Linan Zeng

2 months ago | ebm.bmj.com | Linan Zeng |Monica Hultcrantz |David Tovey |Nancy Santesso

DiscussionThis article has, using hypothetical and actual examples, introduced the challenges that arise when authors initially decide to rate certainty in a non-zero effect and the point estimate falls close to the null. Our suggested approach considers whether the point estimate is certainly or possibly less than the MID: if certainly so (ie, clearly smaller than the smallest plausible MID), we suggest revising to rate certainty in a little or no effect.