1 00:00:01,170 --> 00:00:04,180 Welcome on today. You're here at the Centre for Evidence based Medicine. 2 00:00:04,200 --> 00:00:08,430 I have Professor John Barlow from the from Melbourne with me, 3 00:00:08,460 --> 00:00:16,590 who has a great deal of expertise in diagnostic reasoning and has done a number of research studies with over the last five years. 4 00:00:16,620 --> 00:00:22,919 Welcome, John. Okay. Now what we're going to talk about today is we going to set out and look at some 5 00:00:22,920 --> 00:00:27,090 of the evidence over the last 30 or 40 years relating to diagnostic reasoning. 6 00:00:27,900 --> 00:00:31,590 We are both late to some of the interesting papers we've looked at. 7 00:00:31,590 --> 00:00:37,140 We've looked at a paper by Arthur Elford in thinking about diagnostic thinking, a birth year perspective. 8 00:00:38,310 --> 00:00:43,709 What a paper of research and clinical reasoning by Geoffrey Norman and another paper 9 00:00:43,710 --> 00:00:48,720 about educational strategies to promote clinical diagnostic reasoning by Judith Bone. 10 00:00:49,290 --> 00:00:53,790 And what we'll do then is finish up with some of our own thinking around reasoning. 11 00:00:54,300 --> 00:01:02,130 First, we'll just go over to John, and I'm going to pose some issues about the issues of diagnostic reasoning. 12 00:01:02,970 --> 00:01:10,580 So the first paper that with this paper by author of the and who did work in a hypothetical deductive model in some 30, 13 00:01:10,590 --> 00:01:12,670 40 years ago, which you're well aware of. 14 00:01:13,550 --> 00:01:20,910 And he used this sort of system, which maybe you could just describe, which is this thinking aloud type of work. 15 00:01:21,150 --> 00:01:36,360 And if any research on this, you know, Arthur Holstein and his colleagues and the first group to do this systematically looking at how doctors think. 16 00:01:37,560 --> 00:01:51,510 And I left the inheritance, which to the extent that it's still the most important contribution of I suddenly would start. 17 00:01:52,590 --> 00:02:04,440 And what I did, I collected about 30 or so physicians, physicians paying a mixture of general practitioners and generally not specialists. 18 00:02:04,620 --> 00:02:12,900 Yeah, I gave them a and I asked people to participate who were highly regarded in the group. 19 00:02:12,930 --> 00:02:23,790 Yeah. So they would see a lot of experts and they gave them standardised cases like multiple sclerosis and similar that had a complex process. 20 00:02:23,880 --> 00:02:30,240 Mm hmm. Something as a general practitioner of which they call a physician, would not say very often. 21 00:02:30,510 --> 00:02:44,290 Okay. And they asked them to go through thinking aloud, whatever was going through their mind at the time when they asked questions in person. 22 00:02:44,330 --> 00:02:47,700 Patient was giving a it to them. 23 00:02:51,420 --> 00:03:05,250 And what I found was that within a matter of seconds of minutes, all of these doctors had 3 to 5, what they called hypotheses in mind. 24 00:03:06,210 --> 00:03:11,820 Can I bring you up on that? Just. Yes. I mean, read in this 209 report and I, I see the five I mean, 25 00:03:11,820 --> 00:03:22,140 the first thing that jumped out to me with any major findings were experts and non-expert physicians could not be distinguished. 26 00:03:23,160 --> 00:03:30,899 And that's actually is probably the most important finding in that it's hard to believe a 27 00:03:30,900 --> 00:03:37,920 situation which reflects reality that there's no difference between an expert and a non-expert. 28 00:03:38,760 --> 00:03:48,450 So clearly there was a big question mark about some of the things they were finding in relationship to the method, 29 00:03:49,410 --> 00:04:06,930 because at first people used to think that there's a general, a generalised method for clinical reasoning, but there must be more to it. 30 00:04:07,650 --> 00:04:17,910 And there's another finding which I have some friend that is the same expert given different cases, 31 00:04:18,630 --> 00:04:26,910 some of them that they collected and other cases they were not good at it, saying that there was no generalisability and there was more to it. 32 00:04:27,270 --> 00:04:35,850 I just think trying to solve that problem. So if I come back to a way of doing this was to provide artificial cases. 33 00:04:36,420 --> 00:04:46,890 That's right. But the problem with that approach is, is the artificial breadth of the case doesn't necessarily reflect real world practice. 34 00:04:47,190 --> 00:04:53,490 And that presents a problem then in this artificial nature, either it may be too easy or too hard, 35 00:04:54,030 --> 00:04:57,050 and there's no way of distinguishing expert from non-expert. 36 00:04:58,050 --> 00:05:10,020 I agree that that's. Important to note, the fact that it was a standardised case of the real case takes it away from real reality. 37 00:05:11,010 --> 00:05:15,570 Because one thing we do know these days and become down fairly quickly, 38 00:05:16,200 --> 00:05:23,190 there's a great deal of thinking and behaviour is dependent on the environment in the context of the case. 39 00:05:23,370 --> 00:05:27,990 So a expertise if content or case specific. 40 00:05:28,000 --> 00:05:33,180 So in some areas you may be an expert, particularly in certain areas, but not in all areas. 