Findings
This section synthesises the current focused series by sociotechnical mechanism: how information, work and responsibility can fail across a digital care system. It is not organised by product, clinical specialty or publication date because those groupings can obscure the safety control that needs to be tested.
How to read these findings
The six primary groups combine closely related codes from the sociotechnical coding framework. Human factors is treated separately as a cross-cutting lens rather than as a competing explanation.
| Finding group | Codes | Safety question |
|---|---|---|
| Information visibility and decision support | VIS, ALERT |
Does the right information become perceptible and actionable in the view and workflow being used? |
| Closed-loop work and action ownership | LOOP |
Does a safety-relevant item acquire an owner, deadline, acknowledgement, escalation and closure state? |
| Interoperability, identity and continuity | INT, IDPOP |
Does information follow the correct person, cohort and episode across organisational and technical boundaries? |
| Record integrity and provenance | PROV |
Can users tell what was recorded, changed, transferred or omitted, by whom and when? |
| Resilience and degraded operation | RES |
Does safe work remain possible during outages, delay, after-hours operation and manual fallback? |
| Governance and assurance | GOV |
Are safety controls specified, configured, monitored and preserved across organisations and change? |
| Human-factors lens | HF |
How do interface, workload, staffing, expectations and workarounds shape every mechanism above? |
Codes are non-exclusive. The same report can support more than one mechanism because, for example, information may be hard to see, never become an owned task and also fail to cross an organisational boundary.
Recurrence is not prevalence
The included-case dataset is a targeted series of published Prevention of Future Deaths reports, not a random or exhaustive sample of care. The current review counts demonstrate recurrence within that series; they do not estimate incidence, prevalence, product risk or the probability that a mechanism contributed to a death. Counts overlap and cannot be added together.
Each finding page therefore:
- anchors the mechanism in named reports from the included dataset;
- distinguishes the coroner's concern, respondent evidence and this review's interpretation;
- records important variation and counter-considerations;
- links representative narrative case studies where available; and
- labels Defence implications as analytical translation, not coronial findings.
Evidence path
Read from the source register and included-case dataset, through the coding framework, to these findings. The representative case studies provide readable anchors but are not the complete corpus.