1
00:00:01,320 --> 00:00:05,370
So could you just start by saying your name and what your current position is?

2
00:00:05,850 --> 00:00:12,000
Yes. So my name is Daniel Alhambra, and I am a professor of pharmaco and device epidemiology here at Oxford.

3
00:00:12,420 --> 00:00:17,160
I lead a group of people. It's a section called Health Data Sciences.

4
00:00:17,370 --> 00:00:23,490
So we basically use have data to produce knowledge and to, you know, try and improve human health.

5
00:00:23,600 --> 00:00:29,290
Hmm. Well, we're going to talk about that a lot more. But first, before that, I'd just like a little bit more about your background.

6
00:00:29,310 --> 00:00:37,650
So how did you first get interested in science and into this interesting branch of science, which, to be honest, I hadn't really come across before.

7
00:00:38,130 --> 00:00:42,120
Yes. So I am a medical doctor by training, so I trained as a as a GP.

8
00:00:42,120 --> 00:00:47,270
I practice as a GP for a while and then I kind of realised that, you know,

9
00:00:47,280 --> 00:00:53,280
there's a lot of uncertainty in how we manage disease in primary care and also I would say beyond primary care.

10
00:00:54,840 --> 00:01:03,830
So I started doing see into research methods, epidemiology studies, just to learn the basics of how we, we handle data to generate knowledge.

11
00:01:04,200 --> 00:01:07,620
So I did the first one was I did it in Spain where I'm from in Barcelona.

12
00:01:08,310 --> 00:01:12,220
And then after that I started my Ph.D. and as part of that, I,

13
00:01:12,240 --> 00:01:18,920
I had the fortune to spend a little bit more than a year here at Oxford as a visiting student.

14
00:01:18,950 --> 00:01:29,459
Right. And that was that was basically a project using existing data, routinely collected data, electronic medical records and the like,

15
00:01:29,460 --> 00:01:37,230
from Spain and from the UK to look at a number of musculoskeletal conditions like arthritis,

16
00:01:37,230 --> 00:01:40,410
basically, and also fractures and bone fragility problems.

17
00:01:41,760 --> 00:01:48,389
So yeah, basically, you know, through that kind of route, I ended up finding my passion with small to generate data,

18
00:01:48,390 --> 00:01:52,170
them to consume data as a clinician, thus in their day to day job.

19
00:01:52,560 --> 00:02:00,570
And although I have been practising as a as a physician until very recently, basically I believe that I am,

20
00:02:00,870 --> 00:02:04,650
you know, much better at doing research than not practising as a clinician.

21
00:02:05,070 --> 00:02:09,570
And that's that's where I'm headed. So this is I mean, this is where I think it's really interesting.

22
00:02:09,570 --> 00:02:17,970
This is so, you know, everybody knows about clinical trials where you, you take a treatment or a device and you test it on the population.

23
00:02:18,030 --> 00:02:23,309
Yeah, Prospectively, what you're doing is looking at people who are going through the health care

24
00:02:23,310 --> 00:02:29,190
system and using the data on that that shows what drugs they've had and so on,

25
00:02:29,190 --> 00:02:33,150
analysing that retrospectively to try and come up with answers.

26
00:02:33,300 --> 00:02:41,550
How do I can you give me some idea of the relative and importance of both those two ways of doing things?

27
00:02:42,000 --> 00:02:47,430
Yes. So I think unfortunately there's a little bit of a division in how people see this sometimes.

28
00:02:47,430 --> 00:02:51,360
You know, one is good data, the other is bad data. I think they are very complimentary.

29
00:02:51,360 --> 00:02:51,810
Actually.

30
00:02:52,170 --> 00:02:59,969
The the beauty of clinical trials, of course, is that you have very clean data on whatever outcome or whatever condition you want to look at.

31
00:02:59,970 --> 00:03:02,550
So you collect the information you need and you want.

32
00:03:03,330 --> 00:03:10,139
And additionally you also have the beauty that you are randomising people to receive a treatment or not,

33
00:03:10,140 --> 00:03:14,879
meaning that in the end the treated and untreated groups are comparable and they are basically

34
00:03:14,880 --> 00:03:18,690
the same on everything you want to compare them except the treatment they received.

35
00:03:18,690 --> 00:03:25,379
And therefore any differences in outcomes you know can be referred to the treatment they receive right in the data we use.

36
00:03:25,380 --> 00:03:33,600
That doesn't happen. We look at data as it comes from clinical practice, from routine practice that has limitations,

37
00:03:33,600 --> 00:03:37,589
including the fact that people tend to take treatment for a reason.

38
00:03:37,590 --> 00:03:42,270
So those people are not necessarily comparable to the people who live and receive the same treatment.

39
00:03:42,270 --> 00:03:44,400
That's what we usually call confounding.

40
00:03:45,360 --> 00:03:54,419
But also it has other limitations, including the fact that we use data that was collected in not always ideal conditions.

41
00:03:54,420 --> 00:04:02,670
Right? So, you know, you can imagine in GP practice when things are busy, potentially people will be called things more quickly and less accurately.

42
00:04:03,180 --> 00:04:08,250
So there's a lot of work that needs to be done on kind of curating that data before you can use it for research.

43
00:04:08,580 --> 00:04:13,480
Now, the advantages of this kind of data that I don't think are, you know,

44
00:04:13,590 --> 00:04:19,620
preferable to trial data that are complementary are first that this kind of data includes,

45
00:04:19,620 --> 00:04:26,370
of course, anyone who goes through the health care system here in the NHS, even more so because it's a kind of universal system, right?

46
00:04:27,090 --> 00:04:33,209
So you see basically people that have been excluded from trial participation for whatever reason,

47
00:04:33,210 --> 00:04:38,610
including co-morbidities sometimes, or because they have a shorter life span or because they are pregnant,

48
00:04:38,970 --> 00:04:44,010
pregnant or, or very young, all those people, when exposed to a treatment,

49
00:04:44,010 --> 00:04:51,149
will turn up in our data so we can collect information on what happened to them without exposing them to a risk.

50
00:04:51,150 --> 00:04:56,730
Unless that happened because it was considered necessary through routine clinical practice.

51
00:04:56,970 --> 00:05:00,380
That's one key advantage, I believe, to more care than. It is.

52
00:05:01,700 --> 00:05:07,219
First, The size of the data I can get hold of is, you know, an order of magnitude bigger.

53
00:05:07,220 --> 00:05:07,520
Right.

54
00:05:07,910 --> 00:05:15,890
And that, of course, you know, in some situations it matters, especially when we're looking at safety outcomes that tend to be, of course, rare.

55
00:05:16,700 --> 00:05:25,720
We do need big numbers. And I think, of course, I think a very relevant example is the COVID vaccine site, where we'll get we'll get to that.

56
00:05:26,180 --> 00:05:29,659
So I think that's a very good general principle. Yeah, that's a very good example.

57
00:05:29,660 --> 00:05:34,700
And then the final the final one that I think is not only the number of people,

58
00:05:34,700 --> 00:05:39,580
but also the possibility to follow them up for much longer than a trial.

59
00:05:39,590 --> 00:05:44,270
That's right. So a trial costs a lot of money. You follow people up for a certain time,

60
00:05:44,270 --> 00:05:50,749
but then you basically let go in this kind of data because the data is generated as people go through the health care system,

61
00:05:50,750 --> 00:05:58,670
you can follow people up for as long as they were registered in the GP practice or or for as long as they were having contact with the NHS.

62
00:05:58,700 --> 00:06:02,060
So that can be sometimes ten, 20 years of data available.

63
00:06:02,840 --> 00:06:09,950
Now you've mentioned you use both UK data and Spanish data and perhaps as elsewhere in the world now, and you also used the word cure rate earlier.

64
00:06:10,340 --> 00:06:18,020
So who, who creates these databases and how, how can you ensure consistency across different databases?

65
00:06:18,440 --> 00:06:22,309
That's a very good question. So yeah, indeed.

66
00:06:22,310 --> 00:06:31,850
So we have used data for many different topics, but very, very specially very specifically for COVID from many different parts of the world.

67
00:06:32,300 --> 00:06:42,440
Let's not talk about it just yet. So, yes, so that data is typically initially pseudonyms, you know, for for kind of governance reasons.

68
00:06:42,440 --> 00:06:48,829
The data is it has a fast kind of process that's done by the data you can call it.

69
00:06:48,830 --> 00:06:52,430
Then, though, they three words depending on the country, they use different terminology.

70
00:06:52,430 --> 00:06:56,180
Right. But it's basically the people who made the data available to researchers.

71
00:06:56,720 --> 00:06:59,760
And then these tend to be private companies or are they government based?

72
00:07:00,350 --> 00:07:10,550
So in the UK, in Europe, mostly government based, in the US, mostly private companies and then other parts of the world, it depends on the country.

73
00:07:10,820 --> 00:07:15,800
But I would say in Asia I use some Asian data and with the exception of China,

74
00:07:16,130 --> 00:07:22,100
most of the other countries would be selling or, you know, making the data available through private routes.

75
00:07:22,310 --> 00:07:29,140
But in Europe, it's typically government based institutions or trusted third parties, as they call them, right?

76
00:07:30,590 --> 00:07:32,270
Yeah. So so that's the first process.

77
00:07:32,270 --> 00:07:38,629
The second process is, of course, once you receive the data, you need to do all the checks to make sure that the data is fit for purpose,

78
00:07:38,630 --> 00:07:42,260
that you can use them, that it's got quality completeness and all that.

79
00:07:44,000 --> 00:07:52,969
Typically that will be done by my team if I am directly accessing the data and we do that for data that we've been using for many years,

80
00:07:52,970 --> 00:08:01,190
including for instance, data from Spain or from the UK. Now, the third step in the process that you mentioned already is this interoperability, right?

81
00:08:01,190 --> 00:08:05,540
So making data basically similar across the world.

82
00:08:05,540 --> 00:08:10,730
So you can do international collaboration when that is needed or when that is desirable.

83
00:08:12,410 --> 00:08:16,520
And that is something we've done a lot of work on recently. In the last three, five years.

84
00:08:16,910 --> 00:08:20,239
This was just pure chance or serendipity.

