Assessing methods to specify the targeted difference for a randomised controlled trial

Lead Research Organisation: University of Aberdeen
Department Name: School of Medical Sciences

Abstract

Increasingly it is acknowledged that decisions about what health care to provide should be evidence based. One part of the evidence needed is whether the intervention is a beneficial and worthwhile use of the limited money available to provide health care. Arguably the best way of providing this evidence is by conducting a randomised controlled trial. In a randomised controlled trial individuals, or groups of individuals, are typically randomly assigned to receive either the new treatment (which might be the way in which care is provided) or an alternative treatment. A sufficient number of people is needed so that we can be confident that if a difference between treatments exists we can detect it. The size of difference between the two treatments desired to be identified (the target difference) affects the number of people needed to be recruited and the cost of the study.
If the target difference is not correctly identified then there is a chance that time, effort (by patients, health care practitioners and researchers) and money will be wasted or poorly used. It is also possible that the health of patients may be adverse affected because the target difference was too small and we carry on randomising people to a treatment that we should already be able to say is inferior. Alternatively, if the target is too large we may fail to detect a truly important difference and as a result delay or prevent the introduction of a worthwhile treatment.
In this study we will systematically identify methods of determining the target difference. Scientific papers will be identified that described different methods, experts in the field will be approached, and we will survey organisations and individuals who are actively involved in the design and conduct of randomised controlled tiles. The different methods that we identify will be described, strengths and weaknesses appraised and potential scope for their use assessed. We will develop draft guidance for researchers and funders of randomised clinical trials. This will include separate sections for different types of trials and on different ways in which the outcomes of a treatment might be measured. This guidance will first be discussed amongst the study applicants and an expert advisory group. The recommendations will then presented at a symposium to which those with an interest in the design and conduct of trials and will help further refine recommendations and act as a first method of dissemination.

Technical Summary

Randomised controlled trials (RCTs) are widely used to compare alternative health care interventions. Critical to the validity of a RCT is that the study is of sufficient size to detect a difference between interventions. The required sample size can be calculated once the difference to be identified, statistic analysis and level of statistical certainty is chosen. The difference to be identified is called the targeted difference. From both a scientific and ethical standpoint, selecting an appropriate target difference is of crucial importance. A variety of formal approaches have been proposed for its determination. In practice, the targeted difference, at least from trial reports, to be determined upon convenience or some informal basis. In this study we propose to systematically review methods for specifying the target difference, conduct a survey of experts and produce a guidance document.
Methods for determining the target difference will be identified from: a review of the literature; known experts and those involved in the design and conduct of randomised controlled trial methods for determining the target difference. We will describe each method; appraise its strengths and weaknesses and possible scope.
Information on the methods used by those involved in the design and conduct of randomised controlled trials will be obtained by a web-based survey of UK Clinical Research Collaboration Registered Clinical Trials Units (CTUs) and MRC Hubs for Trials Methodology. The survey will ask what methods of eliciting a target difference these organisations use. The CTUs and the MRC Hubs are centres of excellence and their methods can be thought of as ?best? practice in the UK. To give an international dimension we will also survey members of the Society for Clinical Trials.
The results of the review and survey will be used to develop draft guidance for researchers and funders of randomised clinical trials. Separate guidance sections on methods, study designs (e.g. phase II/III/IV trials) and special topics (e.g. types of outcome and summary measure) will be developed. An expert advisory group will comment and discuss the structured guidance and recommendations during a workshop to help ensure that the recommendations are methodologically sound and can feasibly be adopted by researchers and funders of clinical trials. A symposium will be held with key stakeholders (trial funders, CTUs, trial hubs, patient group representatives) to further refine recommendations and as first method of dissemination.

Publications

10 25 50