More broadly, acute physiological responses to adventure activities have been assessed using salivary cortisol (Draper et al., 2010; Ewert et al., 2016; Magiera et al., 2018), urine catecholamines (Bunting et al., 2000), EEG activity (Bailey et al., 2017), venous blood catecholamines and cortisol (Baláš et al., 2017; Bunting and Gibbons, 2001), heart rate (HR) (Draper et al., 2010; Giles et al., 2020; Magiera et al., 2018; Watts et al., 1999), and respiratory markers (Draper et al., 2010; Gajdošík et al., 2020; Giles et al., 2020; Watts et al., 1999).

While hormones from both the hypothalamic–pituitary–adrenal (HPA) axis and the sympathetic-adrenal-medullary axis reflect the stress response, the use of cortisol can be challenging due to individual variation in reactivity (i.e., how long it takes for an associated peak in cortisol to be reached after a stressor), especially when data are collected at a set time following the event of interest, individual variation in diurnal rhythms, and the expense of processing samples. The use of other catecholamines can be problematic in outdoor settings due to rapid degradation after stress cessation (Ewert, 2015; Zouhal et al., 2008).

Therefore, utilizing heart rate, respiratory markers, and heart rate variability (HRV) is often a more financially and logistically viable way for assessing the stress response in outdoor settings such as adventure activities. For instance, HRV has been shown to reflect vagal activity in these settings (Gladwell et al., 2016), and heart rate is a frequently used measure in sports and exercise sciences and has also been shown to respond to various stressors (Taelman et al., 2009).

Therefore, the current study sought to address the literature gap regarding physiological markers of stress measured during an adventure activity at elevated heights.

Either to focus on the mental aspect or to triangulate and provide a more holistic understanding of the potentially stressful activity, psychological self-report is often employed. Anxiety responses, which are of particular interest in adventure activity-based programs, can be assessed using a range of questionnaires, such as the Competitive State Anxiety Inventory-2 (CSAI-2) (Baláš et al., 2017; Draper et al., 2010; Giles et al., 2020), State–Trait Anxiety Inventory (STAI), Anxiety Thermometer (Pijpers et al., 2003, 2005), and Spielberg State Anxiety Inventory (SSAI) (Bunting and Gibbons, 2001). Anxiety is an emotion characterized by feelings of tension, worried thoughts, and physical changes in the body (Foa et al., 2005). It is made up of multiple components, although the exact number and nature of these components remain a topic of debate in the literature (Brodbeck et al., 2011; Clark and Watson, 1991). The CSAI-2 questionnaire models anxiety as comprising three components—cognitive, somatic, and self-confidence. Although stress and anxiety are interrelated, it makes sense that for adventure activities at height, anxiety—as a future-oriented state—will likely be most pertinent before performing the activity, whereas stress will be more pertinent during the performance.

Although combinations of physiological markers and self-reported anxiety are frequently used in research and practice, very little is currently known about the concordant validity between psychometric tools such as the CSAI and some physiological markers of stress (Goyal et al., 2016), particularly when they are being used in adventure domains. In activity-based adventure programs, participants often report or are observed to experience high anxiety immediately before the activity at height. It would be useful for practitioners to understand the extent to which pre-activity anxiety translates into stress during the activity itself. Therefore, the secondary aim of this study was to assess the concordant validity of the Competitive State Anxiety Inventory-2 Revised (CSAI-2R) questionnaire with respiratory and cardio markers of stress in order to understand the extent to which stress and anxiety are separate ‘constructs’ and whether self-reporting can be used as a proxy measure for physiological markers.

A total of 55 university students (weight 72.2 ± 11.2 kg, height 177.1 ± 8.1 cm, age 20.3 ± 2.1 years) participated voluntarily in this study. They were healthy and asked to refrain from high-intensity exercise (24 h), caffeine (12 h), and heavy meals (2 h) prior to the experiment. All participants signed an informed consent at the beginning of the study. The local university’s ethics committee granted approval for the study (EK238/2020). All participants participated voluntarily.

All participants completed three 4-min walks on a narrow log at two different heights. The first walk took place on a log 0.3 m above the ground at a self-selected speed and was used only for psychometric assessments. The second walk was conducted at 10.5 m above the ground, also at a self-selected speed. The third walk was performed again at 0.3 m above the ground, but this time at the same speed as the high-obstacle trial, ensuring a consistent walking speed for subsequent physiological comparisons.

Randomization of the conditions was not possible for the following reasons:

A portable metabolic system was worn on the chest, and anxiety metrics were recorded during all three conditions, as depicted in Figure 1. Participants completed the CSAI-2R immediately before the first and the second walk.

