Complete clinician-approved specifier definitions and taxonomy restructuring

[BigSimmo]

Current scope — 13 September 2026

Status: open. Retain the display safeguards while completing clinical review and the structural decision.

Reviewed repository baseline: 68041c8c8c03870e6f3fbc3287b1de806e38b9c4 (main). The original issue and the automated comment's blanket claim that all structure/labels are sound must not be treated as current clinical approval. In particular, the original issue itself identifies outstanding taxonomy work.

Implemented safeguards versus unfinished work

• Generated definition text is withheld from the compact client index. Current test definitions assert that generated meaning text is empty there.
• The possession-trance label has corrective wording identifying a separate diagnosis. This is a wording mitigation, not completion of the structural split.
• Qualified clinician approval/authoring of the intended definition prose remains to be evidenced.
• The agreed Trance taxonomy restructuring remains to be verified. Current index entries retain legacy with-without-possession slugs. The existing test checks label/definition wording rather than proving separate diagnosis nodes throughout the taxonomy.

Checked items record the reviewed implementation, not a new test run or clinical validation performed by this edit.

Remaining work

• Inventory the current catalogue and intended definition-review population. Treat 585 items and the historical 68/71 source cross-check as dated observations, not a current comprehensive approval.
• Obtain clinician-authored or clinician-approved definitions with reviewer identity, source/version, decision and review date. Keep source verification separate from clinical approval.
• Confirm the intended Trance/Possession-trance representation against the authoritative classification with a qualified reviewer, then implement the agreed distinct-node relationships. Do not apply the same split blindly to the DID entries described separately in the original issue.
• Add structural acceptance tests for identity, code, parent/group relationship, navigation and search projection. A label containing 'separate diagnosis' alone must not satisfy the structural criterion. Preserve or deliberately redirect existing slugs so the change does not create broken links.
• Confirm that only appropriately reviewed prose can be displayed/published, and that unreviewed generated prose remains withheld. Record scoped test and review evidence and reconcile linked internal tracking.

Closure criteria

Clinical review of the intended prose has auditable evidence, the agreed taxonomy relationships are implemented and tested, and search/detail/publication behaviour is verified. Any residual review work must have an explicit tracked disposition. Do not remove withholding safeguards simply to close this issue.

Ownership and related work

Requires a qualified clinical reviewer and an engineering owner for taxonomy/projection changes. No reviewer acceptance or new assignee is implied. Grouped under area:clinical-quality with #229 and #289, but not merged into either because the datasets and closing criteria differ.

No definitions, taxonomy, production data, clinical statuses or tests were modified by this issue refresh. Existing comments are retained. This current-scope section supersedes the automated comment that says structural changes are unnecessary.

Original issue text, preserved as historical context rather than current approval

Tracking issue for the clinician-facing follow-ups to the DSM-5-TR specifier catalogue (#656), after the interim safety change in #691 (all generated definition text withheld from display) and a source-verification pass of the 71 previously-displayed "source-verified" definitions.

Context

The imported catalogue's auto-generated definitions proved unreliable — automated review repeatedly found clinically wrong text on rows whose source was "verified". #691 now withholds all generated definition/clinical-note text from display; the catalogue shows only structural facts (label, disorder, group, ICD-11 context), review status, provenance badge, and the hand-authored curated guidance. The nested dataset is left intact so clinician-authored definitions can be restored later.

A source cross-check of all 71 previously-displayed definitions against DSM-5-TR and WHO ICD-11 found 68/71 confirmed (the severity thresholds, durations, and subtype codes are accurate) and 3 contradicted.

1. Author clinician-reviewed definitions (the real fix)

Have a qualified clinician write/approve the specifier definitions before any generated definition text is re-displayed. The 585-item structure and labels are sound and can be kept; only the definition prose needs authoring. The full source-verification report (per-item verdicts + cited DSM-5-TR / ICD-11 URLs) is available as a starting point.

2. Trance Disorder possession structure (contradiction from the verification pass)

The catalogue models Trance Disorder with a "with/without possession" specifier. Per ICD-11 this is a structural mismatch: ICD-11 uses two separate diagnoses — 6B62 Trance disorder (without a replacing external identity) and 6B63 Possession trance disorder — not a with/without qualifier of one disorder. This needs restructuring the taxonomy node (split into two diagnoses), which is a clinician/structural decision, not a label edit.

• Note: the parallel Dissociative Identity Disorder "with/without possession form" is correct — ICD-11 6B64 genuinely covers both forms within one diagnosis. Only Trance Disorder needs the split.

Already fixed (no action needed here)

• ARFID third driver relabelled from the ICD-11 exclusion "limited availability" to the genuine driver "apparent lack of interest in eating / low appetite" (6B83) — done in fix(specifiers): withhold all generated definitions pending clinician review #691.
• Gaming online/offline, PTSD/Conduct threshold notes, AMPD provenance, BED/Hypersomnolence severity notes, possession wording — corrected during feat(specifiers): add full DSM-5-TR catalogue with verification gate #656 review.

Related: #656 (catalogue), #691 (withhold definitions).

🤖 Generated with Claude Code

[github-actions]

Summary of GitHub Issue: Specifier Catalogue - Clinician Review of Definitions + Trance Disorder Possession Structure

This issue tracks clinician-driven improvements to the DSM-5-TR specifier catalogue, following prior issues that revealed inaccuracies in auto-generated definitions. Key points discussed:

1. Issue Background:

   • Auto-generated definitions for specifiers in the imported catalogue were found unreliable, containing clinical inaccuracies.
   • As an interim safety measure, the definitions have been withheld from display (fix(specifiers): withhold all generated definitions pending clinician review #691), leaving only structural data visible (e.g., labels, categories, ICD-11 context).
   • A manual source-verification cross-check of 71 previously-displayed definitions confirmed 68 as accurate but found 3 with contradictions.

2. Actions to Address:

   • Definition Review and Authoring: A qualified clinician needs to write or approve the prose definitions for the specifiers. The structural layout of the 585 items is sound and doesn't need changes, just the definitions.

[coderabbitai]

🔗 Related PRs

#446 - Merge branch cleanup and review workflow updates [merged]
#483 - feat(differentials): interactive diagnosis detail page with honest safety snapshot [merged]
#633 - feat(specifiers): add diagnostic decision-support workspace [merged]
#656 - feat(specifiers): add full DSM-5-TR catalogue with verification gate [merged]
#661 - DSM: add diagnosis reference mode [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!