Scholarlink

Background <p>A variety of frailty assessment tools are used internationally, and there is variation in the selection of frailty assessment tools for patients undergoing maintenance hemodialysis (MHD). The most suitable tool for this population remains unclear. This study compared the validity of four frailty assessment tools—the Fried Phenotype, FRAIL Scale, Clinical Frailty Scale (CFS), and Tilburg Frailty Index (T)—among MHD patients. The aim is to provide a reference for healthcare professionals selecting appropriate tools.</p> Methods <p>A convenience sample of 385 patients undergoing maintenance hemodialysis at two hemodialysis centers in Sichuan, China, from January to June 2025 was selected as the study population. All patients were assessed for frailty severity using four instruments. To minimize inclusion bias, the “leave-one-out” composite index (CI) was employed as the reference standard. Kappa statistics and McNemar’s test were used to evaluate agreement between each tool and its corresponding CI. Diagnostic performance was assessed using receiver operating characteristic (ROC) curves, area under the curve (AUC), sensitivity, specicity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. Decision curve analysis (DCA) evaluated clinical utility.</p> Results <p>The Fried Phenotype demonstrated the strongest agreement with its CI (Kappa = 0.789, P &lt; 0.001), indicating excellent concordance and no signicant marginal heterogeneity (McNemar’s P = 1.000). The CFS also showed good agreement with its CI (Kappa = 0.716, P &lt; 0.001) without signicant marginal heterogeneity (McNemar’s P = 0.074). In contrast, while the FRAIL and T showed moderate agreement with their respective CIs (Kappa = 0.705 and 0.537; both P &lt; 0.001), McNemar’s test indicated signicant marginal heterogeneity for both (P &lt; 0.001). The AUC was highest for the Fried Phenotype (0.90), followed by CFS (0.85), T (0.84), and FRAIL (0.82). The Fried Phenotype achieved the best balance between sensitivity (93%) and specicity (86%). Decision curve analysis conrmed its superior net benet across most clinical threshold probabilities.</p> Conclusion <p>The Fried Phenotype demonstrated the highest validity and diagnostic accuracy for frailty screening in MHD patients when benchmarked against a robust “leave-one-out” composite standard. The CFS also performed well with good agreement and discriminative ability. The FRAIL scale showed high sensitivity but lower specicity, while the T exhibited high specicity but poor sensitivity. Tool selection should be guided by clinical purpose—whether for screening, conrmation, or balanced assessment—to optimize frailty management in MHD patients and improve clinical outcomes.</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

Comparison of diagnostic characteristics of four frailty screening tools in maintenance hemodialysis patients

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月16日

查看全文
  • Xuemei Guo,
  • Bingjie Yang,
  • Jingwen Zhang,
  • Jingyan Zhang,
  • Xueming Jing,
  • Min Tan

摘要

Background

A variety of frailty assessment tools are used internationally, and there is variation in the selection of frailty assessment tools for patients undergoing maintenance hemodialysis (MHD). The most suitable tool for this population remains unclear. This study compared the validity of four frailty assessment tools—the Fried Phenotype, FRAIL Scale, Clinical Frailty Scale (CFS), and Tilburg Frailty Index (T)—among MHD patients. The aim is to provide a reference for healthcare professionals selecting appropriate tools.

Methods

A convenience sample of 385 patients undergoing maintenance hemodialysis at two hemodialysis centers in Sichuan, China, from January to June 2025 was selected as the study population. All patients were assessed for frailty severity using four instruments. To minimize inclusion bias, the “leave-one-out” composite index (CI) was employed as the reference standard. Kappa statistics and McNemar’s test were used to evaluate agreement between each tool and its corresponding CI. Diagnostic performance was assessed using receiver operating characteristic (ROC) curves, area under the curve (AUC), sensitivity, specicity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. Decision curve analysis (DCA) evaluated clinical utility.

Results

The Fried Phenotype demonstrated the strongest agreement with its CI (Kappa = 0.789, P < 0.001), indicating excellent concordance and no signicant marginal heterogeneity (McNemar’s P = 1.000). The CFS also showed good agreement with its CI (Kappa = 0.716, P < 0.001) without signicant marginal heterogeneity (McNemar’s P = 0.074). In contrast, while the FRAIL and T showed moderate agreement with their respective CIs (Kappa = 0.705 and 0.537; both P < 0.001), McNemar’s test indicated signicant marginal heterogeneity for both (P < 0.001). The AUC was highest for the Fried Phenotype (0.90), followed by CFS (0.85), T (0.84), and FRAIL (0.82). The Fried Phenotype achieved the best balance between sensitivity (93%) and specicity (86%). Decision curve analysis conrmed its superior net benet across most clinical threshold probabilities.

Conclusion

The Fried Phenotype demonstrated the highest validity and diagnostic accuracy for frailty screening in MHD patients when benchmarked against a robust “leave-one-out” composite standard. The CFS also performed well with good agreement and discriminative ability. The FRAIL scale showed high sensitivity but lower specicity, while the T exhibited high specicity but poor sensitivity. Tool selection should be guided by clinical purpose—whether for screening, conrmation, or balanced assessment—to optimize frailty management in MHD patients and improve clinical outcomes.