1
00:00:00,420 --> 00:00:04,620
Everybody I think I to fly was just drafted by the 40 seconds as well.

2
00:00:05,160 --> 00:00:08,850
So I think we shall make a stop next to the mat.

3
00:00:08,850 --> 00:00:15,210
For those of you who have not met me yet, and I'm currently the course director of the Los Angeles based health care.

4
00:00:15,540 --> 00:00:18,840
And as part of that, this week, we're running for some diagnosis and screening.

5
00:00:19,260 --> 00:00:27,329
And for that, I have asked my very difficult to come along and talk to us about the work they've been doing, talking to industry,

6
00:00:27,330 --> 00:00:35,430
thinking about how to perhaps implement diagnostic tests or strategies and just come up with this very great and interesting title on Beyond Accuracy,

7
00:00:35,820 --> 00:00:40,290
Evidence Gaps and Unintended Consequences. So just one quick note.

8
00:00:40,860 --> 00:00:44,430
I had somebody who left a coat in the other room that we were working in.

9
00:00:45,270 --> 00:00:49,740
Stephanie What have you guys have a chat with Rob and afterwards she can let you have access to that room again.

10
00:00:49,830 --> 00:00:55,850
All right. So, Phil, thank you very much. Thanks very much. Thanks, Annette, for inviting me to come and speak to the group.

11
00:00:55,860 --> 00:01:03,839
So it's actually about just over four years ago that I was sitting where a lot of you were actually, I sort of when I first started in the department,

12
00:01:03,840 --> 00:01:10,800
I took the course as sort of a bit of an introduction to sort of diagnostics in primary care, essentially, to give me a bit of context.

13
00:01:11,460 --> 00:01:16,410
And so there's been quite a lot of water under the bridge since then. And so when asked me to give this talk,

14
00:01:16,890 --> 00:01:22,440
I was quite interested to think about things I've learned in the process of working with industry for the last few years.

15
00:01:22,440 --> 00:01:30,450
And so I came up with this idea of thinking about beyond accuracy, because one of the things you probably learned later in the talk, hopefully,

16
00:01:30,450 --> 00:01:33,510
is that what we find is there's a very big,

17
00:01:34,050 --> 00:01:41,070
big emphasis by the diagnostics industry and sort of innovators developing diagnostic tests on very, very simple metrics.

18
00:01:41,070 --> 00:01:44,280
But they don't always they're only a very, very small part of the story.

19
00:01:44,310 --> 00:01:44,790
Okay.

20
00:01:45,480 --> 00:01:52,620
I think probably another title I could have given it was also sort of observations of a gatekeeper, because that's also been part of my role really,

21
00:01:52,620 --> 00:01:58,590
is that we've as we've been sort of approached by diagnostic companies to come and talk with our organisation,

22
00:01:58,860 --> 00:02:03,870
I've been the person who's generally had to triage out a lot of the companies to make sure that the

23
00:02:03,870 --> 00:02:07,920
people we actually talk to are probably the ones who are developing things which are realistic.

24
00:02:08,850 --> 00:02:14,700
Okay, so I just have to give a declaration because we're funded by the nature and that's

25
00:02:14,700 --> 00:02:18,569
basically that these observations that I'm going to present today on my own,

26
00:02:18,570 --> 00:02:23,070
they're not necessarily those of the NHS, the nature and nature of the Department of Health.

27
00:02:23,430 --> 00:02:30,479
Okay. So, so if I read about certain things and it's all so my own responsibility, okay,

28
00:02:30,480 --> 00:02:35,639
so I'm just going to talk a little bit about the origin of the nature diagnostic infrastructure.

29
00:02:35,640 --> 00:02:42,660
So a number of years ago the notion I think recognised that it was a bit of problem getting technologies,

30
00:02:43,140 --> 00:02:46,980
diagnostic technologies and other technologies into into the health care system.

31
00:02:47,580 --> 00:02:53,730
And really it boils down to a couple of things and a couple of very general barriers and the both around evidence really.

32
00:02:54,000 --> 00:03:00,809
One is that generally a lot of diagnostics and technologies fail because there's insufficient evidence of efficacy,

33
00:03:00,810 --> 00:03:03,810
acceptability and cost effectiveness to persuade stakeholders.

34
00:03:04,380 --> 00:03:10,530
So we're moving beyond that sort of the simple metrics again there and that really we get this very

35
00:03:10,530 --> 00:03:16,829
poor uptake of these technologies when there is a shortfall between the capability and expectations.

36
00:03:16,830 --> 00:03:23,160
So this what we call the technology needs mismatch. So somebody has an idea of what they want to develop as a diagnostic.

37
00:03:23,400 --> 00:03:29,010
It doesn't necessarily meet with what the clinicians expectations and that might be the end user might be at the end of the day.

38
00:03:30,600 --> 00:03:36,810
So the notion of funded originally in 2013 four of what they call the diagnostic evidence cooperatives.

39
00:03:37,320 --> 00:03:45,120
So we were one in Oxford, there's another one up in Leeds and they're in Newcastle upon Tyne and the other one at Imperial College London.

40
00:03:45,120 --> 00:03:50,609
And so we all had slightly different remits. So we're based in the Department of Primary Care.

41
00:03:50,610 --> 00:03:54,900
So our focus is really on primary care, suitable public health, suitable diagnostics,

42
00:03:56,490 --> 00:04:00,300
and that generally restricts kind of what we do down to the point of care tests.

43
00:04:00,600 --> 00:04:07,560
So these are sort of near patient tests that can actually be used by the patient within a surgery or at the patient's bedside.

44
00:04:08,220 --> 00:04:13,200
So taking it further, the Department of Health obviously thought that this was a relatively good thing.

45
00:04:13,200 --> 00:04:16,169
There was quite a lot of good feedback from the diagnostics industry, actually.

46
00:04:16,170 --> 00:04:21,239
So we work quite widely with a lot of companies trying to help with the development of evidence.

47
00:04:21,240 --> 00:04:25,559
So it was funded again. So we bid the other year and we were refunded.

48
00:04:25,560 --> 00:04:31,080
So they changed the landscape a little bit and they changed also the remit really of the.

49
00:04:31,440 --> 00:04:33,419
So originally we had what we were called decs,

50
00:04:33,420 --> 00:04:38,730
the diagnostic evidence cooperatives and also the health technology cooperatives had slightly different foci.

51
00:04:38,970 --> 00:04:46,260
They loved them all together, refunded. And now we have what they call these mix of them and h.R medtech and IVD cooperatives.

