What has EBM done for healthcare?

October 11, 2015



‘Retracing the history of evidence-based medicine (EBM) through conditions such as heart attack provides a useful insight into the influence of EBM on past and current decision-making.’

Carl Heneghan, Professor of EBM: Historical evidence of effective heart attack treatments

cite as. Heneghan C.  (2015) What has EBM done for healthcare? Centre for Evidence-Based Medicine. https://www.cebm.net/  (download pdf of the presentation)

Listen to Podcast of talk: 12th Oct 2015, Rewley House, Oxford University


Analysing the first randomised trials for aspirin, cholesterol lowering, thrombolysis and other relevant treatments poses a question of vital importance – what would treatment look like without evidence? A lack of high quality evidence would certainly lead to dire consequences, kill many people and certainly not have  avoided  many preventable deaths.

In 1974, the BMJ published the first RCT of acetylsalicylic acid – aspirin – for the secondary prevention of mortality from myocardial infarction. [1, 2] Despite Reverend Edmund Stone consuming Willow tree bark in 1758 and publishing his account in the Journal of the Royal Society in 1763 – [1] about the many useful discoveries of the bark of an English tree (useful for curing intermittent disorders such as fever) – it took over 200 years before  aspirin was subjected to the first  clinical trials and found to be useful due its antithrombotic properties.

In the early 80s, Peter Ellwood and Archie Cochrane used the slide (see figure) – what we now know as a meta-analysis – to discuss the effects of combining the first six aspirin trials, including 10,859 patients. [3]

Early effects of aspirin treatment

Elwood & Cochrane. Early effects of aspirin treatment – JR SOC Med [3]

The weighted overall effect was a 23% reduction in mortality; reported in a 1980 editorial by Richard Peto [4] – who went on in 1988 to publish the antiplatelet triallist collaborations findings on 31 RCTs of aspirin in patients with a history of TIA, occlusive/, unstable angina or prior MI. [5]

Peto’s overview emphasized two important EBM principles: first by including large numbers of patients it was possible to reduce the random error associated with any single trial – one of the main purpose of systematic reviews. Second, was to overcome the problem of publication bias; but also, at the time was the problem of marketing and awareness bias: in a pre-internet era some trials were widely known to clinicians, and even despite publication, other trial results were not so. As a consequence results were often skewed in a similar fashion to the effects of publication bias.

By 1994, there had been an explosion of aspirin RCTs. The BMJ overview now contained 145 randomised trials of prolonged therapy and reported a similar 25% risk reduction in mortality. [6]

Because of the properties of aspirin, and its consistent effect in secondary prevention trials, a 2005 BMJ editorial called for – well reported – aspirin to be routine for everyone over the age of 50. However, four years later the aspirin trialist collaboration published its lack of effectiveness in primary prevention. The results questioned the net benefits: any reductions in cardiovascular events were offset by increased Gastro-intestinal bleeds. The conclusions, therefore, reported the use of aspirin in primary prevention was of uncertain net benefit.

Aspirin trials, provide important insights into how the structure of a robust evidence base should look: large trials, which reduce  the random error associated with single trials, followed by replication and provision of individual data for determination of important subgroup effects – to date over 200 trials have been completed on heart disease alone. It also underlines the continual need to question treatment effects, particularly when the baseline risk of an event for an individual is low.

In terms of preventing catastrophic consequences, the 1986 antiarrhythmic’s CAST trial [7] provides an essential lesson on the use of observation over randomisation to establish treatment effects.

In the early 1980s newly introduced antiarrhythmics were found to be highly successful at suppressing arrhythmias – through observation. It was not until a randomised controlled trial – the CAST trial – was performed, was it realised that although these drugs did suppress arrhythmias, they actually increased mortality – by a substantial amount. The CAST trial revealed an excess mortality of 56 per thousand patients treated. Needlessly, it took at least a decade from the onset of use of these drugs to the initiation of the CAST trial for the increased risk of subsequent arrhythmic death to be recognised.

The 1989 trial report, in the New England Journal of Medicine [7] makes for good reading: ‘neither encainide nor flecainide should be used in the treatment of patients with asymptomatic or minimally symptomatic ventricle a rhythmic after myocardial infarction, even though these drugs may be affected initially in suppressing ventricular arrhythmia.’ Thus ending – through randomisation – the devastating effect of a deadly drug. Indeed, the results of this trial likely led to the development of EBM as we now know it.

The accumulating evidence of thrombolysis points to the importance of large-scale trials, but also to the lag between textbook recommendations and research evidence – highlighted so aptly in a 1992 JAMA paper by Antman and Lau. [8]

By the 1980s, 23 small trials of thrombolysis had been completed (including 6000 patients) which showed: bleeding could be a significant side-effect of treatment, and could sometimes be fatal. However, these trial were too small to answer the question as to whether treatment actually saved lives – the benefits suggested about 20 lives were saved for every thousand patients treated. Doctors were unconvinced and treatment was routinely withheld  until two large mega trials, ISIS-2 and GISSI, reported significant effects upon mortality. [9]

And even then, uptake was slow, in response to the evidence.  Doctors’ response, at the time, in the Trent region of the UK illustrated the substantial lag between the development of evidence and uptake into practice. [10]

Screenshot 2015-10-11 16.09.27

Thrombolytic drug treatment in the Trent region [10]

Clearly there is a failure to do trials in a timely manner. These trials – confirmatory trials – act to underpin confidence in treatment effects. In the modern world such trials are often missing and therefore it is harder to establish – with certainty – whether a treatment actually works. Even putting aside that in current times many of these trials are industry funded.

