological arousal. However, if the short-term psychological stress reaction is practically independent of the physiological arousal, one could speculate that any longer-term psychological stress consequences (e.g. psychological strain or psychosomatic consequences) are also not necessarily related to the physiological dysregulation. However, this association is unclear.

Psychological strain can manifest in different areas of life (e.g., at work, regarding the self, or in one’s private life; 27). Occupational problems also include negative feelings related to work. Self-related problems include low frustration tolerance, feelings of depersonalization, or concentration problems. Family-related problems encompass estrangement or decreased participation in familial life. Friends-related problems include reduced interest in friends’ lives and retreating from social contact. Psychological strain can also include psychosomatic symptoms. Vital exhaustion is a psychosomatic state of unusual fatigue, lack of energy, irritability, and demoralization (28). Vital exhaustion is a potential early warning sign of cardiovascular disease (29) and seems closely related to burnout (30).

Individual personality differences can influence stress reactivity (31) and might influence whether stress consequences are evident psychologically. A high core self-evaluation (CSE) can protect an individual against stress (32, 33). CSE is a higher-order construct that includes four personality traits: global self-esteem, generalized self-efficacy, emotional stability, and locus of control. People with high CSE perceive themselves as worthy, capable, and in control of their lives (34). In contrast, neuroticism (i.e., emotional instability) is the disposition to experience negative affectivity (e.g., anger, anxiety, self-consciousness, irritability, depression; (35), and people with high neuroticism interpret ordinary situations as more threatening (36).

Individuals also differ in their strategies to cope with a perceived stressor. However, not all strategies are adaptive in the long run. Approach strategies involve actively confronting the stressor (e.g., active problem-solving, seeking support). In contrast, avoidance strategies aim to withdraw from the stressor (e.g., giving up, social withdrawal), but the problem remains unsolved. Thus, while successful in the short term, the unresolved situation might continue to generate stress. Both personality traits and coping strategies can influence how a stressor is dealt with.

Teaching involves many challenging situations to which the teacher ideally reacts proactively. A high density of social interactions with students, colleagues, or parents, social exposure and evaluation, and classroom disruptions can all contribute to teacher stress (37). However, to the best of our knowledge, no studies have assessed the association between teacher stress and the dysregulation of the physiological stress reaction, so far.

The dysregulation of the physiological stress response has mainly been investigated in laboratory settings using (acute) stress tasks such as the TSST, but not in response to frequently occurring and/or chronic stressors so that results cannot be easily generalized to everyday life. Social settings differ substantially with respect to stress reactions. Thus, particularly for a socially demanding profession such as teaching, physiological stress-sensitive responses should be measured using ambulatory assessments (i.e., during teaching and during leisure time).

Moreover, most studies rely on one or two stress-related outcomes. Even if different concepts of psychological and psychosomatic strain are related, they differ in their underlying models and operationalizations. Thus, it is important to include heterogeneous constructs to assess the consequences of psychological stress in different areas of life. Additionally, individual characteristics such as personality factors and coping styles can serve as risk or protective factors.

To this end, we measured teachers’ physiological stress-sensitive markers, namely sAA and sC, on two workdays and one leisure day. To incorporate both stress systems, we used the sAA/sC ratio to explore a possible dysregulation of the SNS and the HPA axis and associations with personality factors, coping styles, and psychological or psychosomatic strain. Moreover, we assessed teachers’ psychological and psychosomatic strain on two time points, that is, at baseline, and at 2-year follow-up, to consider longitudinal aspects. We pose the following research questions:

The study included 42 apparently healthy teachers (28 female) at baseline. The average age was 39.66 years (SD = 11.99). Participants were recruited via flyers and circular emails sent to school administrations in the canton of Bern, Switzerland. After the first measurement, one teacher moved abroad, and two teachers withdrew their participation due to pregnancy, resulting in a sample of 39 teachers two years later. All teachers were screened for the inclusion and exclusion criteria during a short telephone interview. Inclusion criteria for participation in the study were employment as a primary or secondary teacher in the Swiss canton of Bern and a workload of at least 16 lessons per week (equivalent to at least 60 percent of full-time employment). Exclusion criteria were working outside of the canton of Bern, acute infections, cardiovascular or other chronic diseases, use of cardiovascular drugs or other medication in the past two months (except phytopharmaceuticals), substance abuse, drug use in the last four weeks, more than two standard units of alcohol per day, smoking more than ten cigarettes per day, long-distance flights within the last two weeks, and pregnancy. Participants worked at 39 schools and reported a mean of 13.35 years of teaching experience (SD = 11.07, range = 1–40). Participants had completed standard teacher training and were regular classroom teachers, not teachers for the gifted or students with special needs. The grade levels teachers taught ranged from kindergarten and elementary school (kindergarten to 6th grade; n = 27), over secondary school (7th to 9th grade; n = 12), to high school and vocational school (10th to 12th grade; n = 3). Enrolled participants provided informed consent. The study was approved by the ethics committee of the canton Bern (no. 2019–00787), as well as the Internal Review Board (IRB) of the University of Bern. It was conducted in strict compliance with current data protection laws.

The study comprised a longitudinal multi-method ambulatory design. Assessments were conducted at three time points: Participants completed online questionnaires on personality factors, coping styles, and psychological strain at t0 in December 2019 (before the Covid-19 pandemic).

Saliva samples were collected on two workdays and one leisure day in an ambulatory assessment design between January and November 2020 (t1; before and during the pandemic). When saliva samples were collected, teachers taught in person, not online. The teachers were instructed verbally and in writing to refrain from drinking alcohol starting the evening before each collection day. They were further instructed not to engage in strenuous physical activity on the three ambulatory assessment days. The three saliva collection days for each participant took place within one week. The order (workday or leisure day first) was randomized. However, due to rescheduling among two participants, 23 teachers first collected the leisure day samples, while 19 started with the workday sample collection. When determining the data collection days, we considered the menstrual cycle of our female participants to ensure that the three days of data collection did not fall within the follicular phase, as this could affect cortisol binding and HPA axis reactivity (14).

Two years later (t2), the psychological strain was reassessed using online questionnaires.

All data were analyzed using IBM SPSS Statistics (IBM SPSS Statistics, Version 28). For the self-reported measures, normal distribution was tested using the Shapiro-Wilk test. Skewed variables were then log-transformed using the natural logarithm. Potential confounders, such as age, sex, and teaching experience, were tested for significant associations with the main variables by calculating Pearson correlations. No significant associations were found.

To measure the overall intensity of sAA and sC, the area under the curve with respect to ground (AUCg) was calculated (50) using the raw, untransformed values of the five measurement points from 8 am to 8 pm (resp. +2h and +14h of waking up at the leisure day). AUCg represents both the baseline and the diurnal course in one score. For both sAA and sC, AUCg was calculated twice: first, using the averaged measures of the two workdays, and second, for the leisure day. As the calculated AUCg values were not normally distributed, they were log-transformed using the natural logarithm.

To assess the dysregulation of the SNS and HPA axis, the ratio of sAA over sC was calculated using the calculated AUCg (not yet log-transformed; see Sollberger and Ehlert (51) for a discussion of how to calculate a ratio). The ratio was calculated twice: first, using the AUCg of both sAA and sC of the averaged workdays, and second for the leisure day. Then, according to the procedure Sollberger and Ehlert (51) described, we log-transformed both ratios using the natural logarithm.

We calculated descriptive statistics of all variables and bivariate Pearson correlations. To account for the different saliva collection times, we controlled for the time delay between psychometric assessment at t0 and saliva collection. To assess the longitudinal associations between physiological measures and psychological strain at t2, we calculated partial correlations controlling for the psychological strain variables of t0. A Wilcoxon signed-rank test was performed to investigate any significant differences in the ratio of the averaged workdays and the leisure day. All analyses were two-tailed, with the significance level set at p <.05.

