1 00:00:00,090 --> 00:00:05,130 Hi, everybody. For those I know some of you, but for those of you, I don't know. 2 00:00:05,160 --> 00:00:11,610 My name is Carl Hennigan. I am professor of evidence based medicine and director of the Centre for Evidence based Medicine. 3 00:00:11,610 --> 00:00:14,910 And welcome to many of you to the MFC course. 4 00:00:15,390 --> 00:00:19,230 The title of this talk is What is evidence based medicine done for Health Care? 5 00:00:19,590 --> 00:00:22,560 Well, let's talk through this. It's a sort of illustration. 6 00:00:22,830 --> 00:00:30,840 We often forget about some of the historical perspective of evidence based medicine and how the evidence evolved and what that means. 7 00:00:31,470 --> 00:00:34,500 And so I'm going to particularly just focus on the idea of heart attack. 8 00:00:34,500 --> 00:00:37,980 And you've had a busy day today. Question formulation. 9 00:00:38,560 --> 00:00:43,650 What an abstract thinking about some of the issue searching. And then tomorrow are you going to look at some treatment effects? 10 00:00:43,650 --> 00:00:47,850 So some of this will be really relevant to that. Just to move on to a couple of things. 11 00:00:49,260 --> 00:00:53,430 If you go to the CBM site, I want to do talks now and we're all going to do this. 12 00:00:53,430 --> 00:00:58,860 If you go that there's a blog site and on the blog site here, you can see there's a particular one there. 13 00:00:58,860 --> 00:01:01,590 What has been done for health care? And if you go to that site, 14 00:01:01,800 --> 00:01:08,370 I've actually posted this talk with a summary of the talk and a downloadable PDF of the talk so you 15 00:01:08,370 --> 00:01:12,810 can pull it all off and afterwards and think when you've got nothing to do later on this evening, 16 00:01:13,500 --> 00:01:16,800 you can have a bit of bed night reading and think, Wow, that's fantastic. 17 00:01:17,010 --> 00:01:22,889 Not only that, if you go to the blog at the end of the blog, I you can leave a response or comment. 18 00:01:22,890 --> 00:01:26,250 So at the end of this, we're going to talk about some of the issues, maybe 20 minutes. 19 00:01:26,340 --> 00:01:34,979 We're going to think about just some of the issues. So this talk, basically, I'm going to just take you through this is a timeline from about 1960, 20 00:01:34,980 --> 00:01:37,920 1970, but actually some bits of it go back hundreds of years. 21 00:01:38,220 --> 00:01:44,100 And when you start to think about something just as simple as a heart attack, there are loads of really interesting points in there. 22 00:01:44,550 --> 00:01:54,180 This is the first report in The Lancet. The thrombotic properties of aspirin was published in 6774 with the first ever aspirin trial. 23 00:01:54,780 --> 00:01:57,959 You can move forward here. Editorial by Richard Peto. 24 00:01:57,960 --> 00:02:06,570 I'm going to talk about that. In the 1980s, the in-hospital mortality of a heart attack with about 13% and there's been a whole raft of publication, 25 00:02:06,570 --> 00:02:09,690 31 randomised trials of aspirin there in 1988, 26 00:02:10,050 --> 00:02:14,160 six years later, 145 trials and then moving through, 27 00:02:14,160 --> 00:02:19,200 there's been a raft of evidence carrying on again that brings you right up to date where you can come through. 28 00:02:19,200 --> 00:02:26,280 And I'm going to touch on all of these ACE inhibitors or little bit of beta blockers and some really interesting stuff that's happening right now, 29 00:02:26,520 --> 00:02:34,290 like the use of oxygen in heart attack and the use of adrenaline in clinical a clinical trial that's going on now. 30 00:02:34,560 --> 00:02:41,090 So this actually gives a whole structure when people say to me for many new treatments, what should the evidence base for? 31 00:02:41,090 --> 00:02:45,090 A heart attack is actually almost a seminal understanding of ebben. 32 00:02:45,510 --> 00:02:49,169 So let's move on. First ever trial aspirin, 1974. 33 00:02:49,170 --> 00:02:53,490 It's not that long ago, is it? Only 40 years ago, indeed. 34 00:02:53,760 --> 00:02:57,270 And this is by Peter Ellwood here, published in the British Medical Journal, 35 00:02:57,510 --> 00:03:03,180 a randomised trial of acetyl salicylic acid in the secondary prevention of mortality from me. 36 00:03:04,110 --> 00:03:13,880 But interestingly when you do look at the timeline of aspirin that the first trial of aspirin 1996 FDA approves aspirin for use in suspected I. 37 00:03:14,510 --> 00:03:22,180 That is not that long ago is it. But if you do go back well what was happening if you go right back in time to about 1758, 38 00:03:22,200 --> 00:03:26,520 you get this chap called Reverend Edmund Stone, who was a fellow down here at Wadham. 39 00:03:26,790 --> 00:03:34,229 He found out about the bark of a willow tree in his report in 1763 of the many youthful discoveries 40 00:03:34,230 --> 00:03:38,879 and down here I have found by experience to be a powerful astringent and very efficacious, 41 00:03:38,880 --> 00:03:44,160 incurring anguish and intermittent disorders that fever, I think, you know, language. 42 00:03:45,390 --> 00:03:54,930 But all of that basic sciences for 200 plus years were set on this discovery, and it took till 1974 to move forward to clinical trials. 43 00:03:55,320 --> 00:03:57,180 And when you look at it, it's actually about here. 44 00:03:57,180 --> 00:04:05,520 This is in Mississippi in about 1950s, the first reports of observational studies that were starting to say, 45 00:04:06,060 --> 00:04:09,990 actually, we've got people on aspirin, they don't seem to have a heart attack. 46 00:04:10,320 --> 00:04:16,620 And then beyond that, about 1967 started to see reports about the platelet aggregation effect. 47 00:04:16,920 --> 00:04:20,370 And particularly some of these effects were not aggregated. 48 00:04:20,370 --> 00:04:27,959 Normally, when this is put here, the bleeding tendency after aspirin ingestion that the drug may have anti drop from both properties. 49 00:04:27,960 --> 00:04:32,760 So we're in the sixties now people going we're doing the basic science 200 years after the drug 50 00:04:32,760 --> 00:04:39,030 discovery and then other said we moved to this trial and this was very profound at its time. 51 00:04:39,420 --> 00:04:42,390 And interestingly, if you look at this trial, it's an interesting one. 52 00:04:42,870 --> 00:04:48,839 If you move just a few years later, this is the first trial here, MRC one trial, 1974, 53 00:04:48,840 --> 00:04:55,500 randomised 1239 patients and many of you tomorrow will have heard of the Cochrane Library. 54 00:04:56,190 --> 00:04:59,940 And here's a slide that Peter Elwood and Archie Cochrane you. 55 00:05:00,000 --> 00:05:05,549 The take round, didn't have Internet in them days, didn't have Twitter, didn't have all the TV. 56 00:05:05,550 --> 00:05:11,430 So this was their slide for the dissemination at meetings about the effects of aspirin 57 00:05:11,580 --> 00:05:15,690 when they have a whole bunch of doctors trying to disseminate clinical trial planning. 58 00:05:17,040 --> 00:05:25,590 And what you can see that is they're proposing the thick first trial, all six trials had 10,859 patients. 59 00:05:26,160 --> 00:05:34,620 And the weighted overall effect of aspirin with a 23% reduction in all cause mortality and the P values 0.001. 60 00:05:36,060 --> 00:05:41,970 So that was their way of saying and if you think about it, this is really before Cochrane Cochrane came about six, 61 00:05:43,020 --> 00:05:52,320 15 years later, in effect in thinking this is the start of why we need meta analysis, why we need six systematic reviews. 62 00:05:53,790 --> 00:05:57,659 And moving on, I'm just going to put this up because at the time, Richard Peto, 63 00:05:57,660 --> 00:06:02,129 who's here in Oxford, published an editorial in the month in The Lancet. 