Independent or simultaneous lowering of core and skin temperature has no impact on self-paced intermittent running performance in hot conditions

Ten males (30.5 ± 5.8 years, 73.2 ± 14.5 kg, 176.9 ± 8.0 cm, 56.2 ± 6.6 ml/kg/min; mean ± SD), characterised as performance level 3 athletes (Pauw et al. 2013), volunteered to participate in this study. All participants trained regularly (minimum of five times per week) and were familiar with intermittent exercise (minimum exposure of intermittent exercise including, but not limited to, the following at least once per week; intermittent sports such as football and hockey, or interval training).

Prior to commencement of the study the University Health and Sciences Research Ethics Committee granted ethical approval (project code SH16170014-R) for all methods. All participants completed a health screen questionnaire and provided verbal and written consent to participate in the study. A repeated-measures design was used with each participant completing all interventions in a counter balanced order. Prior to the main experimental trials participants completed a graded exercise test to determine \(\dot{V}{\text{O}}_{{{\text{2max}}}}\) and a separate familiarisation session of the self-paced intermittent exercise protocol. On four separate occasions participants completed an experimental trial in a temperature-controlled room (34.4 ± 1.4 °C, 36.3 ± 4.6% RH). The ambient temperature and relative humidity were similar between conditions (p > 0.05).

Participants completed a 30-min pre-exercise resting period in the temperature-controlled room (34.4 ± 1.4 °C, 36.3 ± 4.6% RH), where they consumed either: (1) 7.5 g/kg of ice slurry (− 0.5 ± 0.4 °C) beverage (INT); (2) a control beverage of 7.5 g/kg of water (23.4 ± 0.2 °C) (CON); (3) wore a cooling garment whilst consuming a control beverage (23.4 ± 0.2 °C) (EXT); (4) wore a cooling garment whilst consuming the ice slurry beverage; (MIX). To ensure consistency across trials, the ice slurry was consumed in 3 equal aliquots of 2.5 g/kg/body mass every 10-min. All trial beverages contained a carbohydrate (CHO) solution (Robinson cordial, UK) to enhance palatability and were matched (0.75 g/kg of body mass) for fair comparison between trials. The cooling garment consisted of a cooling vest (Arctic Heat, Brisbane, QLD, Australia) covering the torso, worn over participants own T-shirt, and cooling towels (Frogg Toggs^®), covering both upper and lower arms. The ice vest was stored at − 80 °C and taken out of the freezer 30-min before application. The surface temperature of the ice vests when donned by the participant was approximately 0.17 °C. The cooling towels were soaked in water and rung out 10-min before application. After 30-min, participants completed a self-paced intermittent exercise test on a non-motorised treadmill.

Participant’s height (Seca, Birmingham, UK) and body mass (Sartorius CAH3G-150IG-H’, Sartorius, Bovenden, Germany) were recorded upon arrival at the laboratory. Participants completed a 5-min self-selected warm-up prior to completing the graded exercise test for the determination of \(\dot{V}{\text{O}}_{{{\text{2max}}}}\) on a motorised treadmill (h/p/cosmos mercury 4.0 h/p/cosmos sports & Medical gmbh, Nussdorf-Traunstein, Germany). Starting treadmill speed was determined from the self-selected warm-up and corresponded to a heart rate value of approximately 130 b/min. The exercise intensity increased every minute by increasing speed by 1 km/h every minute until a comfortable speed was reached and thereafter gradient was increased by 0.5% until volitional fatigue (Hamlin et al. 2012). Respiratory gases were continuously monitored throughout using an online gas analysis system (Cortex Biophysik Metalyzer, Germany) with 10-s averages being recorded. Heart rate (HR) was continuously monitored by telemetry using a HR monitor (Polar FT-1, Kempele, Finland) and rating of perceived exertion (RPE) was recorded during the last 15-s of each stage using the 6–20-point Borg Scale (Borg 1982). Blood lactate was determined from 20μL capillary blood samples (Biosen C-line, EKF Diagnostics) taken approximately 3-min post-test. The following criteria; a plateau in \(\dot{V}{\text{O}}_{{{\text{2}}}}\), HR ≥ 85% age predicted heart rate max, RER > 1.15, RPE ≥ 19, voluntary exhaustion, post blood lactate ≥ 8.0 mmol/L was used to determine \(\dot{V}{\text{O}}_{{{\text{2max}}}}\). This was in accordance with the criteria recommended by BASES (1997). If all criteria were not met then a V̇O_2peak value was recorded as the mean value from the final 30 s.

