Intermittent post-exercise sauna bathing improves markers of exercise capacity in hot and temperate conditions in trained middle-distance runners

This study was approved by the University of Birmingham Ethics Committee (ERN_18-0958), and conformed to the standards set by the Declaration of Helsinki. All participants were informed of the experimental procedures and possible risks involved in the study before providing written consent.

Each participant completed a general health questionnaire and female participants completed a menstrual cycle questionnaire. Throughout the protocol, participants were asked to keep a training diary (including session type, running distance, frequency and session perceived exertion) and female participants were asked to record their menstrual cycle. An overview of the protocol is depicted in Fig. 1. Participants took daily iron tablets (65 mg ferrous sulfate; Nature Made, West Hills, CA, United States) from 2 weeks prior to the first experimental trial until the completion of the protocol.

A battery of experimental trials was completed within a 1-week timeframe at baseline (Pre), following 3 weeks (3-Weeks) and following 7 weeks (7-Weeks) of either normal endurance training (CON) or normal endurance training with the post-exercise sauna bathing intervention (SAUNA). Experimental trials included a running heat tolerance test (HTT), a temperate exercise test (consisting of a lactate profile test to determine running speed at 4 mmol L⁻¹ [La⁻] and a \(V\)O_2max test) and a resting venous blood sample. The temperate exercise test was conducted at Pre and 3-Weeks only.

Experimental trials following three-and seven-weeks intervention were arranged ~ 28 days apart to test female participants during the same menstrual cycle phase, as per a typical menstrual cycle. Participants self-allocated into groups prior to Pre testing. This experiment was conducted in the United Kingdom between the months of October and March to minimise any natural heat acclimatisation.

Fifty-one trained middle-distance and cross-country runners were recruited from the university’s athletics club to participate in this study. However, due to various reasons (injury, scheduling, etc.) only 20 athletes (SAUNA, n = 12 [F, n = 9; M, n = 3]; CON, n = 8 [F, n = 4; M, n = 4]) completed experimental trials at 3-Weeks. Two female participants completed both the SAUNA and CON interventions (in opposite order, with ~ 5 weeks washout), one ‘cross-over’ participant only completing the Lactate Profile test at both Pre and 3-Weeks whilst undergoing the CON intervention. Eight athletes completed experimental trials at 7-weeks, though only the six athletes in the SAUNA group (F, n = 3; M, n = 3) will be reported in the HTT at the final time-point (7-Weeks). The two athletes in the CON group who completed experimental trials at 7-Weeks were not included in the 7-Weeks analysis. An overview of the progression of athletes through the study is detailed in Fig. 2. Participants in the SAUNA and CON group at 3-Weeks had similar baseline characteristics (age 19 ± 1 and 20 ± 2 years, height 169 ± 7 and 172 ± 6 cm, \(\dot{V}\)O_2max 59.3 and 58.4 ± 7.2 ml kg⁻¹ min⁻¹, respectively; all p > 0.05). SAUNA participants who completed experimental trials at 7-Weeks were 19 ± 1 years of age, 171 ± 6 cm tall and had a baseline \(\dot{V}\)O_2max of 60.2 ± 10.1 ml kg⁻¹ min⁻¹. Female participants included in analyses at 3-Weeks were eumenorrheic (n = 7), oligomenorrheic (n = 1), amenorrheic (n = 1), or using various forms of hormonal contraceptives (monophasic oral contraceptive pill, n = 1; progestin-only oral contraceptive pill, n = 1; contraceptive coil, n = 1). Of these, female participants included in analysis at 7-Weeks were either eumenorrheic (n = 2) or amenorrheic (n = 1). Female participants did not report any negative menstrual or Pre-menstrual symptoms that may have affected performance (Giacomoni et al. 2000).

