Go Cammille C, Maldonado David R, Go Beatrice C, Annin Shawn, Saks Benjamin R, Jimenez Andrew E, Domb Benjamin G
From the American Hip Institute Research Foundation, Chicago, IL (CC Go, Maldonado, BC Go, Annin, Saks, Jimenez, Domb), American Hip Institute, Chicago, IL (Domb), and Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA (CC Go, BC Go).
J Am Acad Orthop Surg. 2022 May 1;30(9):e741-e750. doi: 10.5435/JAAOS-D-21-00489. Epub 2022 Feb 24.
Although randomized controlled trials (RCTs) are considered the benchmark study design, a change of outcomes for a few patients can lead to a reversal of study conclusions. As such, examination of the fragility index (FI) of RCTs has become an increasingly popular method to provide further information regarding the relative robustness of RCT results. The purpose of this study was to systematically characterize and assess the predictors of the FI RCTs in total hip arthroplasty literature.
PubMed/MEDLINE, Embase, and Cochrane were systematically searched for all total hip arthroplasty RCTs published between January 2015 and December 2020, which had an equal assignment of participants to a two parallel-arm study design, examined a surgical intervention, and reported on at least one statistically significant dichotomous outcome in the abstract. Potential factors associated with the FI were examined using the Spearman correlation and Mann-Whitney U test.
Thirty-four RCTs were selected, with a median number of study participants of 111 (interquartile range [IQR] 72-171] and a median total number of events of 15 (IQR 9.5 to 29). The median FI was 2 (IQR 1 to 6), while six studies had a FI of 0. In 18 cases (52.9%), the number of patients needed to change from no event to event was less than that loss to follow-up. Larger sample size was found to predictive of a higher FI (rs = 0.367, P = 0.033), but the year of publication, journal impact factor, the calculated power analysis size, and loss to follow-up were not associated with FI.
The FI serves as a useful addition to other more commonly used approaches of quantitative analyses, such as P values, effect sizes, and confidence intervals, and widespread reporting the FI may provide clinicians with further information about RCT results.
Level I, systematic review of Level I studies.
尽管随机对照试验（RCT）被视为基准研究设计，但少数患者结局的改变可能导致研究结论的逆转。因此，检查RCT的脆弱性指数（FI）已成为一种越来越流行的方法，以提供有关RCT结果相对稳健性的更多信息。本研究的目的是系统地描述和评估全髋关节置换术文献中RCT的FI预测因素。
系统检索PubMed/MEDLINE、Embase和Cochrane数据库，查找2015年1月至2020年12月发表的所有全髋关节置换术RCT，这些研究将参与者平等分配到两个平行组研究设计中，研究一种手术干预，并在摘要中报告至少一项具有统计学意义的二分结局。使用Spearman相关性分析和Mann-Whitney U检验检查与FI相关的潜在因素。
选择了34项RCT，研究参与者的中位数为111（四分位间距[IQR]72-171），事件总数的中位数为15（IQR 9.5至29）。FI的中位数为2（IQR 1至6），而六项研究的FI为0。在18例（52.9%）中，从无事件变为有事件所需改变的患者数量少于失访患者数量。发现较大的样本量可预测较高的FI（rs = 0.367，P = 0.033），但发表年份、期刊影响因子、计算的功效分析样本量和失访与FI无关。
FI是对其他更常用的定量分析方法（如P值、效应量和置信区间）的有益补充，广泛报告FI可能会为临床医生提供有关RCT结果的更多信息。
I级，I级研究的系统评价。
