文献速递骨质疏松与代谢性骨病创伤骨科

预测工具识别非创伤性骨折风险个体的性能:系统综述、Meta分析与Meta回归

Performance of predictive tools to identify individuals at risk of non-traumatic fracture: a systematic review, meta-analysis, and meta-regression.
PubMed 文献 · Osteoporos Int · 2019-03-14

该系统综述、Meta分析与Meta回归评估了现有工具预测非创伤性骨折的准确性,共纳入53项验证研究、涉及14种工具。未调整分析中QFracture预测髋部骨折的AUC最高为0.88;调整分析中FRAX联合骨密度(AUC=0.81)与Garvan联合骨密度(AUC=0.79)最高。QFracture、FRAX联合骨密度与Garvan联合骨密度区分能力最佳,但分析存在显著异质性。

要点

  • QFracture、FRAX联合骨密度与Garvan联合骨密度在预测骨折方面区分能力最佳。
  • 未调整分析中QFracture预测髋部骨折AUC=0.88,为所评估工具中最高。
  • 调整分析中FRAX联合骨密度AUC=0.81、Garvan联合骨密度AUC=0.79,为最高。

核心结论

  • QFracture、FRAX联合骨密度与Garvan联合骨密度对骨折的预测区分能力最高。
  • 作者认为可在同一批个体中评估现有工具性能的额外研究以确认该结论。

研究设计与样本

  • 系统综述、Meta分析与Meta回归,检索MEDLINE、EMBASE、Evidence-Based Medicine Reviews与Global Health。
  • 纳入在独立于推导队列的人群中评估区分度的研究,共53项验证研究、评估14种工具。
  • 仅FRAX、Garvan与QFracture因研究数量足够而进入Meta回归模型。
  • 以受试者工作特征曲线下面积(AUC)为效应量,调整变量包括性别、平均年龄、年龄范围与研究质量。

主要发现

  • 未调整分析中,QFracture预测髋部骨折的区分能力最佳,AUC=0.88。
  • 调整分析中,FRAX联合骨密度AUC=0.81、Garvan联合骨密度AUC=0.79,为最高。
  • 预测主要骨质疏松性骨折时,QFracture区分能力最佳,AUC=0.77。
  • 预测骨质疏松性骨折或任何骨折时,FRAX联合骨密度AUC=0.72 vs 不联合骨密度AUC=0.69,Garvan联合骨密度AUC=0.72 vs 不联合骨密度AUC=0.65。
  • 分析中存在显著异质性。

关键数值与推荐

  • QFracture预测髋部骨折AUC=0.88(未调整分析)。
  • FRAX联合骨密度AUC=0.81、Garvan联合骨密度AUC=0.79(调整分析)。
  • QFracture预测主要骨质疏松性骨折AUC=0.77。
  • FRAX联合骨密度AUC=0.72 vs 不联合骨密度AUC=0.69;Garvan联合骨密度AUC=0.72 vs 不联合骨密度AUC=0.65。

临床怎么用

  • 原文未据此给出推荐。

局限性

  • 分析中存在显著异质性。
  • 部分工具的研究数量少,仅FRAX、Garvan与QFracture用Meta回归模型比较。
  • 作者认为需在同一个体中进行额外研究以确认结论。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

UNLABELLED: There is no consensus on which tool is the most accurate to assess fracture risk. The results of this systematic review suggest that QFracture, Fracture Risk Assessment Tool (FRAX) with BMD, and Garvan with BMD are the tools with the best discriminative ability. More studies assessing the comparative performance of current tools are needed. INTRODUCTION: Many tools exist to assess fracture risk. This review aims to determine which tools have the best predictive accuracy to identify individuals at high risk of non-traumatic fracture. METHODS: Studies assessing the accuracy of tools for prediction of fracture were searched in MEDLINE, EMBASE, Evidence-Based Medicine Reviews, and Global Health. Studies were eligible if discrimination was assessed in a population independent of the derivation cohort. Meta-analyses and meta-regressions were performed on areas under the ROC curve (AUCs). Gender, mean age, age range, and study quality were used as adjustment variables. RESULTS: We identified 53 validation studies assessing the discriminative ability of 14 tools. Given the small number of studies on some tools, only FRAX, Garvan, and QFracture were compared using meta-regression models. In the unadjusted analyses, QFracture had the best discriminative ability to predict hip fracture (AUC = 0.88). In the adjusted analysis, FRAX with BMD (AUC = 0.81) and Garvan with BMD (AUC = 0.79) had the highest AUCs. For prediction of major osteoporotic fracture, QFracture had the best discriminative ability (AUC = 0.77). For prediction of osteoporotic or any fracture, FRAX with BMD and Garvan with BMD had higher discriminative ability than their versions without BMD (FRAX: AUC = 0.72 vs 0.69, Garvan: AUC = 0.72 vs 0.65). A significant amount of heterogeneity was present in the analyses. CONCLUSIONS: QFracture, FRAX with BMD, and Garvan with BMD have the highest discriminative performance for predicting fracture. Additional studies in which the performance of current tools is assessed in the same individuals may be performed to confirm this conclusion.

来源:Osteoporos Int,查看原文。版权归原作者所有。

引用

  1. Beaudoin C, Moore L, Gagné M, et al. Performance of predictive tools to identify individuals at risk of non-traumatic fracture: a systematic review, meta-analysis, and meta-regression. Osteoporos Int. 2019. PMID: 30877348. doi:10.1007/s00198-019-04919-6

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。