The same two rounded wooden logs (13 m long, 0.25 m in diameter) were placed at 0.3 m (low-obstacle) and 10.5 m (high-obstacle) above the ground, respectively. The middle 10-m section was marked at 1-m intervals using colored marks. The participants wore a full-body harness during both low- and high-obstacle conditions. However, only during the high-obstacle condition, the participants were belayed by an experienced instructor using a preinstalled rope. The rope was fixed on the participants’ backs so that they could not see it. The rope passed through a carabiner with a pulley, which was running freely on a metallic wire 4 m above the log. From the carabiner, the rope went back to the belayer at the ground level. To control the walking speed during the third walk, the bottom log was highlighted with colored marks every meter, and a researcher verbally guided the participants to accelerate or decelerate their walking pace. At the end of the log, a small balance support attached to a tree was provided to assist with turning.

Physiological responses were assessed using an indirect calorimetry open system and HRV.

Self-confidence, somatic anxiety, and cognitive anxiety were assessed using the CSAI-2R questionnaire. This inventory was selected for its three subscales and its widespread use in outdoor settings, ensuring a comprehensive and contextually relevant anxiety assessment. The CSAI-2R is a 17-item inventory, with each item scored on a Likert scale ranging from 1 (“not at all”) to 4 (“very much so”). Scores for each participant were combined to produce a score for each of the three subscales: (1) somatic anxiety (e.g., my heart is racing), (2) cognitive anxiety (e.g., I am concerned about performing poorly), and (3) self-confidence (e.g., I am confident because I can mentally picture myself reaching my goal). Just before the test during the first and second walks, the 17 statements were read and recorded by an examiner, and the participants verbally answered the appropriate number (1: Not at all, 2: Somewhat, 3: Moderately so, and 4: Very much so) to indicate how they felt at the moment (Cox et al., 2003).

Descriptive statistics (mean ± standard deviation) were used to characterize physiological and psychological responses to low and high obstacles. Differences in the CSAI-2R scores between low and high obstacles were assessed between the first and second walks, while differences in HRV and respiratory markers were evaluated between the second and third walks, when walking speed was controlled. Repeated-measures ANOVA was conducted to assess statistical differences and effect sizes. Furthermore, Pearson correlation analysis was used to assess the relationship between physiological markers and the CSAI-2R results. Statistical significance was set at a p-value < 0.05. The correlations and squared association indices were interpreted as follows: strong effect: R ≥ 0.8; R² ≥ 0.64, moderate effect: R ≥ 0.5; R² ≥ 0.25, and low effect: R ≥ 0.2; R² ≥ 0.04. All calculations were performed using Microsoft Excel and statistical software (IBM Corp., 2019, IBM SPSS Statistics for Windows, Version 28.0, Armonk, NY, USA).

During the high-obstacle condition, the participants significantly (p < 0.01) decreased their self-confidence scores (↓3.0 ± 5.6) and increased both somatic anxiety (↑6.1 ± 5.7) and cognitive anxiety (↑2.9 ± 5.9) compared to the low-obstacle condition (Table 1).

No meaningful correlations were found between HRV, respiratory markers, and psychometric stress markers. Only HR and the SNS index showed a moderate positive correlation with V_E (R > 0.47) (Table 3).

The CSAI-2R is a commonly used and popular inventory for assessing anxiety in sport (Cox et al., 2003; Lundqvist and Hassmén, 2005; Martinent et al., 2010). In the current study, scores from all three dimensions of the questionnaire differed significantly between the low- and high-obstacle conditions, affirming that the CSAI-2R effectively reflects acute stress induced by height. Height induced nearly a threefold greater effect on somatic anxiety compared to cognitive anxiety or loss of self-confidence. However, somatic anxiety demonstrated only a low correlation with respiratory and HRV markers (R = 0.02–0.39), raising questions about the concurrent validity of this CSAI-2R dimension in assessing physiological changes related to the stress response. Similarly, Souza et al. (2019) did not find any meaningful correlation (R = 0.0–0.21) between somatic anxiety from the CSAI-2 and salivary cortisol or LF/HF ratio in the pre-competition state of sub-elite athletes. To the best of our knowledge, the psychometric properties of the CSAI-2R have primarily focused on reliability and factorial validity (Cox et al., 2003; Fernandes et al., 2013; Martinent et al., 2010), while the concurrent validity of the inventory dimensions, notably somatic anxiety, with physiological stress markers has not been established. In the current study, we observed the strongest correlation between VT and cognitive anxiety (R = −0.45) and between V_E and somatic anxiety (R = −0.39). However, these relationships were not consistent across both correlation matrices, and their significance is questionable.