With at least 24-h separating the \(\dot{V}{\text{O}}_{{{\text{2max}}}}\) test, to ensure adequate test–retest reliability (Tofari et al. 2014), participants returned to the laboratory ( ~ 19 °C, 45% RH) to familiarise themselves with the intermittent protocol on the non-motorised Treadmill (Woodway Curve 3.0TM; Woodway, Inc., Waukesha, Wisconsin, USA).

A 46-min modified protocol derived from the 30-min self-paced intermitted protocol, validated by Tofari et al. 2014 was employed, which has a coefficient of variation (with 90% confidence intervals) from a test–retest (n = 3) of 2.9% (3.8–4.1) for total distance covered (km), 4.6% (3.3–8.1) for peak speed (km/h) and 3.8% (0.96–1.1) for sprint distance (km).

The protocol included a 2-min warm-up at a self-selected speed followed by a 46-min intermittent protocol to replicate the first half of a football game. This was based on a previously published protocol (Gerrett et al. 2017). The protocol consisted of three 15.5-min periods separated by a 71 s recovery period and included the following; 10 × 10 s sprints, 3 × 51 s run, 6 × 50 s jog, 6 × 30 s walk, and 2 × 30 s rest period. The high-intensity bouts (run and sprint) were always followed by a low-intensity bout (rest, walk or jog). Participants were informed that the performance indicators were the total distance covered and their average speed during each bout and their maximal speeds during ‘sprint’ and ‘run’ commands. Participants were instructed that ‘sprints’ should be completed at 100% effort, ‘runs’ at 75% effort, ‘jogs’ at 45% effort and ‘walks’ at 10% effort. A screen was placed at eye level and displayed the exercise instruction (e.g., WALK, JOG, RUN, and SPRINT), the total exercise time and a time indicator of when the next exercise instruction would appear. Information regarding speed and distance covered were concealed from view.

Participants were instructed to swallow a telemetric pill (Cortemp; HQ Inc., Palmetto, FL, USA) 7–8 h prior to each test session. On arrival at the laboratory a urine sample was collected prior to measurement of semi-nude body weight. Eight wireless T_sk sensors were attached and a HR monitor worn, after which, a resting fingertip blood lactate sample was taken along with a whole-body thermal sensation (modified ASHRAE scale 2005). Participants then sat for 30-min in temperature-controlled room (34.4 ± 1.4 °C, 36.3 ± 4.6% RH) and completed INT, CON, EXT or MIX, as described above. At 5-min intervals during the pre-exercise period T_gi, HR, and thermal sensation were recorded.

Within 5-min of consuming the final beverage and/or removing cooling garments, participants began the intermittent exercise protocol (as outlined above) in the same temperature-controlled room with a fan set at 1.3 m/s placed 75 cm in front of the participant directed towards the torso. At 5-min intervals T_gi, HR, RPE, and thermal sensation were recorded. During the two 71 s recovery period (mid 1 and mid 2) and at the end of the test (post) a blood lactate sample was taken from the fingertip. At the end of the exercise protocol, participants towel-dried themselves and were weighed semi-nude, prior to the collection of a post urine sample. All trials were completed at the same time of day ( ± 1 h), with at least 5 days separating trials. Participants were asked to replicate their diet prior to each visit and refrain from caffeine 12-hr and alcohol 24-hr preceding the trials.

Running speed and distance covered were recorded continuously (sampling at 100 Hz) during the 46-min intermittent protocol on a non-motorised treadmill. The average speed and total distance covered during each exercise profile (WALK, JOG, RUN, and SPRINT) were calculated. The peak speed achieved during the high-intensity activity profiles (RUN and SPRINT) were identified. The mean velocity in each stage (WALK, JOG, RUN, and SPRINT) was expressed as a percentage of the individuals peak velocity achieved during that trial. The aforementioned data were separated into the first, second, and third (15.5-min) periods. The total distance covered for the entire protocol was also calculated for each condition.