52
00:04:46,920 --> 00:04:56,459
So we're this one down here. So the Oxford Nature Community Health Care MedTech and IVD Co-operative, so a bit of a mouthful.

53
00:04:56,460 --> 00:04:59,660
But again, our sort of remit is focusing on diagnostics.

54
00:04:59,790 --> 00:05:02,910
Technologies that could be used in the in the community.

55
00:05:03,420 --> 00:05:07,049
One of the big things that changed between the two funding schemes was that the first

56
00:05:07,050 --> 00:05:12,330
funding scheme had a focus on us working with HIV AIDS that were commercially available.

57
00:05:12,330 --> 00:05:16,680
So that's basically things that could be bought by a hospital or a clinic and used.

58
00:05:17,040 --> 00:05:18,779
So things already had a CE mark.

59
00:05:18,780 --> 00:05:26,669
So the approval locally, we always found that a bit restrictive because if you think about it often by the stage that it's achieved,

60
00:05:26,670 --> 00:05:33,660
regulatory approval is too late to shape that diagnostic and change how it might work, how it might look, how it might fit into the system.

61
00:05:34,230 --> 00:05:37,830
So we found that to be quite frustrating has quite big knock on effects.

62
00:05:38,220 --> 00:05:44,040
But actually what's happened now is we have a bit more of a flexible remit so we can work with companies in the sort of concept

63
00:05:44,040 --> 00:05:49,800
stage and work with them right from the base that was to hopefully generate things which are suitable for the community.

64
00:05:51,330 --> 00:05:56,430
So the community so we're a sort of remit really is about guiding management decisions in the community.

65
00:05:56,430 --> 00:06:03,660
So we're thinking here about, you know, out-of-hours here, this doctor up at the top left in your standard GP surgery,

66
00:06:04,440 --> 00:06:07,740
managing sort of things within the community like for example,

67
00:06:08,190 --> 00:06:15,389
chronic wounds, ambulatory monitoring of blood pressure could be one example and we're thinking about these particular tests here.

68
00:06:15,390 --> 00:06:19,920
So as I mentioned before, the kind of point of care tests that can be done very rapidly by a patient

69
00:06:20,760 --> 00:06:24,959
thinking about relatively often very relatively simple samples like fingerprint,

70
00:06:24,960 --> 00:06:30,540
blood, blood and the kind of sort of assistance that we've offered as a as a deck.

71
00:06:30,540 --> 00:06:38,249
And also moving on with the mic, is that sort of the sort of clinical input very early on, health economics input, statistical input.

72
00:06:38,250 --> 00:06:42,390
And also importantly, that's not mentioned here on this slide is the qualitative input.

73
00:06:42,780 --> 00:06:49,770
Okay. So exploring clinician and patient, you know, feelings about diagnostic testing and the knock on effects potentially.

74
00:06:51,360 --> 00:06:54,900
So some of the arguments for point of care testing, so these near patient tests,

75
00:06:55,110 --> 00:06:59,100
often we hear this from from industry which is faster is definitely better.

76
00:06:59,520 --> 00:07:05,880
So if you do it fast at the side of the patient, we can make a really rapid clinical decision and change treatment.

77
00:07:05,940 --> 00:07:09,989
Okay, so I've put a big question mark there because I don't think that's always a positive thing.

78
00:07:09,990 --> 00:07:13,319
We'll discuss that a bit later. Could be convenient for patients.

79
00:07:13,320 --> 00:07:18,490
So you go in to see a GP and rather than than taking a sample, send it to the lab, you know,

80
00:07:18,510 --> 00:07:25,500
and waiting for a few days having to call the surgery, get past the receptionist and get a result, you actually get the result than in there.

81
00:07:25,800 --> 00:07:29,760
Okay. So that also what follows from that is that the patient will get the result.

82
00:07:30,240 --> 00:07:36,480
Okay. So if you're on the receiving end of a point of care test, hopefully you're going to receive obviously the right test result at the time.

83
00:07:37,290 --> 00:07:41,310
And I think an important context of the bottom is where there's no laboratory.

84
00:07:41,700 --> 00:07:48,450
So there are contexts, particularly in public health, thinking about the developing world where there isn't any laboratory infrastructure.

85
00:07:48,450 --> 00:07:57,120
So there's nothing to support complex machinery that requires a lot of, you know, input from, you know, a strong support lab and strong support group.

86
00:07:57,570 --> 00:08:01,680
So that's where you perhaps thinking about point of care or diagnostic services,

87
00:08:01,680 --> 00:08:06,569
which are standalone and can support themselves, even things that support themselves on battery power.

88
00:08:06,570 --> 00:08:10,530
We've seen various examples of that during the last couple of years.

89
00:08:12,330 --> 00:08:16,560
So there are some big challenges with the community health care setting.

90
00:08:17,010 --> 00:08:19,050
Probably the first one is patient population.

91
00:08:19,830 --> 00:08:29,430
So if you think about your GP people working in the community context, they're exposed to really the full spectrum of the patient population,

92
00:08:29,430 --> 00:08:33,210
sort of right from the very young to the elderly, and that brings particular challenges.

93
00:08:34,140 --> 00:08:40,500
So you've got this full breadth compliance and to different sampling methods will vary very much so.

94
00:08:40,530 --> 00:08:44,220
You know, obviously it's maybe quite easy to take a venous sample from the lady in the middle,

95
00:08:44,610 --> 00:08:49,169
but you're not going to be able to do that necessarily so easily from the baby or from the elderly person.

96
00:08:49,170 --> 00:08:54,329
Okay. And also, diagnostic performance can differ between patient groups.

97
00:08:54,330 --> 00:08:59,130
So what works for a baby won't necessarily work in an elderly person or have the same thresholds.

98
00:08:59,250 --> 00:09:04,250
Okay. Oh, right.

99
00:09:04,260 --> 00:09:08,790
Sorry. Back again. The other challenge really is about prevalence.

100
00:09:08,910 --> 00:09:15,719
So generally if we think about the GP population, so the patients that will come to the GP, you've got everybody coming through the door.

101
00:09:15,720 --> 00:09:23,790
So it's a very big mixed population and maybe sort of the few people that have a very serious illness, they're like the needles in the haystack.

102
00:09:24,390 --> 00:09:30,180
Whereas often when we think about the context of the acute hospital, this is a population that's usually refined.

103
00:09:30,180 --> 00:09:34,800
So often these are people who've been either brought in by an ambulance, they've been referred in by their GP.

104
00:09:35,220 --> 00:09:38,790
So our population is a more refined population that we're dealing with.