85
00:08:20,240 --> 00:08:28,790
I don't know. You know, just when the pandemic hit, we were already working on this topic of making data look the same across the globe,

86
00:08:29,150 --> 00:08:33,080
because all these data sets in a way contain similar information, right?

87
00:08:33,080 --> 00:08:36,350
It's going to be, you know, your social demographics, what conditions you have,

88
00:08:36,350 --> 00:08:42,170
what medicines you're taking, all that, but maybe recorded in different languages, in different systems.

89
00:08:42,430 --> 00:08:46,640
The format of the data is most likely going to be different as well.

90
00:08:46,940 --> 00:08:50,900
So there is this process of what we call mapping to our common data model,

91
00:08:50,930 --> 00:08:58,489
such a common strand structure or format, which then enables international collaboration without moving data.

92
00:08:58,490 --> 00:09:03,080
So I can analyse data from the UK that I have mapped to this common data model.

93
00:09:03,590 --> 00:09:12,409
In this case we use our common data up and then basically if I have a collaborator in, say, Spain or in South Korea that has data in the same format,

94
00:09:12,410 --> 00:09:19,399
I can share my code and that code should run in their data and then all they can or have to share is results.

95
00:09:19,400 --> 00:09:23,240
Really so aggregated data, we don't move patient level data anymore.

96
00:09:23,540 --> 00:09:31,910
And that has many advantages in terms of, you know, information, governance and security, because you keep the data where it is.

97
00:09:32,090 --> 00:09:38,960
You just you know, you have this version of the data that has been formatted in a very specific way and then we can analyse it,

98
00:09:39,290 --> 00:09:41,059
what we go in a federated manner.

99
00:09:41,060 --> 00:09:48,800
So we basically have someone writing the analytical code and then colleagues or collaborators running that code across the globe.

100
00:09:48,920 --> 00:09:58,190
Mm hmm. So you talk about size, the size of study, typically how many patients are represented in the kind of study you might do so.

101
00:09:59,560 --> 00:10:03,740
For the pandemic. We will be talking typically of tens of thousands of people.

102
00:10:03,760 --> 00:10:07,770
That was the usual. Nowadays, Yeah.

103
00:10:07,870 --> 00:10:10,990
Millions sometimes. Hmm. That's extraordinary.

104
00:10:11,350 --> 00:10:13,840
And before we get to code, we're going to get it in just a minute.

105
00:10:14,020 --> 00:10:18,400
But can you give me a couple of examples of the kind of question that you might have been asking?

106
00:10:18,470 --> 00:10:22,360
No, it is also not related to Katrina, Not least not yet.

107
00:10:22,780 --> 00:10:26,139
Yes. So let's go. I'll give you an example.

108
00:10:26,140 --> 00:10:28,630
Pre-pandemic and post-pandemic not related to COVID.

109
00:10:28,990 --> 00:10:36,219
So an example of something we were doing before the pandemic was looking at and this is NIH funded research that funds,

110
00:10:36,220 --> 00:10:40,060
I felt, a senior fellowship that I that I do a deal for myself.

111
00:10:40,660 --> 00:10:46,510
So we are looking at whether in very old and very elderly people who have lots of conditions,

112
00:10:46,840 --> 00:10:52,870
there's always this question on whether it's worth it's considering at least stopping some of the

113
00:10:52,870 --> 00:10:58,720
treatments they are taking because they are deemed to be unsafe and maybe not that useful for them.

114
00:10:58,960 --> 00:11:02,020
And an example of that could be statins, for instance.

115
00:11:02,020 --> 00:11:07,599
Right. So this is a question that is very tricky to investigate in a trial because you have these people

116
00:11:07,600 --> 00:11:11,620
who are taking the treatment and you have to randomise them to stop taking the treatment or not.

117
00:11:11,620 --> 00:11:15,999
And that's not something that, you know, is easy to to convince people to do.

118
00:11:16,000 --> 00:11:25,510
Right. So we we were looking into that specific question of what happens when you stop one such treatment, what we call preventative therapies.

119
00:11:25,600 --> 00:11:30,010
You know, and in the real world, some doctors might simply decide not to make that change.

120
00:11:30,790 --> 00:11:35,260
And ideally, you know, it's it's kind of an ideal scenario because there's so much uncertainty.

121
00:11:35,500 --> 00:11:39,210
That is it is to be expected in that way that those decisions are small and random.

122
00:11:39,220 --> 00:11:45,430
Right. If you had very clear knowledge that, let's say, stopping therapies causes harm or it doesn't,

123
00:11:45,460 --> 00:11:51,700
then you can expect that everybody would be doing the same. But when you have these situations where you don't have very good evidence,

124
00:11:51,970 --> 00:11:58,510
there's more randomness in those decisions and you can leverage that to do this kind of observational research.

125
00:11:59,020 --> 00:12:02,170
That's an example pre-pandemic post post-pandemic we are doing. And what was the outcome?

126
00:12:02,220 --> 00:12:09,790
Right. So that's something we we we are still working on right now because, yeah, we could talk about priorities, right?

127
00:12:09,790 --> 00:12:12,200
So, you know, that was deprioritized.

128
00:12:12,730 --> 00:12:19,990
It's looking like like if you stop these therapies, you need to be careful because at least for some time after stopping them,

129
00:12:19,990 --> 00:12:23,230
there's going to be an increased risk of the events you were trying to prevent.

130
00:12:23,260 --> 00:12:31,190
So studying, for instance, will be heart attacks, you know, But that's that's something we are writing up now as we speak now.

131
00:12:31,470 --> 00:12:35,670
A post-pandemic example would be a lot of what we're doing on cancer.

132
00:12:35,680 --> 00:12:37,209
So looking at, you know,

133
00:12:37,210 --> 00:12:47,680
different presentations of different types of cancer and the prognosis of cancer and then what does that look like before and after the pandemic.

134
00:12:47,680 --> 00:12:53,740
So that sort of observational study looking at prognosis and and presentation of cancer,

135
00:12:53,770 --> 00:12:57,880
different types of cancer in the population and across Europe in this case.

136
00:12:58,900 --> 00:13:00,219
And you mentioned risks earlier.

137
00:13:00,220 --> 00:13:09,100
So, you know, is that that that just these two main questions you're asking, does this treatment work, this struggle, this device work?

138
00:13:09,100 --> 00:13:15,480
And also, does this treatment or device do harm? Those are the two kinds of big questions like asking, was that it?

139
00:13:15,520 --> 00:13:19,690
I think this three types of research we do with this kind of big data,

140
00:13:19,870 --> 00:13:28,030
because data one indeed is looking at the effects of medical devices, procedures on the treatment medicines.

141
00:13:28,510 --> 00:13:32,049
And you can look at benefits and you can look at risks. Typically,

142
00:13:32,050 --> 00:13:37,660
we do more of the risk bit of the picture because that's what's harder to do with

143
00:13:37,660 --> 00:13:41,440
trial data because these are rare outcomes for the reasons we discussed earlier.

144
00:13:41,440 --> 00:13:47,380
Right. But but we have done also some work on benefits and that's published work.

145
00:13:47,920 --> 00:13:54,399
Now on on top of that, the other two types of studies that we do a lot of one is basic epidemiology.

146
00:13:54,400 --> 00:13:58,120
So characterising a condition, how often does this happen in a population?

147
00:13:58,660 --> 00:14:06,040
In the example I was mentioning, are we seeing, for instance, an increase in the incidence of particular cancer in younger people,

148
00:14:06,040 --> 00:14:13,510
which we will be worried about that sort of thing that we call characterisation or descriptive analysis.

149
00:14:13,870 --> 00:14:16,479
And the third type is what we call prediction.

150
00:14:16,480 --> 00:14:21,910
So this is where we take a group of people and say what happens to people when they get diagnosed with osteoarthritis?

151
00:14:21,910 --> 00:14:26,379
What are the people who are at the highest risk of having a knee replacement or one

152
00:14:26,380 --> 00:14:29,920
of the people who are at the highest risk of falling and having a hip fracture?

153
00:14:30,340 --> 00:14:34,450
So that would be prediction. Prediction work. Mm hmm.

154
00:14:34,510 --> 00:14:36,970
Very interesting. Right. We'll get to Kevin now.

155
00:14:37,210 --> 00:14:46,030
So can you remember where you were or under what circumstances you first heard that something was going on in in China?

156
00:14:46,240 --> 00:14:51,730
And how soon did you realise that that was going to fall into your area of interest?

157
00:14:52,330 --> 00:14:57,550
Yeah. So. So back in December 2019, we were hearing the news, right?

158
00:14:58,450 --> 00:15:03,620
It wasn't very. Solid data yet. So it was very hard to really figure out what was coming.

159
00:15:04,550 --> 00:15:12,500
But yeah, as an epidemiologist, I was kind of, you know, already paying attention to the to the information we were getting in January.

160
00:15:12,500 --> 00:15:20,270
It was already very obvious that this was coming. You know, we were seeing images in China, you know, lockdowns, and we knew that there were still,

161
00:15:20,390 --> 00:15:24,290
you know, planes flying people around the entire world pretty much.

162
00:15:24,290 --> 00:15:30,099
Right. And you were here in Oxford, but I forgot to pick you up on your career steps.

163
00:15:30,100 --> 00:15:34,790
So. No, no, no. I've been here continuously since 2012.

164
00:15:34,820 --> 00:15:39,940
Right. So, yeah, I guess between December and January that year, I still went to Spain for the Christmas May.

165
00:15:41,180 --> 00:15:44,600
That year I managed to get the budget, let's say from January.

166
00:15:44,600 --> 00:15:47,389
I was here in January already.

167
00:15:47,390 --> 00:15:56,600
So there was a possibility that we might be able to contribute some work because of these collaborations that we have internationally.

168
00:15:56,900 --> 00:16:05,330
And I was very, very busy because we had we were due to host a UN international conference, my group and I here at Oxford,

169
00:16:05,840 --> 00:16:09,620
where of course I was in touch with all these colleagues from, for instance,

170
00:16:09,860 --> 00:16:14,090
China or South Korea who were saying we might not be able to trust this event happen.

171
00:16:14,780 --> 00:16:20,660
So which month with it meant to happen in March. So that was due to have it on the 20th of March.

