1
00:00:01,280 --> 00:00:04,820
Hi. Good afternoon, everybody. I think we'll make a start.

2
00:00:05,900 --> 00:00:12,890
Obviously, a few people are still filling themselves up on lunch. I'd like to congratulate you on making it halfway.

3
00:00:14,540 --> 00:00:19,550
I think we may have lost one or two people by now, but some of you are still here for more activities.

4
00:00:20,180 --> 00:00:25,060
I'm going to talk to you about diagnostic reasoning. I think it's interesting.

5
00:00:25,070 --> 00:00:30,230
I have one mission in this talk. You know why some people are going to achieve lots of things.

6
00:00:30,890 --> 00:00:35,870
I'm going to teach you about this and this and this. My mission with Julie Ashwood.

7
00:00:37,340 --> 00:00:44,479
Julie said to me this morning, I'm not really into diagnostics and I'm not really into the teaching of it.

8
00:00:44,480 --> 00:00:48,410
Even more so, my mission, apart from mission two,

9
00:00:48,410 --> 00:00:53,840
I'll tell you Mission two is is to try and change your mind about diagnostics and try and change your mind.

10
00:00:54,620 --> 00:01:00,920
The second one is, at some point I may roam around the room and I have to try and get Olive in the shot at some point.

11
00:01:01,940 --> 00:01:05,059
Okay. About five years ago,

12
00:01:05,060 --> 00:01:08,510
I came on this course and somebody was delivering this talk from diagnostics

13
00:01:08,510 --> 00:01:15,380
about this time in the afternoon and I was sat the a mike Crowley is there so

14
00:01:15,500 --> 00:01:20,270
fiddling around with my bag and after about 5 minutes I started to switch off

15
00:01:20,270 --> 00:01:24,410
and do something else because all the information was going over my head.

16
00:01:25,100 --> 00:01:27,679
Okay. So what I decided then was I thought,

17
00:01:27,680 --> 00:01:33,050
I'm going to start thinking about this and what I'm going to talk to you about today is if you join me on this hour,

18
00:01:33,400 --> 00:01:37,880
I'm just going to talk to you about things I find useful in my practice.

19
00:01:38,960 --> 00:01:42,230
I'm not going to talk to you about fancy calculations and how difficult,

20
00:01:42,230 --> 00:01:46,820
but I'll show you the simple bits are you and I want you to keep that in mind.

21
00:01:47,690 --> 00:01:53,120
Often what happens here? People teach you research methodology and you get lost.

22
00:01:54,380 --> 00:01:57,140
So to illustrate that point, this is my daughter, Katie.

23
00:01:58,310 --> 00:02:03,230
And as you can see there at the back, they were discussing about the calculation, she's only five years old.

24
00:02:03,230 --> 00:02:10,520
And in our house we like to keep things simple. And she's calculated a mean difference there.

25
00:02:11,750 --> 00:02:14,810
And for the sharp people this morning, it was in Amanda's lecture.

26
00:02:15,590 --> 00:02:23,270
You'll realise here she's got the confidence interval in. However, in our house we really like to do it well.

27
00:02:23,840 --> 00:02:28,610
So here at the bottom is my eight year old giving her a bit of advice and how to get it right.

28
00:02:29,510 --> 00:02:35,030
But it's important we have to and I've been doing that in my group is keeping things simple so we can understand them.

29
00:02:36,650 --> 00:02:41,090
What I did, though, is I said to people sometimes and some people want to teach her.

30
00:02:41,750 --> 00:02:47,660
Sometimes I put things on my slide in yellow that allow me to remind to just a reminder to me to do something.

31
00:02:48,380 --> 00:02:50,860
I'm going to give you an illustration of diagnostic reasoning.

32
00:02:50,870 --> 00:02:59,929
Actually, in practice, about two months ago, my daughter was ill in the eye and I've been in training now for eight years.

33
00:02:59,930 --> 00:03:06,290
I've been qualified eight years. And you get to that point where you suddenly think, Oh my God, I'm not quite sure if she's really doing well here.

34
00:03:07,250 --> 00:03:09,410
I'm really quite worried about my daughter.

35
00:03:09,740 --> 00:03:16,460
She had diarrhoea, tummy pain, fever with 41, and I'm going, Oh my gosh, you might want to go and tell me what to do.

36
00:03:17,390 --> 00:03:20,810
Now I'm sure those of you with children will recognise that from position.

37
00:03:22,520 --> 00:03:29,479
So actually what I did do is bring my general practitioners in the morning and board with the receptionist.

38
00:03:29,480 --> 00:03:33,260
Difficult not to mention the word.

39
00:03:33,260 --> 00:03:37,030
I'm a doctor, by the way. And she went, Oh, right.

40
00:03:37,040 --> 00:03:45,469
Okay. What we'll do is we'll get you a phone call. Now, I'm at 19 Beaumont Street and actually the person who then spoke to me on the end of the line,

41
00:03:45,470 --> 00:03:49,880
if polled there at the back it was practising that day, you might have even forgotten it by now.

42
00:03:51,440 --> 00:03:54,979
So he gives me the advice he talked me through when I'm feeling a bit more relaxed.

43
00:03:54,980 --> 00:03:58,220
And then he says to me, What about the jump test?

44
00:03:59,580 --> 00:04:05,760
And suddenly bang in my mind came back. The scenario I remember in my training there was this thing about rebound.

45
00:04:05,760 --> 00:04:11,930
And if you've got this acute abdomen, these ask questions like, What's it like in the car or in a child?

46
00:04:11,940 --> 00:04:18,909
What's it like if they jump up and down? And what happened is a minute earlier, she'd just come down the stairs and,

47
00:04:18,910 --> 00:04:21,940
you know, like children do, they just jump off the end of the steps, don't they?

48
00:04:22,360 --> 00:04:27,549
And I immediately relaxed. I said, Well, gosh, I feel better.

49
00:04:27,550 --> 00:04:32,280
I can manage this. And did the phone call that afternoon.

50
00:04:32,300 --> 00:04:36,440
She got better. Fever subsided. She felt better. You know, children instantly.

51
00:04:36,440 --> 00:04:45,270
The better the running the bell. However, that evening, about 5:00, I suddenly said to myself, where is the evidence for that?

52
00:04:49,370 --> 00:04:53,629
You know, I'm an evidence based practitioner. Let's do some deliberate practice.

53
00:04:53,630 --> 00:04:58,280
So I actually looked it up on PubMed. The advantage of PubMed, you can do it in your own home.

54
00:04:59,060 --> 00:05:05,420
Took me about a minute to find the article. In fact, in the acute assessment of children, the jump test is completely useless.

55
00:05:10,800 --> 00:05:14,270
I thought, right. The next day I'm back to work.

56
00:05:15,290 --> 00:05:20,660
I'm running to the office. By the way, Paul, do you know that assessment of the jump test?

57
00:05:21,890 --> 00:05:26,520
He said, I know. I said, What do you mean, you know?

58
00:05:26,540 --> 00:05:31,190
And he said, Well, I looked it up on my phone. It's completely useless.

59
00:05:33,170 --> 00:05:36,370
That's how we think. Okay. That's evidence based practice thinking.

60
00:05:36,380 --> 00:05:39,620
If feel you use that information.

61
00:05:39,890 --> 00:05:44,210
Questioning your uncertainty. Okay, so we're going to have a bit of fun in this session.

62
00:05:44,420 --> 00:05:51,500
Okay. I will stretch the time to the point when I start to think you're all slightly going to sleep and I'm losing you.

63
00:05:51,710 --> 00:05:54,930
I will finish immediately then, if I think we're doing very well.

64
00:05:54,950 --> 00:05:58,790
I'll carry on. Okay. Two things.

65
00:05:59,060 --> 00:06:00,860
I've been having a lot of fun with Amanda with this.

66
00:06:01,340 --> 00:06:08,720
I think the first thing is in diagnostic skills, we have to accept that there are things we do not know out there.

67
00:06:10,160 --> 00:06:14,210
And the second thing is the way we perceive information can be faulty.

68
00:06:16,550 --> 00:06:20,750
I'm going to deal with the second one and then I'm going to talk about strategies that can improve what we do.

69
00:06:21,290 --> 00:06:24,890
Okay. You've all got this picture up and some of you should have this picture.

70
00:06:26,090 --> 00:06:31,490
Now, if you look at this checkered square squares marked, B, are they the same shade of grey?

71
00:06:35,090 --> 00:06:39,829
How many? No. Hands up. Okay.

72
00:06:39,830 --> 00:06:44,680
Most of you. How many think? Yes. One or two or three.

73
00:06:44,800 --> 00:06:48,560
Okay. This is an illustration of our perception now.

74
00:06:48,580 --> 00:06:52,389
Somebody has got one of these sheets, have the. And I borrow your sheet there.

