Commentary on Lin MF, Hsieh YJ, Fetzer S & Hsu MC (2011) A randomised controlled trial of the effect of music therapy and verbal relaxation on chemotherapy-induced anxiety. Journal of Clinical Nursing 20, 988-999.

摘要

Lin et al. (2011) concluded that 'both music and verbal relaxation therapy are effective in reducing chemotherapy-induced anxiety'. This is plausible, and it might hopefully even be true, but it certainly is not supported by this study. How ironic that the authors decided to use permuted blocks specifically 'to maintain good balance', when this method has been amply discredited in the literature, precisely because it does not maintain good balance across treatment groups when the trial is either unmasked or imperfectly masked (Berger 2005). In fact, permuted blocks are inconsistent with allocation concealment, because they allow for the prediction of future allocations based on knowledge of the past ones, plus the restrictions that force balance in numbers (which, ironically, precludes the possibility of balance in covariates, arguably the more important type of balance for a trial to attain). The prediction allows the investigators to select healthier patients when one treatment is due to be allocated -and sicker patients when another treatment is due to be allocated.