Anchors Away with Professor Pat Croskerry

In the first of two very special episodes, Amie and Ben are joined by Amie's hero, Professor Pat Croskerry, a leading expert on cognitive bias and how doctors think. 

We go back to basics and cover all things critical thinking.

Croskerry, P. (2020) The cognitive autopsy: a root cause analysis of medical decision making. 1st edn. New York: Oxford University Press.

Groopman, J. (2007). How doctors think. Houghton Mifflin Company. To find locally:  https://search.worldcat.org/title/1040496921

Montgomery, Kathryn, How Doctors Think: Clinical Judgment And The Practice Of Medicine (New York, NY, 2005; online edn, Oxford Academic, 31 Oct. 2023), https://doi.org/10.1093/oso/9780195187120.001.0001 To find locally: https://search.worldcat.org/title/58452493

If you want to get in touch contact us at a.burbridge@nhs.net

Disclaimer: All patient stories discussed in Home of Medicine are informed by real patient interactions. However, all identifying details have been removed or appropriately modified to protect patient confidentiality. 

This podcast is intended for education and professional development and should not replace independent clinical judgement or specialist consultation.

Transcript

SPEAKER_02 0:12

Hello and welcome to the Home of Medicine Podcast, a podcast in association with the Royal College of Physicians, Edinburgh. I'm your host, I'm Dr. Amy Burbridge, an acute physician working in the West Midlands in the UK.

SPEAKER_00 0:26

Hi, I'm Ben Lovell. I'm also an acute medic, and I work in London in the UK.

SPEAKER_02 0:30

So for those of you who've been listening to our podcast for for the last few years, you would know that the focus of our podcast is clinical reasoning, biases, and making mistakes, errors, talking about them. Now, one person who I've read so much of his work and learned a huge, huge amount from is Professor Pat Croskery. And he's with us today on the podcast, which is just incredible. Pat, thank you so much for joining us.

SPEAKER_01 1:05

My pleasure.

SPEAKER_02 1:07

So, Pat, tell us where you work. What's your role at the moment?

