What is human factors in healthcare?
Human factors is the study of how people interact with their work: the tasks they do, the equipment and IT they use, the people around them, the physical environment and the wider organisation. The aim is simple. Design work so that the right thing is easy to do and the wrong thing is hard to do.
Healthcare borrowed much of this thinking from aviation, rail and nuclear power. Those industries accepted early that error is a normal part of human performance. Even the most careful expert will misread a label or forget a step at some point. Safety does not come from expecting perfection. It comes from systems that expect error, catch it early and limit the damage.
In the NHS, human factors underpins the Patient Safety Strategy and the Patient Safety Incident Response Framework (PSIRF) in England. It is also why regulators such as the GMC, NMC, HCPC and GPhC expect professionals to report incidents and take part in learning, rather than simply be more careful.
What is the Swiss cheese model?
James Reason described an organisation's defences as a stack of cheese slices. Each slice is a layer of protection: a prescribing system, a pharmacist's review, a second check, a nurse's knowledge, a patient wristband. Every slice has holes. Some holes are short-lived, such as a moment of distraction. Others sit there for months, such as a confusing screen design.
Most of the time a hole in one slice is covered by solid cheese in the next. A wrong dose is spotted by the pharmacist, or the patient says, "I'm allergic to that." Harm happens when holes in several slices line up at the same moment and the hazard passes straight through.
Active failures and latent conditions
Reason split the holes into two kinds. Understanding the difference is the most useful thing you can take from the model.
| Type of hole | What it means | Healthcare examples |
|---|---|---|
| Active failure | An unsafe act by someone at the sharp end, with an immediate effect | Giving a drug to the wrong patient; selecting the wrong dose from a drop-down list; missing a deteriorating observation |
| Latent condition | A weakness built into the system, often by decisions made earlier and away from the bedside, which can lie dormant for a long time | Chronic short staffing; look-alike packaging; no pharmacist cover at weekends; poor locum induction; alerts that fire so often they are ignored |
Latent conditions set the trap. Active failures spring it. If learning after an incident only targets the active failure, for example by retraining one nurse, the latent conditions stay in place and the same incident is likely to happen again with someone else.
Swiss cheese model examples in healthcare
The model makes most sense when you walk through a real-looking incident layer by layer. The two examples below are fictional but typical.
A patient with a documented penicillin allergy is given co-amoxiclav on a Saturday night. The allergy is written in the paper notes but not on the electronic prescribing system, because the two systems are not linked (latent condition). The prescriber is a locum who had no induction to the system (latent condition). There is no ward pharmacist until Monday (latent condition). The nurse giving the dose is covering two bays and does not ask the patient about allergies before administering (active failure, in a context of heavy workload). Four slices, four holes, all lined up.
A post-operative patient's early warning score rises over six hours. The observations are recorded on paper, so the electronic escalation alert never fires (latent condition). The healthcare assistant recording the observations is new and unsure when to escalate (latent condition: training and supervision). The registered nurse notices the score but assumes the doctor has been told at handover (active failure linked to an unstructured handover). The on-call doctor is covering three wards (latent condition). By the time the patient is reviewed, they need critical care.
In both cases, a single action at the bedside could have stopped the harm. That matters, and the individuals involved should reflect on it. But in both cases the organisation had also left holes that made the error more likely and less likely to be caught.
Why do skilled people make errors? Slips, lapses, mistakes and violations
Reason also classified unsafe acts into types. Naming the type helps you pick a fix that actually works, because each type has a different cause.
| Type | What goes wrong | Clinical example | What helps |
|---|---|---|---|
| Slip | Attention: the right plan, carried out wrongly | Picking up the wrong syringe while being interrupted | Fewer interruptions, design that makes the wrong action hard |
| Lapse | Memory: a step is forgotten | Forgetting to restart a feed after a scan | Checklists, prompts, electronic reminders |
| Mistake | Judgement or knowledge: the plan itself is wrong | Prescribing from an out-of-date guideline | Training, decision support, supervision, easy access to current guidance |
| Violation | A deliberate departure from a rule, usually not meant to harm | Skipping a second check because there is never time | Find out why the rule is not followed and fix the conditions; address reckless or repeated violations |
Most violations in healthcare are routine (corner-cutting that has become normal) or situational (the situation makes the rule almost impossible to follow, such as a required second check with only one registrant on shift). Reckless violations, where someone knowingly takes an unjustifiable risk, are uncommon and are treated very differently.
Performance-shaping factors
- Fatigue, especially late in a long shift or on nights.
- Hunger, dehydration and missed breaks.
- Stress, which narrows attention and can cause tunnel vision.
- Interruptions and multitasking during high-risk tasks.
- Unfamiliar equipment, wards or procedures.
- Weak communication at handover or with verbal orders.
Simple self-checks such as HALT (Hungry, Angry, Late, Tired) help you notice when your own slice has bigger holes than usual, so you can slow down or ask for a second check.
Does the Swiss cheese model mean nobody is to blame?
No. This is the most common misunderstanding. You are one of the slices. Your checks, your knowledge and your willingness to speak up are real defences, and regulators expect you to own your part. A good analysis names both the system weaknesses and the individual actions that could have stopped the hazard.
This balance is the idea behind a just culture. NHS England's just culture guide helps managers decide whether a person needs support, learning or a formal process. It asks questions about deliberate harm, health, whether safe procedures existed and were workable, whether another person would have acted the same way, and whether there were mitigating circumstances such as excessive workload. Deliberate harm and reckless conduct are handled very differently from honest error.
