Evidence assurance and uncertainty

This generated view assembles the current assurance position from canonical report, claim, requirement, test and validation records. It is a disclosure dashboard, not a quality score. A linked primary source establishes provenance for what the report states; it does not remove extraction, causal, coverage, response or transferability limits.

Corpus profile · Selection and coverage flow · Counter-positions · Validation queue

Current assurance position

Measure	Current state	Boundary
Included reports with a unique registered primary source	112/112	Source presence is not independent second review
High-confidence extractions	108/112	Project extraction confidence, not causal certainty
Qualified extractions	11/112	Explicitly named below
Medium care-pathway classifications	8/112	Single-reviewer navigation coding
Reports with registered response letters	63/112	Letter presence is not adequacy, implementation or effectiveness
Registered response letters	107	Each is a separate official attachment
Stable claims / reports covered	28 / 22	Selective traceability, not one claim per extraction
Candidate requirements	13	All remain Defence-unvalidated
Designed assurance tests	13	All remain unexecuted
Registered counter-positions or anomalies	10	Preserved without adjudication by preference
Open or partial validation items	16/22	Canonical status remains in the validation queue

Qualified and medium-confidence records

Qualified extractions

Report	Primary source	Extraction confidence	Recorded qualification
Amy Chapman — 2026-0247	`SRC-182`	high	The coroner found gross failures of basic care around record checking and leave decisions; the report raises the alert and audit design as ongoing concerns but does not isolate SystmOne as the cause. The source describes the Haven Unit as an informal community-based placement.
Kiefer Fraser-Phillips — 2026-0216	`SRC-185`	high	The death was recorded as sudden unexplained death in schizophrenia; the report presents observation-recording resilience as a future risk rather than a causal finding. The Judiciary page says 2026-04-14 while the embedded PFD is signed 2026-04-15.
Gareth Chumber-Kelly — 2026-0073	`SRC-193`	high	The jury identified inadequate medical-record review and continuity among multiple probable contributing failures; the report does not isolate the paper or email mechanism. The Judiciary page says 'North London', while the embedded signed PFD says 'Inner North London'.
Nigel Feckey — 2026-0047	`SRC-197`	medium	The digital mechanisms are explicit in the jury narrative and circumstances, but the formal Regulation 28 concern and requested action relate only to offence-neutrality policy; the report found the prisoner was primarily let down by the custodial system
Alan Peet — 2025-0609	`SRC-221`	high	The report says it was likely entries were made under another nurse's details and does not establish exactly who entered each item; the death also involved staffing, training and clinical-care failures
Alan Mitchell — 2025-0577	`SRC-203`	medium	The coroner expressly stated that removal of the prescription played no causal part in this death. Optum's response disputes the reported default mechanism: it says repeat medication does not expire automatically unless an authorised user enables and configures that function, after which the item moves from Current to Past. The mechanism remains a recorded source disagreement
Amy Cross — 2025-0531	`SRC-219`	high	The jury concluded natural causes; the report identifies the record-transfer gap as future risk rather than finding it caused death. Page metadata gives 2025-10-22 while the report is signed 2025-10-20
James Cochrane — 2025-0454	`SRC-209`	high	The report contains extensive clinical-assessment and service concerns; mobile video and SystmOne assurance are future-risk issues and not isolated causal findings
Venetia Pierce — 2025-0427	`SRC-210`	medium	The report found nitrofurantoin-induced pneumonitis more likely than not led to death. EMIS's response says patient-specific warnings were displayed, including an age-related warning to monitor lung and liver function, and that MHRA alerts remained accessible through BNF and Safety Check. This counter-position qualifies, but does not erase, the coroner's concern about the specific alert presented in the prescribing workflow
Myles Scriven — 2025-0357	`SRC-215`	high	Neglect contributed to death through lack of adjustments; the EPR entry itself was accurate. Page metadata gives 2025-07-11 while the embedded report is signed 2025-07-14
Francis Barnes — 2023-0417	`SRC-059`	medium	Delayed transfer and surgery contributed to death; the report did not find the absence of an electronic referral system causal. The response retained telephone referral for emergencies and proposed a Microsoft 365 documentation form

The 4 overall medium-confidence extractions are: Nigel Feckey — 2026-0047, Alan Mitchell — 2025-0577, Venetia Pierce — 2025-0427, Francis Barnes — 2023-0417.

