We read with great interest the randomised controlled trial by Kumar et al., comparing a 10-week (long) versus 6-week (short) prednisolone taper in achieving steroid-free clinical remission in moderate to severely active ulcerative colitis (UC) 1. We commend the authors on this novel study, which aims to address an important clinical question in an evidence-poor grey area. However, we would like to add a few comments regarding the sample size calculation and the interpretation of the statistical significance of the reported outcomes. The authors report that this was a non-inferiority trial, with a 10% non-inferiority margin and assumed 6-month clinical remission rates of 90% (long taper) and 80% (short taper) for sample-size calculation. This was based on a study by Bello et al. 2, which showed a 75% remission at 1 year using Mesalazine alone after a first course of steroids. However, this study included only patients with mild to moderate UC, which is significantly different to the cohort recruited to the current study, moderate to severely active UC, including acute severe UC (> 40% in both arms). It is highly optimistic to expect a 6-month clinical remission rate of 80%–90% in this cohort with just 5-ASA and Azathioprine. As expected, the observed steroid-free clinical remission rates in this study were much lower than projected, 44% in the long-taper arm and 20% in the short-taper arm. Using the same parameters used in this study for sample size calculation, but with conservative estimates of 50% clinical remission in the long taper arm and 40% in the short taper arm, the estimated total sample size is 182 (assuming a 20% drop-out), 91 in each arm. Based on these numbers, we feel that the study is significantly underpowered for the planned non-inferiority objective. In addition to this, the reported clinical remission rates in the long taper arm were significantly better compared to the short taper arm, RR = 2.19 (95% CI 1.08–4.46; p = 0.02). This primary outcome is reported as a relative risk with a p-value in a superiority trial-style comparison, despite the trial design being non-inferiority with expected reporting of risk-difference and confidence interval in relation to the 10% pre-specified non-inferiority margin. Traditional interpretation of a statistical significance of p < 0.05 in a non-inferiority trial means rejecting the null hypothesis and accepting the alternative hypothesis that the observed difference is within the pre-specified non-inferiority margin 3. However, we do understand that this data provides evidence of statistically significant superiority of the long taper over the short taper. We suggest the conclusions be framed as failure to demonstrate non-inferiority of the short taper, with an unexpected signal favouring the longer taper. Hariti Saluja: writing – original draft, writing – review and editing, conceptualization. Mohamed Asif Chinnaratha: conceptualization, writing – review and editing, supervision. The authors have nothing to report. This article is linked to Kumar et al. paper. To view this article, visit https://doi.org/10.1111/apt.70509. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.
Saluja et al. (Wed,) studied this question.