41 00:05:34,110 --> 00:05:40,589 And the other thing that is interesting about these findings, which is that this idea, 42 00:05:40,590 --> 00:05:47,370 which has long pervaded the literature, is a hypothetical deductive method of reasoning, 43 00:05:49,140 --> 00:05:59,720 which is just if you could just briefly explain what that entails and that entails that you recognise something in the patient. 44 00:06:00,240 --> 00:06:07,110 Yeah. And from that you did used that that patient may have a number of other things which make him to have a certain disease. 45 00:06:08,010 --> 00:06:12,440 But we know that this is unlikely to be the case in this case if the patient. 46 00:06:13,170 --> 00:06:21,900 So because it's hard to imagine that you develop a hypothesis in a matter of seconds, it must be a different way of doing it. 47 00:06:22,260 --> 00:06:31,050 Okay. So what I thought were hypotheses developed deduction is probably something else and that's another story later on. 48 00:06:31,500 --> 00:06:40,530 So interestingly so I think there are a couple of interesting things, although think aloud is a useful technique in artificial cases, 49 00:06:40,530 --> 00:06:48,570 it may be somewhat too lacking in content or case specific to be actually elicit true expertise. 50 00:06:49,200 --> 00:06:56,370 And then the second thing is to say in summary, is that potentially the hypothetical deductive model that pervaded for some 30, 51 00:06:56,370 --> 00:07:03,540 40 years is actually in the real world, is not how we operate because you just can't generate you hypotheses and test them. 52 00:07:03,990 --> 00:07:07,260 And it would probably take a really long time if that's how you did practice. 53 00:07:07,560 --> 00:07:10,410 And they use it some in the more complex case. Okay. 54 00:07:11,520 --> 00:07:21,050 And I think it's interesting that even today people talk about hypothetical is the doctor being at all norm and it isn't so interesting. 55 00:07:21,080 --> 00:07:27,390 And then the second thing is, what is don't hear anything is look at the analysis of case reports. 56 00:07:28,500 --> 00:07:36,090 And that's an interesting but one of the cases here says explicitly recalls of who in the era 57 00:07:36,090 --> 00:07:42,510 before the clinical laboratory had developed the urge physicians to listen to their patients. 58 00:07:42,930 --> 00:07:47,550 They are telling you the diagnosis, which is often what we hear. 59 00:07:48,450 --> 00:07:57,180 But is that really helpful, that type of reflection or issue around expert developing expertise? 60 00:07:57,690 --> 00:08:02,730 Well, I think that has a point in that patients often know more than we think. 61 00:08:03,690 --> 00:08:09,000 And I think it's worthwhile listening to the patient. But then you have to check check it by yourself. 62 00:08:09,300 --> 00:08:11,879 Okay. Now, one of the one of the issues here, 63 00:08:11,880 --> 00:08:19,230 which I found the issue with one of the with some of the cases with makes two points relevant to our understanding of diagnose, 64 00:08:19,890 --> 00:08:28,980 diagnostic reasoning, the role of prior clinical experience now in the service of rapid pattern recognition. 65 00:08:30,500 --> 00:08:33,079 And what that goes on in the end. 66 00:08:33,080 --> 00:08:44,660 95 this rapid, poorly verbalised, highly overblown process of intuition or of others prefer it as pattern recognition. 67 00:08:45,680 --> 00:08:51,350 And so within this aspect is that this doesn't clearly fit the hypothetical to this issue 68 00:08:51,350 --> 00:08:56,780 about your prior clinical experience in relation to patterns and what people call intuition. 69 00:08:57,940 --> 00:09:06,560 Yeah, that's an interesting thing to think about because again, people still think about pattern recognition, 70 00:09:07,190 --> 00:09:12,490 which implies that you have looked at a large number of aspects of the case you have passed. 71 00:09:12,830 --> 00:09:22,370 The original so-called hypothesis is usually just dependent on a single look at a single symptom from the patient. 72 00:09:22,880 --> 00:09:36,260 Okay. And I think that is one of the major things to recognise that it is not a person that makes you feel about the idea in the first place, 73 00:09:36,740 --> 00:09:46,160 but is more likely to be a solid feature of a case which you recognised and relationship to experience in experience. 74 00:09:46,160 --> 00:09:53,810 People with medicine still make hypotheses because patients often have hypotheses in medical students. 75 00:09:54,200 --> 00:10:02,300 If you see a patient limping, let's say this patient might have had a stroke because their grandmother's friend had a stroke and she was limping. 76 00:10:02,720 --> 00:10:08,340 So they all have background experiences, but it's a difference of experience from what they expected. 77 00:10:08,870 --> 00:10:13,189 Okay, so that's interesting. So salient features in the role of experience, 78 00:10:13,190 --> 00:10:21,860 but you can have experiences outside of just being trained to be a doctor because they may already be there in the in the background. 79 00:10:21,890 --> 00:10:26,629 Absolutely. Okay. Now, I'm just going to pick out one last point from this paper. 80 00:10:26,630 --> 00:10:34,400 It's a very nice, sort of concise look at some of the work of last 30 years and just come in at the end so 81 00:10:34,400 --> 00:10:39,950 that we know we now know that experienced physicians can and do use all kinds of methods. 