Participants entered the sauna (101–108 °C at ~ 1.8 m, 5–10% RH at ~ 1 m; iButton Hygrochron Logger, Maxim Integrated, California, USA) within ~ 5 min of cessation of outdoor exercise. Sauna bathing typically followed low-intensity, continuous exercise training sessions (i.e., ‘easy runs’ or ‘long runs’). Participants were asked to remain in the sauna for 30-min, as per previous investigations (Scoon et al. 2007; Stanley et al. 2015). In the sauna, participants sat upright and were allowed to drink ad libitum. Peak heart rate (HR) was recorded during the final minute of exposure.

Participants performed all experimental trials at the same time of day (± 2 h). Participants were asked to recall the food they had eaten prior to experimental tests, and to repeat the same diet when re-tested. Participants voided their bladder upon arrival to the laboratory to provide a urine sample, which was analysed for osmolality. If urine osmolality was ≥ 700 mOsm kg⁻¹ (Osmocheck, Vitech Scientific Ltd., West Sussex, United Kingdom; Sawka et al. 2007), participants drank 250 mL of water and did not begin exercising for at least 20 min.

The running heat tolerance test (HTT) was performed in hot conditions (40 °C, 40% RH) in an environmental chamber (TIS Services, Hampshire, United Kingdom) with a fan-generated airflow of ∼ 4 m s⁻¹. Towel-dried, nude body mass was recorded to 0.1 kg using digital scales (Seca 877, Seca, Hamburg, Germany) before and immediately after the HTT to estimate whole-body sweat loss. Participants wore socks, shoes, and either shorts (males) or shorts and a sports bra (females). Participants ran on a treadmill (H/P/Cosmos Quasar 4.0, H/P/Cosmos, Germany) at 9 km h⁻¹ and 2% gradient for 30 min (Mee et al. 2015). Rectal temperature (T_rec), skin temperature (T_sk), and HR were measured continuously. Perceptual measures, which included ratings of perceived exertion (RPE), thermal comfort and thermal sensation, were obtained in the final min of the HTT. Sweat gland activity of the forearm and upper back were recorded immediately following the HTT. Drinking was not permitted during the test.

The lactate profile test was performed in temperate conditions (~ 18 °C). Participants completed a light 10-min warm-up before commencing the step-style incremental test on the treadmill. Participants completed 3-min stages with 30-s stops between stages when capillary [La⁻] was measured. The test began at 1% gradient (Jones and Doust 1996) and a speed 4 km h⁻¹ slower than a recent 5-km race pace, and increased by 1 km h⁻¹ for each stage thereafter. HR and respiratory gas exchange were continuously measured. The test was terminated when [La⁻] exceeded 4 mmol L⁻¹, which occurred following 4–7 stages.

Approximately 10 min after completing the lactate profile test, participants performed a ramp-style \(V\)O_2max test. The test began at 1% gradient and at the speed 2 km h⁻¹ slower than the speed at which the participant had surpassed 4 mmol L⁻¹ [La⁻]. Speed increased 1 km h⁻¹ each minute for the next 2 min, when the speed at which > 4 mmol L⁻¹ [La⁻] had been reached. At this point, the gradient was increased by 1% each minute until volitional exhaustion. Participants were given consistent and loud encouragement during the test. When participants were re-tested at 3-Weeks, the starting speed and progression of the protocol was repeated to allow for measurement of time-to-exhaustion (TTE). HR and respiratory gas exchange were continuously measured. Capillary [La⁻] was measured 5 min from the time of exhaustion.

On a separate visit (between the hours of 9:00–11:00 h), participants rested in a supine position for a minimum of 10 min, at which time resting HR was assessed and venous blood was drawn from the antecubital vein into a K2EDTA-coated vacutainer for measurement of plasma VEGF and an uncoated vacutainer for measurement of serum EPO. Vacutainers were kept on ice (plasma) or at room temperature (serum) for 45 min before centrifugation (4 °C, 10,000 rpm, 10 min) (Heraeus Multifuge X1R, Thermo Scientific, Loughborough, UK) and were frozen at − 80 °C until analysis. Plasma VEGF (Human VEGF DuoSet ELISA; R&D Systems, Abingdon, UK) and serum EPO (Human Erythropoietin DuoSet ELISA; R&D Systems, Abingdon, UK) were determined with ELISA kits.