105
00:09:39,150 --> 00:09:46,710
So the diagnostics for the community setting often have to be extremely strong in terms of the accuracy performance that they can deliver,

106
00:09:47,460 --> 00:09:55,530
okay, if they're to be of any use. And I think the final thing here just in terms of challenges are also people.

107
00:09:55,530 --> 00:09:58,710
So the the challenges of the of the clinicians.

108
00:09:59,600 --> 00:10:05,270
And also the patients, what's acceptable for them. So this is a bit of work we did published in 2016.

109
00:10:05,270 --> 00:10:11,059
So we did a we did a big GP surgery survey in Britain of about just over 1000 GPS and we

110
00:10:11,060 --> 00:10:16,040
collected some information on what GP's sort of impressions were of point of care testing.

111
00:10:16,700 --> 00:10:23,200
And you can sort of see on the sort of left hand column here is about facilitators, you know,

112
00:10:23,220 --> 00:10:29,959
you see the details and the right hand column is about the barriers and I think it's quite evident that there's

113
00:10:29,960 --> 00:10:35,000
a lot of perceived barriers to point of care testing from the GP population compared with the facilitators.

114
00:10:35,000 --> 00:10:40,850
So some of the facilitators are very, very positive of course, and we've got things like improve job satisfaction.

115
00:10:40,850 --> 00:10:43,999
You might not think of that necessarily, maybe because you can give, you know,

116
00:10:44,000 --> 00:10:48,680
you can sort of be holistic in terms of your consultation, reduction of referrals.

117
00:10:48,680 --> 00:10:53,149
So you're targeting people you send up to hospital more effectively and also things like this.

118
00:10:53,150 --> 00:10:57,780
So remote practices maybe can improve care. So, you know, we have that within Britain, don't we?

119
00:10:57,780 --> 00:11:01,040
If you think about the islands and the highlands of Scotland, for example,

120
00:11:01,310 --> 00:11:05,299
there are people who operate in general practice in them very many miles from the acute hospital.

121
00:11:05,300 --> 00:11:10,850
So particular settings where point of care can be positive, but also a lot of barriers.

122
00:11:10,850 --> 00:11:14,750
I mean things like, you know, I mentioned before about the time, you know,

123
00:11:15,080 --> 00:11:20,239
one of the comments we had was that sort of obviously point of care tests eliminates the time for watchful waiting.

124
00:11:20,240 --> 00:11:25,219
So actually, if you given, you know, you do a point of care test, you have a result, you know,

125
00:11:25,220 --> 00:11:30,709
almost compelled to make a decision right then and there in front of your patient that might not necessarily be desirable for you.

126
00:11:30,710 --> 00:11:36,530
You might want to sit and think about it first and also the whole sort of raft of things around clinical governance

127
00:11:36,530 --> 00:11:43,190
and the legal sort of implications of performing point of care tests in general practice in the community.

128
00:11:44,990 --> 00:11:54,950
So I mentioned earlier about the emphasis for test developers and most, if you look at the literature is provided by most point of care manufacturers.

129
00:11:55,340 --> 00:12:00,739
The sort of the metrics you see in general are around this agreement and accuracy.

130
00:12:00,740 --> 00:12:07,340
So you probably recognise this people in here, you know, a bland Altman pop, this is a particularly highly saturated one.

131
00:12:07,340 --> 00:12:13,040
So this is actually creating in point of care test results that were performed at point of care.

132
00:12:13,490 --> 00:12:18,830
So this is where what the people here have done is actually they've compared the

133
00:12:18,830 --> 00:12:22,850
point of care test results with the gold standard of the reference standard test.

134
00:12:22,850 --> 00:12:27,469
So it's a laboratory analyser test. So these are actually data that came out of the community.

135
00:12:27,470 --> 00:12:30,020
So the the these are paired blood, blood samples.

136
00:12:30,020 --> 00:12:35,240
Some of the tests that the point of care by clinicians, the others were sent up to to the lab at the John Radcliffe Hospital.

137
00:12:35,690 --> 00:12:41,860
And we did a bit of a comparison. So you often see a lot of this data presented by point of care test manufacturers.

138
00:12:41,870 --> 00:12:45,349
That's the agreement. And then second thing here is about the accuracy.

139
00:12:45,350 --> 00:12:54,560
So how good does yeah. Is the test at giving a diagnosis or particular or actually picking out particular threshold?

140
00:12:55,160 --> 00:12:58,670
Okay. Sort of for a particular disease, state or condition.

141
00:12:59,360 --> 00:13:09,379
So you see a lot of that. And we also see that along with that data, we get a lot of people who who sort of we get the pragmatists,

142
00:13:09,380 --> 00:13:16,070
people who think that obviously understand that this isn't sufficient evidence for the implementation of a test.

143
00:13:16,550 --> 00:13:22,910
We get others who are very, very resistant to that idea and partly sometimes because of this sort of parachute analogy,

144
00:13:22,910 --> 00:13:29,690
which is that they believe that the the diagnostic they've developed is absolutely so effective that it's like,

145
00:13:29,810 --> 00:13:36,830
you know, essentially it's going to be such so, so good for patient benefit that is comparable with the parachute.

146
00:13:36,830 --> 00:13:41,389
So this is a paper that was published by Pelin Smith in 2003.

147
00:13:41,390 --> 00:13:47,030
So it was one of the in one of the Christmas issues of the BMJ, so slightly tongue in cheek.

148
00:13:47,430 --> 00:13:52,249
And what they argued was that there are some medical interventions which are so clear cut from the outset,

149
00:13:52,250 --> 00:13:55,640
they're going to be definitely positive for the patient.

150
00:13:55,910 --> 00:13:57,739
Okay. There's no point in doing it.

151
00:13:57,740 --> 00:14:02,930
Asked a randomised controlled trial to actually try and elucidate whether there's a positive patient impact or not.

152
00:14:03,620 --> 00:14:12,160
And they're sort of basically they're sort of summary at the end of the paper was that all those people who are strong proponents of asked

153
00:14:12,320 --> 00:14:20,450
you should enter themselves into an act where they which is basically a randomised placebo placebo controlled trial of the parachute.

154
00:14:21,020 --> 00:14:24,110
Okay. So obviously not very many people are going to do that.

155
00:14:24,860 --> 00:14:31,790
So we get a few, we get a few we've had a few companies over time who really think they're onto something like this,

156
00:14:31,790 --> 00:14:38,450
but they don't necessarily think about the kind of more nuanced impacts of introducing tests into a, into a patient pathway.

157
00:14:39,890 --> 00:14:48,350
So just to give a bit of an overview, so since 2013, we've had kind of a rather large number of consultations with the diagnostics industry.