82 00:10:40,010 --> 00:10:45,709 So there's more than one method in effect, in current cognitive, very rapid nonverbal, 83 00:10:45,710 --> 00:10:49,370 intuitive cognition is characterised as system one and system two. 84 00:10:50,240 --> 00:10:54,980 We'll come back to that later, if you don't mind, because that theme being important in some of our work. 85 00:10:56,150 --> 00:11:05,450 But it's interesting here and maybe we should tell you this about how do clinicians decide which approach is best suited for the case at hand? 86 00:11:06,350 --> 00:11:10,880 And it leaves a question mark there. That's a question mark. 87 00:11:10,940 --> 00:11:16,800 Yeah. Even today, if they are so unpopular, if that's how they work on the, you know, 88 00:11:16,880 --> 00:11:24,050 theory we're going to talk about that's to shed some light and a few things I but 89 00:11:24,710 --> 00:11:28,190 okay so that's some of the evidence from nice points that we may touch on later. 90 00:11:28,850 --> 00:11:33,980 Then in the middle here, there's a paper by Judith Boyd in the New England Journal of Medicine, 91 00:11:34,580 --> 00:11:41,000 which is educational strategies to promote clinical diagnostic reasoning, which is sort of aimed at teacher. 92 00:11:41,000 --> 00:11:44,420 This paper is a thought review, look at a few cases and so forth. 93 00:11:46,160 --> 00:11:52,520 And what what she does there is is talk about the expert written brought two sets of skills to the encounter 94 00:11:52,520 --> 00:12:00,290 with the patient looking at a case about Gamze I think it was and looking at how an expert with a non-expert, 95 00:12:00,290 --> 00:12:05,870 a bit like a registrar or consultant versus a sort of medical student, 96 00:12:07,100 --> 00:12:12,290 the resident probably formed an early impression, a mental abstraction of the patient's story. 97 00:12:12,890 --> 00:12:20,390 And then the second thing is the expert residents clinical case presentation, presentation with a succinct summary of the findings, 98 00:12:20,840 --> 00:12:24,640 providing the teacher with a clinical picture of the patient as seen through the residents. 99 00:12:24,650 --> 00:12:28,729 I never thought of what she was trying to say with some aspect of expert. 100 00:12:28,730 --> 00:12:36,020 How did that help us in any way or and it helps in the sense that after it was shown that 101 00:12:36,020 --> 00:12:43,280 the type of thinking from the original study was mostly in the course of the answer, 102 00:12:44,060 --> 00:12:52,640 people started thinking about what sort of knowledge is necessary to make diagnostic decisions 103 00:12:53,930 --> 00:13:01,910 and what sort of schemas is the brain stored in the brain to help you with this work? 104 00:13:02,750 --> 00:13:09,139 And so the basis of what she is talking about is that when you have seen a few patients involved in this 105 00:13:09,140 --> 00:13:15,709 particular case and have seen that possible aspect of that and can differentiate it from something else, 106 00:13:15,710 --> 00:13:17,900 like what you have lost to arthritis. 107 00:13:18,620 --> 00:13:29,690 And you do this because you have a picture in your mind, a complex of knowledge that you've built up from seeing a number of cases in contrast. 108 00:13:30,340 --> 00:13:34,270 The first year medical student and a first year intern. 109 00:13:35,200 --> 00:13:38,470 And since he's actually the patient is God is on his list. 110 00:13:39,470 --> 00:13:52,210 So we know from previous studies that a novice tends to diagnose on lists, often not giving priorities because he doesn't know enough enough about it. 111 00:13:52,900 --> 00:14:00,220 In contrast, the person has got a complex knowledge about it and as much work as when it wants to go. 112 00:14:01,030 --> 00:14:04,720 What do you do next to make it better? To make a better diagnosis? 113 00:14:05,500 --> 00:14:12,430 Often they are referred to as an illness illness script. 114 00:14:12,670 --> 00:14:16,850 Yeah, well, we'll come to that later with Jeffrey. Norman may be the only three route so. 115 00:14:17,020 --> 00:14:24,700 So her position is this position is provides a sort of simple model which looks very appealing but themes what we've just heard 116 00:14:24,700 --> 00:14:32,319 previously and from Elstein where the conclusion a bit too simplistic book talks about the first step is based on knowledge, 117 00:14:32,320 --> 00:14:38,590 experience and other important contextual factors, such as you're in the emergency department, you're in a surgery ward. 118 00:14:38,590 --> 00:14:44,200 What time of day it is? What's the patient want? The second step is this data acquisition. 119 00:14:44,200 --> 00:14:50,470 Depending on the setting elements of the history, findings of physical exam and results of the test. 120 00:14:51,070 --> 00:14:56,320 And then another early step is the creation of a mental abstraction or problem representation. 121 00:14:56,830 --> 00:15:08,020 Usually a one sentence summary. Defining the specific case in abstract and patient has pain early in the morning and it's worth an exercise. 122 00:15:08,050 --> 00:15:16,840 Clearly, this is osteoarthritis. Interesting, though, she goes on to this clinicians may have no conscious awareness of this cognitive step. 123 00:15:17,740 --> 00:15:25,930 The problem representation. That's an interesting there's no conscious awareness in this divide. 