This is part 5 of a series of blogs about human factors and investigations in healthcare. The theme is ‘when’ and that covers ‘when’ to investigate and ‘when’ to try any remedies or interventions your investigation data suggests might prevent the incident occurring again. As this blog can be explained by a photo and a graph, we have some time to recap the story so far and, perhaps, predict a bit of the future.
The story so far...
We investigate an incident to collect facts that will prevent the incident from occurring again (see 'Why investigate?' blog). Facts collected by two or more investigators, with enough time away from the ‘day job’, tend to be of better quality than a single person fitting the investigation in and around their other duties (see 'Who should investigate?' blog).
Human factors is a science done by science types who are trained in understanding how the limited ‘cave dweller’ tries to cope with their environment. Human factors types are not likely to have the title ‘Captain’ and have not just landed at Stansted (see 'Human factors' blog).
Facts are our friend as they allow us to tell people why an incident occurred and, if those facts are accurate, allow us to do an intervention that will prevent the incident occurring again (see 'Where do facts come from?' blog). Good facts and great remedies allow us then to monitor the success of the intervention. But again, we are getting ahead of ourselves by talking about interventions. Sorry. At this stage it might be worth thinking about what we do with all those facts, how we see patterns in the data, what a good intervention might look like, and how and when we monitor success.
As we have seen previously, there are four principal areas where facts come from: the human, the equipment, the environment and the system of working. How the investigation is conducted and by whom, and as we shall see ‘when’, affects these four principal areas of investigation and the three methods of intervention.
So, four areas of investigation with facts emerging from many different sources: from inside the witness’s mind, from ward records, from engineering logs, etc. How these facts come together to form a big picture needs to be considered in terms of the intervention. A later blog will explain these interventions (after we discuss data and analysis – yay statistics), but for now it’s worth saying the three interventions are called ‘the three ‘Es’. Luckily the three words all start with the letter E so it makes sense.
Engineering – The most effective intervention, as the machine keeps the cave dweller from making mistakes.
Enforcement – Where someone polices the method of working or the equipment used, in a given environment.
Education – the least effective method, which relies on a training course or a poster. “Don’t operate on the wrong side of a patient”. Well I never, what a useful reminder in a theatre. My favourite was at a rail depot. A poster (1 of 80 in the area) said. “Be alert and check the doors”. Really closing the train doors is a good idea… More in the intervention blogs.
The why, the who, the what data collected will affect the quality of the facts. The facts collected determine the intervention chosen. The monitoring of success of the intervention is, perhaps, determined by the original hypothesis of the investigator, very early on in the investigation. We may do a blog on bias in investigations.
A word of caution. You don’t always do an intervention. ‘Eek’ I hear screamed from every trust. This is because, as we have discussed, an incident is a rare random event with multiple causes. Sometimes an incident, or series of incidents, have occurred due to the random nature of humans and an emerging pattern of data is thought to have been found. This pattern, and these series of events, are, however, just random. So, very early on in this blog, I introduce the idea that ‘when’ an intervention might occur might be never.
Lots of crashes occurred along a three mile stretch of road. Detailed investigations revealed no pattern in any of the crash’s causation. The local authority had over £3 million to spend and was determined to spend it (they rightly want to keep their community safe). Well what intervention would you do given that there is no pattern? There is no consistency in the facts and the only pattern might possibly be in the investigators' minds. Given accidents are rare random events, if you do an intervention will it not make it worse? If it makes it worse, how do you reconcile your ‘no pattern data’?
A comment was made by the local authority that suggested a pattern existed and we were not good at investigations and human factors. We reviewed the data again and conducted interviews with those involved (at our own expense). Indeed, there was perhaps a pattern. If you were female (most were), you were travelling north (most were), you were in the early stages of pregnancy (most were) – you appear to be involved in a crash. We noted this at home visits and it’s not recorded in the police data.
Upon reporting back, the local authority understood that incidents are indeed rare random events and sometimes data emerges with no explanation. The comment from the authority – “So the only intervention is planned parenthood advice a few days before undertaking any northbound journey?” Indeed, that’s the correct conclusion for the data.
No intervention was undertaken, and seven years later no incidents have occurred, and we understand the northbound mummies and babies are doing fine. The local authority remains a client after 18 years. It might be the case that (as my reviewer points out) that “maybe there was a factor there, but it went away without intervention (sleepwalking cattle randomly moved to another field further from the road)”.
Hopefully, that should show the connection between the philosophy of data collection, its method of collection and by who, and how it affects the intervention and prevention. Also, the benefits of planned parenthood when travelling northbound. Hopefully, I’ve rounded up the last four blogs. So it’s now time to look at the when; like parenthood, it affects the outcomes too!
When to investigate?