SPEAKER_01 1:13

I guess I'm technically retired now. I'm no longer in the trenches with you guys, and most of my post-retirement activities so far has been taken up with trying to write some of this down so that we can spread it to others. We have published uh three books so far, and a fourth one is coming, and the one that's coming is called Critical Thinking in Clinical Practice. And it's an effort to try to reach all the people who are currently involved in thinking activities that influence patient care. So primarily we would be focusing on physicians. From any domain of medicine, one thing that's become fairly clear is that, first of all, the thinking across all disciplines is universal. People don't change their thinking because they're a dermatologist or because they work in the ICU. I mean, they change them to a certain extent, but the basic nuts and bolts of their thinking remains the same. We have models for this, and they've been developed by psychologists and uh to some extent philosophers. In fact, if you go back uh 2,000 years to the Greek philosophers, you can find them talking about very clearly about biases and cautioning their um students or their followers about the traps of cognitive biases. So that's mostly what I'm doing now. We will refer periodically to some of the books that have been published. There are three so far, and there is this four a fourth one coming that is trying to kind of pull it all together for people in medicine. And the reason for this, as as you've already alluded to, is that uh few of us in medicine get any specific training in thinking. Very broadly speaking, there are two types of knowledge that we can acquire in our lifetimes about any particular topic or area. And one is declarative knowledge, where they teach you all the basic facts about something. So, for example, in medicine, it would be teaching about anatomy, uh, physiology, pathophysiology, anything else that you can you can put in there that is factual. That's called declarative knowledge. But what gets us into trouble eventually down the road is something called procedural knowledge. And and the analogy I'll draw is that uh say that you want to make a stew and you ask your mum or your dad, how do I make, say, a seafood chowder? They would say, well, uh, I think they would say, uh, first of all, they would list off all the ingredients. And that is what we do in medicine when we put people through medical school. We list off all the ingredients for what they need. You can't do without that basic declarative knowledge. So they need to know anatomy, physiology, pathophysiology, and a few other bits and pieces in order to sensibly talk about how the whole process works in sickness and in health. So my major concern, I I originally trained as a psychologist before I went into medicine. So I had a background in people thinking about thinking. And when I got into medicine, I I really was quite um appalled. Appalled is a strong word, but I think I was because I didn't see any evidence of people sort of practicing procedural knowledge particularly, except, you know, we had some luminaries in medicine who were known for their thinking, you know, famous uh medical people. But I didn't see much of it in the course. Uh although it may have come out informally where somebody said would say something like, uh, you know, I think you're jumping to conclusions and we should back off a bit, maybe something like that, a little bit of minor editing. But most of the time, um, we didn't put much effort into how people think. And uh really that's been my whole purpose. And it's not just a personal feeling. If you look at where things go wrong, if you look at medico-legal cases and errors, the significant part of the time, the problem doesn't lie with declarative knowledge. It doesn't lie with what you had to learn in medical school. If you end up with a case where the patient suffered significant morbid or mortality, it would not be because you didn't remember what you'd learned in medical school, because most of that stuff was focused on declarative knowledge. So our effort, I suppose, to summarize is just to say, to ask the question, what about procedural knowledge? What about asking ourselves and asking our students uh who we were training in this profession, ask them how they think. And several books have been written about this. There is an excellent book by Catherine Montgomery called How Doctors Think. She was a philosopher in the States and and uh got very interested in how doctors think. The second book was by Jerome Grootman, who was a hematologist from Boston, and he published a book called How Doctors Think, the same type. I don't think he was aware of what Montgomery had done. And that may be because um Montgomery was very focused on how we think, and this was a new area in uh current medicine for Jerry Grootman, and he may have thought that this was the first time that people were asking a question, but she did ask it very clearly, and she had some very clear ideas about where doctors went wrong when they did run into trouble with their thinking. So you can ask these questions, but you've also got to answer the questions, and the questions are centered around how do you think, how do doctors think, how does anybody think? And all of humans and some animals for sure use a dual process model, and uh it is what it implies. It's two ways of making a decision. You can either do it in something called system one or type one thinking, or you can do it in system two. Now, this applies across the board, whether you're a car mechanic or a chef or a pilot or a physician, we all engage this dual process model. So I think you know, a good place to start with thinking about how people think is to think about the model that has evolved through psychology and philosophy, and to some extent now through medicine, of these two systems of making a decision. And some people may have read Daniel Kahneman's book, Thinking Fast and Slow. Carneman is a is a, or was, I think he's retired now, but a psychologist at Princeton in the States. And he did some classic work with a colleague called Amos Toversky. And unfortunately, uh Toversky died prematurely. He's no longer with us. The Nobel Prize was awarded for their work, but they they don't award it posthumously. So Kahneman took the Nobel Prize for this thinking fast and slow book uh that summarized the work that he'd done with Toversky. And uh maybe I could just take a few minutes to summarize that model that Kahneman uh was the underlie underlay um the work that uh Toversky and Kahneman did. Essentially, what Kahneman is referring to now when he talks about thinking fast is system one, and this is a system that all of us have. It's very important because we spend about 95% of our time in system one, and uh it's enough to get by with for most of the things that we have to do in life, but occasionally you you have to