124 00:15:25,930 --> 00:15:29,440 And how do you see what where does that come into the reasoning process? 125 00:15:29,440 --> 00:15:33,250 And if we're not aware of it, how can you improve that or even understand what you do? 126 00:15:33,850 --> 00:15:38,050 Well, a description is appropriate, practice is accurate up to that point, 127 00:15:38,920 --> 00:15:47,800 but it depends on how much introspection the doctor will exercise and how much, 128 00:15:48,250 --> 00:15:54,340 you know, think about it or just accept it coming out of the blue, assuming it's it would be fine. 129 00:15:56,290 --> 00:15:59,470 It's not really. 130 00:16:04,110 --> 00:16:11,310 It's by no means unusual to diagnose this fight. And one of the questions we have, which we still haven't answered completely. 131 00:16:11,880 --> 00:16:20,550 As we said before, it had to change from acceptance to let's stop here and wait and look at it in more detail. 132 00:16:21,570 --> 00:16:24,570 And I think you want to talk a bit later about this. Yeah. 133 00:16:25,480 --> 00:16:32,970 But I believe that's what this paper is about, is how to improve the knowledge in relationship to that particular disease. 134 00:16:34,020 --> 00:16:39,180 But it's not really clinical reasoning. 135 00:16:39,510 --> 00:16:44,790 Is disease recognition and disease oriented as opposed to condition oriented? 136 00:16:45,360 --> 00:16:49,980 So that's interesting. But she does go go back and just relate a bit to this many methods. 137 00:16:50,790 --> 00:17:02,879 She talks about this both non analytical and analytical reasoning strategies are effective and are used simultaneously in either interactive fashion, 138 00:17:02,880 --> 00:17:09,660 so they must be going at the same time non analytical reasoning as exemplified by pattern recognition. 139 00:17:09,660 --> 00:17:13,770 This is the second time we've seen it come up is essential to diagnostic expertise. 140 00:17:15,450 --> 00:17:24,630 Well, no, not analytical reasoning is a demand problem because not all analytical things happen in your brain. 141 00:17:24,870 --> 00:17:30,420 Yes, but it's not to me. My mind and reasoning process reflect on something, okay. 142 00:17:30,630 --> 00:17:36,210 Something that you have built up over time in many cases over here, if you're an expert. 143 00:17:37,110 --> 00:17:44,400 And the interesting thing about it is that it changes all the time a great deal, which I think we all need to come back to later on. 144 00:17:44,820 --> 00:17:48,930 So does that mean so from a non analytical perspective, 145 00:17:49,440 --> 00:17:59,130 the major way to improve in known all I intellect is through experience and that's how we sometimes define expertise in terms of experience sometimes. 146 00:17:59,580 --> 00:18:08,310 However, then she talks about deliberative, analytical reasoning is a primary strategy when a case is complex or ill defined. 147 00:18:09,150 --> 00:18:12,330 And boy, do you see a lot of them when you're not quite sure what's going on. 148 00:18:13,020 --> 00:18:18,540 The clinical findings are unusual or the physician has had a little clinical experience with the particular disease. 149 00:18:18,540 --> 00:18:24,929 And, you know, there must be then a way of reasoning or working out what's going on that's different. 150 00:18:24,930 --> 00:18:31,650 That's and that's me. That where the skill, which is not related to expertise, starts to come in in some way. 151 00:18:32,850 --> 00:18:37,110 It is it is a skill, but it's not just a skeleton. 152 00:18:37,590 --> 00:18:46,780 It's a way of operating, as it were. Some people are more likely to be want to learn, and I think a lot of they think why. 153 00:18:47,520 --> 00:18:54,000 In other words, explore what you're saying and relating to the previous existing language that you have. 154 00:18:55,050 --> 00:18:59,580 And I think that that to me is the issue here. What's what I've learned. 155 00:18:59,610 --> 00:19:05,639 You are also a physician that you going to see, which will determine what sort of expert you have. 156 00:19:05,640 --> 00:19:11,670 And I'd be okay. And I think as you just finish up with learners with strong diagnostic reasoning skills, 157 00:19:11,700 --> 00:19:20,579 often use multiple abstract qualifiers to discuss the discriminating features of a clinical case in comparing 158 00:19:20,580 --> 00:19:26,400 and contrasting appropriate diagnostic hypotheses and leaking each hypothesis to the finding in the case. 159 00:19:27,090 --> 00:19:29,700 So they use in different things. They may say, Well, here's the pattern. 160 00:19:29,700 --> 00:19:32,879 This is what I think this may be, but I going to think about this test result. 161 00:19:32,880 --> 00:19:38,010 It might make it think this and this ruled it out. So they're using different techniques all at once. 162 00:19:38,700 --> 00:19:45,420 I think that I believe they don't use patterns. Okay. I use critical cases that I think are important. 163 00:19:46,080 --> 00:19:59,100 And that's what's one of the early findings in the first phase of this study is that the patient 164 00:19:59,430 --> 00:20:07,760 could doctors do not collect more data than bad actors and America's correlation between apathy, 165 00:20:07,770 --> 00:20:11,670 collect and the rightness or wrongness of your diagnosis. 