FormalPara Key Points

Cold-water immersion (CWI) was more effective than active recovery, contrast water therapy and warm-water immersion for most outcomes, including reducing muscle soreness and improving muscular power.

Water temperature and exposure duration were rarely impactful effect moderators; however there was a dose–response effect of a lower temperature and shorter duration positively influencing the recovery of muscular power after CWI 24-h post-exercise when compared with active recovery.

Air cryotherapy was more effective than CWI for immediately recovering muscular power (1-h post-exercise) and for recovering muscular strength.

1 Introduction

High training and competition loads may induce acute physiological fatigue from which recovery is required to maximise athletic performance in training and competition [1]. As a result, numerous methods to accelerate recovery following training or exercise are commonly utilised with the aim of enhancing the effects of recovery to optimise future performance.

Common recovery methods include water immersion, cold air exposure, massage, and active recovery. Water immersion submerges the body (entire or partial) in cold water (8–20 °C, cold-water immersion [CWI]) [2,3,4,5], warm water (24–38 °C, warm-water immersion [WWI]) [3, 5, 6] or a combination of cold and warm temperatures (contrast water therapy [CWT]) [7,8,9] for durations ranging from 5 to 30 min [5, 10,11,12]. Cold air exposure (air cryotherapy) exposes athletes (either whole body or partial body) to air temperatures ranging from − 85 to − 140 °C for short durations (2.5–3 min) [2, 13,14,15,16,17]. Massage is manual manipulation of specific areas of the body using rubbing, stroking, and kneading techniques [18, 19]. Active recovery is the performance of low-intensity aerobic exercise following strenuous exercise [11, 20]. The mechanisms by which these recovery methods are proposed to accelerate recovery differ but are similarly associated with alterations in post-exercise swelling and oedema [21].

While there have been many reviews examining the effects of various recovery methods on a range of perceptual, physiological and performance outcomes, these reviews have typically only compared one recovery method with passive recovery (i.e., no specific recovery intervention) [22,23,24,25,26]. These reviews have also arbitrarily pooled crossover and parallel studies with no consideration for the statistical differences between study methodologies [22,23,24,25,26]. For example, many crossover studies only report mean values for each treatment group, with no consideration for within-participant differences; this oversight reduces the precision of the results [27]. In addition, reviews that have compared more than one recovery modality have typically only analysed one or two outcome variables (i.e., delayed-onset muscle soreness [DOMS] or physiological markers of muscle damage such as creatine kinase [CK]) [24, 28], and there have been limited comparisons of the effects of multiple recovery modalities on exercise performance [29,30,31]. Furthermore, the limited number of reviews that have compared multiple recovery modalities on subsequent exercise performance did not report heterogeneity or did not attempt to reduce heterogeneity through subgroup analysis [29,30,31]. These reviews also made recommendations based purely on the effect size of the outcome measure. However, when making recommendations in relation to efficacy of recovery protocols, the evidence presented should incorporate measures of heterogeneity, number of participants evaluated, and level of bias as recommended by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) method for grading evidence quality and strength of recommendations [32]. This level of scientific rigour has not been applied in previous reviews.

CWI is one of the most commonly used recovery methods by physically active individuals [21]. Therefore, this review applied GRADE criteria to compare the effects of CWI with other commonly used recovery methods on perceptual, physiological and exercise performance outcomes following strenuous exercise in physically active participants. Additionally, this review compared the time course of recovery and evaluated dose–response effects. Identifying protocols that aid recovery following strenuous exercise will inform appropriate prescription of recovery modalities for physically active individuals.

2 Methods

2.1 Design

This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for the reporting of systematic reviews and meta-analyses [33]. This review was prospectively registered with Open Science Framework (10.17605/OSF.IO/NGP7C). No amendments to the protocol occurred after registration, and no protocol was prepared.