you have to go into system two, which is a deliberate analytical type of thinking, which is characteristically slower because you you can't just go on your impulse or your intuition, you have to slow down and say, I have to work out here what's going on, you know, and you start summarizing facts and you start trying to describe the relationship between different things that you've identified to be relevant to the problem and so on. That's called system two, and it's much slower, uh, it's generally speaking more reliable because a lot of system one doesn't reach the level of consciousness. That's a very critical part of the way that we approach thinking in medicine. Because if we're going to talk about biases, and and the psychologists do say that bias is the greatest threat to our thinking, you know, uh of all the things that can interfere with your thinking. Say that you're working in emergency and and you're overwhelmed, and there's a a wait time of of uh six hours or something to be seen, and and there's lots of activity and commotion and busyness uh and all of that. That certainly interferes with the way that you think, and it will tend to push you into system one, that is that fast way of thinking or reflexive, intuitive way of thinking. And it in in some ways, uh it system one is not a bad place to be. For example, if you're skiing down the side of a hill and uh you come around a bend and there's suddenly an avalanche has occurred in front of you, then you have to make some pretty quick decisions whether you're just going to plow yourself into the obstacle, or can you find a way around it, or can you can you do something to lessen your impact that could be dangerous to you? So understanding when system one is appropriate and what's likely to push you into system one is an important part of um understanding how you make decisions in medicine. And and we find that that there are three general areas where errors will arise when we're dealing with patients. The first one is the influence of these biases that interfere with the way that we think. The second one is the conditions under which we work, you know, and some of the conditions today in emergency departments, for example, are just horrendous uh with the wait times and what it does to people and patients and the people who work there. And then the third area is actually knowledge deficits, you know, that people simply don't know enough about what they're dealing with. That gets back to the declarative knowledge. But usually in medicine, if we get to a point where we don't know enough, we tend to refer the patient to a specialist who knows more than we do. So you can deal with you can deal with the knowledge deficit fairly easily. You can deal with bad working conditions to some extent. You could say, um, you know, we need bigger departments or we need more personnel or we need to give people more rest or whatever. You can think of solutions to the poor workplace scenario. But the third scenario, the we make biases and we have failures in the way that we think, those procedural errors, you pretty much have to learn about how we think and the biases that we make. You have to understand that process in order to deal with it. That's not surprising. So so we start off with um uh medical students at Dauhousie here in Canada. We teach them the sort of nuts and bolts of the jaw process model. What's involved in Kahneman's fast and slow thinking? That model is called the jaw process model. So we teach them how that model works, where you spend most of your time in the model, what system one can do to system two, and what system two can do to system one, and other aspects of that model that help you understand the process of thinking. And uh, because this particular podcast is aimed at biases, you you would want to know a lot about system one, because system one is where most of those biases are. I should start by saying biases generally have a negative reputation. People don't think very highly. Oh, if you if I say uh to a colleague, I think you're biased on this, bias implies some sort of fault in reasoning, some negative attribute that um is is undesirable. That's a little bit unfair to biases, because biases can actually help us in some ways. Certainly they can allow us to make extremely fast decisions when we need to under adverse conditions. So, in some ways, biases give us a certain flexibility. It's a bit like if you think of a lawn bowl, where you you have a spherical ball that is used by people on lawn, very nice lawn surfaces, and and essentially they want to get as I don't know the rules of bowling, but essentially they want to get as close as possible to the the little white ball up the other end of the green. Now, if if that bowl is perfectly spherical and round, then if you roll it, it will go in a straight line. And that's all the bowl will do. But any self-respecting lawn bowler will tell you that the bowls have various biases built into them, which allows them to go in more than a straight line. So, for example, I could have picked up a ball that has a bias in it that says, when you release me, I'm gonna veer to the left, or another one that says, I'm gonna veer to the right. And this allows you to sort of uh come at the little white ball that you have to get close to. It allows you to adopt a curved trajectory to get there through this built-in bias. And so there are ways of using bias to help you. I'll give you a quick example. Um, most psychiatric patients, in say in the emergency in AE, the emergency department, most psychiatric patients will have a bias against them. As soon as you say, um, I have this um schizophrenic patient who uh, as soon as you say schizophrenic, you've introduced the bias. If or I I have a bipolar patient here, or I have a whatever, the psychiatric patient automatically introduces a bias. And not surprisingly, throughout medicine, psychiatric patients do not receive the same quality of care as a non-psychiatric patient, right? Just a it's a simple fact. You can show that psychiatric patients, if a psychiatric patient is coming in with chest pain, the psychiatric patient fares less well than the non-psychiatric patient coming in with chest pain, right? And so we know that this bias is actually harmful to the safety of those patients. So there are certain things you can do. You can say, for example, okay, if this patient is identified as psychiatric, or it emerges that they have a psychiatric history during the during the history taking, then I can do certain things to make sure that I don't commit the usual errors that are made on psychiatric patients. So, for example, uh I might take a more thorough history, or I might do a more thorough examination, or I may clarify some things and just make sure I've understood them fully from the patient. So you can you can use the bias to say it will make me a more thorough diagnostician, something like that.

