Cold-Water Immersion and Lower Limb Muscle Oxygen Consumption as Measured by Near-Infrared Spectroscopy in Trained Endurance Athletes

Ten men (age = 23.4 ± 3.4 years, height = 1.8 ± 0.1 m, mass = 68.8 ± 10.7 kg) volunteered for this study. Mean adipose tissue thickness beneath the NIRS device was 6.7 ± 2.7 mm. All participants were long-distance runners competing for the same National Collegiate Athletic Association Division II team. They refrained from consuming caffeine and alcohol for 48 hours before each testing session. All individuals provided written informed consent, and the study was approved by the University of Essex research subcommittee.

A randomized crossover design was implemented to assess the influence of CWI on mVO₂. Each person was required to participate in 2 testing sessions, separated by at least 24 hours. Participants attended the university sports therapy center at the same time of day, where they were randomly assigned to 1 of 2 recovery interventions, 10°C or 15°C CWI, for 20 minutes. Before and after CWI, AO of the leg undergoing NIRS assessment was performed. An overview of the experimental protocol is shown in Figure 1.

The NIRS device was placed on the midportion of the medial gastrocnemius muscle belly. To ensure accuracy of repeat device placement, we drew an outline of the device on the limb using a surgical marker pen. Participants were instructed to maintain the outline during the testing period. Limb skinfold thickness was measured at the site of device attachment using Harpenden skinfold calipers (Baty International). The limb undergoing assessment was randomly assigned.

Participants wore shorts and sat in an upright position in an immersion bath (model S-85-S; Whitehall Manufacturing), which was filled with water to the level of the navel. Water temperature was monitored using a digital aquarium thermometer, maintained within ±0.3°C of 10°C or 15°C, and stirred every 2 minutes using an inbuilt whirlpool machine. Individuals were instructed to keep as still as possible and not activate their lower limb muscles during immersion. After immersion, they exited the recovery bath and quickly towel dried before transitioning to the post-CWI occlusion. This transition period took ≤120 seconds.

Arterial occlusions were performed for 5 minutes at a pressure of 250 mm Hg¹⁹  before (baseline) and after CWI. A pressure cuff (model SC10D; DE Hokanson, Inc) was attached to the participant’s limb just above the knee joint and inflated to the corresponding pressure. The occlusion protocol was identical for both testing sessions, whereby participants lay supine on a therapy plinth next to the CWI bath. The protocol consisted of 1 AO (baseline), followed by a 5-minute recovery before transitioning to CWI. Post-CWI, each person returned to the therapy plinth, and the procedure was repeated.

A portable NIRS device (model PortaMon; Artinis Medical Systems BV) was fixed with black adhesive tape over the marked measurement site to prevent signal contamination from external light sources. The device was made waterproof as previously described.²⁰  The same researcher (R.S.E.) attached each device and applied the adhesive tape for all testing procedures. To account for the external pressure applied to the muscle belly, we used a standardized approach.²¹  The PortaMon is a dual-wavelength, continuous-wave system that simultaneously uses the modified Beer-Lambert law and spatially resolved spectroscopy methods. Changes in tissue oxyhemoglobin (O₂Hb), deoxyhemoglobin (HHb), and tHb were measured using the difference in absorption characteristics of wavelengths at 760 and 850 nm. Values for O₂Hb, HHb, and tHb are reported as a concentration change from baseline (30-second averaging before each test with participants in the supine position) in micromolar units, using a differential path length factor of 4.0. Hemoglobin difference (Hb_diff) was calculated as follows: [Hb_diff] = [O₂Hb] − [HHb]. The TSI % was expressed as a percentage and calculated as follows: [O₂Hb]/([O₂Hb] + [HHb]) × 100. It is independent of the NIR photon path length in muscle tissue and was calculated using the spatially resolved spectroscopy method. Chromophore concentrations were detected using the farthest light-emitting diode (40 mm). Data were collected offline at a sampling rate of 10 Hz.