75
00:06:52,390 --> 00:07:00,370
Sir, if you don't mind, I want you to just do something. If you start to follow the sheet over and bring A to B and then match the shading,

76
00:07:02,020 --> 00:07:07,870
what you should start to see is that actually A and B are identical shades of grey.

77
00:07:10,680 --> 00:07:13,770
You can take this away with me. Great.

78
00:07:14,140 --> 00:07:17,880
Olive. Disagree. Can you just come down here with your paper, please?

79
00:07:19,140 --> 00:07:28,470
On the right. Now this. Might you take it away? And however much I looked at this left down slide, I still think I used different to be.

80
00:07:30,550 --> 00:07:37,430
Okay. You can take that away with. You can. Okay.

81
00:07:37,480 --> 00:07:41,790
I'm going to show you a second one. That's even more fun. Okay. Okay.

82
00:07:41,800 --> 00:07:48,190
The car picture. Are these cars the same size?

83
00:07:50,830 --> 00:07:55,330
Now, you all think I'm playing tricks now? I could be doing this. Did I look the same size?

84
00:07:56,710 --> 00:08:01,210
Okay, I think that's important because this is how we see information.

85
00:08:01,220 --> 00:08:07,870
So when I look at my when I make mistakes in diagnosis, I can't understand sometimes how I've made the mistakes,

86
00:08:08,500 --> 00:08:15,370
but information is presented to it's the same information. So when I measure these cars now watch where the wheelbase.

87
00:08:15,380 --> 00:08:19,330
What happens to the wheelbase? You see that?

88
00:08:20,440 --> 00:08:26,380
I'm going to move to the side as the wheelbase got shorter or longer.

89
00:08:28,980 --> 00:08:39,700
And what about that? So say no matter how much I look at this, I cannot explain why that the same.

90
00:08:40,650 --> 00:08:47,730
However, the information surrounding the cars is in perspective and different, and it's making you look at things differently.

91
00:08:48,240 --> 00:08:52,770
So the context we see a person affects how we see them in terms of diagnosis.

92
00:08:53,610 --> 00:09:00,780
If I see them in general practice, they're less likely to have a heart attack than if I see them in in the emergency department.

93
00:09:01,380 --> 00:09:05,160
The context of the information is actually going to change what we do.

94
00:09:05,820 --> 00:09:11,430
So what we did about three years ago is actually on a on a train about three years ago,

95
00:09:11,430 --> 00:09:15,630
I thought there has to be more to diagnosis than what is just a two by two box.

96
00:09:16,950 --> 00:09:20,490
And I was on a train with Paul and we started to flesh out what the strategies are.

97
00:09:21,330 --> 00:09:27,630
And I want you to think of the strategies just like I did then as measuring effects, trying to improve the way we do things.

98
00:09:28,710 --> 00:09:35,980
And I'm going to talk to you about a piece of work that we've been ongoing for about three years, about what I think evidence based diagnostics have.

99
00:09:37,910 --> 00:09:43,370
I may lose you at some points. And if I'm losing you, you can do what Amanda did and just put your hand up and shout at me.

100
00:09:44,570 --> 00:09:46,130
Okay, so let's stop.

101
00:09:47,930 --> 00:09:58,880
I want you to just imagine this is me as a GP Monday morning in the clinic, feeling a bit tired in need of my first cup of coffee of the day.

102
00:09:59,390 --> 00:10:05,210
And I have a message on my screen. There is a child coming in who is on well with a fever.

103
00:10:05,570 --> 00:10:09,080
I have booked her into your emergency slot now.

104
00:10:09,230 --> 00:10:15,470
I want you to just write down the one thing or two things that come to mind in terms of what the diagnosis is.

105
00:10:17,030 --> 00:10:23,500
I'm going to give you 10 seconds just to write this down. If you don't know anything, that's fine.

106
00:10:29,160 --> 00:10:29,610
Okay.

107
00:10:31,680 --> 00:10:38,250
As you pop out for your next patient, hoping to see a cup of coffee out there, you ask the receptionist if there was anything else about the show.

108
00:10:40,360 --> 00:10:45,070
Nice receptionist says to you. Oh, yes. Her mom said she had a rash or has a rash.

109
00:10:46,670 --> 00:10:52,460
Okay. Write down the one thing that comes to your mind. You don't have to be a clinician here because remember, we're all parents.

110
00:10:53,540 --> 00:10:58,700
We all do this diagnostic every day. Okay.

111
00:10:58,910 --> 00:11:05,330
Number three, a child comes in to see a three year old is cries all for those?

112
00:11:05,330 --> 00:11:09,569
That's a runny nose. It's had a temperature for two days of 39 degrees.

113
00:11:09,570 --> 00:11:16,100
Three. The mom tells you that you check the temperature. The rash began this morning and is on the body and it's on the face.

114
00:11:17,750 --> 00:11:26,860
Write down what comes to mind now. Lots of you writing down only a few people activity wise.

115
00:11:27,790 --> 00:11:32,709
Okay, that is the rough. You say I want to look at the rough. That's the rough bit.

116
00:11:32,710 --> 00:11:36,340
Slightly out of focus. Write down.

117
00:11:36,340 --> 00:11:44,080
What? What? Some people are looking at that going, Oh my gosh, the GPU's at the back of the room are going.

118
00:11:44,350 --> 00:11:53,080
There we go. Okay. Before you discuss with the mother what you consider the diagnosis to be, you ask her what she thinks it might be,

119
00:11:54,400 --> 00:11:59,770
and the mother says to you, Well, I'm particularly concerned because at nursery at the moment, there's a lot of children with chickenpox.

120
00:12:01,390 --> 00:12:15,870
Write down what you think the diagnosis is now. Okay.

121
00:12:16,200 --> 00:12:20,230
So let's start here. What did people put down here as a possible diagnosis?

122
00:12:20,250 --> 00:12:29,760
Only one piece of information such as urinary tract infection, infection, flu.

123
00:12:32,040 --> 00:12:35,310
Tonsillitis. Meningitis.

124
00:12:37,360 --> 00:12:43,520
Anyone with any viral illness. Okay, so simple pieces of information.

125
00:12:43,540 --> 00:12:49,510
Single pieces of information. However much you think nothing's going on is setting off a trigger already in your mind.

126
00:12:50,920 --> 00:12:53,950
Immediately. We're starting to think diagnostic strategies.

127
00:12:56,810 --> 00:13:03,720
Okay. You got the second piece of information. And what do people think now?

128
00:13:04,140 --> 00:13:08,010
Measles. Measles, meningitis.

129
00:13:10,340 --> 00:13:18,200
Okay. Wherever I've been. That's about the two two particular diagnosis that everybody jump forward is particularly the meningitis.

130
00:13:19,340 --> 00:13:22,970
Particularly apart from emergency departments, where now they're completely lost.

131
00:13:23,480 --> 00:13:28,400
But everybody goes in that direction. Okay. Got this additional information.

132
00:13:30,050 --> 00:13:35,570
What do people think then? Chicken pox.

133
00:13:35,870 --> 00:13:39,980
Measles. What did you think? Chicken pox. Anybody else think chicken pox at this time?

134
00:13:42,320 --> 00:13:45,560
One, two, three, four, five. Okay.

135
00:13:45,740 --> 00:13:49,910
That's a classical textbook definition of how you diagnose chickenpox.

136
00:13:51,220 --> 00:13:54,530
Okay. That's exactly how it appears in the textbook.

137
00:13:54,550 --> 00:13:59,410
That's what we teach people. Only four or five of you actually could recognise a.

138
00:14:00,460 --> 00:14:04,210
What about this? Okay, a lot of you are now more certain.

139
00:14:05,350 --> 00:14:09,570
Okay. And what about when the mother said she had chickenpox?

140
00:14:11,390 --> 00:14:15,390
Oh, she was worried about the nurse. Does anybody what do people think it potentially of now?

141
00:14:16,140 --> 00:14:25,320
Okay, I'm going to illustrate. So what we've done there is we've used what are called the initiation strategies, initiation of the diagnosis.

142
00:14:25,590 --> 00:14:31,440
You've used for possible strategies to initiate the diagnosis, the spot diagnosis with the picture.

143
00:14:31,590 --> 00:14:34,530
Some people can immediately key into the spot diagnosis.

144
00:14:34,950 --> 00:14:42,180
The self labelling patients or the professionals will ask you or tell you there is a particular thing they are concerned about.

145
00:14:42,420 --> 00:14:49,540
Can you look at this patient? I'm concerned. They have cellulitis. I'm concerned about having a chest infection in this case.

146
00:14:49,750 --> 00:14:57,370
The mother was concerned about other children with chickenpox presenting complain does set off of actual strategy.

147
00:14:57,370 --> 00:15:02,170
But what you could see at the presenting complaint stage is you got a wide variety of possibilities.