166 00:20:11,850 --> 00:20:14,550 Okay, interesting. We'll come back to that and come back to that. 167 00:20:15,000 --> 00:20:21,239 Interestingly, interestingly as well in some of the letters that you're looking at responses, 168 00:20:21,240 --> 00:20:25,410 but some of the criticism and one of the writers said Berlin's review of 169 00:20:25,410 --> 00:20:28,920 educational strategies can be that can be used to promote diagnostic reasoning, 170 00:20:28,920 --> 00:20:32,580 does not sufficiently emphasise the concept of premature closure. 171 00:20:33,560 --> 00:20:42,000 And I thought that was interesting as well. And I'll come back to that when we come throughout some of the work you've done over the last five years. 172 00:20:42,630 --> 00:20:48,300 But let's just before we get there, Jeffrey Norman Hey, I did a paper here in 2005, 173 00:20:48,300 --> 00:20:51,810 which is the research in clinical reasoning, past history and current trends. 174 00:20:53,550 --> 00:21:00,629 I think it's interesting to know of the the 40 million papers published in PubMed every year, there actually isn't much research in this area. 175 00:21:00,630 --> 00:21:06,270 There's there's only a couple of. Hype is out there. Really? And the other problem is it's published all over the place. 176 00:21:06,300 --> 00:21:15,810 Yeah. So it's published from psychology to medicine to surgery to education and so on and so on. 177 00:21:16,260 --> 00:21:20,850 And many in the big journals, I mean, it's quite nice to see the New England Journal in Medicine. 178 00:21:20,850 --> 00:21:28,800 This is 2006, actually don't really push publish much about teaching or reflecting or reasoning around diagnostics, do they? 179 00:21:28,980 --> 00:21:37,410 That's right. Which seems to me a sort of paradox, because this is an important concept of medicine. 180 00:21:37,410 --> 00:21:40,830 If you get the diagnosis right, we can go from there. If you get it wrong. 181 00:21:41,530 --> 00:21:42,450 We're in deep trouble. 182 00:21:42,450 --> 00:21:50,460 Yet nobody wants to emphasise they're more likely to publish a result of a test single test property and to think about how doctors may learn better. 183 00:21:50,910 --> 00:21:55,020 But that's the difference between disease or education and conceptualisation. 184 00:21:55,710 --> 00:21:56,850 So what is just fine? I mean, 185 00:21:56,970 --> 00:22:02,340 there was a number of themes that actually Norman I quite like and I'll maybe I'll take each one and then you could your thoughts on the first one. 186 00:22:03,180 --> 00:22:08,760 There is little evidence that reasoning can be characterised in terms of a general growth variables. 187 00:22:10,120 --> 00:22:14,730 Yeah. Just for the people listening. What does that mean, general process variables? 188 00:22:15,420 --> 00:22:18,210 Well, what it means is that they started with all this. 189 00:22:18,810 --> 00:22:27,240 The idea was that, you know, you've got that system of reasoning and we can teach it to every student for every type of case. 190 00:22:27,870 --> 00:22:32,130 But even the early studies have showed this is not the way it works. 191 00:22:33,540 --> 00:22:42,450 Doctors do it in a very different way, and every doctor will have a different way of doing it himself or herself. 192 00:22:42,900 --> 00:22:50,700 Okay. That's interesting. And that's why is evidence that expertise is associated not with a single basic reputation, 193 00:22:50,970 --> 00:22:57,240 but with multiple coordinated representations in memory from causal mechanisms to prior examples. 194 00:22:57,240 --> 00:23:00,330 And that's what he's trying to get that more than one. Okay. 195 00:23:00,690 --> 00:23:07,290 Different representations may be utilised in different circumstances, but he goes on, this is again. 196 00:23:07,290 --> 00:23:14,190 And there are things that if little is known about the characteristics of a particular situation that led to a change in strategy. 197 00:23:16,320 --> 00:23:21,720 So this is a bit like when we don't quite know which one the presented when why you change the way you do. 198 00:23:22,890 --> 00:23:31,500 Yeah, I have to go back to the beginning and say that they thought that they know very little about this. 199 00:23:32,370 --> 00:23:38,730 There there's some studies recently which addressed a few ideas, but. 200 00:23:41,880 --> 00:23:55,870 We can really divide what's in a concept in the brain into three different categories, again mentioned by Norman Rockwell. 201 00:23:56,260 --> 00:24:04,979 Now, maybe we'll come to that in a minute or come that sort of a nice little narrative through the thought 1780s and 202 00:24:04,980 --> 00:24:13,280 nineties of all things in the crazy how things work and talked about the hypothetical and deductive were interesting. 203 00:24:13,500 --> 00:24:22,530 We talked a little bit in the eighties about the issues about chess and the growing lecture on expertise in other domains. 204 00:24:22,530 --> 00:24:28,319 Particularly, chess was quite influential in the 18th and interesting to note know the single best measure of 205 00:24:28,320 --> 00:24:34,620 expertise in chess was recall of a typical mid-game position where after a five second exposure, 206 00:24:34,620 --> 00:24:39,030 experts would typically recall the exact position of about 80% of the pieces. 