All results are reported as mean ± SD. Normality and outliers were verified using the Shapiro-Wilk test and by observing the normality plots and residual plots. Preimmersion data of participant 3 were identified as outliers for both temperatures. We calculated imputed estimates using the fully conditional specification method to account for this.²⁴  Mean values for O₂Hb, HHb, tHb, Hb_diff, and TSI % were determined for the 20-minute CWI and for 5-minute segments. Two-way repeated-measures analysis of variance was conducted for all NIRS measures obtained in each 5-minute segment of immersion; within-participant variables were time (5, 10, 15, and 20 minutes) and temperature (10°C and 15°C). When a main effect of time was present, post hoc analysis was performed using a Bonferroni correction factor. Effect sizes were calculated using partial eta squared (η_p²). Values were interpreted as small (η_p²  = 0.01), medium (η_p²  = 0.06), or large (η_p²  = 0.14) effect sizes. Two-way repeated-measures analysis of variance was computed for mVO₂; within-participant variables were time (pre- and post-CWI) and temperature (10°C and 15°C). Further post hoc analysis was conducted on mVO₂ values for pre- and post-CWI at 10°C and 15°C. We set the level of significance at P < .05. All statistical analysis was performed using SPSS (version 25; IBM Corp).

Representative examples of the O₂Hb, HHb, tHb, and Hb_diff traces and TSI % during the experimental protocol (10°C) are displayed in Figure 2A and B, respectively. A similar response was seen in the group data.

Pre- and post-AO elicited similar responses. At the onset of AO, a steady and progressive decrease in O₂Hb concomitant with an increase in HHb concentration was present throughout the 5-minute occlusion, which reflects depletion of local O₂ stores in the static compartment of blood. A more rapid decrease in Hb_diff concentration occurred at the onset of AO. The Hb_diff represents the difference in concentration between O₂Hb and HHb and therefore provides the extent of desaturation taking place during occlusion. The TSI % signal largely followed the same trend as the O₂Hb and Hb_diff, with large decreases in saturation during both pre- and post-AO and minimal change during the 20-minute immersion period. The NIRS trace responses were similar for 10°C and 15°C, with no main effects (P > .05) across all NIRS measures for temperature or temperature × time interactions. A main effect of time was noted for O₂Hb (F_3,27 = 14.227, P = .001, η_p²  = 0.613), HHb (F_3,27 = 5.749, P = .009, η_p²  = 0.344), tHb (F_3,27 = 24.786, P = .001, η_p²  = 0.734), and Hb_diff (F_3,27 = 3.894, P = .020, η_p²  = 0.302) during the immersion, whereby they decreased over the 20-minute immersion period. No effect of time was evident for TSI % (F_3,27 = 2.859, P = .056, η_p²  = 0.241).

A nadir or flattening was reached in the O₂Hb, HHb, and Hb_diff signals during AO, indicating complete desaturation (Figure 3). The overlapped O₂Hb, HHb, tHb, and Hb_diff signals pre- and post-AO for the 10°C immersion trial are shown in Figure 4. The rate of decrease in O₂Hb ([A] in Figure 3) and Hb_diff ([D] in Figure 3) concentrations was noticeably slower over the 5-minute occlusion post-AO (the slope value for the line, as calculated using linear regression, confirmed this). Similar end values were achieved. This may reflect a slower depletion of local O₂ stores during the post-AO. After cuff release of both AOs, rapid resaturation occurred and was seen as an increase in O₂Hb ([A] in Figure 3) and Hb_diff ([D] in Figure 3) and a concomitant decrease in HHb ([B] in Figure 3), which also displayed an attenuated response during post-AO. Restoration of all signals was demonstrated after 1 to 2 minutes.

Post hoc analysis revealed reductions in the last 5-minute period of immersion (15–20 minutes) versus earlier periods of immersion for O₂Hb (10–15 minutes, P = .005; 5–10 minutes, P = .03; and 0–5 minutes P = .007) and tHb (10–15 minutes, P = .001; 5–10 minutes, P = .002; and 0–5 minutes, P = .001). In addition, we observed reductions in tHb at 10 to 15 minutes (5–10 minutes, P = .040; 0–5 minutes, P = .008) and 5 to 10 minutes (0–5 minutes, P = .02) compared with earlier periods of immersion. The reductions in O₂Hb and tHb represented a change of about 2 to 4 μM (Table).

A main effect of time for mVO₂ pre- and post-CWI for both temperatures (F_1,9 = 27.7801, P = .001, η_p²  = 0.755) was observed: mVO₂ decreased as a result of CWI (Figure 5). We did not find a main effect of temperature (F_1,9 = 1.996, P = .19, η_p²  = 0.181) despite the large effect size. We observed a temperature × time interaction (F_1,13 = 13.302, P = .006, η_p²  = 0.592), with a large effect size.