64 00:06:02,130 --> 00:06:08,820 And Jeff nodding there, because Jeff has probably read The Lancet in 1980, but most people in this room did not read The Lancet. 65 00:06:08,880 --> 00:06:15,360 I was still at school from a book, but actually he put this and said, Oh, this is a really important stuff. 66 00:06:15,690 --> 00:06:22,110 Large trial. Putting them together gives you a systematic effect, how you look at treatment effect. 67 00:06:22,500 --> 00:06:27,390 And importantly, if you take Richard Pinto's work, what he did is eight years later, 68 00:06:27,630 --> 00:06:36,150 then publish the overview of 51 trials in patients with a history of a stroke, unstable angina or prior asthma. 69 00:06:36,150 --> 00:06:39,150 And this is antiplatelet trials collaboration. 70 00:06:39,570 --> 00:06:44,550 He still exists today in Oxford and still work out the clinical trial for this unit. 71 00:06:45,660 --> 00:06:48,750 Really interesting and so published in the BMJ. 72 00:06:48,900 --> 00:06:53,760 And when you look at this, it's really interesting. If you take these articles, these old articles, boy, can you learn a lot. 73 00:06:54,330 --> 00:06:59,010 And if you look in the first point, two main purposes overview. 74 00:06:59,490 --> 00:07:04,139 They include far large number of patients and individual trials do and hence your results 75 00:07:04,140 --> 00:07:08,880 that are far less subject to randomised error compared to the single trial alone. 76 00:07:09,750 --> 00:07:15,900 Now that's an incredibly important issue when you think about some of the trials today that come out with one single trial, 77 00:07:16,260 --> 00:07:19,710 maybe manufactured, funded, maybe very small in nature. 78 00:07:20,100 --> 00:07:24,120 And what are you saying that if you do that you get a lot of random error around that result? 79 00:07:25,560 --> 00:07:29,970 That might become a bit more obviously tomorrow, but over a lifetime of evidence based medicine, 80 00:07:30,240 --> 00:07:36,210 this random error is really important when we sit there and go, Oh, this could be false with positive finding. 81 00:07:36,540 --> 00:07:39,329 The second issue is No, actually, well, I'll read this out. 82 00:07:39,330 --> 00:07:44,520 They are about the substantial systematic bias that may be engendered when dozens of related trials 83 00:07:44,520 --> 00:07:50,249 have been conducted and just a few become well known for trials may tend to become well known, 84 00:07:50,250 --> 00:07:55,230 partly because their results are unusually promising or unusually unpromising. 85 00:07:55,920 --> 00:08:03,780 And that issue is quite interesting. We would call this today if you take away the public, it's like publication bias in effect. 86 00:08:03,780 --> 00:08:13,040 But I consider this is in the olden days before the Internet area, there was a real marketing and bias towards positive results. 87 00:08:13,290 --> 00:08:17,220 So not quite publication both, but had the same effect with publication both. 88 00:08:17,850 --> 00:08:23,670 And that's a really interesting issue that we still suffer with today is about half of all trials go unpublished. 89 00:08:24,210 --> 00:08:27,210 There was not as many trials as it was then in about the seventies. 90 00:08:27,210 --> 00:08:32,280 It was probably a couple of thousand trials each year. Now we're talking about 40,000 trials each year. 91 00:08:32,970 --> 00:08:39,150 So if you take them results, you'll see tomorrow you'll learn about this is a forest plot and here down are all the single trials. 92 00:08:39,780 --> 00:08:48,030 These were the bigger arms of a wider confidence interval, the smaller trials, and these are the bigger trial with a smaller confidence. 93 00:08:48,030 --> 00:08:53,820 And if you follow this down, you start to get this 25% across all trials in all cause mortality. 94 00:08:54,270 --> 00:09:01,170 And that's the forest plot. That's the purpose of systematic reviews, allows you to give rise to a precise estimate. 95 00:09:02,850 --> 00:09:05,040 And moving on from that, I mean, it's really interesting. 96 00:09:05,040 --> 00:09:15,420 There was obviously an explosion in trials of aspirin because in a six year window, we went from 51 trials to 145 trials. 97 00:09:15,810 --> 00:09:22,140 That's a huge amount of trials. Now, if you think about it now, when we looked at drug treatment, we often seen I mean, 98 00:09:22,350 --> 00:09:26,399 we did some work in Tamiflu, which you'll all be aware of is a really important issue. 99 00:09:26,400 --> 00:09:32,850 And it's been a thick seven years in my life. But actually the number of trials in Tamiflu is actually incredibly small. 100 00:09:33,780 --> 00:09:40,050 It's about 14 treatment trials, four or five prophylaxis trials, and in in children's about four clinical trials. 101 00:09:40,890 --> 00:09:47,590 But here you see for a large effect, 545 12 I think there were about three learning points. 102 00:09:47,610 --> 00:09:52,470 And maybe the more one is it was easier to do trials then than it is now. 103 00:09:53,190 --> 00:09:59,820 We've made it much harder to do trials. The regulatory pathway means it's really difficult to get people in clinical trials and it. 104 00:09:59,890 --> 00:10:04,290 Cost a lot of money. So in the old days you would probably think 30 year old days. 105 00:10:04,300 --> 00:10:07,740 It sounds like you've done it 30 years ago. You're probably four. 106 00:10:07,750 --> 00:10:13,360 I'm not quite sure about the effect of this treatment. Maybe I'll set up a trial and do my own trial in my own hospital. 107 00:10:13,960 --> 00:10:20,440 You could probably get ethical approval by the end of the week, set up a simple randomisation system and probably run a trial within a couple of week. 108 00:10:20,890 --> 00:10:25,120 Fair enough. Just a little bit extra, maybe a few a month. 109 00:10:25,330 --> 00:10:27,880 But around the globe, you would be able to do it really simple. 110 00:10:28,510 --> 00:10:33,580 And you would just do a simple test, like flip a coin, maybe a to accept flipping a coin in the order. 111 00:10:34,090 --> 00:10:42,280 I think that's that's one to is the element of replication makes you less uncertain. 112 00:10:42,280 --> 00:10:48,670 In fact, if there's a bias in one or two trials, you will find that out with all the different trials that occur. 113 00:10:49,360 --> 00:10:57,040 And I think that's an important because if you did one large trial with 100,000 people and you had an inherent bias in the design of that trial, 114 00:10:57,460 --> 00:11:03,850 it would systematically affect all of that trial. Whereas if you do a number of independent trials, it's highly unlikely. 115 00:11:04,020 --> 00:11:07,370 For instance, they have one issue within the randomisation. 116 00:11:08,420 --> 00:11:14,059 And then I think my third point is to say having 145 trials and we'll see later allows you 117 00:11:14,060 --> 00:11:19,139 to do lots more interesting subgroup analysis if you can get the individual data to them. 118 00:11:19,140 --> 00:11:25,820 I three point a moving on. So aspirin still remained a really interesting topic I think throughout its lifecycle. 119 00:11:26,420 --> 00:11:31,910 In 2005, an editorial in the BMJ said that everybody over the age of 50. 120 00:11:32,270 --> 00:11:36,560 Not quite there yet but will be soon will have to take an aspirin. 121 00:11:37,040 --> 00:11:40,399 And that's a good idea. Indeed. You know, 122 00:11:40,400 --> 00:11:45,920 and it's interesting and I put this point that whatever comes about is the continual need 123 00:11:46,190 --> 00:11:50,300 to look at some of the issues about what we do and question what we do all the time. 124 00:11:51,020 --> 00:11:56,630 And that's what happened. The aspirin trial is collaboration have now changed their name to Antithrombotic trial. 