Waiting for the results of randomised trials of public health interventions can cost hundreds of lives, especially in poor countries. Researchers in this week's BMJ argue that, if the science is good, we should act before the trials are done.
Randomised controlled trials are usually required before new interventions are implemented. But in 2003, the BMJ published an article about the lack of trials on parachutes to show that situations still exist where such trials are unnecessary.
This week, US researchers use examples from poor countries to show how an overemphasis on randomised controlled trials poses important ethical and logistical problems and may incur avoidable deaths.
In all three examples, the parachute approach has the potential to save tens of thousands of lives, say the authors. For instance, many lives could have been saved if male circumcision had been provided on a large scale from the early 1990s, when data showed that it protected against HIV, instead of waiting 10 to 15 years for more evidence.
Similarly, thousands of women could have been saved if a drug to control blood loss after childbirth was made available in remote areas. But the World Health Organisation has not added it to the essential drugs list because of a lack of evidence from randomised trials in home settings.
The authors 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.
The yardstick for decision making should take into account the risks and benefits in the local conditions, not those of an ideal situation, they conclude.