The low correlations between somatic anxiety and physiological markers of acute stress responses suggest that self-reports using the CSAI-R do not reliably reflect objective physiological changes in the body. This finding supports the notion of a dissociation between cognitive-affective and autonomic elements, thereby contributing to debates about the extent to which subjective and physiological stress responses operate via partially independent pathways depending somewhat on individual differences, such as interoceptive accuracy (Thayer and Lane, 2000; Thayer and Lane, 2009; Critchley et al., 2004; LeDoux, 1996). Therefore, integrating psychometric tools with multi-dimensional physiological monitoring may provide a more accurate assessment of acute stress responses.

HR increased significantly during the high-obstacle task (103 ± 15 beats.min⁻¹ for the low-obstacle task and 135 ± 18 beats.min⁻¹ for the high-obstacle task, respectively). This increase was accompanied by a marked reduction in RR intervals (597.1 ± 91.8 ms to 451.5 ± 61.7 ms), reflecting heightened cardiac workload and a shift toward sympathetic dominance during the more stressful task, while intensity was held constant. These two HR markers showed the greatest differences between low- and high-obstacle conditions based on the η_p² magnitude.

Time-domain HRV measures showed significant reductions, with SDNN decreasing from 23.4 ± 10.8 ms to 9.9 ± 6.9 ms and RMSSD dropping from 16.1 ± 10.7 ms to 6.1 ± 4.2 ms, indicating diminished parasympathetic activity and reduced autonomic flexibility.

Frequency-domain HRV measures showed a slight decrease in parasympathetic activity and a minor increase in sympathetic modulation during the high-obstacle task. The LF/HF ratio also increased slightly, indicating a trend toward sympathetic dominance. However, the lack of significant changes suggests that sympathetic and parasympathetic shifts occurred proportionally, maintaining relative autonomic balance. These results emphasize that task difficulty has a stronger impact on time-domain HRV measures than on frequency-domain proportional metrics.

High-obstacle tasks led to a significant reduction in the PNS index (−2.4 ± 0.8 to −3.6 ± 0.6) and a nearly threefold increase in the SNS index (4.7 ± 2.7–11.9 ± 5.4). The stress index (SI) doubled during high-obstacle tasks (22.4 ± 9.8–45.3 ± 20.7). These results demonstrate a pronounced shift toward sympathetic dominance, consistent with the “fight-or-flight” response to the heightened physical and mental demands of the task.

High-obstacle tasks significantly increased respiratory and metabolic demands, specifically with V_E rising from 22.5 ± 5.3 L.min⁻¹ to 40.5 ± 8.5 L.min⁻¹ from the low- to high-obstacle condition. This increase in V_E is much higher than the 7.7% increase reported during high versus low overground climbing (Gajdošík et al., 2020) in a similar population. It appears that the stress response to height is more pronounced when arm contact with a support is not possible and free space under the feet is present throughout the entire exercise. It should be noted that the increase in respiratory and metabolic responses was due to height exposure alone, as the speed of walking was kept the same as in the low-obstacle task. These responses also show that stress itself may increase the metabolic cost by 63%, as measured by VO₂.

There are several limitations that must be acknowledged. The sample consisted of a relatively homogenous group of Czech university students, which limits the study’s generalizability to the broader population and to the general domain of education or therapeutic programs. Future research should include participants with greater heterogeneity to enable cross-cultural comparisons in the analyses presented.

Across all measures, there is a possibility for learning effects across multiple trials. Habituation, increased task familiarity, and task sequence may have combined to sequentially reduce physiological and psychological stress measures, most notably during the second low-obstacle trial. Furthermore, in the long term, it is unknown how longitudinal repeated exposures might influence stress and anxiety responses. Indeed, substantial variability in stress responses was observed across all variables, reflecting individual differences in stress adaptation and perception. This variability is likely influenced by factors such as prior experience with height (Ewert et al., 2016; Gajdošík et al., 2020; Giles et al., 2014), personality traits (Brouwer et al., 2015; Luo et al., 2023; Penley and Tomaka, 2002), social perceptions (Gillman et al., 2023), and other individual characteristics. In addition, external factors such as temperature, wind, light rain, humidity, and physiological needs (e.g., hunger, thirst, fatigue) may have also affected the outcomes (Ewert et al., 2016). Future research should explore the mediating or moderating roles of these confounding factors in stress responses and examine how adventure-based programs influence changes in stress markers. Although our study focused on HRV, respiratory measures, and self-reported anxiety to assess acute stress responses, future research could benefit from incorporating a broader set of physiological and micro-behavioral indicators. For example, facial micro-expression analysis (e.g., orbicularis oculi electromyography) and posture or gait dynamics may offer more nuanced real-time insights into participants’ experiences of height. Integrating these measures into multimodal models, along with temporal self-reporting, HRV, and ventilatory measures, would enable continuous, ecologically valid stress monitoring. This approach would allow for improved temporal resolution in detecting stress responses and could enhance intervention strategies in high-stress environments such as outdoor adventure programs.