Issue No. 003: Passive Control

Recovery is specified as physiology. What the evidence establishes is a null.

Nothing beat lying down

A network meta-analysis in competitive and elite endurance athletes pooled sixteen randomised trials, 273 participants and ten modalities across eight outcomes, with passive recovery as the reference. Active recovery, cold-water immersion, whole-body cryotherapy, compression garments, pneumatic compression, deep and superficial massage, electrical stimulation and far-infrared were all in the room.

The verdict. Across 47 comparisons against passive recovery, no modality was shown to outperform doing nothing on any outcome examined. Three intervals excluded the null and the authors discard all three: two rest on single-trial nodes, and all three come from networks pooling measures that are not commensurable.

The certainty. Thirty-one of the 47 comparisons were rated very low, fifteen low, and exactly one moderate.

The one certain finding is a null

Most empty results mean nobody looked hard enough. A few mean the looking was good enough to rule out an effect worth having. This review contains one of the second kind, and it is the more useful sentence.

Moderate certainty, once. Compression garments do not improve sport-specific performance: SMD 0.04, 95% CI -0.37 to 0.45, against a pre-specified minimal important difference of 0.5. It is the only comparison in the entire review rated above low.

Two more join it. Active recovery at -0.01 (-0.39 to 0.36) and neuromuscular electrical stimulation at 0.00 (-0.40 to 0.40). All three intervals sit entirely inside the range the authors pre-specified as clinically unimportant.

Why it holds. This is the best-supported network in the review, eleven trials, with heterogeneity of exactly zero. In the authors' words, it is “the one network in which the absence of an effect is informative rather than merely uninformative”.

The sham out-ranked the machines

One included trial ran a thermoneutral immersion at 27 degrees as a sham. Entered as its own node, it did not behave like a control.

Where it finished. Highest of five in the flexibility network, P-score 0.683 against cold-water immersion at 0.516 and deep massage at 0.439. Third of eleven for sport-specific performance at 0.671, above deep massage, active recovery, superficial massage and compression garments. Above both massage arms for countermovement jump.

The authors' reading. “The sham was not merely non-inferior to the active modalities, it out-ranked most of them. This is what an expectancy account of this literature predicts, and it is difficult to reconcile with an account in which the active modalities exert modality-specific physiological effects of the size usually claimed.”

Why it cannot be resolved yet. Physical modalities cannot be blinded; the athlete always knows. The outcomes reported most often, soreness and perceived exertion, are the ones most sensitive to expectation. On these designs, physiology and expectation are indistinguishable.

What the review refuses to say

Anyone about to run this as proof that cold water does nothing for soreness should read the soreness network first.

It is uninterpretable. Heterogeneity of 90.2 per cent, a between-study standard deviation of true effects of 1.32 SMD units, and intervals roughly four SMD units wide. The spread of true effects is larger than any point estimate in the network.

It reverses. Remove two of its five trials and four modalities become superior by around three standard deviations. The authors report both analyses and conclude from neither, which is the right call.

The base is thinner than sixteen sounds. Thirty-five of the 55 nodes across the eight networks, 64 per cent, rest on a single trial. Four of the sixteen cohorts were adolescents, mean ages 13.6 to 17.7. Only seven trials reported VO2max at all.

It cannot survive itself. Enforce the review's own elite eligibility criterion and four of the eight outcomes have no estimable network left. The authors call that “among the more informative results of this review”, and they are right. They also document and correct a coding error from their own submitted version, caught at peer review, with the code and arm-level data published. That is a paper behaving well.

The point: the recovery literature cannot presently separate a physiological effect from an expectation, and the one thing it has established with reasonable confidence is that compression garments will not make the next performance better.