158
00:14:48,350 --> 00:14:53,210
So these have been various point of care tests, people that have come to us to sort of seek a bit of advice.

159
00:14:53,630 --> 00:14:56,900
And that's gone right from infection right through to SGI. Okay.

160
00:14:56,900 --> 00:15:05,390
So and the sort of point of note. Here is that very few of them we thought actually would have the suitability.

161
00:15:05,400 --> 00:15:10,650
So we've done a of when we scrutinise, we thought about what the test might offer in a patient pathway.

162
00:15:11,010 --> 00:15:13,680
There are very few that we've actually thought would have a positive impact.

163
00:15:14,120 --> 00:15:18,300
Okay, so we've worked with a few and some of them have gone on and potentially could be very, very positive.

164
00:15:18,660 --> 00:15:22,290
But the vast majority we think are have some sort of shortfall.

165
00:15:22,290 --> 00:15:23,430
And partly that, as I say,

166
00:15:23,620 --> 00:15:28,950
is due to the fact that we were seeing people at the point where they've already got a test that's been that's been approved.

167
00:15:29,610 --> 00:15:32,819
So in current practice, they're very minimal point of care tests available.

168
00:15:32,820 --> 00:15:36,980
So there's been very poor dissemination and adoption, adoption in primary care.

169
00:15:36,990 --> 00:15:42,149
So hence the sort of doctor's bag hasn't really changed a whole lot in probably 50 years.

170
00:15:42,150 --> 00:15:45,990
So we still see the stethoscope. Obviously, we've got urine, lipsticks.

171
00:15:46,440 --> 00:15:50,120
A few people these days carry around a pulse oximeter, but there's not a lot else there.

172
00:15:50,220 --> 00:15:56,100
There's not a lot in the primary care doctors armoury to use in terms of diagnosis.

173
00:15:56,910 --> 00:16:04,320
And I think that's partly because new diagnostic tests can be very, very disruptive to patient pathways and not always in a predictable manner.

174
00:16:04,680 --> 00:16:08,579
Okay. So that's we're going to talk about following on from this. Okay.

175
00:16:08,580 --> 00:16:15,300
So a bit of a challenge for you. Right. So just to sort of think a little bit about this before we move on.

176
00:16:16,050 --> 00:16:20,670
So a bit of a case study. So this is entirely this is entirely fictional.

177
00:16:20,670 --> 00:16:25,910
There's lot of truth in it as well. Okay. So we think about diarrhoeal disease in rural South Africa.

178
00:16:25,920 --> 00:16:33,540
So I've got my wife is South African and she's got an uncle and a cousin who are both up in the north of South Africa.

179
00:16:33,550 --> 00:16:36,810
So this is they deal with very often. Okay.

180
00:16:36,810 --> 00:16:42,740
So we're thinking about a rural GP in the B district which has a very high caseload of diarrhoeal disease.

181
00:16:42,750 --> 00:16:47,250
So the water supply there is a bit unreliable. So transmission is very, very high.

182
00:16:47,490 --> 00:16:50,490
There's also people who use a lot of informal sources of water.

183
00:16:50,490 --> 00:16:55,290
So, you know, using streams as a source of water for animals, etc., at times.

184
00:16:56,550 --> 00:17:04,040
So the disease that people pick up there is an attribute attributable to a huge array of bacteria and parasites.

185
00:17:04,050 --> 00:17:07,200
So there's a whole bucket load of them that you can pick up.

186
00:17:07,530 --> 00:17:09,479
And actually, last time I was there in South Africa,

187
00:17:09,480 --> 00:17:14,730
my son was unfortunate enough to pick up one of these as well, and we had to deal with the consequences.

188
00:17:15,660 --> 00:17:22,830
So in this scenario, local parcel services turn around comprehensive results for stool samples in about 24 to 48 hours.

189
00:17:23,550 --> 00:17:34,379
At about 80 runs for a sample. Okay. And this guy is being approached by the manufacturer of a new diagnostic test, which is for giardia, intestinal.

190
00:17:34,380 --> 00:17:39,090
This, which is one of the common parasites. There is a very common parasite in the water in this region.

191
00:17:39,360 --> 00:17:45,240
It's very, very accurate. This test results in about 10 minutes and it's about 40 runs for a test.

192
00:17:46,140 --> 00:17:49,170
So the question is, what is the likely impact of this test?

193
00:17:49,170 --> 00:17:52,440
Does anybody have a feeling any comments at all?

194
00:17:55,490 --> 00:18:06,380
There's no right or wrong answer, really. So while lots of possible right answers, I think for your ends are still a lot of money in that area.

195
00:18:07,550 --> 00:18:12,800
So of the people probably needed it would have fought year end.

196
00:18:13,340 --> 00:18:17,020
Yes. So. Well, I'm thinking more so that's a possibility.

197
00:18:17,030 --> 00:18:23,660
I'm thinking more of so you know, say they've got private health insurance over say the money to the actual patients, not the issue.

198
00:18:25,550 --> 00:18:29,570
What's the boss? There's a blindingly big problem with this.

199
00:18:29,910 --> 00:18:33,250
So they get medications like three, three, four that they have. Yeah.

200
00:18:33,260 --> 00:18:36,470
So that's a good point. So there is medication for this metronidazole.

201
00:18:36,980 --> 00:18:40,640
You can treat other names like ill. So that's not a problem.

202
00:18:40,820 --> 00:18:45,570
Yep. You just. They're not a company that's selling the initial diagnostic test.

203
00:18:45,570 --> 00:18:49,340
He doesn't want this competitor to get into the market. No.

204
00:18:49,340 --> 00:18:53,870
I mean, the thing is, the laboratory test is sort of a is not a point of care test.

205
00:18:53,870 --> 00:18:59,660
So that's not something you could use in the doctor's surgery yet. Mostly the fact that the order is common.

206
00:18:59,840 --> 00:19:03,030
So if you have diarrhoea. Yep. You have a doctor.

207
00:19:03,290 --> 00:19:07,130
See, is likely to possibly be giardia.

208
00:19:07,610 --> 00:19:10,730
But so no way to treat it thinking is that.

209
00:19:11,030 --> 00:19:16,250
Well if you do test for that it who's Ed. In which case that's 4041 which is fine.

210
00:19:16,610 --> 00:19:20,390
Or it moves out, in which case it's still going to have to include 80 round.

211
00:19:20,930 --> 00:19:27,709
Exactly. Okay. So there's all as I mentioned, there's a whole load of parasites and bacteria, bacterial species that can cause there.

212
00:19:27,710 --> 00:19:34,850
So you're only ruling out one thing by doing this test. It's the high likelihood, as you've pointed out, is that you're going to do this test.

