Editor's Choice


BMJ 2006; 333 doi: https://doi.org/10.1136/bmj.333.7572.0-f (Published 12 October 2006) Cite this as: BMJ 2006;333:0-f
  1. Fiona Godlee, editor (fgodlee{at}bmj.com)

    There are some lively discussions on bmj.com. One that makes it on to this week's letters pages is about whether it's ever OK to assume that an intervention is effective based on observational studies alone. Are there times in medicine when we should act without waiting for randomised trials? Malcolm Potts and colleagues argued that there were (30 September, p 701), citing oral rehydration therapy, circumcision to prevent HIV infection, and misoprostol for postpartum haemorrhage. Some responders accepted aspects of their argument, but most expressed a mixture of outrage and scorn (p 807). What about adverse effects? Why should people in the developing world be treated on a lesser standard of evidence? Richard Lehman concludes in his journal blog on bmj.com, “This article is so bad it may actually do some good.” Presumably the potential good lies in getting us to think about what we base our actions on.

    One intervention whose effectiveness may seem obvious, if only in biological terms, is calcium supplementation to strengthen bones. In fact, as Tania Winzenberg and colleagues show in this week's BMJ (p 775), there is no evidence of a clinically important decrease in fracture risk during or after supplementation in children. In an accompanying editorial (p 763), Amy Joy Lanou asks whether we might in fact be doing harm by encouraging healthy children to consume dairy products in view of our obesity epidemic and high rates of lactose intolerance globally. And what of the opportunity costs? “The focus on calcium… draws attention away from more comprehensive research on how to promote long term bone health among young people,” she says.

    Of course, even randomised controlled trials can be misleading. Carl Heneghan and colleagues call for caution in interpreting the results of the recently published diabetes reduction assessment with ramipril and rosiglitazone medication (DREAM) trial of drug treatments to prevent diabetes (p 764). The reported positive effect of rosiglitazone looks less good at three years, when the rate of all cardiovascular events was higher in the intervention group, and if rates of heart failure are taken into account. Like Lanou, they conclude that we might do better to focus attention on evaluating pragmatic lifestyle measures rather than expensive and potentially harmful drugs.

    You may notice that none of these articles end by saying “More research is needed.” That's because the phrase has been banned at the BMJ for as long as I can remember. Stephen Lock, one of my predecessors as editor, viewed it as a fatuous conclusion to any piece of work because more research is almost always needed. Polly Brown and colleagues take up his banner (p 804), calling on researchers to be more specific with their recommendations for more research. After all, they will have looked at the literature, designed their own research, and learnt from doing it. The least they can do, say Brown and colleagues, is take the trouble to record their views on what the next research steps might be. I agree. But your views via bmj.com will be welcome.

    View Abstract