2.2 Search Strategy and Selection Criteria

The MEDLINE, SPORTDiscus, Scopus, Web of Science, Cochrane Library, EmCare, and Embase databases were searched from inception until 12 September 2022 using the following search strategy, which was adapted for each database:

athlet* or sport* or exerci* or football* or soccer or hockey or basketball* or netball* or volleyball* or "track and field" or cycli* or running or runner* or swim* or handball or softball* or tennis or baseball or cross country or cricket or surf* or skiing or golf or hurdling or bicycling or boxing or gymnast* or martial arts or racquet sports or badminton or jogg* or walk* or weight lifting or lift* weights or weight?lift* or wrestling or resistance train* or endurance train* or interval train* or climb* or strength* train* or strength* program and (cold* or ice* or low* temp*) adj3 (bath* or hydrotherap* or immers* or submers* or submerg*)

Database search results were exported to Endnote© (version 20; Thomson-Reuters, Toronto, CA, USA) and then uploaded to Covidence© Systematic Review software (Veritas Health Innovations, Melbourne, VIC, Australia). All duplicates were removed before two reviewers independently screened titles and abstracts for eligibility (EM, SS). Full texts were obtained for the remaining articles and independently assessed for eligibility by two reviewers (EM, SS). Results from each reviewer were compared after each stage and any discrepancies were resolved by an independent reviewer (JB). Reference lists of all eligible studies and any previous systematic reviews were checked to identify any additional eligible studies that were not identified by the primary search.

Inclusion criteria were (1) peer-reviewed randomised controlled trials published in the English language; (2) participants were competing at any level of sporting competition or deemed physically active; (3) protocols that used CWI following an acute bout of strenuous exercise (defined by the authors as exercise that would induce muscle damage) with further immersions permitted to be completed on subsequent days; (4) used varying recovery modalities as the comparator intervention; and (5) outcome measures included recovery of exercise performance (flexibility, muscular strength [including maximal voluntary contractions or 1RM testing], muscular power [including jump performance, anaerobic power performance of < 10 s or sprint performance]) or physiological (CK) and perceptual markers of recovery (DOMS, perceived recovery). Studies were excluded if they used combined treatments that may confound CWI results (e.g., combining CWI with compression garments, CWI with active recovery, CWI with nutritional supplements), or utilised training interventions involving more than one session of exercise. Data published as theses or conference abstracts were excluded.

2.3 Risk of Bias

An assessment of methodological quality for the selected studies was undertaken using the Randomised Controlled Trial (RCT) checklist from the Scottish Intercollegiate Guidelines Network (SIGN) [34]. The SIGN RCT checklist was developed to ensure a balance between methodological quality and practicality of use for authors and was used in this review as it is specific to the design of the studies included. Before commencing assessment, definitions provided by SIGN were clarified by the review team. Two reviewers appraised each study based on these appraisal definitions (EM, SS). Any discrepancies were resolved by an independent reviewer (JDB). A grade of ‘yes’, ‘no’, ‘can’t say’ or ‘not applicable’ was issued for each appraisal item. ‘Yes’ and ‘not applicable’ answers were indicative of a lower risk of bias, therefore the total frequency of ‘yes’ and ‘not applicable’ was tallied to indicate overall methodological quality. Quality of the studies was labelled as ‘high quality’, ‘acceptable’, ‘low quality’ or ‘unacceptable’ [34].

2.4 Data Extraction

Data were extracted by one reviewer (EM) and entered in a standardised Microsoft Excel© spreadsheet (V2105, Microsoft Corporation, Redmond, WA, USA). These data were independently cross-checked by another reviewer (SS) and any discrepancies were resolved through discussion. Further information was sought from study authors if all information could not be obtained from the full-text article. The extracted information included publication details (author information, publication date, country of origin), study methodology (sample size, exercise intervention, study type, assessment measures, comparison intervention), participant information (age, sex, height, body mass, sport, training history), CWI protocol (temperature, duration, number of immersions, depth of immersion, body position during immersion, timing of immersion post-exercise), comparator recovery protocol (recovery method, type [if applicable], intensity [if applicable], temperature [if applicable], duration, body position during protocol [if applicable], multiple applications [if applicable]), and assessment measures (test, units, measurements at various timepoints, effect sizes, confidence intervals [CIs], p values).

2.5 Statistical Considerations

Standardised mean difference (SMD) with Hedges' g correction for positive bias was used to determine the effect sizes for comparing the effect of recovery modalities and facilitating data synthesis. For the purpose of this review, effect sizes were presented for each study and were considered trivial (SMD < 0.20), small (SMD 0.20–0.60), moderate (SMD 0.61–1.20), large (SMD 1.21–2.00) and very large (SMD > 2.00) [35]. Effect size precision was described using 95% CIs whenever sufficient information was provided by the study authors.

