# Adversarial review log — ccsvi-ms-liberation

- **Authoring model:** Sol 6.1
- **Adversarial reviewer:** Claude Opus 5.5 (Anthropic)
- **Date:** 2026-10-14
- **Canonical file:** `hsef_v53_harm_related_three_2026-10-12\ccsvi-ms-liberation\ccsvi-ms-liberation.review.md`
- **Level before → after:** Β — Associational → **Β — Associational** (unchanged)
- **Confidence before → after:** Moderate → Moderate (band unchanged; rationale rescoped, see F8)
- **Causal role / M / S / D:** epiphenomenon · Little new work · Stable · Contested (all unchanged)
- **PMIDs checked:** 71/71 resolved (69 references + correction notices 28072565, 31896294); 0 failures, 0 retractions or expressions of concern
- **Load-bearing claims spot-checked:** 16
- **Validator:** FAIL (1) — `v5.lay_summary` "no lay_summary; mandatory from v5.4", 31 checks pass. **Identical on the pre-edit file**, so this is a pre-existing v5.4-versus-v5.3 mismatch and not a regression. Output saved to `validator.txt`; the original package report is preserved at `sol61-original/validation-report.json`.

## Findings by severity

| Severity | Count |
|---|---|
| Critical | 0 |
| Major | 1 |
| Minor | 5 |
| Editorial | 2 |
| **Total corrections applied** | **8** |

## Findings table

| ID | Severity | Location | Correction | Evidence |
|---|---|---|---|---|
| F1 | Major | Evidence and causal interpretation; item [2]; card [2] | Serious-adverse-event counts by arm (5 of 49 venoplasty, 1 of 55 sham) now stated before the blinded adjudication | PMID 30266886 abstract. Reporting only the "unrelated" adjudication understated the harm signal. |
| F2 | Minor | Item [22]; card [22] | "surgical removal of a damaged angioplasty balloon" → "surgical femoral venotomy to retrieve an angioplasty balloon" | PMID 21107001 abstract reports femoral venotomy to remove a balloon; it does not say the balloon was damaged. |
| F3 | Minor | Item [30]; card [30] | CoSMo comparator "other neurological disorders" → "other neurodegenerative diseases" | PMID 24014572 abstract. |
| F4 | Minor | Cards [1], [7], [19] | "ErratumIn: PMIDnot supplied." → "ErratumIn: PMID not supplied in the PubMed record." | Known builder defect; no substitutable identifier exists in PubMed. |
| F5 | Minor | REFERENCES 12, 13, 27, 28, 51, 53, 54, 55, 65 | "DOI: ." → "DOI: none in PubMed record." | PubMed confirms these nine records carry no DOI. |
| F6 | Editorial | Item [4]; card [4] | "RR1.42, 95%CI1.00-2.01, p.032" → "RR 1.42, 95% CI 1.00-2.01, p = 0.032" | Values match PMID 31735108. |
| F7 | Editorial | REVIEWER NOTES; Consensus reconciliation | "blocked403" → "blocked (HTTP 403)"; "amended target300" → "amended target of 300" | Rendering defects. |
| F8 | Minor | Score card `final_score.confidence_rationale` | Rationale rested partly on "incomplete access" → rewritten to rest only on evidence density and reproducibility | The framework ties confidence to evidence density and reproducibility, not reviewer access. Band re-derived and **stays Moderate**: three independent sham-controlled trials and several blinded diagnostic programmes replicate the null (an argument for High), but every trial was small or under-recruited with incomplete target engagement and the diagnostic literature remains operator- and protocol-dependent — the Moderate text. |

Rejected findings, the full evidence column and residual limitations are recorded in the
`## ADVERSARIAL REVIEW AND APPROVAL` section of the review itself.

## Files changed in this case folder

- `ccsvi-ms-liberation.review.md` — F1-F8, banner, `## ADVERSARIAL REVIEW AND APPROVAL`
- `ccsvi-ms-liberation.scorecard.json` — F1-F4, F6-F8 mirrored in the matching evidence-card and access-status strings (no level, confidence, subclaim, aggregation or M/S/D field changed)
- `validator.txt` — new
- `ADVERSARIAL_REVIEW.md`, `adversarial-findings.json` — new
- `sol61-original/` — pre-edit copies of the review, the scorecard and `validation-report.json`

## Propagation

The finished review and scorecard were copied byte-for-byte to
`hsef_v53_remaining_priority_20_22_2026-10-12\ccsvi-ms-liberation\`, and the hashes were compared.

## Approval

Following adversarial review and the corrections listed above, the revised document is approved.
— Claude Opus 5.5 (Anthropic), 2026-10-14.