The telemetric pill was used to measure gastrointestinal temperature (T_gi), which was used as an indicator of core temperature. This was measured every 5 min during the pre-exercise and exercise period. A safety stop criteria was set at 39.5 °C but this was not exceeded by any participants in any conditions. Based on the recommendations by Hunt et al. (2017) all telemetric pills were checked against a reference mercury thermometer in a water bath and where appropriate corrections were applied to reduce any systematic bias. T_sk was measured at eight sites (forehead, chest, upper back, upper arm, forearm, hand, thigh, and calf) using iButtons^® (Maxim Integrated Products, Inc., Sunnyvale, California, USA). Occasionally some data (1 or 2 locations) were missing due to equipment errors so an unweighted mean T_sk was calculated. T_sk was continuously recorded at a sample rate of 1 per second with data averaged over a 5-min period. Mean body temperature (T_b) was calculated as 0.79 ×  T_gi  + 0.21 × mean T_sk (Colin et al. 1971).

HR was recorded using a wireless HR monitor (Polar FT-1), sampling every second and averaged every 5-min during rest and exercise. Thermal sensation was rated using a scale ranging from + 10 (extremely hot) to − 10 (extremely cold) with 0 indicating thermal neutrality (Ashrae 2005) and was recorded at 5-min intervals during rest and exercise. Participants were instructed to report the number (corresponding to the thermal sensation description) that best represented their whole-body thermal status at that moment in time and were not permitted to recall their previous sensation. RPE was recorded using the 6–20-point Borg Scale (Borg 1982) at 5-min intervals during exercise only. Participants were instructed to report the number (corresponding to the exertion description) that best represented how hard they felt they worked for the preceding 5-min period. TS and RPE were collected by the same researcher and accompanied by standardised instructions and the memory-anchoring procedure (Haile et al. 2014). Capillary blood samples (20 µL) were taken from the fingertip and analysed for blood-lactate concentration.

Urine samples were assessed for urine osmolality using a portable osmometer (Vitech Scientific, West Sussex). Changes in semi-nude body mass were used to estimate gross sweat loss (g) adjusted for fluid intake.

Before fitting statistical models, a data exploration was undertaken following a protocol to visualise and examine the data for outliers in both response and explanatory variables, homogeneity in the response variables, collinearity between explanatory variables, and the nature of the relationships between response variables and explanatory variables (see Zuur et al. 2009). No influential outliers were found and the data met the assumptions required for the chosen statistical models. Data exploration was performed using R statistical software (R Core Team 2017). Performance variables (distance covered, average speed and maximum speed, and percentage of peak speed) were investigated using linear mixed effects models. Condition (INT, CON, EXT or MIX), activity (WALK, JOG, RUN, SPRINT) and period (first, second, third) were treated as fixed effects. The model included an interaction term that allowed the quantification of the interaction between conditions, activities and periods, and a random term was used to estimate variation between participants whilst also controlling for the treatment and activity effects.

Changes in the physiological responses were used for the main part of the analysis and the average absolute data are presented in supplementary files. Generalised Additive Mixed Models (GAMMs) were fitted to the data for the various physiological responses to treatment and activity due to non-linear patterns in the temperature change-response variable identified at the data exploration stage. GAMMs incorporate non-parametric smooth terms with mixed effects (Wood 2017). Condition and time were treated as fixed effects, including a time × condition interaction term. A random term was used to estimate variation between participants whilst also controlling for the treatment and activity effects. The model was fitted to the data to test the effects of different pre-cooling treatments on changes in the physiological responses of the participants. Smooths were fitted as interaction terms with different treatments to investigate if patterns in physiological responses were significantly different between treatments. 95% CIs were used as a way of confirming significance during the trials whereby if the intervals do not overlap at all then a real effect has occurred (O’Brien and Yi 2016). All models were fitted using the gamm4 package (Wood and Scheipl 2017) in R. The assumptions for all models were checked by extracting residuals and plotting these with fitted values to check for non-linear patterns and homogeneity of variance. Residuals were also plotted against covariates used within the model as a further check for homogeneity and any potential patterns. Residual patterns were absent and homogeneity of variance satisfied.

Comparison between pre-cooling protocols was standarised using Cohens d effect sizes (ES) with the following descriptive criteria; an ES of < 0.2 is classified as ‘trivial’, 0.2–0.4 as ‘small’, 0.5–07 as ‘moderate’ and > 0.8 as a ‘large’ effect.

