1
00:00:00,210 --> 00:00:12,050
To introduce. Dr. Merlin Wilcox is an academic, clinical lecturer and a practising GP and he's based in Oxford and Southampton.

2
00:00:12,060 --> 00:00:13,830
So having to share him for the time being.

3
00:00:15,030 --> 00:00:19,920
And Mel is going to talk to us about trials and tribulations and explain his work that he's been doing there.

4
00:00:20,190 --> 00:00:23,760
And I think you're happy for people to interact.

5
00:00:24,120 --> 00:00:28,290
Yeah, yeah, absolutely. Yes. What you make of it.

6
00:00:28,770 --> 00:00:35,580
Thank you very much. Thank you very much. Thank you for asking me. It's great to be here and nice to meet you.

7
00:00:36,300 --> 00:00:43,379
So as Clare said, I want to make this interactive, so please just interrupt if you've got questions or thoughts or whatever.

8
00:00:43,380 --> 00:00:45,480
And I'm going to be asking you some questions as well.

9
00:00:46,830 --> 00:00:53,550
So as Clare mentioned, I'm a GP and an academic clinical lecturer and I've been one of my main interests is global health.

10
00:00:53,560 --> 00:00:57,690
I've been involved in a variety of different research projects in Africa,

11
00:00:57,690 --> 00:01:03,479
and I'm going to give you a flavour of some of those with a particular focus on illustrating different study designs,

12
00:01:03,480 --> 00:01:06,300
because I understand that's what you're doing this week.

13
00:01:07,440 --> 00:01:12,950
So before I get started, I'm going to show you a picture of a couple of children because I like children.

14
00:01:12,960 --> 00:01:16,950
This is my daughter blowing out the candles on her fifth birthday cake.

15
00:01:17,550 --> 00:01:23,700
And this is a patient called Abu, who I treated in Mali a few years ago.

16
00:01:24,180 --> 00:01:29,130
And I want to start off by asking you a question which only four isn't allowed to answer because he knows the answer.

17
00:01:30,180 --> 00:01:36,479
How many times more likely is a child to die before their fifth birthday in Mali, which is a country in West Africa?

18
00:01:36,480 --> 00:01:40,890
In case you didn't know compared to the UK, it's a multiple choice question.

19
00:01:41,220 --> 00:01:47,610
If you think the answer is three times, put up your hand. If you think the answer is ten times, put up your hand.

20
00:01:48,750 --> 00:01:51,780
If you think the answer is 30 times, put up your hands.

21
00:01:52,920 --> 00:01:57,060
Well, unfortunately, those of you who've just put up your hands are correct.

22
00:01:57,930 --> 00:02:03,720
A child in Mali is 30 times more likely to die before their fifth birthday than a child in the UK.

23
00:02:04,050 --> 00:02:09,330
So the under-five mortality rate is one in 200. In the UK and Mali it's one in six.

24
00:02:10,170 --> 00:02:18,780
So I think a lot of us realise the world is an unequal and unfair place, but not many people realise quite how unequal and unfair it is.

25
00:02:19,500 --> 00:02:21,540
You might be wondering why I'm starting off with this,

26
00:02:21,540 --> 00:02:27,269
but this is my sort of premise for my passion in life, which is how do we reduce child mortality?

27
00:02:27,270 --> 00:02:34,530
And these are my research questions, very broad global research questions, which I'm not going to be able to answer completely tonight.

28
00:02:34,530 --> 00:02:39,840
But this is sort of the reason behind all of the studies which I'm going to be talking about.

29
00:02:40,200 --> 00:02:45,450
Why are so many children dying in Africa? What can be done to reduce childhood mortality?

30
00:02:45,450 --> 00:02:49,290
And how should we spend our money to save as many lives as possible?

31
00:02:49,620 --> 00:02:54,179
I'm not going to really answer fully any of these questions tonight, but as I say,

32
00:02:54,180 --> 00:02:58,230
this is just to give the context of the reason for the studies that I am showing.

33
00:03:00,390 --> 00:03:07,500
So this is an outline of what I'm going to talk about. I'm going to start off with why bother with different study designs?

34
00:03:07,500 --> 00:03:13,960
And then I'm going to give you examples of different observational study designs and experimental study designs

35
00:03:13,980 --> 00:03:18,780
of studies illustrating these various designs which you may or may not have already been talking about.

36
00:03:18,780 --> 00:03:23,160
And some of them are a bit innovative or wacky or different, probably from the standard ones.

37
00:03:24,780 --> 00:03:34,650
So just to start off with the hierarchy of evidence, which I'm sure you've all seen before, looking at this, one might think, well,

38
00:03:35,190 --> 00:03:42,600
the only thing that policy makers and decision makers ever look at is actually a systematic reviews of randomised controlled trials.

39
00:03:42,810 --> 00:03:48,540
So why bother with any of the rest? Why not just do randomised controlled trials of everything?

40
00:03:48,540 --> 00:03:55,140
Because that's the only thing that matters. Surely cohort studies and case control studies and case reports and all the rest of it or waste of time?

41
00:03:55,860 --> 00:04:02,820
Well, part of what I'm going to try and show you this evening is that actually those studies are quite useful and they do have their place.

42
00:04:04,080 --> 00:04:12,330
So bear with me on that. So let's start off then with the top of the pyramid of the evidence based medicines that we're talking about.

43
00:04:12,540 --> 00:04:19,980
Since I'm talking about child mortality in Africa, there are lots of systematic reviews, and this is one from The Lancet published a few years ago,

44
00:04:20,280 --> 00:04:25,590
which found that actually there's quite a lot of interventions which have very good evidence that they work.

45
00:04:26,250 --> 00:04:33,090
And basically, if they were all implemented, 63% of childhood deaths could be prevented.

46
00:04:34,290 --> 00:04:39,450
So maybe we should just leave it at that and start implementing what we already know.

47
00:04:39,960 --> 00:04:50,700
This is the summary of the studies from the this systematic review showing that for the biggest causes of under-five death, there are,

48
00:04:50,700 --> 00:04:59,700
for most of them, level one, which means sufficient evidence of interventions which are effective and could be rolled out to prevent.

49
00:05:00,010 --> 00:05:07,990
Most of these causes of death. So there are those who say we should just implement what we already know.

50
00:05:08,680 --> 00:05:12,490
But interestingly, this is a thing that was published a few years ago in The Lancet.

51
00:05:12,500 --> 00:05:17,430
UNICEF attempted to do just that in several countries in Africa.

52
00:05:17,440 --> 00:05:21,969
They tried to implement this accelerated child survival and development program,

53
00:05:21,970 --> 00:05:25,910
which is basically trying to implement all of those evidence based interventions.

54
00:05:26,320 --> 00:05:30,610
And they found that actually it didn't, unfortunately, seem to make much of a difference.

55
00:05:30,970 --> 00:05:34,810
So the areas where they implemented it, yes, the mortality reduced,

56
00:05:35,110 --> 00:05:44,470
but it wasn't significantly greater than the control areas in the in the same countries.

57
00:05:45,400 --> 00:05:49,150
And why? Well, it's to do with implementation problems, etc.

58
00:05:49,960 --> 00:05:54,640
So maybe randomised trials and systematic reviews aren't enough on their own.