172
00:16:20,810 --> 00:16:27,730
So so yeah, at that time already, you know, we were concerned, but yeah,

173
00:16:27,740 --> 00:16:32,870
it wasn't clear whether we would have cases, although, you know, we could see it coming.

174
00:16:32,870 --> 00:16:35,120
Right. So we were already looking into what happened.

175
00:16:35,120 --> 00:16:40,400
If we cancel, you know, we lose all the money and that kind of thinking of preparing for the worst.

176
00:16:41,080 --> 00:16:47,360
So that was January. And then in February it was already very obvious that we had to cancel the conference.

177
00:16:47,630 --> 00:16:51,080
You know, the minute we saw what was happening in Italy and in Spain.

178
00:16:52,100 --> 00:16:57,800
And, you know, we have colleagues there who were working in in the hospitals and telling us what was happening.

179
00:16:58,070 --> 00:17:03,320
And so Spain, I mean, we we had a lot Italy was very high profile, but I don't remember very much about Spain.

180
00:17:03,320 --> 00:17:06,500
But did they also have a problem before it came to UK?

181
00:17:06,860 --> 00:17:10,309
So happens. So Italy and Spain are very connected countries.

182
00:17:10,310 --> 00:17:14,900
Yeah. Because they are. Yeah. They speak very similar languages and there's a lot of mobility between both countries.

183
00:17:15,230 --> 00:17:21,920
So the minute it started getting a little bit better for that first wave in Italy, it started getting really bad.

184
00:17:21,950 --> 00:17:30,350
In Spain, especially in big cities like Madrid and Barcelona. I think the first recognised case is 23rd, 24th of February in Spain,

185
00:17:31,340 --> 00:17:35,720
but here like a week later, they were already, you know, thinking of locking down and so on.

186
00:17:36,110 --> 00:17:41,210
So it was, yeah, maybe a couple of weeks earlier than the UK when things had to be done very rapidly.

187
00:17:42,230 --> 00:17:47,240
And at that point, you know, it was very obvious to us who were talking to these people all the time.

188
00:17:47,270 --> 00:17:52,580
And I was also talking to people in Asia that there was no way we were going to do it.

189
00:17:53,810 --> 00:17:57,920
So I guess that's that's the summary of those first few weeks.

190
00:17:59,390 --> 00:18:06,980
And I mean, one of the first responses or a fairly early response was the idea of trialling repurposed drugs.

191
00:18:07,400 --> 00:18:12,230
Was that something that you felt you could contribute to in your approach?

192
00:18:12,320 --> 00:18:20,809
Yes. So so at that time, already in late February, we decided we had to cancel that conference I mentioned and we decided that instead of doing that,

193
00:18:20,810 --> 00:18:24,200
we would have something that we're very used to now.

194
00:18:24,200 --> 00:18:30,049
But at the time sounded quite novel, a kind of virtual we called it study athon,

195
00:18:30,050 --> 00:18:34,640
which was basically let's gather all our colleagues internationally and hold.

196
00:18:34,640 --> 00:18:42,560
I think it was a four day, you know, 24 seven activity because of course you could have people awake during the whole day looking

197
00:18:42,560 --> 00:18:47,100
at a number of things and the number of things we we and this was all on an online screen.

198
00:18:47,190 --> 00:18:51,860
Online. Yes, yes, yes. On teams, which at the time was, you know, something we had not used that much.

199
00:18:52,610 --> 00:18:57,710
So we basically, you know, all the people who had registered for this conference, we had to tell them we're going to cancel.

200
00:18:58,010 --> 00:19:00,979
And then we opened, you know, we told all of them, if you want,

201
00:19:00,980 --> 00:19:04,820
we are going to be hosting this event that's going to be online and we're going to be working on research.

202
00:19:04,820 --> 00:19:08,840
Right. And meeting and discussing other things, which was the plan originally.

203
00:19:09,170 --> 00:19:15,500
And it was very nice to see. Of course, you know, all these collaborators from around the world rallying together.

204
00:19:16,640 --> 00:19:22,520
This was a with a community called Odyssey Observational Data Sciences and Informatics,

205
00:19:22,520 --> 00:19:28,339
or DSA, where, you know, we were going to host the Odyssey Europe meeting.

206
00:19:28,340 --> 00:19:32,180
And and these colleagues, you know, from all over the world said, Yeah, let's work together on this.

207
00:19:32,570 --> 00:19:37,100
And then, you know, late February we were writing on writing the protocols on the studies we wanted to do,

208
00:19:38,030 --> 00:19:44,120
and the most important ones were, first, really characterising the condition because there was a lot of uncertainty.

209
00:19:44,120 --> 00:19:49,669
You would remember some people were saying this is like the flu, and we published What If He's a seminal paper?

210
00:19:49,670 --> 00:19:52,490
The first international characterisation of the condition,

211
00:19:53,120 --> 00:19:59,150
looking at people hospitalised with COVID versus people hospitalised with the flu in previous years just to show.

212
00:19:59,310 --> 00:20:04,400
Differences, Right? We can get into a little if you want, but that was a very important piece of work.

213
00:20:04,760 --> 00:20:08,659
Second piece of work was looking at repurposed medicines or medicines people were

214
00:20:08,660 --> 00:20:15,050
talking about because at the beginning we didn't have a lot of data on COVID patients.

215
00:20:15,830 --> 00:20:23,659
We were interested to learn on medicines that were heavily used, but maybe we didn't know so much about in terms of safety,

216
00:20:23,660 --> 00:20:28,400
because, again, you know, some of these medicines have been I'm talking now hydroxychloroquine in particular.

217
00:20:28,820 --> 00:20:37,250
And we wrote a paper looking at some concerns with the use of hydroxychloroquine in combination with antibiotics,

218
00:20:37,250 --> 00:20:40,879
which is, again, another advantage of this data is that you see everything that happened,

219
00:20:40,880 --> 00:20:46,040
right, is not like a trial where you have one treatment only here, if someone is taking two or three, you see them in combination.

220
00:20:46,460 --> 00:20:53,240
And we found a very strong signal for a cardiovascular safety risk when people combined hydroxychloroquine with azithromycin,

221
00:20:53,240 --> 00:21:01,280
which is an antibiotic that was being heavily used at the time, and this made it all the way to the regulators nationally and internationally.

222
00:21:01,280 --> 00:21:08,630
There were changes in clinical guidelines, change changes or, you know, letters and talking each and say don't combine these two treatments.

223
00:21:09,020 --> 00:21:12,469
And that was using not really using COVID patient data.

224
00:21:12,470 --> 00:21:18,740
This was using retrospective data that we had risk. Yes, but there was indeed a plan then also to look at ones.

225
00:21:18,740 --> 00:21:23,090
We had more data on COVID patients, look at whether the treatments also had any benefits.

226
00:21:23,090 --> 00:21:26,270
Right. Because that was, of course, an important consideration.

227
00:21:26,270 --> 00:21:27,680
You need to balance risk and benefit.

228
00:21:28,940 --> 00:21:38,330
And the last bit of work was trying to identify who was a very high risk of getting, you know, severe COVID when they got the infection.

229
00:21:38,420 --> 00:21:48,350
Right. And that that's again, information we produced and we shared what here nationally with with Sage and and the advice of the government,

230
00:21:48,350 --> 00:21:53,989
but also internationally when everybody was interested because we found that it was

231
00:21:53,990 --> 00:21:57,770
relatively straightforward to predict who was going to get in trouble based on,

232
00:21:57,770 --> 00:22:03,030
you know, five, six conditions, age and sex, as we probably know very well known.

233
00:22:03,410 --> 00:22:07,590
But at the time, that was a big question. And then did ethnicity come up it in that?

234
00:22:07,910 --> 00:22:10,670
So ethnicity didn't come up in that first bit of work,

235
00:22:11,330 --> 00:22:17,420
probably because we were mostly relying on data from countries where there's not a lot of variation in ethnicity.

236
00:22:17,780 --> 00:22:25,249
But we had data at the time from South Korea and Spain, I think I say, and while Spain has some more variation,

237
00:22:25,250 --> 00:22:33,559
but probably not for the ethnicities that lead to a high risk of of severe COVID and South Korea has very little ethnic collaboration.

238
00:22:33,560 --> 00:22:42,560
So we didn't see it at the time, but we still saw that we could predict very accurately who would get severe COVID based on five, six, seven features.

239
00:22:42,560 --> 00:22:46,520
You could basically create a very simple, you know, equation and calculate.

240
00:22:46,820 --> 00:22:54,229
So things like age and age six obesity, diabetes, cancer, hypertension.

241
00:22:54,230 --> 00:22:55,820
I think that was one of the needs.

242
00:22:58,680 --> 00:23:10,159
And and was there any overlap between or relationship between what you were able to do with databases and what the recovery trial was doing,

243
00:23:10,160 --> 00:23:13,880
which was a prospective trial of treatments? Yes.

244
00:23:14,180 --> 00:23:17,960
So I think I think hydroxychloroquine is a beautiful example.

245
00:23:19,520 --> 00:23:28,599
We we did observe and this was already made available to key stakeholders in during that first week that we work together with colleagues.

246
00:23:28,600 --> 00:23:37,249
So late March, very early April, we wrote already the manuscript with these findings that were very important and and I

247
00:23:37,250 --> 00:23:42,140
think so that showed that hydroxychloroquine was not necessarily as safe as we thought.

248
00:23:43,610 --> 00:23:46,999
And especially when you combine that with these antibiotics and, you know,

249
00:23:47,000 --> 00:23:53,000
people are using these combinations very often just to protect against bacterial pneumonia and there was a lot of uncertainty.

250
00:23:53,720 --> 00:23:58,430
So if you remember then in June, recovery showed that the treatment was also not effective.

251
00:23:59,150 --> 00:24:05,240
So we knew by June with two seminal papers from Oxford that hydroxychloroquine wasn't

252
00:24:05,240 --> 00:24:08,570
a good idea because it wasn't as safe as we thought and also it wasn't converted.

253
00:24:08,750 --> 00:24:14,450
Yeah, I think it's a perfect example of complementary information coming from these two types of data.

254
00:24:16,330 --> 00:24:22,100
Yeah, and I think that was beautiful to observe how these different studies can complement each other.

