Sixteen strength trained males volunteered to take part in the study (Table 1). Participants were considered to be resistance-trained if they had performed ≥3 resistance sessions per week for ≥2 years with a minimum of one session per week including exercises that targeted the lower limbs (Buckner et al., 2017). Prior to any experimental procedures, written informed consent was obtained from all individual participants and the study conformed to the latest revision of the Declaration of Helsinki (World Medical Association, 2013). All participants completed a health screening questionnaire and were excluded from the study if the investigator deemed they were contraindicated to the study procedures. The study was granted ethical approval (application number: 1,686) by the London Sports Institute Ethics Sub-Committee at Middlesex University.

Using a between-subject design, participants were pair matched for baseline strength (1 repetition maximum [RM] back squat), and body composition (body surface area to body mass ratio) in line with previous research (Roberts et al., 2015) and assigned to HWI (n = 8) or passive recovery (PAS) (n = 8) groups. Previous research has shown the relationship between body surface area relative to body mass to be an important influencer on thermal and physiological responses to hydrotherapy (Stephens et al., 2014). Participants attended the laboratory on four occasions; during the first visit, anthropometric data was collected (height and body mass), before participants were familiarised with experimental procedures. Additionally, participants performed a strength assessment to determine a 6 RM for the exercise techniques used in the exercise session (back squat, front squat, good morning, Bulgarian split-squat). The second visit involved a body composition assessment prior to a further familiarisation with experimental procedures. Participants then performed a strength assessment to determine a 1 RM for the back squat.

Visit 3 formed the start of the experimental period and required participants to perform baseline assessments in the following order: muscle soreness, blood sample, intramuscular temperature and maximal voluntary isometric contraction (MVIC). Following baseline assessments, participants performed a resistance exercise session and then completed either the HWI or PAS interventions. Intramuscular temperature assessments were repeated: post-exercise, post-intervention, 1 h and 2 h post-exercise. Muscle soreness, blood sample and MVIC were repeated at 2 h post-exercise. During visit 4, participants returned to the laboratory to perform 24 h post-exercise assessments for muscle soreness, blood sample and MVIC (Figure 1).

Prior to visit 3, participants were provided with a standardised meal (27 g porridge oats, 180 mL semi-skimmed milk, 200 g high protein yoghurt) to consume 2 h before arrival at the laboratory. Participants also consumed a ready to drink protein milk (30 g protein) following completion of the exercise session and after the 2 h post-exercise assessments, between which they were required to be fasted (Roberts et al., 2015). Additionally, participants were asked to refrain from food consumption in the 2 hours prior to any testing procedures. Aside from the control measures, participants were instructed to maintain their habitual dietary intake throughout the study. Participants were required to avoid the following throughout the study: exercise external to the protocol, any therapeutic interventions or nutritional supplements, alcohol and non-steroidal anti-inflammatory drugs.

Strength assessments were performed using a 6 RM testing protocol in accordance with recognised guidelines (Haff and Triplett, 2015). Following a standardised warm-up (Table 2), participants performed three sets of six repetitions of back squat with a progressively increasing load that corresponded to 50, 75% and 90% of their perceived 6 RM. Participants then performed sets of six repetitions with an increasing load for the determination of 6 RM with 2 min rest afforded between attempts. All 6 RM determinations were made within four attempts which were deemed successful by an investigator if a participant had reached a position in which the thigh was at least parallel to the floor. Participants repeated the above procedure for the determination of 6 RM on the front squat, good mornings and Bulgarian split-squat exercises with successful attempts determined by an investigator against standardised techniques. During the Bulgarian split-squat, participants placed the top of the toes of the trail leg on a 12 inch platform (McCurdy et al., 2010). The lead leg was placed approximately 39–45 inches from the front edge of the platform supporting the trail leg. Participants were then required to squat to a depth where the thigh of the lead leg was parallel to the ground before returning to the start position. These exercises were chosen to target a range of lower limb musculature and are commonly included in strength and conditioning programmes (Haff and Triplett, 2015). All resistance exercises were performed using free weights and a standard 20 kg bar (ELEIKO SPORT, Illinois, United States).

For the purposes of matching groups for baseline strength, maximal lower-body strength was assessed by 1 RM testing in the back squat consistent with recognized guidelines established by the National Strength and Conditioning Association (Baechle and Earle, 2008).