Descriptive statistics for the physiological stress-sensitive measures (AUCg of sAA and sC, and their ratio) are depicted in Table 1 . Results regarding the diurnal course and the awakening reaction of both alpha-amylase and cortisol are reported elsewhere (40). Intercorrelations are reported in the Appendix (Table A1) and in detail in Wettstein et al. (39).

A Wilcoxon signed-rank test was performed to investigate differences in the assessment setting of the physiological measures sAA, sC, and their ratio. Results revealed no significant difference between sAA of the averaged workdays (Mdn = 1029.17) and sAA of the leisure day (Mdn = 1443.16, z = -0.57, p = .572), as well as no significant difference between the sAA/sC ratio of the averaged workdays (Mdn = 3.34) and the sAA/sC ratio of the leisure day (Mdn = 3.74, z = -1.40, p = .162). However, there was a significant difference between sC of the averaged workdays (Mdn = 35.91) and sC of the leisure day (Mdn = 29.85, z = -1.98, p = .048, r = .33).

Bivariate correlations between assessed psychological variables and the physiological markers on workdays are depicted in detail in Table 2 .

The present study has several limitations. First, the sample was relatively small and consisted of apparently healthy and medication-free teachers, limiting generalizability. Second, the saliva sampling relied solely on our participant’s accuracy in following the sampling instructions, as there was no objective control of compliance. Thus, it is possible that the results were influenced by non-adherence to the sampling protocol, limiting their ecological validity. Third, participants gave stimulated (chewing on cotton rolls) in contrast to non-stimulated (passive drooling) saliva samples. This could have possibly altered results concerning sAA, as different salivary glands may have different salivary secretion rates, which could influence the amounts of sAA delivered into the oral fluid (55, 56). Fourth, there is some evidence that sAA can also be influenced by parasympathetic activity (56). Thus, it might not be a pure marker of SNS activity. Fifth, salivary measures were assessed before and during the COVID-19 pandemic. Although teachers were teaching face-to-face in the classroom as usual during the data collection period, we cannot exclude that factors related to the pandemic might have influenced the results. While it is still interesting that most of the associations were only statistically significant using workday saliva samples, we did not control what kind of stressors led to this result. Thus, the reported results should be interpreted accordingly. Last, our results are based on correlations and should not be interpreted causally. Future research should replicate the presented finding with a larger, less homogenous sample.

Nevertheless, this study has several strengths. First, the combination of self-reports and an ambulatory assessment of stress-sensitive biomarkers increases the ecological validity of the presented results. Few studies assess the response of the physiological stress systems outside of the laboratory. However, ambulatory assessment is necessary to understand stress occurring in natural settings (57). Moreover, the longitudinal assessment of psychological strain allows for a more nuanced statement regarding long-term stress consequences. Thus, this study contributes to the scarce literature on physiological stress in natural settings and its (longitudinal) associations with strain. Second, whereas much is known about cortisol, alpha-amylase and its associations are less studied. Moreover, the ratio of alpha-amylase and cortisol promises to be a lean way to assess a possible asymmetry of the physiological stress systems. Thus, the present study can help researchers to identify important factors linked to alpha-amylase and the SNS.

The present study combines a longitudinal multi-method design with ambulatory assessment and provides crucial information on the interplay of the two physiological stress systems, the SNS and HPA axis, in teachers. Results show that, cross-sectionally, sAA was an important indicator of increased levels of psychological and psychosomatic strain and was positively related to individual characteristics of risk. Longitudinally, the sAA/sC ratio as a reflection of an asymmetry between sAA and sC was positively related to psychological and psychosomatic strain. These results underline the importance of including measurements of both physiological stress systems. Moreover, the physiological response becomes evident also on a psychological level in that it is related to higher levels of strain.