125 00:11:56,650 --> 00:12:05,270 A collaboration of a mouthful looked at this in the primary and secondary prevention 2009 now. 126 00:12:05,330 --> 00:12:10,489 So this is only a few years ago and asked this question about some of these effects. 127 00:12:10,490 --> 00:12:16,280 And so if you look at the effects in primary and secondary prevention, what happens is if you take, say, for instance, 128 00:12:16,280 --> 00:12:23,359 males over the age of age, 52 and 59, in secondary prevention, you get this vascular death quite a big difference. 129 00:12:23,360 --> 00:12:27,419 But if you look at primary prevention, the difference is only about 0.5 percent. 130 00:12:27,420 --> 00:12:33,530 The absolute the difference. You could still do a relative effect and you still see that's great in the 10%. 131 00:12:33,530 --> 00:12:37,960 So relative looks okay but is offset by this increase in GI bleed. 132 00:12:38,480 --> 00:12:45,200 So any small reduction is offset by the increase in gastrointestinal bleeds and that's exactly the outcome. 133 00:12:45,500 --> 00:12:48,800 So we do not recommend aspirin in primary prevention. 134 00:12:49,160 --> 00:12:55,910 Ten years ago we were going to give it everyone. But because of this difference between primary and secondary effect, absolute effect, not relative. 135 00:12:56,570 --> 00:13:02,330 It's a really important issue to understand. When you look at absolute and at the baseline, risk comes lower. 136 00:13:02,840 --> 00:13:04,130 You often get this problem. 137 00:13:04,790 --> 00:13:11,420 We could turn and say, that's exactly the same issue we're now looking at with cholesterol lowering advice that we've had in the UK. 138 00:13:11,630 --> 00:13:15,740 What we want to do with lower the risk which will give treatment. 139 00:13:15,740 --> 00:13:19,760 And people are probably thinking, well, haven't we been there before to aspirin? 140 00:13:19,760 --> 00:13:24,640 The really interesting. Okay, I'm going to move on now. Aspirin is a quick tool. 141 00:13:24,650 --> 00:13:26,900 Here's another one. And I want you to think of as I'm talking. 142 00:13:27,170 --> 00:13:34,640 Imagine if we didn't have any evidence based medicine and we just decided we were going to do treatment based on our own observations, 143 00:13:34,970 --> 00:13:40,640 our own experiential learning, because that's what we did feel about the seventies, in effect, because with no clinical trial, 144 00:13:40,970 --> 00:13:45,710 you come in with your heart attack and I might make a decision probably up until about 1982. 145 00:13:45,980 --> 00:13:48,850 I'm just going to treat you as I think my peers would treat you. 146 00:13:49,100 --> 00:13:53,960 So think, although I wonder what it would look like the world if we didn't have an IBM approach. 147 00:13:54,560 --> 00:14:00,800 So this is a book by Thomas Moore Why tens of Thousands of Heart Patients Died in America's Worst Drug Disaster. 148 00:14:01,100 --> 00:14:01,969 Very good book. 149 00:14:01,970 --> 00:14:08,930 You can buy it for a few pounds, about £7 on Amazon, although every time I recommend it, the price goes up for a few weeks and then it comes down. 150 00:14:10,550 --> 00:14:12,320 You will have heard of the case trial, 151 00:14:12,620 --> 00:14:20,120 the really interesting trial in effect because of what it revealed in terms of excess mortality for these drugs. 152 00:14:20,120 --> 00:14:25,120 Antiarrhythmic And if you look at it, it's really profound. 153 00:14:25,170 --> 00:14:30,410 You could argue that this trial started the whole of the evidence based medicine ideology, 154 00:14:30,980 --> 00:14:38,240 because observation said that clinicians use this drug because it stopped the arrhythmias in the heart, 155 00:14:38,870 --> 00:14:45,020 the electrical conduction problems that come after heart attack because they've killed a bit heart, they can be seen an EKG. 156 00:14:45,170 --> 00:14:50,030 These are you stop them. They leave me. But unfortunately, some people literally drop dead. 157 00:14:51,080 --> 00:14:56,660 And when you look at it, at least a decade before the initiation of KAUST, 158 00:14:57,440 --> 00:15:04,970 it was recognised that these had an increased risk of subsequent arrhythmic death as compared with patients without these arrhythmias. 159 00:15:05,720 --> 00:15:12,320 But that recognition was based on observational data and we continue I continue to get this in. 160 00:15:12,320 --> 00:15:17,960 There's a real mantra around now that we should use big data and observational data to inform treatment effect. 161 00:15:18,470 --> 00:15:25,760 And that's a big, interesting issue actually. But actually this should warn you off immediately if you go this is a bit of a bad news. 162 00:15:25,970 --> 00:15:31,280 So for about ten years, people use these treatments widespread until the trial was published. 163 00:15:31,850 --> 00:15:41,839 And this is a preliminary report in 89 in the New England Journal of Medicine, which clearly shows whether the therapy is you should not use. 164 00:15:41,840 --> 00:15:45,950 We conclude that neither Encalade nor Fleck and I should be using the treatment for patients 165 00:15:45,950 --> 00:15:51,770 with asymptomatic or minimally symptomatic ventricular arrhythmia after myocardial infarction. 166 00:15:52,730 --> 00:15:57,110 Even though these drugs may be effective initially in suppressing ventricular arrhythmia. 167 00:15:58,070 --> 00:16:05,020 Ten years. Lots of death. More death. On that treatment then I killed in the whole of the Vietnam War. 168 00:16:05,320 --> 00:16:08,710 And that's what the book, Thomas Moore's book. So I think that's a profound. 169 00:16:09,640 --> 00:16:14,970 Interesting issue about the use of randomised trial evidence versus the use of observation. 170 00:16:14,980 --> 00:16:19,990 And I'll come back to that point. And sometimes statistics is not very difficult. 171 00:16:20,020 --> 00:16:27,130 You don't need the numbers there and this green line is a survival and that's the people on placebo. 172 00:16:27,400 --> 00:16:33,340 And boy, do you wish you were in the placebo group, because you can see it's a dramatic difference. 173 00:16:33,340 --> 00:16:37,040 Very rapid. Okay. Accumulating the evidence. 174 00:16:37,040 --> 00:16:42,699 Here's another one from Blythe. If that's the treatment we've used in in heart attack and we continue to use and 175 00:16:42,700 --> 00:16:47,860 this is a really interesting paper by chaps called Antman and Lao in JAMA in 1992. 176 00:16:48,130 --> 00:16:52,540 You probably see this in the course. Somebody will probably use it at some point, but it's a really interesting piece. 177 00:16:52,900 --> 00:16:59,170 What happened is you had these initial early trials and you can look at the recommendations in textbook reviews. 178 00:16:59,170 --> 00:17:03,399 His 21 not mentioned or not recommended experimental. 179 00:17:03,400 --> 00:17:06,549 And so here but here's the small trials and when you look at them small trials, 180 00:17:06,550 --> 00:17:13,090 what it was basically saying by the 1980s there were 20 very small trials and bleeding could be a side effect. 181 00:17:13,510 --> 00:17:18,130 There were too small to answer the question Does Thrombolysis save lives in patients with heart attack? 182 00:17:18,610 --> 00:17:24,670 And they thought about suggested that the benefit might be about 25 per thousand patients treated. 183 00:17:25,780 --> 00:17:29,560 So actually quite a big benefit, but nobody's using it. Doctors are not convinced. 184 00:17:29,800 --> 00:17:30,850 Nobody use the treatment. 185 00:17:31,120 --> 00:17:40,990 So what happened from there is basically in the next the next few years, what we saw is a massive increase in very large scale trials of this to GC. 186 00:17:41,140 --> 00:17:44,440 And these trials came on board and clearly answered the questions. 