1 00:00:00,810 --> 00:00:06,240 Welcome to the Oxford University Psychology Podcast series, today I have Andrea Cypriote, 2 00:00:06,240 --> 00:00:12,870 who's associate professor here at Oxford University and editor of the Journal for evidence based Mental Health. 3 00:00:12,870 --> 00:00:17,790 Good afternoon, Andrea. Good afternoon. Thank you for coming today. 4 00:00:17,790 --> 00:00:29,640 Your research is about Masroor Analysis Network, Métro analysis about synthesising data to provide helpful information to to doctors. 5 00:00:29,640 --> 00:00:36,390 So maybe we could begin by just talking about how you got here in the first place. 6 00:00:36,390 --> 00:00:47,460 How did you start thinking about these these things? Evidence synthesis is mainly what I'm dealing with in terms of my research. 7 00:00:47,460 --> 00:00:57,120 And I started many years ago, about 11 years ago when I was here in Oxford as a resident in psychiatry. 8 00:00:57,120 --> 00:01:02,610 I trained in Italy, in Verona, but I spent nine months here in Oxford. 9 00:01:02,610 --> 00:01:11,830 And I have to say that it was John Carlos who inspired me and told me all about evidence synthesis. 10 00:01:11,830 --> 00:01:22,590 So that was where it started. OK, and you began by looking at meta analysis of medications. 11 00:01:22,590 --> 00:01:35,430 Is that right? Yes. I like the idea of doing something systematic and so collect to have a comprehensive use of all available evidence, 12 00:01:35,430 --> 00:01:43,860 but at the same time trying to synthesise all these evidence based studies into one pooled estimate. 13 00:01:43,860 --> 00:01:52,900 Because this is something that can really inform clinicians because you have to have one fear to work with and not a lot of different small studies. 14 00:01:52,900 --> 00:01:56,550 So that's the thing I like in evidence synthesis. 15 00:01:56,550 --> 00:02:04,380 So methodologically rigorous, but at the same time having a bottom line message to give to clinicians. 16 00:02:04,380 --> 00:02:09,240 Can you give us an example of that? Well, I have in mind two example. 17 00:02:09,240 --> 00:02:20,610 One was the piece of research I did in 2000 and three and four, and it was published in 2005 with John and Keith Horton about suicide and lithium. 18 00:02:20,610 --> 00:02:32,040 So what we had was a old and quite a numerous, many trials about lithium in affective disorder, mood disorders, lithium versus placebo. 19 00:02:32,040 --> 00:02:35,670 Even the most famous have been lithium versus other drugs. 20 00:02:35,670 --> 00:02:48,090 And we focus on the prevention of suicidality, which means suicide and deliberate self-harm in people allocated to lithium or other compounds. 21 00:02:48,090 --> 00:02:55,560 And what we found here was that even if lithium, the difference was not significant at the state level, 22 00:02:55,560 --> 00:03:07,370 when we pulled all the studies together, we found a striking effect of lithium in preventing both suicide and also the level of self-harm. 23 00:03:07,370 --> 00:03:14,700 The second example is many years later, when we address the issue of antidepressant for major depression, 24 00:03:14,700 --> 00:03:24,630 what we wanted to have was not a simple standard metabolises A versus B, but we wanted to compare everything versus everything. 25 00:03:24,630 --> 00:03:31,650 And that's the so-called network with analysis and collecting data. 26 00:03:31,650 --> 00:03:42,330 It was about 27000 patients, 120 studies and having a clinical clear bottom line that something was better than other drugs or 27 00:03:42,330 --> 00:03:48,600 other drugs were clearly worse than others was was really a good example of the powerful evidence. 28 00:03:48,600 --> 00:03:56,370 And you can have in this field as a psychiatrist working with these medications, 29 00:03:56,370 --> 00:04:04,380 I would definitely use that evidence in my practise and discussed the evidence that you've published with with patients directly. 30 00:04:04,380 --> 00:04:08,460 And it's been very helpful to have that data synthesis. 31 00:04:08,460 --> 00:04:12,030 But if this isn't an easy thing to do, 32 00:04:12,030 --> 00:04:20,110 is that this isn't something that you can just take off the shelf and and do as a staff research because the methodology is very complicated. 33 00:04:20,110 --> 00:04:22,440 So, I mean, 34 00:04:22,440 --> 00:04:31,780 how is it that you've developed these these techniques that can test such a vast array of different interventions and compare them with each other? 35 00:04:31,780 --> 00:04:36,750 It's a minefield because it's very, very easy to make mistakes. 