59
00:05:54,940 --> 00:06:03,060
Maybe we need some research into how to implement things and scale them up effectively because clearly it's not yet working properly,

60
00:06:03,070 --> 00:06:11,320
otherwise all these kids would be dying. So my first question is why is so many children still dying in Mali?

61
00:06:11,350 --> 00:06:17,260
Mali is one of the countries in West Africa with probably to the highest under-five mortality rates,

62
00:06:17,410 --> 00:06:20,560
one of the countries with the highest and five mortality rates in the world.

63
00:06:21,910 --> 00:06:29,770
So a good study designed for answering this question is a confidential inquiry, which is a sort of mixed methods approach.

64
00:06:30,520 --> 00:06:35,680
It includes quantitative information on numbers of deaths and causes of death,

65
00:06:36,100 --> 00:06:44,500
but it also has qualitative information on things that are avoidable factors, recommendations of what could be done to avoid these deaths.

66
00:06:45,850 --> 00:06:52,479
So we asked village health workers, which are basically volunteers in villages,

67
00:06:52,480 --> 00:06:59,980
to report all the deaths of children under five within the selected study areas over a period of two or three years.

68
00:07:00,610 --> 00:07:06,730
And then we asked Fieldworkers to interview the families, visit the families, invite them to be interviewed.

69
00:07:07,510 --> 00:07:14,799
Obviously, we asked for them for the informed consent and the fieldworkers conducted what's called a verbal autopsy interview,

70
00:07:14,800 --> 00:07:23,350
which is a standard questionnaire. But there's also open ended questions in there asking about the story of what happened before the child died,

71
00:07:23,530 --> 00:07:27,820
asking about their symptoms, but also where they went for treatment, what treatments they took, and so on.

72
00:07:28,450 --> 00:07:33,340
And we also attempted to interview any health workers involved in looking after those children.

73
00:07:34,720 --> 00:07:42,430
And all of this information was summarised and brought together to a panel of doctors, nurses.

74
00:07:42,730 --> 00:07:46,510
We even it sometimes included village health workers, traditional healers.

75
00:07:46,960 --> 00:07:51,040
And they discussed the case and tried to decide what was the most likely cause of death.

76
00:07:51,880 --> 00:07:59,500
What were the avoidable factors, and what recommendations could be made to avoid similar deaths in the future?

77
00:08:01,330 --> 00:08:05,950
And so these are some of the results. And this is just the one family.

78
00:08:05,950 --> 00:08:09,580
We did this in Uganda as well, but I'm not going to show those results now.

79
00:08:10,690 --> 00:08:14,530
And these are these are the top causes of death, according to our study.

80
00:08:15,010 --> 00:08:18,750
So interestingly, malaria comes out as a clear winner.

81
00:08:18,760 --> 00:08:21,820
46% of the deaths were attributed to malaria.

82
00:08:22,180 --> 00:08:25,370
Interestingly, that's very different from the W.H.O. statistics.

83
00:08:25,660 --> 00:08:29,500
That puts malaria in third place with only 14% of deaths.

84
00:08:30,190 --> 00:08:33,920
So maybe our areas are different from the ones that W.H.O. sampled.

85
00:08:33,940 --> 00:08:40,600
We don't really know how W.H.O. got those figures, but anyway, they were not applicable really for our study area.

86
00:08:41,470 --> 00:08:47,200
Also interestingly, malnutrition came up as a quite an important direct cause of death,

87
00:08:47,500 --> 00:08:52,660
which doesn't even appear in the W.H.O. list of causes of death.

88
00:08:53,020 --> 00:08:57,610
So already this has given us some new and different information from the standard statistics.

89
00:08:58,870 --> 00:09:06,850
And you might think and a commonly made assumption is that if you know the causes of death, then you can work out what to do about it.

90
00:09:07,420 --> 00:09:10,120
Because if the top cause of death is malaria,

91
00:09:10,390 --> 00:09:18,000
then we know from our systematic reviews that mosquito nets are an effective intervention for preventing deaths from malaria.

92
00:09:18,010 --> 00:09:22,920
So probably we should roll out mosquito nets if we take that assumption to be true.

93
00:09:22,930 --> 00:09:29,050
But is it true? So these are the avoidable factors from our Confidencial inquiry.

94
00:09:29,770 --> 00:09:35,679
So this is the percentage of deaths in which the intervention could have prevented

95
00:09:35,680 --> 00:09:42,310
the fatal illness and in which it wasn't already being used by the person who died.

96
00:09:43,300 --> 00:09:48,490
And interestingly, mosquito nets are right at the bottom only 5% of deaths.

97
00:09:49,210 --> 00:09:55,480
So that's bizarre. Why? Well, it turns out that the majority of kids who died of malaria,

98
00:09:55,930 --> 00:10:00,550
their parents claimed at least that they had already been sleeping under a mosquito net.

99
00:10:01,720 --> 00:10:05,410
Now, from this study, we don't have enough information to unpick that.

100
00:10:05,650 --> 00:10:10,120
There could be a variety of different reasons why they weren't working.

101
00:10:10,240 --> 00:10:14,440
Maybe they weren't sleeping under them every night. Maybe the parents were lying.

102
00:10:14,680 --> 00:10:23,469
Maybe the mosquito nets had holes in them. Maybe mosquitoes changed their biting habits so they bite earlier than children.

103
00:10:23,470 --> 00:10:28,629
Go to bed. You laugh. But it's true. There's evidence not from here, but from Indonesia.

104
00:10:28,630 --> 00:10:37,990
I think in some places the selective pressure of mosquito nets, because they're so widely used, has actually selected mosquitoes that bite.

105
00:10:37,990 --> 00:10:42,280
It's a different time of day because they are so good at killing mosquitoes that bite during the night.

106
00:10:42,700 --> 00:10:46,150
It could be that mosquitoes have become resistant to insecticides.

107
00:10:46,810 --> 00:10:57,010
There's also evidence that that's happening in various places. So we don't know why mosquito nets seem to not be working in this group.

108
00:10:57,850 --> 00:11:05,829
But what we can say from this is that continuing to roll out mosquito nets in the way that it's being done now probably

109
00:11:05,830 --> 00:11:13,480
actually isn't going to have as big an impact as one might predict just by looking at the cause of death statistics.

110
00:11:15,040 --> 00:11:19,810
The other interesting thing is that nutrition, as in good nutrition and family planning,

111
00:11:20,110 --> 00:11:24,820
come quite high up as things that could prevent quite a lot of deaths.

112
00:11:25,240 --> 00:11:27,910
And you wouldn't really predict either of those.

113
00:11:28,330 --> 00:11:33,850
Maybe nutrition, but you wouldn't predict that it was so important by looking at this list of causes of death.

114
00:11:34,810 --> 00:11:39,820
So interviewing people about avoidable factors, which has a qualitative component to it.

115
00:11:40,330 --> 00:11:44,469
This was all a lot of it was derived qualitatively by what the panels thought,

116
00:11:44,470 --> 00:11:50,650
etc. and then we quantified it afterwards is quite a useful thing to do as well.

117
00:11:51,910 --> 00:11:53,560
Now where did the children die?

118
00:11:54,370 --> 00:12:03,550
This picture actually is taken from an article that was published in 2001 using statistics from the 1990s about where children died.