Following the standardised warm up (Table 2), participants performed three sets of six repetitions of back squat with a load corresponding to 50, 75% and 90% 6 RM. Participants then performed four sets of six repetitions with a load corresponding to 6 RM for the following exercises: back squat, front squat, good mornings and Bulgarian split-squat. The intensity (100% 6 RM or ∼85% 1 RM) and volume (12 sets targeting the quadriceps muscle group) of the session were selected based upon recommendations that loads of 80%–95% 1 RM elicit maximal gains in strength (Peterson et al., 2005), and hypertrophy (Fry, 2004). The performance of at least 8–10 weekly sets per muscle group has also been suggested to be required to maximise increases in muscle strength (Peterson et al., 2005) and size (Schoenfeld et al., 2016) in trained individuals. Participants were instructed to perform the eccentric phase of the exercises in a controlled fashion lasting approximately 2 s, whilst the concentric phase was to be performed with maximal acceleration. This method of lifting was chosen given the suggestion that it is the intended rather than the actual velocity that determines the velocity-specific training response (Behm and Sale, 1993). Two minutes rest was afforded between sets and exercises, which has been recommended as a minimum for maximising gains in muscle size (Schoenfeld et al., 2015).

The HWI or PAS interventions were performed within 10 min post-completion of the resistance exercise session. Participants in the HWI group sat in an inflatable bath (iSprint, iCoolsport, Miami, Australia) and were required to submerge their legs in the water up to their waist in a seated position (hip angle of ∼90°), with their legs outstretched and relaxed. Water temperature was maintained at 40°C using a circulatory heating unit (iCool dual temperature LITE, iCoolsport, Miami, Australia) and participants were required to remain immersed for 10 min. Participants in the PAS group sat on a physiotherapy bed and were required to remain still in a seated position (hip angle of ∼90°), with their legs outstretched and relaxed for 10 min. To avoid any external confounding influences on temperature, all participants were required to remain in the laboratory until after all the assessments were complete on visit 3.

In order to match groups for body surface area to body mass ratio, body composition was assessed by dual-energy x-ray absorptiometry (DXA; GE Lunar Prodigy, GE Healthcare, Bucks, United Kingdom) in accordance with previous research (Bell et al., 2016).

Intramuscular temperature was measured using a re-useable sterile needle thermistor (MKA08050-A, Ellab A/S, Rodovre, Denmark) with data read via a thermocouple system (E-Val Flex, Ellab A/S, Rodovre, Denmark). Previous studies have reported precision of 0.1°C for this system (Mohr et al., 2004). The needle thermistors were sterilised using an autoclave prior to each use according to manufacturer guidelines. The site of insertion was identified as the mid-point of the vastus lateralis muscle of the right limb, between the superior border of the patella and the inguinal fold and was marked using a pen for consistency between assessments with the site of insertion sterilised using a topical antiseptic (Betadine, Purdue Products LP, CT, United States). To ensure consistency in the depth of insertion, skin thickness and adipose tissue was measured using a caliper and a piece of medical tape was placed from the end of the needle at a distance which corresponded to 3 cm plus half of the skinfold thickness. The needle was then inserted into the vastus lateralis muscle until the tape contacted the skin surface, ensuring an intramuscular temperature of 3 cm. Once the reading had stabilised (approximately 2 s), the temperature was recorded. The needle was then manually removed to a depth of 2 cm and the temperature recorded once the reading had stabilised. This process was repeated for a depth of 1 cm prior to the needle being removed. Previous research has used this method to determine the effect of water immersion on intramuscular temperature following resistance exercise (Mawhinney et al., 2017).

Participants were seated on the dynamometer chair (Biodex 3, Biodex Medical Systems, NY, United States) with a hip joint angle of 90° and a knee joint angle of 70° (Eddens et al., 2017), set by the investigator using a goniometer. A knee joint angle of 70° has been shown to be sensitive to detect reduced muscle function following eccentric exercise, with no difference between this angle and the torque produced at 90° (McHugh and Tetro, 2003). Participants completed a standardised warm-up consisting of efforts at 50, 75% and 90% of perceived maximal force. Participants then performed three maximal voluntary isometric contractions (MVIC) of the right limb, each lasting 3 s, with standardised verbal instruction and encouragement provided throughout. Sixty seconds rest was afforded between attempts with peak force (N) recorded and the best attempt used for subsequent analysis.