207 00:24:39,990 --> 00:24:43,049 It is basically showing the expertise. 208 00:24:43,050 --> 00:24:50,820 Is it based on underlying relation between memory performance and expertise and that characterises all of the thinking around that time? 209 00:24:51,060 --> 00:24:57,000 That's right. Interest in a chess master is remembered about 50,000 game positions to get there. 210 00:24:58,530 --> 00:25:03,810 However, what you say, medicine, there is a little gain from gathering and remembering extensive amount of patient data, 211 00:25:04,080 --> 00:25:09,510 which I think you've already mentioned. Consequently, foreignness is a poor index of expertise. 212 00:25:09,870 --> 00:25:18,570 That's right. So this is where you see some connections there for ten, 15 minutes with the patients and others in their 60 seconds, 90 seconds. 213 00:25:19,230 --> 00:25:24,270 And this concept of some people will write a whole ream of history and other people write two 214 00:25:24,300 --> 00:25:30,510 lines and say there's a diagnosis and in some way the expertise of the ones who come through. 215 00:25:30,900 --> 00:25:37,920 Would you agree more quicker to get into what the nub of the issues in the case was? 216 00:25:37,920 --> 00:25:42,329 Quite okay. You can disagree me if you want. 217 00:25:42,330 --> 00:25:52,409 That's okay. It's been shown that experts with original first thought think of is more likely to be arrived in. 218 00:25:52,410 --> 00:25:57,720 Someone is not an expert. So that's true. On the other hand, just how much data you collect. 219 00:26:00,030 --> 00:26:04,360 Can depends very much on that person's inclination. 220 00:26:04,930 --> 00:26:11,860 Some experts just spend a lot of time and some experts do not have to point it all to risk taking. 221 00:26:12,760 --> 00:26:21,400 Basically, if you make a diagnosis is a matter of seconds yet quite at risk to the outside world and also to yourself, 222 00:26:22,060 --> 00:26:25,510 because you didn't have to to check on your thinking. 223 00:26:25,900 --> 00:26:28,000 In other words, you did not go on to system. 224 00:26:28,000 --> 00:26:37,659 Do as we are okay to check on your system one, but it's a matter of choosing the place that you make and you don't have to do. 225 00:26:37,660 --> 00:26:42,310 This is the issue that you're talking about. Okay, that's interesting. 226 00:26:42,320 --> 00:26:46,840 I'm going to come back to that, but just a bit in this Norman paper that really interested me, 227 00:26:46,840 --> 00:26:53,799 which was research during the 1990s focussed on forms of mental representations going back to 228 00:26:53,800 --> 00:26:59,740 the old pattern recognition and experience and can be divide roughly into broad categories. 229 00:26:59,740 --> 00:27:06,490 And there was a Jack Goldsmith who did a lot of the work and they said and it comes into basic science or causal knowledge, schemas, 230 00:27:06,490 --> 00:27:14,110 scripts and other representations of the relation between signs and symptoms and diagnoses and exemplars based on experience with past cases. 231 00:27:14,680 --> 00:27:25,480 So the example here is that by teaching basic science, you create the ability to have the ability to dip into the exemplar and go back. 232 00:27:26,590 --> 00:27:35,440 And but interestingly, there's a very pertinent quote which is going to read Given the amount of educational time devoted to learning basic sciences, 233 00:27:36,130 --> 00:27:42,910 it is reasonable to assume that expert clinicians exemplify the application of scientific concepts to clinical problems, 234 00:27:43,600 --> 00:27:51,460 yet primarily that expert clinicians make little use of biomedical science in daily reasoning. 235 00:27:52,570 --> 00:27:58,690 Now there's a conflict there in terms of what we teach and what you use in practice. 236 00:28:00,610 --> 00:28:12,280 I think the answer to that is that the basic science knowledge is something which you sometimes isn't used, 237 00:28:12,940 --> 00:28:17,820 and it depends up to the point on the speciality point on the patient, evidently. 238 00:28:21,050 --> 00:28:31,770 It is a concept to be shaped when people talk about encapsulated knowledge of knowledge that's so sticks to the script, for instance, or Schmidt. 239 00:28:32,550 --> 00:28:36,540 And it's the knowledge that you have picked up somewhere else that you stick onto it. 240 00:28:37,740 --> 00:28:44,040 And how much of that is okay is taking place? Because they're very much dependent on the types of cases you are saying. 241 00:28:45,390 --> 00:28:54,010 And so when you say you do not use basic science knowledge, that is not entirely true because you can't, 242 00:28:54,140 --> 00:28:58,320 because you can show that some of it is important to you and some of it is not. 243 00:29:00,630 --> 00:29:08,370 On the other hand, when you look at a student or watch, you have a doctor seeing a patient for the first time. 244 00:29:08,970 --> 00:29:11,700 They use a great deal of basic science. 245 00:29:11,910 --> 00:29:18,690 Yeah, that's how they read things that the expert doesn't have to replace the good doctor with all that somewhere. 