148
00:15:03,200 --> 00:15:08,600
And then actually we got this pattern recognition trigger. It relies on a particular memory.

149
00:15:08,840 --> 00:15:15,020
Only a few people have some memory of that. Recognises the actual pattern of chickenpox.

150
00:15:16,670 --> 00:15:24,170
Okay. So we've actually done this, we've gone forward as an actual piece of work actually with our GP's in Oxford to say do these exist?

151
00:15:24,590 --> 00:15:30,500
And actually this is just a piece of diagram where they collected 50 new cases and said What initiates the strategy?

152
00:15:30,770 --> 00:15:38,839
Each circle represents one of the gaps and as you can see, if they add up to more than 100%,

153
00:15:38,840 --> 00:15:43,960
so you use more than one strategy at any one time to trigger the possibility.

154
00:15:45,660 --> 00:15:54,030
So you can see here. Now, this is important. About 20% of diagnoses in primary care rely on a diagnosis.

155
00:15:55,570 --> 00:15:58,900
Now, Mark, you're an undergraduate or a teacher in general practice.

156
00:15:59,260 --> 00:16:03,190
How much diagnosis do you think we actually take in the medical education?

157
00:16:05,360 --> 00:16:10,280
A law that I never had any teaching of spot diagnosis.

158
00:16:11,270 --> 00:16:16,130
One in five of my presentations rely on some instantaneous recognition of the problem.

159
00:16:19,630 --> 00:16:28,690
I like this self labelling. In fact, actually and I'll come back to this put in the when I did this with the emergency department registry,

160
00:16:28,690 --> 00:16:32,800
I actually believe that you would ask the patient in some way they would be correct.

161
00:16:35,140 --> 00:16:39,310
Okay. They said the only time they took notice of self labelling was when you actually

162
00:16:39,310 --> 00:16:44,020
got right to the end of the consultation and you had no idea what was going on.

163
00:16:45,280 --> 00:16:54,730
Then it was worth actually asking the patient. You can see here, presenting complaint always triggers off some hypothesis almost quite a lot of time,

164
00:16:54,910 --> 00:16:58,660
and the pattern recognition is actually a smaller component, but it still exists.

165
00:17:00,910 --> 00:17:04,040
Okay. So I want you to do this one more. Go at this.

166
00:17:04,060 --> 00:17:08,950
This is actually this is last week in practice. Last Monday, this is Miriam, an eight year old.

167
00:17:09,850 --> 00:17:18,100
So I want you to think of the four possible strategies. And then soon as you come up with a diagnosis in your mind, I want you to put your hand up.

168
00:17:19,840 --> 00:17:26,530
So Miriam comes in to see me. She thought, what's? And you know, and you go out to get the patient and she walks in a little bit slowly.

169
00:17:26,770 --> 00:17:31,749
Not too bad. Doesn't look, but, you know, hunched over a little bit, comes and sits down.

170
00:17:31,750 --> 00:17:39,399
Okay. And I say, you sit there as a GP and you try and not say too much and you sit there and I go, okay, Miriam, what can I do for you?

171
00:17:39,400 --> 00:17:44,440
And she said, I've got a problem with diarrhoea and a bit of flu.

172
00:17:46,440 --> 00:17:51,480
And I've not been feeling too well. I'm listening, and I'm trying to encourage you not to interject.

173
00:17:52,530 --> 00:17:58,530
And I've got back pain. I'm going.

174
00:17:59,400 --> 00:18:02,400
Okay. Tell me about the back page. That's been pretty painful, actually.

175
00:18:03,930 --> 00:18:07,270
It's down in the lower bit here. Nobody's got the handle yet.

176
00:18:07,290 --> 00:18:10,700
You're all cagey. So I've got this. Oh, Paul's got his hand up.

177
00:18:10,710 --> 00:18:15,420
Good. I've got the back pain here, and it comes around the side.

178
00:18:17,020 --> 00:18:20,770
I think diarrhoea, gastroenteritis, old lady back pain.

179
00:18:20,770 --> 00:18:24,579
And it comes around the side. I said anything funny about this?

180
00:18:24,580 --> 00:18:29,540
She says, yes, and I've got a rash as well. Another person putting their hand up.

181
00:18:30,680 --> 00:18:34,400
Okay. I said, What about this? She said, Well, it's tingling.

182
00:18:36,300 --> 00:18:41,590
Okay. The more hands are going up. And so I said to her, So what do you think?

183
00:18:41,610 --> 00:18:45,840
I said, Can I have a look at this rash, if you don't mind? Okay, there's the rash.

184
00:18:48,510 --> 00:18:53,520
And then I asked her, What do you think it is? He said, My husband had something similar to this.

185
00:18:55,640 --> 00:18:59,370
I think it was called shingles. Okay.

186
00:18:59,370 --> 00:19:03,150
So some of you then. And that's the example there. That's how I think.

187
00:19:04,410 --> 00:19:10,400
Did you see what was happening? Was going in the wrong direction in the first place with the diarrhoea and the gastroenteritis?

188
00:19:10,410 --> 00:19:14,280
I was thinking, this patient has gastroenteritis. I shifted focus.

189
00:19:15,650 --> 00:19:18,709
Because she said back pain. Back pain can come with abdominal pain.

190
00:19:18,710 --> 00:19:22,310
You start to get a bit more worried. But then she starts a rash.

191
00:19:22,310 --> 00:19:26,600
We get in a pattern, a period. And I had the tingling and some people went bang.

192
00:19:27,350 --> 00:19:32,960
Some people can look at that and go, that is shingles can look at a patient and say, that is shingles.

193
00:19:33,560 --> 00:19:39,550
Some people have never seen it before. And then when she qualified it by saying, My husband have it.

194
00:19:40,640 --> 00:19:47,090
We can build the information off of how I think initially and that's the end of my diagnosis, in fact,

195
00:19:47,090 --> 00:19:51,700
because I shut my mouth for 2 minutes and thought about where my strategy for that's the end.

196
00:19:51,710 --> 00:19:55,430
I can then move into the consultation about what we're going to do for this patient.

197
00:19:56,390 --> 00:20:01,969
So that's the initiation. Okay, I'm going to get you to think now,

198
00:20:01,970 --> 00:20:06,680
but you've got some I'm going to give you an activity for a few minutes just to illustrate some of these points.

199
00:20:07,190 --> 00:20:10,519
So at the moment, you've all got this little quiz. So I'm going to give you about 2 minutes.

200
00:20:10,520 --> 00:20:12,410
I'd like you to just have a go at this quiz.

201
00:20:17,280 --> 00:20:26,160
Now for the highest scoring person in the quiz, I have two copies of the User's Guide to the Medical Literature to give away.

202
00:20:26,550 --> 00:20:32,010
So there's an incentive. These are not by Blackwell's, by the way.

203
00:20:32,610 --> 00:21:12,750
They can't buy them at the bookshop. Okay.

204
00:21:13,140 --> 00:21:19,470
Good, good. Now I need to remind myself of the questions.

205
00:21:21,210 --> 00:21:24,750
Question one Was this here is this.

206
00:21:27,120 --> 00:21:32,520
John four Give yourself a month. If you put John for all, inspect them off.

207
00:21:34,230 --> 00:21:40,170
I'll give you a mark for inspect them. All he's got here is John four.

208
00:21:40,170 --> 00:21:48,990
You'll have to agree with that. What neurological condition does John feel forth have stuck with about Parkinson's?

209
00:21:49,860 --> 00:21:55,680
Because I call foot drop. Give yourself a mark. If you've got foot drop, how many people got foot drop?

210
00:21:57,390 --> 00:22:02,640
All people. How many people? Watch Inspector Morse. Now, you could have a mark for polio.

211
00:22:02,640 --> 00:22:05,760
Lots of you. Now, John, for every episode walks like this.

212
00:22:05,910 --> 00:22:11,010
The cameras caught him off, but he walks like this. He had polio as a child and he got dropped.

213
00:22:11,430 --> 00:22:16,650
And if you're watching another edition, you'll see walks like this. The camera often cuts him off there.

214
00:22:16,920 --> 00:22:20,540
But if you don't know that to spot that, you will not be able to get that.

215
00:22:23,140 --> 00:22:26,380
Okay. The next question. Question three What's this?

216
00:22:27,990 --> 00:22:32,340
PADGETT Well, we know we're going to get full.

217
00:22:32,670 --> 00:22:37,980
The question for what bridge is this, which is okay.

218
00:22:38,160 --> 00:22:46,050
And the question five is, which is the nearest pub, my favourite pub, apart from the white horse next to Blackwell, the Turf Tavern.

219
00:22:48,330 --> 00:22:55,920
If you haven't been down that street, you can't recognise it. You see that if you have been down that street, you can instantly recognise that.