The main effect for time can be seen in the decrease in mVO₂ from preimmersion to postimmersion (P = .001; Figure 5). Figure 5 also shows a decrease in both temperature conditions, and this decrease was bigger postimmersion at 10°C than 15°C. The slope of the line for preimmersion versus postimmersion mVO₂ was clearly steeper at 10°C than at 15°C, as implied by the effect size and difference of the time × temperature interaction effect (P = .006). The large main effect for time describes the decline in mVO₂ from preimmersion to postimmersion at both temperatures (ie, the interventions were effective across both temperatures). The magnitude of the interaction effect illustrates the difference that temperature had on the preimmersion to postimmersion decline between conditions. Together, the large main effect for time and large interaction effect showed that immersion at 10°C to 15°C reduced mVO₂ and immersion at 10°C resulted in a larger reduction in mVO₂ than immersion at 15°C.

We investigated the effect of 20-minute CWI at 10°C and 15°C on gastrocnemius mVO₂ measured by NIRS. To address the limitations of previous research, we combined NIRS measures of hemodynamic response with the AO procedure to more robustly quantify mVO₂ post-CWI. The main findings were that CWI decreased mVO₂ after 20-minute immersion at 10°C and 15°C. A greater reduction in mVO₂ was seen at the colder temperature (10°C). These results disagree with our hypothesis that CWI would cause a postimmersion reduction in mVO₂ at 10°C but not 15°C. However, the finding that a colder immersion temperature led to a larger reduction in mVO₂ was in line with our premise.

Muscle oxygen consumption was reduced after 20-minute CWI at 10°C and 15°C, indicating the interventions were effective across both temperatures. The reduction was greater at 10°C. This outcome suggests that CWI at 10°C and 15°C reduced O₂ use in the muscle, which was consistent with the mechanism of recovery by which CWI is proposed to prevent inflammation and secondary hypoxic injury.^25,26  To accurately quantify mVO₂, we performed AOs at baseline and post-CWI. These AOs temporarily arrest blood flow, creating a “static” compartment of blood in the muscle in which the linear decrease in O₂Hb or Hb_diff directly reflects the rate at which the preexisting O₂ store in the blood compartment is consumed.¹⁶  As such, this method has the advantage of quantifying the O₂ use element of the TSI % measurement while eliminating the influence of further O₂ supply. Researchers have proposed that the reduced tissue metabolism associated with CWI is mediated through reductions in tissue temperature.^5,27  While we did not measure tissue temperature, a reduction likely did occur, as authors using similar CWI protocols observed reductions in tissue temperature during and after immersion.^5,27  A 24% reduction in post-CWI mVO₂ was seen after immersion at 10°C, with a 13% reduction at 15°C, reflecting the influence of the 5°C temperature reduction on mVO₂. Changes in muscle blood volume (discussed later) also occurred with CWI at both 10°C and 15°C but only during the last 5 minutes of immersion. Therefore, an interaction between temperature and duration should be considered when interpreting muscle metabolic responses. Indeed, the greater reduction in mVO₂ at 10°C is an interesting finding and suggests a possible threshold temperature at which reductions in tissue metabolism occur and below which colder temperatures lead to greater reductions in metabolism. In fact, Wakabayashi et al²⁸  observed the effect of decreasing muscle forearm temperature (via cooling pads) on NIRS-derived mVO₂ and reported higher mVO₂ values at higher tissue temperatures. Yet focusing solely on immersion temperature is potentially misleading, as shown by the large interaction effect, because both time and temperature play roles in reducing tissue metabolism. The benefit at colder immersion temperatures likely relates to efficiency (ie, reduced immersion periods). However, warmer immersions for longer durations could result in the same reduction in metabolism. To the best of our knowledge, we are the first to use the AO technique with NIRS measurements to establish mVO₂ changes in response to different CWI temperatures. Our findings of reduced mVO₂ support earlier reports of reduced muscle metabolic activity.^7,8  Although an accurate quantification of muscle metabolic activity is certainly useful and extends prior results, the threshold or extent of reduced muscle metabolic activity required to aid recovery is still unknown. An optimal reduction threshold may exist, but any quantification is outside the constraints of our study.

