REF-1706 · Hou, Yin & Qiao, 2026 · Healthcare · Paper · PubMed · Free full text

After a match or hard team-sport session, recovery tools helped mostly in the first 24 h and faded by 48-72 h: in a network meta-analysis of 22 small randomised trials (about 417 mostly male team-sport players), active recovery (SMD 0.94, 95% CI 0.20-1.69) and cold-water immersion (0.45, 0.12-0.77) restored jump height at 24 h better than passive rest, massage and cold water lowered soreness at 0 and 24 h (massage SMD −2.62 and cold water −2.19 at 24 h), and only cold water still lowered soreness at 48 h (−1.57); no tool helped jump height or CK at 48-72 h. No certainty rating was possible, soreness and CK heterogeneity exceeded I² 75%, and by the study table the active-recovery and massage estimates each rest on a single trial

Main caveat: The authors': sparse networks for several tools (foam rolling, thermoneutral water), so the absence of inconsistency does not prove consistency.
moderate evidenceMeasured on recreational to elite athletesRead in full

Hou, Yin and Qiao asked whether the common physical recovery tools work on the same clock: whether a strategy that helps right after exercise still helps two or three days later.

They searched five databases in January 2026 for randomised trials in team-sport athletes or active people doing team-sport-like work, and included 22 trials comparing cold-water immersion, massage, active recovery, contrast water, stretching, foam rolling, compression, light or infrared therapy, garments and other options with passive rest or with each other.

They analysed jump height, soreness and CK separately at each time point in a network anchored on passive rest and ranked options by P-score. In the first 24 h, active recovery and cold water restored jump height at 24 h, cold water and massage lowered soreness immediately and at 24 h, and cold water lowered CK at 0 and 24 h.

By 48-72 h no option helped jump height or CK, and only cold water still went with lower soreness at 48 h.

The authors conclude that recovery effects are strongly time-dependent: active recovery for short-term explosive performance, massage for early soreness and cold water for broader short-term support, with natural recovery likely enough when 48 h or more is available; they add that repeated cold water in training blocks may blunt strength and muscle gains.

Our reading: the time course is the useful finding; the per-tool rankings are fragile, because by the trial table the active-recovery jump result at 24 h comes from one trial and the massage soreness results from another.

Participants

22 randomised trials, 6 to 45 participants each; about 417 in total (our sum of Table 1; the paper gives no total).
Mostly men; one basketball trial included men and women (split not given) and one field-hockey trial 14 men and 8 women.
Levels from club, collegiate and youth players to professional soccer, futsal and Australian football players; ages not tabulated.
Exercise: soccer, futsal, rugby, basketball and Australian football matches or training, simulated team-sport circuits, Wingate tests, drop or box jumps and eccentric knee extensions; one multi-day preseason block.

Protocol

Search: PubMed, Web of Science, SPORTDiscus (EBSCOhost), Cochrane Library and Scopus on 24 January 2026, terms built around team-sport athletes, recovery interventions, recovery outcomes and randomised designs; reference lists screened. 3,086 records, 1,480 after duplicates, 86 full texts, 22 included.
Inclusion: team-sport athletes or healthy active people doing team-sport-relevant exercise; a defined post-exercise physical recovery intervention; passive rest, seated rest, no intervention, sham or another strategy as comparator; extractable CMJ, DOMS, CK or other recovery data; randomised parallel or crossover design. Nutritional supplementation as the main intervention excluded.
Interventions seen: cold water mostly 10-15 °C for 10-15 min (one 5 °C for 5 min); massage 30 min (one trial); active recovery 20 min (one trial with a 24-h outcome) and one with outcomes at 0 and 48 h; contrast water, stretching, foam rolling, whole-body vibration, electrostimulation, light therapy, infrared or bioceramic garments, cold with compression, and a multimodal package.
Analysis: SMD (or MD for CK in U/L at 24 h) with 95% CI, random effects; P-scores; I²; node-splitting and comparison-adjusted funnel plots; the "acute" (0-24 h) and "delayed" (48-72 h) phases are labels for reporting, each time point analysed separately; disconnected subnetworks left out of the primary analysis.
Risk of bias: RoB 2 by two reviewers; results shown only in Figure 2 (not in the text read).
Funding and interests: no external funding; no conflicts of interest declared.