The metafor statistical package in R software (version 4.1.0, R Foundation for Statistical Computing, Vienna, Austria) was used to perform random-effects meta-analysis and meta-regression. Restricted maximum likelihood estimation was used for model fitting and the inverse variance method was used to weight the study effects. Separate analyses were performed for each recovery timepoint (1 h, 24 h, 48 h, 72 h, 96 h). The primary comparison was CWI compared with all other recovery methods. Subgroup analyses were undertaken for CWI compared with each specific type of other recovery method where possible. Water temperature and exposure duration were explored as potential continuous moderator variables. A unique identification number was assigned to each study and included as a random factor in the meta-analysis. Outcomes from studies that reported multiple CWI versus other recovery methods comparisons were assigned to the same study identification number due to the lack of independence of those observations.

Crossover studies were combined with parallel studies using the approach described by Elbourne and colleagues [27]. This approach required crossover studies to report a CI, standard error, or p value from a paired t test in addition to mean treatment effects or mean and standard deviation for each condition. Crossover studies that did not provide this information were still included if values could be estimated using information available from other included studies that considered the same outcomes and comparison conditions, as described by Elbourne and colleagues [27]. The most conservative estimate was used in all cases where estimation was required.

I2 statistics were used to explore statistical heterogeneity within each meta-analysis and indicated the consistency of effect sizes between the included studies [36]. Statistical heterogeneity was considered low (I2 < 25%), moderate (I2 = 25–49%) or high (I2 > 50%) [36]. The GRADE system was used to rate the overall quality of evidence synthesis as high, moderate, low, or very low [32]. Specifically, the quality rating was downgraded one level from high for each of the following limitations: total number of unique participants < 100 (imprecision), high statistical heterogeneity and more than 50% of the studies in the meta-analysis deemed to be low quality.

3 Results

3.1 Search Results

The database searches identified 6255 potential studies. Following the removal of duplicates and ineligible articles, 28 studies were included in the pooled meta-analyses, while 26 were included in the stratified meta-analyses. Two studies were included in the pooled meta-analysis only, as there were insufficient comparators in their recovery or outcome subgroup to be included in the stratified meta-analyses [37, 38]. Eleven studies were unable to be included in either meta-analysis due to a lack of comparators [39,40,41,42,43,44,45,46,47,48,49]. A complete overview of articles included in the review can be found in Table 1. A complete overview of the screening process can be found in Fig. 1.

Table 1 Overview of studies included in the review
Full size table
Fig. 1
figure 1

PRISMA flowchart for screening of articles

Full size image

3.2 Risk of Bias

Only one (3%) study was classified as being of high quality, 24 (86%) were classified as being of acceptable quality, and three (11%) studies were classified as being of low quality. The most common issues identified from the risk-of-bias analysis was that concealment of the treatment from the researchers was rarely completed, with only one study concealing treatment [51]. Randomisation of treatment groups was also poor for most studies, with only two studies adequately randomising participants [15, 51]. Individual results of the risk of bias separated by category can be found in Online Supplement 1.

3.3 Meta-Analysis of All Recovery Methods Compared with Cold-Water Immersion (CWI)

3.3.1 Pooled Effects on DOMS

CWI had a limited effect on the recovery of DOMS compared with the other recovery methods. At 1 h, there was a non-significant trivial effect in favour of CWI (GRADE = high) (Table 2). At 24 h (Fig. 2) and 48 h, there were small significant effects in favour of CWI (24 h GRADE = moderate; 48 h GRADE = high) (Table 2). At 72 h and 96 h, there were small and trivial non-significant effects, respectively, in favour of other recovery methods (GRADE = moderate) (Table 2).

Table 2 CWI compared with all other recovery methods meta-analysis summary
Full size table
Fig. 2
figure 2

Forest plot illustrating the influence of CWI compared to other recovery methods 24 hours post exercise on muscle soreness (stratified by recovery method). CI confidence interval, CWI cold water immersion, deg degrees, Ecc eccentric, Ex Mode exercise modality, HIT high intensity training, min minutes, SMD standardised mean difference

Full size image

3.3.2 Pooled Effects on Muscular Power

CWI had no effect on the recovery of power performance compared with other recovery methods. At 1 h and 24 h (Fig. 3), there were small and trivial non-significant effects, respectively, in favour of CWI (GRADE = moderate) (Table 2); at 48 h, there was a non-significant trivial effect that did not favour any recovery method (GRADE = moderate) (Table 2); and at 72 h, there was a trivial non-significant effect in favour of other recovery methods (GRADE = low) (Table 2).

