Do we need this task? Honest sizing review (Round 2 contingency re-opened)

Status (2026-04-21): Written after the user asked whether this task genuinely closes a gap, and flagged that the task CLAUDE.md was seeded from lightly corrupted copies of the bsi-clinical-auditor and celine-clinical-consultant agent briefs. The user also flagged the Slack decision recorded at https://legithealth.slack.com/archives/G01FNQNCHCZ/p1776764305591299 (parent thread p1776763246590659), which consolidates the clinical-evaluation roster on Antonio Martorell as the sole SME plus Celine Horiana as external consultant reviewer. This memo re-scopes the task in the light of that decision.

TL;DR

• The task is worth keeping, but only at ~15 % of the original scope. Options B, C and D in the original brief are permanently off the table — they are not regulatory obligations, not Round-2 contingencies, and will not be executed.
• The single principled action that remains is Option A: add Céline Fabre (Horiana, Health Data Consulting) to the CEP and CER as External Methodological Reviewer, in a dedicated subsection under the §Responsibilities / §Qualification-of-the-responsible-evaluators section. CV and Declaration of Potential Conflict of Interest to be added to Annex I as housekeeping once collected.
• Option A is principled, not cosmetic: Horiana delivered a formal methodological review on 2026-04-17 (methodological-expertise report V1.0 + recommendations document V1.0, both PDF deliverables on file), led by Céline Fabre, with a 2026-04-20 follow-up alignment call; Saray Ugidos translated the framework internally on 2026-04-18. The audit-visible §Responsibilities section is the notified-body-visible roster; Horiana's work was previously only visible in internal files.
• Options B (named dermatopathologist), C (named paediatric dermatologist) and D (named rare-disease consultant) are rejected on regulatory grounds: the MDR / MDCG 2020-13 Section A / MEDDEV 2.7/1 Rev 4 §6.4 do not require named subspecialty consultants within a specialty. A single board-certified general dermatologist anchoring the medical-knowledge competence, combined with evidence-based subspecialty coverage (biopsy-confirmed reference standards in pivotal malignancy investigations, peer-reviewed severity-validation literature, per-epidemiological-group technical-performance V&V, MAN_2025 Fitzpatrick V–VI MRMC, legacy-device RWE + PMS), is a defensible regulatory posture. If BSI raises SME breadth in Round 2, the Round-2 response defends this posture with the applicable regulatory clauses; it does not recruit.

The four facts that drive the re-sizing

Fact 1 — The real textual defect was already closed under task-3b11

Task-3b11 (../completed-tasks/task-3b11-sme-coverage-subspecialty-documentation/) closed on 2026-04-20 under Option D-lite. The CEP phrase "the Subject Matter Expert's documented consultation network for ad-hoc subspecialty input on case-level adjudication where required" (which promised an Annex I roster that did not exist) was deleted and replaced with a structured six-row subspecialty-coverage table citing only audit-visible artefacts (MC_EVCDAO_2019, AIHS4_CSP_2025, APASI/AUAS/ASCORAD literature, per-epidemiological-group ICD V&V in R-TF-028-006, MAN_2025 Fitzpatrick V–VI MRMC, R-TF-015-012 legacy RWE, R-TF-007-003 legacy PMS). The same rewrite was applied to the CER at line 2850.

The internal-consistency bug — the hard part — is gone. Anything this task does is on top of a base that already reads cleanly to a notified-body PDF reviewer.

Fact 2 — BSI did not raise SME breadth in Round 1

A verbatim grep across every question.mdx under clinical-review/round-1/ plus the 2026-03-25 Erin + Nick clarification-call transcript returns zero hits for "evaluator qualifications", "team competence", "SME coverage", "subspecialty expertise", "consultation network", or "clinical evaluation team". Round 1 addresses Items 1–7 (CER update frequency, device description, clinical data / equivalence / PMS / sufficiency / gaps, usability, PMS Plan, PMCF Plan, risk). Annex XIV Part A §1(d)/(f) evaluator qualifications and MDCG 2020-13 Section A team-expertise are not in that list.

The Round-1 preamble warns that later rounds may probe CEAR Section A more aggressively, so Round-2 escalation is plausible, but Round-2-plausible is not Round-1-required.

Fact 3 — The finding comes from the internal bsi-clinical-auditor agent rubric, not from BSI

Line 72 of .claude/agents/bsi-clinical-auditor.md hard-codes this exact pattern as a rubric finding: "SME coverage must be proportionate to the breadth of clinical indications: a single SME for a multi-domain MDSW (malignancy + severity assessment + paediatric + rare-disease) is a finding." The rubric is genuinely backed by MDCG 2020-13 Section A and MEDDEV 2.7/1 Rev 4 §6.4, so the concern is not fabricated. But "our internal agent would flag this" is not equivalent to "BSI flagged this". The agent is doing its job by surfacing the risk; the company decides whether to act on it now or on BSI trigger.