Key Results

Against passive rest (Table 2; positive SMD = better jump; negative = less soreness or lower CK):

Outcome, timeOptionEffect (95% CI)
CMJ, 0 hnoneno significant difference
CMJ, 24 hActive recoverySMD 0.944 (0.201 to 1.686), ranked first
CMJ, 24 hCold-water immersionSMD 0.447 (0.124 to 0.770)
DOMS, 0 hMassageSMD −2.237 (−3.995 to −0.479)
DOMS, 0 hCold-water immersionSMD −1.147 (−2.106 to −0.188)
DOMS, 24 hMassageSMD −2.616 (−4.427 to −0.804), ranked first
DOMS, 24 hCold-water immersionSMD −2.187 (−3.193 to −1.182)
CK, 0 hCold-water immersionSMD −0.494 (−0.944 to −0.044), the only option that differed
CK, 24 hCold-water immersionMD −177.1 U/L (−343.3 to −10.9)
DOMS, 48 hCold-water immersionSMD −1.568 (−2.578 to −0.558), the only option that differed
CMJ and CK, 48-72 hnoneno significant difference for any option
Heterogeneity and consistency: I² above 75% in the soreness and CK networks; node-splitting found no significant disagreement where loops existed; funnel plots looked broadly symmetrical (visual only).
Where the single-tool estimates come from (our reading of Table 1): the only trial with active recovery and a 24-h jump outcome is Rey 2012 (31 professional soccer players); the only massage trial is Delextrat 2013 (16 basketball players; 0 and 24 h). Cold-water estimates draw on about a dozen trials.
Not reported: effect sizes for the options that did not differ, the full rankings (in the supplement), any MCID anchoring.

Practical Application

With a hard session or race again within a day, a recovery tool may help; with 48 h or more, natural recovery did about as well. In 22 team-sport trials, active recovery and cold water restored jump height at 24 h (SMD 0.94 and 0.45), massage and cold water lowered soreness at 0 and 24 h, and by 48-72 h no option improved jump height or CK; only cold water still went with lower soreness at 48 h (SMD −1.57).
Consistent with [SYNTH-70]'s modality table and [SYNTH-11]'s massage row: massage eases soreness, active recovery cuts soreness and costs nothing, and cold water lowers soreness and CK in the short term (REF-385, REF-250).This review adds the time course and a jump result for active recovery at 24 h, but both the active-recovery jump estimate and the massage soreness estimates rest on one trial each (our reading), so the ranking between tools is not settled. It did not measure stiffness, so it adds nothing to SYNTH-11's point that massage leaves measured stiffness unchanged (REF-317).
Cold water is a short-turnaround tool, not a training-block habit. The authors themselves caution that repeated cold water after training may blunt strength and muscle gains, as SYNTH-70 holds from training studies (REF-011, REF-914); this review measured 0-72 h only.
Inference, not tested here: for a cyclist or runner between race days or stages, an easy spin or jog is the default recovery (SYNTH-70); these trials were team-sport players and measured jumps and soreness, not endurance performance.

Limitations

The authors': sparse networks for several tools (foam rolling, thermoneutral water), so the absence of inconsistency does not prove consistency; heterogeneity above I² 75% for soreness and CK and no certainty rating (GRADE or CINeMA could not be computed); clinical heterogeneity in protocols, levels and doses; no sensitivity analysis without high-risk trials and no formal publication-bias test; no MCID anchoring; protocol archived after extraction began; CK pooled as raw U/L across laboratories at 24 h.
Single-trial estimates (our reading of Table 1): active recovery at 24 h and massage at 0 and 24 h each rest on one trial, so their large effects and top rankings are one study's results.
Reporting: the Results describe prolonged endurance running among the protocols, but Table 1 lists none; several trial years differ between Table 1 and the Figure 2 legend (for example Barber 2020 and 2017, Delextrat 2013 and 2012); the risk-of-bias judgements are only in a figure.
Transfer: team-sport players, nearly all men; no endurance athletes; outcomes jump height, soreness and CK only.
Not read: Figure 2 (risk of bias), the forest and network plots, and the supplementary league tables, node-splitting and funnel plots.
No external funding; no conflicts of interest declared.
How this entry was checked and changed
Why this level

a systematic review of randomised trials read in full, but the RoB 2 judgements are shown only in a figure that was not in the text we read and the authors could not rate certainty, so the rubric's condition for A- is not met; in addition heterogeneity exceeded I² 75% for soreness and CK, the trials are small, and the protocol was archived after extraction began. The CK results are a muscle-damage marker and stand one step lower (B)