T_rec was measured using a rectal thermistor inserted 10 cm past the anal sphincter (Mon-a-Therm, Covidien, Mansfield, MA, United States). Weighted mean skin temperature was recorded using skin thermistors (Squirrel Thermal Couples, Grant Instruments, Cambridge, UK) attached to four sites on the left side of the body: pectoralis major (T_chest), biceps brachii (T_arm), rectus femoris (T_thigh), and gastrocnemius lateral head (T_lowerleg). Skin and rectal temperatures were continuously logged at 30-s intervals (Squirrel 2020 series, Eltek, Ltd., United Kingdom). HR (Polar Electro, Kempele, Finland) was recorded continuously (sampling rate of 1 Hz) on the Polar Beat application (Polar Beat, Kempele, Finland).

Active sweat glands were quantified using a modified-iodine paper technique with blinded computer aided analysis (Gagnon et al. 2012). Samples were collected by the same researcher from the dorsal side of the thickest segment of the forearm and mid-scapula on the upper back. Training session RPE was measured using a 1–10 point scale (RPE_1–10; ‘very light session’ to ‘max effort session’) and RPE during HTTs was measured using the 6–20 point Borg Scale (RPE_6–20; Borg 1982). Thermal sensation and thermal comfort were measured using modified 13-point and 10-point scales, respectively (Gagge et al. 1967). Measures of [La⁻] were taken from a fingertip capillary sample and immediately analysed using a Biosen C-Line Lactate analyser (EKF Diagnostics, Penarth, UK), which was quality checked each day and calibrated every 60 min. Respiratory gas exchange was sampled breath-by-breath using open-circuit spirometry (Vyntus CPX, Jaeger, Wuerzberg, Germany).

Peak T_rec was the highest T_rec value measured during the HTT. T_recRISE was calculated as the change in T_rec from 0–30 min during the HTT. T_sk was calculated as a weighted average according to Ramanathan (1964):

Estimated sweat loss was calculated as the difference between Pre- and post-HTT nude body mass. Running speed at 4 mmol L⁻¹ [La⁻] was determined using a custom Matlab script (The Mathworks Inc, Natick, USA) to fit a third-order polynomial curve to each individual dataset and interpolate the running speed at 4 mmol L⁻¹. Respiratory gas exchange data were exported as 5-s values and used to calculate oxygen consumption, running economy and respiratory exchange ratio (RER) during the lactate profile and \(V\)O_2max tests. Running economy (mL kg⁻¹ km⁻¹) was calculated as oxygen consumption (mL kg⁻¹ min⁻¹) divided by treadmill speed (km h⁻¹) (Barnes and Kilding 2015). During the lactate profile test, submaximal RER, running economy and HR data were averaged across the final minute of the stage where participants exceeded 4 mmol L⁻¹ [La⁻] at baseline, with the same running speed compared at 3-Weeks. \(V\)O_2max was calculated as the highest rolling 30-s average attained during the test. Successful attainment of \(V\)O_2max required meeting two of the following three criteria: (1) [La⁻] ≥ 8 mmol L⁻¹ (Howley et al. 1995), (2) RER ≥ 1.10 (Edvardsen et al. 2014), (3) maximal HR ≥ 90% of age-predicted maximal HR (220 − age). Additionally, a plateau was confirmed both visually and systematically by a 5-s average value ≥ 2 standard deviations lower than the linear forecast of \(V\)O₂ increase. For individual T_sk sites, if a time-point was lost, the missing data were modelled according to the slope/pattern of other data points from that participant during the previous test. Data at multiple T_sk sites were lost for two SAUNA group participants in the HTTs at 7-Weeks, and therefore SAUNA T_sk at 7-Weeks (n = 4) was deemed incomplete and not reported herein. Training data and resting HR were not collected in n = 1 SAUNA participant for both 3- and 7-Weeks.

