Inattentional blindness or "What's that gorilla doing there?"

There has been much media attention recently on the fact that radiologists are unable to spot gorillas. For most people, the first encounter with this gorilla will have been the "passing the ball" video which demonstrates a study by Daniel Simons and Christopher Chabris in 1999 at Harvard University. This video shows the limits of human attention.

All of us will have experienced inattentional blindness in our everyday lives, perhaps while driving a car or watching TV. Things which are occurring within our field of vision (and therefore are being picked up by the retina) are being ignored in preference to something else.

What does the study about radiologists add to our understanding about this phenomenon? 83% of radiologists failed to detect an image of a dancing gorilla on a CT scan when they were asked to look for nodules. 100% of untrained observers failed to spot the gorilla. The authors admit that if the radiologists had been told to look for something unusual then they would probably have spotted more gorillas. What the study doesn't tell us is the difference between those radiologists who did or did not spot the gorillas. Did those who spotted the gorillas spot fewer lung nodules because they were less focused on their primary task?

In terms of clinical practice and understanding of human factors, the study adds to the evidence that if you are focused on one task you cannot also focus on something else. This would suggest that a team leader does not get involved in a task. It also suggests that one should be precise when asking someone else to perform a task. For example if you ask someone to keep an eye on a patient's saturations then don't be surprised if they fail to pick up the ectopics on the ECG... you didn't ask them to look at the ECG trace.

Inattentional blindness is a side-effect of our amazing ability to focus on something and stay focused on it. The best defences against are to be aware of its existence and to have another background mental process running which is asking you whether there is something else you are missing while you are focusing on a given task.

How real does simulation need to be?

The question of how realistic a simulation needs to be is a perennial one. For a long time time I tended to ignore the question altogether because, as a mentor once told me, "It's all about the debrief!" The view from this camp is that the equipment is secondary to the primary educational intervention of the debrief.

Then I was swayed by Roger Kneebone's argument that the simulation just needs to be "real enough". So in his mobile inflatable simulator the anaesthetic machine is replaced by a couple of posters, with the sounds of the machine playing in the background. (This, I must add, is for simulations aimed at surgeons, not anaesthetists.)

Recently I have been thinking about our IV access arm. We use it in some of our medical undergraduate scenarios if the mannequin does not have IV access at the start of the scenario (e.g. they have just arrived at the emergency department.) Our rationale for this is, I think, two-fold; the students get to appreciate how long IV access can take to achieve (which lets us explore team dynamics, delegation, situational awareness, shared mental model, etc.) and they are provided with the opportunity to carry out a technical skill.

In our more advanced courses the IV arm is no longer used. Either the mannequin already has IV access ("someone" has placed a cannula) or when the participant tells the confederate in the room that they would now obtain IV access, the IV cannula appears on the mannequin. Why is there a change in practice in the more advanced courses? Partly because we no longer want people to "waste" valuable time trying to obtain access, partly because we think that at this stage of their career the participants will be well-trained in IV access and partly because having to go to a separate IV arm some distance from the body may impinge on the realism of the scenario. (I appreciate that we could stick the IV access arm onto the mannequin or use the expensive IV access available on the mannequin but I will try and explain why this is not really the solution below.)

Our final use of the IV access arm is in one-hour workshops we run for medical undergraduates to explore IV fluids and IV access. And it is here where my concerns about the IV arm are greatest. In order to improve our skills at something, we need to have deliberate, coached practice. I am happy to agree with anyone who tells me that the IV arm is useful for teaching students how to prepare for IV access (wash hands, gloves, prepare skin, etc.) but I would have trouble agreeing with anyone who tells me that the actual insertion of the IV cannula into the arm is useful. My reasoning here is that the feel of the needle as it pierces the plastic, the amount of force required and the angle of approach is very different from what is required in a patient. I would worry, in fact, that if someone became very good at obtaining IV access in the IV arm they would be not very good at obtaining IV access in a real person. (Although I have no research to back this up.) As a corollary, if someone is very good at IV access on people then the IV arm can present them with difficulties (which is why we don't get people to obtain access on the more advanced courses.)