Which fixes actually work? The hierarchy of controls
Once you know why the holes lined up, choose actions that really close them. Human factors specialists rank interventions by strength. Notice where "staff reminded to be vigilant" sits.
| Strength | Type of action | Example |
|---|---|---|
| Strongest | Remove or design out the hazard | Taking concentrated potassium out of general ward stock |
| Strong | Forcing functions and automation | An e-prescribing hard stop for a drug the patient is allergic to |
| Moderate | Standardise, simplify, use checklists | One model of infusion pump across a hospital; the WHO Surgical Safety Checklist |
| Moderate | Independent checks | Independent second checks for high-risk medicines |
| Weaker | Warnings, labels, reminders | Posters and alert stickers |
| Weakest | Training, policy, telling people to be careful | Re-issuing a policy |
Training is still needed, but on its own it relies on the same memory and attention that failed in the first place. Strong action plans mix personal learning with at least one system change higher up the list.
How to use the Swiss cheese model in a reflection
If you have been involved in an incident, the model gives your reflection a clear structure. It shows that you understand both your own role and the system around you, which is a large part of what regulators mean by insight.
- Describe what happened factually: Say what you did and what the result was, without patient-identifiable details and without minimising.
- Name the type of error: Was your action a slip, lapse, mistake or violation? Say why. Note any performance-shaping factors such as fatigue or interruptions.
- Map the slices: List each defence that should have caught the problem and the hole in it. Separate active failures from latent conditions.
- Own your slice: State clearly what you could have done differently and how your behaviour may have weakened other defences, for example by discouraging a colleague from calling you.
- Set out changes at two levels: Describe specific personal changes, and what you have done or proposed to fix the system, using the hierarchy of controls.
- Show evidence over time: Add supervision, audit or feedback that someone else has verified, and revisit the reflection after a few months.
I gave the wrong dose because the ward was short-staffed and the pharmacist wasn't there. The system let me down. I will be more careful.
I gave a dose ten times higher than prescribed. It was a slip: I was interrupted twice while drawing it up and did not re-check the label before giving it. Two other defences failed, as there was no second checker on shift and the stock strengths look almost identical. I told the patient, reported the incident and took part in the review. I now use the do-not-disturb tabard for every high-risk medicine, and I raised the packaging issue, which led pharmacy to separate the two strengths. My ward manager has observed three of my medicine rounds.
Learning more about human factors
The Swiss cheese model is a starting point, not the whole of human factors. Related tools include SEIPS (a systems model now widely used in NHS learning responses), structured communication such as SBAR, read-back of verbal instructions, briefings and debriefings, and safety huddles. If you work in Scotland, Wales or Northern Ireland, check your local incident framework, as PSIRF applies in England.
If you want a structured way to apply these ideas to your own practice, our course Human Factors and Learning from Incidents covers the Swiss cheese model, error types, just culture and how to build a reflection and remediation plan. If your incident involved medicines or documentation, Medication Errors: Reflection and Remediation and Record Keeping and Documentation: Remediation Course go deeper on those areas. If you are under investigation, also seek advice from your defence organisation, trade union or legal representative. Completing a course does not decide the outcome of any process.
Frequently asked questions
Who created the Swiss cheese model?
The model was developed by James Reason, a British psychologist who studied human error. He set it out in his work on human error in the 1990s and in a widely read BMJ paper in 2000. It was designed for high-risk industries in general and has since become one of the most used patient safety models in healthcare.
What do the holes in the Swiss cheese model represent?
The holes are weaknesses in each layer of defence. Some are active failures, such as a slip by a clinician. Others are latent conditions, such as understaffing, confusing equipment or missing training. Holes open, close and move over time. Harm occurs when holes in several layers line up so a hazard passes through every defence.
What is the difference between active failures and latent conditions?
Active failures are unsafe acts by people at the front line, with immediate effects, such as giving a drug to the wrong patient. Latent conditions are weaknesses built into the system by earlier decisions, such as rota design, look-alike packaging or poor IT. Latent conditions make active failures more likely and less likely to be caught.
What are the 4 types of human error in healthcare?
A common classification is slips (attention failures during a familiar task), lapses (memory failures, such as forgetting a step), mistakes (the plan itself is wrong because of knowledge or judgement), and violations (deliberate departures from a rule, usually without intent to harm). Each needs a different fix, which is why naming the type is useful.
What are the criticisms of the Swiss cheese model?
Critics say it can make incidents look like a straight line of events when real healthcare is more complex and adaptive. It can also be misused to shift blame onto the system. Many NHS learning teams now pair it with systems models such as SEIPS and ideas like work as imagined versus work as done.
How do I use the Swiss cheese model in a reflective account?
Describe the incident factually, name your error type, then list each defence that should have caught it and the hole in it. Be clear about your own part as well as system issues. Finish with specific personal changes, any system improvements you contributed to, and evidence such as supervision or audit.
Is human factors training mandatory in the NHS?
In England, the NHS patient safety syllabus has free e-learning, with level 1 (Essentials of patient safety) aimed at all NHS staff and covering systems thinking and human factors. Whether a module is mandatory for you depends on your employer. Check your trust's training requirements and what your regulator expects for your role.
Sources and further reading
- Human error: models and management (2000). BMJ.
- The NHS Patient Safety Strategy. NHS England.
- Patient Safety Incident Response Framework. NHS England.
- A just culture guide. NHS England.
- SEIPS quick reference guide and work system explorer. NHS England.
- Good Medical Practice (2024). General Medical Council.
Practitioner Insight is an independent education provider. It is not affiliated with or endorsed by the GMC, MPTS, NMC, HCPC, GPhC or any NHS organisation, and this information is not legal advice. If you have been referred to a regulator, speak to your defence organisation, trade union or legal representative.