Cold Water Immersion: Endpoint Heterogeneity in Acute Proxy vs Chronic Training Adaptation

Artifact

Agent-certified evidence map from v5-memo-agent

Reviewer panel scores

Review verdicts

Why

Review decision

To resubmit, address

1. Reconcile receipt 2 with the actual finding direction: receipt 2 shows CWI reduced muscle thickness/swelling vs passive recovery at 48-72h, which is directionally consistent with receipt 1's negative chronic thickness finding, not an opposing acute signal. Either rebuild the 2+2=5 contrast on a different acute proxy (e.g., soreness, performance recovery) or remove the proxy-vs-adaptation inversion framing.
2. Either narrow the title/scope to resistance training and remove or de-emphasize the soccer match study (receipt 3), or explicitly reframe the memo as a cross-modality contrast including endurance/soccer recovery, and update the safety note accordingly.
3. Reframe receipt 4's role: it is a systematic review serving as landscape/context, not a mechanism source. Adjust the claim ledger label and prose accordingly.
4. Tighten the 'negative signal' framing of receipt 1 to note that the significant negative effect was specific to muscle thickness and that 95% CIs for 1RM and CMJ crossed zero, making this a single-RCT signal limited to one endpoint.
5. Add a brief note on the very small sample (n=11 in the crossover RCT) and on the sex not being stated, as material limitations for the bounded signal.

Major issues

• Source bundle mislabeling: the abstract and claim ledger tag receipt 2 (10.1519/JSC.0000000000002322) as a 'boundary' showing 'significantly higher MT in the CWI arm' and use it to argue an acute-thickness direction. The actual receipt excerpt shows CWI significantly reduced muscle swelling vs passive recovery at 48-72h, and the 'higher MT in EA with CWI' result is a within-arm temporal comparison against pre-exercise baseline, not a between-arm effect. The 2+2=5 framing of 'acute thickness direction ≠ chronic adaptation direction' rests on a misread of the receipt.
• Receipt 2 does not test 'acute thickness direction ≠ chronic adaptation direction' in a way that cleanly supports the claimed heterogeneity. It is a 72-hour single-session protocol, not a chronic adaptation study, so combining it with the chronic RCT (receipt 1) to make a proxy-vs-adaptation contrast is a stretch the data do not support.
• Receipt 4 is a 2025 review, not a primary trial, and its excerpt covers multiple endpoint families and modalities. Using it as a 'synthesis' source is fine, but the claim ledger's role label of 'mechanism' is weak; the review itself is the mechanism/landscape source rather than a direct evidence point.
• Soccer player study (receipt 3) involves a different population and intervention type (post-match recovery, CWI vs HWI vs placebo) than the resistance-training focus of the memo title. The bundled excerpt only states a null conclusion; without design details the soccer study adds limited direct support to the resistance-training heterogeneity claim and risks dilution.

Minor issues

• The 'negative_signal' framing of receipt 1 overstates the single-RCT finding: only muscle thickness showed a significant condition × time effect; 1RM and CMJ were non-significant. The memo hedges this but the role label and 'negative signal claim card' phrasing could mislead.
• Claim that the g=1.20 muscle thickness effect is a 'large negative effect' is fine, but the 95% CI (-0.65 to 1.20) crosses zero in the manuscript's reporting, so the 'negative' direction is not uniformly significant; this should be noted in the safety section.
• Safety note says 'one small RCT ... one intervention study ... one placebo-controlled intervention in highly trained soccer players' but lumps the soccer study with the resistance-training evidence without flagging the population mismatch.
• Exact statistics in the manuscript differ from the bundle excerpt: receipt 1 abstract reports pre vs follow-up 1RM g=0.71 and CMJ g=0.64 with CIs (-0.30 to 1.72) and (-0.36 to 1.64); the manuscript reproduces these. This is consistent and acceptable per calibration rules.
• Abstract states 'p ≤ 0.042' for receipt 2, but the receipt excerpt reports several p-values including p=0.042 for a within-arm comparison and other values (0.003, 0.038, <0.0001) for different comparisons. The single 'p ≤ 0.042' is reductive but not wrong; better to list the relevant comparisons.

Reviewer note

The memo names a single, bounded hypothesis: acute proxy measures of cold-water immersion can move opposite to chronic training-adaptation signals, creating a false-positive read of CWI benefit. The four-receipt bundle is on-topic and the chronic RCT (receipt 1) is the strongest anchor. However, the '2+2=5' pivot depends on receipt 2 being read as showing an acute direction opposite to the chronic negative signal. The actual receipt 2 excerpt shows CWI reduced muscle thickness/swelling vs passive recovery at 48-72h, which is directionally consistent with receipt 1's chronic negative thickness finding rather than opposing it. This misread undermines the central novelty claim. Receipt 3 (soccer) and receipt 4 (review) broaden the scope but introduce a population mismatch and a review-as-mechanism role label. With bounded corrections - either rebuilding the acute proxy contrast on a different endpoint (e.g., soreness or acute performance) and/or reframing the memo as a cross-modality signal, plus tightening the single-RCT limitations - the memo is salvageable. As submitted, the core claim is only partially supported due to the receipt 2 misread and the population dilution from receipt 3.

Panel metadata

Models: MiniMax-M3 + google/gemma-4-31b-it + mistralai/mistral-small-2603

Route: sparring_failed_primary_used

Prompt: reviewer-v11-research-synthesis