36 00:04:36,750 --> 00:04:47,400 It started many years ago when I met Georges Allante, a statistician coming from with a strong background in mathematics, pure mathematics. 37 00:04:47,400 --> 00:04:52,230 And I talked to her about which was the clinical problem. 38 00:04:52,230 --> 00:04:59,740 I had so many antidepressants and what to do. Is there anything we can use in terms of evidence synthesis to. 39 00:04:59,740 --> 00:05:11,890 Grasp a very robust message, so we started developing the methodology, it was a methodology used in other fields of medicine, Nevine Psychiatry. 40 00:05:11,890 --> 00:05:16,330 So it was a sort of pioneering the field. 41 00:05:16,330 --> 00:05:22,550 And we did it with many trials and errors. And in the end, I learnt the methodology. 42 00:05:22,550 --> 00:05:29,080 But what I know now is that you need a proper statistician to do this afterimage analysis. 43 00:05:29,080 --> 00:05:35,860 And also a statistician skilled in natural talent is not a simple statistician on. 44 00:05:35,860 --> 00:05:43,390 The second thing is it's very important to be absolutely sure about the results because when you do electrometer analysis, 45 00:05:43,390 --> 00:05:53,230 it's a process three to four years every instrumentalise, you collect the data and the statistician runs all the analysis of the syntaxes. 46 00:05:53,230 --> 00:06:01,270 But many times the first, the prelaw preliminary results are wrong because so there's a it's not because as team work, 47 00:06:01,270 --> 00:06:07,270 you have to have the knowledge of the clinical understanding of the methods and they apply. 48 00:06:07,270 --> 00:06:15,400 The statisticians are playing with numbers. And what I do is to do it blind so they don't know the names of the drugs. 49 00:06:15,400 --> 00:06:20,680 They know the ABC and drug Ajram B2C. So it's completely blind. 50 00:06:20,680 --> 00:06:23,470 But when we unblind the results, 51 00:06:23,470 --> 00:06:33,730 we need to talk with statisticians and clinicians of different with different expertise to understand whether it is correct or not. 52 00:06:33,730 --> 00:06:40,010 It sounds like you've learnt a lot about the process of developing that methodology. 53 00:06:40,010 --> 00:06:43,420 What would you say or what advice would you give? 54 00:06:43,420 --> 00:06:47,350 You mentioned that there's there's not always good Métro analysis out there. 55 00:06:47,350 --> 00:06:57,100 There are some bad meta analysis how to tell the difference. I think all clinicians should be able to understand the difference. 56 00:06:57,100 --> 00:07:06,340 And there are a few key issues. One is, first of all, to look at the evidence, the primary evidence, primary studies, 57 00:07:06,340 --> 00:07:11,680 which means table one, the table of the paper should report all the studies included. 58 00:07:11,680 --> 00:07:17,440 And it's a very useful starting point to understand whether that methodology is 59 00:07:17,440 --> 00:07:23,140 really answers the question that they report at the beginning of the paper. 60 00:07:23,140 --> 00:07:31,930 The second thing is how they report the data, because as long as the data in the first place are reported in a transparent way, 61 00:07:31,930 --> 00:07:38,830 which means replicable, we can you can you have the figures, the numbers, the denominators, and you can redo the analysis. 62 00:07:38,830 --> 00:07:44,380 This is a good proxy that they didn't respond way and it's in a reliable way. 63 00:07:44,380 --> 00:07:53,900 The other thing is the journal. So a good journal tends to publish very good stuff, but it is not the case. 64 00:07:53,900 --> 00:07:59,890 So it's not necessarily the case. It's not always the case. So really what you're doing is, 65 00:07:59,890 --> 00:08:10,540 is providing clinicians with a series of very helpful studies that can synthesise a lot of data and sort of give give clinicians an answer. 66 00:08:10,540 --> 00:08:17,410 Use this antidepressant medication. Use this dose of antipsychotic, for instance. 67 00:08:17,410 --> 00:08:21,130 So that's really the building blocks of evidence based medicine. 68 00:08:21,130 --> 00:08:28,900 And I was just wondering what you think is of evidence based medicine, more evidence based practise that's occurring in psychiatry. 69 00:08:28,900 --> 00:08:36,250 What do you think of it? Do you think it's this happening? Do you think psychiatrists are operating within the evidence base? 70 00:08:36,250 --> 00:08:44,350 Well, I think this is the best framework we have, even though it's not ideally the optimal situation. 71 00:08:44,350 --> 00:08:51,550 But evidence based practise, evidence based medicine is what I'd like to have as a patient. 