Evidence-map and thematic-synthesis method

The product is a targeted evidence map and thematic case synthesis of PFD reports. It borrows transparent reporting and mapping principles from JBI, PRISMA and NHS patient-safety thematic-review guidance (SRC-030 to SRC-032), but does not claim full systematic- or scoping-review conduct.

Scope

Framework period: 25 July 2013 through the search date, with the focused series running from 1 January 2018. The 2018 backfill used a bounded publication-year PDF/OCR pass and a limited later-migration check; 2019 onward remains a targeted search requiring a final rerun. Include England and Wales PFD reports where a coroner explicitly identifies an EHR, digital record, interoperability, electronic messaging, alerting, tasking, display, health IT workflow, or transferable paper–electronic boundary in the circumstances, concern, or preventive action. The narrow telecare boundary includes a system only when the report expressly concerns its care-information, digital-workflow or health-IT design mechanism; a generic alarm or property technology mention remains out of scope.

Discipline

1. Record discovery leads separately from included reports.
2. Prefer the Judiciary report and official recipient response.
3. Assign every primary report a stable source ID; assign a CLM- ID when explicit claim traceability is required.
4. Extract the coroner's concern, the digital mechanism, organisations, response, and limitations.
5. Double-check borderline inclusions and do not infer causation beyond the report.
6. Track search terms, dates, exclusions, duplicates, and unavailable records.
7. Treat current-system claims separately from historical report findings.
8. Use report reference for deduplication; do not count official responses separately.
9. Preserve causal caveats, especially when a report identifies future risk without finding that the digital mechanism caused or contributed to the death.
10. Generate descriptive counts from the canonical CSV rather than manually transcribing them.
11. Keep controlled mechanism, workflow, setting and boundary facets separate from free-text extraction.
12. Record coroner statements, recipient accounts, reviewer coding, synthesis and Defence translation as different statement types.
13. Map report-level evidence before drawing cross-case themes; look for variation and counter-considerations as well as recurrence.
14. State the last search, last integration, open rechecks and next update on reader-facing status pages.
15. Preserve a publication-date/report-date distinction and audit later-created or migrated pages separately; see the 2018 search audit.

Extraction confidence is operational rather than global: high means the full primary report was checked and the digital mechanism is explicit, with its causal qualification retained; medium means the source is linked but the digital mechanism, remedy or causal relevance is indirect, contested or substantially dependent on respondent context. Neither label is an independent second-reviewer check, a certainty-of-evidence grade or a measure of source authority.

Evidence model

The authority chain is:

primary source → included report → claim → hazard mechanism → candidate requirement → assurance test

Generated catalogues, tags, maps and graphs are derived routes. They do not add evidence authority.

Synthesis model

• Inductive layer: specific digital failure descriptions extracted from reports.
• Deductive layer: the nine sociotechnical codes in the coding framework.
• Context facets: controlled workflow, care setting and boundary groups retained alongside narrative descriptions.
• Cross-case synthesis: recurring mechanisms, variations and limitations within the bounded corpus.
• Translation: Defence requirements and tests labelled as analytical until validated with DPHC/CORTISONE evidence.

No aggregate count is interpreted as incidence, prevalence, causation or comparative product safety.