Credibility of claims of subgroup effects in randomised controlled trials: systematic review

Sun et al.(1) studied subgroup analyses in published randomised controlled trials. One of their results was that essential statistical findings such as interaction tests were often not reported. The outcomes of subgroup analyses therefore need to be interpreted with caution, as was also emphasised by Oxman(2) in the related editorial. However, subgroup analyses are crucial for enhancing our knowledge and this was not discussed by the authors.

Only average results are reported in randomised controlled trials and it follows from this that a treatment effect in a randomised controlled trial only implies that there was a subgroup in which the treatment has been successful(3). Trial participants are often recruited from a very heterogeneous population. It could well be that treatment did not work or even had an adverse effect in a number of participating patients. The problem is that one has to identify the non-responders, consider adjusting the theory and plan new studies. Cartwright and Munro(4) argued that theoretical and practical knowledge related to the particular study area was crucial for doing this.

Sun et al.(1) rightly mentioned that treatment should not be withheld to patients only because statistical analysis has shown that it was not effective in their subgroup. However, it is important to report subgroup findings and plan new studies to test further claims.

References
1. Sun X, Briel M, Busse JW, You JJ, Akl EA, Mejza F, et al. Credibility of claims of subgroup effects in randomised controlled trials: systematic review. BMJ. 2012;1553(March):in press.

2. Oxman AD. Subgroup analyses The devil is in the interpretation. BMJ. 2012;2022(March):in press.

3. Cartwright N. Are RCTs the Gold Standard? BioSocieties. 2007;2:11–20.

4. Cartwright N, Munro E. The limitations of randomized controlled trials in predicting effectiveness. Journal of Evaluation in Clinical Practice. 2010;16:260–6.