All data were analysed using SPSS statistical software (SPSS version 25.0.0, SPSS, Chicago, IL, United States). To address the main aims, physiological data from HTT, lactate profile and \(V\)O_2max tests were compared between SAUNA and CON groups using a one-way analysis of co-variance (ANCOVA), with change scores (∆; difference between Pre and 3-Weeks) entered as the dependent variable and Pre-intervention absolute data entered as a co-variate. This allowed all changes to be covariate-adjusted for the baseline values (Vickers and Altman 2001) and yielded a single p value to represent the between-group comparison. This p value reflects the effect of the intervention. Because sex affects cardiorespiratory fitness, sex as a variable was tested for violations related to collinearity and entered as co-variate in analyses of \(V\)O_2max and running speed at 4 mmol L⁻¹ [La⁻]. Effect sizes derived from between-group comparisons are presented as Cohen’s d, whereby d > 0.8 are considered large, 0.5–0.8 moderate and 0.2–0.5 small. Normality of these data was assessed using Levene’s test of equality of error variances, which was not violated on any occasion.

To address the subsidiary aim, a one-way repeated-measures analysis of variance (ANOVA) was used to compare HTT physiological data from the SAUNA group participants at Pre, 3-Weeks and 7-Weeks (time [3 levels]).

Normality of data analysed by ANOVA was assessed using Mauchly’s test of sphericity, and Greenhouse–Geisser corrections were applied where assumptions of sphericity were violated. When a significant main effect was found, Dunn-Bonferroni-corrected post hoc comparisons were made. Main effect sizes for ANOVAs were calculated using partial eta-squared (η_p²), whereby η_p² > 0.14 are considered large, 0.06–0.14 moderate and < 0.06 small (Cohen 1988).

Ordinal data, including average running session RPE_1–10, and peak RPE_6–20, thermal comfort and thermal sensation during HTTs, were compared between groups (3-Weeks) using a Mann–Whitney U test or within-group (7-Weeks) using a Friedman’s test with post hoc analysis by Wilcoxon sign-rank tests.

3-Weeks: Training sessions were performed 7 ± 2 times per week (SAUNA: 6 ± 1; CON: 7 ± 2) in the weeks between Pre and 3-Weeks testing. Average weekly running distances were 53.3 ± 23.3 km and 54.5 ± 28.2 km in SAUNA and CON, respectively. Training was not different in type, frequency, running distance or perceived exertion between the SAUNA and CON groups at 3-Weeks (all p > 0.05; Fig. 3).

7-Weeks: Participants in the SAUNA group trained 6 ± 1 times per week, and ran 59.1 ± 21.8 km week⁻¹_. Training was not different in type, frequency, running distance or perceived exertion across the 7 weeks prior to 7-Weeks testing in the SAUNA group (all p > 0.05).

3-Weeks: Participants in the SAUNA group attended sauna sessions 3 ± 1 times per week, accumulating 9 ± 1 sauna sessions before completing the lactate profile and \(V\)O_2max tests, and 10 ± 1 sauna sessions before completing the HTT. Participants remained in the sauna for 28 ± 2 min each session, totalling 290 ± 48 min of sauna exposure. Peak HR reached 127 ± 10 beats min⁻¹ whilst in the sauna.

7-Weeks: Participants in the SAUNA group attended sauna sessions 3 ± 1 times per week, accumulating 19 ± 1 sauna sessions before completing the final HTT at 7-Weeks. These participants remained in the sauna for 28 ± 1 min each session, totalling 528 ± 29 min of sauna exposure. Peak HR reached 120 ± 14 beats min⁻¹ whilst in the sauna. Frequency, duration, and peak HR were not different between the sauna sessions completed before 3-Weeks testing as compared to those completed between 3-Weeks and 7-Weeks testing (all p > 0.05).