SPEAKER_02 19:41

So, what's interesting there is I actually had a psychiatric patient who presented to the emergency department. I was a very young doctor and he was schizophrenic and he was confused. I assumed it was psychosis. So in the UK, we we have a system where we section patients who have mental health problems because I assumed his behaviour was all due to schizophrenia. He went to the psychiatric hospital the next day. I didn't do any investigations, and then he came back the next day and he had a large subdural hemorrhage and I completely missed it. Um I'd been a doctor for about a year, um, but it it stayed with me, and I always ask myself why I made the assumption that the schizophrenia was the cause of the presenting complaint, and I completely stopped thinking.

SPEAKER_01 20:39

Right. So that that that is exactly the era that I'm talking about. So so the strategy for debiasing yourself is to say, this is I mean, to explicitly say to yourself, this is a psychiatric patient. This patient has an increased likelihood of having something missed or uh or having substandard care. Therefore, I will do um extra whatever to try to make sure that doesn't happen and keeping that bias in front of me all the time. So that's a that's a specific bias. Uh uh it's called a um, it's had has uh the sorry the debiasing procedure has various names. It's called debiasing or counter-biasing. Um, but in effect, you try to reduce the chance that some error will be made with that patient because of some implicit assumption that you've made. And and and most of the time our implicit assumptions are okay. But in certain cases, I mean, there are there is a litany of horrendous things that have been missed in psychiatric patients. And and so um we can do something about bias in that particular Case by rebiasing ourselves to do a more extensive examination, a more extensive history, and and maybe ask a few more questions than you would otherwise ask. So that that's an introduction to debiasing. We've talked about two systems. One gets you to the answer very quickly, that's thinking fast, and the other one is more slower. And so you should also know that almost 95% of our time we're in the fast system. Okay? Um, so the m the the idea being here that the more you understand about that dual process model, the more you will be able to anticipate some of its properties and and and when things go wrong. Most of the shortcuts in our thinking occur in system one. So so for example, uh I live in the country and I uh I work in town, it's about a 40-mile drive, and I can do that drive in the morning and get to work and not really have any recollection of the journey. And the reason for that is because most of the driving is done in system one. 95% of the driving is done in system one. If if a sign comes up on the road that says caution hazard ahead or something, you switch into system two. You start looking for trouble and you look for some variation in the pattern that serves you very well the rest of the time. So so 95% of your driving is in system one. And it sounds a bit dangerous, but but it works quite well, providing you keep in mind that system one is not very good for dealing with emergencies or distractions or interruptions or things that are um unplanned, basically. Um most of the errors that we see in human thinking uh seem to occur in system one. And that's largely, I think, just because we spend 95% of our time there. Um, repetitive operations in system two can get you into system one. So say that I say, uh, you've never driven a car before. Okay, uh, this is a car, there's the driver's seat, here's all the instruments, and here's what they do. And I teach you how to drive, we can then, after you've done a significant amount of driving, we can come out of system two and you'll go into system one automatically. So repetitive operations in system two eventually get you into system one. If I'm teaching you how to intubate, I I start off by saying this is a laryngoscope and it is hinged here, and there's a little light to tell you what's going on, and you have to you have to use your non non-dominant hand. And I go through all the steps of intubation, but after you've done about a hundred, you don't, well, hopefully long before that, you will be able to intubate almost intuitively. You know, you'll you'll set yourself up and and perform all the movements uh in a smooth sequence, uh, and and it's not broken up into those training uh segments that I I originally started with you. So so eventually, if we spend enough time in system two, however complex the act, we sooner or later will end up in system one doing it fairly smoothly and and without error.

SPEAKER_02 26:00

So you do you said that most of the errors occur in system one.

SPEAKER_03 26:06

Yeah.

SPEAKER_02 26:06

Yeah. So is that where most of the biases are found then, would you think, in our thinking?

SPEAKER_01 26:12

Yes.

SPEAKER_02 26:13

Yeah. And as a clinician on the shop floor, um it's busy, it's chaotic. We've described this before. It's really loud, it's really noisy. When I'm seeing a patient, what other biases as a clinician do I need to be aware of? So you've talked about the psychiatric bias. What's the one what's the other really common ones that we tend to see quite a lot of?

