Study outcome measurements: statistically insignificant or non-inferior? A reply

Abstract

We thank Professor Sethuraman for his comments [1] on our article [2]. Regarding the concern that the assumption of non-inferiority lacked justification, we believe it is important to consider the evidence available at the time the study was designed, rather than applying current standards retrospectively. Our trial was registered in December 2021, at which time there were no published studies directly comparing the incidence of delirium between remimazolam and propofol. Furthermore, since there had been reports suggesting that benzodiazepines could increase the risk of delirium in patients who are critically ill [3, 4], adopting a non-inferiority design was a reasonable choice at the time. When the study was completed and the article was written, several relevant studies had been published, and we considered it appropriate to reflect their findings in the introduction [5].

The second point concerns whether our study adhered to the principles of non-inferiority trial design and reporting. We believe this may reflect a misunderstanding. The study was, in fact, conducted within a non-inferiority framework. The absence of an explicit statement of non-inferiority in the Clinical Research Information Service registry reflects a limitation of the platform itself, which classifies studies only as interventional or observational and does not include an item to indicate non-inferiority designs. At the time of planning, no previous trials had compared remimazolam and propofol directly in this specific context. Consequently, sample size estimation had to rely on pilot data. Figure 1 supports our conclusion of non-inferiority.

The third point relates to the clinical interpretation of the non-inferiority design, specifically the omission of pharmaco-economic considerations. We acknowledge that this aspect was not addressed. However, we believe that while cost is an important factor, it is not a required component in the interpretation of every non-inferiority trial. As mentioned in the introduction, remimazolam offers several clinical advantages over propofol, including no injection pain, greater haemodynamic stability, and more rapid recovery [2]. For clinicians who place greater value on these features, the higher cost of remimazolam may be acceptable. We believe that our findings may serve as a meaningful reference when selecting anaesthetic drugs for older patients.

Non-inferiority trials are generally more complex and methodologically demanding than conventional randomised trials, but they are feasible within clinical research. Our study adhered to the principles of non-inferiority trial design, and we consider the interpretation of the results to be appropriate in showing non-inferiority.