119
00:12:04,030 --> 00:12:11,140
And the article was called The Heirs of the Hippopotamus. And it was making the point that if you work in a hospital in Africa,

120
00:12:11,560 --> 00:12:18,370
you only see the ears of the hippopotamus because fewer than 5% of children at that time were coming to hospital.

121
00:12:18,910 --> 00:12:25,320
There are no icebergs in Africa. So what were are comparable figures?

122
00:12:25,330 --> 00:12:28,600
So in 2012, these are the statistics from our study.

123
00:12:28,810 --> 00:12:33,970
It hasn't really changed very much. You can see a little bit more of the years, 12%.

124
00:12:34,360 --> 00:12:41,080
But still, the majority of the children are dying at home or on the way to a health facility.

125
00:12:42,040 --> 00:12:46,870
So really, if you're just focusing on hospitals, you're going to miss the majority of the problem.

126
00:12:48,610 --> 00:12:57,219
Okay. So summarising the advantages and disadvantages of this study design method of a confidential inquiry, the advantage is all that.

127
00:12:57,220 --> 00:13:05,400
It does give you very detailed information about the deaths. It enables you to prioritise the problems that need to be addressed.

128
00:13:05,410 --> 00:13:12,489
And it's very good for generating recommendations and hypotheses, but it does have some disadvantages.

129
00:13:12,490 --> 00:13:15,490
We don't actually know what happened to the children who didn't die.

130
00:13:15,940 --> 00:13:23,140
So, you know, we don't know if there was a difference in how many of those were using the various interventions that we were talking about.

131
00:13:23,770 --> 00:13:28,860
And we don't necessarily know which interventions are most effective for tackling the problems identified.

132
00:13:28,870 --> 00:13:32,799
So for example, for malaria, we don't actually know in this context.

133
00:13:32,800 --> 00:13:36,400
Well, you know, just the net program needs to be modified.

134
00:13:36,400 --> 00:13:41,740
Or is it about improving treatments when there is a lot of information in here about treatment,

135
00:13:41,740 --> 00:13:47,799
seeking poor quality of care, which I haven't gone into and I haven't got time to go into, but it's a cut.

136
00:13:47,800 --> 00:13:53,230
Long story short, that was a big part of the problem as well. People not getting the right treatments and the right time.

137
00:13:54,790 --> 00:13:58,690
Okay. But moving on. So the next question,

138
00:13:59,290 --> 00:14:08,170
which sort of comes out of this is how is malaria being managed in Mali focusing on malaria since that's the most important cause of death?

139
00:14:08,410 --> 00:14:10,600
Which treatments are being used? Do they work?

140
00:14:12,130 --> 00:14:19,810
So the study I'm going to tell you about next is something that we called the Retrospective Treatment Outcome Study.

141
00:14:19,810 --> 00:14:23,950
And I'm pretty sure that's not something that you will have covered or will cover in this course,

142
00:14:23,950 --> 00:14:27,070
because it's a slightly different innovative study design.

143
00:14:27,430 --> 00:14:29,860
It's basically a sort of cross-sectional study.

144
00:14:30,550 --> 00:14:40,390
But the idea was to interview parents of children about specific cases of malaria within a short record period.

145
00:14:40,600 --> 00:14:46,030
So the idea was the fieldworkers would go to the village and look for families where a child had

146
00:14:46,030 --> 00:14:51,700
had malaria and then interview them about what treatment they had taken within a short period.

147
00:14:51,700 --> 00:14:59,260
That means within the last month to three months, basically, so that the parents are likely to remember what happened and what they did.

148
00:15:00,850 --> 00:15:02,470
Because it's retrospective.

149
00:15:02,480 --> 00:15:10,000
We had to use a syndromic definition of malaria, which means basically fever because you can't do blood tests retrospectively.

150
00:15:11,020 --> 00:15:15,250
And the idea was to analyse what happened to the patients taking different treatments.

151
00:15:15,490 --> 00:15:22,360
Were there any that were associated with particularly good or bad outcomes and adjust for confounding factors?

152
00:15:24,130 --> 00:15:29,680
So in this study we looked at 952 case histories of children.

153
00:15:29,690 --> 00:15:39,040
The majority of them had been treated at home, 40% had taken only modern medicine, 33% and combined modern and traditional medicine.

154
00:15:39,340 --> 00:15:42,490
And 27% had taken only traditional medicine.

155
00:15:43,630 --> 00:15:53,530
So what happened to them? Well, not surprisingly, those with uncomplicated malaria, virtually none of them died, but those with severe malaria.

156
00:15:54,130 --> 00:15:57,670
The interesting thing is, if you look at the percentages who died.

157
00:15:58,030 --> 00:16:07,450
26% of those who took modern treatment died and 11% of those who had taken traditional treatment died.

158
00:16:09,130 --> 00:16:11,530
And you think hang on a minute. Isn't that the wrong way round?

159
00:16:13,780 --> 00:16:20,330
Surely those who took traditional treatment should be more likely to die than those who took modern treatments.

160
00:16:20,350 --> 00:16:27,910
If you assume that modern treatment works and traditional treatment is rubbish, but that those are the right numbers.

161
00:16:28,870 --> 00:16:36,400
So that's a bit puzzling. And with this study design, you can't explain those numbers.

162
00:16:37,060 --> 00:16:42,310
They are what they are. So we're going to go on to this poses some more questions.

163
00:16:43,060 --> 00:16:50,049
So before I pose more questions, I'm going to pause and think about the advantages and disadvantages of this method,

164
00:16:50,050 --> 00:16:51,970
since that's the sort of theme of tonight.

165
00:16:53,260 --> 00:17:01,240
So it's quite a good method for measuring treatment, seeking behaviour in the whole population if you're interested in a specific illness.

166
00:17:01,570 --> 00:17:04,870
We can look for associations between treatments and outcomes.

167
00:17:06,310 --> 00:17:08,830
And it's quite good for generating hypotheses.

168
00:17:08,860 --> 00:17:15,999
So we could hypothesise that modern medicine is rubbish or modern medicine is not being effectively applied,

169
00:17:16,000 --> 00:17:21,850
or maybe the medicines are of bad quality or or people are just going there too late.

170
00:17:22,210 --> 00:17:27,040
We could hypothesise that maybe the traditional medicine is working or some of them are working.

171
00:17:27,490 --> 00:17:36,969
There are lots of hypotheses. We don't know if any of them are true. Disadvantages are that because it's retrospective, it's a presumed diagnosis.

172
00:17:36,970 --> 00:17:41,350
So we're not actually sure that these patients have malaria. Maybe most of them didn't even have malaria.

173
00:17:41,350 --> 00:17:47,530
Maybe it was something else. And also the major problem is that differences in our.

174
00:17:47,640 --> 00:17:50,640
Comms could be due to differences between the patients.

175
00:17:51,060 --> 00:17:54,990
Maybe those going to modern medicine are sicker than those who went to traditional medicine.

176
00:17:55,860 --> 00:18:04,129
Not just differences between the treatments. So those are that's a common factor for observational studies.

177
00:18:04,130 --> 00:18:09,680
You can't really attribute causation to any associations that you might find.

178
00:18:11,150 --> 00:18:17,360
So the next question that comes out of this is why are children dying of severe malaria in spite of modern treatment?

