LAUSR

Dear Editor: With great interest, we read the recent article by Sun et al. entitled, BLocking plates versus intramedullary nails in the management of displaced proximal humeral fractures: a systematic review and meta-analysis^ [1]. In the study, authors revealed that locking plates and intramedullary nails have similar performance in terms of the functional scores and total complication rate. We applaud and congratulate their valuable work for clinical practice, but there are some potential methodological pitfalls which may affect the creditability of the conclusion. We would like to address several queries for the authors. Firstly, the meta-analysis was performed following the guideline of PRISMA, but the pre-defined protocol was not registered in any platforms, such as the Cochrane Library, PROSPERO, and BMJ Open, which may affect its transparency and reproducibility. Secondly, the authors mentioned that PubMed, EMBASE, Clinical Trials Registry and Cochrane Central Register of Controlled Trials, and Google Scholar were searched in this study, but these databases seemed to be not enough to retrieve all the eligible studies. Alternatively, some additional databases, such as NLM Gateway, and BIOSIS previews as well as unpublished data like grey literature should be included, which may contribute to get a more comprehensive collection of eligible studies. Thirdly, it is general that RCTs and observational studies cannot be pooled except that outcomes are involved in harm/adverse effects of the intervention according to the Cochrane Handbook. Furthermore, when comparative studies (RCTs and observational studies) are included in a meta-analysis, it is more appropriate to use BORs,^ not BRRs^ to act as the pooled effect estimate of dichotomous variables. Thus, it may be amethodological error to pool RCTs and observational studies in the currentmeta-analysis. A better choice was to perform pooled analysis based on study design respectively, or perform subgroup analysis based on study design (RCTs vs. cohort studies). Fourthly, Jones and colleagues revealed that age (60 years old or older) may be a key risk factor to cause functional loss in patients with proximal humeral fractures undergoing locked plating [2]. Therefore, it is better to evaluate the influence of the age (younger than 60 and 60 years old or older) on the overall effect in the current study. Actually, considering the substantial clinical heterogeneity of included studies, the Bleave-one-out^ influence analysis and subgroup analysis based on various clinical parameters, such as fracture types, mean age, sample size, quality of studies, and follow-ups should be conducted to increase the robustness and reliability of the current available evidence. Finally, the authors evaluated the publication bias of the current meta-analysis by using visualization of the funnel plot of 13 comparative studies for total complication rate. It would be better if the publication bias was assessed by Begg’s test and Egger’s test. Moreover, the conclusion of the current metaanalysis was merely based on 13 studies with small sample size ranging from 22 to 211. It was hardly to exclude potential Bsmall-sample effect^ and Bfalse negative,^ so the conclusion should be interpreted with caution [3]. At least, authors should emphasize this limitation in the Discussion section. Anyway, thank Sun and colleagues for valuable clinical evidence on a comparison of locking plates versus intramedullary nails in the management of displaced proximal humeral fractures. However, owing to the limitation of the current meta-analysis, further high-quality multicenter RCTs with larger sample size should be warranted to clarify the optimal surgical intervention for displaced proximal humeral fractures. * Hua-Ding Lu [email protected]