187 00:17:44,440 --> 00:17:52,380 And here what you can say is this is a cumulative meta analysis and as you go down the confidence intervals getting smaller by tomorrow, 188 00:17:52,390 --> 00:17:55,240 you should all know this today you might be like, where what we aren't about. 189 00:17:55,570 --> 00:18:03,250 But by here you can see a very narrow confidence interval, 48,000 people in the randomised trial, huge amount of people randomised. 190 00:18:03,610 --> 00:18:08,340 Importantly the textbooks still some not mentioned, some experimental. 191 00:18:08,350 --> 00:18:14,770 It takes quite a long time for textbooks to appear. You can see routine in textbooks and actually that's really interesting. 192 00:18:14,770 --> 00:18:18,219 It's reflected in the evidence base because if you look at the evidence, 193 00:18:18,220 --> 00:18:24,670 both this is in Lancet shows in the Trent region, the uptake of Thrombolysis and there's the GC trial, 194 00:18:24,670 --> 00:18:30,159 the large trial, those are these two still no take takes about another six months to a year and 195 00:18:30,160 --> 00:18:34,930 then we see a dramatic uptake over the next three years of the evidence comes in. 196 00:18:36,280 --> 00:18:43,089 There is a lag. But you know what? One thing of all with being convinced by is that actually clinicians are really good. 197 00:18:43,090 --> 00:18:48,070 When the evidence base is good, they don't need guidelines when the evidence is strong. 198 00:18:48,640 --> 00:18:55,180 Often what we get is guidelines when the evidence is poor, quality, uncertain, and people tell you what to do and we all argue and moan about it. 199 00:18:55,420 --> 00:19:02,290 But actually what we really want is guidelines that say here's really important evidence and actually what we should do. 200 00:19:02,290 --> 00:19:04,750 If you want to get the most benefits out of health care, 201 00:19:05,560 --> 00:19:14,650 we should be doing the things that are evidence based in 100% of the patients all the time optimising treatments that are known to be effective. 202 00:19:14,860 --> 00:19:20,229 And we often don't do that. We discard that. And a good example of that is this paper. 203 00:19:20,230 --> 00:19:25,150 Then moving on is what's called the golden hour. And if you look at the golden hour, 204 00:19:25,390 --> 00:19:31,660 if looking at the Thrombolysis papers and taking that evidence and looking at 205 00:19:31,660 --> 00:19:36,889 the relationship of the time to onset of treatment to that effect on mortality. 206 00:19:36,890 --> 00:19:38,799 And when you do that, you end up with this, 207 00:19:38,800 --> 00:19:44,980 which is called a meta regression and you see the red line absolute benefits per thousand treated start up here about eight. 208 00:19:45,490 --> 00:19:50,080 And if you get down here in a window of treatment delayed by hours, you can see it goes down to less than ten. 209 00:19:51,010 --> 00:20:02,170 That means if you delay treatment by one hour from 0 to 1 to 1 to 2 hours, that will cost you around about 30 lives lost. 210 00:20:02,620 --> 00:20:10,540 The best place to have your heart attack in the whole of the UK is in reading right now and they we still publish a 90 minute target. 211 00:20:11,290 --> 00:20:13,269 We don't publish 60 minute targets. 212 00:20:13,270 --> 00:20:20,470 I don't know why, but if you wanted to save a lot of lives, this is the one intervention that would be amazing to optimise within an hour. 213 00:20:20,740 --> 00:20:24,970 But nobody ever talks about optimising them. And if you did that, you'd have to buy in everybody. 214 00:20:25,000 --> 00:20:28,630 You have to buy in patient because they can't sit around going to got bit of chest pain. 215 00:20:28,840 --> 00:20:33,159 Half an hour later, maybe I'll ring the doctor. You have to buy everybody in the sequence. 216 00:20:33,160 --> 00:20:35,740 Don't you operationalise that and do that really well. 217 00:20:36,490 --> 00:20:44,200 We never talk about doing the things that are on the periphery, but we certainly could push this line back in here in a one hour window. 218 00:20:45,280 --> 00:20:48,370 Okay. We've had primary angioplasty. 219 00:20:48,370 --> 00:20:49,059 I won't touch on that. 220 00:20:49,060 --> 00:20:57,459 But that's sort of in the same issue has replaced the concept of Thrombolysis and the evidence for that has strong 23 randomised trials. 221 00:20:57,460 --> 00:21:02,930 So lots of evidence there. So moving on. Loading lots of evidence again in cholesterol lowering. 222 00:21:02,940 --> 00:21:09,630 When you look at cholesterol, there's a lot of research, actually a very huge amount of evidence here. 223 00:21:09,640 --> 00:21:14,100 The fourth trial, one of the early that's in chronic coronary heart disease. 224 00:21:14,400 --> 00:21:18,750 His Wolfcamp stone in Scotland in primary prevention. And then we've had these yellow ones. 225 00:21:18,750 --> 00:21:23,910 Have we changed the definition? We had to go, oh, my gosh, we've got to start again and do it in acute coronary syndrome. 226 00:21:24,180 --> 00:21:27,270 But a huge amount of trials have been gone, have occurred. 227 00:21:27,480 --> 00:21:34,020 These are the large scale trials. There are many small if you take all the small trials, it's about 135 trials. 228 00:21:34,020 --> 00:21:37,200 Again, these are the ones with over a thousand patients in them. 229 00:21:38,580 --> 00:21:45,270 And here's the fourth trial. That's a 30% relative risk reduction from 11 and a half to 8.2%. 230 00:21:45,930 --> 00:21:50,910 So take that out of 111 and a half to about eight people's lives saved. 231 00:21:51,330 --> 00:21:56,730 And the interesting bit about that is the risk reduction is sustained over the long term, 232 00:21:57,540 --> 00:22:03,620 six years of follow up since randomisation and then not moving on to evidence and 233 00:22:03,630 --> 00:22:07,050 people have not stood still because in the nineties when this evidence come out, 234 00:22:07,230 --> 00:22:10,890 you had to have an elevated cholesterol to be treated after a heart attack. 235 00:22:11,370 --> 00:22:16,559 And Hfpef, which was done here in Oxford, coordinated in Oxford, looked at a different that, 236 00:22:16,560 --> 00:22:21,180 looked at 20,000 people and randomised them to statin or placebo, 237 00:22:21,420 --> 00:22:27,630 but then looked at the baseline cholesterol levels, LDL and looked at the effect across them LDL. 238 00:22:27,930 --> 00:22:31,410 And basically what it showed is irrespective of where you start, 239 00:22:31,470 --> 00:22:37,830 the benefit is consistent and that again change practice radically going from achieve had a heart 240 00:22:37,830 --> 00:22:42,420 fat will measure your cholesterol we know you've had a heart attack actually this is 2002. 241 00:22:42,570 --> 00:22:49,320 We're going to treat you with a cholesterol lowering agent because on average you'll get this 25% reduction in mortality. 242 00:22:49,710 --> 00:22:53,910 So again, evidence radically change practice in the way we thought about it. 243 00:22:54,210 --> 00:22:58,200 And it's interesting nowadays when I teach medical students, we just take this as granted. 244 00:22:58,470 --> 00:23:01,920 This is the way we do these things, don't we? But actually, when you think about it, 245 00:23:01,920 --> 00:23:09,719 what's interesting about this whole talk is actually every single decision is based on really sound evidence and you can move on. 246 00:23:09,720 --> 00:23:12,420 We looked at cholesterol and this is part of the issue now. 247 00:23:12,420 --> 00:23:19,020 They've looked at 90,000 participants in 14 trials and started to look at people primary and secondary prevention. 248 00:23:19,800 --> 00:23:23,850 They're looking at the IPD collaboration. And right now you can look at cholesterol. 249 00:23:23,850 --> 00:23:30,000 The issue with the guidance is now I said let's lower the risk to 10% to give people cholesterol. 