Eighteen males, who were part of the Military Firefighters Corps with at least 3 years of profession, were recruited. For convenience, the sample consisted of firefighters taking the sergeant training course. Exclusion criteria were: have visible or known cardiovascular, mental, or metabolic diseases; have infectious or inflammatory processes; have muscle or joint damage; smokers; or take a drug that could influence the analyzed variables. Two firefighters were excluded because they used drugs that could influence the variables assessed. One participant was excluded from the analysis of autonomic modulation during the cold pressor test (CPT) due to recording failure. We registered the protocol in the Brazilian Registry of Clinical Trials (ReBEC; n^o. RBR-24db7p7). The Human Research Ethics Committee of the Federal University of Santa Catarina approved the experimental procedures (protocol n. 4.942.301). The participants signed the Free and Informed Consent Term.

This experimental randomized crossover study was conducted at the Military Firefighters Corps in Florianopolis, Santa Catarina, Brazil. Participants made three visits to the laboratory. At visit one, participants received information about the research and signed the informed consent form. In visits two and three, the following interventions were performed randomly: a) Firefighting Simulation; b) Control: at rest for 10 min (average Simulation time). Randomization was performed using software with random allocation sequence generation function (Microsoft Excel^®, Microsoft^®, Redmond, WA, USA) so that half of the participants started with the Control and half with the Simulation on each data collection day. All the participants performed the interventions in the afternoon starting at the same time. The interval between visits was 7–15 days. In addition, at visit two, before randomization, participants answered questionnaires on mental health, sleep quality, and physical activity levels and underwent an anthropometric assessment (waist circumference, height, and body mass). At visits two and three, before starting the Simulation or Control, the participants remained seated, at rest, for 5 min. Then, they received a call to start the intervention. After the interventions, we assessed cognitive performance using the Trail Making Test, the Stroop Test, and the Paced Auditory Serial Addition Test (PASAT). Finally, the vascular function was assessed through carotid artery reactivity (CAR) to CPT using a Doppler ultrasound. At Simulation and Control visits, the HR and cardiac intervals were recorded continuously using a chest strap connected to a smartphone application for 5 min before and during interventions, and during CPT. Also, the AP was measured through an automated oscillometric device before starting the interventions to evaluate the anticipatory effect. An extra AP measurement was performed only in the Control visit after the rest period to measure the basal AP value for sample characterization (Figure 1). We provided Methods details below.

The firefighting Simulation was developed based on a training protocol used in the Military Fire Corps of Santa Catarina, Brazil. The Simulation was composed of five firefighting tasks (Figure 2) and was performed in an open environment with a temperature of 21.1°C ± 2.2°C. After collecting the baseline parameters, the participant remained seated until he received a verbal prompt from a designated firefighter to start the Simulation. The tasks were: I) Getting ready. Participants should wear firefighting clothes and shoes and an oxygen cylinder (totaling ∼26 Kg; distance from rest to task I: ∼30 m). II) Climbing with equipment. Fetch a hose roll (∼10 Kg) and go up three flights of stairs taking it. Then, leave the hose and go back to get a second hose, go up the ladder again, leave the hose, and go down the ladder (distance from task I to task II: ∼30 m). III) Rescue. Drag a dummy (∼70 Kg) for 40 m (distance from task II to task III: ∼40 m). IV) Route. Travel 510 m as fast as possible (distance from task III to task IV: ∼40 m). V) Active recovery. Five min consisted of removing firefighting clothing followed by light walking (distance from task IV to task V: ∼40 m).

Participants were instructed to perform the tasks as if they were in a real firefighting situation. Immediately after the Simulation, the Borg Scale (6–20) was used to verify the subjective perception of exertion during the intervention (Borg, 1977; Eston and Williams, 1988).

To characterize firefighters in terms of sleep quality, physical activity level, and mental health status, we used the Pittsburgh Sleep Quality Index (PSQI), the International Physical Activity Questionnaire (IPAQ), and the Depression, Anxiety, and Stress Scale-21 (DASS-21), respectively.