Female participants completed all experimental trials at Pre, 3-Weeks and 7-Weeks 28 ± 2 days apart. Normally menstruating participants (n = 8) reported menstrual cycles ranging between 26–31 days and five completed all tests in the same phase of their menstrual cycle. Of the remaining three, one SAUNA participant completed all Pre tests in the follicular phase and all 3-Weeks tests in the luteal phase, one SAUNA participant completed temperate exercise tests at Pre in the luteal phase and at 3-Weeks in the follicular phase (though HTTs were completed in the same phase), one CON participant completed temperate exercise tests at Pre in the follicular phase and at 3-Weeks in the luteal phase (HTTs completed in the same phase). Participants using oral contraceptives (n = 2) completed all tests in the active pill-taking phase.

3-Weeks: Changes in resting HR were not different between groups at 3-Weeks as compared to Pre (SAUNA: 58 ± 6 vs 55 ± 9 beats min⁻¹; CON: 54 ± 6 vs 51 ± 4 beats min⁻¹, at Pre vs 3-Weeks, respectively; p = 0.90, d = 0.06). Changes in body mass were not different between groups (SAUNA: 61.0 ± 6.3 vs 61.2 ± 6.4 kg; CON: 63.8 ± 5.5 vs 63.5 ± 5.4 kg, at Pre vp 3-Weeks, respectively; p = 0.43, d = 0.45). Serum EPO was detectable in 17 out of 20 samples (SAUNA, n = 12; CON, n = 5). Changes in resting serum EPO were not different (p = 0.82, d = 0.24) between groups (SAUNA: 3.8 ± 2.7 vs 3.1 ± 1.6 mIU mL⁻¹; CON: 2.8 ± 2.3 vs 2.6 ± 1.9 mIU mL⁻¹, at Pre vs 3-Weeks, respectively). VEGF was detectable in 10 out of 20 samples (SAUNA, n = 5; CON, n = 5). Changes in resting plasma VEGF were not different (p = 0.09, d = 0.35) between groups (SAUNA: 128.5 ± 121.9 vs 117.9 ± 106.4 pg mL⁻¹; CON: 183.0 ± 234.4 vs 179.8 ± 274.7 pg mL⁻¹, at Pre vs 3-Weeks, respectively).

7-Weeks: Participants in the SAUNA group showed a trend (p = 0.06) for a decrease in resting HR across time, though this was not significant (57 ± 6, 53 ± 10 and 53 ± 7 beats min⁻¹ at Pre, 3-Weeks and 7-Weeks, respectively). Serum EPO did not significantly change across time (3.6 ± 1.6, 3.1 ± 1.8 and 4.4 ± 2.8 mIU mL⁻¹ at Pre, 3-Weeks and 7-Weeks, respectively; p = 0.18). VEGF was detectable in n = 3 samples and therefore not statistically compared (77.9 ± 55.6, 77.5 ± 39.4 and 76.1 ± 51.6 pg mL⁻¹ at Pre, 3-Weeks and 7-Weeks, respectively).

Physiological responses during the lactate profile test at Pre and 3-Weeks are detailed in Table 2. The SAUNA group exhibited a 4 ± 3% improvement in running speed at 4 mmol L⁻¹ [La⁻], equating to a 0.6 km h⁻¹ [0.1, 1.0] greater improvement than the CON group (0 ± 3%) at 3-Weeks as compared to Pre (p = 0.01, d = 1.19; Fig. 5b).

During the lactate profile test, there was a trend (p = 0.06; d = 1.40) for a greater reduction in submaximal HR in the SAUNA group than in the CON group at 3-Weeks as compared to Pre (− 7 beats min⁻¹ [0, − 14]). Neither changes in submaximal running economy (p = 0.73, d = 0.27) nor RER (p = 0.14, d = 0.30) were different between groups.

The SAUNA group exhibited an 8 ± 12% improvement in \(V\)O_2max, equating to a 0.27 L min⁻¹ [0.05, 0.49] greater improvement than the CON group (+ 2 ± 8%) at 3-Weeks as compared to Pre (p = 0.02, d = 0.69; Fig. 5a; Table 2). Similarly, the SAUNA group exhibited a 12 ± 6% improvement in TTE during the \(V\)O_2max test, equating to a 54 s [17, 90] greater improvement than the CON group (− 1 ± 11%) at 3-Weeks as compared to Pre (p < 0.01, d = 1.56; Fig. 5c; Table 2).