Fact 4 — The team has already taken a deliberate decision on roster shape

The user's Slack anchor (https://legithealth.slack.com/archives/G01FNQNCHCZ/p1776764305591299, parent thread p1776763246590659) documents the decision to consolidate on a single SME (Antonio) and add Celine as external consultant reviewer, after removing the earlier Constanza Balboni and Maria Belen Hirigoity reviewer directories (still visible in git history under docs/legit-health-plus-version-1-1-0-0/clinical/Evaluation/R-TF-015-007-constanza-balboni and …-maria-belen-hirigoity, both deleted from the working tree). That is a team decision, not a default, and it constrains the option space:

• Recruiting brand-new named subspecialty consultants (Options B, C, D in the original brief) contradicts the direction of travel the team chose in that thread. Doing it now, under Round-1 pressure, would reverse a deliberate decision without a BSI finding to justify the reversal.
• Celine, by contrast, is already on record. Surfacing her in the §6.4 table is consistent with the Slack decision, not a departure from it.

Why the original task brief is too big

The original CLAUDE.md frames four options and presents them as roughly peer-weighted choices:

• Option A — Celine on §6.4 (methodological reviewer).
• Option B — named dermatopathologist.
• Option C — named paediatric dermatologist.
• Option D — named rare-disease / genodermatoses consultant.

Three of the four (B, C, D) require recruiting external board-certified subspecialists into audit-visible documents with signed CVs and DOIs. That is:

1. Slow. Realistic lead time 6–8 weeks per named consultant under a Round-2 clock. Fragile under the resubmission window.
2. Load-bearing against a soft finding. BSI has not asked for this. If Round 2 does not raise it, the recruitment cost was unnecessary.
3. Inconsistent with the Slack decision. The team chose a deliberately lean roster. Swelling it in response to an anticipatory internal finding undoes that choice.
4. Prone to fabrication risk. Under time pressure the temptation to put "Dermatopathology consultant: [TBC]" into Annex I grows, which is strictly worse than omitting the row.

None of those objections apply to Option A.

Why Option A is worth doing now

Option A (Celine on §6.4 + CV + DOI in Annex I) is the only one of the four options where every precondition is already satisfied:

• Formal contribution on record. Celine delivered a dated methodological review on 2026-04-17 with three artefacts (expertise-and-recommendations.md, key-insights-and-action-items.md, and the scope-limitation addendum). Saray Ugidos' internal translation dated 2026-04-18 is on record. This is a documented consultancy engagement, not a retrospective naming.
• CV and DOI obtainable without recruitment. Celine is an existing external consultant; the CV and DOI are administrative collection, not hiring.
• Role is distinct and non-overlapping. External Methodological Reviewer is a clearly-titled, narrow role — not a "silent dermatopathologist". It anchors the MDCG 2020-1 three-pillar framework, pillar mapping, Rank-vs-Pillar orthogonality and indirect-benefit causal-chain integrity. It does not claim subspecialty clinical adjudication in dermatopathology, paediatrics or rare-disease dermatology — those remain discharged through the Option D-lite structured-evidence scaffolding already in place.
• Consistent with the Slack decision. The roster shape the team chose is Antonio + Celine. Option A surfaces that shape in the audit-visible document.
• Genuine de-risking on a latent Round-2 exposure. Erin and Nick, if they probe §6.4 in Round 2, read a five-person roster (four internal + one external methodological reviewer) rather than a four-person roster. That is a better starting posture for a Round-2 conversation without committing the company to anything it cannot defend.

What Option A does NOT do

Stating the limits honestly so the scope does not quietly creep:

• Option A does not close the single-SME-vs-346-ICD-11-breadth concern. Celine is a methodological reviewer, not a subspecialty clinical adjudicator. If Round 2 specifically flags that Antonio alone is insufficient for dermatopathology / paediatrics / rare-disease, Option A will not satisfy that finding. The Option D-lite structured-evidence scaffolding is still the primary answer to the subspecialty-breadth concern, and Options B, C, D remain available as escalations.
• Option A does not change the Option D-lite subspecialty-coverage table. That table (six rows in the CEP, mirror in the CER) is already the subspecialty-breadth answer; nothing in this task re-opens it.
• Option A does not re-introduce Constanza Balboni or Maria Belen Hirigoity. The Slack decision to remove them stands.

Recommended execution scope

Keep the task. Drop it to Option-A-only. Concretely:

1. Draft reviewer-roster-rewrite.md — the proposed new row in the §Responsibilities four-competence table of R-TF-015-001 (after the four existing rows at lines 372–377) naming Celine Horiana as External Methodological Reviewer with her academic degree, relevant years of experience, and justification of suitability (methodological anchor for the MDCG 2020-1 three-pillar framework, pillar mapping, indirect-benefit causal-chain integrity). Include the mirror row in R-TF-015-003.
2. Update §MEDDEV 2.7/1 Rev 4 §6.4 four-competence coverage (lines 388–395) to add Celine explicitly under Research methodology and Information management where her expertise applies, with no contamination of the Medical-knowledge anchor competence (that remains Antonio).
3. Draft annex-i-addition.md — the CV + Declaration of Potential Conflict of Interest package for Celine, to be attached to Annex I of both the CEP and the CER once the user confirms the materials are on file.
4. Run the three-agent reviewer pass (bsi-clinical-auditor, audit-deliverable-reviewer, celine-clinical-consultant) on the rewritten §Responsibilities and §6.4 sections before edits are applied to the audit-visible documents. Celine's own agent will check that her role is described accurately (external methodological consultation, not primary clinical evaluation).
5. Close out Options B, C, D permanently. They are not Round-2 contingencies — they are rejected on regulatory grounds. The MDR, MDCG 2020-13 Section A and MEDDEV 2.7/1 Rev 4 §6.4 do not require named subspecialty consultants within a specialty. If BSI raises the subspecialty-breadth concern in Round 2, the Round-2 response defends the current structured-evidence posture against the applicable regulatory clauses; it does not recruit. Any archival-level option descriptions in the main CLAUDE.md or in completed-tasks/task-3b11/CLAUDE.md that frame B/C/D as "escalation" or "Round-2 contingency" should be read as superseded by this 2026-04-21 regulatory-reading decision.

Effort and sequencing

Route	Effort	Human dependency	Round-1 closure	Residual Round-2 exposure
Original brief (all four options evaluated)	2–6 weeks	High (recruit three named consultants)	Uncertain	Low
Option-A-only (recommended)	2–4 hours	None (Celine already engaged; need CV + DOI)	Same-day	Medium (same as after task-3b11 closed; Option A modestly improves posture)
Do nothing further	0 hours	None	Same-day	Medium (Option D-lite stands; §6.4 roster remains four internal team members only)

Option-A-only is a ~15 % scope against the original brief with the majority of the defensible Round-2 de-risking captured. Options B, C, D collectively cost 10–20× more effort, contradict the Slack decision, and mitigate a concern BSI has not raised.

Residual risk after Option-A-only

Two Round-2 risks remain, both defended on the regulatory reading rather than closed by recruitment:

1. Single-SME footprint may read, to a notified-body reviewer, as a single clinical SME for 346 ICD-11 categories. The Option D-lite structured subspecialty-coverage table (task-3b11 output, already in the CEP and CER) discharges the evidentiary side. The team-breadth side is a regulatory-reading disagreement, not a regulatory gap: MDR Annex XIV Part A §1(d), MEDDEV 2.7/1 Rev 4 §6.4 and MDCG 2020-13 Section A do not require named subspecialty consultants within a specialty when the medical-knowledge anchor is a board-certified specialist with adequate experience. If BSI raises the concern in Round 2, the response defends the posture; it does not recruit.
2. Nick's "346 indications is unusual in my experience" stance interacting with the single-SME footprint. Documented Nick position; independent of this task. Mitigation is on the evidence-breadth side (task-3b5 triangulation, task-3b7 per-epidemiological-group V&V, MAN_2025, R-TF-015-012), not on the team-breadth side. Option-A-only does not change Nick's exposure and does not need to.

Neither residual risk is load-bearing enough to justify recruiting subspecialty consultants — in Round 1, in Round 2, or at any subsequent point.

What NOT to do

• Do not recruit named dermatopathology, paediatric dermatology or rare-disease dermatology consultants — ever. The MDR does not require this. Even if BSI raises the concern in a later round, the correct response is to defend the structured-evidence posture against the applicable clauses of MDR Annex XIV Part A §1(d), MEDDEV 2.7/1 Rev 4 §6.4 and MDCG 2020-13 Section A, not to cave and recruit.
• Do not add Celine to the table without her CV and signed Declaration of Potential Conflict of Interest on file. Option A is principled only if the Annex I package is real. If the DOI cannot be collected before the Round-1 submission window closes, Option A defers to Round 2 as well.
• Do not re-open the §Subject Matter Expert coverage across the indication scope table (lines 399–410 at 2026-04-21). That is task-3b11's output and is the primary subspecialty-breadth answer; leaving it alone is the point.
• Do not describe Celine as a clinical subspecialty adjudicator. Her role is external methodological reviewer; the celine-clinical-consultant agent will catch any mis-framing.
• Do not re-introduce Constanza Balboni or Maria Belen Hirigoity. The Slack decision to remove them stands and is not being reversed by this task.
• Do not cite task-3b11-…, task-3b15-…, completed-tasks/, any CLAUDE.md or any Slack URL from the CEP / CER / Annex I rewrites. Cite only audit-visible R-TF-… records and the applicable MEDDEV / MDCG / MDR references.

Ownership

• User: confirms that Celine's CV and signed Declaration of Potential Conflict of Interest are on file (or obtainable within the Round-1 submission window), then signs off on Option A.
• Claude, in a separate session: reads this memo and the current §Responsibilities and §6.4 sections of R-TF-015-001 and R-TF-015-003, drafts reviewer-roster-rewrite.md and annex-i-addition.md, runs the three-agent reviewer pass, and applies the edits to the audit-visible documents once the user confirms sign-off.