255
00:24:23,810 --> 00:24:29,960
And I mean, similarly dexamethasone, which turned out to be of benefit, was that something that you were looking at in the data?

256
00:24:30,410 --> 00:24:38,450
So for Sun at that time, we were not looking at that because there wasn't a lot of use of steroids.

257
00:24:38,450 --> 00:24:43,550
I don't know if you remember, but there was a lot of, well, colleagues because again, we had a lot of data from Asia at the time.

258
00:24:45,680 --> 00:24:52,100
The thing is, from the experience that colleagues from China and South Korea had from us,

259
00:24:52,310 --> 00:24:58,910
one, so, you know, 20 years before they thought that steroids were not a very good.

260
00:24:59,220 --> 00:25:04,910
Then for this condition and they were not using. It is quite interesting how these things, you know, basically.

261
00:25:05,040 --> 00:25:08,069
So without I guess with without recovery,

262
00:25:08,070 --> 00:25:17,100
we would probably have never recommended this treatment because previous similar conditions taught had told us not to use them.

263
00:25:17,100 --> 00:25:25,409
Right. So it's great that that recovery happened and we managed to see that's the only thing we did on Dexamethasone was this published in the BMJ.

264
00:25:25,410 --> 00:25:29,680
We one of my students at the time who is now post-doc out of bed price,

265
00:25:29,970 --> 00:25:34,230
wrote a paper with colleagues from all over the globe looking at what treatments were

266
00:25:34,230 --> 00:25:39,000
being used routinely when people get admitted with COVID 19 during the first wave,

267
00:25:39,390 --> 00:25:43,330
and you could see a lot of variation, as you can imagine, but you could see things like this.

268
00:25:43,360 --> 00:25:48,690
You could see like in Asia, very, very little steroids, very, very little hydroxychloroquine.

269
00:25:49,110 --> 00:25:55,710
Other countries like the US were using a lot of hydroxychloroquine here in the UK was halfway through.

270
00:25:55,950 --> 00:26:01,530
So you could see this variation that usually demonstrates that you don't really know how to manage the condition, right?

271
00:26:02,220 --> 00:26:08,160
And then these different clusters of, you know, countries that do things similarly, which is quite interesting as well.

272
00:26:08,520 --> 00:26:15,600
And in there we saw for steroids in particular how, you know, very, very little use in Asia, more so in Europe and in the US.

273
00:26:16,950 --> 00:26:25,559
Mm hmm. And and so should I jump to vaccines or was there another important study that

274
00:26:25,560 --> 00:26:31,680
happened before we had the the opportunity to think about thinking about vaccines.

275
00:26:32,790 --> 00:26:37,990
So I guess before we jump into vaccines, the other bit of that was very important was this work on prediction, right?

276
00:26:38,020 --> 00:26:41,790
Oh, yes. Yeah. Because if you classify you again into these three categories.

277
00:26:42,960 --> 00:26:49,890
Yeah, we found that it was relatively easy to predict somebody's risk of of having severe COVID.

278
00:26:50,280 --> 00:26:54,530
And that was very important information because some governments with planning, you know,

279
00:26:54,540 --> 00:27:00,540
shielding strategies or thinking how to roll out their vaccines when they were available.

280
00:27:00,960 --> 00:27:07,620
In the end, it wasn't rocket science because it was relatively intuitive, I guess, but it wasn't quite like the flu.

281
00:27:07,620 --> 00:27:11,429
Like if you try to see who is going to get very high risk of, you know,

282
00:27:11,430 --> 00:27:17,850
ICU or hospital admission with the flu, you get like two pigs with young children and and elderly people.

283
00:27:17,850 --> 00:27:24,060
With COVID, you had, you know, children were okay. Very elderly people also had problems.

284
00:27:24,300 --> 00:27:29,459
But then you also had this group of people who were relatively healthy but had obesity, diabetes,

285
00:27:29,460 --> 00:27:38,550
a very different metabolic profile that made them, you know, for some reason prone to having very severe complications from COVID 19.

286
00:27:38,550 --> 00:27:44,430
And that was quite a finding than whatever people did with that information, different governments and different things.

287
00:27:44,430 --> 00:27:48,570
Right. But I thought at the time that was very important information that we had to make available.

288
00:27:49,530 --> 00:27:53,160
And again, that that I think informed some of the policies internationally.

289
00:27:54,600 --> 00:28:01,170
And what was the mechanism quite interested in that kind of detail of how information flowed between researchers and the government.

290
00:28:01,980 --> 00:28:05,730
Did you have a kind of hotline to the Department of Health? Yeah.

291
00:28:05,730 --> 00:28:11,220
So, so I guess there were a number of ways things happened.

292
00:28:12,060 --> 00:28:23,370
We were making all our data available in the form of Preprints in med archives mostly, and some governments and some even international stakeholders,

293
00:28:23,370 --> 00:28:30,509
like the European Medicines Agency or the FDA, were very good at reviewing the literature every day and identifying studies that were relevant.

294
00:28:30,510 --> 00:28:33,810
And we would get an email saying, Oh, we would like to know more about these.

295
00:28:34,140 --> 00:28:39,570
Or sometimes the manuscript had enough information, then more nationally.

296
00:28:39,570 --> 00:28:44,459
I also had, you know, been in touch with some of the colleagues in the Department of Health,

297
00:28:44,460 --> 00:28:52,830
and I was very closely talking to them and telling them, look, there's this paper coming in Morocco tomorrow, see if it's of interest.

298
00:28:52,830 --> 00:28:57,510
And if you need any clarifications, let me know. Yeah.

299
00:28:57,570 --> 00:29:04,830
So there was basically this, you know, more proactive or more reactive attitude depending on what the information you were producing.

300
00:29:04,830 --> 00:29:13,110
And I guess also the information that each country had already available because some countries were very data poor while the UK was very data rich.

301
00:29:13,110 --> 00:29:20,250
So the Department of Health, I believe, had a lot of information coming from UK HCA as well as researchers from all over the country.

302
00:29:21,300 --> 00:29:28,860
So I you know, we were very proactive in providing the information, but I'm sure they had much more data than many other countries did.

303
00:29:30,700 --> 00:29:36,020
So it's about at what point do you think it became obvious that there was going to be a vaccine?

304
00:29:36,640 --> 00:29:39,940
You know, it was looking good. We haven't actually got the results of the trials yet,

305
00:29:39,940 --> 00:29:45,489
but the possibility that there would be a vaccine within a year of the pandemic beginning was looking looking good.

306
00:29:45,490 --> 00:29:49,600
And did that influence your thinking about how to address that? Yes, very much so.

307
00:29:49,840 --> 00:29:56,340
I think by the summer of 2020, which, you know, thinking about it is quite early on, amazingly fast.

308
00:29:56,350 --> 00:30:02,259
Yeah, maybe. Maybe, of course, because I live in Oxford and then you, you know, if someone who is in a phase one trial,

309
00:30:02,260 --> 00:30:07,300
you know, like I had a neighbour who was one of the first 50 or 100 people who were testing one of the vaccines.

310
00:30:07,870 --> 00:30:19,300
Yeah. So, so we knew by summer 2020 that there were candidate vaccines that were, you know, going through a kind of trial process.

311
00:30:20,170 --> 00:30:26,380
I don't think we knew exactly when we would have enough data to approve them, but but we knew, you know, this was moving very quickly.

312
00:30:26,830 --> 00:30:35,290
So what that meant for us was that we needed to be prepared to help the regulators with their work on safety.

313
00:30:35,410 --> 00:30:43,899
Yeah. And when it comes to vaccines, safety is paramount because, you know, if you're giving someone who is sick a treatment,

314
00:30:43,900 --> 00:30:49,690
you're giving them that treatment to allow something to improve the quality of life.

315
00:30:49,990 --> 00:30:52,870
With vaccines, we are basically trying to prevent something from happening.

316
00:30:52,870 --> 00:30:58,370
And typically, not only people who are very high risk, but also potentially very healthy people.

317
00:30:58,370 --> 00:31:02,380
Right. So so safety concerns are very important.

318
00:31:03,250 --> 00:31:11,140
And so the way he works with with regulatory work is that once a product is approved based on the clinical trial data,

319
00:31:11,440 --> 00:31:14,260
there are all these commitments for post-marketing surveillance.

320
00:31:14,530 --> 00:31:20,859
And typically the first signals that the regulators get come from yellow cards and that's that sort of information.

321
00:31:20,860 --> 00:31:27,150
They call it spontaneous reports of patients or physicians reporting that something has happened.

322
00:31:27,170 --> 00:31:30,580
It doesn't need to be causal, but they just report something that wasn't in the label.

323
00:31:31,480 --> 00:31:34,070
And it could be it could be a sprained ankle, it could be anything.

324
00:31:34,130 --> 00:31:41,770
Yeah, they need to look at the narratives in those reports and then assess, you know, if it's let's use the sprained ankle example,

325
00:31:42,040 --> 00:31:47,290
is this happening more than for people who are being vaccinated with flu vaccines, for instance?

326
00:31:47,290 --> 00:31:51,130
Right. So they do a little bit of what they call a disproportionality metric.

327
00:31:51,820 --> 00:31:55,960
Yeah. In any case, when they see something that they think could be potentially causal,

328
00:31:56,350 --> 00:32:06,130
they then what they do is they they look at previous knowledge on how often that sort of thing, that event happened in the general population.

329
00:32:06,460 --> 00:32:10,690
And how often are we seeing that in the vaccinated population. And they compare both, right?

330
00:32:10,690 --> 00:32:19,750
And they do this type of analysis they call observed versus expected analysis, which I think is an interactive terminology in this case.

331
00:32:20,080 --> 00:32:23,230
So of course, to have an observed effect is expected analysis.

332
00:32:23,380 --> 00:32:27,430
You're going to need the observe. You also need the expected, which comes from historical data.

333
00:32:27,430 --> 00:32:35,320
Right. And although you would think that we have information on the rates of any condition you can imagine, that is not true.

334
00:32:35,710 --> 00:32:40,030
There are many things that are very important for vaccine surveillance that we did not

335
00:32:40,060 --> 00:32:45,970
know quite how often they happened in the historical data in the general population.