The mean 0.3 °C reduction in peak T_rec observed in the SAUNA group at 3-Weeks surpasses the coefficient of variation of the HTT (0.34%, corresponding to 0.1 °C in the current dataset) as validated by Mee and colleagues (2015). Furthermore, the 0.2 °C difference in the change in peak T_rec between the SAUNA and the CON groups meets the physiologically meaningful threshold identified in a meta-analysis by Tyler and colleagues (2016). This reduction in peak T_rec in the SAUNA group may have resulted from an increased heat dissipation, possibly by means of increased skin blood flow combined with greater peripheral sweat gland activation to improve evaporative heat loss efficiency (Taylor 2014; Smith and Havenith 2019), given we did not observe increased whole-body sweat loss. The 0.8 °C greater reduction in peak T_sk observed in the SAUNA group at 3-Weeks as compared to the CON group is consistent with other heat acclimation interventions, as Tyler et al. (2016) reported a 0.5 ± 0.5 °C reduction in peak T_sk across 67 interventions. Similarly, the 11 beats min⁻¹ greater reduction in peak HR observed in the SAUNA group at 3-Weeks as compared to the CON group is consistent with other heat acclimation interventions, which report a 16 ± 6 beats min⁻¹ reduction in peak HR across 118 interventions (Tyler et al. 2016). This reduction in peak HR in the SAUNA group was likely mediated by increased plasma and/or blood volume, as was demonstrated using a similar post-exercise sauna bathing protocol by Scoon et al. (2007), though these variables were not measured in the current study. Though endurance training may (Nadel et al. 1974; Ichinose et al. 2009; untrained cohorts, 10 days to ~ 3 months training) or may not (McGarr et al. 2014; moderately-trained cohort, 2 week training) improve thermoregulation during exercise heat stress, the changes in T_rec, T_sk and HR in the SAUNA group were greater than those observed in the CON group. Overall, this indicates that the improvements observed in the SAUNA group during the HTT were as a result of the post-exercise sauna intervention as opposed to normal training. Therefore, intermittent post-exercise sauna bathing appears to induce thermoregulatory and cardiovascular adaptations that are superior to 3-weeks normal training, and of a similar magnitude to those observed in other heat acclimation protocols in the literature (Tyler et al. 2016). These improvements occurred despite using only ten 30-min exposures and did not cause any disruptions to normal training.

Sweat loss during the HTT differed between groups. In the CON group, there was a reduction in sweat loss during the HTT at 3-Weeks. Conversely, there was no change in sweat loss in the SAUNA group during the HTT at 3-Weeks, despite their lower body temperature (i.e., T_re and T_skin). This indicates a possible seasonal change in sweat rate for the CON group (Matsumoto et al. 1990), whereas the SAUNA group showed an enhanced evaporative heat loss for a given body temperature. Further, at 3-Weeks, the SAUNA group exhibited a substantially greater increase (+ 54%) in sweat gland activation on the forearm than the CON group (− 6%). Though sweat gland activation is not indicative of sweat gland output, these findings are supported by a recent study that showed a greater relative sweat output on the limbs following heat acclimation (Smith and Havenith 2019). Notably, the increase in sweat gland activation on the forearm without increases in total sweat loss (as exhibited by the SAUNA group in the current study) may be a more efficient adaptation for athletes competing in hot-humid environments, as it would improve evaporative capacity without exacerbating dehydration (Alber-Wallerström and Holmér 1985).

Besides the further ~ 0.1 °C reduction in peak T_rec at 7-Weeks in the SAUNA group, physiological adaptations appeared to plateau by 3-Weeks, despite nearly doubling the participants’ total sauna exposures by 7-Weeks. This plateau is consistent with data from active heat acclimation interventions, whereby changes in heart rate and body temperatures are nearly maximised by ~ 7 days (Griefahn 1997; Periard et al. 2015). Interestingly, RPE_6–20 was only decreased at 7-Weeks. Whether this would coincide with the anticipated steady increase in self-paced exercise capacity in the heat with a long-term intervention (Periard et al. 2015; Racinais et al. 2015b) is unknown, as the HTT was completed at a fixed workload.