179
00:18:19,010 --> 00:18:23,960
So in order to try and answer this question, we decided to go to the hospital,

180
00:18:24,500 --> 00:18:30,260
and this is the Castle Hospital and the regional capital of this area of Mali.

181
00:18:30,650 --> 00:18:38,870
And if we zoom in on these panels, which you see when you come into the hospital, you'll see that the exit so means exits.

182
00:18:38,870 --> 00:18:48,410
It's a francophone country and the exit sign points to an advert for the cemetery, which is not a bad summary of what happens in the hospital,

183
00:18:49,070 --> 00:18:53,959
because an audit in 2002, which was a couple of years before we started working there,

184
00:18:53,960 --> 00:18:59,510
found that the inpatient mortality was 24.3% in the paediatric ward.

185
00:18:59,510 --> 00:19:03,620
That means if you have a child that goes in, that will there's a one in four chance they're going to die.

186
00:19:04,070 --> 00:19:14,090
That's pretty bad. And 42% of those deaths were due to malaria, which was the number one cause of death.

187
00:19:15,500 --> 00:19:22,910
Okay. So how to unpick this? Well, the first thing we decided to do was a case control study.

188
00:19:23,780 --> 00:19:32,960
So we decided to look at 50 children who died and 50 survivors, and they were selected at random and the year 2005.

189
00:19:33,290 --> 00:19:38,090
And we extracted the data retrospectively from their medical records and tried to compare

190
00:19:38,360 --> 00:19:43,760
the key characteristics between the two groups to see if there was any obvious difference.

191
00:19:44,750 --> 00:19:50,420
So this is a shortened version of the table from that study.

192
00:19:50,990 --> 00:19:56,300
There wasn't any difference in the age. There wasn't a significant difference in the sex.

193
00:19:57,350 --> 00:20:03,560
There was a significant difference on the type of malaria. So if you had severe anaemia, you were less likely to die.

194
00:20:04,040 --> 00:20:09,110
If you had neurological, that means coma and convulsions, you were much more likely to die.

195
00:20:10,130 --> 00:20:15,290
Duration of illness didn't seem to make much difference, so it's not necessarily that they're presenting late.

196
00:20:15,740 --> 00:20:20,380
If anything, the ones who survived perhaps have have been ill slightly longer.

197
00:20:20,390 --> 00:20:24,680
That's not significance. Length of hospital stay was very significant.

198
00:20:24,680 --> 00:20:28,970
But then that's sort of obvious, isn't it? Because if you die, you're not going to be in hospital for so long.

199
00:20:29,600 --> 00:20:33,470
And most of those many of those who died, died within 24 hours.

200
00:20:34,310 --> 00:20:38,510
And the treatment they'd received before coming to hospital, there was no significant difference.

201
00:20:38,510 --> 00:20:41,060
And again, interestingly, although it's not significant,

202
00:20:41,520 --> 00:20:48,260
slightly more of the survivors appeared to have used traditional medicine, which is a bit odd and counter-intuitive.

203
00:20:49,400 --> 00:20:53,090
Anyway, it is what it is. It doesn't give you any explanations.

204
00:20:53,090 --> 00:20:56,960
It just shows you some interesting associations.

205
00:20:57,980 --> 00:21:03,770
And the only ones that really come out are things that are sort of blindingly obvious, and I guess we knew already.

206
00:21:04,580 --> 00:21:08,570
But it's interesting that this wasn't significant because we thought it might have been.

207
00:21:10,400 --> 00:21:14,360
So advantages and disadvantages of the case. Control methods.

208
00:21:15,340 --> 00:21:18,770
Yeah. And malaria with a clinic.

209
00:21:20,610 --> 00:21:28,980
Was it? So this thing in the hospital, one would have hoped that they'd had blood tests to confirm the diagnosis.

210
00:21:29,580 --> 00:21:41,340
Having said that, when we went to the hospital lab to look at the quality of the of the lab tests, that could be questionable.

211
00:21:42,330 --> 00:21:45,600
I mean, the microscope was not in the best state.

212
00:21:45,600 --> 00:21:49,290
And I mean, obviously, this was retrospective.

213
00:21:49,290 --> 00:21:52,589
So we couldn't we couldn't check any of the any of the results.

214
00:21:52,590 --> 00:21:55,890
But the following year, we did another study, which I'm about to tell you about.

215
00:21:56,430 --> 00:22:01,140
And yeah, let's say that the hospital lab wasn't 100% accurate.

216
00:22:01,530 --> 00:22:08,190
Hmm. But it's probably true that most of them probably did have malaria because there's a lot of malaria in that.

217
00:22:08,470 --> 00:22:14,160
But. But there could have been some false positives included in that.

218
00:22:17,160 --> 00:22:24,660
Yeah. So does anyone have any thoughts on what are the advantages or disadvantages of using a case control design?

219
00:22:26,390 --> 00:22:30,620
In this case or in any case, why do you think we bother doing it?

220
00:22:30,620 --> 00:22:35,610
Because it didn't actually give us anything very useful in the end. Well, we did it.

221
00:22:35,630 --> 00:22:40,640
And I guess the reason a lot of people do case control is it's a quick and easy thing to do is quick and dirty.

222
00:22:40,640 --> 00:22:46,910
You can the data's there. You can easily extract it. You know, it didn't take much time or effort or money to do it.

223
00:22:48,260 --> 00:22:51,379
In this case, I'm not saying all case control studies are like that, but in this case,

224
00:22:51,380 --> 00:22:54,560
that was certainly the reason, and we didn't end up publishing it, to be honest.

225
00:22:54,650 --> 00:22:56,930
It was quite small numbers. We could have done bigger numbers.

226
00:22:56,930 --> 00:23:03,050
But the other main advantage of it is that you can look at risk factors for a rare outcome such as death,

227
00:23:03,650 --> 00:23:07,630
which in this context actually unfortunately wasn't that rare.

228
00:23:08,510 --> 00:23:15,860
But in theory, if you have got a rare outcome, it does enable you to do things that you otherwise might not be able to do.

229
00:23:16,640 --> 00:23:20,360
The disadvantages specifically, particularly for this study,

230
00:23:20,360 --> 00:23:24,499
we were limited to the information contained in the medical records because it was retrospective.

231
00:23:24,500 --> 00:23:28,969
But actually I think a lot of case control studies are probably like that. You're using someone else's data.

232
00:23:28,970 --> 00:23:32,150
You can't always define what you actually want in there.

233
00:23:32,900 --> 00:23:38,690
You can't derive figures on prevalence of different risk factors because you haven't got the

234
00:23:38,690 --> 00:23:45,259
whole population and you can't do any interesting analysis like regression or survival analysis.

235
00:23:45,260 --> 00:23:47,480
So you're fairly limited as to what you can do.

236
00:23:47,660 --> 00:23:52,640
But it's a quick and dirty way of getting an idea of what, you know, might be some interesting things to look at.

237
00:23:53,300 --> 00:23:57,530
So the next thing we decided to do was a prospective cohort study.

238
00:23:58,010 --> 00:24:04,250
So in this we decided to follow up. Every child admitted with severe malaria over a two month period in this hospital.

239
00:24:05,450 --> 00:24:11,719
And we would because it's prospective, we were able to interview the parents at the time when they came into the hospital,

240
00:24:11,720 --> 00:24:18,890
we were able to ask them questions about all the things we were interested in and we were able to follow them up to death or discharge.