When to investigate is determined by the facts you want to collect, where those facts come from and whether those facts are time sensitive, and your availability and the accessibility of the location. In broad terms, the ‘when’ is affected by two types of evidence: physical stuff and human witness stuff.
Let’s start with a photograph. (Warning the image below contains graphic depictions of an older man in shorts!)
Image 1: Older man finds the remains of an aircraft converted to a bar and restaurant. Copyright: User Perspective Ltd.
Recovering engineering or physical evidence is less time sensitive than information from witnesses. Ward records can last a long time and engineering logs can as well. If you collect evidence from CCTV – that has a life span of 30 days. Generally, in medicine physical evidence is not time sensitive. However, like the image above, it shows that if you leave evidence for long enough someone will change it. In this case they make it into a restaurant. I eat elsewhere as I was sure a fellow human factor person was looking for the crash site!
Human stuff – witnesses
Most of the facts you collect come from witnesses, aka humans, aka cave dwellers. As we shall see in the ‘how to interview’ blogs, the facts are contained in the mind and it’s not easy to get them out. As you can see in Image 2 below, the decline in the availability of facts is very severe after 20 minutes. In later blogs we can discuss how to interview witnesses and how to get good quality data.
Image 2: The Forgetting curve 1885. Copyright: User Perspective and HM Government (for this version).
The important bit now is to think about the basic processes of human memory, which are:
- Perception – information gets into the mind.
- Encoding – its related to other facts and ‘digitised’.
- Storage – we need to keep it somewhere.
- Retrieval. Unless it can be extracted, it’s not useful.
Each of these stages is associated with a decline in the quality of data and its retrieval is based on the ability of (in this case) the interviewer extracting it. As we are talking about when, the important thing is to get to those memories as quickly as possible and, certainly in medicine, to ensure that witnesses don’t get to chat to each other. If you want evidence from humans – get it quickly and ensure they don’t talk to each other. How quickly? It’s called the golden 24 hours in accident investigation – even though the graph from 1885 suggests a lot shorter time span. Incidentally, the person(s) reporting the incident needs acknowledgement a lot quicker than the 24 hours.
Your availability and access
Ideally you are a human factors person with a ‘go bag’. I’ve several ‘go bags’ that contain equipment needed for each domain (road, rail, security) I work in. The road one has green high vis, the rail has orange. The security one has assorted passes and body armour.
This may be different in medicine. You might not be the first person called, and you work shifts, the chances of a call in the middle of the night is most likely rare. In other domains access is aided by blue lights and the possibility of handcuffs. Healthcare is different – remember these blogs are about prevention rather than prosecution. However, the point is that every second counts and the sooner you are there the better the data.
The facts you collect, how those facts are collected, and by who and when, affects the conclusions you can draw about the incident. Physical data lasts longer than human memory data but, as the picture of the ‘converted’ aircraft shows – things change.
Who and when the facts are collected affects the interventions you can use, and the reliability of testing those interventions you trial / test. Human factors people or psychologists are a vital part of the team. They are only part of the team.
You should see a pattern. What evidence (when it’s done and by whom) you collect, affects the intervention and its success. With no data you should not do any interventions. Indeed, without data you may not wish to.
Remember it’s about outcomes and not just documented processes. In the words of the philosophers – Metallica – “nothing else matters”.
Next time... Human factors part 2, or should we do interventions? Like the Star Wars films, these blogs may appear in the wrong order but the final box set hopefully makes sense! Comments welcome young Skywalker.
Read the other blogs in this series
- Why investigate? Part 1
- Why investigate? Part 2: Where do facts come from (mummy)?
- Who should investigate? Part 3
- Human factors – the scientific study of man in her built environment. Part 4
- When to investigate? Part 5
- How or why. Part 6
- Why investigate? Part 7 – The questions and answers
- Why investigate? Part 8 – Why an ‘It’s an error trap conclusion’ is an error trap
- Why investigate? Part 9: Making wrong decisions when we think they are the right decisions
- Why investigate? Part 10: Fatigue – Enter the Sandman
- Why investigate? Part 11: We have a situation
- Why investigate? Part 12: Ethics in research
About the Author
Martin is a topic leader for the hub.
He founded the Human Factors group at the University of Sussex (1999), which became User Perspective Ltd in 2003. Martin, User Perspective MD and Chief Scientist, aided by his team, has undertaken almost 600 research and forensic investigation projects. He is interested in human error and human factors.
Martin is a research auditor for the UK government, EU academic networks and many governments worldwide. Within healthcare he has investigated matters as diverse as neonatal safety in transport, unexplained injuries in the hospital mortuary, sepsis diagnosis and retained instruments. Martin co-authored the very first Healthcare Safety Investigation Branch (HSIB) report that investigated orthopaedic surgery in the UK and Europe. His interest in the law and justice extends to his voluntary role as a justice of the peace (JP) in the Magistrate and Crown courts.