220
00:22:57,120 --> 00:23:01,290
If you haven't been to the Turf Tavern, what you should consider going there before you leave.

221
00:23:02,580 --> 00:23:10,290
It's a very nice place to go for a drink if you can find it. But that's a good illustration of what I call a spot diagnosis.

222
00:23:11,160 --> 00:23:15,660
There's no either or. You can't make it up. I can't give you additional information.

223
00:23:16,080 --> 00:23:23,100
You either do know it or you don't. So there are lots of areas in your in your when you go back to your health care,

224
00:23:23,100 --> 00:23:28,979
whether it's midwifery or nursery or it's in general practice, where you can say, actually, what are the spot diagnoses?

225
00:23:28,980 --> 00:23:33,840
Our students need to know. Remember the 20% an experience will gain them.

226
00:23:34,710 --> 00:23:37,800
Question six What's this? Okay.

227
00:23:38,640 --> 00:23:42,720
Question seven Is the building what's this?

228
00:23:43,320 --> 00:23:52,500
Divinity school? The divinity school. It's part of the Bodleian Library and it's been in Harry Potter.

229
00:23:54,190 --> 00:23:58,950
Okay. I, too, have some discriminatory questions in case you're really smart and you've done this before.

230
00:23:59,520 --> 00:24:04,360
And the last one. It's conjunctivitis.

231
00:24:05,920 --> 00:24:10,570
Okay. Anybody score ten, nine, eight.

232
00:24:12,310 --> 00:24:21,219
Gotcha. I'll make an easier test. Seven, seven, seven, six, six.

233
00:24:21,220 --> 00:24:24,400
For Mike and another chap at the back there.

234
00:24:24,820 --> 00:24:28,780
Congratulations. Now, if you give that one to Mike as well.

235
00:24:28,930 --> 00:24:33,960
Now, it's an important point. I wanted to do that just to reinforce about the issue of spot diagnosis.

236
00:24:33,970 --> 00:24:37,840
It's about experience. You can all see that you have to see these things.

237
00:24:38,050 --> 00:24:42,090
You have to see these things in practice to know what you're doing. Okay.

238
00:24:42,520 --> 00:24:47,830
Self labelling by patients in GP consultations. Let me get up to 50 cases.

239
00:24:48,100 --> 00:24:51,400
This is what patients often came in with tonsillitis.

240
00:24:52,810 --> 00:24:58,270
I've had it before, Doctor, and I've got tonsillitis again. I have a chest infection, Doctor.

241
00:24:58,270 --> 00:25:02,620
I'm sure we'll all recognise. I don't. I have a groyne strain.

242
00:25:02,950 --> 00:25:06,110
I have. Most people know when they've got growing ankle sprain.

243
00:25:06,120 --> 00:25:11,080
That's primary cold. I have asthma. Even new people who've never had it before.

244
00:25:11,470 --> 00:25:15,220
And I have a UTR, just like the last time. This is women.

245
00:25:16,300 --> 00:25:17,560
So this is an interesting one.

246
00:25:17,740 --> 00:25:24,550
If you have a patient who labels themselves as having a urinary tract infection, a woman who's had a recurrent urinary tract infection.

247
00:25:24,970 --> 00:25:32,860
How often are they likely to be correct? So that's this is going to be now an interesting question that I asked, because this is what happened.

248
00:25:34,360 --> 00:25:38,780
Okay. We formulated a question like we know know how to do it slightly different.

249
00:25:38,800 --> 00:25:46,270
Here we have the indicator or the intervention or the test. And this was the question I was interested in, in a woman with recurrent uterine.

250
00:25:47,500 --> 00:25:52,900
Can she treat herself? And how well does that compare to a microbiological culture?

251
00:25:53,290 --> 00:25:56,589
And is she correct in the diagnosis and treatment, though?

252
00:25:56,590 --> 00:26:01,989
Basically, can a woman who's had a UTI before come into my surgery tell me she's got a urinary tract

253
00:26:01,990 --> 00:26:07,420
infection again and I feel certain or secure that she's pretty accurate in what she's doing.

254
00:26:08,820 --> 00:26:10,620
Okay. So we looked on PubMed.

255
00:26:10,620 --> 00:26:18,900
Now you're all familiar with and there's a particular trial in 2001 in the Annals of Internal Medicine that actually matches our question.

256
00:26:20,020 --> 00:26:23,290
There's a recurrent urinary tract infections in young women.

257
00:26:23,980 --> 00:26:29,020
Patient initiated treatment of uncomplicated OC.

258
00:26:29,170 --> 00:26:35,620
And here's a measurement of the accuracy of self-diagnosis determined by evidence of a definite culture,

259
00:26:35,620 --> 00:26:40,900
positive or probable sterile period and no alternative diagnosis.

260
00:26:44,240 --> 00:26:50,160
Okay. And here's the outcome. 88 of 172 women self-diagnosed.

261
00:26:50,160 --> 00:26:59,300
The total of 172 UK laboratory evaluation showed a euro pathogen in 144 cases 84%.

262
00:27:00,860 --> 00:27:08,450
So if a woman comes in to see me tomorrow is had a urinary tract infection and says, I have another one, what do you think I do?

263
00:27:10,810 --> 00:27:15,220
Do you think I send off a culture? You think I just relax and go?

264
00:27:15,230 --> 00:27:17,650
Here you go. Here's your prescription. I believe you.

265
00:27:17,650 --> 00:27:23,590
I make sure she's telling me the right story of the symptom, but in effect, she's about 85% correct.

266
00:27:24,280 --> 00:27:30,970
However, the thing is, with self labelling, lots of areas where we don't know where people are very good at self labelling.

267
00:27:32,620 --> 00:27:37,720
Yeah, all these other areas. But in that area I feel I have some evidence to back me up.

268
00:27:38,110 --> 00:27:41,130
I've changed the way I practice in effect. Okay.

269
00:27:41,980 --> 00:27:45,400
I'm going to move on now to what we call refinement of the diagnosis.

270
00:27:45,400 --> 00:27:48,640
I've showed you areas where you can simply jump in and go there.

271
00:27:48,730 --> 00:27:55,000
The diagnosis, chickenpox, shingles, urinary tract infection.

272
00:27:55,210 --> 00:28:00,400
There are quite a lot of them. But some areas what we end up with is uncertainty.

273
00:28:00,550 --> 00:28:09,310
We have a certain idea that there are two or three possibilities now, and what we'd like to do is to refine that to a single most likely cause.

274
00:28:10,000 --> 00:28:14,260
So the second area we found is what we call refinement of that diagnosis is you set

275
00:28:14,260 --> 00:28:18,309
off your hypothesis like at the beginning when we said this could be chicken pox,

276
00:28:18,310 --> 00:28:25,090
measles, it could be upper respiratory tract infection, but you need a number of rules or ideas to be able to help you.

277
00:28:25,330 --> 00:28:33,490
And these four, five rules we came apart particularly one is that all strategies are wanting to restrict the rule to is a stepwise refinement.

278
00:28:33,760 --> 00:28:38,740
The third is probabilistic reasoning and so forth with pattern recognition.

279
00:28:38,980 --> 00:28:46,260
Fit on the 50. The clinical prediction now. I'm going to explain some of these, how we use them and how we might use them.

280
00:28:47,310 --> 00:28:52,140
Okay. Restricted rule out. This is very interesting. This is new to me in the last two years.

281
00:28:52,890 --> 00:28:55,860
Is it a learned diagnostic strategy for each presentation?

282
00:28:56,430 --> 00:29:01,950
And there's a particular guy down the bottom here called John Murtagh from Australia who wrote a whole book about this.

283
00:29:02,340 --> 00:29:05,520
It's about common problems and a safe diagnostic strategy.

284
00:29:06,270 --> 00:29:09,990
And what you do is you try and find out what the common causes of the disease.

285
00:29:10,770 --> 00:29:15,330
And also what you're really interested in is what's going to kill you or cause you real harm.

286
00:29:16,950 --> 00:29:22,350
So, for instance, in a headache, it's very difficult in a headache to jump in and say there's a single core.

287
00:29:23,670 --> 00:29:29,340
But what you would do is to check for particular common causes, migraine tension type headache.

288
00:29:29,700 --> 00:29:37,140
But you also want to rule out the the other causes high blood pressure, temporal arthritis and subarachnoid haemorrhage.

289
00:29:39,280 --> 00:29:48,760
And in fact, Amanda talked to me the other day about your dad or his good friend and his temporal doctor and his headache and the pulsatile vein.

290
00:29:49,180 --> 00:29:53,020
And a restricted rule would have allowed you to pick that up because although

291
00:29:53,020 --> 00:29:57,310
you'd have got into the Common Core and remember at the beginning I said in you,

292
00:29:57,580 --> 00:30:04,690
we make mistakes, we look at things, and we go in and we jump in and we say, this is a migraine, but rules.