250 00:23:30,270 --> 00:23:37,500 What's been the problem? The problem is the people who've done this work in Oxford never collected some of the adverse event data. 251 00:23:37,800 --> 00:23:45,720 So they've got to go back to the drawing board and say, could we have your adverse event data for this individual data collaboration without problem. 252 00:23:45,720 --> 00:23:50,340 If you've got we were looking at this just this afternoon, if you've got a 10% risk. 253 00:23:50,490 --> 00:23:56,040 Well, actually, I looked at my own risk. My own risk in the next ten years came a 5%. 254 00:23:57,460 --> 00:24:00,700 Not that I don't smoke, have not got diabetes to try and keep myself fit. 255 00:24:01,060 --> 00:24:05,140 If I take a cholesterol lowering treatment, I can lower that from 5 to 4%. 256 00:24:06,520 --> 00:24:13,450 But I've got to take the tablet every single day and it didn't quite have all the adverse effect data because my risk of diabetes, 257 00:24:13,450 --> 00:24:19,560 about half a percent. And we can't quantify. The adverse effects in terms of muscle fatigue. 258 00:24:20,190 --> 00:24:26,759 But there is studies out there suggest these equal. But some of this is not reported from these large trials. 259 00:24:26,760 --> 00:24:32,520 It's not been reported. The question we had is at what point and I think this is an important one. 260 00:24:32,940 --> 00:24:38,310 When would you what risk would we as clinicians forget patients, as clinicians or as health professionals? 261 00:24:38,970 --> 00:24:43,530 What risk would you decide that it's important for me to take a treatment? 262 00:24:43,860 --> 00:24:48,640 Because remember, the relative effect stays the same. The absolute effect gets greater and greater as you go up. 263 00:24:48,660 --> 00:24:56,220 So when you get to about 20%, your risk can come down to about 14, 15% of a treatment of an absolute event. 264 00:24:56,230 --> 00:25:01,410 So you get much greater absolute effect. But actually, we haven't even asked ourselves what we would tolerate. 265 00:25:01,620 --> 00:25:07,020 And then we were expected to communicate this with patients. And I'm not sure what we use. 266 00:25:07,230 --> 00:25:13,470 We like these programs and these diagrams, but our patients just want sometimes a simple communication for myself. 267 00:25:14,850 --> 00:25:18,670 I think the field of evidence based communication is right up for grabs. 268 00:25:18,690 --> 00:25:24,930 Now it's really interesting time because the Internet area has made our patients much more knowledgeable. 269 00:25:25,780 --> 00:25:29,130 Okay, beta blockers, another one, really interesting. 270 00:25:29,280 --> 00:25:32,399 We're nearly there now, just through the the bit. You're the beta blockers. 271 00:25:32,400 --> 00:25:41,060 Again, loads of evidence about beta blockers. This is a two in 1999 BMJ paper by Nick Freemantle showing all the absolutely facts. 272 00:25:41,070 --> 00:25:48,030 But what's interesting about beta blockers is actually the first ever paper trial was in 1965 before aspirin. 273 00:25:48,030 --> 00:25:51,480 And this is the effect of propranolol in me in The Lancet. 274 00:25:51,990 --> 00:26:00,800 So actually quite a long time ago. But one of the interesting things about beta blockers is this trial, the comment trial. 275 00:26:00,810 --> 00:26:08,100 So we know beta blockers using this. What was helpful about the comment trial 40,000 people randomised is it clearly showed 276 00:26:08,520 --> 00:26:13,020 that in people with cardiogenic shock or who have any hemodynamic instability, 277 00:26:13,350 --> 00:26:21,270 you should withhold the beta blockers. And actually they're the people who do really badly and consider starting in hospital only when they die. 278 00:26:21,270 --> 00:26:27,360 My condition after me is stabilised so that changed the way we treatment or other evidence also changed the way we think. 279 00:26:27,360 --> 00:26:37,080 We used to not give people with COPD beta blockers, but actually in 1998, trial 201,000 patients with acute MRI. 280 00:26:37,560 --> 00:26:46,110 This is the difference. Actually, they do tolerate beta blockers, so actually we shouldn't withhold and that of the treatment ACE inhibitors. 281 00:26:46,410 --> 00:26:52,760 I was at medical school when we looked at the Hope trial. And if you think I'm a pro player, 282 00:26:52,760 --> 00:26:59,870 an important difference in terms of the relative risk effect and the absolute effect in people with coronary heart disease. 283 00:27:00,620 --> 00:27:05,540 Okay. And finally, clopidogrel. So these are all the treatments and capital is probably a lateral. 284 00:27:05,810 --> 00:27:10,040 This is part of the comment trial. 40,000 people. Very small reduction. 285 00:27:10,040 --> 00:27:12,140 Only a 1% absolute reduction. 286 00:27:12,500 --> 00:27:22,310 So you need an awful lot of people to say with certainty a 9% relative risk is actually only a 100 patients will benefit from clopidogrel. 287 00:27:22,400 --> 00:27:26,540 So you do need a lot of people to understand that size of effect. 288 00:27:27,800 --> 00:27:30,080 Okay. And then this really does interest I mean, 289 00:27:30,080 --> 00:27:37,850 these last two just finishing up this is a Cochrane review on the routine use of oxygen in patients with heart attack. 290 00:27:38,720 --> 00:27:42,500 And when you look at that Cochrane review and this has been in the news, in fact, 291 00:27:42,500 --> 00:27:52,820 actually it looks like if there is about a 25 to 30% more heart damage in patients, not given oxygen is three small trials. 292 00:27:53,360 --> 00:27:57,830 So we're in the same position as you. When you have benefit or harm, you've got random error. 293 00:27:58,220 --> 00:28:02,959 So what we need now is large trials and there are two trials I'm pretty sure going on at the moment. 294 00:28:02,960 --> 00:28:10,850 I haven't checked on trials scope to answer this question, which I think again would profoundly change the way I think about mechanisms, 295 00:28:11,150 --> 00:28:20,510 expert opinion, because when I was in training oxygen with the wonder drug and actually every the wonder drug, it looks like no oxygen. 296 00:28:20,510 --> 00:28:25,160 And you can think about some of the mechanism. And then here's another one. 297 00:28:25,760 --> 00:28:33,620 I got asked about this in the news. Oxford Ambulance Service is accused of playing God and putting life at risk in new drug trial. 298 00:28:35,180 --> 00:28:42,890 Chairwoman of Oxford Health Group patient voice Jackie Pierce said What they are doing is basically playing God by enrolling people into this trial. 299 00:28:43,160 --> 00:28:48,800 I think it's a very worrying trial. I'm highly suspicious when health care services don't publicise trials. 300 00:28:50,780 --> 00:28:55,669 Director of the Oxford based Centre for Evidence based Medicine says there are areas of clinical 301 00:28:55,670 --> 00:29:00,380 practice where the effects of what we do are uncertain and drugs like a generally may cause more death. 302 00:29:01,220 --> 00:29:05,900 And here we're going to A.V in the eighties. I don't know about the case trial, but I did say Cath, 303 00:29:06,410 --> 00:29:11,660 so I must say I heard about the cath and killed more Americans and the number of it died in the Vietnam War. 304 00:29:11,750 --> 00:29:16,420 Actually, just the cast because the team stands for trial. 305 00:29:16,430 --> 00:29:21,020 Yeah. So it's cath trial? Yeah. And finally, though, 306 00:29:21,020 --> 00:29:25,579 his the paramedic to trial is looking at whether a general adrenaline is helpful 307 00:29:25,580 --> 00:29:28,820 or harmful in treatment of cardiac arrest that goes outside of hospital. 308 00:29:29,090 --> 00:29:34,550 The key about this trial is they are randomising the ambulances to adrenaline or not. 309 00:29:34,880 --> 00:29:42,740 So if you have a cardiac arrest at the roadside, you they will be blinded and they won't know what treatment you get so you don't get consented. 