Fig. 3
figure 3

Forest plot illustrating the influence of CWI compared to other recovery methods 24 hours post exercise on muscular power (stratified by recovery method). CI confidence interval, CWI cold water immersion, deg degrees, Ecc eccentric, Ex Mode exercise modality, HIT high intensity training, LED light emitting diode, min minutes, SMD standardised mean difference

Full size image

3.3.3 Pooled Effects on Strength

CWI had no effect on the recovery of strength compared with other recovery methods. There were small to large non-significant effects in favour of other recovery methods at all timepoints (1 h, 24 h, 48 h, GRADE = moderate; 72 h, GRADE = low) (Table 2, Fig. 4).

Fig. 4
figure 4

Forest plot illustrating the influence of CWI compared to other recovery methods 24 hours post exercise on muscular strength (stratified by recovery method). CI confidence interval, CWI cold water immersion, deg degrees, Ecc eccentric, Ex Mode exercise modality, min minutes, SMD standardised mean difference

Full size image

3.3.4 Pooled Effects on Perceived Recovery

CWI had no effect on perceptions of recovery compared with other recovery methods. At 24 h, there was a trivial non-significant effect in favour of CWI (GRADE = high) (Table 2, Fig. 5). At 48 h, there was a non-significant null effect that did not favour any recovery method (GRADE = moderate) (Table 2).

Fig. 5
figure 5

Forest plot illustrating the influence of CWI compared to other recovery methods 24 hours post exercise on perceived recovery (stratified by recovery method). CI confidence interval, CWI cold water immersion, deg degrees, Ecc eccentric, Ex Mode exercise modality, min minutes, SMD standardised mean difference

Full size image

3.3.5 Pooled Effects on Flexibility

CWI had no effect on the recovery of flexibility compared with other recovery methods. There was moderate to very large non-significant effects in favour of CWI at all timepoints (GRADE = low) (Table 2, Fig. 6).

Fig. 6
figure 6

Forest plot illustrating the influence of CWI compared to other recovery methods 24 hours post exercise on flexibility (stratified by recovery method). CI confidence interval, CWI cold water immersion, deg degrees Ecc eccentric, Ex Mode exercise modality, HIT high intensity training, min minutes, SMD standardised mean difference

Full size image

3.3.6 Pooled Effects on Creatine Kinase

CWI had a limited effect on reducing CK levels in the blood compared with other recovery methods. At 1 h, there was a small non-significant effect in favour of other methods (GRADE = low) (Table 2). At 24 h, there was a small effect with a minor degree of uncertainty in favour of CWI, as indicated by the 95% CI (effect size =  − 0.58 (− 1.17, 0.01); p = 0.06; GRADE = high) (Table 2, Fig. 7) and a small non-significant effect in favour of CWI at 48 h (GRADE = moderate) (Table 2).

Fig. 7
figure 7

Forest plot illustrating the influence of CWI compared to other recovery methods 24 hours post exercise on creatine kinase (stratified by recovery method). CI confidence interval, CWI cold water immersion, deg degrees, Ecc eccentric, Ex Mode exercise modality, LED light emitting diode, min minutes, SMD standardised mean difference

Full size image

3.3.7 Meta-Regression Outcome

The meta-regression run in conjunction with the pooled meta-analysis found that water temperature and exposure duration did not significantly moderate effects at any timepoint for any outcome measure.

3.4 Meta-Analysis Stratified by Recovery Intervention

3.4.1 CWI Compared with Active Recovery

CWI had limited effect on DOMS compared with active recovery. At 1 h, there was a small significant effect in favour of CWI (GRADE = moderate) (Table 3), and at 24 h (Fig. 2) and 48 h, there were trivial non-significant effects in favour of CWI (GRADE at both timepoints = moderate) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

Table 3 Meta-analysis summary stratified by recovery method
Full size table

CWI had no effect on promoting recovery of power performance compared with active recovery. At 24 h, there was a trivial non-significant effect in favour of CWI (GRADE = low) (Fig. 3, Table 3). There were significant moderating effects at 24 h of both water temperature and exposure duration, whereby for every 1-min increase in duration, the effect size decreased by 0.07 (− 0.12, − 0.01; p = 0.01) and for every 1° increase in temperature, the effect size decreased by 0.06 (− 0.10, − 0.01; p = 0.007). At 48 h, there was a trivial non-significant effect in favour of active recovery (GRADE = moderate) (Table 3).

