Definition
A learning response is the structured activity an organisation runs after a patient safety incident to understand what happened and improve care. Under PSIRF the recognised responses include the after action review, the swarm huddle, the multidisciplinary team review and the patient safety incident investigation (PSII), supported by NHS England's learning response toolkit.
Source: NHS England learning response toolkit, August 2022 · Last reviewed 6 July 2026
Choosing a proportionate response
Under the old Serious Incident Framework, one method fitted all: a root cause analysis for anything crossing the "serious" threshold. PSIRF asks a different question. What is the learning opportunity, and what response fits it? The answer is set in advance in the organisation's patient safety incident response plan, which maps its main incident types to intended responses.
Proportionality cuts both ways. A cluster of low-harm medication incidents might justify a full investigation because the learning potential is high. A serious outcome from a risk that is already well understood, with improvement work under way and safety actions being monitored, may justify no individual response at all. NHS England's guidance says that explicitly, and it is the change that most surprises teams arriving from the old framework.
A small set of incidents keep a nationally mandated response regardless of the local plan, including deaths thought more likely than not due to problems in care, which always require a PSII.
The learning responses
- 01After action reviewthe workhorse: a facilitated team discussion built on four questions
- 02Swarm huddlesame-day learning at or near where the incident happened
- 03MDT reviewbreadth across professions, teams and pathways
- 04PSIIthe deepest response, reserved for the greatest learning potential
After action review (AAR)
The after action review is the workhorse of PSIRF, and for many organisations the most used response. It is a facilitated team discussion built around four questions: what was expected to happen, what actually happened, why was there a difference, and what can we learn? The method predates healthcare (it was developed for military debriefs) and earns its place through speed and candour. A typical AAR takes an hour or two with the people who were there, facilitated by someone trained in the method, and completes within days or a couple of weeks of the incident.
An AAR works best when the event is recent, the team is intact and the ground to cover is contained: a deteriorating patient spotted late, a medication error caught at the last check, a discharge that went wrong. The output is a short record of the learning and any actions, not a lengthy report. Because it examines the gap between expected and actual, it surfaces the workarounds and system pressures that written procedures hide, which is exactly the material safety improvement needs.
Swarm huddle
The fastest response in the toolkit. Immediately after an incident, staff gather at or near where it happened to establish what occurred, why it may have occurred, and what needs to change now to reduce risk. The value is in the immediacy: memories are fresh, the environment is unchanged and unsafe conditions get addressed the same day rather than after a report lands. A huddle typically runs 15 to 30 minutes, led by someone trained to keep it systems-focused rather than person-focused.
Multidisciplinary team review
A structured review bringing together the professions involved in the pathway, suited to incidents that cross team or departmental boundaries, or to reviewing a theme across several incidents. An MDT review takes longer to convene than an AAR (weeks rather than days) and trades speed for breadth: it can examine how handoffs, escalation and shared processes behaved across the whole pathway rather than within one team.
Patient safety incident investigation (PSII)
The deepest response: a formal, system-based investigation led by trained investigators, reserved for incidents with the greatest learning potential and for those where it is nationally required. PSIIs replaced the root cause analysis report, and they differ from it in method as well as name. The PSII page covers scope, timescales and how the investigation runs.
SEIPS, the model behind the tools
The toolkit's methods share a common analytical base: SEIPS, the Systems Engineering Initiative for Patient Safety. SEIPS describes care as a work system made up of people, tasks, tools and technology, the environment and the organisation, all interacting to shape how work actually gets done. The learning response tools use it to keep attention on those interactions rather than on the last person who touched the incident.
The practical consequence is a shift in the questions asked. "Who made the error?" becomes "what made that error likely, and what usually catches it?" The distinction between work-as-imagined (the procedure) and work-as-done (the wet Tuesday night reality) runs through every tool in the kit, and NHS England's versions were developed with human factors specialists for that reason.
Recording the outputs
Whatever the response, the learning only counts if it is captured and tracked. Incidents themselves are recorded through the national Learn from Patient Safety Events (LFPSE) service, and the response outputs (findings, safety actions and their owners) need a home where completion can be monitored, because supportive oversight under PSIRF looks at whether safety actions worked, not whether a report was filed. Templates for each response type sit in NHS England's toolkit, covered on the PSIRF in practice page, and the people running them need the training described on the training page.