SPEAKER_01 26:42

The most common um in in the book that we published called The Cognitive Autopsy, we actually identified the hierarchy of biases, uh, which is the most common and which uh uh and and where are they likely to occur in the sequence when you're seeing the patient. Um so so we in the book we give each bias a color, and then we unravel a case, a typical case, and what is the first bias to appear, the second, third, fourth, and fifth. Um and uh uh uh certain biases will occur early on. So the first the first bias that tends to occur is called anchoring. And uh it it just makes sense, you know, when you first run into a problem, the first thing you do, you you you pay attention to its salient features, you know. What are the what are the you know the the child nurse may say to you, I have a patient here who's overdosed on um a tricyclic, say something like that. And and immediately your brain is focusing on tricyclic overdose. You anchor onto the key features of that presentation. You haven't even asked if the patient is male or female or 80 or 8. Um, the only thing that matters to you at the outset is that the most critical uh element that's been presented to you is is actually the substance that we that they overdosed on. And that and so you can see that in some ways that's that's a very important bias to have is to anchor onto something that's highly relevant and highly uh influential. Um the the bias itself is called anchoring and adjustment, uh originally by Kahneman and Tzversky. And and and they added that little caveat just to make sure that you understood where the problem was coming from. Because there's nothing wrong with anchoring. If I go into emergency and I say I'm having chest pain, and and you don't immediately start examining the foot or the arm or the back or anything else, you go straight to the chest and and perhaps uh order a cardiogram right away and put your stethoscope on the chest and so on. Um, so anchoring itself is really very helpful. But the problem is that um you you have to be willing to make an adjustment if other information comes into you that says, you know, the patient may say, they may start off by saying, I have some chest pain. And and later, as you go through the exam, they may say, um, and it's it's since I drank uh this toxic substance or something. So so then you can adjust your anchor and say, I'm going down the cardiac route, but maybe I should be thinking GI and toxicology routes. So an important um adjunct to the to the anchoring bias is adjustment. You you must be prepared. If you want to use anchoring well, you must be prepared to keep adjusting as other information comes in. There's nothing wrong with anchoring. In fact, anchoring is is the right thing to do, but it's not the right thing necessarily to stick with. So anchoring is a very common bias.

SPEAKER_00 30:40

Um what might happen, of course, is that when the patient says, uh, and by the way, I've had this chest pain since I ingested this substance, and the clinician, maybe me, says, Yeah, yeah, yeah, yeah. We're not dealing with that now, but just get back to the chest pain and we'll that that's not relevant to and then discounting it because I've moved on to another bias. I can't know if that's confirmation or ascertainment bias where I'm starting to discount things that I don't that don't fit into my my neat anchor already. Yeah. Sort of uh discarding new information so it comes on because it's it's cluttering up my thinking process. And then I can I can potter off down the wrong diagnostic pathway like that.

SPEAKER_01 31:16

That's right. And and and that's exactly what we'd like people to do. We'd like we'd like the clinician to be kind of monitoring themselves um and what sorts of biases they may be vulnerable to as you go along. And and it's not a tall order, you know. You can um you can teach people the top 20 biases, you can teach them how they work and and where they're likely to occur, anchoring very likely to occur early on. Um and and and it it it is really a great endpoint to have the clinician monitoring what they're doing, monitoring what they're thinking, and and that is called metacognition. And and it's also called reflect reflective thinking, but you know, thinking about what you're thinking. And if you have a clinician like that, um, and and uh and we we did we did some exercises with medical students where we asked them to identify the particular bias that might be influencing them at particular points, and and that is quite a useful exercise as well. And I and I went to uh um MM RANDs. Do you have MM RANDs? Yeah.

SPEAKER_02 32:31

We do, yeah. Yes.

SPEAKER_01 32:33

At uh I went to one in uh Boston at Harvard, and the case was presented um in this way. They uh the presenter said, This is a uh 27-year-old uh uh patient who came in with chest pain, say. And um almost immediately their practice was to try to identify the biases that as you went along. So it was a a kind of running commentary on the various biases that that may have been relevant to the case. And I I thought it was very useful exercise. A lot of the time we simply don't have the time to do that, and we just want to get on and go for where the money is. Um, but but that's got to do with how busy you are and why biases go up when you're very busy. There is there is less opportunity to be thoughtful and and to be mindful and to be reflective uh and to practice metacognition when when you don't have much in the way of personal resources.

SPEAKER_02 33:46

And I think we as doctors, we make a diagnosis, like we set down the anchor, and then it's really difficult sometimes to accept that we're wrong. So you often see it written in notes or on the um electronic patient record, the same diagnosis every single day, and it's not challenged because we don't like to challenge other clinicians. So, for example, if I saw a patient tomorrow and Ben had seen them today, I wouldn't really probably want to challenge him because of that sort of professional respect. But actually, I probably should, you know, take a hit.

SPEAKER_00 34:23

Feel free. I was in system one mode the whole day yesterday. I've talked to tons of them and say, please do check my homework if you ever come round after me, Amy, on the wards.

SPEAKER_02 34:32

So, and I think it'd be really it's really interesting because when people change my diagnosis, I'm like, oh God, I really messed up and you really start to think and challenge yourself. And I think that that is diagnostic momentum, isn't it? Where that diagnosis never changes, it gains momentum, but there's no evidence. And it just carries on and carries on and carries on. And I'm sure we've all been there with cases where we've particularly in neurological conditions, I tend to find they're they're often missed spinal cord compression or Gillian Barray, really difficult diagnoses to pick up. And somebody on day one has said something, can we just go on with that?

