Letters The truth about sports drinks

Laboratory controlled trials are different from clinical trials

BMJ 2012; 345 doi: https://doi.org/10.1136/bmj.e5425 (Published 14 August 2012) Cite this as: BMJ 2012;345:e5425
  1. James A Betts, senior lecturer1
  1. 1Department for Health, University of Bath, Bath BA2 7AY, UK
  1. j.betts{at}bath.ac.uk

Heneghan and colleagues deemed many highly regarded studies to be of low quality and at high risk of bias.1 It is therefore essential to consider whether they used appropriate criteria to evaluate studies. Their method to determine risk of bias is more commonly used to evaluate clinical trials so it places heavy emphasis on controlling for deciphering treatment or group allocation, blinding, and thus avoidance of demand characteristics or placebo effects. Such problems are extremely important in large clinical trials under free living conditions but are not always so relevant for acute laboratory controlled trials (where blinding alone is often sufficient and knowledge of treatment can be integral to study design). Even so, they labelled some of the studies included in their analysis as “non-randomised trials,”2 3 4 or with a “lack of blinding,”4 5 6 when both randomisation and blinding were clearly reported in the relevant papers.

Another issue was the perplexing view that studies with fewer than 100 participants should be considered small. This statement was supported by reference to a clinical review that derived this number on the basis of the specific variability associated with therapeutic interventions for osteoarthritic joint pain. Major journals in the exercise sciences call for sample size estimates to determine whether the outcomes specific to each study require 10s, 100s, or 1000s of participants (the authors state that only four studies they reviewed provided power calculations, yet they overlooked others in their list).7 8 9

Given the tight laboratory controls applied in acute experiments, we usually find that 10-20 participants in a crossover design are adequate to detect effects (as shown by the numerous positive effects reported). It therefore seems excessive (unethical?) to test 100 or more people in studies that typically involve intense physical exertion and multiple tissue samples when hypotheses can be supported or refuted with just 10 or 20.


Cite this as: BMJ 2012;345:e5425


  • Competing interests: Several of JAB’s published papers were funded by GlaxoSmithKline and/or included their products. He also produced a report submitted to EFSA on behalf of the British Specialist Nutrition Association to evaluate claims regarding oral rehydration solutions.


View Abstract

Log in

Log in through your institution


* For online subscription