The NIRS-derived measurements of muscle blood volume (ie, tHb) have shown blood volume reductions during^7,8  and after postexercise CWI,^7,29  with ΔtHb used as an indirect and proxy measure of mBF. Collectively, previous investigators^6,7,30  reported decreases in mBF during and after CWI. We saw an approximately 4-μM reduction in tHb during 20-minute CWI. Specifically, tHb was lower at 15 to 20 minutes during immersion compared with all other measured points, and the reduction in tHb was similar at both temperatures (10°C and 15°C). More substantial decreases in tHb (approximately 20 μM) have been noted over a 20-minute recovery period incorporating three 4-minute CWIs.³¹  The different experimental design might be a reason for the disparate findings. Our participants performed CWI in a rested rather than immediate postexercise state, which may more accurately typify the extent of tHb change. Large increases in muscle and skin blood flow are seen postexercise due to exercise hyperemia. As such, a reduction in tHb relative to postexercise values occurs during and after CWI.^7,8  Given that our participants did not exercise before CWI, the change in tHb would be expected to be smaller and provide a greater indication of blood volume changes seen when using CWI at 24 or 48 hours postexercise. During our CWI protocol, peak changes in tHb occurred within the last 5 minutes of immersion (15 to 20 minutes), with minimal changes observed earlier. Furthermore, the 5°C difference in immersion temperature did not influence the magnitude of tHb reduction. Our results, therefore, suggested that a small reduction in muscle blood volume (tHb) can be achieved in either 10°C or 15°C water after at least 15 minutes of immersion. These outcomes may have implications for selecting immersion durations, which currently range from 30 seconds to 30 minutes.¹⁷ 

Authors of a small number of studies have used the TSI % response during and after CWI, with ΔTSI % considered a proxy for changes in muscle metabolic activity.^7,8,13,29  We identified no changes in TSI % during the 20-minute CWI and marginal reductions in tHb at the end of immersion. No change in TSI %, despite a small reduction in local muscle blood volume (tHb), tentatively suggests decreased O₂ use during this period, possibly reflecting a reduction in muscle metabolic activity. This finding and interpretation have been reported previously.⁸  Other researchers⁷  detected no change in TSI % values during CWI despite a reduction in tHb, which the authors interpreted as a decrease in O₂ use because ΔtHb is considered a proxy for mBF. However, these and other interpretations should be viewed with caution, as the TSI % measurement is a combination of muscle O₂ supply and use, and thus inferences regarding use cannot be made with any certainty.

This study had several limitations. First, the portable NIRS device measures only a small (2–6 cm³) volume of muscle tissue. The values calculated for mVO₂ are assumed to represent the entire muscle, but given the restricted coverage of the gastrocnemius and its heterogeneity of fiber type and muscle metabolism,³²  the changes likely did not represent the entire muscle. To obtain representative estimations, multisite mVO₂ measurements should be assessed. Second, skin blood flow can influence the NIRS (O₂Hb and tHb) signals. For example, during whole-body heating, elevated skin blood flow influenced the NIRS signal,¹⁴  and it has been well documented that cooling reduced cutaneous blood flow.^5–7  However, the effects of cooling on the NIRS signal have not been fully established. Our measurements of mVO₂ used the Hb_diff signal, which is less influenced by skin blood flow and use of AO, which creates a closed loop system, to remove the influence of skin blood flow. Finally, the study sample was exclusively male and comprised well-trained long-distance runners; therefore, extrapolation to female and other sporting populations requires further research.

We demonstrated that the NIRS-occlusion approach can detect changes in mVO₂ pre- and post-CWI and could provide future investigators with a simple tool to assess muscle metabolic activity. Although our results demonstrated that CWI influenced mVO₂, the significance of a reduction due to CWI is yet to be established. A key finding was that mVO₂ decreased CWI at both temperatures after 20 minutes of immersion. This reduction was most pronounced after the colder immersion temperature. This indicated that colder immersion temperatures elicited a greater decrease in muscle metabolic activity.

The role of mVO₂ in postexercise recovery protocols must be established before the optimal immersion time or temperature for CWI can be determined. Indeed, the application of CWI as a recovery tool is still controversial, with evidence to suggest its regular use can ameliorate training-induced changes after resistance training, with more beneficial findings associated with endurance-based training.²  The NIRS-occlusion technique may offer further insight into muscle metabolic responses, beyond what is attainable from the NIRS primary signals.

We thank the participants for taking part in the study.