Scholarlink

Background <p>Hospital-acquired infections (HAIs) are a major global health concern, increasing patient morbidity, mortality, and healthcare costs. While randomized controlled trial (RCT)-based meta-analyses (MAs) have played a critical role in guiding infection control strategies, their methodological quality and robustness remain uncertain. The fragility index (FI), an event-level metric, offers a complementary perspective to conventional sensitivity analyses by quantifying the stability of statistical significance against small changes in event counts.</p> Methods <p>A systematic search of PubMed, EMBASE, and Web of Science was conducted for RCT-based MAs on HAI management published between January 2020 and June 2025. Eligible MAs reported pooled binary outcomes as RR, OR, or RD. FI was calculated using an online calculator. Study characteristics were extracted, and potential determinants of robustness were examined using LASSO-based variable selection followed by multivariable logistic regression, and further interpreted using Shapley additive explanations (SHAP).</p> Results <p>A total of 171 studies comprising 440 MAs were included. The overall median FI was 7, with 61.14% of MAs demonstrating FI &gt; 5, indicating moderate robustness. Logistic regression identified the number of included RCTs (OR: 1.06, 95% CI: 1.02–1.11) and total number of events (per 50 events) (OR: 1.04, 95% CI: 1.01–1.07) as significant factors associated with robustness. SHAP analysis confirmed their predominant influence on FI.</p> Conclusion <p>Most RCT-based MAs in HAI management exhibited moderate robustness, with FI strongly driven by the number of included RCTs and events. Routine reporting of FI in future MAs is recommended to enhance transparency and reliability. Integration with complementary approaches such as trial sequential analysis and bias adjustment may further strengthen evidence synthesis in infection management.</p> Clinical trial number <p>Not applicable.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fragility index–based robustness assessment of RCT-driven meta-analyses in hospital-acquired infection management

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月27日

查看全文
  • Jiacheng Li,
  • Yi Guo,
  • Mengmeng Zhou,
  • Kaixuan Zhao,
  • Yinhua Zheng,
  • Wei Lu

摘要

Background

Hospital-acquired infections (HAIs) are a major global health concern, increasing patient morbidity, mortality, and healthcare costs. While randomized controlled trial (RCT)-based meta-analyses (MAs) have played a critical role in guiding infection control strategies, their methodological quality and robustness remain uncertain. The fragility index (FI), an event-level metric, offers a complementary perspective to conventional sensitivity analyses by quantifying the stability of statistical significance against small changes in event counts.

Methods

A systematic search of PubMed, EMBASE, and Web of Science was conducted for RCT-based MAs on HAI management published between January 2020 and June 2025. Eligible MAs reported pooled binary outcomes as RR, OR, or RD. FI was calculated using an online calculator. Study characteristics were extracted, and potential determinants of robustness were examined using LASSO-based variable selection followed by multivariable logistic regression, and further interpreted using Shapley additive explanations (SHAP).

Results

A total of 171 studies comprising 440 MAs were included. The overall median FI was 7, with 61.14% of MAs demonstrating FI > 5, indicating moderate robustness. Logistic regression identified the number of included RCTs (OR: 1.06, 95% CI: 1.02–1.11) and total number of events (per 50 events) (OR: 1.04, 95% CI: 1.01–1.07) as significant factors associated with robustness. SHAP analysis confirmed their predominant influence on FI.

Conclusion

Most RCT-based MAs in HAI management exhibited moderate robustness, with FI strongly driven by the number of included RCTs and events. Routine reporting of FI in future MAs is recommended to enhance transparency and reliability. Integration with complementary approaches such as trial sequential analysis and bias adjustment may further strengthen evidence synthesis in infection management.

Clinical trial number

Not applicable.