Supplied reference-document audit

Conclusion

The supplied Reg 28 PFD Reference material.docx is useful as a discovery and analytical prompt, but it is not reliable enough to function as source evidence. Its 60 nominal case/profile headings all map to reports that were already present in the included-report dataset, although the Darren Carrington and Valerie Gibson narrative content does not match the named primary reports. At the initial reconciliation, 11 of the then-71 Wiki reports were absent from the document. The same-day update screen added 41 further primary-sourced reports, all also absent from the document, so the current difference is 52. Absence from the document is therefore not evidence that Wiki content was invented.

The reconciliation also found two genuine Wiki metadata errors. The canonical dataset now records Sebastian Daniels under Hampshire, Portsmouth and Southampton, and Jennifer Trigger under North Wales (East and Central), as stated by their official Judiciary pages. No case mechanism was changed on the authority of the supplied document alone.

Source status

The document is registered as SRC-169, a project-supplied secondary synthesis and discovery artefact. The reviewed file is identified by SHA-256 checksum d28e500e3bd5e06f248e36e65bbe105d56ac3ef682f41c124e96c7bd16f20083.

The review rendered all 400 pages and extracted approximately 97,700 words. The analytical framework occupies pages 1–78, introductory material for the case library pages 78–81, and 60 case profiles pages 81–400. The framework has no bibliography or citations. Only 26 of the 60 profiles contain a clickable hyperlink; 167 hyperlink elements resolve to 50 unique URLs, and 33 of those URLs carry a utm_source=chatgpt.com parameter. Those parameters are evidence of the discovery route, not evidence for the underlying propositions.

The document contains no comments, tracked-change record, footnotes or endnotes. Its named author and last modifier are Alexander Nicholls; its core properties record creation on 21 July and modification on 22 July 2026. These properties identify the file, not the authority of its claims.

Bidirectional case reconciliation

All 60 document profile identities or headings have a corresponding included Wiki report after allowing for spelling and metadata errors. That identity match does not validate the profile narrative: Darren Carrington and Valerie Gibson contain substantive wrong-case content. The document added no previously unknown report.

The following 11 reports formed the initial Wiki-only set against the then-71-report corpus:

Report	Registered primary source
Caliel Smith-Kwami (`UNREF-2018-CALIEL-SMITH-KWAMI`)	`SRC-106`
Pauline Pryor (`2018-0009`)	`SRC-107`
David Buttriss (`2018-0010`)	`SRC-108`
Grahame Searby (`2018-0162`)	`SRC-115`
Joan Lunt (`2018-0164`)	`SRC-116`
Michael Drewell (`2018-0259`)	`SRC-121`
Paul Price (`UNREF-2018-PAUL-PRICE`)	`SRC-126`
Karl Willis (`2018-0256`)	`SRC-128`
Morris Reddington (`2021-0312`)	`SRC-004`
Avery Hall (`2026-0048`)	`SRC-056`
Francis Barnes (`2023-0417`)	`SRC-059`

These are independently checkable in the source register and included-report dataset. The later update added another 41 official-primary-source reports, bringing the current Wiki-only set to 52; they are listed in the generated all-reports catalogue, and their update dispositions are recorded in the screening log. The detailed update audit is retained at research/audits/judiciary-update-screen-2026-07-22.md. The document is not used to validate any of them.

Conflicts resolved against primary sources

Profile	Document problem	Primary-source disposition
Darren Carrington, p. 211	Described as a 2024 hospital discharge-summary case, without an exact report reference or source link	The official report is `2018-0181` and concerns prescribing alerts, online repeat requests and pharmacy supply (`SRC-118`). The document account is a substantive wrong-case narrative.
Valerie Gibson, pp. 134–139	Described as an antipsychotic population-monitoring case with no EHR failure	The official `2025-0630` report concerns contradictory Omnicell and ePMA dispensing/administration records (`SRC-043`).
Marian Grant, from p. 200	Name given as “Marion”; the EPR mechanism is omitted	The official report explicitly concerns EPR VTE-warning timing and unresolved action after repeated record access (`SRC-125`; `CLM-026`).
Theo Tuikubulau, p. 371	Contains conversational drafting residue and presents an algorithm-only explanation as settled	The official report raises a prospective categorisation concern, while the registered NHS England responses give a counter-position that both systems reach category 1 when the full presentation is elicited (`SRC-070`, `SRC-100`, `SRC-101`). Both positions remain visible in the Wiki.
Paula Doreen Hughes, from p. 139	Gives 10 October 2025 and Liverpool/Wirral metadata	The official report gives 14 October 2025 and Inner South London (`SRC-041`). The Wiki metadata was already correct.
Cuthbert Hingert, pp. 205–212	Repeats `2018-0280` while linking an attachment whose filename contains `2018-0279`	The official page/attachment discrepancy remains open as `VAL-018`; the document does not resolve it.

Several apparent date conflicts were publication-date/report-date confusions. Primary-source rechecking confirmed the existing report dates for Steven Welch, Brian Bicat, Jacob Sulaiman, Akhona Moyo and John Spencer, and the existing coroner area for Greg Hutchins. The audit did identify and correct the two coroner-area errors described above.

Provenance reconfirmation

The reconciliation checked both link structure and substance:

every then-included row had a registered source identifier and its source_url matched the URL in the source register; the same automated linkage checks apply to the expanded dataset;
every structured claim source, report and mechanism reference resolved;
every case used by a candidate requirement existed in the included dataset, and every requirement/test reference resolved;
conflict cases were reopened at the official Judiciary page or report rather than decided by whichever secondary narrative appeared more detailed.

These checks establish traceable provenance, not independent replication of every extraction. Second-reviewer checks remain open in the validation queue.

Safe use of the document

The document may be used to suggest a question, lifecycle stage or search lead. It must not be used to establish a coronial finding, current product behaviour, Defence-system fact, causal conclusion or validated programme requirement.

Its eight-stage information lifecycle has been retained only as a labelled analytical lens in the information-lifecycle worksheet. Its case narratives, confidence labels, hazard lists and “Defence relevance” passages are not imported as findings. New source and response leads are independently verified before registration.

The detailed reconciliation record is retained in research/audits/reference-document-reconciliation-2026-07-22.md.