REF-1090 · Jin et al., 2026 · Frontiers in Sports and Active Living, · Paper · PubMed · Free full text

Physical recovery modalities in competitive and elite endurance athletes

Network meta-analysis of 16 RCTs (273 participants): no modality outperformed passive recovery on soreness, creatine kinase or sport-specific performance; compression garments do not improve performance (SMD 0.04, moderate certainty), a single-trial thermoneutral-immersion sham ranked as high as or above most active modalities, and 64% of network nodes rested on a single trial

A-fairly strong evidenceMeasured on highly trained to world-class athletesWritten before October 2026, being re-checked

The authors searched PubMed, Embase, CENTRAL and Web of Science to 1 April 2026 for RCTs of physical recovery modalities in competitive endurance athletes, fitting random-effects network meta-analyses for eight outcomes with passive recovery as the reference. Certainty was rated with CINeMA against a minimal important SMD of 0.5. Open-access full text read; numbers marked "(full text)" are not in the abstract.

Participants

Sixteen RCTs (273 participants); fourteen contributed to at least one network. Competitive, elite or professional endurance athletes; some cohorts adolescent (full text).

Protocol

Modalities (full text): cold-water immersion, whole-body cryotherapy, active recovery, superficial and deep massage, compression garments, intermittent pneumatic compression, electrical stimulation, far-infrared therapy; one trial had a thermoneutral-immersion sham.
Eight outcomes; primary: delayed-onset muscle soreness (DOMS), creatine kinase (CK), sport-specific performance. SMDs with 95% CIs and P-scores.

Key Results

Three of the 47 comparisons with passive recovery had intervals excluding the null, none of them robustly: two depended on nodes built from a single trial, and all three on networks that pooled measures that cannot be compared.
No modality outperformed passive recovery for DOMS (5 trials; I² 90.2%), creatine kinase (4 trials) or sport-specific performance (11 trials; I² 0%).
A single trial was the only support for 35 of 55 nodes (64%); limiting the analysis to confirmed elite participants meant four of the eight outcomes had no network that could be estimated.
On all three outcomes it entered, a thermoneutral-immersion sham ranked level with or above every active modality — flexibility, sport-specific performance, countermovement jump, not the primary trio (full text). In the body it ranked first for flexibility but third of eleven nodes for performance (behind whole-body cryotherapy), above massage, active recovery and compression; one trial in adolescent swimmers, no sham contrast significant (full text).
Of 47 comparisons, 31 had very low certainty and 15 low; the one finding with moderate certainty was that compression garments give no performance benefit (SMD = 0.04, with 95% CI −0.37 to 0.45).
Performance intervals for active recovery (SMD −0.01, 95% CI −0.39 to 0.36) and electrical stimulation (0.00, −0.40 to 0.40) also lie within the clinically unimportant range (full text). Cold-water immersion, three trials: SMD −0.27 (−0.70 to 0.16) for performance, not significant (full text).
Taking out two of the five trials reverses the muscle-soreness network.

Practical Application

"Do compression, cold water, massage or other recovery tools help me perform better in the next session?" A 2026 network meta-analysis of 16 RCTs in 273 competitive and elite endurance athletes found none shown to outperform simply resting; compression garments, with moderate certainty, do not improve sport-specific performance (SMD 0.04), and a sham bath (one trial) ranked as well as most of the real tools.
Use these tools for comfort if the athlete likes them, not as a performance lever: the trials cannot be blinded, and a sham doing as well as the modalities means expectancy cannot be separated from physiology.
For soreness and creatine kinase the evidence here is too thin to say either way; the broader meta-analyses in the base (REF-385, REF-387, REF-250) still describe soreness and marker effects, mostly in non-elite populations after damage.
Not from this paper; the base's own practice (SYNTH-70): easy movement, food and sleep come first; tools are an optional extra in race blocks.

Limitations

Small, unblinded trials; 64% single-trial nodes; several networks with no estimable heterogeneity.
The DOMS result depends on two trials whose values were estimated rather than extracted; with them removed it flips to large effects driven by one anomalous control arm — the authors trust neither (full text).
Networks pooled incommensurable measures (e.g. different lactate and jump constructs).
Some adolescent cohorts; restricting to confirmed elite participants left little.
How this entry was checked and changed
Why this level

SR/MA of controlled trials, but mostly small trials, 35 of 55 nodes resting on one trial, extreme heterogeneity for soreness and very low certainty for 31 of 47 comparisons (rubric: SR/MA with serious heterogeneity or mostly small trials = A-)