72 00:08:51,550 --> 00:09:01,240 And I want I'd like my psychiatrist and my physician to use this kind of approach because it's a combination of the clinical circumstances, 73 00:09:01,240 --> 00:09:10,330 circumstances, the patient's values and preferences and the best available evidence, which means not only the most robust evidence, 74 00:09:10,330 --> 00:09:17,920 but also the most up to date, because the results change over time, may change over time. 75 00:09:17,920 --> 00:09:25,000 I think there's a gap in the implementation of evidence based data into practise. 76 00:09:25,000 --> 00:09:33,700 And I think psychiatry is particularly difficult because we might be psychiatrists, we might be a bit ideological. 77 00:09:33,700 --> 00:09:40,750 On one hand, there's a lot of debates about more ideological things rather than real clinical problems. 78 00:09:40,750 --> 00:09:47,680 And what I also worried is that some people have a too simplistic approach. 79 00:09:47,680 --> 00:09:57,880 They tend to simplify too much. So we I like the idea of having a bottom line message, but at the same time, clinical practise is so complex. 80 00:09:57,880 --> 00:10:08,520 But you need to have some. Points and references, but the actual decision is between the Medick or the mental health professional and the patient, 81 00:10:08,520 --> 00:10:17,280 some of your recent publications have been about the doses of antipsychotics and antidepressants. 82 00:10:17,280 --> 00:10:21,930 Do you think that's that sort of level of detail should be left up to clinical 83 00:10:21,930 --> 00:10:27,570 practise and and titrating doses according to the individual patients they see? 84 00:10:27,570 --> 00:10:40,020 Or should this all be regulated through guidelines? I think what medicine and psychiatry, the degree of freedom of clinician has to be preserved. 85 00:10:40,020 --> 00:10:44,880 Definitely what we wanted to address is a twofold question. 86 00:10:44,880 --> 00:10:50,860 One is, can we give a sort of general advice about doses, 87 00:10:50,860 --> 00:10:57,690 say not to the press and also in terms of comparability between different depressants and different those, 88 00:10:57,690 --> 00:10:58,530 on the other hand, 89 00:10:58,530 --> 00:11:06,930 is the never ending story of the clinical trial using different doses versus placebo favouring the comparator of investigational drugs. 90 00:11:06,930 --> 00:11:15,270 So it was also a work we wanted to do to clarify some issues in terms of regulatory policies. 91 00:11:15,270 --> 00:11:25,380 So a bit of both really, Muzaffer, clinical expertise, but we need to have the evidence base to back it up in your understanding of, you know, 92 00:11:25,380 --> 00:11:31,800 expert knowledge about the way the current evidence is for prescribing and in mental health, 93 00:11:31,800 --> 00:11:39,840 what would you say the gaps are or where do you think we need to understand more at the moment what we are trying to do? 94 00:11:39,840 --> 00:11:47,640 And when I say we is a group of people, a network of people interested in this synthesis and measurement analysis, 95 00:11:47,640 --> 00:11:58,710 we are trying to summarise all the evidence for pharmacological but also non pharmacological interventions in the main disorders in psychiatry, 96 00:11:58,710 --> 00:12:04,920 unipolar depression, bipolar disorder, schizophrenia, PTSD, anxiety, panic disorder. 97 00:12:04,920 --> 00:12:15,840 And the idea is then to use this data to build a sort of algorithm, because what we can do is to use the natural metabolises technique, 98 00:12:15,840 --> 00:12:21,000 not only with summary data from primary studies, but also with individual patient data. 99 00:12:21,000 --> 00:12:28,890 And we use the individual patient data. We can build these algorithms, try to find the treatment indication, 100 00:12:28,890 --> 00:12:36,840 according to the number of people, that the severity of illness and know country or whatever it is. 101 00:12:36,840 --> 00:12:50,160 And then the big plan is to start monitoring real patients, real world settings, monitoring outcome after prescribing treatments. 102 00:12:50,160 --> 00:12:56,340 According to this algorithm, at the end of a follow up period could be months or years. 103 00:12:56,340 --> 00:13:04,290 We can use this data via a learning machine process to implement and feedback the original algorithm. 104 00:13:04,290 --> 00:13:16,320 And so the idea is to overcome the dichotomy between randomised data and observational data and try to merge them into a real world scenario. 