Active muscle soreness was determined using a 200 mm visual analogue scale (VAS) with “no pain” indicated at one end and “pain/soreness as bad as it could be” at the other (Bell et al., 2014). Participants were instructed to stand with hands on hips and feet shoulder width apart prior to performing a squat to a depth whereby the thigh was parallel to the floor. Upon completion, participants indicated the pain felt in the lower limbs by drawing a line on the VAS.

Venous blood samples were collected using the venepuncture technique from a vein in the ante-cubital fossa region by a trained phlebotomist. Blood was collected into two 5 mL serum separator tubes and left to clot for 30–60 min prior to being centrifuged at 3,000 g, 23°C for 8 min. The serum was then removed and immediately stored in aliquots at −80°C for later analysis. Blood samples were analysed for markers of: muscle cell disruption (creatine kinase [CK]) and inflammatory processes (interleukin-6 [IL-6], high-sensitivity c-reactive protein [hsCRP], matrix metalloproteinase-9 [MMP-9]). The time of collection for all blood markers was based upon likely known time-course responses, and peak changes post-exercise (Table 3).

Serum CK and hs-CRP were determined by electrochemiluminescence using an automated analyser (Roche c702 chemistry module, Roche Diagnostics Ltd., United Kingdom). Serum IL-6 (Invitrogen Corporation, California, United States, MAN0006706) and MMP-9 (Thermo Scientific, Maryland, United States, BMS 2016-2 and BMS 2016-2TEN) were determined by an enzyme-linked immunosorbent assay (ELISA) using commercially available kits.

Raw data is reported as mean ± SD. All data (except soreness) were checked for normality using Shapiro-Wilk, which is recommended for sample sizes less than 30. Data identified as non-normal was log-transformed and all further data analysis was conducted on the transformed data. A mixed model ANOVA was used to analyse each dependent variable with a between-subjects factor of group (HWI and PAS) and a within-subject factor of time (baseline and post resistance exercise timepoints). Mauchley’s test of sphericity was used to assess the homogeneity of variance and, where necessary, Greenhouse-Geisser corrections were applied. Significant main and interaction effects were analysed using Bonferonni post hoc pairwise comparisons. Partial eta squared (ηp²) was used to indicate the effect sizes for main and interaction effects with ≥ 0.01, ≥ 0.059, ≥ 0.138 indicating small, moderate and large effects, respectively (Cohen, 1988). SPSS (IBM Corp., IBM SPSS Statistics for Windows, Version 28, Armonk, NY) was used for statistical analysis with a significance level of p < 0.05.

The mean ± 95% confidence interval (CI) for the differences between groups for changes between baseline and post measurement points were calculated (HWI minus PAS group). ‘Unclear’ was used to describe 95% CI’s that crossed 0. Cohen’s d effect sizes were used to determine the magnitude of the difference between the groups at each time point with < 0.20 indicating a trivial effect, 0.20—0.49 a small effect, 0.50—0.79 a medium effect, 0.80—1.19 a large effect, and ≥ 1.2 a very large effect (Cohen, 1988).

There was a significant main effect of time (F_{1.348, 18.870} = 13.005, P = <0.000, ηp² = 0.482), and post hoc pairwise comparisons determined significant decreases in MVIC between baseline-2 h, and −24 h post resistance exercise. There was no significant interaction effect (F_{1.348, 18.870} = 0.041, p = 0.904, ηp² = 0.003) or main effect of group (F_1,14 = 0.347, p = 0.565, ηp² = 0.024) (Figure 2A). Comparisons for the group difference in the change from baseline to 2 h post (−14 N, 95%CI [-51, 22], d = 0.4) and baseline to 24 h (−9 N, 95% CI [-32, 14], d = 0.4) were unclear.

There was a significant main effect of time (F_2,28 = 17.632, P = <0.001, ηp² = 0.557) with post hoc comparisons showing significant increases in soreness from baseline-2 h, and −24 h after resistance exercise. No significant interaction (F_2,28 = 1.287, p = 0.292, ηp² = 0.084) or group effect (F_1,14 = 2.064, p = 0.173, ηp² = 0.128) was determined (Figure 2B). Comparisons for the group difference in the change from baseline to 2 h post (−3 mm, 95% CI [-48, 42], d = 0.07) and baseline to 24 h (−32 mm, 95% CI [-84, 20], d = 0.7) were unclear.