The PSQI assesses the quality of sleep in the previous month. The higher the score, the worse the sleep quality (Buysse et al., 1998). The final score classifies sleep as good (0–4 points), poor (5–10 points), and presence of disturbance (>10 points). We also used the sleep efficiency provided by the PSQI, which results from dividing the number of hours slept by the number of hours in bed multiplied by 100. The higher the percentage, the greater the sleep efficiency. That is, >85% (score 0), 75%–84% (score 1), 65%–74% (score 2), and <65% (score 3) (Bertolazi et al., 2011).

The IPAQ infers the levels of physical activity from the previous week (Matsudo et al., 2001). Participants who practiced at least 150 min per week of moderate physical activity or 75 min per week of vigorous activity were considered sufficiently active (World Health Organization, 2020).

The DASS-21 assesses the mental health status in the previous week, using the Likert scale, where the number 0 indicates “not applied at all”; 1, “applied to some degree or for a short time”; 2, “applied to a considerable degree or for a good part of the time”; 3, “applied a lot or most of the time”. The scale contains 21 questions, divided into three categories (stress, depression, and anxiety), with seven questions for each (Parkitny and McAuley, 2010; Vignola and Tucci, 2014). The final score is counted and multiplied by two. For depression, it is considered normal (0–9 points), minimal (10–13), moderate (14–20), severe (21–27), and very severe (28+). For anxiety, normal (0–7), minimal (8–9), moderate (10–14), severe (15–19), and very severe (20+). For stress, normal (0–14), minimal (15–18), moderate (19–25), severe (26–33), and very severe (34+).

The Trail Making Test, Stroop Test, and the PASAT were used to assess cognitive performance.

The Trail Making Test, is an adaptation of Partington’s Test of Distributed Attribution, first developed by John E. Partington (Partington and Leiter, 1949). The test contains two parts, A and B, and evaluates attention, processing speed, cognitive flexibility, and visual search. In part A, the participant has to link a sequence of 25 numbers in ascending order. In part B, the task consisted of interspersing a sequence of letters in alphabetical order and numbers in ascending order, totaling 25 digits (Llinàs-Reglà et al., 2017).

The Victoria version of the Stroop Test, developed by Regard (1981), assesses executive functions through selective attention skills and inhibitory behavior. Three cards were used. With card 1, the participant names the color of colored rectangles. With card 2, he identifies the ink color in which the words are printed. With card 3, the participant names the ink color of words, but the words’ names correspond to different colors, generating an incongruous stimulus (Campanholo et al., 2014). In this test, we used color naming time to assess cognitive performance.

The PASAT, developed by Sampson (1958), consists of a sequence of 61 numbers presented audibly, separated by 3-s of interval. The participant has to add the last number heard to the next and answers the result before the following number is presented (Tombaugh, 2006). The number of correct answers was considered as an assessment criterion.

AP was measured in sitting using an automated oscillometric device (Omron HBP 1100, Kyoto, Japan). The acquisition and recording of HR and cardiac intervals were performed using a Polar H10 chest strap (Polar Electro, Kempele, Finland) connected to the Elite HRV smartphone application (Asheville, NC, USA) at 1,000 Hz. The application was previously validated and showed a high correlation with electrocardiogram (Gambassi et al., 2020; Moya-Ramon et al., 2022).

Cardiac interval series were exported from the smartphone application, and HRV analyses were performed using the Kubios HRV Standard software (version 3.5.0; Kubios Oy, Kuopio, Finland). Before the cardiac autonomic modulation analysis, the cardiac interval series were examined, and artifacts and ectopic beats (i.e., non-sinusal beats) were removed using Kubios’s digital filter, considering a maximum acceptable loss of 5% of beats. Since the participants were standing still before the interventions and during CPT, only a very low digital filter was applied. None of the cardiac interval series exceeded 5%.