310 00:29:43,280 --> 00:29:50,149 And that's a is in terms of like stroke or in certain situations when we've done the crash trials, 311 00:29:50,150 --> 00:29:56,030 when people have had a head injury, that they're out of a completely ethical way to answer questions of uncertainty. 312 00:29:56,690 --> 00:30:03,319 So and lastly, there's a big movement now, which I think this presents an interest in this with just out this year. 313 00:30:03,320 --> 00:30:05,430 This is from the American College of Physicians. 314 00:30:05,780 --> 00:30:10,610 And if you talk to me, the Americans 3 to 5 years ago were coming out with evidence based statements like this. 315 00:30:10,880 --> 00:30:14,720 I would have thought maybe we'd lost the war or something had gone badly wrong. 316 00:30:15,020 --> 00:30:22,280 But what they're saying is we've looked at the evidence and they have understood the principle of overdiagnosis and too much medicine. 317 00:30:22,580 --> 00:30:30,320 And in looking at the evidence, they said there is no evidence that cardiac screening improves patient outcomes in asymptomatic, low risk adults. 318 00:30:30,710 --> 00:30:34,580 The number of false positives is too high. You end up in a treatment cycle. 319 00:30:34,580 --> 00:30:38,480 That's a disaster and it costs a lot of money for no benefits. 320 00:30:38,810 --> 00:30:44,200 And that basically is overdiagnosis. And that's a really important issue that the Americans are grasping. 321 00:30:44,570 --> 00:30:52,520 They do about twice as many diagnostic tests for, say, the UK for no added value, and the cost of that is crippling their health. 322 00:30:53,300 --> 00:30:57,830 So finally, think about the discussion I want to have and we just have five or 10 minutes. 323 00:30:57,830 --> 00:31:05,420 All you might want to think about it and comment. Consider the consequences of a health system without an evidence based approach. 324 00:31:06,050 --> 00:31:09,380 What do you think it would look like if we stripped all of that evidence out? 325 00:31:09,890 --> 00:31:14,120 And then what would the treatment of heart attack look like in an evidence void? 326 00:31:14,570 --> 00:31:20,959 How do you think we would track practice in hospital or when people came out of hospital with all that evidence, 327 00:31:20,960 --> 00:31:25,010 not in front of us and in in place to make treatment decisions. 328 00:31:25,370 --> 00:31:29,540 Thank you for listening. Far off with with a question comment. 329 00:31:30,140 --> 00:31:33,620 Can I hear more about the lipid lowering, your cholesterol lowering? 330 00:31:34,010 --> 00:31:35,330 So you're saying that no matter what, 331 00:31:35,480 --> 00:31:43,630 whenever you start the treatment effect of that with a statin drug is the same that that that it reduces your of the relatives. 332 00:31:43,910 --> 00:31:48,440 Yeah. So the relative effect of a starting is about 30% reduction in mortality. 333 00:31:48,890 --> 00:31:53,160 Relative effect. Right. And so if you start your baseline and that's what they've looked at, 334 00:31:53,160 --> 00:32:00,420 vote for consistently across all the different baselines and said, so if you start at 20% absolute your risk. 335 00:32:00,420 --> 00:32:09,750 And so if you're when I'm about 65 with the same blood pressure, if I don't smoke and if I'll have diabetes, I'll be a 20% baseline risk. 336 00:32:09,960 --> 00:32:15,570 Right? If I then take the statin, I can reduce my risk from about 20 to 14%. 337 00:32:16,800 --> 00:32:23,130 So that's an absolute effective fix for ten years of treatment with a small increase in diabetes and I might get some muscle pain. 338 00:32:24,780 --> 00:32:30,329 The same issue now if I take that stuff in at 5%, a 30% is around about ground in oh four. 339 00:32:30,330 --> 00:32:33,540 If I go to ten, that will come to about seven, maybe just a little bit more. 340 00:32:34,680 --> 00:32:40,649 So that's what the argument is. So the key is for that relative effect. 341 00:32:40,650 --> 00:32:44,520 At some point you say the absolute benefits are so great, I'm going to take the treatment. 342 00:32:44,790 --> 00:32:50,160 Well, if you have a heart attack tomorrow, your risk of another heart that goes immediately beyond 30%. 343 00:32:50,550 --> 00:32:56,370 Right. So you can see beyond 30%. The risk are significant because you come from 30. 344 00:32:56,580 --> 00:33:02,040 That's ten lives, every hundred people treated. So we've gone from a secondary prevention to a primary prevention. 345 00:33:03,000 --> 00:33:09,750 What we don't know, and this is in the discussion, I think if we start to treat a whole wave of population at low risk, 346 00:33:10,050 --> 00:33:15,209 I think we should do the trials again because we don't understand what type of 347 00:33:15,210 --> 00:33:19,650 behaviours they may adopt if they suddenly go well if they came on a statin, 348 00:33:19,970 --> 00:33:23,010 I think I think I'll take up smoking again actually. What's the problem? 349 00:33:23,010 --> 00:33:27,899 The doctor said, I'm going to live forever. I might eat more, I might get diabetic, I might stop exercising. 350 00:33:27,900 --> 00:33:28,709 So we don't know. 351 00:33:28,710 --> 00:33:35,730 You can't just say blinder we would like to say let's just blindly because actually the number of people in them trials were some time ago. 352 00:33:36,480 --> 00:33:40,170 And actually we may be different as the population now in some way and behave differently. 353 00:33:40,470 --> 00:33:44,370 So I think it is crucial that we should still consider doing the clinical trial. 354 00:33:45,660 --> 00:33:54,900 The reason why you get the same percentage reduction no matter where you are is a consequence of the low dose response curve, 355 00:33:55,860 --> 00:34:00,240 which is a basic pharmacological phenomenon. It applies to all drugs. 356 00:34:01,620 --> 00:34:12,749 If you draw the relation between the effect of the compound and the concentration of drug producing and therefore the dose you get, 357 00:34:12,750 --> 00:34:16,920 the thing that looks like that, it's a rectangular hyperbola and it plateaus. 358 00:34:17,260 --> 00:34:25,680 So above a certain dose should get no more thing. If you transform the x axis to have a rhythmic axis, you get a sigmoid curve. 359 00:34:27,150 --> 00:34:33,450 And between 20 and 80% of the maximum effect, that sigmoid curve is approximately linear. 360 00:34:34,800 --> 00:34:42,420 And that tells you that for every unit change in the concentration of the drug, you get the same percentage change. 361 00:34:42,420 --> 00:34:45,720 In effect, the same thing is happening with cholesterol. 362 00:34:46,110 --> 00:34:49,070 To reduce your cholesterol by one, to remove 32, 363 00:34:49,320 --> 00:34:56,460 you reduce your risk of a coronary event or stroke by 10% no matter where you are because it's a logarithmic. 364 00:34:58,410 --> 00:35:02,270 So what you need to do is go and read it. 365 00:35:02,280 --> 00:35:06,500 We have to put that up in the reading list, read the IPD paper because that's all in there, 366 00:35:07,230 --> 00:35:12,630 that transformation, you know, and that's moving on to another level of understanding any of the fourth. 367 00:35:13,590 --> 00:35:18,989 So what do you think? My question to everybody, what do you think would have been different if we would get rid of the beam? 368 00:35:18,990 --> 00:35:22,710 But the evidence from the many of you might have thought about that. 369 00:35:23,490 --> 00:35:27,360 I just don't come from your experience and from the other piece in this room. 370 00:35:27,660 --> 00:35:36,239 What's the percentage of GP's who keep up to date with the latest outcomes from evidence based research? 