293
00:30:04,690 --> 00:30:10,930
And there are a number of these that can be applied. When you've got serious issues, you have to stop and think of the one or two causes.

294
00:30:15,080 --> 00:30:18,860
Prediction role. Okay. This is an interesting one.

295
00:30:18,860 --> 00:30:25,280
Prediction. Which ones do you use in practice? How many people who use a prediction rule?

296
00:30:27,950 --> 00:30:32,360
To people at the back. I'm going to exclude them because I know that I might use this one.

297
00:30:32,990 --> 00:30:36,410
Might urine three. Okay. Okay.

298
00:30:36,440 --> 00:30:44,510
Clinical prediction rules are like at the beginning when I did with the car, we're an attempt to put measurement in some way to help with what we do.

299
00:30:46,920 --> 00:30:56,829
There are a number of the GP's used. When we did this ultra uncool streptococcal full throat rule ABCD school for

300
00:30:56,830 --> 00:31:02,890
stroke risk had school for depression well rule for DVT and chest infection.

301
00:31:04,640 --> 00:31:08,310
Now this is slightly skewed population because all these are academic GPAs.

302
00:31:09,530 --> 00:31:15,770
So when you get a bunch of academic GP's together in a university department, you can get to about six prediction rules that are helpful.

303
00:31:17,430 --> 00:31:22,700
So it's not surprising when you're in a non-academic setting. But the number is often zero.

304
00:31:24,200 --> 00:31:29,090
But some of you may have heard of some of these. In your mind, you might have heard of things like the auto rental world.

305
00:31:30,620 --> 00:31:35,360
So the auto rentals that will take you, for instance, can you predict which patients have an ankle fracture?

306
00:31:37,560 --> 00:31:43,470
You can use clinical queries here again. In fact, I went into the systematic review here for ankle and rule.

307
00:31:45,580 --> 00:31:51,940
And before their BMJ systematic review the accuracy of of proof to exclude fractures of the ankle and midfoot.

308
00:31:52,210 --> 00:31:56,170
Systematic Review. There's a 22nd search for you.

309
00:31:56,770 --> 00:32:00,250
Okay. That's how useful clinical queries can be.

310
00:32:01,490 --> 00:32:05,380
You can look for systematic reviews in diagnosing. Okay.

311
00:32:05,560 --> 00:32:14,770
Let's go to Uncle Ruth. I put this up in an emergency department where they all said they'd heard of the old one call rules.

312
00:32:15,580 --> 00:32:24,610
I did what we did now, and I did that. Okay. I removed it from the speech, and I asked people to say, tell me, what is your uncle rule?

313
00:32:27,690 --> 00:32:31,460
This is what worried me now. Not one single person in the room.

314
00:32:31,470 --> 00:32:36,480
This is about 25 people could actually tell me the rule or the same variant of the rule.

315
00:32:38,440 --> 00:32:42,310
So we have these things that are incredibly helpful and I'll show you how helpful they are.

316
00:32:42,550 --> 00:32:46,810
But we have no strategy and have not worked out any way how to get them into practice.

317
00:32:47,650 --> 00:32:52,740
And all we ever get is how do we change practice at the back here? Do you see the problem?

318
00:32:53,100 --> 00:32:56,220
We've got simple things that will help us really well, and they're being ignored.

319
00:32:56,850 --> 00:33:00,930
We don't need to have fancy change. However, we need to develop this from the bottom up.

320
00:33:00,960 --> 00:33:06,720
You need to go back to your own practices, and if you did one thing, you could say, well, which room would help us in our practice?

321
00:33:07,470 --> 00:33:15,840
But let's look at how helpful the outward ankle rule with this instrument has a sensitivity of almost 100% and a modest specificity,

322
00:33:15,870 --> 00:33:21,330
and its use should reduce the number of unnecessary radiographs by 40% 30 to 40%.

323
00:33:23,830 --> 00:33:28,510
So if you apply that rule, you actually say there's a lot of money and a lot of hassle for patients.

324
00:33:29,650 --> 00:33:33,070
However, a GP is pretty smart in our place called Peter Roth.

325
00:33:33,070 --> 00:33:38,709
What he does is he actually keeps about these on his desk in a little car and he pulls

326
00:33:38,710 --> 00:33:41,740
out the wheels when he needs it and he pulls out the picture of this when you need it.

327
00:33:42,280 --> 00:33:44,710
Because if you get it wrong, you don't get the benefits you'd expect.

328
00:33:46,200 --> 00:33:52,230
So you could go back and you can look at this in any setting and determine which prediction rules might help you.

329
00:33:52,440 --> 00:33:59,309
But here's the problem. Now, if this goes out live, I might get in trouble with the if it went out on the net.

330
00:33:59,310 --> 00:34:01,560
But essential evidence plus is out there for sale.

331
00:34:01,620 --> 00:34:09,060
Some people of you might have bought this info point book and I was looking at this is a particular chap in our department called Matthew Thompson,

332
00:34:09,330 --> 00:34:15,720
who is reviewing it at the moment for us. And I don't know if you can see that just at the top, just there.

333
00:34:17,660 --> 00:34:26,660
How many? This is a clinical prediction rules that are 291 of these prediction rules in the essential evidence plus.

334
00:34:30,030 --> 00:34:33,389
You think that's a lot? Maybe not a lot of you.

335
00:34:33,390 --> 00:34:44,820
Not for a lot. What do people think? So I've given you 300, probably overwhelms you in some way.

336
00:34:46,770 --> 00:34:49,440
We haven't worked out of them, which would be useful.

337
00:34:51,560 --> 00:34:56,030
People have actually sat in their setting and said, Well, which ones do we think would help us in our practice?

338
00:34:56,360 --> 00:35:00,080
And just supplying free prediction rules is a waste of time.

339
00:35:01,010 --> 00:35:04,820
Any instrument that just froze them in your face is a waste of time because it's too many.

340
00:35:05,060 --> 00:35:11,660
And you haven't looked at the evidence and looked at your patient population and how it might help you and how it might affect how you practice.

341
00:35:12,890 --> 00:35:15,890
So just giving you a top down approach is a waste of time and prediction.

342
00:35:16,070 --> 00:35:19,550
You need to do it from the bottom up. Okay.

343
00:35:20,570 --> 00:35:24,290
I particularly like this one. This was the bit where we talked about pattern recognition.

344
00:35:27,480 --> 00:35:35,290
This was in the BMJ. Actually, I pointed out the BMJ, a 49 year old with severe flare up of ulcerative colitis,

345
00:35:36,010 --> 00:35:41,500
gets admitted to a hospital and she receives intravenous hydrocortisone.

346
00:35:42,460 --> 00:35:47,860
Amino salicylate and cyclosporine treatment because she's pretty and well and all that was started.

347
00:35:48,850 --> 00:35:53,830
And after seven days she was then given oral prednisolone at about 60 milligrams per day.

348
00:35:55,570 --> 00:35:59,560
And every day after that she started to write this note to the junior doctor.

349
00:36:01,870 --> 00:36:07,330
Every time you turn off, we get this pitch, this card. Day seven, day eight, day nine, day ten.

350
00:36:09,030 --> 00:36:14,350
What for? Diagnosis mania. And what's causing the mania.

351
00:36:15,190 --> 00:36:23,980
Steroids. Do I need to have probabilistic reasoning or likelihood ratios now that the pattern recognition fit?

352
00:36:24,520 --> 00:36:29,590
She's gone manic. She's gone psychotic, if you like, on a background of too many steroids.

353
00:36:30,470 --> 00:36:37,540
There's a pattern recognition that, okay, so we build in the number of different strategies that you can employ and think about and teach about.

354
00:36:38,380 --> 00:36:42,220
You don't need to know a two by two table. You don't need a calculation.

355
00:36:43,600 --> 00:36:47,420
You just need to know that these sort of patterns exist and they're helpful.

356
00:36:47,440 --> 00:36:50,740
You won't see a lot. Okay.

357
00:36:52,840 --> 00:36:56,260
This is the panacea, I guess, of all evidence based practice.

358
00:36:56,260 --> 00:37:03,190
And this is about five years ago, and this is the only bit of what people used to teach.

359
00:37:03,490 --> 00:37:08,750
Okay? He used to teach what's called probabilistic reasoning in a two by two box and it's in you.

360
00:37:09,040 --> 00:37:13,540
And it used to be if people could grind their way through this two by two box and understand this,

361
00:37:14,410 --> 00:37:19,240
then, boy, they want one of you in the room would stand up at the end and think, I am an IBM God.

362
00:37:20,290 --> 00:37:24,160
But most of us would go, Oh my God, what's this going to help us in practice?

363
00:37:24,790 --> 00:37:29,440
However, we're going to have a bit of fun for about ten, 15 minutes if you want to keep going.

