Rename decision — "Safety confirmation" column label

STATUS (2026-04-20): Decision = KEEP the label "Safety confirmation". Do not rename.

Decision

The label stays as "Safety confirmation" — in the CEP §Planned evidence classification per study table, in the CEP §Evidence hierarchy narrative, in the <EvidenceRankMatrix> summary cross-tab, and everywhere the term appears in the CER and in the data file.

Rationale

Three considerations, in order of weight:

1. "Safety confirmation" is the term the anchoring regulatory guidance uses. MDCG 2020-6 §6.1 names safety as a first-class dimension of the clinical evaluation; MEDDEV 2.7/1 Rev 4 §11 and MDR Article 83(1) describe the post-production information cycle as "confirming safety" across the lifecycle. Renaming to "Safety endpoints captured" (one of the task-brief candidates) narrows the scope to study-level endpoints and loses the post-market-surveillance and benefit-risk connotations. Renaming to "Contributes safety evidence" is vague and would invite exactly the ambiguity the task is trying to remove. Renaming to "Safety data source" is closer but still loses the confirmatory framing (MDCG 2020-6 §6.3 specifically requires that PMS and PMCF confirm safety conclusions, not just contribute data).

2. Once the definition is added, the ambiguity collapses. All three readings that surfaced in the 2026-04-20 meeting — post-market vigilance (Taig); long-term post-endpoint follow-up (Jordi); redundant because all studies confirm both (Saray) — resolve to the same ticking rule under the proposed definition in proposed-definition.md. The column label was not the problem; the missing definitional text was. Renaming without adding a definition would leave the ambiguity; adding the definition makes renaming unnecessary.

3. Rename blast radius is disproportionate to the value. A label change propagates to: the Pillar type and PILLAR_LABELS map in clinicalEvidenceRankMap.ts; the <EvidenceRankMatrix> cross-tab header; the <EvidenceRankMatrix> rank-ordered-table per-row Pillar badge; the CEP §Planned evidence classification per study row for Legacy PMS data; the CEP §Evidence hierarchy Rank 7 row; the CEP §Route B narrative; the CER §Route B narrative; the CER §Assessment of the combined evidence portfolio bullet; the internal docs/bsi-non-conformities/clinical-review/evidence-rank-and-phases.mdx page; the GP-015 T-015-003 CER template. Every one of those is a potential build error or silent inconsistency the day before the 2026-04-21 re-submission window. The task brief's own §What NOT to do explicitly warns against renaming "merely to signal activity".

Counter-arguments considered and rejected

• "Saray actively dislikes the label." Saray's objection is to the framing (reading C — redundant with performance), not to the two words themselves. The definition answers the framing objection directly by establishing safety as a dimension distinct from performance. A rename would not address the framing objection any better than the definition does, and would add rework without adding clarity.
• "Auditors read MDCG 2020-1 first and see three pillars; a fourth column under the same heading looks like we invented a fourth pillar." Valid concern — and handled not by renaming but by the introduction paragraph added before the <EvidenceRankMatrix> render, which states explicitly that the fourth column is not an MDCG 2020-1 pillar and cites the separate regulatory anchor (MDCG 2020-6 §6.1 + MDR Annex I §§1, 8, 14). The column header in the cross-tab keeps its existing text; the surrounding narrative is what changes.
• "A rename makes the column self-describing and removes the need for a footnote." No — any label short enough to be a column header will always need a footnote at regulatory level. Long descriptive labels ("Safety-relevant endpoint collection & post-market surveillance dimension") make the table unreadable. The trade-off is: short label + definitional footnote vs long label + no footnote. The short-label route is the regulatory convention (MEDDEV 2.7/1 Rev 4 is full of short column headers with footnotes).

Revisit trigger

If BSI explicitly flags the label in Round 1 feedback — e.g., "what is meant by Safety confirmation; how does it relate to the three MDCG 2020-1 pillars" — and the definition we add is judged insufficient, reconsider the rename at that point. Candidate substitutes to have ready: "Safety evidence contribution"; "Residual-risk confirmation"; "MDR Annex I safety dimension". Record the auditor's exact phrasing when reopening.