213
00:19:35,210 --> 00:19:38,270
It becomes negative. You're going to have to do the other test as well.

214
00:19:38,870 --> 00:19:45,019
Okay. I think the other thing as well, from a general practice perspective is this number here and now in Britain.

215
00:19:45,020 --> 00:19:48,889
That would be a massive problem, I think, in the within the private health care system in South Africa.

216
00:19:48,890 --> 00:19:51,950
It's not such an issue. Yeah, I think it's common.

217
00:19:51,950 --> 00:19:58,070
20 minute appointments are common, but in the public health system there is very similar to Britain and people are very time limited.

218
00:19:58,490 --> 00:20:02,780
Okay. So it's you know, this the ten minute turnaround time is not particularly good.

219
00:20:03,410 --> 00:20:10,820
Okay. So we're just move on now. So how do we decide actually if a diagnostic technology is any good and what evidence is required?

220
00:20:11,450 --> 00:20:24,170
It's the next question. Okay. So back in 2014, Andrea Horvath, they proposed this cycle for evidence that you should try and collect for a diagnostic.

221
00:20:24,560 --> 00:20:31,490
Okay. This sort of holistic approach. And so we think of it as a bit of a cycle with the clinical pathway in the middle.

222
00:20:31,520 --> 00:20:38,330
Okay. So, so the patient goes along that pathway and we think about the different points where this data is collected.

223
00:20:38,750 --> 00:20:42,799
So the top one here, we think here is we've mentioned before about the analytical performance.

224
00:20:42,800 --> 00:20:46,670
So that's how well a test detects a particular analyte. Okay.

225
00:20:46,670 --> 00:20:50,180
So that's the sort of thing that you might do in the laboratory under very controlled conditions.

226
00:20:50,960 --> 00:20:56,810
The next one here is the clinical performance. That's how well the device detects the condition or the physiological state.

227
00:20:57,320 --> 00:21:04,639
So, you know, often with manufacturers, the data they collect on this and that they present in terms of the diagnostic accuracy,

228
00:21:04,640 --> 00:21:09,709
the very often do this in the lab as well. And that's a bit of a problem because we know that in context.

229
00:21:09,710 --> 00:21:18,200
So where the test is actually used so by the busy clinicians in the context out in well in a public health setting or in a GP surgery,

230
00:21:18,560 --> 00:21:25,639
the results can be very, very different. Okay. Because people every now and again mess the test up, you know, they might make a mistake.

231
00:21:25,640 --> 00:21:28,940
The conditions that they're operating on might be slightly suboptimal.

232
00:21:29,270 --> 00:21:32,450
So that has an impact on the results that you get. Okay.

233
00:21:33,320 --> 00:21:37,879
The next one, very important, is about the clinical effectiveness. So that's downstream.

234
00:21:37,880 --> 00:21:42,770
So by introducing the test, what effect does that have on the outcome for the patient?

235
00:21:43,370 --> 00:21:50,900
So does that mean that the the patient has a better outcome the same, or does it have a deleterious effect on the patient?

236
00:21:52,100 --> 00:21:56,299
The next one along is about whether it's value for money, obviously cost effectiveness of the test.

237
00:21:56,300 --> 00:22:03,230
So is there a point? Yeah. If we introduce this, this is going to cost us an absolute fortune and I could only deliver a negligible benefit.

238
00:22:03,590 --> 00:22:08,210
Or is it going to sort of be a negligible sort of cost and very good patient benefit?

239
00:22:08,870 --> 00:22:14,540
And then this sort of rather nebulous thing at the end is this sort of concept of broad impact of the test.

240
00:22:14,990 --> 00:22:20,750
So that's things like, for example, does it have an impact on the drugs that are prescribed?

241
00:22:20,750 --> 00:22:26,360
You know, how many are prescribed? Does it have a positive impact on the patient's state of mind?

242
00:22:26,570 --> 00:22:30,680
What impact does it have on the doctor who's doing the test, etc., etc.?

243
00:22:30,680 --> 00:22:40,999
So it's a whole sort of big catch all. So what's the state of evidence for Point of Care Diagnostics so that for a number of

244
00:22:41,000 --> 00:22:45,920
years it's been running the Horizon scanning program within the diagnostic evidence group.

245
00:22:46,460 --> 00:22:54,320
And so what this is aimed to do is to produce kind of rapid, short, systematic reports on new developing diagnose.

246
00:22:54,550 --> 00:22:58,959
It's basically trying to sum up the current evidence and also try and picture

247
00:22:58,960 --> 00:23:02,650
that particular diagnostic within a pathway within the National Health Service.

248
00:23:03,160 --> 00:23:07,390
Okay. So there's about at the moment, I think we may actually have more than 48 now.

249
00:23:07,420 --> 00:23:11,800
I think it's about 48, 49 different reports that you can freely download.

250
00:23:11,810 --> 00:23:14,230
So if anybody's interested, they can go to our website.

251
00:23:15,370 --> 00:23:21,960
So sort of most recent one we did was on Helicobacter pylori infection and point of care testing for that.

252
00:23:21,970 --> 00:23:25,840
So it's this kind of sort of short, short form,

253
00:23:26,470 --> 00:23:32,020
short form publication and some of them are being converted into for publications and picked up by

254
00:23:32,380 --> 00:23:38,470
the British Journal of General Practice into these short sort of clinical intelligence reports.

255
00:23:39,100 --> 00:23:43,490
They've also gone on to form some health technology assessment programs as well within the CHA.

256
00:23:43,780 --> 00:23:48,009
So they've been quite pivotal and important for us. So what did we do?

257
00:23:48,010 --> 00:23:50,950
So we had an idea to think about what kind of evidence.

258
00:23:50,950 --> 00:23:55,760
So obviously we know there's a bit of a problem with these tests, point of care tests getting through into practice.

259
00:23:56,230 --> 00:23:59,490
So we were trying to get an idea of what the issue was with the evidence.

260
00:23:59,500 --> 00:24:04,750
So we had our own inklings about what the evidence looked like, but we tried to sort of quantify it a bit more.

261
00:24:05,530 --> 00:24:13,269
So we took all of the all of the Horizon scan reports that the Annette had overseen, and we did a big data extraction.

262
00:24:13,270 --> 00:24:20,440
So we basically looked at all of the primary data sources that had gone into the all of the different horizon scan reports.

263
00:24:20,740 --> 00:24:27,069
We extracted all of those and then we categorised all of those papers basically into a huge spreadsheet.