371 00:35:36,240 --> 00:35:40,200 Yeah. No, that's an interesting what we haven't worked how to keep people up to date in any sense. 372 00:35:41,280 --> 00:35:45,540 I don't actually think you do need to keep up with so many things that we think. 373 00:35:46,410 --> 00:35:51,180 I think there are actually a small number of treatments that we should just optimise and do really well. 374 00:35:52,290 --> 00:35:56,249 And actually we have all these new treatments and new potential. 375 00:35:56,250 --> 00:36:02,130 But actually I think we have to devise a self service that says, actually, we're not going to do this. 376 00:36:02,490 --> 00:36:07,880 This is still uncertain. We're going to hold this back. Our job is to think about the bits of it. 377 00:36:07,890 --> 00:36:14,070 So the problem is, and this is the problem we go into primary prevention is we still know there's about 20% of the 378 00:36:14,070 --> 00:36:18,300 population walking out with a heart attack today who have not got their treatment optimised. 379 00:36:18,960 --> 00:36:23,010 Our public health want to do anybody about that. We haven't got enough time to do anything about that. 380 00:36:23,220 --> 00:36:30,090 And so we just say, let them walk around and do that. And that's what we should be optimising and telling other people until we've done that. 381 00:36:30,090 --> 00:36:34,980 Sorry, we ain't going to your primary prevention and your lower we're going to get the maximum benefit here. 382 00:36:35,940 --> 00:36:43,500 I think the favour the reason I like this story that the failure for many other interventions and treatments to do as much research as this. 383 00:36:44,310 --> 00:36:49,800 But I think you can basically say when the research is sound, everybody would go, well, actually in an MRI. 384 00:36:50,070 --> 00:36:58,320 We understand, based on the evidence, why we do these five things treatments really well, and we would put that to our patients. 385 00:36:58,320 --> 00:37:02,150 But lots of other guideline stuff is based on much poorer quality evidence. 386 00:37:03,000 --> 00:37:09,450 And I think our job is to start to reject treatments much more forcefully when the evidence is not good enough or insufficient. 387 00:37:09,750 --> 00:37:13,920 And I think there are many areas we would be able to say, sorry, we're not going to treat this in this way. 388 00:37:15,270 --> 00:37:23,370 The problem at the moment is we perceive any new piece of evidence is just as valid as of 145 trials of aspirin, isn't there? 389 00:37:23,820 --> 00:37:26,639 So the next time you see a new intervention, as you said, for this question, 390 00:37:26,640 --> 00:37:31,799 I wonder how it would stack up against aspirin in terms of an evidence base. 391 00:37:31,800 --> 00:37:32,490 How would it fit? 392 00:37:32,490 --> 00:37:41,459 What would it make me feel like compared to all then trial for all that work, I'd bet most interventions wouldn't even match up to our within. 393 00:37:41,460 --> 00:37:46,410 Cochrane The regional dissemination of trial going here is a really important effect. 394 00:37:46,530 --> 00:37:55,319 Yeah, difficult to quantify the effect of a new intervention when you already have therapy in the background that is taken, 395 00:37:55,320 --> 00:38:01,979 whatever reduction that you were going to have in that you know what little you have left here. 396 00:38:01,980 --> 00:38:07,980 So it is good that if it had been done earlier on other therapies. 397 00:38:09,090 --> 00:38:14,010 So there are two important yet. So there are about three or four important points there. 398 00:38:14,010 --> 00:38:16,770 The first is you've got much smaller benefit to work with. 399 00:38:17,310 --> 00:38:22,379 So that tells you you need a bigger and bigger trial, more larger trials than you did originally. 400 00:38:22,380 --> 00:38:31,860 So most trials now are underpowered quite considerably and you might think six 7000 people the about of error that's potentially still huge. 401 00:38:32,160 --> 00:38:37,649 So most trials the new trials are underpowered. The second issue that if you do see this at the FDA a lot, 402 00:38:37,650 --> 00:38:42,510 you think how is this drug got away with going to placebo when there are established treatments already? 403 00:38:43,230 --> 00:38:49,620 And that's a real problem. And then the third issue is and we've looked at this, we Jeff, we ended up writing a piece for the MRC. 404 00:38:49,620 --> 00:38:55,680 Here is comparative effectiveness. No drug companies want to do treatment against treatment, right? 405 00:38:56,400 --> 00:39:00,810 So now you've seen an establishment of a new methodological approach called network meta analysis. 406 00:39:02,250 --> 00:39:05,250 Really interesting, quite nice for the mathematicians in the room. 407 00:39:05,430 --> 00:39:09,420 They really love it. It's a new way of doing something. They're all excited about it, 408 00:39:09,720 --> 00:39:14,670 but basically they're putting in a load of rubbish into the machine and it creates huge issues 409 00:39:14,670 --> 00:39:18,600 that really still want the randomised trial of treatment to the established treatment. 410 00:39:18,600 --> 00:39:22,950 B So you're right, aspirin would probably have had to have been a bigger trial. 411 00:39:22,980 --> 00:39:27,120 It might not be as effective now in secondary prevention if you've got the new agent. 412 00:39:28,140 --> 00:39:33,370 Really important issue. But you said a couple of questions on for all treatments, however, 413 00:39:33,650 --> 00:39:38,940 like this which is good for heart disease which is common and I would paediatric surgery 414 00:39:38,940 --> 00:39:43,850 for small congenital diseases in order to get 400,000 patients talking about that. 415 00:39:44,550 --> 00:39:49,260 Yeah. Yeah. How speciality how are we ever getting to evidence based. 416 00:39:49,470 --> 00:39:51,870 Yeah, that shouldn't still I mean they've done it in cancer haven't they. 417 00:39:51,870 --> 00:39:56,880 So what they've done in cancer in paediatric is create a whole network not only in the country but across the EU. 418 00:39:57,330 --> 00:40:03,389 And I agree with you, it's difficult, but it's not impossible. And so you should be asking yourself the question. 419 00:40:03,390 --> 00:40:07,620 What you're really looking for is dramatic effect. If you don't do surgery. 420 00:40:07,620 --> 00:40:14,729 This is really important and it's not. So number one is you don't need randomised trials if you get dramatic effects and a dramatic effect 421 00:40:14,730 --> 00:40:18,780 is you don't need a randomised trial to show you that if you jump out a plane you need a parachute. 422 00:40:19,770 --> 00:40:26,760 Okay, that's the best one. But there are things like if you don't stop somebody's bleeding because it's all about confounders and bias. 423 00:40:26,790 --> 00:40:31,770 If so, if you have children, if you don't operate a birth, they die and you do that live. 424 00:40:31,980 --> 00:40:38,400 That's a dramatic effect. But actually, there'll be many things you might do in practice that you think, this is really interesting. 425 00:40:38,820 --> 00:40:42,420 Can we do children get pulmonary embolism or are we going to reduce infection? 426 00:40:43,620 --> 00:40:49,380 And you could just end up doing more and more work on the basis you haven't got the clinical trials to underpin what you're doing. 427 00:40:49,740 --> 00:40:56,370 And this is an important area which is in orphan drug orphan drugs are about what, less than one in 1500 people. 428 00:40:56,670 --> 00:41:01,229 And there's a massive explosion in these drugs now because actually you get something. 429 00:41:01,230 --> 00:41:09,870 If you get exclusivity, you get tax breaks, and you're the only person who is allowed to market that drug for that indication for ten years. 430 00:41:10,950 --> 00:41:18,569 And so you get things like ibuprofen is remodelled and has an exclusive license for patent doctors. 