246 00:29:19,260 --> 00:29:22,020 I suppose I shouldn't say. Well, some of it. Is there some? 247 00:29:22,770 --> 00:29:29,640 Because certainly learning how many things you've got in your brain that Ossify is not going to help you. 248 00:29:30,360 --> 00:29:36,719 So interesting thought is that I'm going to move forward now because into and we've got 249 00:29:36,720 --> 00:29:40,799 a paper here identifying the early warning signs for diagnostic errors in primary care. 250 00:29:40,800 --> 00:29:47,150 And this is John Barlow and actually myself here and Clare Guide and Matthew Thompson in BMJ Open. 251 00:29:47,160 --> 00:29:52,590 This is actually a paper that looked at 25 experienced general practitioners. 252 00:29:53,310 --> 00:29:58,860 But this is built on a couple of papers that basically try to bring a model together that 253 00:29:58,860 --> 00:30:04,180 says part of this reasoning process and we we heard earlier right beginning from our 254 00:30:04,230 --> 00:30:09,420 Eltham with around the system want to system to rethink you want to just explain what that 255 00:30:09,420 --> 00:30:15,150 is a little bit and then we'll look at how that applies to the process of diagnosis. 256 00:30:19,160 --> 00:30:27,530 Arthur Elston said in this and other paper that we were referring to him that one of the biggest problems they have is 257 00:30:27,770 --> 00:30:36,650 how to thank you going to system two if you need to make what matches you need to as opposed to you don't need to. 258 00:30:37,490 --> 00:30:46,130 So let me just about health system. One reason is the fast intuitive reasoning and system two is useful to the liberal analytical reasoning. 259 00:30:46,670 --> 00:30:53,330 I mean, in some of the books, like I tell you this and they thought to say a system one reasoning so if I say to you to close to 260 00:30:53,660 --> 00:30:59,960 immediately comes to your mind and people listening more you can't stop but go for can you in your mind. 261 00:31:00,470 --> 00:31:08,209 But if I go 35 times 23, then you've got to go into a deliberative, analytical mode to try and work out. 262 00:31:08,210 --> 00:31:13,650 You can still work it out, but you have to think about it. And that's what's going on here up to a point. 263 00:31:14,360 --> 00:31:18,259 So how does that then apply to the diagnostic process? 264 00:31:18,260 --> 00:31:23,990 How do clinicians work in relation to this fast and slow thinking for. 265 00:31:27,250 --> 00:31:34,500 They use mental pictures. I mean, to what's in your knowledge base? 266 00:31:35,550 --> 00:31:41,640 And they have different names for it. The name I like best is Personal Knowledge. 267 00:31:41,880 --> 00:31:48,240 It's okay. And personal knowledge is based on a number of things. 268 00:31:49,260 --> 00:31:56,190 Basic science, in other words, was referred to quite often as the theory of the profession. 269 00:31:56,790 --> 00:32:01,280 This is how she talks about it is what you were talking about before. 270 00:32:01,410 --> 00:32:03,930 Yes. And there's we there. 271 00:32:04,820 --> 00:32:17,280 The second thing that there is the disease oriented schemas that we were talking about that Schmidt and and a number of other things like heuristics, 272 00:32:17,280 --> 00:32:25,560 they're going to be talked about. It used to being sort of rules of thumb, for instance, about if if you do this, you do that. 273 00:32:27,090 --> 00:32:31,020 But there are a couple of other things, and one of which are very important. 274 00:32:31,530 --> 00:32:35,520 And probably the most important is has been provided to things you have seen before. 275 00:32:36,360 --> 00:32:45,270 And you've got a large number of things and then you switch it on to your present knowledge, it becomes your personal knowledge. 276 00:32:46,290 --> 00:32:52,170 And one of the problems we have is we all have different ways of approaching this because we are different people, 277 00:32:52,710 --> 00:33:00,110 so somehow more likely to want to incorporate it and others that do not get much interested in it. 278 00:33:01,340 --> 00:33:08,309 And so what, what I'm talking about is that that is personal knowledge that we have, 279 00:33:08,310 --> 00:33:16,620 which depends on the person himself or herself, as well as the experience and as well as the okay theory. 280 00:33:17,580 --> 00:33:26,010 And as another thing that is very important to recognise is the environment that you doing your work in, 281 00:33:26,670 --> 00:33:31,680 which is you are referring to before that the original cases were not real cases. 282 00:33:31,800 --> 00:33:33,960 Yeah, we are the real world. Yeah. 283 00:33:34,380 --> 00:33:43,980 And one of the most important decisions to make about that I spent time in, as true or not, is coming from the outside world. 284 00:33:44,790 --> 00:33:55,560 And I think you have to think of it as for instance, a typical example is Friday afternoon, everything is closing up and again the next day today. 285 00:33:55,920 --> 00:33:59,969 Is that tomorrow? Because I guess I will be closed tomorrow. Okay. 286 00:33:59,970 --> 00:34:07,620 Yeah, now, now. But when? So in the context of important but we also had this bit earlier which you mentioned and which comes up a bit, 287 00:34:07,980 --> 00:34:16,410 this idea that in this process that there's not so much pattern recognition in, but it's critical queue. 