Skip navigation

The Anti-Weasel Campaign: Greyness, NNTs and Life

The greyness of medicine upsets managerial talent - why do we always have to add so many caveats to our guidelines? Reality is that clinical decisions are necessarily context-dependent. Nonetheless we need ways to encapsulate the evidence from RCTs and systematic reviews, ways which are useful in making both clinical and policy decisions, and which make the context explicit. Numbers-needed-to-treat (NNT) is Bandolier's current favourite encapsulating tool (see #11 [Jan 95] on), and readers have asked for a how-to-do-it repeat.

The example is taken from a survey of the risk that epidurals in childbirth are associated with an increased risk of long term back problems [1]. Long term back problems after:

The example is not from RCT data, and so should carry a greyness warning that it may be incorrect. Bandolier uses it because it is a lovely example of the greyness - when is the data true? - and, more importantly, because childbirth groups use it in their literature and the professionals have to answer it, optimally by doing the RCT to rebut the allegation, difficult though this may be.

The current professional argument is that women with more difficult labours are more likely to have epidurals. The difficult labours are themselves more likely to result in long term back problems.


  1. RJ Cook, DL Sackett. The number needed to treat: a clinically useful measure of treatment effect. British Medical Journal 1995; 310: 452-4.
  2. R Russell, P Groves, N Taub, J O'Dowd, F Reynolds. Assessing long term backache after childbirth. British Medical Journal 1993; 306: 1299-1303.

previous or next story in this issue