364
00:37:30,190 --> 00:37:34,929
Nobody fell asleep yet. That's good. We can have ten or 15 minutes where we're going to.

365
00:37:34,930 --> 00:37:38,830
Really? I'm going to really test you. I'm going to ask you to just think.

366
00:37:39,250 --> 00:37:46,660
And if you can get these one or two principals on board, actually, you'll be far superior than most people out there in the health professional.

367
00:37:46,930 --> 00:37:52,300
But before I do that, I'm just going to give you a 20/32 break so you can get yourself ready to concentrate again,

368
00:37:52,630 --> 00:37:56,530
so you can have 30 seconds just to talk with your partner, and then I'm going to start again.

369
00:38:11,480 --> 00:38:18,920
I want you to get better now.

370
00:38:20,840 --> 00:38:29,110
I need you down in a minute. Okay.

371
00:38:29,120 --> 00:38:36,470
Good. 30 seconds gone. We've got a lot to achieve.

372
00:38:36,490 --> 00:38:40,719
Okay. You've all seen this wonderful graph and this is the graph.

373
00:38:40,720 --> 00:38:45,310
And you can you can. It's in your book. So I'm not going to go I'm just going to talk to you about a couple of bits.

374
00:38:45,880 --> 00:38:49,930
I've got this extra hour in there called Reproducible. This is not in the workbook.

375
00:38:50,830 --> 00:38:54,910
I want it in the next edition of the workbook Pull because it's a lot of fun.

376
00:38:55,150 --> 00:38:58,510
And here's something that you can all try out when you get back to your workplace.

377
00:38:59,540 --> 00:39:04,459
Okay. When we normally do this, have a lot of fun.

378
00:39:04,460 --> 00:39:10,730
When I'm in a big arena with lots of tables and I get you to all do this on each other and you can try this out when you get back.

379
00:39:10,730 --> 00:39:16,940
I get three or four students and I get you to measure one head. However, the problem is a bit confined and I can't get you all to do this.

380
00:39:17,360 --> 00:39:20,439
However I'd like. Mike, can you just come down?

381
00:39:20,440 --> 00:39:23,540
Might from it. Can you just illustrate the principle here for me?

382
00:39:24,080 --> 00:39:30,920
So what we do normally is in a big setting, in an open plan, I get three or four, even if I can imagine one head and just not tell him the result.

383
00:39:30,920 --> 00:39:34,370
So I, I might just imagine my whole life is yeah, I know.

384
00:39:34,400 --> 00:39:38,890
Just the head job.

385
00:39:39,020 --> 00:39:45,979
You take a few inches off like usual. Oh you. Oh, dear.

386
00:39:45,980 --> 00:39:49,820
That's changed. Okay. How about 59.5?

387
00:39:50,030 --> 00:39:54,260
Okay. 59.5 centre. Thanks, Mike. Give Mike a round of applause.

388
00:39:55,340 --> 00:40:02,030
Okay. Some of you may notice that coffee time I came around asking people to give me their measurement of my head.

389
00:40:02,570 --> 00:40:07,969
Okay. And I did this.

390
00:40:07,970 --> 00:40:11,330
What I called Amanda gave me this idea this morning. It was fantastic.

391
00:40:11,330 --> 00:40:15,440
One of the potters. I won't share the Potters, but I thought, okay, I coffee time.

392
00:40:15,440 --> 00:40:20,389
I'm going to do my own study. What I'm going to do is I'm going to take my tape measure.

393
00:40:20,390 --> 00:40:25,730
And that's why I started with mine. And I'm going to take the tape, measure enough people to measure my head circumference,

394
00:40:26,120 --> 00:40:30,860
and I'm going to stop doing that when somebody gets the same phone answer.

395
00:40:32,070 --> 00:40:39,720
Okay. So if you do 59.5, then did you if the next person had got 59.5, I would have stopped.

396
00:40:44,080 --> 00:40:47,500
And how many people do you think it would take before I got the same measurement?

397
00:40:50,060 --> 00:41:02,180
Within exactly the same measurement. Well, people were measured at 201.5, although doing it 12 1225.

398
00:41:02,800 --> 00:41:06,250
Right. Okay. Let's look at the look how we did.

399
00:41:06,820 --> 00:41:12,750
Okay, Mike, you're out again. Mike got 59 centimetres and I went to Etna.

400
00:41:13,240 --> 00:41:16,360
I'm not sure if I spelled that right, but I did. I did some odd.

401
00:41:16,360 --> 00:41:20,770
She went in two inches 23 and a half inches.

402
00:41:20,780 --> 00:41:24,400
So I didn't do the conversion factor there, but I was already in total.

403
00:41:25,340 --> 00:41:28,719
Okay, pull that went to 60.5 centimetres.

404
00:41:28,720 --> 00:41:34,810
So we went up. Okay. And Julie then went 63 centimetres.

405
00:41:35,800 --> 00:41:40,060
Mike had said at the beginning, didn't he? Are we doing that? He said to me, We're doing this because your head's getting bigger.

406
00:41:41,940 --> 00:41:49,269
Have to prove mine. However, Katherine said something very important to me.

407
00:41:49,270 --> 00:41:55,300
She said, Does it matter where I measure it? I'm going to let you think about that yourself.

408
00:41:55,480 --> 00:41:58,860
Actually, my head size went down at that point. 61 nine.

409
00:41:59,470 --> 00:42:04,480
Okay, then bang. He said to me, What is happening with your head?

410
00:42:07,880 --> 00:42:12,950
59.5. Does anybody know what's happening with my head?

411
00:42:14,480 --> 00:42:19,130
Because according to me, it gets bigger and smaller after a cup of coffee.

412
00:42:20,330 --> 00:42:26,930
Okay, Amanda, 61.5. Now, Amanda had seen these measurements because I said this is what I'm doing.

413
00:42:27,170 --> 00:42:30,800
So she turned it over to the inches and said, Well, I'll just do it inches and then convert it back.

414
00:42:32,360 --> 00:42:36,979
Still no help. Claire. Claire at the back there, 60 centimetres.

415
00:42:36,980 --> 00:42:41,900
So I'm up to number eight person, I think. And Jean, I'm not going to get back to my group now.

416
00:42:42,410 --> 00:42:46,600
And luckily down there came in at 59.5, which is the same as bang, bang.

417
00:42:48,320 --> 00:42:54,310
Peter. I don't want to confuse my group anymore.

418
00:42:56,470 --> 00:43:00,500
Important principle. Do you see that about measurements? Very important.

419
00:43:00,520 --> 00:43:04,570
There are some nice statements there. People are asking, do you want to send me to the trenches?

420
00:43:04,780 --> 00:43:08,410
Does it matter where I measure it? What is actually happening with my head?

421
00:43:10,460 --> 00:43:15,820
You know, they used to use a tape measure in children when they had meningitis and it got up by seven millimetres.

422
00:43:16,570 --> 00:43:22,550
Then they'd ordered an urgent three CT scan. I'm not sure what they'd order on me.

423
00:43:24,080 --> 00:43:29,660
However, a tape measure I got taught to use this in obstetric practice.

424
00:43:30,680 --> 00:43:34,569
It varies by up to five centimetre. In fact,

425
00:43:34,570 --> 00:43:40,780
a friend of mine this happened to who actually ended up with this situation and said you

426
00:43:40,780 --> 00:43:44,620
need a scan because I'm concerned that your baby is undersized by five centimetres.

427
00:43:45,490 --> 00:43:51,460
However, the scan was not available for one week. Do you think she felt bad and anxious in that period?

428
00:43:52,500 --> 00:43:58,480
You know, I think it's a it's a slight scandal that we do these things and then don't realise the consequences of what we've got to offer next.

429
00:43:58,840 --> 00:44:02,950
So if you teach people to measure and to do it this way, yes, we try and standardise it,

430
00:44:03,220 --> 00:44:06,970
but we should use these measurements to then say, what's our next action?

431
00:44:07,180 --> 00:44:11,170
How do we remember at the beginning about the cost, the problems we face?

432
00:44:11,560 --> 00:44:16,750
And so you often see this measure, which is not the measure in these tests, is about measure of agreement.

433
00:44:16,960 --> 00:44:22,660
How well do we agree? And at the bottom, you get this cap of value, bit like very good agreement.

434
00:44:22,660 --> 00:44:26,470
We can look at an x ray all the time and agree on what it failed at the top.

435
00:44:26,710 --> 00:44:33,490
0.2 we get the poor agreement and generally you look in this, but that's what you've just done.

436
00:44:33,700 --> 00:44:40,690
I think you'd agree that the nine people would probably be somewhere quite high up there in terms of agreement.

437
00:44:41,470 --> 00:44:46,720
So would we say that using a tape measure is a useful way of measuring the head circumference?

438
00:44:48,400 --> 00:44:53,170
Okay, so we wouldn't even move forward and that's an important defence.