Our data demonstrate that 3 weeks of intermittent post-exercise sauna bathing is ergogenic for exercise performance in temperate conditions in well-trained participants. The training status of the athletes in the current study make these results especially meaningful. The 8% mean improvement in \(V\)O_2max observed in the current study (~ 6% more than the CON group) is comparable to the 5% observed by Lorenzo and colleagues (2010) when using a 10-day active heat acclimation protocol. Possible mechanisms facilitating this improvement in aerobic capacity with heat acclimation include a greater maximum cardiac output, as demonstrated by Lorenzo et al. (2010) in their study, or increases in total haemoglobin mass, which are possible in an intervention stretching over multiple weeks, as more recently demonstrated by Rønnestad et al. (2020). The iron supplements given to participants in the current study may have helped to facilitate improvements in oxygen carrying capacity, though participants were not given iron supplements by Rønnestad et al. (2020) or by Scoon et al. (2007), who both observed haematological adaptations following heat acclimation.

Three weeks of post-exercise sauna bathing also increased running speed at 4 mmol L⁻¹ [La⁻] by ~ 4%, equating to a 0.6 km h⁻¹ greater improvement than in the CON group. These results were similar to the 5% improvement in cycling power output at 4 mmol L⁻¹ [La⁻] reported by Lorenzo et al. (2010), and are supported by other observations of reduced lactate accumulation during exercise following heat acclimation (Young et al. 1985). Although this improvement in running speed at 4 mmol L⁻¹ [La⁻] was accompanied by a trend for a greater reduction in submaximal HR in the SAUNA group, running economy was not different during the lactate profile test at 3-Weeks. Thus, changes in [La⁻] may have been due to short-term adaptations such as a decreased rate of glycogenolysis (Febbraio et al. 1994; Kirwan et al. 1987), or due to better splanchnic perfusion (and thus improved lactate removal) and haemodilution via increased blood and plasma volume (Scoon et al. 2007; Lorenzo et al. 2010). Alternatively, the multi-week design of the current intervention might have allowed time for heat acclimation-induced mitochondrial biogenesis to manifest (Tamura et al. 2014; Hafen et al. 2018).

The SAUNA group’s ~ 12% increase in TTE following 3-Weeks post-exercise sauna bathing was superior to the CON group’s ~ 1% reduction in TTE. The mean improvement in the SAUNA group was lower than the 32% increase observed by Scoon et al. (2007), however there were important differences in the TTE tests used, making comparisons difficult. Scoon and colleagues implemented a ~ 15-min TTE test at the runners’ best 5-km race pace and 0% gradient. In the current study, the TTE test increased in either speed or gradient each minute and on average participants reached volitional exhaustion at ~ 8 min, but at a running speed at which they had reached 4 mmol L⁻¹ [La⁻] and at a 4% incline. Despite using a different style of TTE test, this study builds on evidence for the ergogenic benefits of post-exercise sauna bathing provided by Scoon and colleagues, with the additions of the gold-standard test of aerobic capacity, a robust submaximal exercise test and a larger, mixed-sex cohort.

Of the 37 participants who began the experiment, 19 athletes succumbed to an injury (ranging from minor injuries that prevented training for ~ 1 week, to more severe injuries such as stress fractures or sprains) and did not complete the study. This is consistent with data reported in a review by van Gent et al. (2007), whereby incidences of lower extremity running injuries ranged between 19 and 79% in distance runners, with the highest rates observed in athletes reporting weekly running distances and training frequencies similar to those in the current study (i.e., females running 48–63 km week⁻¹, males running 64+  km week⁻¹, and athletes training 6–7 days week⁻¹). For those who completed the intervention, training was consistent across the 7 weeks of post-exercise sauna bathing. Importantly, unlike active heat acclimation protocols that require changes to an athlete’s training to accommodate heat acclimation sessions, there were no differences in training between the CON group and SAUNA group in this study. This included similar perceived training intensity and training volume (i.e., frequency and type of sessions and weekly running distance) across the intervention period.