CWI had no effect on feelings of perceived recovery compared with active recovery. At 24 h, there was a trivial non-significant effect in favour of CWI (GRADE = moderate) (Fig. 5, Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

3.4.2 CWI Compared with Contrast Water Therapy

CWI had limited effect on DOMS compared with CWT. At 1 h and 24 h (Fig. 2), there were trivial to small non-significant effects in favour of CWI (GRADE at both timepoints = moderate) (Table 3), and at 48 h, there was a small significant effect in favour of CWI (GRADE = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

CWI had no effect on promoting recovery of power performance compared with CWT. At 1 h, 24 h (Fig. 3) and 48 h, there were moderate non-significant effects in favour of CWI (1 h GRADE = low; GRADE for 24 h and 48 h timepoints = moderate) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

CWI had no effect on promoting recovery of flexibility compared with CWT. At 1 h, 24 h (Fig. 6) and 48 h, there were moderate to very large non-significant effects in favour of CWI (GRADE for all timepoints = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

3.4.3 CWI Compared with Warm-Water Immersion

CWI had no effect on DOMS compared with WWI. At 1 h, 24 h (Fig. 2) and 48 h, there were trivial to moderate non-significant effects in favour of CWI (1 h GRADE = moderate; 24 h and 48 h GRADE = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

CWI had no effect on promoting recovery of power performance compared with WWI. At 24 h, there was a small non-significant effect in favour of WWI (GRADE = low) (Fig. 3, Table 3), and at 48 h, there was a trivial non-significant effect in favour of CWI (GRADE = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at either timepoint.

CWI had no effect on promoting recovery of strength performance compared with WWI. At 1 h, there was a small non-significant effect in favour of WWI (GRADE = low) (Table 3), and at 24 h (Fig. 4) and 48 h, there were small to trivial non-significant effects in favour of CWI (GRADE = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

CWI had no effect on reducing CK concentration compared with WWI. At 1 h, there was a small non-significant effect in favour of WWI (GRADE = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects.

3.4.4 CWI Compared with Air Cryotherapy

CWI had no effect on DOMS compared with air cryotherapy. At 24 h, there was a trivial non-significant effect in favour of CWI (GRADE = high) (Fig. 2, Table 3), and at 48 h and 72 h, there were trivial to small non-significant effects in favour of air cryotherapy (48 h GRADE = moderate; 72 h GRADE = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

CWI had no effect on promoting recovery of power performance compared with air cryotherapy. At 1 h, there was a moderate significant effect in favour of air cryotherapy (GRADE = moderate) (Table 3). At 24 h (Fig. 3), 48 h and 72 h, there were trivial non-significant effects in favour of CWI (GRADE for 24 h and 48 h = moderate; GRADE for 72 h = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

CWI had no effect on promoting recovery of strength performance compared with air cryotherapy. At 1 h, 24 h (Fig. 4) and 48 h, there were large to very large non-significant effects in favour of air cryotherapy (GRADE at all time points = low) (Table 3). Water temperature and exposure duration did not significantly moderate effects at any timepoint.

3.4.5 CWI Compared with Massage

CWI had no effect on promoting recovery of power performance compared with massage. At 24 h, there was a trivial non-significant effect in favour of CWI (GRADE = low) (Fig. 3, Table 3). Water temperature and exposure duration did not significantly moderate effects.

4 Discussion

The aim of this review was to examine the efficacy of CWI for promoting recovery of performance, perceptual and physiological outcomes compared with commonly used recovery modalities following strenuous exercise. A second aim was to evaluate dose–response effects of water temperature and/or duration of exposure during CWI. The majority of findings favoured CWI compared with other recovery modalities, but few results reached statistical significance. CWI was more effective than other recovery modalities for improving DOMS at 1 h, 24 h and 48 h post-exercise. There was a dose–response effect of a lower temperature and shorter duration positively influencing the recovery of muscular power after CWI 24 h post-exercise when compared with active recovery, with shorter and colder exposures facilitating greater recovery. However, air cryotherapy was more effective than CWI in the recovery of muscular power performance 1 h post-exercise.

