Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Can we really be expected to look for randomised controlled trials to test the effectiveness of large scale community interventions? Is it really ethical to educate only some children specifically about the dangers of traffic, or to randomly allocate some children at high risk of abuse/neglect to health visiting and ignore others, or to treat some meningococcal contacts with ceftriaxone instead of rifampicin, in order to “prove” that an intervention works?
There is a school of thought that places the randomised controlled trial as the only way to show that an intervention is effective. Many books about evidence based medicine take this as a central truth. There is evidence to support this claim—Schultz has shown the powerful effect of randomisation in trials of obstetric care.1
However, when you move from the tidy world of pill giving (or inhaler spraying, or cream rubbing) into the land of free roaming children, a number of things change. The intervention is often not a neat package with a quality control; who teaches the educational programme may affect its outcome. The intervention has a tendency to dissipate; if one new mother acquires some effective coping strategies for infant crying, she might well share them with her “control group” friend. The analysis of the data isn't always straightforward (for example, if it is villages that have been randomised to get insecticide coated mosquito nets, then you should analyse by number of villages, not number of people infected or number infected with malaria). And who gives consent to do this experiment on a whole community (such as fruit and veg education for weight reduction)?
Despite these problems, many randomised trials have been conducted on large scale interventions—all of those mentioned in this piece, in fact. They are not always large enough to show a convincing effect (or lack of one2), and this fundamentally questions their utility.3 The question of whether we should be expecting these trials is still unanswered, but while they are being performed we should try to assess them in attempting to practice in the most evidence based way we can.