241
00:24:20,870 --> 00:24:30,829
So sorry, this is rather small, but this is basically the results of regression analysis and it's cut a long story short,

242
00:24:30,830 --> 00:24:39,980
it confirmed what the case control had shown, that there was no significant difference as to what treatment they taken before coming to hospital.

243
00:24:42,320 --> 00:24:45,799
It did confirm that the type of malaria made a big difference.

244
00:24:45,800 --> 00:24:51,860
So if you were in coma six times more likely to die, respiratory distress was significant and blood sugar level.

245
00:24:51,890 --> 00:24:55,040
Having a low blood sugar level was very significant as well.

246
00:24:56,150 --> 00:25:01,910
But interestingly and completely unexpectedly, one of the risk factors was being female.

247
00:25:02,000 --> 00:25:05,270
So girls were twice as likely to die as boys.

248
00:25:05,810 --> 00:25:12,800
And again, there's no explanation for that. But that is what we found in this particular cohort.

249
00:25:13,880 --> 00:25:22,459
So we did a survival analysis as well, and the only thing that came out, we didn't include in this the type of malaria,

250
00:25:22,460 --> 00:25:27,560
but we were looking particularly interested in pre-hospital treatment and pre-hospital risk factors.

251
00:25:27,980 --> 00:25:32,630
And the only thing that came out as being highly significant was being female.

252
00:25:33,020 --> 00:25:36,470
So girls were much more likely to die than boys.

253
00:25:39,020 --> 00:25:45,140
And we also did a regression analysis on glucose blood sugar at admission,

254
00:25:45,470 --> 00:25:52,040
and that we found there was an almost linear relationship between blood sugar on admission and death,

255
00:25:52,640 --> 00:26:00,230
which is quite interesting because W.H.O. guidelines and other guidelines sort of assume that there's a CUT-OFF at 4.4.

256
00:26:00,230 --> 00:26:03,559
But this actually showed that, you know, actually the higher your blood sugar,

257
00:26:03,560 --> 00:26:08,480
the more greater chance you have of surviving, which is quite interesting.

258
00:26:10,340 --> 00:26:13,010
The one thing we couldn't do, unfortunately,

259
00:26:13,010 --> 00:26:20,870
which we which is the thing that we really wanted to do is a before and after comparison of our treatment package.

260
00:26:20,870 --> 00:26:25,549
Because together with this cohort study, when we were admitting the children,

261
00:26:25,550 --> 00:26:32,720
we were implementing an improved protocol because we suspected that the quality of care was probably not as good as it could be.

262
00:26:33,260 --> 00:26:37,520
And therefore we wrote a protocol on improved quality of care.

263
00:26:37,880 --> 00:26:41,900
We provided free treatment for the children that were included in the study,

264
00:26:42,170 --> 00:26:50,450
and we were hoping that we might be able to show a reduction in in-hospital case fatality compared to previous years.

265
00:26:51,440 --> 00:26:59,840
And we assumed stupidly, very stupidly with hindsight that the hospital we could use the hospital statistics on mortality.

266
00:27:00,110 --> 00:27:07,220
That was a bad idea because it turns out that the hospital statistics on deaths were completely inaccurate.

267
00:27:07,610 --> 00:27:15,259
And if a child died 5 minutes after coming into hospital or before receiving any treatments,

268
00:27:15,260 --> 00:27:25,700
even if that that was a few hours after coming into hospital or if it was a death that the junior doctor would rather not be criticised about by his.

269
00:27:25,930 --> 00:27:35,830
Sultan's. The file was conveniently ripped up and disappeared into the dustbin and did not enter into the hospital statistics prior to our study.

270
00:27:36,430 --> 00:27:39,579
So the red line is the mortality rate.

271
00:27:39,580 --> 00:27:48,520
In the year when we were doing the cohort study, which we started in July and August and the previous years, are here.

272
00:27:48,730 --> 00:27:55,780
But actually the before data is probably a massive underestimate of what the mortality really was.

273
00:27:56,050 --> 00:27:58,570
If they had been using the same measure that we were using,

274
00:27:58,570 --> 00:28:04,870
which was every child who arrived breathing and alive in the paediatric departments, how many of them died?

275
00:28:05,650 --> 00:28:10,210
So this unfortunately we weren't able to publish or use.

276
00:28:10,960 --> 00:28:14,620
So there's a lesson in that. Okay.

277
00:28:14,620 --> 00:28:23,620
So the next question then, which came out is going back to that treatment outcomes study,

278
00:28:23,980 --> 00:28:31,750
we've sort of looked at the modern medicine side and we've worked out that probably although it was difficult to show it using the cohort study,

279
00:28:31,750 --> 00:28:37,299
but probably the quality of care was not as good as it should have been in the hospital,

280
00:28:37,300 --> 00:28:42,610
which might explain parts of partly why children die in spite of receiving more than treatments.

281
00:28:43,280 --> 00:28:50,260
The second question, which we were interested in, is, are any of the traditional medicines actually associated with good outcomes?

282
00:28:50,680 --> 00:28:54,310
And obviously that group of traditional medicines is large.

283
00:28:54,760 --> 00:28:55,990
So how do you unpick that?

284
00:28:56,770 --> 00:29:04,900
Well, let's go back to the retrospective treatment outcomes study and think for a minute about how we might look at the data.

285
00:29:05,980 --> 00:29:11,230
So it's essentially a population survey looking at what treatments people have taken.

286
00:29:12,250 --> 00:29:15,909
And in this hypothetical example, lots of people have taken treatment.

287
00:29:15,910 --> 00:29:19,900
Three, a few have taken treatment one, and not many have taken treatment two.

288
00:29:20,920 --> 00:29:26,739
And people who are interested in traditional medicine tend to do what they call an ethno botanical survey,

289
00:29:26,740 --> 00:29:31,120
where you go and interview people and find out what treatments they take for this illness or that illness.

290
00:29:31,420 --> 00:29:37,510
And traditionally, they think, well, you know, the treatment that's taken by most people must be the most interesting one because most people use it.

291
00:29:38,440 --> 00:29:41,020
But is that assumption correct? Maybe not.

292
00:29:41,350 --> 00:29:49,600
Maybe patients have different outcomes, you know, that aren't correlated with how many people have taken them.

293
00:29:49,600 --> 00:29:54,940
So in this hypothetical example, treatment three is taken by lots of people, but most of them don't get better.

294
00:29:55,810 --> 00:30:00,460
Treatment two is taken by a small number of patients, but most of them actually do get better.

295
00:30:00,880 --> 00:30:03,940
So that's the treatment that we want to find, not this one.

296
00:30:05,770 --> 00:30:10,299
So we analysed the results from the Retrospective Treatment Outcome study in this way,

297
00:30:10,300 --> 00:30:15,130
and this is a summary of the top three traditional medicines, herbal medicines.

298
00:30:15,310 --> 00:30:21,670
The full table included 66 or 66 different plants being used in this area, mainly by different people.

299
00:30:22,540 --> 00:30:29,950
And the top one, everyone who took it, only 30 people took it, but everyone who took it said they got better.

300
00:30:30,880 --> 00:30:36,310
So we thought that looks quite interesting. And that was significantly better than the next ones down.