336
00:32:46,270 --> 00:32:53,379
So we thought the first thing we had to do was basically do this very accurate analysis,

337
00:32:53,380 --> 00:32:57,370
stratifying, you know, into different groups of age and sex and so on.

338
00:32:58,030 --> 00:33:05,079
On how often do you see these conditions that were potentially identified as potential safety signals to keep an eye

339
00:33:05,080 --> 00:33:11,920
on before we had the trials so that the regulators had all that information by the time the vaccines were approved,

340
00:33:12,040 --> 00:33:14,230
and this is people who had vaccinations for other things,

341
00:33:14,410 --> 00:33:20,239
this would be either people who had been vaccinated for something else or in general in the whole population.

342
00:33:20,240 --> 00:33:27,390
Oh, I see. So if you want to say, oh, I'm seeing more blood clots in this population that has been vaccinated than in the general population,

343
00:33:27,400 --> 00:33:30,730
you need to know how often that happens in that general population,

344
00:33:31,720 --> 00:33:35,950
and that is not always available maybe for blood clots it was, but for other things it wasn't,

345
00:33:36,130 --> 00:33:40,930
because some of these things are very specific and not very common in general.

346
00:33:41,110 --> 00:33:43,809
And we did a piece of work again with this community,

347
00:33:43,810 --> 00:33:50,200
with the Odyssey community and colleagues across the globe looking at basically estimating all these rates

348
00:33:50,200 --> 00:33:57,760
that we call them background rates or expected rates from historical data 2017 to 2019 before the pandemic.

349
00:33:58,690 --> 00:34:02,860
And we had data from 16 or 17 countries that was published in the BMJ.

350
00:34:02,860 --> 00:34:08,829
And I know from I've heard from regulators that they use that all the time because that was the first point of contact.

351
00:34:08,830 --> 00:34:12,819
You know, when they were looking at blood clots, they would go and say, well, what is the expected number?

352
00:34:12,820 --> 00:34:19,990
And they could do that calculation very quickly. And we did that for 15 or 16 conditions that they had.

353
00:34:19,990 --> 00:34:26,680
I think the FDA had listed US adverse events of a special interest for surveillance of COVID vaccine safety.

354
00:34:28,000 --> 00:34:29,770
So one of the other things I've heard people talk about.

355
00:34:29,840 --> 00:34:38,600
What is the anxiety that you might get an exacerbation of the condition with the vaccine that it could make?

356
00:34:38,870 --> 00:34:42,559
They could actually make the symptoms. Yes. I forgot what the proper term for that is.

357
00:34:42,560 --> 00:34:46,760
Yeah. There is a term for that. And I know it's not quite getting to me now.

358
00:34:48,950 --> 00:34:52,250
Yeah, I know. I know that is a concern.

359
00:34:53,720 --> 00:34:57,230
I think that was relatively clear.

360
00:34:57,230 --> 00:35:04,700
It was theoretically clear early on from the phase two trials that that was not happening for the vaccines that were coming up.

361
00:35:04,700 --> 00:35:13,130
So that's always a concern. But the good thing with that particular problem is that you typically find out very early on the trial because, you know,

362
00:35:13,850 --> 00:35:19,790
yeah, so having COVID, of course, is not a rare outcome compared to, you know, having a blood clot or something else.

363
00:35:20,240 --> 00:35:24,730
So that was I think that was data that we didn't produce because it was already available in trials.

364
00:35:25,160 --> 00:35:29,030
Yeah. So that was a the first kind of decision.

365
00:35:29,030 --> 00:35:37,549
We had to produce that data as early as possible. And then the other decision was to start preparing, identifying candidates across the world,

366
00:35:37,550 --> 00:35:45,140
but also across Europe, more specifically of people who we knew would have really good data on vaccine exposures.

367
00:35:45,330 --> 00:35:53,280
Yeah, So, you know, I'm talking about people around the continent and also in the UK who had access to data with, you know,

368
00:35:53,300 --> 00:36:01,370
vaccine brand vaccine batch, because there could be issues with specific batches, you know, specific date when a vaccine was given.

369
00:36:01,640 --> 00:36:07,760
And that was we knew would be hard because we knew this vaccination campaign would be universal and

370
00:36:07,760 --> 00:36:13,219
people would be going to the Kasama Stadium in Oxford or to become known in Barcelona to get vaccinated.

371
00:36:13,220 --> 00:36:18,140
Right. So it's not like we would just go to the GP records and find out that we needed, you know,

372
00:36:18,140 --> 00:36:23,510
data sources that had that ability to link or to identify people getting vaccinated.

373
00:36:24,500 --> 00:36:31,940
And that was that is where it was very obvious that European data would be the best in the world for that because of the way we organise healthcare,

374
00:36:32,120 --> 00:36:38,899
right? So a country like the UK or a country like Spain that have universal healthcare systems where you

375
00:36:38,900 --> 00:36:45,260
can track a patient across the healthcare is much better because you have a unique number NHS,

376
00:36:45,470 --> 00:36:48,470
because you have a unique number, and Scandinavia will be another example, right?

377
00:36:48,920 --> 00:36:54,290
Compared to, for instance, you know, in the US where people basically have a health insurance,

378
00:36:54,650 --> 00:37:01,430
but then that doesn't always necessarily link to a vaccination in a stadium or in some sort of centre.

379
00:37:04,120 --> 00:37:10,190
So yeah, we prepared for that, knowing that at some point there will be a vaccine safety issue and we would need to produce later.

380
00:37:10,300 --> 00:37:13,570
And I know if you want to discuss that in more detail, but that's, you know,

381
00:37:13,810 --> 00:37:18,070
I guess the third bit of work that we did that was very impactful was investigating

382
00:37:18,070 --> 00:37:25,360
some of the events as they as they were observed by some regulators and and those.

383
00:37:25,390 --> 00:37:33,130
So you were once the vaccine rollout started, you were you you were monitoring continuously in these.

384
00:37:34,420 --> 00:37:40,840
How does it work? Do you have software set up to to pick out any of these signals that something's going on?

385
00:37:40,990 --> 00:37:46,600
So that's the work that the regulators do. All right. I see. They're the ones who typically will modify a signal.

386
00:37:46,600 --> 00:37:53,050
And usually the happens before in the spontaneous report data that happens in the data I have access to.

387
00:37:53,470 --> 00:37:58,780
But what we have is software that was capable of software and colleagues that

388
00:37:58,780 --> 00:38:03,309
were capable of curating this data all the time so that when an issue happened,

389
00:38:03,310 --> 00:38:10,670
we had access to recent data because otherwise, yeah, you basically don't have enough information to do the research.

390
00:38:11,170 --> 00:38:16,479
And of course the the data that the regulators have, these spontaneous reports are very limited in that,

391
00:38:16,480 --> 00:38:20,500
you know, they are low in numbers, they don't have enough information.

392
00:38:20,500 --> 00:38:25,840
They don't have either nominate or you don't know how many people look like that person that reported that problem.

393
00:38:26,500 --> 00:38:30,070
There are issues with completeness. Not everybody reports when they have an issue,

394
00:38:30,520 --> 00:38:36,909
but they are very useful as a kind of signal detection system where you can pick up something that's surprising.

395
00:38:36,910 --> 00:38:39,690
Right. And then we were basically ready to go.

396
00:38:39,770 --> 00:38:46,360
I mean, there was something that was indeed surprising to, you know, a more elaborate kind of evaluation of that specific signal.

397
00:38:46,420 --> 00:38:50,170
Mm hmm. And how did that play out?

398
00:38:50,630 --> 00:38:58,690
Um, but obviously, there was I think everybody remembers that there was this sudden concern about the possibility of blood clots.

399
00:38:58,720 --> 00:39:01,410
Yeah, the numbers were tiny. Yeah.

400
00:39:01,420 --> 00:39:09,130
So I think in relation to the AstraZeneca vaccine, so the blood clot one is one that I think we contributed a lot of data to.

401
00:39:10,240 --> 00:39:18,190
So there was, yeah, basically, I guess mostly the UK national regulator and the European Medicines Agency were the first

402
00:39:18,190 --> 00:39:22,900
ones to see this signal because these vaccines were not heavily used in the US initially.

403
00:39:23,830 --> 00:39:29,470
And then they basically the minute they said there was an issue with that, we had the background rates that I mentioned before.

404
00:39:29,620 --> 00:39:40,500
We could we could see whether there was a very clear increase already and we didn't see something very, I'd say, spectacular.

405
00:39:40,510 --> 00:39:42,520
It wasn't something that I mean,

406
00:39:42,520 --> 00:39:48,249
they were talking about something very clear from the spontaneous report data that we were not seeing something that clear from our data,

407
00:39:48,250 --> 00:39:55,600
which was reassuring in a way. But then, of course, we needed to then say, okay, let's do a cohort study where we followed people up,

408
00:39:55,870 --> 00:40:02,020
you know, and observed how many of them had one of those blood clots in their first 28 days after a vaccine.

409
00:40:02,620 --> 00:40:11,110
And yeah, so we did that in collaboration with partners across Europe again, and we were the first to produce that international bit of work on that.

410
00:40:11,800 --> 00:40:22,510
That was the first bit of more, let's say, you know, rapid evaluation and using these, you know, expect to observe the expected type of analysis.

411
00:40:22,840 --> 00:40:29,140
And then after that, we also said, look, if if this signal is going to be observed for people vaccinated with these specific vaccines,

412
00:40:29,470 --> 00:40:33,210
why not compare them to the people vaccinated with the other vaccines to see whether,

413
00:40:33,520 --> 00:40:38,049
you know, they will be more comparable than comparing them to, you know, people from 2019.

414
00:40:38,050 --> 00:40:38,290
Right.

415
00:40:38,650 --> 00:40:46,600
So that was the second bit of work that we also completed using data from, I think, six European countries and US, and that was published in the BMJ.

416
00:40:46,600 --> 00:40:58,660
And and very, very welcome. We did observe little signal for thrombocytopenia, these reduction in blood clots, in reduction in platelet counts.

417
00:40:58,990 --> 00:41:04,420
And then a little bit of a potential signal for blood clots, which kind of, I think, you know,

418
00:41:05,950 --> 00:41:10,210
reassured us that the data we had produced before was was in line with that as well.