This brings me on to my thoughts about the realism in the simulator. Because of time constraints or mannequin fidelity we have to (or think we have to) sacrifice aspects of realism. A septic patient may, in reality, have a few hours between starting to show signs of sepsis and becoming comatose with septic shock. In a simulated scenario we compress the intervening time period down to 15 minutes or so (often the mannequin starts off relatively "well" to allow the candidates to make a diagnosis/initiate treatment and then the mannequin quickly becomes unwell and requires further interventions.) My concern is that, when candidates then go on to see sepsis (for example) outside the simulator, the cues and timings are so different from the simulation that it does not trigger the behaviours and actions we want. Perhaps we are failing in our role as educators for certain conditions because the simulation is insufficiently realistic around the most crucial aspects (whatever they may be)? Answers on a postcard please...

When things go wrong

Many of us who are involved in simulation-based medical education (SBME) continue to carry out clinical duties. There are arguments in favour and against maintaining a clinical role, which I will not go into in this post. However, for those of us who are still clinically active there will come a time when, despite the human factors training we have received and deliver, we will be involved in a serious adverse event. I thought I would share some thoughts with you on what to do/not to do and how we can use these experiences to enrich SBME.

Step 1: Take ownership of your omissions and commissions
The best way not to learn from an adverse event is to deny that it has anything to do with you or that it was not your "fault". If you were in the room when the event was happening then there will have been steps you could have taken to prevent or mitigate the event. Ensure that the patient and/or family know that you are sorry for what has happened and that the event will be investigated.

Step 2: Write down a full timeline of the events from your perspective (and ask others to do the same)
Doing this as soon as possible after the event means that you will have the best chance of remembering things.

Step 3: Analyse the timeline and add human factors commentary
Consider at all stages and from as wide a view as possible what the circumstances were which led to the event. Were there gaps in knowledge? Did fatigue play a role? Was communication an issue? Were there any error traps such as confirmation bias, loss aversion, or recency bias?

Step 4: Debrief
Use the timelines from as many people as possible to create a "master timeline" (which may have contradictive events) and assign a non-involved person versed in human factors to debrief. Remember to list all the things that went well. Identify changes in practice which may attenuate or prevent a similar adverse event.

Step 5: Initiate and sustain changes in practice
As a person who was involved in the adverse event you have a duty to initiate and sustain changes in your workplace (e.g. use of WHO checklist, time-outs, encouraging people to speak up)

Step 6: Use the increased understanding of this adverse event in your delivery of SBME
Generally speaking I would discourage the exact "re-run" of a particular adverse event in the simulator, however many of the circumstances identified in step 3 will be applicable (with perhaps minor changes) to the courses you currently run.

Step 7: Inform the patient/family
Let the patient and/or family know about all of the above and how the lessons learnt are being applied.


I appreciate that the above is not a perfect sequence but it is a good starting point. Lastly, if you are involved in a serious adverse event, remember that you too are human, that you too will make mistakes and that the best possible outcome from a mistake is that you learn from it.

Please tell us how we did.

As I mentioned in my previous post, we collect feedback from the participants on all our courses. We don't (yet) have a generic feedback form so courses which are scenario-based will ask how each scenario went, courses such as faculty-development will ask what we did well and what we could do to improve, etc.

The more feedback forms I see and the more feedback forms I fill out myself the more I realise that our feedback forms are (generally) not fit for purpose. (For a similar point of view, see this article from the BBC.) In particular the tick boxes of "pre-course administration" "administration during the course" "catering" etc. actually give us very little information. The majority of participants tick "very good" or "good", while the occasional "poor" or "very poor" remains unexplained.

Even specific questions such as "Was the duration of the course: a) Too short b) About right or c) Too long" can be difficult to interpret. When people were ticking "Too short", I wasn't sure if they meant that the actual course on the day was too short or that they would have preferred a 2 day course. (When I asked that anybody who ticked "Too short" explain what they meant in the comments section, it turned out that they meant that they would like to have a 2 day course, not that they wanted to finish at 5:30pm or minded if the course finished "early" at 4:45pm)

Currently we also ask people to fill out the feedback on a piece of paper which our administrator then transcribes onto an excel spreadsheet, quite a time-consuming task.

The temptation then is to just ask two questions at the end of the course:
1) What did we do well?
2) What could we do better?
Might these two questions get to the heart of the matter? Would the lack of numerical data make it difficult to show funders that the courses are value-for-money?

I would be interested to hear from anybody who has cracked the feedback puzzle.