The Northern Ireland local group of the RSS held a meeting on Wednesday, 4 December, 2019 at 1pm, in the Wellcome-Wolfson Centre, BCH site, Queen’s University Belfast. The speaker was Professor Mike Clarke, director of the Northern Ireland Methodology Hub at Queen’s University Belfast and of the Northern Ireland Clinical Trials Unit in the Belfast Health and Social Care Trust.
Professor Clarke gave a general talk on the logic of randomised controlled trials (RCTs), arguing that large trials were preferable. He started off by explaining the scientific purpose of the RCT and took the audience on an ’evidence-based journey’ about ovarian ablation as a treatment for women with breast cancer.
Starting in 1889 he reviewed the history of the types of study which had been used to evaluate this treatment; moving from the initial proposal, through case reports, observational studies until the first RCT held in the Christie hospital, Manchester, in 1948. This, post-WW2, trial of ovarian irradiation versus none, randomised patients using envelopes. Later, in 1957, in Oslo, there was a further trial of ovarian irradiation using telephone randomisation - an early example of a now standard method of randomisation. In 1996 a systematic review of trials with a total of 2102 women (under 50) found that 47.6% of those allocated ovarian ablation had died by 15 years, compared to 53.9% in the control group: a 6.3% (SD 2.3) increase in survival (2p=0.001) after ovarian ablation.
Next he argued for ’evidence-based health care - an entirely natural, and now well-established, extension of RCT methodology into the health care sector. Such trials were to be designed in light of systematic reviews to enable decision makers to distinguish between interventions, actions and strategies that were beneficial, harmful and unproven: and to make informed choices by having quantified the magnitude on any potential benefit. Larger trials would have the effect of improving reliability by minimising bias and therefore were to be encouraged.
He dealt then with the impact of trials on clinical practice giving two interesting examples. The first was the ISAT trial (1994) which dealt with finding the better method of treating sub-arachnoid haemorrhage (Endovascular coiling versus Neurosurgical clipping). This trial randomised c1070 patients to each arm and first reported in 2002 (Lancet). The evidence was in favour of coiling and practice had really changed: in 1990 there was no coiling, but by 2007 75-85% of patients were being treated by Endovascular coiling, and by 2017 the figure was 95%.
The second example which involved multiple trials was in breast cancer treatment - Tamoxifen versusno immediate Tamoxifen. There had been 30 years of research and Professor Clarke’s overview contained 88% of worldwide data in the 56 trials presented. These trials had enrolled a total of 48,000 women whohad experienced 18,000 deaths. The evidence was heavily in favour of Tamoxifen which had the effect of reducing recurrence and mortality (UK and US data).
Finally, Professor Clarke wrapped up his talk on a favourite theme, ’Research on Research’. He was a well-known advocate of studies within a trial (SWATs) and once again set about persuading the audience of the virtues of this relatively novel approach.
This applied talk was received by acclaim by a large audience (more than 70 people were in attendance). Professor Clarke dealt with questions at the end. It was pointed out that medicine relied on accumulating evidence and it was rare for one trial to change practice. Thus, in terms of size, 'Large' should really mean 'Large enough', in order that the study remained ethical. After some discussion there was general agreement on this point. Professor Clarke also dealt with some questions about the details of the examples presented. It was noted that Professor Clarke’s approach to trials fell squarely in the 'Oxford Model' stable, but that there were other competing approaches to RCTs which were potentially more efficient.
The audience thanked Professor Clarke for a stimulating talk.