Four US-based medical researchers have made a controversial call for health interventions to be used on the basis of "good science" instead of waiting for more conclusive evidence from randomised clinical trials.
They say that fast-tracking treatments for which evidence of effectiveness exists would prevent avoidable deaths, particularly in developing countries.
But other researchers point to the dangers of making public health recommendations before trials have been conducted in resource poor settings.
Malcolm Potts and colleagues at the University of California's School of Public Health made the call in the British Medical Journal on Friday (29 September).
They highlighted examples of evidence-based interventions that could have saved many lives if they had been implemented before trials were conducted in poor settings.
These include oral rehydration to treat diarrhoea, male circumcision to prevent HIV infection and the drug misoprostol to prevent haemorrhage in women who have just given birth.
Potts's team stress that randomised controlled trials are needed and, when appropriate, should be part of the empirical evidence necessary for decision making.
"The question is how much evidence is needed to move from research to practice, when the matter is life-saving interventions in poor settings," they write.
They conclude that health policies should take account of the risks and benefits relevant to the local conditions, rather than an ideal situation.
But biomedical researchers interviewed by SciDev.Net warn against following their advice.
"This is an unfortunate editorial," says Scott Ramsey, director of cancer outcomes research at US-based Fred Hutchinson Cancer Center. "It implies that we should begin treatment programmes based on suboptimal evidence simply because of high mortality rates."
"Resource reallocation based on flimsy evidence that turns out later to be wrong seems less ethical," says Ramsey. "We make such mistakes time and again in the developed world. Let's not repeat those mistakes in poor countries."
Zachary Bloomgarden, a clinical professor at the Mount Sinai School of Medicine in New York City, United States, points out that there are "countless examples of treatments that have been recommended only to be shot down by randomised controlled trials".
"Before making potentially immensely important recommendations for public health interventions, we must be certain that they work," he told SciDev.Net.
Link to full article by Potts and colleague in the British Medical Journal
Reference: British Medical Journal 333, 701 (2006)