431 00:41:18,570 --> 00:41:24,390 Our theory health, if the price has gone up 8000 times on that basis because of one license. 432 00:41:25,050 --> 00:41:32,490 But your your point is you will come to a point in surgery is coming to this point is it's a lot of areas are going we 433 00:41:32,490 --> 00:41:42,560 need some evidence that what we do my thoughts are in a generation you will have changed the way the old system works. 434 00:41:42,570 --> 00:41:47,100 There are people at the old end who are still going to end their careers like this. But you knew it will come in and go. 435 00:41:47,100 --> 00:41:52,500 We have to question what we do. So it's only if only 40 years ago we were asking that about aspirin. 436 00:41:53,280 --> 00:41:57,659 In the modern world, you might have these global networks where you all work together and go, Actually, we need a big trial. 437 00:41:57,660 --> 00:42:02,160 We're going onto this. I think we will come to this understanding. 438 00:42:02,160 --> 00:42:10,200 We can't afford not to. We can't afford you to be wasting a load of money and spending hundreds of thousands on some new way of doing surgery. 439 00:42:11,760 --> 00:42:17,970 I'll give you a good example when we go as I go as to look at the data and published on the metal hips, I couldn't believe how bad the data was. 440 00:42:18,600 --> 00:42:24,060 You know, you come from a 3% failure rate, a ten year to a 20% failure rate, 441 00:42:24,480 --> 00:42:31,620 free for a change in somebody when we're going to just change the way we do this operation because people get back to work quicker. 442 00:42:32,430 --> 00:42:40,649 And that was really interesting in that area. So I suspect there will be also mandatory registries for all surgical interventions, implants. 443 00:42:40,650 --> 00:42:47,280 There will be mandatory registries and implants that are implanted in the body will have four clinical trials going forward. 444 00:42:47,970 --> 00:42:53,640 We can't afford it otherwise. What about really well-designed observational studies? 445 00:42:53,940 --> 00:42:59,770 Worrying about potential confounders and the like when you can't do drugs? 446 00:43:00,210 --> 00:43:00,590 Yeah. 447 00:43:00,600 --> 00:43:12,270 So, I mean, I guess the best observational study that we've ever seen was the one in Oxford done by Richard Doe and I think the number one priority. 448 00:43:12,360 --> 00:43:19,110 So the smoking cessation, 50 years follow up, 99.8% of all doctors followed up this amazing study. 449 00:43:19,320 --> 00:43:25,379 Most of what we do now is crap because we do retrospective studies on databases just cos they exist. 450 00:43:25,380 --> 00:43:28,470 And that's big data, if you like. It's very dangerous. 451 00:43:28,470 --> 00:43:32,790 However, when you do prospective studies, you can set them up really well. 452 00:43:33,360 --> 00:43:36,509 You can start to important things called propensity scores. 453 00:43:36,510 --> 00:43:41,100 You can talk about trying to understand the risk stratification of the individuals at the outset. 454 00:43:41,580 --> 00:43:47,879 You then follow all the patients thought and create a reality to look out that data going forward. 455 00:43:47,880 --> 00:43:51,400 And that's exactly how Thomas Charlie invented the hip in the first place. 456 00:43:51,720 --> 00:43:57,330 He was a master evidence based medicine person. He followed everybody up and he refined the intervention. 457 00:43:57,660 --> 00:44:01,460 So I think and then plug in some of the criteria from Bradford Hill. 458 00:44:02,520 --> 00:44:03,299 And if you do that, 459 00:44:03,300 --> 00:44:11,390 you'll have a potential to look at what you're doing without randomisation is still will be problematic one more will take one more. 460 00:44:11,760 --> 00:44:16,920 We talked about I think when I guess was at the Centre for evidence based Medicine when a park was taking around, 461 00:44:17,950 --> 00:44:24,779 a lot of a large percentage of decisions were changed. Was there any follow up data as far as outcomes like overall outcomes? 462 00:44:24,780 --> 00:44:27,780 I'm sure it's a broad range. No, no. 463 00:44:27,780 --> 00:44:30,920 And I think that was, again, an interesting issue. 464 00:44:30,930 --> 00:44:35,999 Nobody's ever been shown that following an evidence based approach, 465 00:44:36,000 --> 00:44:40,350 using evidence to inform your decision makes a difference to people population outcomes. 466 00:44:41,430 --> 00:44:45,540 But I think to me, taking a side over busy rate, 467 00:44:46,830 --> 00:44:53,760 it really worries me when I see people who don't practice in some evidence based way because the decisions they make 468 00:44:53,760 --> 00:45:01,139 sometimes are really dangerous and often are stressed and overworked because they can't reel back from ideas and go, 469 00:45:01,140 --> 00:45:03,720 Well, it's okay. We don't need to use that treatment evidence that we're okay. 470 00:45:04,050 --> 00:45:09,030 And it's a really interesting way of working that I can't understand how anybody would adopt that because I would find 471 00:45:09,030 --> 00:45:14,130 it incredibly stressful because I am making decisions here and I've no idea what I'm doing and why I'm doing it. 472 00:45:14,610 --> 00:45:19,500 Apart from a colleague told me it was a good idea. And if that's what you're really saying, aren't we. 473 00:45:20,160 --> 00:45:25,050 Now what David thing with the Commonwealth actually we were in the business of everything 474 00:45:25,380 --> 00:45:30,330 actually about half of things we do we could change if we had some evidence to hand at the time. 475 00:45:30,750 --> 00:45:38,670 Nobody still really should solve that issue because we still no better off in terms of the technology. 476 00:45:38,670 --> 00:45:43,000 And there's a problem now. The amount of information is not getting more and. 477 00:45:43,150 --> 00:45:45,430 Life and better from is getting bigger and bigger within it. 478 00:45:46,060 --> 00:45:55,060 So every time you look at it, most people in this room and everybody outside of this room uses Google as their number one search engine. 479 00:45:55,750 --> 00:46:02,490 And we've got a really smart I.T. chap working with a dphil who can show us that 97% of people will 480 00:46:02,770 --> 00:46:09,010 only use the first page of the only ever resource for what they use to inform their decision making. 481 00:46:09,430 --> 00:46:10,810 And that's how it generally works. 482 00:46:11,950 --> 00:46:18,099 And so when you think of all that stuff going on and what we did, I think it's a very interesting time, but it's not got easier. 483 00:46:18,100 --> 00:46:22,990 I think it's got actually a bit harder and nobody solved the point of care sourcing. 484 00:46:24,310 --> 00:46:31,330 And I don't know if you have any thoughts about how you would practice in an evidence board for difficult issues and. 485 00:46:33,140 --> 00:46:40,030 Okay. But you've got a long week ahead of you and this should set you up for some of the things tomorrow. 486 00:46:40,030 --> 00:46:45,550 But have a think about when you're going through. What does it mean when you've got evidence to hand compared to not? 487 00:46:46,060 --> 00:46:49,360 What would it look like the world if you didn't have any evidence in any area? 488 00:46:49,360 --> 00:46:52,899 Think about the areas like you said in paediatrics where there's low evidence, 489 00:46:52,900 --> 00:46:57,070 how you might use some of the levels of evidence is still low with evidence there. 490 00:46:58,090 --> 00:46:58,930 There's no such thing. 491 00:46:58,930 --> 00:47:06,790 I've never come across anywhere where there's no evidence because if if the best evidence is opinion, that's what you have to go with. 492 00:47:07,960 --> 00:47:10,450 Have a good week. Thanks, everybody. That's coming.