419
00:41:11,080 --> 00:41:20,530
So that was for the blood clots. And the other issue that I don't think is entirely resolved yet is this problem with Guillain-Barre syndrome.

420
00:41:20,530 --> 00:41:24,420
I don't know if you've heard right. Yes, I have. That's a neurological inflammatory condition.

421
00:41:25,390 --> 00:41:32,590
And there were some signals coming up again from spontaneous reports on Guillain-Barre, but also on Encephalomyelitis,

422
00:41:32,800 --> 00:41:40,390
another inflammatory condition of a different part of the brain and also Bell's palsy or facial paralysis.

423
00:41:40,750 --> 00:41:46,850
So we did an analysis of those three in data from the UK and Spain, which were the later we had more access to one and you know,

424
00:41:47,050 --> 00:41:57,010
that were of better quality and found in our analysis, which I think were quite, quite accurate.

425
00:41:57,430 --> 00:42:02,410
We didn't find a signal for any of those three. But interestingly, we found that when people have.

426
00:42:02,790 --> 00:42:07,140
The risk of having those conditions is increased by like three fold.

427
00:42:07,410 --> 00:42:15,570
All right. Which I think was reassuring because at the time, again, the counterfactual of not getting vaccinated was getting it right.

428
00:42:15,600 --> 00:42:19,380
Yes. I was going to ask you about that in relation to the blood clots, because, yes, I looked at that.

429
00:42:19,500 --> 00:42:23,020
We will look at that and we saw a similar thing. I think we knew that already.

430
00:42:23,050 --> 00:42:29,370
I think so that the risk from getting COVID of these adverse events is much higher than the risk.

431
00:42:29,370 --> 00:42:36,749
Much, much higher. Yes, I think maybe I didn't point that out earlier because I think we kind of knew that all of us who are

432
00:42:36,750 --> 00:42:41,399
clinicians knew that because one of the reasons people were dying when they had COVID was blood clots.

433
00:42:41,400 --> 00:42:43,950
Right? So we were very worried that, you know,

434
00:42:43,980 --> 00:42:50,790
people were scared of having a blood clot in half a million people when actually when people got COVID, it was like one in a thousand maybe.

435
00:42:50,800 --> 00:42:54,120
Right. So that was but, you know, on paper, that was very, very clear.

436
00:42:54,120 --> 00:43:00,670
There's a yeah, five, ten fold increased risk of things like, you know, pulmonary embolism,

437
00:43:00,680 --> 00:43:10,620
some blood clots in your lungs with COVID compared to a potential increase, which wasn't even significant in this case when you had the vaccine.

438
00:43:11,100 --> 00:43:14,129
So the kind of. Yeah. Risk benefit again. Right.

439
00:43:14,130 --> 00:43:22,470
So the the the problem when when you had the infection was much, much worse than when you had the vaccine.

440
00:43:22,800 --> 00:43:29,790
But I guess the in the public mind, having a vaccine is something you choose to do to yourself or have somebody do to you.

441
00:43:29,880 --> 00:43:33,660
Yeah. And in a way, you feel you've got control over that.

442
00:43:33,660 --> 00:43:40,799
You can take a decision whether or not to get it and with it, whereas catching an illness is just bad luck, I suppose.

443
00:43:40,800 --> 00:43:48,090
I don't know. I'm trying to account. Yeah. So the way people seem to be more fearful of possible risks from vaccines than they are the disease.

444
00:43:48,300 --> 00:43:58,860
So I guess my argument against that would be actually in a way, you are choosing whether you have the vaccine and the vaccine protects your arms.

445
00:43:58,860 --> 00:44:02,429
ABC is right. So it's not only bad luck, it's also if you choose not to have the vaccine,

446
00:44:02,430 --> 00:44:07,860
then you are much, you know, at a much higher risk of having a severe form of the disease,

447
00:44:07,860 --> 00:44:10,440
which is what leads to these complications, because, of course,

448
00:44:10,920 --> 00:44:20,040
we also did a study looking at what happens if you have COVID in the ambulatory setting, like, you know, in a home.

449
00:44:20,040 --> 00:44:22,799
Right. Or in hospital.

450
00:44:22,800 --> 00:44:30,030
And the people who are in hospital that had a much, much, much higher risk of having blood clots when they had this severe form of COVID 19.

451
00:44:30,510 --> 00:44:33,630
So in a way, you know, yeah, I do appreciate what you mean.

452
00:44:33,640 --> 00:44:41,969
Like when you obviously put in your arm there to get a vaccine whilst people probably had this idea that, well, COVID is just bad luck and you get it.

453
00:44:41,970 --> 00:44:48,680
But basically there was this intervention that was preventing you from getting it, or at least clearly preventing you from having a severe.

454
00:44:48,690 --> 00:44:53,099
I'm not saying it's rational. No, no, no. I know. I know how to work with the kind of rationality about it.

455
00:44:53,100 --> 00:44:55,740
But it's it's just this, you know. Yeah.

456
00:44:55,770 --> 00:45:01,649
I guess as scientists and especially in epidemiology, we always work with counterfactuals, but we don't like to give you one number.

457
00:45:01,650 --> 00:45:07,500
We like to say this is the number if you do this, but this is the counterfactual if you don't do it.

458
00:45:08,190 --> 00:45:13,810
So, you know, to us it was important to produce this knowledge on what happens when people have COVID without being vaccinated,

459
00:45:14,070 --> 00:45:22,620
you know, And that was that was our our thinking, both with the blood clot signal and with the neuro inflammatory signals.

460
00:45:22,620 --> 00:45:28,979
And again, with those, to my surprise, because I didn't know what to expect, really, we we did see, again,

461
00:45:28,980 --> 00:45:37,950
a very high risk of having these newer inflammatory complications when people had COVID and were not vaccinated.

462
00:45:37,950 --> 00:45:43,409
So, again, that was very reassuring data to say if you are scared of blood clots or if you are scared of having Guillain-Barre,

463
00:45:43,410 --> 00:45:45,650
what you should do is actually get vaccinated. Yes, yes, yes.

464
00:45:46,320 --> 00:45:52,979
And and that, I think, was important not only to reassure the regulators and, you know, to inform future research,

465
00:45:52,980 --> 00:45:58,920
but also hopefully to inform the population of the importance of of vaccines getting getting vaccinated.

466
00:46:00,150 --> 00:46:05,340
And have you looked at long COVID at all? So that's an ongoing project we're doing right now.

467
00:46:07,080 --> 00:46:14,940
We are looking at whether and this is and again, NIH, National Institute for Health Research Grants that we got about,

468
00:46:15,150 --> 00:46:21,060
I want to say nine months ago to look at whether being vaccinated prevents long COVID.

469
00:46:21,420 --> 00:46:29,010
And this is something someone could have looked at in the vaccine trials, but long COVID wasn't a concern when we did those trials.

470
00:46:29,010 --> 00:46:34,649
I mean, you know, it's going to be very hard to think that somebody wouldn't do a trial of vaccination versus not vaccination of age.

471
00:46:34,650 --> 00:46:39,060
Right. So we are doing this with the kind of data we have information on.

472
00:46:40,530 --> 00:46:47,610
We are finalising those analysis right now. So I'm hoping we will have results by the end of this month, but it's looking quite promising.

473
00:46:48,480 --> 00:46:56,430
I think we will have very, very good, very compelling evidence that vaccination is also a good idea to prevent long-covid,

474
00:46:56,430 --> 00:47:00,840
which is what we kind of expected based on anecdotal evidence.

475
00:47:01,590 --> 00:47:09,819
But there is. There's a systematic review of the literature by UK agency showing that there is a need for this specific type of study,

476
00:47:09,820 --> 00:47:14,810
and that's why we we are doing it. So there's no question now.

477
00:47:14,840 --> 00:47:25,280
I mean, I think early on when Long-covid was being talked about, I think even the term long COVID was developed by patient groups because there

478
00:47:25,280 --> 00:47:32,930
was a lot of scepticism in the medical community about whether it was real. And is that something your studies have addressed?

479
00:47:33,170 --> 00:47:36,600
So this is something we have looked at already, what we've done.

480
00:47:36,770 --> 00:47:42,170
I don't think we can with the data I have, I don't think we can look at whether long-covid is a thing.

481
00:47:42,180 --> 00:47:44,770
You know, I don't think that is something we can do.

482
00:47:44,780 --> 00:47:51,170
I tend to believe that if patients explain an experience they are probably having, and so I never hesitated of that.

483
00:47:51,680 --> 00:47:57,620
But what's really important, I believe, and what we need and is research the all this research on long COVID,

484
00:47:57,620 --> 00:48:00,949
we've been working with long COVID patients very closely.

485
00:48:00,950 --> 00:48:03,470
So it's been very much informed by patients themselves.

486
00:48:03,950 --> 00:48:10,790
But one of the things was, you know, we have a list of 25 symptoms that the W.H.O. identified us,

487
00:48:11,240 --> 00:48:15,740
you know, to be used as a definition of long COVID clinically.

488
00:48:15,740 --> 00:48:21,110
So this is what clinicians do nowadays, is they see whether people had those symptoms for more than three months.

489
00:48:21,110 --> 00:48:25,669
And if you have one or more of those for more than three months, then you officially have long-covid.

490
00:48:25,670 --> 00:48:29,510
And that type of new is it called post-COVID syndrome.

491
00:48:29,510 --> 00:48:34,280
Now post COVID? Yeah, I think they could pass quite post-acute COVID syndrome as well.

492
00:48:34,280 --> 00:48:38,520
As long COVID remains the the the I guess, yeah,

493
00:48:39,110 --> 00:48:44,959
less formal terminology and the most official one is that it's passed, but it has this definition by the W.H.O.,

494
00:48:44,960 --> 00:48:51,980
which comes from an expert group and, you know, who sat together and said, this is what my patients have and that's what we are using.

495
00:48:51,980 --> 00:48:58,160
But I think that that can be refined and the kind of data we have access to can be very useful to really find that.

496
00:48:58,460 --> 00:49:02,360
So before looking at, you know, do vaccines prevent long COVID,

497
00:49:02,360 --> 00:49:07,399
we did a lot of work to look at what that long COVID actually look like in the data and what we did,

498
00:49:07,400 --> 00:49:10,010
which is quite interesting and has not been done before.