SPEAKER_01 35:13

Yeah. That's right. So so um transitions of care where one patient is moved to another is another um source of errors. That's a um a hot point, hot point for error. And to you can't do this constantly because it's so cognitively demanding, but to keep challenging, to make yourself um what was the word they used to use for um uh Richard Dawkins? He had a he had a um um not skeptical, um, but but his mind was always asking questions. And yeah. If you do that, um uh you you you can come across as a bit tedious and and people people may not approve of you, but a certain amount of challenge. Oh, I've I've just thought of the word, it's contrarian, and it would that it was actually uh Hitchens who described himself as a contrarian, but that is really challenging everything, yeah. And and and and being totally skeptical. And I don't think we we benefit by being that extreme. Uh in medicine you have to take a certain amount of things for granted because or assume that they're correct in order that you're you maintain some optimum um level of performance. You you can't challenge everything. Now and again, you simply do accept what you've been told. Um but to know the vulnerable times is important. For example, if a department is extremely busy, or if this patient is being transferred from one team to another or one physician to another, then uh it recognize that those times are dangerous because labels, you know, are very sticky once they're applied. And uh, you know, and it's one of the reasons why we shouldn't um uh if possible avoid referring, I mean, in emergency you would get something like sometimes like the chest pain in bed four or or the kidney stab the kidney stone in three or something. And so we we were trying to discourage that because the labels can seal the patient's fate, you know, if you don't if if if you if you don't accept that when you when you do that, you end up inheriting somebody else's thinking.

SPEAKER_02 37:51

It's the framing effect, isn't it?

SPEAKER_01 37:53

Is that the It is certainly framing.

SPEAKER_02 37:55

Yeah. Yeah, and you find that in um for example, if I want to maybe refer a patient to a specialty or somebody refers a patient to me, it's almost like it's been sold to you.

SPEAKER_01 38:08

Yeah.

SPEAKER_02 38:09

Um, because they want you to buy a certain diagnosis, which I'd say.

SPEAKER_01 38:13

And and that there is actually a literature on this in emergency physicians where they they have learned certain tricks for saying patients to other specialties in order that you can move them. I mean, not in any callous way, but you can unload you can unload the work from your side by getting the other person to accept responsibility for it. And at times like that, people don't want to run into a contrarian or a skeptic who's gonna challenge everything that they say. But it does happen in medicine, and it's a specific game that uh AE physicians can play. You know, for example, if I was referring to a patient to plastics, I remember being told once um uh because they'd suffered a skin evulsion or something. And a colleague said to me, don't ever try to make a referral if the amount of skin that's been lost is less than three square centimeters or something. But so you would say, I have a patient here with a uh an avulsion of skin that that's probably about four square centimeters, but you go above you go above the threshold that that will allow them to accept the referral. So there's a bit of that goes on in this. Um so I I I like the uh I I like I mean you've already started identifying some of the biases, anchoring, framing, diagnosis, momentum. But we do know we we're not making this up. We we when we did the analysis in the book uh the cognitive autopsy, um, we did identify the top 20 biases in emergency medicine. And it it it would be useful if you just had that page of 20 biases up on the wall somewhere, or if your patient is going to suffer from a bias, it most likely falls into this top one of these. And to know what the biases are in the top 20 as well, I think is useful. But but you've already started to pick up on it. Anchoring is number one, for sure. You know, ascertainment bias is up there, number two or number one, uh seeing what you expect to see, and confirmation bias, and and you can you can uh where we've where we've um uh uh developed this plot. Um, I don't know if I'm just looking to see if I've got this in the uh in the book, but we have a diagram showing different colors for all the biases.

SPEAKER_02 40:51

Yeah.

SPEAKER_01 40:52

And and as they appear. Oh, there it is.

SPEAKER_02 40:55

Yeah, I can see it, yeah. Yeah.