Time domain indices were analyzed: standard deviation of normal cardiac intervals (SDNN) and the square root of the mean square of the differences between adjacent normal cardiac intervals (RMSSD). In the frequency domain, the time series were transformed into uniformly spaced series by interpolation of cubic splines (4 Hz) and were distributed in overlapping semi-sets of 300 points (Welch periodogram). A Hanning window was used to mitigate the side effects and the interpolated time series had the spectra calculated by the Fast Fourier Transform (FFT) algorithm. The spectra were integrated into the low-frequency bands (LF; 0.04–0.15 Hz), inferring a mixed sympathetic and parasympathetic modulation, but with sympathetic predominance; high frequency (HF; 0.15–0.40 Hz), which correlates with parasympathetic modulation to the heart. Data were presented as natural logarithm (Ln)-transformed and normalized units (n.u.). The LF/HF ratio was analyzed to assess sympathovagal balance (Task Force, 1996; Formolo et al., 2022).

We used a 5-min cardiac interval series to analyze the anticipation of interventions, obtained immediately before starting the Simulation or Control for all time and frequency domain variables. For CPT analyses, we used 1-min cardiac interval series (baseline, 1st, 2nd, and 3rd min of test). In this case, due to the short recording period, we only used the HF (Ln) data, as suggested previously (Task Force, 1996).

The CPT was performed with the participants in the supine position on a stretcher, positioned close to one of the right sides, ensuring hand movement without significantly moving the neck. During CPT, the right hand was submerged in cold water (∼4°C). The water temperature was measured with a digital thermometer. Participants were instructed to breathe normally and not speak during the test. After a 1-min baseline, the participants were instructed to place their right hand in the cold water for 3 min (Peace et al., 2020).

The diameter of the left common carotid artery was assessed by ultrasound (Toshiba Viamo, Toshiba Medical Systems Corporation, Japan) by an experienced evaluator. The image of the common carotid artery, proximal to the carotid bulb, was optimized so that the artery walls were well-defined. Doppler velocity was also recorded at the smallest possible insonation angle (always <60°). Carotid artery diameter was evaluated at rest (1 min) and during CPT (3 min). The mean carotid diameter at rest was considered the basal diameter. During the CPT, the carotid diameter was averaged every 10 s, and the value of the greatest relative change in diameter, above or below the diameter at rest, was used to calculate the CAR in percentage (Peace et al., 2020). The video analyses were performed by a blinded evaluator using FloWave software run in MATLAB 2012b (Coolbaugh et al., 2016).

We presented the description of the variables as the mean ± standard deviation of the mean. We analyzed the data using Graphpad Prism 6.0 statistical software (Graphpad Inc.; La Jolla, USA). We verified the data normality using the Shapiro-Wilk test. For Control and Simulation visits comparisons, we used the Student’s T-test with paired measures for normal distribution data and Wilcoxon for non-normal distribution data. Since the Shapiro-Wilk test did not show evidence of non-normality in HF (Ln) data (Control: W = 0.9803, p = 0.9661; Simulation: W = 0.9761, p = 0.9254), we used the analysis of variance (ANOVA) for repeated measures to assess the HF (Ln) as a function of time during CPT. For all tests, a significance level of p < 0.05 was adopted.

Table 1 presents the characteristics of the firefighters, including age, anthropometry, baseline cardiovascular parameters, levels of physical activity, sleep quality, and mental health.

Table 2 presents the firefighting Simulation characterization.

On Simulation, the firefighters performed the Stroop Test, part B, faster than Control. There was no difference between the interventions in the Trail Making Test, PASAT, and Stroop Test, part A (Figure 3).

Analysis of carotid artery reactivity to CPT showed greater carotid dilation in the Control than in the Simulation. In the Control, all participants exhibited carotid dilatation during the CPT. However, in the Simulation, four participants showed vasoconstriction. There was no difference between interventions in the baseline diameter and time-to-peak measurements (Table 3).

Ln HF analysis showed an intervention effect and an intervention vs time interaction, indicating a lower parasympathetic modulation to the heart during CPT after the Simulation compared to the Control (Figure 4).

The mean and maximum HR were higher before the Simulation than the Control. Measurements of HRV and AP did not show differences between the interventions (Table 4).