499
00:49:10,370 --> 00:49:17,160
And this is something that will hopefully be out in the literature very soon because we've submitted the paper and everything is looking at, okay,

500
00:49:17,180 --> 00:49:24,979
what happens if I get a person who had who tested positive for COVID and I compare them to a similar person with the same age,

501
00:49:24,980 --> 00:49:29,210
gender and so on, who tested on the same week but tested negative.

502
00:49:29,360 --> 00:49:33,620
Yeah, there are some of the symptoms that are listed in the W.H.O. list,

503
00:49:33,620 --> 00:49:40,459
like chai tea that I think all of us were probably experiencing to some degree when we were in the middle of lockdown.

504
00:49:40,460 --> 00:49:47,330
Right? So we really wanted to see whether there was which of those 25 symptoms had really difference in the

505
00:49:47,330 --> 00:49:53,180
duration or the severity or their presentation in people tested positive versus tested negative,

506
00:49:53,180 --> 00:49:54,320
who looked alike.

507
00:49:54,830 --> 00:50:04,580
And we've done that work and we've I think we're going to provide very important information on what are the symptoms that are more differential.

508
00:50:04,610 --> 00:50:12,739
I'm not saying that patients with long COVID didn't have on chai tea. What I'm saying is if we know which of those symptoms are more specific,

509
00:50:12,740 --> 00:50:18,889
maybe it's the words to Long-covid that could help us do research on those specific mechanisms.

510
00:50:18,890 --> 00:50:28,610
Right? So imagine if we just said, well, people with long COVID have ten times more risk of having cardiac symptoms like palpitations,

511
00:50:28,610 --> 00:50:32,209
chest pain and something else than people who tested negative.

512
00:50:32,210 --> 00:50:36,530
Clearly, there's at least a group of people who have a heart condition that should be investigated.

513
00:50:36,530 --> 00:50:44,419
Right? What if we say in chai it's not differential, for instance, then maybe that is not something we should be investigating so much.

514
00:50:44,420 --> 00:50:47,750
We should treat it, of course, because we should treat everyone with that condition.

515
00:50:48,110 --> 00:50:54,200
But I don't think that is then something we should think is involved in the issue of pathogenesis,

516
00:50:54,200 --> 00:51:03,770
as we say in the causal, in the mechanistic, uh, processes that happen before people have have long COVID.

517
00:51:04,250 --> 00:51:11,030
So that's what we've done and we're going to publish very soon. And it's very exciting to see how this kind of analysis can, you know,

518
00:51:11,300 --> 00:51:16,520
and rather than new knowledge and new mechanisms to understand more about this

519
00:51:16,520 --> 00:51:21,829
condition that's intriguing and affecting loads and loads of people in that analysis,

520
00:51:21,830 --> 00:51:30,620
we also looked at whether there was a way, a wave of long-covid every time we had a wave of COVID and whether that changed over time.

521
00:51:31,230 --> 00:51:34,610
And we did see that that was true for the first few waves.

522
00:51:34,970 --> 00:51:40,790
But it looks like at the end, like from Delta onwards, probably because people were already mostly vaccinated,

523
00:51:41,150 --> 00:51:47,629
we do see less of a wave of long-covid after each wave of COVID in the community.

524
00:51:47,630 --> 00:51:51,260
So that again, is, I think, very important planning for what right.

525
00:51:51,260 --> 00:51:54,950
And how many people come we expect will get COVID after having COVID?

526
00:51:55,700 --> 00:52:01,700
And do you still have questions to ask about COVID? I mean, we're I think we're being encouraged to think, oh, it's all finished now.

527
00:52:01,700 --> 00:52:05,479
But, you know, you look at the data and I mean, there are still quite a lot of people around.

528
00:52:05,480 --> 00:52:11,930
You have it. Yeah. So we are doing a fair bit of work on maybe smaller things now.

529
00:52:12,140 --> 00:52:21,080
Maybe. On such big data in terms of numbers of people, but rich of data in terms of numbers of variables in a given dataset,

530
00:52:21,080 --> 00:52:26,719
we're using a lot of UK Biobank data, for instance, now looking at genetic traits associated.

531
00:52:26,720 --> 00:52:30,320
We, for instance, published a paper showing that there are some,

532
00:52:30,320 --> 00:52:35,990
some genetic traits that predict who is going to have a blood clot after having COVID, for instance,

533
00:52:35,990 --> 00:52:41,870
which is very important because you could potentially consider doing prophylaxis prevention

534
00:52:41,870 --> 00:52:46,130
of sorts in people who have genetic markers of high risk when they get COVID right.

535
00:52:47,030 --> 00:52:54,320
So we're doing work on that. We're doing work on genetic markers of breakthrough infection, genetic markers of long COVID.

536
00:52:54,680 --> 00:52:59,360
So all all of that is work ongoing in the team right now as we speak.

537
00:53:00,020 --> 00:53:05,150
And in the meantime, if you you've you're still doing your predictive work or going back to your predictive work on cancer.

538
00:53:06,050 --> 00:53:10,460
In the meantime, the team is also looking at controls in cancer,

539
00:53:10,970 --> 00:53:15,200
predicting cancer and looking at the use of machine learning and artificial

540
00:53:15,200 --> 00:53:20,930
intelligence to classify cancer being better so that we can treat it more accurately.

541
00:53:20,960 --> 00:53:25,820
I guess is the is the way to say it. I mean, of course, the team has grown quite a bit.

542
00:53:25,820 --> 00:53:30,410
My team has grown quite a bit since the pandemic because, you know, the amount of work that we're doing is,

543
00:53:30,410 --> 00:53:35,090
I think, probably three fold what we were doing as a consequence of the pandemic.

544
00:53:35,180 --> 00:53:41,150
I would probably say so because of course all this work on COVID didn't exist, right?

545
00:53:41,390 --> 00:53:49,340
So we have to do that that we were not doing before. And there was this deep prioritisation exercise that we lead for a while where we said, Well,

546
00:53:49,340 --> 00:53:54,979
the most important thing now is to figure this out, but now we have to go back to also answering other questions that matter.

547
00:53:54,980 --> 00:54:00,350
Like, you know, this issue I was talking about in elderly people who are taking these medicines, Right?

548
00:54:00,830 --> 00:54:04,940
So we are now, I think, trying to answer all those things in parallel.

549
00:54:04,940 --> 00:54:08,720
And that requires more resources. And and we do have that.

550
00:54:10,040 --> 00:54:16,189
But do you think the value of your own group and the net, the wider global network has been recognised?

551
00:54:16,190 --> 00:54:20,200
Has your profile been raised? Absolutely. Which will help with getting more funding for people?

552
00:54:20,300 --> 00:54:27,590
Yeah. Yeah, absolutely. So we we have so in I was going to say in the middle of the pandemic, by the end of 2021,

553
00:54:27,590 --> 00:54:33,230
we had a very large European grant to look at this issue with cancer that I was mentioning.

554
00:54:33,710 --> 00:54:45,470
And more recently just a year ago, we we got a very large contract from Erasmus Medical Centre and the European Medicines Agency to do this work on,

555
00:54:45,770 --> 00:54:49,550
you know, looking at the safety of medicines and so on internationally.

556
00:54:50,450 --> 00:54:57,170
So so at the international level, 100%, very much increase at the national level, I would say so as well.

557
00:54:58,670 --> 00:55:05,059
And I mean obviously a work question perhaps doesn't apply to you because your work has always been very collaborative.

558
00:55:05,060 --> 00:55:06,650
It's nationally and internationally.

559
00:55:07,280 --> 00:55:17,899
But did you have a sense that working during the pandemic was that people were more prepared to collaborate and you wanted to work together?

560
00:55:17,900 --> 00:55:23,000
And so that the normal kind of cut and thrust of academic competition was less less of an issue?

561
00:55:23,420 --> 00:55:30,860
Yes, definitely. I think I think specifically for I mean, to be honest, at the beginning, COVID was not bringing any funding to anyone.

562
00:55:30,860 --> 00:55:32,930
We were just doing it because we had to do it right.

563
00:55:33,830 --> 00:55:40,400
So when I was talking about, you know, growing the size of the team at the beginning, it was basically the same people working more hours.

564
00:55:41,750 --> 00:55:47,569
So it wasn't a funding competition as such. It was more, you know, I guess, competition to produce useful knowledge.

565
00:55:47,570 --> 00:55:52,219
Yes. And I think, yeah, I want to be optimistic.

566
00:55:52,220 --> 00:56:00,350
I think I think the pandemic in a way catalysed a move towards more collaborative work, at least in my field, in heavily the sciences.

567
00:56:00,350 --> 00:56:05,549
I would say probably, Yeah, I think so. Okay.

568
00:56:05,550 --> 00:56:10,470
I'm going to come along to to talk a little bit about how it impacted on you personally and how how you work.

569
00:56:10,470 --> 00:56:16,020
So, I mean, first of all, how threatened did you feel by the virus itself yourself personally?

570
00:56:17,250 --> 00:56:22,049
Yeah. So let's say probably because I had access to a lot of information, I was, I think,

571
00:56:22,050 --> 00:56:26,610
more scared than most of the population much earlier than most of the population.

572
00:56:29,580 --> 00:56:38,730
Yeah. So, you know, I think this is a very nasty virus that before vaccination is still nowadays very risky and very dangerous.

573
00:56:39,810 --> 00:56:43,170
So, you know, not specifically because I have any particular condition or anything,

574
00:56:43,170 --> 00:56:49,680
but I knew it was a was a virus that was causing, you know, even in people like me, healthy and relatively young.

575
00:56:50,110 --> 00:56:57,500
A hospital admission in every 100 which is you know, if you are if you do the math it's a lot of people in hospital.

576
00:56:57,510 --> 00:57:04,950
Right. I, I am also someone who travels a lot for work, who interacts with a lot of people because of collaboration,

577
00:57:05,400 --> 00:57:08,070
who meets people in a room like this all the time.

578
00:57:08,490 --> 00:57:15,149
So at the beginning when when we were, you know, starting to say, okay, it's clearly here because it's until the end,

579
00:57:15,150 --> 00:57:21,450
it's in Spain, and we're getting like, you know, hundreds of people moving in in both directions every day.

