We appreciate the interest of Dr. Essex in our work. Dr. Essex raises concern about the −15% noninferiority limit chosen in our recent publication comparing face-down positioning and nonsupine positioning (NSP) for macular hole repair.1 Although we see benefits of a narrower noninferiority margin, we argue against the claim that a −15% margin is unsupported. First, the noninferiority limit must be interpreted in relation to the discomfort and harm caused by face-down positioning. Second, the only noninferiority limit used in a randomized trial comparing positioning in macular hole repair is a −15% limit used by Tadayoni et al.2 Although a −5% noninferiority limit has recently been used in a large registry-based study by Essex et al,3 no randomization was used in this study.
In our publication, we deemed 33 of 34 patients (97.1%) successful in the NSP group (95% confidence interval: 84.7–99.9); however, in fact 34 of 34 patients (100%) in the NSP group achieved hole closure with a single operation (95% confidence interval: 89.7–100.0). In a worst-case scenario, 18.9 patients (Equation 1) would be able to avoid face-down positioning in order that a single additional patient did not achieve hole closure. This number needed to treat calculation is based on an 89.7% lower confidence interval limit in the NSP group (taking into consideration that no reoperations were required in 34 prospectively evaluated NSP participants). Choosing a noninferiority limit requires a weighing of advantages and disadvantages. The threshold level for acceptability of NSP is also not static but varies from patient to patient. Ultimately, positioning is an issue for surgeons to discuss with patients and evaluate on an individual basis which regimen is most appropriate. We maintain that our study with its −15% noninferiorty limit (i.e., the maximum difference we were prepared to tolerate) is of value to patients considering the discomfort and harm caused by face-down positioning.
The 798 patients required for a randomized −5% noninferiority study as suggested by Yorston et al4 would contribute to our understanding of positioning; however, perhaps contrarily this is a case where “perfect is the enemy of good.” It is important not to lose sight of the goal of improving patient care. In this light, we believe that it is of greater value to fully understand the physiology of macular hole closure, which includes understanding the covarying effects of positioning compliance and intraocular gas fill rather than merely increasing the sample sizes and narrowing error margins. Perhaps, the surgeon's positioning recommendation is not as important as the patient's ability to avoid the face-up position and the amount of intraocular gas fill achieved. Until questions like these are answered, we cannot justify efforts to study 798 patients. However, we do agree that ongoing efforts are needed to address this unanswered question.