Further thrombolysis evidence, in 1996, illustrated the importance of early treatment in MI – the golden hour. Analysing the time to treatment showed that if 1000 patients are treated within the 1st hour of symptom onset (as opposed to the second hour) 28 extra lives would be saved. [9]

Further evidence, for the use of angioplasty, cholesterol-lowering, beta-blockers, clopidogrel and ACE inhibitors –  in the associated presentation – point to the need for high quality robust evidence to underpin treatment effects.

Of importance, though, is the continual need to ask questions about current clinical care, and in response update the evidence-base. As an example, a 2013 Cochrane review questions the use of oxygen in heart attack patients; early trials have shown those treated with oxygen had about 30% more heart damage than patients not given oxygen, leading to the development of ongoing large-scale trials to answer this question. [11]

Further to this, in 2015, the PARAMEDIC 2 trial [12] asked the question – despite decades of treatment – as to whether adrenaline is helpful or harmful. Its long-term use in cardiac arrest is based solely on poor quality observational data –  a similar narrative to the historical use of antiarrhythmics. The results of this trial will influence out-of-hospital treatment, and it wouldn’t surprise me if we found out the treatment doesn’t work; but also, that it is harmful.

Finally, we have developed evidence to prevent overdiagnosis and overtreatment of individuals. In 2015, the American College of Physicians released guidelines on the screening for heart disease. These guidelines showed there was no effect of screening and the rates of false positives were high to recommend intervening. [13]

Screenshot 2015-10-11 16.09.58

ACP guidelines on screening for coronary heart disease [13]

Periodically, it is worth revisiting important treatments and the establishment of their effectiveness. Not least it allows us to understand how to establish a robust evidence-base for treatment effects in the future that can be applied to patient care. Do I think the evidence will look different in the next five year for practice? It depends…… on the number of confirmatory trials, their size and robustness and if they look half as good as the evidence for aspirin.


cite as. Heneghan C.  (2015) What has EBM done for healthcare? Centre for Evidence-Based Medicine. https://www.cebm.net/what-has-ebm-done-for-healthcare/ DOI: 10.13140/RG.2.1.2538.5680 
2015-10-12 T 14:05:57 UTC





  1. Fuster V , and Sweeny J M . Aspirin: a historical and contemporary therapeutic overview. Circulation. 2011 Feb 22;123(7):768-78. doi: 10.1161/CIRCULATIONAHA.110.963843.
  2. Elwood PC, Cochrane AL, Burr ML, et al. A Randomized Controlled Trial of Acetyl Salicyclic Acid in the Secondary Prevention of Mortality from Myocardial Infarction. British Medical Journal. 1974;1(5905):436-440. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1633246/
  3. Elwood P. The first randomized trial of aspirin for heart attack and the advent of systematic overviews of trials. J R Soc Med. 2006 Nov;99(11):586-8. Review. No abstract available.
  4. Peto R. Aspirin after myocardial infarction. Lancet 1980;215: 1172-3 [Editorial]
  5. Secondary prevention of vascular disease by prolonged antiplatelet treatment. Antiplatelet Trialists’ Collaboration. Br Med J (Clin Res Ed). 1988 Jan 30;296(6618):320-31. http://www.bmj.com/content/308/6921/81.long
  6. Collaborative overview of randomised trials of antiplatelet therapy–I: Prevention of death, myocardial infarction, and stroke by prolonged antiplatelet therapy in various categories of patients. Antiplatelet Trialists’ Collaboration. 1994 Jan 8;308(6921):81-106. http://www.bmj.com/content/308/6921/81.long
  7. Preliminary report: effect of encainide and flecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infarction. The Cardiac Arrhythmia Suppression Trial (CAST) Investigators. N Engl J Med. 1989 Aug 10;321(6):406-12. http://www.ncbi.nlm.nih.gov/pubmed/2473403
  8. Antman, Lau et al. A comparison of results of meta-analyses of randomized control trials and recommendations of clinical experts. Treatments for myocardial infarction. 1992 Jul 8;268(2):240-8. http://www.ncbi.nlm.nih.gov/pubmed/?term=1535110
  9. Early thrombolytic treatment in acute myocardial infarction: reappraisal of the golden hour. 1996 Sep 21;348(9030):771-5 http://www.ncbi.nlm.nih.gov/pubmed/?term=8813982
  10. Ketley D, Woods KL. Impact of clinical trials on clinical practice: example of thrombolysis for acute myocardial infarction. Lancet. 1993 Oct 9;342(8876):891-4. http://www.ncbi.nlm.nih.gov/pubmed/?term=8105166
  11. Routine use of oxygen in people who have had a heart attack. Cochrane library http://www.cochranelibrary.com/editorial/10.1002/14651858.ED000065
  12. PARAMEDIC 2 TRIAL. http://www2.warwick.ac.uk/fac/med/research/hscience/ctu/trials/critical/paramedic2/
  13. ACP guidelines. https://www.acponline.org/clinical_information/guidelines/guidelines/




Leave a Reply

Your email address will not be published. Required fields are marked *

* Checkbox GDPR is required


I agree