301
00:30:38,440 --> 00:30:46,360
So it was tested. Then we got it tested in the laboratory, in the test tube to see if it killed malaria parasites.

302
00:30:46,540 --> 00:30:52,150
And it did. That was one of four plants that had a good inhibitory concentration.

303
00:30:53,080 --> 00:31:00,160
And as I said, it was associated with good clinical recovery. So we thought, well, maybe this is interesting, maybe we should look into this further.

304
00:31:00,170 --> 00:31:09,820
So what should we do next? You can't really do a clinical trial in something that's, you know, that has you've only got retrospective data wrong.

305
00:31:10,780 --> 00:31:20,170
So we thought, why don't we actually observe and see what happens to those patients who take only Mexicana Tea in home based management of malaria?

306
00:31:20,590 --> 00:31:24,610
Is it safe? If we were to do a trial, what dose should be used?

307
00:31:27,010 --> 00:31:31,450
So we decided to do an observational study.

308
00:31:31,960 --> 00:31:38,230
I've put stroke phase two because it's almost like a face to face to uncontrolled clinical trial.

309
00:31:39,400 --> 00:31:42,970
You'll see one hesitating between the two in a minute.

310
00:31:43,570 --> 00:31:52,660
So our original idea was to simply observe patients for taking this, but to do it prospectively and to confirm the diagnosis of malaria.

311
00:31:52,690 --> 00:31:58,120
Because obviously, as I mentioned, the retrospective one, we're not sure if they did have malaria because it was retrospective.

312
00:31:59,470 --> 00:32:03,700
So we had to find a place where people were using this traditional medicine.

313
00:32:04,180 --> 00:32:10,930
And the place we came upon was this village called McDougall, which was about 40 kilometres from the nearest health centre.

314
00:32:11,410 --> 00:32:21,550
There's no electricity or running water. This is the road which during the malaria season gets blocked by rivers that appear out of nowhere.

315
00:32:22,120 --> 00:32:25,390
So as you can imagine, access to the health centre is not very.

316
00:32:25,650 --> 00:32:30,720
See, and most people don't bother going to the health centre unless they're at death's door.

317
00:32:32,640 --> 00:32:40,170
The village chief was a traditional healer who had learned about the use of this plant from his own grandfather.

318
00:32:40,200 --> 00:32:52,439
So there was three generations of experience of using this plant in his village and he welcomed us with open arms and gave us this little house,

319
00:32:52,440 --> 00:32:56,760
lent us this little house to be our laboratory and study centre.

320
00:32:57,720 --> 00:33:00,030
So with some solar panels on the roof,

321
00:33:00,030 --> 00:33:07,889
we were able to run a microscope and a little centrifuge so we could do haematocrit of white blood, white blood cell counts.

322
00:33:07,890 --> 00:33:14,460
We could do EKGs. And the basic idea was to observe patients who came to the traditional healer

323
00:33:16,050 --> 00:33:20,340
and then do some blood tests to confirm the diagnosis and follow them up.

324
00:33:21,900 --> 00:33:27,420
So we included everyone who the traditional healer had decided to treat with his herbal medicine,

325
00:33:27,420 --> 00:33:31,540
who we agreed had a diagnosis of malaria, where that was confirmed on a blood test.

326
00:33:31,560 --> 00:33:35,140
There was no other obvious cause of fever. They didn't have severe malaria.

327
00:33:35,160 --> 00:33:39,090
Those we sent straight to the hospital and where the parents gave their consent.

328
00:33:40,110 --> 00:33:48,270
And we had a pharmacy students who helped a lot with recruiting the patients, and he observed the doses being given by the traditional healer.

329
00:33:49,410 --> 00:34:00,500
So when we followed people up to date 28 by doing consultations, clinical examination, blood tests, etc., now this is where it gets interesting.

330
00:34:00,510 --> 00:34:07,919
We have plans to have different groups of doses, but after the first group of patients,

331
00:34:07,920 --> 00:34:12,480
we were starting to be rather disappointed because most of them weren't getting better.

332
00:34:13,470 --> 00:34:21,960
But we'd noticed that the traditional healer was giving them one glass once a day for three days of his decoction.

333
00:34:22,620 --> 00:34:27,750
And so we decided to have a meeting with him when we asked him, Is this really what you would normally do if we weren't here?

334
00:34:28,410 --> 00:34:35,220
And he said, Well, no, of course not. If you weren't here, I would just give them the stuff and tell them to go home and drink as much as they can.

335
00:34:35,790 --> 00:34:42,110
And so and we said, so why are you giving them one class a day for three days?

336
00:34:42,120 --> 00:34:47,610
And he said, Well, that's much more scientific, isn't it? That's how you give chloroquine one tablet a day for three days.

337
00:34:48,000 --> 00:34:57,090
Okay. So we decided then to try and standardise what he meant by go home and drink as much as you can.

338
00:34:57,600 --> 00:35:04,680
So we decided that the second group, which we called Group B, to give them one glass twice a day for seven days.

339
00:35:05,310 --> 00:35:11,490
And then Group C got one glass four times a day for the first four days, and then twice a day, up to seven days.

340
00:35:13,410 --> 00:35:18,660
And they weren't randomised, but the characteristics were fairly similar between the groups.

341
00:35:18,660 --> 00:35:28,470
The majority of patients were children, equal proportions of girls and boys, and there was a big difference in the percentage who got better.

342
00:35:28,560 --> 00:35:31,770
So the first group, only 39% got better.

343
00:35:32,130 --> 00:35:37,890
The second group, 72 and a half percent got better. And the third group, interestingly, was no different.

344
00:35:38,220 --> 00:35:44,140
So there was a big difference, a significant difference between these two groups, which was quite interesting.

345
00:35:44,160 --> 00:35:50,130
And the parasite counts came down almost to zero, but not completely to zero in most of the patients.

346
00:35:51,840 --> 00:35:57,329
There were a few side effects, nothing very serious, except at the highest dose.

347
00:35:57,330 --> 00:36:07,319
A couple of patients had EKG changes. So we concluded from this study that it appeared to be appeared to be effective in a dose escalating study.

348
00:36:07,320 --> 00:36:12,900
So it it sort of morphed from an observational study into a dose escalating study.

349
00:36:14,400 --> 00:36:20,460
The optimal dosage was twice a day for one week, and that dosage appeared to be safe and well tolerated.

350
00:36:22,200 --> 00:36:25,890
So the advantages and disadvantages of this sort of study,

351
00:36:26,400 --> 00:36:32,460
it's quite useful for establishing the best dose and it's useful for establishing a safety profile.

352
00:36:32,910 --> 00:36:39,150
But you can't really be certain about the efficacy because we don't have a control group and in a setting,

353
00:36:39,180 --> 00:36:46,320
especially where there's lots of immunity to malaria, it's possible that all these patients might have got better anyway.

354
00:36:47,280 --> 00:36:50,880
We don't know. But that's, you know, so we can't be sure that it's effective.

355
00:36:50,930 --> 00:36:58,440
I mean, I guess what's in our favour is the fact that the first group who took an undue dose, actually most of them didn't get better.

356
00:37:00,600 --> 00:37:07,380
So with hindsight, this is probably what we were trying to do and perhaps what we would do in a more structured way another time,

357
00:37:07,770 --> 00:37:12,120
which is that if you've got a range of doses traditionally recommended,

358
00:37:12,120 --> 00:37:17,609
if you're looking at a traditional medicine, you might start with the lowest dose and then look at what the clinical results are.

