Durgical
EN
Open the app

Learning from incidents: reporting, root cause analysis and barriers

  • Concept

A report is only useful if it changes something. Analysing incidents well turns one bad day into a stronger system for every later patient.

Updated

What it is

An incident reporting and learning system collects reports of safety events, analyses the most serious or frequent ones and feeds back changes that reduce risk. Analysis uses a systems view (the Swiss cheese model), a structured method such as root cause analysis or the London Protocol, and a just culture that separates human error from reckless choices. The output is a set of barriers, ranked by strength, plus open communication with the patient.

Key facts

  • Report incidents, near misses and adverse events; near misses show the same holes at no cost.
  • Decide which events get a full analysis: severe harm, high risk of recurrence or new types of failure.
  • The analysis team includes people from the units involved but not those directly involved in the event, plus someone trained in the method.
  • Gather the record, interviews and the setting; build a timeline before asking why.
  • Care delivery problems are what went wrong in the care; contributory factors are why (patient, task, individual, team, environment, organisation).
  • Each action has an owner, a date and a measure; feedback goes to the reporter and the team.

Worked example

  1. A made-up laboratory finds 14 mislabelled blood samples per 10 000 in a year; one led to a near-miss transfusion of the wrong group.
  2. Chronology: samples from 2 patients with the same surname were labelled at the desk after drawing, in a batch of 6 tubes.
  3. Contributory factors: labels printed in batches (task), no bedside labelling (organisation), a night shift with 1 nurse for 18 patients (team).
  4. Actions ranked: bedside labelling with a wristband scan (strong), 2 identifiers at the bedside (medium), reminder posters (weak).
  5. The strong and medium actions are implemented; the posters alone are dropped as too weak.
  6. One year later: 3 mislabelled samples per 10 000, a fall of (14 − 3) ÷ 14 ≈ 79 %, with no wrong-group near misses.

Just culture: the response depends on the behaviour, not the outcome

BehaviourExampleResponse
Human errorPicks the wrong vial from two identical-looking onesConsole; redesign the system
At-risk behaviourSkips the second identifier because everyone does on busy nightsCoach; remove the incentive to drift
Reckless behaviourKnowingly ignores a substantial risk, such as working impairedSanction, fairly and in proportion

Markers

  • Swiss cheese model
  • Just culture
  • London Protocol
  • Barrier hierarchy
  • Learning system

In practice

  • When you witness an incident, make the patient safe first, then report it the same shift with facts, not opinions.
  • Disclose early to the patient and family: what happened, what is being done and who to contact; keep a note of the conversation.
  • Support the clinician involved: the ‘second victim’ needs help, not isolation.
  • In meetings, ask ‘what made this possible?’ rather than ‘who did it?’

Common pitfalls

  • Stopping the analysis at the person: ‘retrain the nurse’ leaves the latent condition for the next patient.
  • Recommending only weak actions (training, memos), because they are easy to approve.
  • Collecting reports without feedback: when nothing changes, staff stop reporting.
  • Judging the behaviour by the outcome: the same slip deserves the same response whether or not harm followed.

Common traps

  • Retraining is a strong corrective action: training and warnings sit at the bottom of the barrier hierarchy; design changes and forcing functions work even when people are tired
  • Just culture means no one is ever accountable: it consoles error and coaches drift, but reckless disregard of a substantial risk is still sanctioned
  • A root cause analysis finds the single root cause: serious events have several contributory factors at different levels, and each needs its own action
  • Telling the patient must wait until the analysis is finished: disclose the facts early and update the patient as the analysis progresses

Test yourself

Think of the answer, then tap the question to see it.

A nurse gives a medicine from a vial that looks identical to another on the same shelf. Under just culture, what is the response?

Human error: console the nurse and redesign the system, for example by separating or relabelling look-alike vials. Punishing would not remove the hazard.

An analysis proposes: a memo, a training session and a barcode check before dispensing. Which is the strongest barrier?

The barcode check: it is a forcing function built into the process. Memos and training rely on memory and attention, the weakest barriers.

Staff report many incidents but never hear what happened next. What will happen to reporting, and why?

It will fall: reporting takes effort, and without feedback or visible change staff conclude it is useless. A learning system must close the loop.

In the London Protocol, what is the difference between a care delivery problem and a contributory factor?

The care delivery problem is what went wrong in the care (a sample labelled away from the patient); contributory factors explain why (batch printing, staffing, no bedside scan).

Add to my cards (4)

Elsewhere in DurgiPedia

Connections

← All of epidemiology

Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.

Sources: Patient safety incident reporting and learning systems: technical report and guidance (World Health Organization, 2020); Global Patient Safety Action Plan 2021–2030 (World Health Organization, 2021); Medical Error Prevention and Root Cause Analysis (StatPearls (NCBI Bookshelf), 2026).

Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake

Representative model, for educational purposes only.