SPEAKER_01 40:58

And uh we did that uh to to try and make the concept of bias more concrete, to say uh this is what it is, this is where it tends to appear, and you should know the properties of it. If you if you are influenced by the bias, here's the sorts of things that are gonna happen um with it. Um so uh uh the the other the other things we do is we coach students on the properties of the model. We've we've already mentioned several of them. Um system two can override system one, which is a good thing. Uh, and that's what happens if you engage metacognition uh and some sort of reflective or mindful practice. You can say, well, it looks like that, but it could also be this or this. Um can override system two, and that you can predict which conditions will predispose you to that, you know, if you're extremely busy, uh change over times, and so on. And fatigue, sleep deprivation will increase your vulnerability to biases. Um and the importance of calibration. Um, if I I don't know what strategies your departments have for dealing with things when they go wrong, but but in when you do that, it's helpful if you can um get as much information to people as possible about what went wrong, because most of the time we are calibrated by our mistakes. You know, we it it it's uh it's better to be uh it it's better to face up to a mistake because you can learn more from it sometimes than your successors.

SPEAKER_02 42:53

So just to conclude then, we've learnt a huge amount about your processing theory, about some of the biases. I'm gonna go to work tomorrow and I'm gonna meet maybe three fresh-faced resident doctors or medical students. What can I do in a clinical setting to introduce them to this concept? Do we know what works when we try and educate people about this?

SPEAKER_01 43:24

If it's a real clinical situation, this they're seeing patients here, it's a bit awkward. Uh uh I we advocate trying to learn about the properties of biases and how they work beforehand. And then when you're in a clinical situation, you're teaching a recent or a student, you can say, um if I do this or this, um, what bias does that indicate? Maybe at play, or the patient said this, what might that do to my thinking? Thinking, but try to you it's difficult just to teach it all from scratch in a in a fresh clinical situation because you wouldn't do that if if with uh declarative knowledge, you wouldn't say to a medical student, um, so this patient has a breathing problem, uh, which organ system does that involve? I mean, uh so it's nice to get all the as much as possible done beforehand. So you have seminars, you have um uh MM rounds which may be focused on certain biases. It's nice in MM if you can pick a case that illustrates uh a bias, because the there is a tremendous power in in clinical cases. I mean doctors have learned that over the years that if if you put the thing in a clinical text, in a clinical setting, then people remember it better. So it'd be nice uh rounds if you could identify the bias and say, you know, it's clear here that we um we we were vulnerable to anchoring or cognitive bias of some kind, you know, premature closure, search satisfacing. Um people remember it a lot better. I I had a medical student once who who um had a patient come in with a um uh hand injury and she presented the case to me and said, um this this uh guy has got uh uh an injury to his um middle finger, and um I've ordered an x-ray on it. So she ordered the x-ray, and uh the guy had a small avulsion fracture of the of the proximal phalanx or something. And uh and and that was uh uh and and then so I said, so what do you want to do here? And she said, Well, I think we should uh you know make this kind of splint and send him to a clinic and so on. I said, No, the the first thing you've got to do is look for the second fracture. Because after the first fracture, you've got a tendency to close off the search. That's called search satisfaction, and it's a big problem in Medsa. So she went back to the x-ray and she looked again and she found a second fracture, and and she was just amazed and said, Wow, uh there was a second fracture there. So she so I said, What do you want to do now? And she said, Well, uh we we'll get into orthoclinic and so on. I said, No, what do you do now is you found a second fracture, you go back and you look for the third. Because as long as you're finding fractures, then you can't call off the search. And in the end, she found three fractures in this guy's hand, and they were all minor, but they were all there and they were all relevant. Uh and uh I saw the reason I mentioned her is I saw her about two months ago, and she said she just went search satisfacing, right? She remembered that case. But search satisfacing is a very powerful bias because if we lose something, if we if, for example, we lose our car keys, uh, we go on a search for the car keys, and as soon as we find the car keys, we call the search off. We don't go looking for another set of car keys. We say, that's what I wanted. And and we're reinforced for terminating the search after we found something. So it's a hard lesson to learn sometimes, but you do have to remind people that as soon as you find something, look for the other one. And and there's so many good examples of it in medicine. Uh, I can remember one case where we got caught up in a uh a lady came in with an empty bottle of aspirin and said, I took all of these. So, you know, we immediately launched into an ASA toxicity uh toxidrome treatment. And uh and then she wasn't getting much better, and in fact, she was getting worse. And then her husband wandered in uh and said, She's done this before. And we say, Yeah, I we've uh discovered that she has taken a bottle of aspirin. And he said, Yeah, she'd tell you about these, and pulls out an empty bottle of tricyclics. So we'd called off the search as soon as we thought we were dealing with one toxied ramp but then suddenly her QRS intervals are all over the place, and and we realize we had something else on our hands. So big bias in medicine is search satisficing, and give people examples of it and and how they can be uh how they can how they can occur across different areas of emergency medicine, like orthopedics in that case, to uh a a toxydrome where you may be dealing with more than one um problem. Because nobody ever goes looking for a second set of khaki's in real life.