Changes in MMP-9 were significant over time (F_2,28 = 33.777, P = <0.001, ηp² = 0.707) with post hoc comparisons identifying significant increases between baseline and 2 h post resistance exercise. There was no significant interaction (F_2,28 = 0.582, p = 0.565, ηp² = 0.040) or main (F_1,14 = 1.620, p = 0.224, ηp² = 0.104) effect of group. Comparisons for the group difference in the change from baseline to 2 h post (−98 ng mL⁻¹, 95% CI [-410, 215], d = 0.3) and baseline to 24 h (81 ng mL⁻¹, 95% CI [-209, 370], d = 0.3) were unclear.

Due to high baseline values in two participants (one per group) data were analysed for seven participants per group. There were significant increases between baseline and 24 h in CK (F_1,12 = 56.402, p = <0.001, ηp² = 0.825). There was no significant interaction effect (F_1,12 = 0.189, p = 0.672, ηp² = 0.015) or main effect of group (F_1,12 = 0.039, p = 0.848, ηp² = 0.003). There were no clear differences for group comparisons between baseline and 24 h (316 U.L⁻¹, 95% CI [-2,967, 3,599], d = 0.1).

Due to a high baseline value in one participant in the PAS group, data were analysed as seven (PAS) versus eight (HWI). There were no significant main effects of time (F_1,13 = 1.339, p = 0.268, ηp² = 0.093) or group (F_1,13 = 0.260, p = 0.618, ηp² = 0.020) and no significant interaction effect (F_1,13 = 2.312, p = 0.152, ηp² = 0.151). Group comparisons between baseline and 24 h were unclear (−0.8 mg.L⁻¹, 95% CI [-2.0, 0.4], d = 0.8).

Due to high baseline values in two participants (one per group) data were analysed for seven participants per group. There was a significant main effect of time (F_2,24 = 4.783, p = 0.018, ηp² = 0.285) with post hoc tests determining a significant increase between baseline and 2 h post. There was no significant interaction effect (F_2,24 = 0.093, p = 0.911, ηp² = 0.008) or main effect of group (F_1,12 = 1.286, p = 0.279, ηp² = 0.097). Effects were unclear for group comparisons between baseline and 2 h post (−0.7 pg mL⁻¹, 95%CI [-1.8, 0.4, d = −0.8), and baseline and 24 h (−0.3 pg mL⁻¹, 95%CI[-2.1, 1.5), d = −0.2).

There were significant main effects of time for intramuscular temperature (IMT) at 1 cm (F_{2.232, 31.250} = 9.144, p = <0.001, ηp² = 0.395), 2 cm (F_{1.866, 26.124} = 29.305, p = <0.001, ηp² = 0.677) and 3 cm (F_4,56 = 60.753, p = <0.001, ηp² = 0.813). Post-exercise resulted in the highest IMT at all depths, with significant increases at this point from baseline at 2 cm and 3 cm, and significant decreases from post-exercise to post-intervention (2 and 3 cm), 1 h post (2 and 3 cm) and 2 h post (all depths). Significant increases from baseline to intervention were only observed at 3 cm.

There were main effects of group at all IMT depths (1 cm: F_1,14 = 9.758, p = .007, ηp² = 0.411; 2 cm: F_1,14 = 11.866, p = 0.004, ηp² = 0.459; 3 cm: F_1,14 = 10.321, p = 0.006, ηp² = 0.424) with HWI eliciting higher intramuscular temperatures versus PAS.

There were no significant interaction effects at 1 cm (F_{2.232, 31.250} = 1.745, p = 0.188, ηp² square = 0.111) or 2 cm (F_{1.866, 26.124} = 2.735, p = 0.087, ηp² = 0.163). However, at 3 cm there was a significant interaction effect (F_4,56 = 3.419, p = 0.014, ηp² = 0.196). Post-hoc tests showed that at both post-intervention and 1 h post intramuscular temperature in HWI was significantly greater than PAS, furthermore between baseline and post-intervention there was only a significant increase in HWI.

Comparisons for the group differences are shown in Table 4. All effects were unclear except between baseline and post-intervention HWI resulted in a greater increase in intramuscular temperature compared to PAS at all depths.