439
00:44:53,920 --> 00:44:58,030
O. O Is the significance of the difference?

440
00:44:58,330 --> 00:45:02,380
Okay. Good point. Exactly. So in children it makes a difference.

441
00:45:02,800 --> 00:45:07,600
I think we showed my head could vary by about five centimetres. It's about the same as what happens with the following.

442
00:45:08,900 --> 00:45:14,360
And it's significant because it changes what you do next. You see that?

443
00:45:14,360 --> 00:45:20,420
And even if you same person come back to the same measurement, 2 hours later something different happened.

444
00:45:23,180 --> 00:45:26,960
So it's even standardising the person can have an effect on the agreement level.

445
00:45:28,100 --> 00:45:32,330
Okay. But I'm going to do the last thing now.

446
00:45:33,530 --> 00:45:39,170
Okay. Interpreting sensitivity and specificity. Panacea and remember is only one small strategy.

447
00:45:40,070 --> 00:45:44,780
And in fact, it's only about 5% of what we do in primary care.

448
00:45:44,990 --> 00:45:52,690
In hospital medicine, it's more. But there is this particular term sensitivity and specificity that probably,

449
00:45:52,770 --> 00:46:00,130
as everybody heard of them to anybody not heard of them since Amanda's not heard of them.

450
00:46:00,310 --> 00:46:03,910
Okay, good. Amanda, theme after class.

451
00:46:04,960 --> 00:46:09,400
Okay, good. Let's do this. We're going to do this now. I'm going to speed up a bit because we're running out of time.

452
00:46:09,550 --> 00:46:15,850
I think we're running out of time of about 5 minutes. Okay. I want you to spend you do this together with your partner on the side.

453
00:46:16,120 --> 00:46:21,309
And I want you to just think in small groups, what is the chance this patient of having this patient has?

454
00:46:21,310 --> 00:46:25,360
The disease. A disease with the prevalence of 3% must be diagnosed.

455
00:46:25,930 --> 00:46:32,890
There is a test for this disease. It has a sensitivity of 50% and a specificity of 90% given a positive test.

456
00:46:33,190 --> 00:46:37,389
What are the chances of the disease? Okay, I'll repeat that.

457
00:46:37,390 --> 00:46:43,120
While you think it through. Somebody rings you out like my dad on the end of the phone and said he's had a PSA

458
00:46:43,120 --> 00:46:49,870
test done and he's come back positive and his chances of having cancer are about 3%.

459
00:46:51,310 --> 00:46:55,840
And you look up and you see that the sensitivity of this PSA test is 50%.

460
00:46:57,700 --> 00:47:07,130
And the specificity of the test is 90%. Given a positive test, what are the chances of having that disease?

461
00:47:10,100 --> 00:47:14,290
Now. How many people would like to. How many people feel they can just help me?

462
00:47:14,480 --> 00:47:18,830
How many people feel I can really answer that question? One, two.

463
00:47:19,130 --> 00:47:22,340
Okay, good. All right. Well, won't do it then. Stop. Stop.

464
00:47:22,370 --> 00:47:30,320
Okay, relax. If only two people will move on. But generally, what happens is people generally answer about 50%.

465
00:47:31,190 --> 00:47:37,460
The high proportion of people will at once, and maybe some of you will and for never and some of you will answer all the way.

466
00:47:38,630 --> 00:47:46,100
It's quite difficult, isn't it, to do that? Very difficult to think through something about a strategy, about how you look at this problem.

467
00:47:48,300 --> 00:47:50,290
But that's the information we're asking you to look at.

468
00:47:50,310 --> 00:47:54,930
If you look in your books, it's going to tell you the sensitivity of this, this and the specificity of that.

469
00:47:56,220 --> 00:48:02,560
And no wonder everybody finds it difficult. So don't be worried, because when you do do this, the answer is 13%.

470
00:48:02,580 --> 00:48:08,130
And I'll show you doctors. With an average of 14 years expense experience answers range from 1 to 99%.

471
00:48:08,760 --> 00:48:16,020
Half of them estimating the probability of 50% is about a 1% chance of a clinical person getting it right.

472
00:48:16,260 --> 00:48:20,990
Okay. All right. Did you see that?

473
00:48:21,000 --> 00:48:24,270
I said I want to run a few tests on you just to cover my ass.

474
00:48:25,060 --> 00:48:29,220
That's the main reason people test. They don't understand. I'm going to walk you through this.

475
00:48:29,520 --> 00:48:33,479
And if you want, you can ignore me now. Or you can. Look, I have the sheet for you to take back.

476
00:48:33,480 --> 00:48:38,640
Just. You can use these spelling sheets to work through the thinking if you want to look at diagnosis.

477
00:48:38,940 --> 00:48:42,280
What? I'm going to show you how I think. Okay.

478
00:48:42,580 --> 00:48:46,930
The first thing is, what do we use a two by two table for? I don't use it for the calculations.

479
00:48:46,930 --> 00:48:50,570
I use it for thinking. I think my brain is here.

480
00:48:50,590 --> 00:48:54,910
Imagine if this is my brain. Okay. On the top is always disease.

481
00:48:55,870 --> 00:49:00,160
Okay. On the side is always test. I never move them around.

482
00:49:01,480 --> 00:49:05,709
Somebody today in our group wanted to move them around in terms of how we look at tables.

483
00:49:05,710 --> 00:49:09,130
And I was like, Don't do that to me. I've been having this way around for 14 years.

484
00:49:09,970 --> 00:49:15,670
I wanted to put the disease on the side. You're going to kill me. So this is an important thing in in your teaching, though.

485
00:49:15,700 --> 00:49:19,690
Whichever way you teach something, keep it the same amongst all of you.

486
00:49:20,350 --> 00:49:23,920
If you start to do things different ways, you'll confuse everybody.

487
00:49:24,250 --> 00:49:29,230
If we teach different ways in this group, one group will start to talk to the other group and get confused.

488
00:49:29,530 --> 00:49:33,070
But what I use this book for is actually for the definition.

489
00:49:34,180 --> 00:49:39,270
Okay. I draw a line down and I think to myself, disease.

490
00:49:40,400 --> 00:49:42,260
Going down is sensitivity.

491
00:49:42,980 --> 00:49:51,140
So when I talk about sensitivity, I'm talking about the proportion of people with the disease and I'm thinking about how many of them test positive.

492
00:49:52,070 --> 00:49:59,360
Do you see that? However, with the difficult thing about what I ask you is I asked you about what's called the positive predictive value.

493
00:49:59,840 --> 00:50:03,500
What happens if you have a positive test and you're all lost?

494
00:50:04,460 --> 00:50:10,310
But what I do is to remember the definition of sensitivity. I throw that and I just said, I gave to you.

495
00:50:10,460 --> 00:50:19,910
If you've got the disease and you test positive sensitivity, that the first piece of information is to use that sensitivity.

496
00:50:20,570 --> 00:50:24,890
Now, my group will know this because we spend about 2 hours on what I call natural frequency.

497
00:50:25,190 --> 00:50:33,860
I think in terms of treatment effects, in terms of diagnosis, in terms of 100 people, nothing else apart from sometimes in a thousand.

498
00:50:34,520 --> 00:50:39,740
But I think often the first bit of information I gave you there was a prevalence of 3% in every patient.

499
00:50:41,060 --> 00:50:44,930
And I split that into actually. What does that mean in terms of 100 patients?

500
00:50:45,980 --> 00:50:49,100
Three people had the disease. 97 did not.

501
00:50:50,570 --> 00:50:54,590
Okay. Everybody see that? Happy with that?

502
00:50:54,740 --> 00:50:58,150
Anybody not see that? Okay.

503
00:50:58,450 --> 00:51:03,550
Then I gave you a piece of information. I said the sensitivity was 50%.

504
00:51:05,690 --> 00:51:10,010
Which group of these people have the belief? The top bit.

505
00:51:10,940 --> 00:51:18,550
How many of them? Top people who test positive? One and a half.

506
00:51:20,210 --> 00:51:26,270
You see that? I'll go back. Three people had the disease.

507
00:51:26,600 --> 00:51:30,800
Sensitivity, 50%. Remember them going down the column. People who have the disease.

508
00:51:31,100 --> 00:51:35,480
The proportion of test positive was 50%. That's one and a half people.

509
00:51:36,470 --> 00:51:40,610
Now, in effect, I could stop there because that's probably more information than anybody ever needs,

510
00:51:41,450 --> 00:51:46,100
because now you can realise if you have a positive test, some people still will be left behind with the disease.

511
00:51:47,840 --> 00:51:55,880
And some people will put the important feeling is to get a feeling of how these tests work in terms of of the prevalence of the disease.

512
00:51:58,790 --> 00:52:02,510
Okay. We've still got these 97 people without the disease.