105 00:13:16,320 --> 00:13:20,670 So starting from randomised evidence, having treatment indications, 106 00:13:20,670 --> 00:13:27,840 then observational data to reinforce implement change, which is the original thing. 107 00:13:27,840 --> 00:13:30,540 So we you're suggesting this is actually quite radical. 108 00:13:30,540 --> 00:13:38,340 You're suggesting that we might begin to have evidence about what medications a patient should use if they're female, 109 00:13:38,340 --> 00:13:44,760 if they're middle aged, if they're planning on what race they are. 110 00:13:44,760 --> 00:13:50,020 So all this kind of all this kind of information is going to change clinical practise quite significantly? 111 00:13:50,020 --> 00:13:56,520 Well, I think it's what we as doctors and what we what we see in our clinics almost every day. 112 00:13:56,520 --> 00:14:00,580 So people respond differently. People do not respond. 113 00:14:00,580 --> 00:14:10,650 So and everybody has different ideas. So if we can collect all this data and look at the data we have already collected, that's really informative. 114 00:14:10,650 --> 00:14:21,510 The potential and we have examples in neurology stroke cancer anthologies are using data of previous studies. 115 00:14:21,510 --> 00:14:24,430 They found incredibly powerful differences. 116 00:14:24,430 --> 00:14:37,020 Try to find treatments according to the degree of to nose's to gender differences, no age or different things, I think. 117 00:14:37,020 --> 00:14:40,360 Yes, in psychiatry we need innovation. We need new treatment. 118 00:14:40,360 --> 00:14:50,040 We need to and academia is should have this mission trying to discover new treatment and better treatment for our patients at the same time, 119 00:14:50,040 --> 00:14:55,740 or the old treatments like lithium or other drugs we know are really effective. 120 00:14:55,740 --> 00:14:59,390 We need to understand why they work, but at the same time try. 121 00:14:59,390 --> 00:15:10,670 To understand how we can use that better, it sounds really interesting that that that work and I look forward to hearing more results in the future, 122 00:15:10,670 --> 00:15:17,840 but maybe we can end by just some thoughts from you about, 123 00:15:17,840 --> 00:15:22,790 well, what you'd like to say to potential junior psychiatrists or medical students who 124 00:15:22,790 --> 00:15:28,730 might be interested in a career in academic psychiatry or career in psychiatry. 125 00:15:28,730 --> 00:15:33,890 What what advice or words of wisdom would you have for them? 126 00:15:33,890 --> 00:15:40,730 I would encourage anyone to to do this kind of career because it's fascinating. 127 00:15:40,730 --> 00:15:46,430 It's exciting. And it's also very important to find someone. 128 00:15:46,430 --> 00:15:56,640 I was lucky to find someone who inspired me and taught me a lot of things about which is the mission of people in academia. 129 00:15:56,640 --> 00:16:01,970 The you have to be free thinkers, a completely open mind. 130 00:16:01,970 --> 00:16:12,500 At the same time, there's a dose of courage and investment in things you think are important and not necessarily what the average people say. 131 00:16:12,500 --> 00:16:22,820 So to be quite innovative in your way of thinking and doing and this is from the personal point of view, 132 00:16:22,820 --> 00:16:32,150 from a more general point of view, I think we people who work in the academia and who try to work with the younger generation, 133 00:16:32,150 --> 00:16:42,170 we need to fill these gaps and we need to be better example, better models, because there's a lot of bad models and it's difficult to find someone. 134 00:16:42,170 --> 00:16:48,770 And it's also a matter of availability time to have the time to spend with the younger generation. 135 00:16:48,770 --> 00:16:56,720 As editor of the journal Evidence based Mental Health, what I want is to be able, 136 00:16:56,720 --> 00:17:02,660 with a boring thing like evidence based synthesis and data synthesis, 137 00:17:02,660 --> 00:17:09,920 tried to show how rigorous a methodology can inspire clinical practise on an everyday basis. 138 00:17:09,920 --> 00:17:14,780 And what I'd like to do is engage younger generation with examples. 139 00:17:14,780 --> 00:17:20,000 And this is the reason why we started the Google Hangouts to treat a child's new use, 140 00:17:20,000 --> 00:17:28,790 the social media to engage young people because we need someone with fresh ideas and with the energy. 141 00:17:28,790 --> 00:17:33,260 Interesting. Great speech Tuesday. Thank you for your time. Thanks. Thank you. 142 00:17:33,260 --> 00:17:38,930 Thank you for tuning in to another podcast with the Oxford University Psychology Podcast series. 143 00:17:38,930 --> 00:17:40,284 Goodbye.