Medium care-pathway classifications

Report	Primary source	Classification note
Amy Chapman — 2026-0247	`SRC-182`	The PFD expressly describes the Haven Unit as an informal community-based placement at Millview Hospital; the physical configuration is not further resolved, so it is not coded as an inpatient unit.
Alan Peet — 2025-0609	`SRC-221`	The electronic care and medication entries were created and used inside one nursing care home; the source qualifies authorship as likely rather than certain.
Anthony Card — 2026-0068	`SRC-195`	The police report was intended for whichever primary or mental-health provider held responsibility; the source does not identify one receiving care service.
James Cochrane — 2025-0454	`SRC-209`	Video and family observations arose in the home and were offered to a community mental-health assessment; the exact SystmOne assurance path is not fully specified.
Chloe Burgess — 2025-0121	`SRC-055`	The report identifies the absence of an interaction alert across primary-care prescribing and community-pharmacy checking, but it does not support a single setting where the safety-relevant work item originated. Action was expected in both contexts, so destination and failure point are coded as multiple while origin remains unclear.
Greg Hutchins — 2018-0129	`SRC-112`	The telephone-triage contact arose in a community mental-health service, was not recorded in RiO and could not be shared rapidly for a patient from outside the area. The canonical extraction includes GP and mental-health services but the primary report does not identify one later accountable destination.
Patricia Heslop — 2018-0102	`SRC-110`	Changes in presentation and care-plan information arose principally in the care home and were needed by care-home staff, primary care and external reviewers. The source does not isolate a single recipient or failure point across the fragmented forms.
David Buttriss — 2018-0010	`SRC-108`	Material risk and recent-contact information was held across GP, out-of-hours, mental-health and acute services. The report does not support one directional pathway, so origin and destination remain multiple; cross-service record access was the predominant failure.

These classifications support browsing only. They remain subject to the independent recoding gate in VAL-020.

Response and counter-position coverage

Dataset response status	Reports
No official response located	4
Not assessed in this dataset	45
Response evidence noted	63

The response catalogue has two distinct gaps: 45 reports are not assessed in this dataset, while 4 record that no official response was located in the existing search. Neither status proves that a recipient failed to respond. The current data is report-level and attachment-level; it does not yet normalise every recipient, deadline, promised action, completion claim or independent implementation check.

Registered counter-position classes:

Class	Records
Compatible clarification	3
Counter-risk	1
Documentary anomaly	1
Material disagreement	5

Review the complete response catalogue · Review counter-positions and anomalies

Selective claim and Defence trace

The claims register contains 28 stable claims covering 22 included reports. It is intentionally selective: report-specific fields remain in the included-case dataset, while stable claim IDs are used where explicit cross-case or downstream traceability is needed. 23 claims feed 13 candidate requirements.

Candidate requirement status	Requirements
`candidate_partial_trace_defence_unvalidated`	2
`candidate_trace_complete_defence_unvalidated`	11

Assurance-test status	Tests
`designed_not_executed`	13

No candidate status establishes Defence applicability, approval or implementation. No designed-test status establishes conformance or clinical safety.

