Metformin Reduces the Incidence of Shoulder Stiffness After Arthroscopic RC Repair: Response

Authors’ Response:

We sincerely appreciate the insightful comments and constructive suggestions from Qin et al ¹ regarding our trial. Their concerns about methodological and interpretive aspects have helped refine our understanding of the research, and we now clarify the key points as follows to strengthen the validity of our findings.

First, all 146 randomly assigned participants were included in the initial analysis framework. We processed the missing outcome data of 20 patients lost to follow-up using the multiple imputation method before conducting inter-group comparisons. Additionally, to assess the potential impact of loss to follow-up on the results, we conducted a sensitivity analysis using the pattern-mixture method. Assuming all patients lost to follow-up developed shoulder stiffness, the recalculated 3-month incidence of shoulder stiffness was 27.4% (20/73) in the metformin group and 41.1% (30/73) in the control group, with a still statistically significant difference between the groups (P = .048). Assuming none of the patients lost to follow-up developed shoulder stiffness, the incidence rates in the 2 groups were 9.6% (7/73) and 23.3% (17/73), respectively, with a significant difference (P = .019). This indicates that even under extreme assumptions, the study's conclusions remain stable and that no serious bias was introduced by loss to follow-up, meeting the verification standards for sensitivity analysis. ³

Second, as listed in Table 1 of the original article, the proportion of patients undergoing concomitant biceps tenodesis was balanced between the 2 groups. In addition, based on our surgical routine, acromioplasty was performed in all patients with no between-group difference in implementation rate. ² Thus, these adjunct procedures did not introduce confounding to our primary findings.

Third, we fully agree that systemic plasma biomarkers cannot be equated to the local glenohumeral joint capsule microenvironment. In fact, our study framed these biomarkers as exploratory, hypothesis-generating indicators rather than confirmatory mechanistic evidence. That is also why we did not apply multiple comparison correction for these biomarkers. Notably, plasma biomarker analysis was a secondary outcome, and our core clinical conclusion (metformin reduces early postoperative shoulder stiffness) is independent of these biomarker results.

We again thank Qin et al ¹ for their valuable comments, which highlight important details for improving research transparency and point out key directions for our follow-up work.