COD Snapshot Export

Data Management COD Snapshot Export

Purpose

This policy defines the baseline behavior for the data-management coded COD snapshot export and its backing reporting materialized view.

Baseline

  • the coded COD snapshot export is submission-granular: one row per va_sid
  • it reports only active/current COD-facing records, not history
  • it preserves three distinct human-COD views when available:
    • latest active coder data
    • latest active reviewer data
    • authoritative final COD outcome
  • coded_at_authoritative means the saved timestamp of the authoritative final COD outcome, not merely the latest coder Step 2 save

Included Data

The reporting snapshot may include:

  • project, site, form, submission, workflow, and demographic context
  • active payload unique_id and survey_block values when present
  • narrative text and narration language
  • latest active coder Step 1 and Step 2 COD fields
  • latest active reviewer final COD fields
  • authoritative final COD fields
  • active SmartVA result fields
  • active Narrative Quality Assessment fields, including the individual scored NQA question columns
  • active Social Autopsy fields, including the app-owned analysis summary and the active payload's raw sa* questionnaire fields
  • active coder/reviewer assignment names where available

WHO 2022 Bucket Mapping

Buckets come from the WHO_2022_VA_2026 scheme for every death (owner, 2026-09-24; docs/policy/icd10-to-icd11-transition.md decisions 6 and 7). Each value is bucketed through the rows of its own classification, decided by its code shape: ICD-10 values through the icd10 rows, ICD-11 values through the icd11 rows (exact code, then nearest catalogue ancestor; a post-coordinated value by its first stem). SmartVA ICDs are ICD-10. No crosswalk is used. Each bucket carries a provenance value: icd10, icd11_native, unmapped (a value with no bucket), or empty (no value).

WHO 2022 VA COD bucket mapping must be derived independently for:

  • latest coder final ICD
  • latest reviewer final ICD
  • authoritative final ICD
  • SmartVA cause 1 ICD
  • SmartVA cause 2 ICD
  • SmartVA cause 3 ICD

If no WHO 2022 mapping exists for a given ICD, the export must leave the bucket fields empty rather than infer a fallback.

For Step 1 reporting:

  • the export should include the preserved latest coder Step 1 assessment for the submission when one exists
  • final COD submission does not itself invalidate Step 1 reporting data

When a historical ICD is present in coding data and map_icd10_legacy_reporting_aliases defines a reporting-code replacement, the bucket lookup must use that reporting ICD. This reporting-only normalization must not overwrite the raw coder, reviewer, authoritative, or SmartVA ICD columns exposed by the snapshot export.

Authority Semantics

  • authoritative COD follows the existing final-COD authority model
  • reviewer authority takes precedence when explicitly set
  • if no explicit authority row exists, reporting may fall back to the latest active reviewer final COD, then the latest active coder final COD
  • this export does not change any workflow or authority behavior; it only reports the current state