359
00:37:17,610 --> 00:37:26,910
If it's good and it's safe, then that's fine. If it's if it's good, but it's not safe, then you would decrease the dose.

360
00:37:27,090 --> 00:37:32,520
If it's not good, but it's safe, then you can increase the dose and you keep going around until you hit one of these two.

361
00:37:34,530 --> 00:37:43,170
So at this stage, people in Mali started to get quite interested and this chap was the head of the malaria control program at the time.

362
00:37:43,190 --> 00:37:48,920
He actually came to visit the village and look at the herbal medicine, etc. and we had a meeting with the villagers.

363
00:37:49,440 --> 00:37:52,649
So I asked them, what what do you think we should do next?

364
00:37:52,650 --> 00:37:56,850
What we do best way forward. And everyone agreed they wanted to carry on with the research.

365
00:37:57,330 --> 00:38:03,930
And the question we really wanted to ask is, is this tea effective in home based management of malaria?

366
00:38:05,340 --> 00:38:09,930
So in order to answer that question, we decided we needed to do an AZT.

367
00:38:11,430 --> 00:38:16,740
And this is our Pico question. It looks from the board that you've been doing Pico today.

368
00:38:17,090 --> 00:38:20,780
Is this you? Yeah. Okay. So the of.

369
00:38:20,870 --> 00:38:25,880
Patients with those with presumed uncomplicated malaria of all ages,

370
00:38:26,090 --> 00:38:31,639
because that would be the group for home based management of malaria presumed because there are no blood tests in the village.

371
00:38:31,640 --> 00:38:35,930
So we include everyone, even those who may not actually have malaria.

372
00:38:37,580 --> 00:38:43,310
The intervention was the argument in Mexico on a decoction twice a day for 7 to 14 days.

373
00:38:44,030 --> 00:38:50,810
The comparison was assassinated mode you couldn't, which at the at the time was the standard treatment for malaria.

374
00:38:51,680 --> 00:38:58,250
And the outcome measures that we were interested in were clinical recovery, which we defined as no need for a second treatment.

375
00:38:58,610 --> 00:39:02,720
Incidence of new clinical episodes of malaria and incidence of severe malaria.

376
00:39:04,880 --> 00:39:13,190
There were, of course, some ethical issues. This is an unproven herbal remedy, although it is widely used and widely available.

377
00:39:14,210 --> 00:39:21,980
Our testing etymology coin is known to be effective and was recommended by W.H.O. but wasn't yet widely available in this area.

378
00:39:22,400 --> 00:39:31,770
Malaria is potentially fatal, so if you were on the Ethics Committee, would you give ethical approval to this trial in the UK?

379
00:39:31,790 --> 00:39:33,650
The answer is probably, almost certainly no.

380
00:39:34,670 --> 00:39:42,320
But in Mali, if the Ethics Committee said yes because they thought this was an important question that needed to be answered.

381
00:39:44,660 --> 00:39:54,500
So they gave us approval to go ahead with this study. Sample size, we I don't know if you've talked about this yet,

382
00:39:54,500 --> 00:40:00,080
but we were assuming that at least 85% of patients on the modern medicine would not need three treatments.

383
00:40:01,490 --> 00:40:07,670
And we hypothesise that the herbal medicine is noninferior to the artemisinin combination therapy,

384
00:40:07,670 --> 00:40:12,260
which is the standard treatment, and we'd need to detect a difference of greater than 15%.

385
00:40:13,350 --> 00:40:20,630
And so we plugged the numbers into the standard sample and we came up with a sample size of 119 per group.

386
00:40:21,440 --> 00:40:26,250
So we randomised patients in a 2 to 1 ratio to the herbal medicine versus the ACTU.

387
00:40:26,270 --> 00:40:31,639
The reason for that is we felt there wasn't much information about the herbal medicine at all,

388
00:40:31,640 --> 00:40:34,370
whereas there was already quite a lot of information about the act.

389
00:40:34,520 --> 00:40:40,129
So that's why we wanted to put twice as many patients into the herbal group and we stratified it into

390
00:40:40,130 --> 00:40:45,800
different age groups because it's known that children under five are more likely to get severe malaria.

391
00:40:47,390 --> 00:40:53,480
We discussed it with the community who were very happy and interested to go ahead.

392
00:40:54,140 --> 00:41:00,980
So the study worked like this. Patients first came to see a village health worker who happened to be the traditional healers son.

393
00:41:01,520 --> 00:41:04,700
He sort of delegated his role to his his son.

394
00:41:06,350 --> 00:41:12,110
Anyone who he thought had fever was referred to the study team who were asked for consent.

395
00:41:12,740 --> 00:41:16,790
And then we examined the patients, took the history, took blood tests, etc.,

396
00:41:16,790 --> 00:41:22,850
and then they were randomised to herbal treatments or AZT and we followed them up for three months.

397
00:41:24,230 --> 00:41:30,350
So out of 313 patients who came to see the healer with presumed malaria,

398
00:41:30,710 --> 00:41:38,360
we only excluded 12 most of them because they had severe malaria or we live too far away, couldn't come back for follow up.

399
00:41:38,930 --> 00:41:43,909
So 301 were included, 87% were positive for the malaria parasites.

400
00:41:43,910 --> 00:41:54,020
So that's good. That means that most of them did have malaria and about 200 were randomised to the herbal medicine and about 100 to the active group.

401
00:41:55,790 --> 00:42:04,729
And this is what happened. So the herbal medicine was slightly inferior to the modern one.

402
00:42:04,730 --> 00:42:11,270
89% got better compared to 95% and 12.8% had a recurrence.

403
00:42:11,660 --> 00:42:19,020
By day 28, compared to 9.9%. But, you know, it's they are fairly small differences.

404
00:42:20,760 --> 00:42:27,240
And over the first 28 days, we had no deaths in either group, no severe malaria in those over five.

405
00:42:27,420 --> 00:42:32,520
And in the under-fives, it was broadly the similar incidence of severe malaria in both groups.

406
00:42:32,520 --> 00:42:40,470
No one had coma convulsions and there was slightly more adverse effects in the modern medicine group than in the herbal medicine group.

407
00:42:42,090 --> 00:42:48,719
And then we followed them up to three months to look for severe malaria, and there was basically no difference between the groups,

408
00:42:48,720 --> 00:42:54,380
although the study wasn't powered for this outcome because it's quite a rare outcome.

409
00:42:55,770 --> 00:43:00,450
But the incidence is less than one would expect with no treatments, with no treatment.

410
00:43:00,450 --> 00:43:05,280
You would expect in this population maybe 11% of severe malaria.

411
00:43:07,440 --> 00:43:09,540
And this is quite an interesting graph.

412
00:43:10,830 --> 00:43:21,690
So W.H.O., in its standard malaria protocols, looks for parasite clearance as part of the main outcome measure, which we explicitly didn't include.

413
00:43:21,720 --> 00:43:26,310
We were interested in whether people got better clinically and whether they got severe malaria.

414
00:43:26,970 --> 00:43:33,960
So these two lines show the percentage of patients who have any malaria parasites in their blood over the course of follow up.