264
00:24:27,070 --> 00:24:34,450
So several, many, many hundreds of papers, actually. One of the sort of important things where we were looking for the sample size,

265
00:24:34,780 --> 00:24:43,390
we were looking for where the actual test was used because often all of these point of care tests were appraised in secondary care, not primary care.

266
00:24:43,400 --> 00:24:50,320
So we tried to think about where the actual data came from. And then we also looked to see whether it fulfilled the horvath's sort of cycle.

267
00:24:50,320 --> 00:24:55,750
So which which bit of evidence in that cycle was provided by the particular paper.

268
00:24:57,310 --> 00:25:04,060
So we did that and then we did a bit of a data analysis and plotted the data to have a have a look at what the data profile looked like.

269
00:25:04,630 --> 00:25:08,800
And in a way, it was quite gratifying because it confirmed what we suspected,

270
00:25:08,800 --> 00:25:16,120
which was that we had a huge emphasis on the sort of early stage of the cycle, so the analytical performance data,

271
00:25:16,120 --> 00:25:19,689
so that very often the stuff that's carried out in the lab, the clinical performance,

272
00:25:19,690 --> 00:25:27,310
but then there was a huge drop away really in the studies that covered the clinical effectiveness, the cost effectiveness and the broader impact.

273
00:25:27,760 --> 00:25:33,940
And I think what was also interesting was that this was a phenomenon that you could see across the different clinical domains.

274
00:25:33,940 --> 00:25:39,280
So we split by clinical domain. There you see the different domains at the bottom representing that.

275
00:25:39,310 --> 00:25:44,470
So these different colours and we could see that that that sort of cycle, that profile was repeated across.

276
00:25:44,980 --> 00:25:52,600
Okay. So there's sort of basically a huge gulf here of evidence which is sort of potentially lacking for these tests.

277
00:25:53,320 --> 00:25:58,180
And interestingly, this is the sort of test profile that's adopted largely by Nice as well.

278
00:25:58,180 --> 00:26:05,650
So if you think about nice evaluation in Britain for diagnostics and technologies, they tend to sort of roughly follow this evidence cycle of Horvath.

279
00:26:07,180 --> 00:26:12,669
So the other thing that was interesting was about how people had approached developing evidence.

280
00:26:12,670 --> 00:26:21,040
So we also looked at the different sort of types of evidence within the cycle, and we looked at the sequence in which people had done this.

281
00:26:21,040 --> 00:26:27,609
So you kind of almost imagine that people would start with the analytical performance and then they work around the cycle, okay?

282
00:26:27,610 --> 00:26:29,979
From the sort of analytical performance,

283
00:26:29,980 --> 00:26:36,790
the diagnostic accuracy assessment through to the impact and right through to the broader impact to the test, but actually was very haphazard.

284
00:26:36,790 --> 00:26:43,720
So in terms of this of length of time over which the studies were carried out, see that people did it in all sorts of random order.

285
00:26:43,730 --> 00:26:51,190
So not necessarily very logical. And also, I think the other important thing here was that we found that this whole cycle too,

286
00:26:51,190 --> 00:26:57,999
so to complete a full evidence cycle for a particular point of care test, took a median time of about nine and a half years.

287
00:26:58,000 --> 00:27:03,390
So is comparable really with drugs. So think about how long it takes to get a drug through this.

288
00:27:03,400 --> 00:27:06,700
A process of evidence right to the point where it can be approved.

289
00:27:06,700 --> 00:27:10,870
It's about nine, ten years time and it's very similar for diagnostic diagnostics.

290
00:27:10,870 --> 00:27:16,420
But obviously the the sort of yield in terms of capital for company is much, much lower, usually for a diagnostic.

291
00:27:17,980 --> 00:27:21,879
So the neglect to the elements, clinical effectiveness, the cost effectiveness,

292
00:27:21,880 --> 00:27:26,950
the broader impact and as I mentioned, so that's all these things like acceptability, social.

293
00:27:27,330 --> 00:27:32,350
So the impact of a test psychological so you can think about if somebody has some sort of,

294
00:27:33,100 --> 00:27:36,730
you know, point of care test done in the community for some sort of genetic disorder.

295
00:27:37,150 --> 00:27:41,350
Obviously, finding out the result of that is not necessarily going to have a very positive impact on the patient,

296
00:27:41,350 --> 00:27:45,159
for example legal, ethical, societal and organisational.

297
00:27:45,160 --> 00:27:49,270
So how these tests fit in can disrupt an organisation financially.

298
00:27:49,870 --> 00:27:53,680
Okay, so I'm just going to go through sort of.

299
00:27:53,800 --> 00:28:01,660
Finish off a couple of case studies. So I'll call this sort of unexpected or less easily predicted impact and unexpected benefit.

300
00:28:02,050 --> 00:28:05,140
Okay. So we've got two case studies here to talk through briefly.

301
00:28:05,830 --> 00:28:10,030
So the first one is on rapid antigen diagnostic tests for malaria.

302
00:28:10,060 --> 00:28:18,430
So back in 2010, who the World Health Organisation generated new guidance for,

303
00:28:19,090 --> 00:28:25,150
for basically sort of the the prescribing prescribing of artemisinin combination therapy for malaria.

304
00:28:26,200 --> 00:28:31,750
So what they said was that they wanted to move away from presumptive prescription.

305
00:28:31,900 --> 00:28:37,540
Okay. So they wanted to move away from that to evidence based prescription by the introduction of these rapid tests.

306
00:28:38,110 --> 00:28:41,830
So the rapid test is very much like a pregnancy test.

307
00:28:41,840 --> 00:28:45,250
There's lots of different versions of this cards and they sort of sticks.

308
00:28:45,640 --> 00:28:50,890
And essentially what you do is with a patient, you take a finger prick sample, you take it up in a pipette,

309
00:28:51,340 --> 00:28:57,370
the bit of blood sample goes into the hole here and then a buffer sample is added here to flow the sample through.

310
00:28:57,850 --> 00:29:01,870
Okay. And then if the if malaria, the malaria parasite is present,

311
00:29:02,110 --> 00:29:07,390
it cross reacts here and you get a control line, which is some ubiquitous factor from the from the blood.

312
00:29:08,140 --> 00:29:11,230
And you get another line here to say that the test is positive. Okay.

313
00:29:11,230 --> 00:29:18,160
So there they're actually very, very these have been around for donkey's years and they're usually very, very accurate and they're pretty cheap.

314
00:29:18,160 --> 00:29:19,580
They're about less than $10.

315
00:29:20,050 --> 00:29:25,990
So there is a bit of an issue with that because at the moment, artemisinin combination therapy is actually a couple of dollars.