580
00:57:22,380 --> 00:57:27,890
It to me was very scary because I was kind of very confident that I had all the chances to get it.

581
00:57:28,380 --> 00:57:32,190
And because there was this, you know, uncertainty on how long the incubation period was,

582
00:57:32,190 --> 00:57:36,240
I was also scared to pass it on to my partner and to my children, right?

583
00:57:38,310 --> 00:57:48,209
Yeah. So it was it was very scary at the beginning. And I think it was still scary for a very long time for me until we were vaccinated.

584
00:57:48,210 --> 00:57:53,850
And at that time I knew it seemed quite different and we were moving into a completely new stage of the pandemic.

585
00:57:54,510 --> 00:57:58,560
And Did your team work at home? Mostly because, I mean, most of your work is with computers.

586
00:57:58,710 --> 00:58:01,740
Yeah. Most of the work you don't. There's no work.

587
00:58:01,740 --> 00:58:07,110
There's no patients. Exactly. So, yeah, so the only value, which is a lot,

588
00:58:07,320 --> 00:58:15,510
but the only value of being here in the partner Research Centre where we work is that we meet all day long, we go to each other to ask questions.

589
00:58:15,510 --> 00:58:22,829
There's much more opportunity for solving problems together and and for brainstorming and things like that.

590
00:58:22,830 --> 00:58:26,640
But the work can be done mostly online from home, really.

591
00:58:27,840 --> 00:58:35,459
So yeah, I mean, the way that we knew this was coming and there was going to be a lockdown, I basically told everyone, Feel free to work from home.

592
00:58:35,460 --> 00:58:38,850
And, you know, I knew I knew that that was coming.

593
00:58:38,850 --> 00:58:44,970
And me what's going to happen, you know, earlier or later. So so we all worked from home.

594
00:58:45,480 --> 00:58:51,990
The team was impressive in their, you know, in how they worked and delivered everything.

595
00:58:51,990 --> 00:59:00,000
Despite the massive change and adaptation, I have a lot of colleagues who are, you know, from other parts of the world.

596
00:59:00,000 --> 00:59:06,000
Some of them went back home or, you know, or what their families wear, and that still worked out really well.

597
00:59:07,650 --> 00:59:15,150
So yeah, I think it demonstrated the resilience of, you know, a team of people that really wanted to make a difference, despite all the odds.

598
00:59:15,750 --> 00:59:23,390
For me, uh, maybe it made personal, but for me particularly was also a way to kind of keep busy, you know,

599
00:59:23,430 --> 00:59:30,030
sometimes of feeling useful and, and probably as a physician myself, you know, feeling that you're doing something to help.

600
00:59:30,540 --> 00:59:33,029
Did you consider going back into clinical practice?

601
00:59:33,030 --> 00:59:40,409
I considered that every single day, and I hadn't been involved in the management of an acute patient for a very long time,

602
00:59:40,410 --> 00:59:43,650
so I knew I wasn't the best person to do that, really.

603
00:59:44,130 --> 00:59:49,740
And I knew that there weren't that many people in the world who could do the research we weren't doing.

604
00:59:49,740 --> 00:59:55,830
So I you know, but I still every day I woke up in the morning thinking, should I just, you know, go and help in the hospital.

605
00:59:55,830 --> 01:00:04,140
Right. And there was a point in January 21 or December 20, maybe when I was very close to saying, yeah, this is the right thing to do now.

606
01:00:06,120 --> 01:00:09,149
But yeah, I mean, I managed to help clinically in other ways.

607
01:00:09,150 --> 01:00:17,820
So I was helping with the rheumatology here and all the junior doctors of younger people who probably are, you know,

608
01:00:17,850 --> 01:00:24,089
more up to date with the management of acute patients despite being rheumatologists were deployed in the cold wars and we were

609
01:00:24,090 --> 01:00:32,550
basically covering mix so that you know in a way I was doing my bit clinically and did you have to get into full PPE and everything?

610
01:00:32,640 --> 01:00:36,480
Was it was it was it it was mostly this hospital COVID free zone?

611
01:00:36,720 --> 01:00:43,799
Yeah, it was this hospital, the National Orthopaedic Centre is, you know, it was relatively safe space or all the time,

612
01:00:43,800 --> 01:00:48,360
but also a lot of the rheumatology work could be done online because sometimes, you know,

613
01:00:48,510 --> 01:00:52,620
it's a consultation that's more about, you know, you have inflammation in your joints and so on.

614
01:00:52,920 --> 01:00:56,909
I'm not saying it's the same as seeing a patient face to face, but, you know, again,

615
01:00:56,910 --> 01:01:03,780
considering the risks and the benefits of seeing someone relatively, you know, with on treatment with immune suppressants and the kind of treatment.

616
01:01:03,880 --> 01:01:07,960
Use in the clinic wasn't desirable.

617
01:01:07,990 --> 01:01:11,590
Also, you know, most patients didn't want it either at the time.

618
01:01:13,240 --> 01:01:23,950
So, yeah, you know, I had that every day. But but in a way, the amount of research I was doing was also so, you know, so, so big that I was busy.

619
01:01:23,950 --> 01:01:30,720
I was working every single weekend. Yeah. I thought that was the other thing I was going to ask about how your hours changed.

620
01:01:31,320 --> 01:01:34,420
I mean, I'm sure you work long hours anyway, but yeah, I mean,

621
01:01:34,420 --> 01:01:41,380
it's it's hard to say how many hours we work because we don't have a kind of office, especially for us or, you know, mostly MIA.

622
01:01:42,970 --> 01:01:46,410
But at that time it was crazy and unsustainable.

623
01:01:46,420 --> 01:01:50,080
Clearly, it was just something temporary that we thought, you know,

624
01:01:50,110 --> 01:01:55,210
I guess we we kind of kept going because we thought it would be another two weeks and then it would be another two months.

625
01:01:55,780 --> 01:02:01,149
But I guess recognition should go to our families and the people who supported us during that period,

626
01:02:01,150 --> 01:02:04,420
because in my case, it was yeah, it was very, very crazy.

627
01:02:04,420 --> 01:02:10,569
I still remember the first time I was watching a movie on the weekend with my kids and feeling guilty that I wasn't working,

628
01:02:10,570 --> 01:02:16,990
you know, And I was like, well into 2021 already or the previous year.

629
01:02:16,990 --> 01:02:21,740
And it was basically just work as much as you can because this needs doing.

630
01:02:21,740 --> 01:02:24,150
Before yesterday, everything COVID was urgent, right?

631
01:02:24,160 --> 01:02:30,160
There wasn't anything that you could, you know, basically try and and, and take more time to produce.

632
01:02:30,730 --> 01:02:33,730
So yeah, that was clearly unsustainable and desirable.

633
01:02:33,730 --> 01:02:38,680
But the only thing we could do at the time and you said you hinted that the the

634
01:02:38,680 --> 01:02:42,310
fact that you had an important project to work on did support your wellbeing.

635
01:02:42,970 --> 01:02:47,799
Yes. So for me it was important. And I and I think if you if you interviewed my partner, she would say the same.

636
01:02:47,800 --> 01:02:50,980
It was my I guess it was one of my May I also exercise talent.

637
01:02:51,040 --> 01:02:57,980
I didn't did Yeah. We were able to but it was clearly a mechanism for me to Yeah.

638
01:02:58,000 --> 01:03:05,020
To basically, you know think more of the others than of myself you know, and, and create this knowledge that was heavily needed.

639
01:03:05,530 --> 01:03:09,360
And rather than worrying about, you know, what happened if I get the virus.

640
01:03:09,370 --> 01:03:16,690
Um, also it's true that of course working from home as I could do for all the work that we were doing, felt very safe, really.

641
01:03:17,170 --> 01:03:21,580
Uh, although sometimes boring and, you know, and not very interesting.

642
01:03:22,840 --> 01:03:29,290
Yeah, I think we more or less got to the end with COVID. Just about covered everything there.

643
01:03:30,100 --> 01:03:40,570
Um, so yes, final question. Has the experience of working on COVID related research topics changed your thinking about how you work and what you do?

644
01:03:41,320 --> 01:03:44,750
And is there anything you'd like to see change in the future? It has.

645
01:03:44,830 --> 01:03:50,710
It has a lot. It has changed the way we do things in my team quite fundamentally, actually.

646
01:03:51,220 --> 01:03:57,010
So in the past we were more reactive and I think we are now much more proactive.

647
01:03:57,370 --> 01:04:05,860
Like we want to make sure that instead of having a question and then writing analytical code to crunch the numbers,

648
01:04:06,310 --> 01:04:12,580
we now are producing software that will be ready to go to produce that data that we will need.

649
01:04:13,420 --> 01:04:21,520
Basically with, you know, a line of code and a click rather, and, you know, three days of work coding and then quality checking and all that.

650
01:04:21,880 --> 01:04:29,920
So we've moved into a space that probably is closer to, I guess, software engineering where we try to have pipelines, as we call them,

651
01:04:30,190 --> 01:04:36,370
to be able to produce those data or that information very quickly in a very reliable manner,

652
01:04:36,370 --> 01:04:40,480
rather than being reactive and saying, Oh, now we need to do this, let's see how we do it.

653
01:04:41,440 --> 01:04:46,690
So that's we've been able to find commonalities between different kinds of questions and yes,

654
01:04:46,960 --> 01:04:50,710
streamline your your software and the standardisation of data.

655
01:04:50,810 --> 01:04:55,210
The mapping of data that we discussed earlier facilitates that a lot because, you know,

656
01:04:55,570 --> 01:05:01,960
if you if you know that the data is always going to look like this, you can always write code towards that format, right?

657
01:05:02,200 --> 01:05:04,660
And then tested in different datasets and so on.

658
01:05:05,050 --> 01:05:16,720
So it has changed fundamentally the way we do, you know, have data sciences, as we call them, to basically be faster and more scalable, if you like.

659
01:05:16,780 --> 01:05:19,990
Scalability is probably the word that we use in this community.

660
01:05:20,410 --> 01:05:24,010
Mm hmm. Terrific. Thank you very much.