Open and partial validation gates

ID	Priority	Task	Needed evidence	Current status
VAL-003	1	Complete and document the current discovery and screening pass	Search log, inclusion/exclusion table, deduplication	Partial — 165 retained candidates logged, 112 included and 53 excluded; the 2018 publication cohort is completely enumerated and extracted in a 379-row manifest, but only 36 cohort rows have retained screening decisions and the other 343 lack row-level contextual dispositions; 2019–2026 discovery remains targeted despite the reproducible 2025–2026 publication-date update screen, and later-migration checking remains limited
VAL-004	2	Verify present-day EMIS/SystmOne screening behaviour	Current vendor/NHS documentation or assurance evidence	Open
VAL-005	2	Test the claim of increasing digital PFD frequency	Yearly counts with stable definition and denominator	Open
VAL-006	1	Evidence DPHC wrong-practice registration risks	DMICP workflow, cohort/search tests, incidents or safety cases	Awaiting sponsor material
VAL-009	1	Resolve conflicting official ICMM submission deadlines	Direct submission portal check and, if needed, organiser confirmation	Open — 3 August vs 14 July 2026
VAL-010	1	Close rapid-review search and release audited counts	Completed screening log, deduplication, full-text decisions and code recount	Current pass complete; final pre-submission rerun remains open
VAL-011	1	Independently recheck inclusion, extraction and mechanism coding	Second-reviewer decisions with adjudication record and source-lineage audit	Open
VAL-012	1	Validate candidate Defence requirements and assurance scenarios	DPHC workflow evidence, CORTISONE baseline, clinical-safety review, representative users and authorised test results	Open — all `REQ-` items are unvalidated and all `AST-` items unexecuted
VAL-013	2	Complete official-response coverage for the remaining reports and implementation follow-up	Judiciary attachments, recipient publications and dated implementation evidence	Open — 49 report rows have no registered response letter; landing-page coverage does not establish that none exists elsewhere, and catalogue status does not infer adequacy or implementation
VAL-014	2	Recheck controlled workflow, setting and boundary facets	Independent facet assignment and adjudication against full reports	Open — current primary facets are conservative single-reviewer browsing aids
VAL-015	1	Re-screen exclusions and borderline reports under the non-circular inclusion rule	Independent full-text decisions based on the digital mechanism, without requiring downstream requirement utility	Open — protocol corrected 2026-07-15
VAL-016	1	Capture a reproducible pre-submission search transcript	Literal queries, surface/database, filters, execution time, result coverage and deduplication output	Partial — the bounded 2018 archive enumeration, attachment hashes, extraction routes and base-family hits are row-reproducible in `2018_publication_cohort_manifest.csv`; complete contextual dispositions for all 379 rows and the wider 2019–2026 literal final-search transcript remain open
VAL-017	1	Complete a full later-migration search for reports dated in 2018	All later-created/migrated Judiciary pages with 2018 report dates, report attachments, OCR where required and row-level dispositions	Open — supplementary `Ref: 2018` search found pages outside the publication-year cohort; only clearly eligible `2018-0405` was integrated in this backfill
VAL-018	2	Resolve Cuthbert Hingert reference discrepancy	Judiciary metadata or corrected official attachment confirming whether `2018-0280` or `2018-0279` is authoritative	Open — page shows `2018-0280`; report-attachment filename shows `2018-0279`; dataset provisionally follows the page
VAL-019	1	Validate the current DMICP deployment and clinical-safety evidence boundary	Current EMIS PCS/DPCS version/adaptations and central/deployed hosting topology; DMICP route across MCN/MODNet and the security boundary; fixed/deployed configuration baseline; authoritative integration/access catalogue distinguishing native interfaces, browser-only services and unavailable national platforms; onboarding/deployment approvals; adopted clinical-safety regime and role allocation; applicable DCB0129 artefacts from every Manufacturer/adaptor/integrator; local DCB0160-equivalent file, plan, Hazard Log, Clinical Safety Case Reports, incident log and residual-risk acceptance	Open — sponsor confirms the product family, MCN terminology, central hosting/client model, browser-only e-RS access and named unavailable national services, but cannot presently confirm the clinical-safety regime or artefacts. These are recorded as unknown, not absent/non-conformant. Public evidence establishes current use, historical lineage/hosting/MODNet access, PDS integration and dated pathology messaging, but does not independently confirm the exact version/adaptations, complete topology/interface catalogue, direct DMICP–MCN path, control effectiveness, Defence applicability or DCB status.
VAL-020	2	Independently validate the care-pathway setting recoding	Second-reviewer check of settings/participants, information origin and destination, non-exclusive origin/destination care contexts, care-setting span, sector boundary, failure point, confidence and note against each primary report, with adjudication of primary/secondary, mental-health, maternity, diagnostic, ambulance, private and secure-care ambiguities	Open — all 112 included rows have a structured single-reviewer care-pathway classification; aggregate counts and directional pathways remain descriptive browsing aids pending independent review.

The living update and release policy defines when these gates require a new evidence release, when an editorial correction is sufficient and what must be disclosed.