Georgetown University School of Medicine, Washington, DC, USA.
Department of Orthopaedic Surgery, MedStar Union Memorial Hospital, Baltimore, MD, USA.
J Shoulder Elbow Surg. 2022 Aug;31(8):1610-1616. doi: 10.1016/j.jse.2022.01.141. Epub 2022 Mar 1.
Proximal humeral fracture represents an increasingly common pathology with evaluation and treatment often guided by evidence from randomized controlled trials (RCTs), but the strength of an RCT must be considered in this process. The purpose of this study was to evaluate the strength of outcomes in RCTs on the management of proximal humeral fractures using the fragility index (FI), a method used with statistically significant dichotomous outcomes to assess the number of patients that would change an outcome measure from significant (P ≤ .05) to nonsignificant if the patient outcome changed. We also aimed to correlate the FI with other measures of study strength.
A systematic review was performed using Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines to evaluate RCTs on the management of proximal humeral fractures. The PubMed, Ovid MEDLINE, Web of Science, and Embase databases were searched from database inception to May 2021. RCTs with at least 1 statistically significant (P ≤ .05) dichotomous outcome were included. The FI was calculated for each included trial using the Fisher exact test. The FI was correlated with the study sample size and journal impact factor.
Ten RCTs reporting on 656 patients and published between 2011 and 2020 were included. The median patient sample size was 67 (mean, 65.6; range, 40-86). Complications were the most commonly reported dichotomous statistically significant outcome. The median FI was 1 (mean, 2.6; range, 0-18), with 4 studies having an FI of 0. A median FI of 1 indicates that 1 patient experiencing an alternative outcome or having not been lost to follow-up could have changed the pertinent conclusions of the trial for a given outcome. The median number of patients lost to follow-up was 3 (mean, 4.9; range, 0-16) and exceeded the FI in 50% of studies. There was no correlation between the FI and sample size (Spearman coefficient = 0.0592, P = .865) or between the FI and journal impact factor (Spearman coefficient = -0.0229, P = .522).
In most studies of proximal humeral fractures, only 1 or 2 patients experiencing an alternative outcome or lost to follow-up would change the conclusions for the dichotomous outcome studied. Although the FI cannot be used to assess continuous variables, which are often the primary outcome variables of RCTs, it does offer an additional unique measure of study strength that surgeons should consider when evaluating RCTs.
肱骨近端骨折是一种越来越常见的疾病，其评估和治疗通常以随机对照试验（RCT）的证据为指导，但在这个过程中必须考虑 RCT 的强度。本研究的目的是使用脆弱指数（FI）评估肱骨近端骨折治疗的 RCT 中结局的强度，该方法用于具有统计学意义的二分类结局，以评估如果患者的结局发生变化，有多少患者会使结局测量值从有意义（P≤0.05）变为无意义。我们还旨在将 FI 与其他研究强度的衡量标准进行关联。
采用系统评价的首选报告项目（PRISMA）指南，对肱骨近端骨折治疗的 RCT 进行了系统评价。从数据库成立到 2021 年 5 月，在 PubMed、Ovid MEDLINE、Web of Science 和 Embase 数据库中进行了检索。纳入至少有 1 项统计学意义（P≤0.05）二分类结局的 RCT。使用 Fisher 精确检验为每个纳入的试验计算 FI。FI 与研究样本量和期刊影响因子相关联。
纳入了 10 项 RCT，共纳入 656 例患者，发表于 2011 年至 2020 年之间。中位患者样本量为 67（平均 65.6；范围，40-86）。并发症是最常见的报告二分类统计学显著结局。FI 的中位数为 1（平均 2.6；范围，0-18），4 项研究的 FI 为 0。中位数 FI 为 1 表明，1 例患者出现替代结局或失访，可能会改变试验对特定结局的相关结论。50%的研究中失访患者的中位数为 3（平均 4.9；范围，0-16），超过了 FI。FI 与样本量之间无相关性（Spearman 系数=0.0592，P=0.865），FI 与期刊影响因子之间也无相关性（Spearman 系数=-0.0229，P=0.522）。
在肱骨近端骨折的大多数研究中，只有 1 或 2 例患者出现替代结局或失访，才会改变对研究的二分类结局的结论。尽管 FI 不能用于评估通常是 RCT 的主要结局变量的连续性变量，但它确实提供了一个评估 RCT 时外科医生应考虑的研究强度的额外独特衡量标准。