SPEAKER_02 49:14

So um Ben, I'm gonna come to you and we've learned so much. What I want you to do, Ben, is give us three key learning points that uh that our listeners can go listen to and go, right, yeah, absolutely. What have you taken away from that?

SPEAKER_00 49:34

Uh well you might have warned me you were gonna do that, Amy, but unfortunately I was taking notes as we went through. Um I could see you taking notes, so I knew it'd be fine. Oh, you now knew you assumed I was taking notes. I could draw a cat. No, I was taking notes. Yes. Um, so what have I written down here? Anchoring itself is not a bad thing, it can guide you towards the right diagnosis in a pinch, but you need to be prepared to adjust your anchor. I've written that. We should call this episode of the podcast Anchors Away or something clever like that.

SPEAKER_02 50:03

Oh, we like it, yeah. Yeah.

SPEAKER_00 50:05

Um I wrote down the next one and I might have spelt it wrong. Did you call it satisficing? How did you put how did you call that um bias pad?

SPEAKER_01 50:14

It it it's it's a um combination of two words. Um satisfy.

SPEAKER_00 50:22

Uh God, I've forgotten the two words now. It wasn't just me then. It's I was like I it's a really difficult word, isn't it? It's a difficult word to say. Yeah, yeah. I guess it applies in the premature closure bit. It's don't stop looking because you found one thing, because there are other things associated with it. And if you're still finding stuff, the search isn't over. So I've I've written that down. I probably could have phrased it a bit better than that, but that's that's what I wrote down. Um and you wanted three things anyway else is in my scroll here. Um, 95% of thinking is done in system one. And you know, that really makes me think because not only do I come into work and I'm I'm faced with, okay, Ben, you've got to you've got to see 30 patients. They're all in the corridor and AE because of no space, and you've got to do it by four o'clock. There's a meeting then where we have a big, you know, what workout how big the disaster is. And simultaneously I receive texts on my phone from the trust saying, alert, black alert, black alert, please discharge as many patients as possible. And I feel like these messages are basically saying, Ben, system one thinking only today, please. That's all you've got time for. And so I oblige. I'm an obliging person. And I I do make very rapid decisions based on a combination of surface analysis and a little bit of gut and diagnostic momentum, hoping someone's done a bit of thinking before me. Um, and I and I think mostly it goes okay, but I'm more and more aware of when I'm doing it, and there's a slight unease at bay when I'm doing it. So just writing down this difference between system one and system two, if I were to say to the powers that be, hang on, slow down. I need to do some system two thinking here. There's something going on, and I need to sit down and work it out. That's almost frowned upon because you're gonna, it doesn't look like you're doing much. Me sitting there scratching my beard and drinking coffee and having to think doesn't look like I'm practicing medicine. And I try not to do it too much, but I think that's how I exhibit system two thinking sometimes. So that's giving me something to think about as well.

SPEAKER_02 52:16

And I really like the sticky labels because I think that's I've not ever thought about it like that before, but people come in and it's almost like they've got a label that they they've got on their top that says ACS or gastric or or migraine or something. Um, and it's really difficult. That sticky label's really sticky, it's like stuck on with super glue. Um, and I think it's really important that that label should just be very lightly attached and that we can easily take it off.

SPEAKER_01 52:47

And that includes labels like frequent flyer.

SPEAKER_02 52:50

Oh yeah.

SPEAKER_01 52:51

One US department I was involved with had for frequent flyer patients, they had a little airplane beside their name. I mean, that is terrible. I've just remembered I've just remembered the uh origin of search satisfacing. It's a portmanteau of um satisfy and sufficient. So you put the two together and you have satisficing. I am satisfied I have done sufficient work with this. Brilliant, thank you.

SPEAKER_02 53:28

Thank you. That was brilliant. Um I've made I've always made loads of notes that I can never really understand later. Um, I know that all of our listeners will absolutely love that episode. Um, a massive thank you to all of our listeners, a massive thank you to the Royal College of Physicians of Edinburgh, and we will be back soon. Thanks for listening. Goodbye.