316
00:29:26,350 --> 00:29:30,909
So you've always got that sort of when you introduce the diagnostic, you've often got this conflict,

317
00:29:30,910 --> 00:29:36,190
which is diagnostics are often relatively expensive compared with the drugs that they're trying to guide.

318
00:29:36,190 --> 00:29:40,690
So. But anyway, so who decides to make this recommendation?

319
00:29:41,380 --> 00:29:46,510
And so they collected some surveillance data from a number of settings.

320
00:29:46,510 --> 00:29:51,459
So there were eight settings in subsaharan Africa and two in Afghanistan.

321
00:29:51,460 --> 00:29:56,200
So they were very, very variable in terms of the prevalence or the incidence of the disease.

322
00:29:56,830 --> 00:30:02,850
So you've got areas like this. So this is Uganda in the Great Lakes region, and you can see here on this side.

323
00:30:02,890 --> 00:30:09,910
So this is shows map. This is where you've got a greater than 300 confirmed cases per thousand in the population.

324
00:30:10,570 --> 00:30:16,150
So actually, probably most people coming through your door who who are febrile got a temperature.

325
00:30:16,780 --> 00:30:20,169
There is a very strong likelihood they're going to have the disease, isn't there? Okay.

326
00:30:20,170 --> 00:30:27,490
We're in a situation like that, whereas at the other end of the extreme, you have places in Afghanistan where incidence is very, very low.

327
00:30:27,610 --> 00:30:31,720
Okay. So very, very different settings. So.

328
00:30:36,470 --> 00:30:41,120
Okay. So they introduced this and then they followed up to look at the impact of the test.

329
00:30:42,230 --> 00:30:46,910
Okay. So they were looking for a whole range of impacts, but there are a couple of important ones.

330
00:30:47,540 --> 00:30:50,630
Okay. So they introduced it. And so what happened next?

331
00:30:50,660 --> 00:30:52,190
Anybody think about what would happen?

332
00:30:52,200 --> 00:31:00,050
So they're basically introducing the diagnostic to try and control the prescription of artemisinin combination therapy.

333
00:31:01,040 --> 00:31:05,790
So anybody got any thoughts about what might have happened? Yep.

334
00:31:06,390 --> 00:31:11,320
One of the big challenges is that the patient comes to that kind of when you diagnose and they get.

335
00:31:12,500 --> 00:31:17,990
I didn't pay the consultation to be speaking for colleagues and yet I have seen that it is not.

336
00:31:20,860 --> 00:31:26,260
Okay. So it's a good point. And you've picked up on some the inadvertently we'll discuss in a second.

337
00:31:26,980 --> 00:31:29,980
So what they found was pretty much across all the sites. So.

338
00:31:30,790 --> 00:31:37,360
Okay. So on this side, this is the prescription of AZT without the diagnostic test.

339
00:31:38,110 --> 00:31:42,670
And these are the same sites later with the introduction of the diagnostic test.

340
00:31:43,000 --> 00:31:49,260
So pretty much uniformly, they saw that there was a reduction in the prescription of artemisinin combination therapy.

341
00:31:49,270 --> 00:31:53,870
So that was kind of one of their goals. They had a bit of a flawed goal as well.

342
00:31:53,890 --> 00:31:56,500
I mean, when you read the read the original guidance,

343
00:31:56,500 --> 00:32:04,540
what they said was that they wanted to reduce the use of AZT to try and combat resistance in Plasmodium falciparum.

344
00:32:05,440 --> 00:32:07,120
But I think that's a big, big issue with that,

345
00:32:07,120 --> 00:32:12,700
which is that you don't drive resistance in an organism by giving drugs to people who don't have the disease.

346
00:32:13,540 --> 00:32:18,670
So they were slightly flawed in that respect, but anyway, they succeeded in that.

347
00:32:18,760 --> 00:32:22,180
So they actually got their AZT prescriptions down.

348
00:32:22,210 --> 00:32:27,860
Obviously, there are other positive benefits of that in terms of, you know, there are side effects associated with these treatments.

349
00:32:27,860 --> 00:32:31,540
They're generally well tolerated. But so it's not not all negative, of course.

350
00:32:32,200 --> 00:32:39,400
So what was the next thing that happened? Anybody think back to the gentlemen's point at the back?

351
00:32:40,150 --> 00:32:44,110
Can I really have a guess? Yeah, but this went up.

352
00:32:44,410 --> 00:32:53,560
Yeah, exactly. So they found pretty much across all the sites, there was a modest increase in most of the sites in terms of antibiotic prescriptions.

353
00:32:53,920 --> 00:32:57,250
Exactly. As the gentleman there said, which is that you've got a patient in front of you.

354
00:32:57,730 --> 00:33:00,640
They're febrile, but they don't have malaria. So what do you do?

355
00:33:01,060 --> 00:33:08,379
So you're thinking about facilities here that had community health care workers, so they don't necessarily have an in-depth clinical knowledge.

356
00:33:08,380 --> 00:33:12,730
So the prescribing, you know, they feel under pressure and they want to do something for a patient.

357
00:33:13,210 --> 00:33:16,750
So they also found as well similarly that there were antipyretic.

358
00:33:16,960 --> 00:33:24,070
So there was a slight increase in Antipyretic. So things like paracetamol, ibuprofen prescribed for people as well in certain areas yet.

359
00:33:24,520 --> 00:33:28,000
But were they concerned about false negative results of the test?

360
00:33:28,720 --> 00:33:31,720
I don't think particularly because these tests are pretty good.

361
00:33:31,990 --> 00:33:37,210
I think the sort of the performance is not quite so good in children as it is in adults.

362
00:33:38,110 --> 00:33:42,490
But you would also found within this group that people would have been presumptively prescribing as well.

363
00:33:42,520 --> 00:33:47,810
I think even people who got a negative test result would probably have been exercising clinical judgement.

364
00:33:47,830 --> 00:33:53,680
So it's not purely based on the result of the test. Okay, so that's that one.

365
00:33:53,680 --> 00:33:59,680
So. And another one is about the instability of serum potassium from primary care samples.

366
00:33:59,680 --> 00:34:04,840
So this is another case study. So this is thinking about where a test actually might have a benefit.

367
00:34:05,020 --> 00:34:10,540
Okay. So point of care test. So there's quite a well documented phenomenon in primary care samples.

368
00:34:10,540 --> 00:34:15,160
So samples that come out of the community, which is not seen in hospital samples where there's a lab on site.

369
00:34:15,760 --> 00:34:18,100
Okay. And what you see is that there's this temperature dependence.