This is the first review to compare CWI with other recovery modalities and their effects on physiological, perceptual, and physical performance measures at specific time points following differing exercise interventions in physically active populations. Despite a more specific search strategy than previous reviews, this review identified a greater number of studies for inclusion in the analysis. This is also the first review to use meta-regression to evaluate dose–response effects of water temperature and/or exposure durations on outcome measures. Furthermore, this is the first review to account for methodological variations within parallel and crossover study designs to increase the precision of the results reported.

4.1 CWI as a Recovery Method

Pooled effects comparing CWI with all other recovery methods examined showed that CWI was as effective as, and sometimes superior to, other recovery methods for the recovery of many performance outcomes (Fig. 8). Coaches and athletes should strongly consider its use as part of their recovery process during competitive phases. However, it should be acknowledged that CWI may blunt training adaptations during preparation phases, particularly for resistance-based training programmes due to attenuated changes in the muscle [1, 57].

Fig. 8
figure 8

Summary of stratified review outcomes presented to allow practitioners easy interpretation. AR active recovery, CK creatine kinase, CWI cold water immersion, CWT contrast water therapy, DOMS delayed onset muscle soreness, h hours, PR perceived recovery, WWI warm water immersion

Full size image

Strength outcomes showed that there was no difference between CWI and other recovery methods. This finding is consistent with a previous meta-analysis comparing CWI with passive recovery, which found that CWI was ineffective for the recovery of strength performance [58]. Recovery methods that promote cooling may not be effective for the recovery of strength, as cooling the neuromuscular system may inhibit isometric strength (a measure most studies in this review used to indicate strength performance). However, dynamic power and strength performance (eccentric and concentric movements) are improved due to reduced neuromuscular fatigue [59].

CWI was more effective at reducing DOMS 24 and 48 h post-exercise compared with other methods. This may be due to the hydrostatic pressure of the water reducing swelling and inflammation, and colder temperatures having an analgesic effect [1, 21]. CWI was potentially more effective at reducing CK concentrations 24 h post-exercise when compared with other recovery modalities. Vasoconstriction of the blood vessels induced by the colder temperatures of CWI may contribute to this accelerated clearance [60]. The cooler temperatures also slow the delivery of inflammatory markers, which may reduce secondary tissue damage and lower inflammation [61].

4.2 Stratified Effects of CWI Compared with Other Recovery Methods

Low GRADE scores across many of the subgroup analyses indicate the limited amount of evidence available to substantiate results with a combination of low study numbers and subsequently low participant numbers, as well as high heterogeneity. An overview of the stratified results can be found in Fig. 8. More research is needed to be able to confirm the effects of each recovery method. Another consideration for recovery studies is the effect that belief in the method may have on the outcome measures. To account for placebo effects, researchers should include an additional group that receives a placebo condition that participants are led to believe is as effective as the intervention groups. Belief could also be quantified and used as a randomisation factor in parallel studies or as a covariate in crossover studies. The placebo effect and belief of recovery effects have been demonstrated successfully where participants receiving a sham recovery method recovered better than participants receiving accepted recovery methods [12].

4.2.1 The Effects of CWI Compared with Active Recovery

Active recovery is thought to accelerate the body’s return to homeostasis following strenuous exercise through the enhanced removal of blood lactate, restoring muscular energy supplies and reducing the severity of muscular injury and soreness [62]. However, CWI was found to be more effective at reducing DOMS 1 h post-exercise compared with active recovery. This could be due to the hydrostatic pressure during water immersion reducing swelling in the periphery, which may in turn reduce muscular pain. The analgesic effect of the cold temperature also lowers the activation threshold of tissue nociceptors and slows the conduction of nerve pain signals [63].

CWI had a limited effect on muscular power. However, meta-regression showed that decreasing the temperature and duration of CWI exposure was associated with greater recovery of muscular power compared with active recovery. This may be attributable to colder water temperatures reducing musculo-tendinous stiffness, which promotes performance of movements that utilise the stretch–shortening cycle [64].

4.2.2 CWI Compared with Contrast Water Therapy

CWT has been proposed to positively influence recovery due to the ‘pumping’ mechanism created by alternating between cold (promoting vasoconstriction) and hot (promoting vasodilation) temperatures [65,66,67]. The pumping mechanism may assist in the reduction of oedema, spasm and inflammation, as well as improve range of motion [65,66,67]. However, the present pooled recovery modality results for both muscular power and flexibility showed moderate to large effects across multiple time points in favour of CWI. These findings suggest that constant cold temperatures may be more effective at influencing recovery of performance than intermittent cold and hot temperatures, although these effects were not statistically significant.

