Writing a Hazard Log That Actually Works
Beyond the template. How to identify, articulate and quantify clinical hazards so your DCB0129 hazard log withstands audit — and actually prevents harm on the ward.
DCB0129ISO 14971
For NHS Clinical Safety Officers
Practical clinical safety training for NHS AI deployment.
DCB0129, DCB0160 and DTAC applied where they actually get applied — on the ward, in procurement, under real deployment pressure. Clinician-led. Independent. Built for the work, not the certificate.
The gap
DCB theory is not enough for real-world AI deployments.
Certified CSOs understand the standards. Where training falls short is application: writing a hazard log that stands up to audit, challenging a supplier safety case that reads like marketing, and recognising where AI systems violate the deterministic assumptions DCB0129 was built on.
Every NHS Trust deploying AI is quietly discovering the same thing — the controls that worked for traditional health IT do not cleanly transfer to systems that learn, drift, and fail in distributional ways. This resource closes that gap, module by module, scenario by scenario.
The three launch modules
Where most CSOs get stuck — and what to do about it.
Evaluating a Supplier Safety Case
Reading between the lines of a DCB0129 Safety Case Report. What questions to ask a vendor. What absence of evidence actually means when you sign the Clinical Safety Case.
DCB0129DTAC
Why AI Breaks DCB Standards
DCB0129 assumes deterministic software behaviour. AI systems don't deliver it. What changes when you are deploying a model — drift, distributional shift, opacity — and how to adapt your safety case.
DCB0129BS AAMI 34971MHRA AIaMD
Launch simulator
The Hazard Log Builder.
Work through a realistic AI deployment scenario. Identify hazards, apply controls, and generate a DCB0129-aligned hazard log entry by entry. Structured feedback at every step — modelled on how an experienced CSO would challenge your thinking.
- Realistic NHS deployment scenarios, not sanitised case studies.
- Identify, articulate, and quantify clinical hazards with structured feedback.
- Export an audit-ready hazard log for review and iteration.
Hazard Log — draft
Scenario 01
- Hazard
- AI triage model deprioritises frail elderly patients with atypical presentation.
- Clinical consequence
- Delayed recognition of sepsis. Severity: 4. Likelihood: 3.
- Control
- Triage nurse override required for any patient > 75 with NEWS2 ≥ 5.
- Residual risk
- Acceptable with monitoring
Frameworks covered
The standards CSOs actually have to apply.
Every module is anchored to the frameworks that define clinical safety for NHS digital health. We cover twelve — from the familiar to the emerging.
DCB0129
Clinical Risk Management — Manufacturer
Defines the obligations on health IT manufacturers to operate a clinical risk management system and produce a Safety Case.
DCB0160
Clinical Risk Management — Deployment
The deploying organisation's counterpart to DCB0129. Governs how Trusts implement and monitor health IT safely.
DTAC
Digital Technology Assessment Criteria
NHS England's baseline assessment covering clinical safety, data protection, technical assurance, interoperability and usability.
PSIRF
Patient Safety Incident Response Framework
The NHS's approach to learning from patient safety incidents. Replaces the Serious Incident Framework.
ISO 14971
Risk Management for Medical Devices
The international standard for applying risk management across the medical device lifecycle.
BS AAMI 34971
AI Risk Management Guidance
Guidance on applying ISO 14971 to machine learning enabled medical devices, addressing AI-specific failure modes.
MHRA AIaMD
AI as a Medical Device
The MHRA's evolving regulatory approach to software and AI as a medical device in the UK.
UK GDPR / DPIA
Data Protection
Data Protection Impact Assessments where AI processing presents risks to patient rights and freedoms.
NICE ESF
Evidence Standards Framework
NICE's framework for evaluating digital health technologies, including AI-driven tools.
EU AI Act
High-Risk AI Regulation
EU-wide regulation of high-risk AI systems, including health applications. Relevant to UK suppliers operating cross-border.
AMLAS
Assurance of Machine Learning for Autonomous Systems
A structured argumentation-based methodology for assuring the safety of machine learning components.
ECSF
England Clinical Safety Framework
Emerging national framework coordinating clinical safety practice across NHS England.
Waitlist
Get the launch modules when they go live.
One email when Module 5, Module 7 and Module 11 are published. No newsletter. No marketing. You can leave at any time.
Form placeholder — not yet connected to a backend.
Disclaimer. Independent educational resource — not affiliated with NHS England or any regulatory body. For educational use only.