370
00:34:18,100 --> 00:34:24,580
So if you take a a blood sample from a patient and you send it up to the hospital lab, what?

371
00:34:24,790 --> 00:34:28,780
Sinclair, with quite a number of studies that show this isn't restricted to there.

372
00:34:28,810 --> 00:34:36,910
So this particular paper, what they showed was that when the ambient temperature increases, you get a decrease in serum potassium.

373
00:34:37,240 --> 00:34:40,930
Okay. So and this phenomenon you actually see in hospital.

374
00:34:42,370 --> 00:34:48,310
So what? This is the only bit of basic science we can actually have in this talk.

375
00:34:48,790 --> 00:34:56,589
So it's to do with really the so the instability of the potassium in the serum of the of the blood sample is down to this phenomenon.

376
00:34:56,590 --> 00:35:01,360
So you've got if this is our serum here and this is a red blood cell,

377
00:35:02,620 --> 00:35:07,230
you've got a protein on the membrane of the red blood cell, which is a sodium potassium ATP.

378
00:35:07,270 --> 00:35:11,560
So it basically moves sodium out of cell and potassium in.

379
00:35:11,590 --> 00:35:16,660
Okay. So it sort of helps to maintain the resting state, what we call the resting potential of the cell.

380
00:35:17,200 --> 00:35:22,209
Okay. So it's part of this of homeostatic mechanism. This is temperature dependent.

381
00:35:22,210 --> 00:35:27,880
So if you have a sample that's been transported potentially over a long distance from a GP practice,

382
00:35:28,510 --> 00:35:34,210
the warmer it is, the the sort of more this process is driven okay within the sample.

383
00:35:34,750 --> 00:35:40,480
So you get this movement of potassium so loading into the erythrocytes and then when you actually do a measurement of the serum potassium,

384
00:35:41,050 --> 00:35:44,860
it potentially has an effect of lowering the serum potassium. Okay.

385
00:35:44,860 --> 00:35:49,959
If the sample and this actually this process can be blocked so pharmacologically,

386
00:35:49,960 --> 00:35:55,000
you can isolate this by applying a compound called wiper in which blocks the sodium potassium atpase.

387
00:35:55,330 --> 00:36:01,910
So this is partly how this sort of mechanism was elucidated. So the relationships are pretty, pretty clear, actually.

388
00:36:01,930 --> 00:36:06,579
So this is data that's come out of the John Ratcliffe Hospital and was provided by a colleague of ours.

389
00:36:06,580 --> 00:36:11,140
So what they showed here was the mean weekly temperature in degrees and the serum potassium.

390
00:36:11,440 --> 00:36:16,530
So basically the higher the temperature, the lower the mean potassium levels were in the samples.

391
00:36:16,600 --> 00:36:21,470
So you can see that this. It is is kind of quite quite stocks, quite clear.

392
00:36:22,400 --> 00:36:30,350
Physical potentials have quite big impacts because obviously if you've got a patient who's somewhere near the sort of cusp of being hyperkalemia,

393
00:36:30,470 --> 00:36:33,140
so they've got maybe too much potassium in their blood.

394
00:36:33,950 --> 00:36:39,169
You know, through this actual phenomenon, you've got the potential where a patient is driven below this line.

395
00:36:39,170 --> 00:36:45,860
Okay? And that might affect the decision of the clinician to treat or not or to escalate this person to hospital or not.

396
00:36:46,550 --> 00:36:49,490
So it could exert a negative influence on the treatment treatment decision.

397
00:36:49,490 --> 00:36:52,610
So you could actually have a very negative patient outcome in this circumstance.

398
00:36:53,990 --> 00:36:57,500
So but one of the things with point of care testing, obviously,

399
00:36:57,500 --> 00:37:02,780
if you're doing this point of care tests on a patient, you're not going to have that issue of transport.

400
00:37:02,780 --> 00:37:06,769
So you have much more you have much greater stability in the potassium levels in the sample.

401
00:37:06,770 --> 00:37:11,860
So it's not something that's a problem. And there are currently a number of platforms that are available or do this.

402
00:37:11,870 --> 00:37:17,269
So actually within the trust at the moment on several sites, there are a number of point of care tests that will do blood potassium.

403
00:37:17,270 --> 00:37:24,470
So blood electrolyte. So we don't have this phenomenon well, so we don't have the problem with this phenomenon within these community settings.

404
00:37:25,700 --> 00:37:31,459
So just to summarise so I think we probably say so medical practices are actually parachutes.

405
00:37:31,460 --> 00:37:38,210
So going back to the original comment about the people being totally convinced that a test is going to have a very positive impact,

406
00:37:38,630 --> 00:37:41,870
that's really the case because very often it's a lot more complicated than that.

407
00:37:41,870 --> 00:37:47,809
And this is actually an infographic that was produced by Vinay Prasad over in in Oregon and his group.

408
00:37:47,810 --> 00:37:52,310
So they published a paper and that followed up from the original 2003 paper.

409
00:37:52,310 --> 00:38:01,520
And what they did was they they looked and searched for all of the papers that subsequently referenced the original 23 parachute paper.

410
00:38:02,000 --> 00:38:09,380
They found 35 papers that actually claimed that the practices that they were trying to evaluate were equivalence of the parachutes.

411
00:38:09,770 --> 00:38:16,970
So there are so, so clear in terms of their perceived benefit that they were as good as a parachute in terms of risk reduction.

412
00:38:17,900 --> 00:38:27,110
So 22 of these were actually evaluated by a randomised controlled trial and only six of them came out as being positive trials.

413
00:38:27,260 --> 00:38:34,010
Okay. So, so that's from basically 35 different groups being convinced that something was going

414
00:38:34,010 --> 00:38:37,730
to be overwhelmingly positive and only six of them came out with positive trials.

415
00:38:38,060 --> 00:38:46,970
Okay. So I think the sort of to take home message is really that parachute diagnostic interventions don't really exist or are extremely unlikely.

416
00:38:47,540 --> 00:38:53,899
So it's very vital to consider diagnostic or any kind of diagnostic intervention in the context of the full clinical pathway.

417
00:38:53,900 --> 00:39:01,940
So that's why we need to generate this cycle of evidence if we could understand the full impact of the test and to achieve successful implementation.

418
00:39:03,350 --> 00:39:06,919
So just like to acknowledge a few people because a few people who've commented on

419
00:39:06,920 --> 00:39:11,660
this and provided some bits and pieces and also just to give our contact details,

420
00:39:11,900 --> 00:39:16,010
if anybody wants to talk to us about this further, I'd be very happy to to discuss it.

421
00:39:16,080 --> 00:39:16,880
Thank you very much.