As expected T_gi did not change during pre-cooling for CON or EXT but by the end of pre-cooling T_gi had dropped by − 0.79 ± 0.9 °C and − 0.96 ± 0.8 °C for INT and MIX, respectively. The additive effect of combining ice-slurry ingestion with an external cooling garment resulted in a 0.17 °C reduction in T_gi in comparison to internal cooling alone. This suggests that the addition of an external cooling garment alongside internal cooling can provide a small but meaningful change in T_gi. However, after 15-min of self-paced intermittent exercise T_gi was similar between conditions. Unlike endurance performance studies where core temperature is usually higher at the end of exercise in the pre-cooled compared to controlled condition (Siegel et al. 2010, 2012), T_gi at the end of the intermittent exercise protocol in our study was similar between all conditions. This seems to be a common trait amongst studies that successfully lowered core temperature prior to intermittent exercise (Aldous et al. 2018; Duffield and Marino 2007; Gerrett et al. 2017).

Whilst we were able to successfully lower mean T_sk using the external cooling techniques (MIX and EXT), this only lasted somewhere between 10 and 20-min during exercise. Mean T_sk was similar after only 10-min of exercise during MIX compared to CON but had the longest effect (20 min) for EXT compared to INT. From our results and previous literature, it seems that with the start of intermittent-based exercise the benefits of external cooling on mean T_sk are short lived with convergence to a similar mean T_sk occurring within the first 10 min (Duffield and Marino 2007; Price et al. 2009; Skein et al. 2012). A common trend that appears across external cooling studies is that the rate of rise in T_sk is typically greater for an external cooling technique compared to a control or internal cooling technique (Castle et al. 2006; Duffield and Marino 2007; Minett et al. 2011; Stevens et al. 2017). This greater rate of change in T_sk will increase the firing rate of warm thermoreceptors; resulting in stronger thermal perceptual responses (de Dear et al. 1993; Zhang et al. 2004). This is further supported from our thermal sensation responses where we typically observed that the lower the thermal sensation score at the end of pre-cooling, the greater the increase (and larger ES) in thermal sensation upon the initiation of exercise. This increase in perception of warmth may bestow no performance benefit via behavioural thermoregulation (Schlader et al. 2011). Internal cooling has been reported to delay the onset time of thermo-effector responses (i.e. sweating/vasodilation starts later into exercise) leading to a more rapid accumulation of metabolic heat during the early phase of exercise despite a lower absolute core temperature (Jay and Morris 2018). Whilst no sudomotor or vasomotor responses were measured in the present study and we also observed no differences in GSL between conditions, we did observe a greater rise in T_gi for the two conditions that employed an internal cooling protocol (INT and MIX) compared to external only and control conditions. The external cooling techniques also resulted in a greater rate of change in T_sk for EXT and MIX conditions. Maintaining a lower T_sk is important to provide a more efficient heat transfer gradient from the core.

Aldous et al. (2018) maintained the effects of mixed-method pre-cooling (7.5 g/kg of ice slurry at − 1 °C and ice vest and towel to torso and arms and ice packs around the upper legs) on core temperature, T_sk and thermal sensation throughout the first half (45-min) of an intermittent protocol compared to a control condition. Whilst in our study, these thermophysiological responses were similar to the control within 10–15 min of exercise. We conducted our pre-cooling within the temperature-controlled room (34.4 ± 1.4 °C, 36.3 ± 4.6% RH), whereas Aldous et al. (2018) conducted all cooling manoeuvres in a temperate condition (18 ± 0.9 °C, 50.3 ± 4.7% RH) before moving into hot conditions for exercise (30.7 ± 0.3 °C, 50.9 ± 4.2% RH), which may have offered a prolonged cooling advantage. This suggests pre-cooling in an already cool room may be more effective than cooling in the same hot environment. The influence of the environmental conditions where pre-cooling techniques are administered requires clarification.

The role of T_sk on behavioural regulation has been documented in endurance-based exercise activities (Schlader et al. 2011) but not intermittent exercise. However, we speculated that by lowering both T_gi and T_sk a colder thermal sensation would be experienced which would benefit behavioural thermoregulation and allow for improved performance compared to a control condition. However, the benefits of lowering thermal sensation with our external cooling techniques were short-lived, converging within 10 min and having the greatest rate of change for MIX compared to CON and INT. Although MIX lowered both T_gi and mean T_sk, the thermal sensations were similar to EXT which lowered mean T_sk only. External cooling is therefore required to alter thermal sensation but combining it with an internal cooling technique did not seem to enhance thermal sensation in our study. Favourable thermal perceptions may need to be strong enough to attenuate performance decrements in the heat, as such, maintaining a lowered core temperature, mean T_sk and thermal sensation may be required for intermittent exercise performance in the heat. This warrants investigation.