Serum EPO changes in the SAUNA group were not significantly different from the CON group at 3-Weeks, and did not significantly change in the SAUNA group by 7-Weeks. As EPO is inherently variable in response to environmental stimuli (Chapman 2013; Płoszczyca et al. 2018), additional data may be required to fully understand this effect. Furthermore, serum EPO reflects a balance between renal EPO production and EPO consumption in the bone marrow (Rusko et al. 2004), making it difficult to discern the influence of either production or consumption by measurement of circulating EPO levels alone. Although the subsequent haematological responses of either increased consumption or production of EPO (i.e., increased red blood cell production) would coincide with the improved aerobic capacity observed at 3-Weeks in the current study and the increased red cell volume observed by Scoon et al. (2007), we were unable to measure red blood cell mass in the current study. VEGF was largely undetectable across the study, as a number of participant samples fell outside the ELISA kit’s limits of detection for the measurement, providing little insight into any possible angiogenesis with post-exercise sauna bathing.

These results contribute to the growing body of research supporting favourable health and performance outcomes following passive heating interventions. The investigation of the efficacy of post-exercise sauna bathing to induce heat acclimation is especially timely as the Tokyo 2020 Olympic games (though postponed) are predicted to be the hottest Summer Olympic Games to date. We observed that heat acclimation was nearly maximised by three weeks of intermittent post-exercise sauna bathing and 7 weeks did not induce any further meaningful physiological improvements. Thus, athletes need only incorporate a post-exercise sauna regime for 3 weeks to achieve beneficial adaptations.

This study did not exclude participants based on menstrual cycle or hormonal contraceptive use, and instead included a random sample of female athletes in random phases of their menstrual cycle. The diverse menstrual cycle patterns and contraceptive choices observed in this cohort is representative of the female athlete population. Whilst menstrual cycle was self-reported, the study design successfully allowed for tests and re-tests to be in the same phase for most female participants, making it unlikely that results were skewed or biased by menstrual cycle.

Participants self-selected into the SAUNA or CON groups. However, participants in both groups were given performance feedback following temperate exercise tests, which they discussed with their coach and used to inform their training. Given the highly competitive nature of the athletes recruited and the training ethos of the running club they were recruited from, the SAUNA and CON groups were equally invested in their training and performance during the exercise tests. This is supported by the similar SAUNA and CON training profiles (intensity, frequency, type and duration). Furthermore, the objective nature of the \(V\)O_2max and lactate profile tests reduces the impact of extrinsic, motivational factors on temperate performance results.

Although alternative methods of lactate profile testing may be preferred for determining lactate threshold (i.e., maximal lactate steady-state, or the D-max method described by Cheng et al. 1992), these methods significantly underestimate ‘race pace’ for ~ 5-km races (Palmer et al. 1999; Chalmers et al. 2015). The training aims of the running club that participants were recruited from for the current study were largely centred around cross-country races (typically ranging between 4 and 6 km). Thus, running speed at 4 mmol L⁻¹ [La⁻], observed to closely reflect 5-km race pace (Fay et al. 1989), had important ecological relevance for this cohort. In addition, though the results of this study certainly imply an improved exercise capacity, the extent at which the observed adaptations would impact self-paced performance in either the heat or in temperate conditions is unknown.

Finally, the repeated-measures aspect of the two participants who undertook both the SAUNA and CON intervention are not statistically accounted for. However, as depicted in figures where individual data from the cross-over participant(s) are indicated, this exerted a negative bias for most measures. A sensitivity analysis was undertaken whereby the cross-over data for the two cross-over participants was removed (e.g., participants’ data from the intervention they completed first were only included, and data from their cross-over intervention were removed; see supplementary material).