LAUSR

We thank Zhang et al. for their meticulous comments regarding our article and for giving us the opportunity to provide further clarification [1]. First, addressing our clinical success inclusion criteria of an Eckardt score of≤3, Zhang et al. argued that the studies by McKay and Mondragon should have been excluded because it was stated that their definition of clinical success was a score of < 3 [2, 3]. We agree that the most consistently used endpoint measure to define clinical success in achalasia trials is an Eckardt score of ≤3 [4]. On closer examination of the respective studies by McKay and Mondragon, however, it can be noted that they defined “clinical failure” (the opposite of success, as it is reasonable to deduct) as an Eckardt score of > 3, which, by exclusion, is consistent with our definition of success as a score of ≤3. We noted this contradictory aspect and, although we did not specify it in the original manuscript, in these two studies we extracted the numerical data of patients with clinical failure, rendering our inclusion criteria of an Eckardt score of ≤3 still applicable. Second, Zhang et al. claimed that heterogeneity had not been appropriately addressed in our study. In detail, they argued that clinical success rates could be lower in studies with sample sizes of fewer than 100 patients, citing learning curve data as supportive evidence. This observation may not be appropriate because, for example, in the study by McKay et al., it was disclosed that the outcomes of the first 100 patients were published in a previous study [5]. The remainder of the studies did not specify whether included patients belonged to the “learning curve” population (i. e. the first 100 patients treated at each center). In other words, the observation does not apply if a center had already treated a fair number of patients before starting to enroll patients in the long-term outcomes study. However, to fully respond to this observation by Zhang et al., we performed new subgroup analyses, which did not show a substantial difference: the pooled clinical success rate in studies including fewer than 100 patients was 82% (95%CI 75.9%–88.1%, I2=39.8%), whereas in studies with 100 or more patients it was 91.1% (95%CI 87.9%– 94.4%, I2=65.5%). In response to the comment that studies with shorter follow-up may have higher clinical success rates, further subgroup analyses showed that the pooled clinical success rate in studies with less than 48 months of follow-up was 87.2% (95%CI 81.2%– 93.2%, I2=72.8%), with a comparable rate of 86.2% (95%CI 79.4%–92.9%, I2= 76.6%) in studies with a median followup longer than 48 months. In conclusion, in response to the comments from Zhang et al., we have clarified and justified the clinical success inclusion criterion, and provided data from new subgroup analyses that substantially confirm the stability of the results shown in our meta-analysis.