Rapid Responses to:

LETTERS: Peter H Canter and Edzard Ernst Scientific evidence was ignored in CAM and the NHS BMJ 2005; 331: 1024-b [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Cost-effectiveness of CAM: pragmatic trials are good science

Hugh MacPherson, Kate Thomas, Professor, Complementary and Alternative Medicine Research   (21 November 2005)

Cost-effectiveness of CAM: pragmatic trials are good science 21 November 2005
Hugh MacPherson, Senior Research Fellow Department of Health Science, University of York, York YO10 5DD, Kate Thomas, Professor, Complementary and Alternative Medicine Research Send response to journal: Re: Cost-effectiveness of CAM: pragmatic trials are good science	Editor, We would like to respond to Canter and Ernst’s letter 1 in which they state that pragmatic randomised controlled trials are “methodologically weaker” than “gold standard” RCTs. In this response we are assuming that by “gold standard” they mean placebo-controlled trials, even though this limitation to the concept of the “gold standard” is questionable. Our main point however is that, as any good undergraduate student should know, the research methods you choose should be determined by the nature of the research question you are asking. If you are asking questions about the overall clinical impact of a package of care, perhaps in order to facilitate policy decisions involving the allocation of scarce resources, then a pragmatic RCT is not only appropriate, it is also good science. If you are asking questions about the whether a simple intervention or an element of a complex intervention has a specific effect per se, then a placebo-controlled trial is appropriate, and it is also good science. Canter & Ernst make the elementary error of judging the research methods independent of the research question the method was designed to address. In other words criticising a pragmatic trial for not identifying a specific effect, a question it was never designed to answer, is obviously missing the point. Like Canter and Ernst, we argue for the application of scientific method to medicine, however for us good science is based on a clear understanding that different research questions require different methods, and may well provide different answers. 2 Canter and Ernst further claim that pragmatic RCTs do not resemble real life treatments any more than “gold standard” (their short-hand for placebo controlled) RCTs. This opinion lacks credibility. When considering a parallel arm placebo-controlled trial where the primary comparison is between a verum treatment and a placebo treatment, all measurements in the placebo arm relate to an artificial intervention, all differences in outcomes between arms are also artificial. One consequence is that the results are uninterpretable in terms of what they might mean in a real world context. As a placebo controlled trial can only ever provide evidence under experimental conditions there can be no guarantee that the results will transfer to practice. Moreover, there is no way that the results can tell us whether the intervention under examination performs better or worse than current best practice or other available alternatives. Finally, because the costs associated with the placebo arm are also artificial, such trials are useless when it comes to establishing the cost effectiveness of an intervention. 1. Scientific evidence was ignored in CAM and the NHS Peter H Canter, Edzard Ernst BMJ 2005;331:1024 2. Understanding controlled trials: What are pragmatic trials? Martin Roland, David J Torgerson . BMJ 1998;316:285 Competing interests: None declared