For flexibility (only two to three studies included in each analysis), the 95% CIs were extremely wide, indicating that more RCTs comparing CWI and CWT are required. Muscular power analyses had more studies included (5–7 studies) and produced narrower 95% CIs with less overlap in favour of CWT. More studies would allow for the consideration of the moderating variables to indicate whether CWI is truly more effective than CWT.

CWI was also found to better influence recovery from muscle soreness over multiple timepoints compared with CWT, indicating that constant cold temperatures are more effective at reducing the oedema and inflammation that causes pain [21] than intermittent cold and hot temperatures [67].

4.2.3 CWI Compared with Warm-Water Immersion

WWI has been found to increase blood flow to the deep muscles (through vasodilation), which improves oxygen flow to these areas and may promote healing of the tissues post-exercise [68].

CWI was found to be more effective in reducing the effects of DOMS compared with WWI. The analgesic effect of the cold temperatures is likely to contribute to this; however it might also be attributable to a placebo effect due to athletes believing CWI will be more likely to improve recovery [69]. The null effects of the recovery of strength and power performance are in line with previous research where neither cold nor warm temperatures were more effective at recovering physical performance [70, 71]. The results of the present analysis suggested that WWI may be more effective at removing CK from the blood 1 h post-exercise compared with CWI. The warmer temperatures and vasodilation of the vessels may be more effective at clearing the metabolic byproducts than colder temperatures and vasoconstriction.

4.2.4 CWI Compared with Air Cryotherapy

Air cryotherapy was found to be more effective than CWI in some areas of performance recovery, but CWI was more effective for the recovery of DOMS within the first 24 h post-exercise. This suggests that the hydrostatic effect of CWI may be more important for recovery of DOMS than the impact of cold temperature.

Air cryotherapy is conducted at extreme temperatures (− 85 to − 110 °C) for short time periods (2.5–3 min exposure) and showed non-significant moderate to very large effects on the recovery of both strength and power. It is possible that the extreme cold conditions are beneficial to muscular performance; however previous studies have found that air cryotherapy increases muscular stiffness as well as stiffness in the connective tissues, which leads to increased risk of muscle damage [72] and decreased strength performance [64]. Therefore, caution should be used when considering the application of air cryotherapy for recovery of strength and power.

4.2.5 CWI Compared with Massage

CWI and massage as recovery methods have rarely been compared using RCT study designs despite a large amount of anecdotal evidence involving massage in particular [73]. CWI was found to have a trivial effect on power recovery; however evidence was graded as low due to high heterogeneity and low study numbers. Massage is believed to have the same benefits to athletes that CWI has, and includes reducing swelling and pain in the muscle, as well as enhancing the clearance of metabolic byproducts [74]. However, despite the similar benefits, the mechanism of each modality differs. Massage relies on biomechanical mechanisms where pressure exerted on the tissues reduces passive and active stiffness and increases range of motion, which positively influences athletic performance [75]. Massage also increases blood flow to the skin and muscles and increases the release of relaxation hormones, which assists with decreasing pain and perceived fatigue [29, 75]. CWI uses the analgesic effects of the cold temperature to decrease pain [63], but the cold temperature also reduces musculo-tendinous stiffness, which positively influences athletic performance [64].

4.3 Limitations and Future Research

This review was influenced by some limitations. First, the low study numbers (with small sample sizes) and high heterogeneity has led to several low GRADE scores. Second, some studies were unable to be included in the meta-analyses due to differences in study designs, without adequate reporting of data that would allow pooling across designs. Finally, low study numbers also impacted the ability to distinguish between eccentric exercise recovery and high-intensity exercise recovery, and therefore specific recovery recommendations were unable to be discerned using these data. To improve this, more high-quality research (with larger sample sizes) is required comparing recovery modalities on athlete recovery.

5 Conclusion

As a recovery method, CWI is as effective as other recovery modalities for recovery following strenuous exercise in physically active individuals. CWI was more effective than active recovery, CWT and WWI for most outcomes, including reducing DOMS and improving muscular power. Air cryotherapy was more effective than CWI for immediately recovering muscular power (1 h post-exercise) and for recovering muscular strength.