513
00:52:03,290 --> 00:52:08,720
Okay. Now, this is the real trick. We've actually.

514
00:52:10,390 --> 00:52:20,650
Set back diagnostic thinking in probabilistic reasoning many years by giving people the specificity with the sensitivity because the specificity,

515
00:52:20,650 --> 00:52:24,790
if you want, you can think of the arrow and m got the arrow. It is going, Oh.

516
00:52:27,430 --> 00:52:33,440
The proportion of people without the disease who test negative. But what I'm interested in is this box.

517
00:52:35,010 --> 00:52:40,680
You see that? Because I wanted to go across. Now this is on.

518
00:52:40,680 --> 00:52:47,910
I can't do this. It seems to be we've got this very difficult thinking that because we've got these two boxes here,

519
00:52:48,330 --> 00:52:52,680
we can't mentally adjust and come up with a figure for that predictive value.

520
00:52:54,000 --> 00:52:57,180
And so what's gone wrong in our wire is too difficult for us to understand.

521
00:52:58,230 --> 00:53:01,770
However, what I do is I use what's called the false positive rate.

522
00:53:02,280 --> 00:53:07,950
And if I go back, what I do is I do one minus the specificity immediately in my mind.

523
00:53:09,700 --> 00:53:13,960
Or 100%, minus the 90%. I gave you a specificity of 90%.

524
00:53:14,500 --> 00:53:18,010
You see that? And so I am interested in the false positive rate.

525
00:53:21,030 --> 00:53:26,110
I'm interested in this box. The people who test positive but don't have the disease.

526
00:53:26,850 --> 00:53:30,840
That is not the specificity either. I'll repeat that.

527
00:53:31,020 --> 00:53:36,360
The people who test positive do not have the disease. That's what I'm interested and more interested in a positive result.

528
00:53:38,190 --> 00:53:42,719
Now, when that comes out, that was 10% of 97. That's pretty easy math for me to do.

529
00:53:42,720 --> 00:53:46,320
That's about ten people I do. I don't want to be exact. 9.7 is meaningless.

530
00:53:47,130 --> 00:53:54,090
Okay. But what I've done is I've assembled all the people who have a positive test now.

531
00:54:01,170 --> 00:54:08,070
And of them, 11 and a half, only one and a half, one and a half will have the disease that with the 15% chance.

532
00:54:13,090 --> 00:54:23,250
And I repeat this point again. Well, all these patients now on this right hand side are the people who've tested positive.

533
00:54:24,920 --> 00:54:30,020
But it is important to recognise that the majority of people who tested positive are these who didn't have the disease.

534
00:54:31,170 --> 00:54:39,680
That's the one minus the specificity. And I think the terms of the positive tests, the specificity is of no benefit to you.

535
00:54:42,220 --> 00:54:46,810
The smart people here will start to realise I'm using what's called the likelihood ratio, but I won't go there.

536
00:54:47,380 --> 00:54:53,140
But what I will do is that's how I describe the result to my patient back in the natural frequency of 100.

537
00:54:53,530 --> 00:54:55,659
Both of you want to have a go at this in your group.

538
00:54:55,660 --> 00:55:00,580
I have these little handout sheets where you can just fill the book in, but I'll show you what it feels like.

539
00:55:01,270 --> 00:55:04,509
Here's how you do it quick. You can do this and I can do this quick in my mind now.

540
00:55:04,510 --> 00:55:10,270
But the most important is being in the ballpark. And I don't care if you get 15% or 10%.

541
00:55:10,480 --> 00:55:17,350
What worries me is if the first three people would have got an answer, like 70% or always I start to worry.

542
00:55:18,610 --> 00:55:22,870
I've changed the prevalence of the disease to 30% and the same test.

543
00:55:23,320 --> 00:55:27,490
You see that same test prevalence is 30%.

544
00:55:27,880 --> 00:55:33,460
This is like going from a GP setting with a probability of 3% to now in hospital.

545
00:55:33,970 --> 00:55:36,700
The prevalence has gone up or the pre-test probably has gone up.

546
00:55:38,560 --> 00:55:44,470
Now our previous competitor, in terms of a hundred patients, how many people have the disease?

547
00:55:45,790 --> 00:55:49,230
There's 30. Okay.

548
00:55:50,540 --> 00:55:58,970
Pretty happy with that. How many do not have the disease centre OC sensitivity of 50%.

549
00:56:01,070 --> 00:56:04,340
Three 5050 okay.

550
00:56:05,060 --> 00:56:09,680
Forget the specificity. You want the hundred percent minus the 90, which is 10%, don't you?

551
00:56:10,460 --> 00:56:16,670
The false positive rate. You're interested in the proportion of people who don't have the disease who test positive.

552
00:56:17,330 --> 00:56:24,060
How many people with a. 10% of. Seven and seven.

553
00:56:24,750 --> 00:56:27,910
Okay, now I've just assembled you very quickly.

554
00:56:27,940 --> 00:56:31,299
All the people have a positive test. 22 positive testing.

555
00:56:31,300 --> 00:56:38,860
A total of 15 have the disease. The chances of you now have the disease of 70% have gone from 15 to 70%.

556
00:56:45,390 --> 00:56:51,090
That's much more informative in terms of disease, in terms of thinking, in terms of what you're going to do next.

557
00:56:52,320 --> 00:56:58,650
You didn't have to get it right, but you did have to be able to think mentally that you can get in the ballpark of where you're going.

558
00:56:59,790 --> 00:57:01,290
I'm pleased that some of you got there.

559
00:57:01,290 --> 00:57:06,269
And if you want to go back to your groups, when you look at this piece in your family, you might not do diagnostics from here on.

560
00:57:06,270 --> 00:57:10,200
That fine is quite challenging. I find it a lot of fun.

561
00:57:11,510 --> 00:57:16,550
I think we have to think about how we change, how we view things. However, I'm going to finish now.

562
00:57:17,240 --> 00:57:21,320
But the important thing is to recognise. I like this paper. This is a Dutch paper.

563
00:57:22,190 --> 00:57:26,060
When they looked at children with serious, serious infection, very low rate,

564
00:57:26,570 --> 00:57:31,160
about one less than 1% of children had serious infection from all like my daughter.

565
00:57:31,160 --> 00:57:38,240
Remember I told you at the beginning about this? Well, high temperature, less than full accuracy of signs and symptoms was fairly low.

566
00:57:38,810 --> 00:57:42,650
The sign, Paramount in all trees was the position statement.

567
00:57:42,650 --> 00:57:48,100
Something is wrong. Just like we sat there going, I think there's something wrong.

568
00:57:48,370 --> 00:57:51,190
So probabilistic reasoning is only one part of a strategy.

569
00:57:51,520 --> 00:57:57,150
We've put this together and I haven't given you the whole bit, but here's all the strategies and the people who've been involved at the top.

570
00:57:57,160 --> 00:58:00,549
There are people, some people who like Paul and Dan and Rafael,

571
00:58:00,550 --> 00:58:04,880
and we've been collecting patients and this is going to be in the BMJ, which we're very pleased about.

572
00:58:05,110 --> 00:58:10,270
We've actually got this to a point where we're going to start explaining this in a whole series of articles.

573
00:58:10,840 --> 00:58:16,900
So watch out. It's about 4 to 6 weeks away from going into the BMJ, and then we're going to take each one and say,

574
00:58:17,050 --> 00:58:21,100
once we describe this, we're going to take the spot diagnosis and illustrate it more.

575
00:58:21,850 --> 00:58:26,230
So once that's come out, I'll put this talk up, put this talk up until that's come out.

576
00:58:26,650 --> 00:58:31,040
But that'll be about six weeks ago. Okay. And I particularly like this.

577
00:58:31,040 --> 00:58:32,780
I'm going to finish on this by Jeff Norman.

578
00:58:33,320 --> 00:58:38,690
Jeff Norman's a very willed, eminent person on education and diagnostic reason, and I've been looking it for many years.

579
00:58:39,770 --> 00:58:44,140
Diagnosis is not a matter of acquiring some kind of all inclusive reasoning strategy.

580
00:58:44,540 --> 00:58:47,990
To some extent, several strategies can lead to the same diagnosis.

581
00:58:49,700 --> 00:58:54,890
The recognition of the strategy should encourage the use of experience to guide our search for the correct diagnosis.

582
00:58:55,730 --> 00:59:02,150
So there are more ways of getting the count. And I hope that sort of makes you feel that some of the bits you can be comfortable with and say,

583
00:59:02,330 --> 00:59:06,750
Maybe I'll just improve the way we do spot diagnosis. Maybe I'll look at prediction.

584
00:59:07,520 --> 00:59:11,520
Maybe I'll forget the probabilistic reasoning. Like that.

585
00:59:11,600 --> 00:59:12,500
I'll stop. Thank you.