Comment on "The Unbearable Lightness of Difference Between Statistical and Clinical Significance".

Abstract

We are glad to see that the short-term results of our RVlob trial comparing robotic-assisted lobectomy and video-assisted lobectomy recently published on Annals of Surgery1 have aroused wide discussion. We deeply appreciate the comment from Bertolaccini et al, which reminds us to prudently interpret the statistical significance. Regarding one of our primary endpoints, the extent of lymph nodes (LNs) dissection, Bertolaccini et al pointed out that the reported difference of one more LN dissected by robotic-assisted lobectomy was not clinically significant. Indeed, we are fully aware of the gap between statistical and clinical significance and regard the survival benefit as the most solid evidence. As mentioned in the Discussion section of our original article, the correlation between an improved number of LNs and long-term survival is insignificant according to some published retrospective studies.2,3 However, beneficial effects of more LNs dissected on proper staging and prognosis may be shielded by the limited sample size. Furthermore, consensus on the optimal number of resected LNs in lung cancer surgery is still lacking. The threshold values vary from 8 to 12 in different studies.4–6 Ultimately, since no surrogate endpoints can perform as well as the overall survival, we expect the further follow-up of our RVlob trial to demonstrate a convincing argument for the controversy. When it comes to the comparison of direct and indirect costs, heterogeneity between countries, and medical systems should never be neglected. According to a review written by Jacobs and Fassbender,7 there is a wide variation among the measurements and analyses of indirect costs. In our study, we have defined the indirect cost as the overhead costs and amortization of surgical equipment,1 which is inconsistent with the definition provided by Bertolaccini et al. We will appreciate if they could specify the reference for the “widely used guidelines for the economic evaluation of healthcare interventions” so that we could reflect on the suitability of our definition. Despite the different reimbursement conditions of various medical systems, the current high cost of robotic surgery remains a common issue worldwide. No matter how much of the robotic surgery fee is reimbursed by the public health service, breakthrough in technical barriers, and innovative inventions will undoubtedly improve the feasibility of the robot-assisted surgery. As we have mentioned in a previous letter8 discussing a propensity score-matched study of robotic segmentectomy,9 the continued debate on robotic surgery will urge us to provide more convincing evidence. Fortunately, we have already stepped out to provide the perioperative results of a prospective randomized-controlled trial rather than a retrospective comparative study. Meanwhile, prudent reflection on methodology and statistical analysis should always be promoted when evaluating new treatments, just as the letter from Bertolaccini et al.