288 00:34:16,410 --> 00:34:23,490 That's right. And what's the difference between these? See the difference between pattern recognition and critical cues, 289 00:34:24,750 --> 00:34:30,810 a better position to make things up, a picture where if you have a patient with a cough, 290 00:34:31,290 --> 00:34:39,780 you have to ask questions about how long you've had it and what time is that cough around the district and so on and so on. 291 00:34:40,590 --> 00:34:45,600 But the experienced doctor will just ask one or two questions about the most complex case. 292 00:34:46,260 --> 00:34:51,660 If you come along to me and you say I had a headache, which was like a thunderclap. 293 00:34:51,870 --> 00:34:55,050 Yeah, you don't ask any more questions then you're more or less not. 294 00:34:56,040 --> 00:34:59,700 That's a single, I think, or a red herring quite often. 295 00:34:59,940 --> 00:35:12,809 Okay. Because she was what's happening. And one of her suggestions will be that certain red herrings that says the opposite red flags, 296 00:35:12,810 --> 00:35:17,790 in other words, suddenly telling you that there's something to look up to, to be worried about. 297 00:35:18,180 --> 00:35:24,860 So let's think a bit harder about this. Okay. But the environment, if you know that you're in there about should your you're ache. 298 00:35:24,870 --> 00:35:33,390 Right. And again the anxious about it, that's the time to make sure that you think about it another way and stop and get back again. 299 00:35:34,050 --> 00:35:46,709 Another one might be and that that is why this work is important that we attach to specific diagnostic not diagnostic readings, 300 00:35:46,710 --> 00:35:57,270 but blues critical views and restaurant red flags and special steps of the diagnostic process. 301 00:35:58,290 --> 00:36:05,820 And if you look at the beginning, the initiation phase, yeah, they keep talking about salient feature. 302 00:36:06,390 --> 00:36:12,630 Maybe I'll just come into this salient feature in this initiation while you're talking then because this paper, 303 00:36:13,020 --> 00:36:21,870 the initiation closure of the cognitive process, that's the beginning in the end are the most exposed to risk of ever error. 304 00:36:23,040 --> 00:36:27,290 Cognitive biases developed early. The pro athletes were at the end. 305 00:36:28,580 --> 00:36:35,629 And so somehow you either are at the beginning of a diagnostic process and this is where that work about heuristics you anchor 306 00:36:35,630 --> 00:36:42,890 in you think it's too early in the or at the end closure because you may be rushing or you're not quite sure and you say, 307 00:36:42,900 --> 00:36:48,530 well the patient's well and actually you should have carried on correcting some information. 308 00:36:50,090 --> 00:36:53,330 How does that help the clinician then think about improving their performance? 309 00:36:54,980 --> 00:37:01,640 And if you have used a certain salient feature for a long time, you know, and based your diagnosis on it, 310 00:37:01,760 --> 00:37:11,530 which we tend to do, then from time to time, you should take time off, time out and look it up on the literature. 311 00:37:12,170 --> 00:37:18,590 So what we're saying is because we know that I know I mean, I've seen that some people develop rule based where they're clearly wrong. 312 00:37:19,190 --> 00:37:23,130 I mean, I've seen one. He can't have pulmonary embolism unless you got a company. 313 00:37:23,240 --> 00:37:26,510 What is that? Somebody that articulated to me. And that's clearly wrong. 314 00:37:26,510 --> 00:37:30,889 There are many of these theories. So we're going to finish up now. 315 00:37:30,890 --> 00:37:34,490 I just want what do you think, all the take home messages? 316 00:37:35,150 --> 00:37:43,219 I think for two types of people. One is for the clinician out there who's trying to improve with reasoning and diagnostics. 317 00:37:43,220 --> 00:37:47,840 And second, for those that try to teach this type of work to medical student, 318 00:37:47,840 --> 00:37:55,069 these issues are the best thing to do would be to work in an environment which supports this in 319 00:37:55,070 --> 00:38:03,649 that I think there's a lot of a lot of writing about this and we know a lot about this as people. 320 00:38:03,650 --> 00:38:05,990 We have it differently in different environments. 321 00:38:06,740 --> 00:38:12,350 And if you work in a supportive environment where you can talk with colleagues and review their activities 322 00:38:12,350 --> 00:38:19,490 with discussing cases and then looking at as to what should be happening and work with a facilitator. 323 00:38:19,850 --> 00:38:23,360 That would be a number one, in my opinion. Okay. Okay. 324 00:38:24,080 --> 00:38:29,510 Okay. Well, I'm going to let conclusion from Alpha Alpha and finish of this conversation. 325 00:38:29,510 --> 00:38:35,240 I thought, this is a nice finish, but if diagnostic errors can never be entirely eliminated. 326 00:38:35,780 --> 00:38:40,610 Human reasoning is not perfect. And so mistakes in interpretation in parents will be made. 327 00:38:41,060 --> 00:38:45,830 Sooner or later situations will be encountered where one's knowledge will be incomplete. 328 00:38:46,730 --> 00:38:49,610 Feedback in this clinical setting setting is spotty. 329 00:38:50,660 --> 00:38:56,750 Second, even if our knowledge were complete and our inference process with perfect clinical evidence is not. 330 00:38:57,410 --> 00:39:00,830 Thank you very much, John Barlow, here today at the Centre for Evidence based Medicine. 331 00:39:00,830 --> 00:39:03,020 It's been a pleasure. Thank you.