415
00:43:34,290 --> 00:43:37,019
And you can see there's a massive difference between the herbal medicine,

416
00:43:37,020 --> 00:43:43,829
where most of the patients still had some malaria parasites in their blood and the artesunate,

417
00:43:43,830 --> 00:43:49,770
where most of the patients cleared the parasites, although they gradually started creeping back over the course of follow up.

418
00:43:50,340 --> 00:43:58,800
So if you were to use that, if we had used that as our primary outcome measure, the herbal medicine would have looked way worse than the modern one.

419
00:43:59,280 --> 00:44:07,500
When you look at the clinical outcomes, new clinical episodes, there were more at day 28 in the Herbal Medicine Group.

420
00:44:08,190 --> 00:44:11,549
We know that, but it's the second and the third month.

421
00:44:11,550 --> 00:44:20,459
There was no difference between the groups. So it looks like having parasites in your blood is not as important as people might think,

422
00:44:20,460 --> 00:44:23,730
especially in this sort of population where there are high levels of immunity.

423
00:44:24,030 --> 00:44:27,210
And what perhaps really matters is the clinical outcomes.

424
00:44:29,160 --> 00:44:36,299
So if you look at the cost of the two strategies, using the herbal medicine as first line with a conventional medicine as a backup.

425
00:44:36,300 --> 00:44:41,790
Second line is obviously a lot cheaper than using the modern medicine as first line for everyone.

426
00:44:43,140 --> 00:44:51,300
So it could be a useful first line home based treatment for uncomplicated malaria, especially in patients aged over five in this area.

427
00:44:54,730 --> 00:45:03,250
But yeah, I mean, it's a strategy for maybe sparing modern medicines where they're not available or

428
00:45:03,250 --> 00:45:07,329
if one has to give recommendations to people out of your 66 herbal medicines,

429
00:45:07,330 --> 00:45:10,420
which one do you use? This one is probably better than nothing.

430
00:45:12,250 --> 00:45:18,520
So reflecting back on the advantages and disadvantages of doing an r, c, t with the advantages,

431
00:45:18,520 --> 00:45:23,049
obviously it's the gold standard for establishing efficacy disadvantages.

432
00:45:23,050 --> 00:45:30,180
It's quite expensive and complicated to do. It's not very efficient for developing or optimising an intervention you couldn't

433
00:45:30,190 --> 00:45:35,049
really do in our seats on each different dosage regimen or each different preparation.

434
00:45:35,050 --> 00:45:45,640
You have to be quite fairly convinced before you start that you've got the best possible thing to trial and it's not always possible to do an c t.

435
00:45:46,090 --> 00:45:49,150
I'm going to give you an example of that in a minute.

436
00:45:50,380 --> 00:46:00,340
But just for a moment, reflecting about drug development, does anyone know how long it takes to develop a conventional, modern medicine?

437
00:46:01,930 --> 00:46:08,110
Starting from the chemistry and whips. So after skin, you know it's starting from the chemistry and all of that.

438
00:46:08,110 --> 00:46:19,870
It takes takes 15 years on average apparently and you know how much it costs be well the figure given in this particular article is $800 million.

439
00:46:20,680 --> 00:46:22,959
And unfortunately, most of the time,

440
00:46:22,960 --> 00:46:29,470
the end product is often unaffordable and unavailable to the poor people who really need it, especially in the case of malaria.

441
00:46:29,470 --> 00:46:35,080
So actually, there are no drug companies investing their own money in developing new antimalarials all the new drug development

442
00:46:35,080 --> 00:46:44,920
is being publicly funded and in comparison to that or scheme of finding a new antimalarial took six years,

443
00:46:44,920 --> 00:46:50,950
starting with this retrospective treatment outcome study, then the dose escalating study than the randomised controlled trial.

444
00:46:51,100 --> 00:46:56,740
We sort of did it backwards, we called it reverse pharmacology and at the end we started to look for active compounds,

445
00:46:56,890 --> 00:46:59,200
not so much because we wanted to purify them,

446
00:46:59,200 --> 00:47:07,210
but because we wanted to look at a tool for quality control, etc. But that's a whole other story that I haven't got time to go into now.

447
00:47:08,200 --> 00:47:13,360
It cost about €0.4 million and the end product is easily affordable and available.

448
00:47:14,170 --> 00:47:18,700
So it may not be such a bad way of doing things.

449
00:47:20,470 --> 00:47:24,820
Going back to the confidential inquiry, which I mentioned at the beginning,

450
00:47:26,080 --> 00:47:34,000
we did do a before and after study on that because we couldn't do a randomised controlled trial with the resources available

451
00:47:34,810 --> 00:47:44,139
and we found that compared to 3.9% reduction in under-five mortality at the national level and most of our study sites,

452
00:47:44,140 --> 00:47:50,799
there was a much greater mortality reduction except in one of them went up slightly.

453
00:47:50,800 --> 00:47:56,830
But overall the under-five mortality reduced by 18% across the study sites.

454
00:47:57,850 --> 00:48:02,140
Now this is a before and after study. It's not a randomised controlled trial.

455
00:48:02,470 --> 00:48:09,100
So one might argue or what we could be criticised that maybe this reduction was due to other things going on.

456
00:48:09,520 --> 00:48:13,149
Maybe there was some other reason. And it's true that in two of our study sites,

457
00:48:13,150 --> 00:48:20,890
these two we were aware of another NGO that had come along halfway through the study and started giving free treatment for malaria and malnutrition.

458
00:48:21,550 --> 00:48:27,610
And the other study, science. As far as we're aware, there wasn't anything else going on that could account for this reduction.

459
00:48:28,120 --> 00:48:35,170
But if we really wanted to be sure about whether this is an effective thing to do, we would need to do a cluster randomised trial.

460
00:48:36,010 --> 00:48:41,950
But when you start doing the power calculations and the budgeting it becomes very big and very expensive.

461
00:48:42,310 --> 00:48:50,800
I think we worked out we would need to have at least 30 subdistricts and when we started costing it it was millions of pounds.

462
00:48:51,100 --> 00:48:53,589
And we put we've put in several funding applications,

463
00:48:53,590 --> 00:48:59,590
but we haven't yet found a funder who's willing to spend that amount of money on doing that randomised controlled trial.

464
00:48:59,770 --> 00:49:04,690
So that's an example of where it hasn't been possible so far to do a randomised controlled trial.

465
00:49:06,340 --> 00:49:16,360
So take home messages, choose the most important question to research and then pick the best design to answer this question in a way that's feasible,

466
00:49:16,360 --> 00:49:20,410
unaffordable because you might not always be able to afford the best study design.

467
00:49:21,340 --> 00:49:27,010
Don't be frightened to modify existing designs or even invent new designs.

468
00:49:27,010 --> 00:49:32,350
If you can't find one that fits what you want to do and study designs at the bottom of the

469
00:49:32,350 --> 00:49:37,930
evidence hierarchy are actually essential for generating hypotheses and developing interventions.

470
00:49:38,110 --> 00:49:41,800
They might not be the final proof that people look at when they do meta analysis,

471
00:49:42,100 --> 00:49:47,830
but you can't really get to a randomised trial until you've done all those bits before or some of those bits before.

472
00:49:49,270 --> 00:49:51,610
So I think that's all I wanted to say.