Wu Tongyang, Li Bowen
Rehabilitation Medical Center, Affiliated Hospital of Shandong Second Medical University, No.2428 Yuhe Road, Weifang, 261000, Shandong, China.
Department of Physical and Rehabilitation Medicine, Tianjin Medical University General Hospital, No. 154 Anshan Road, Tianjin, 300052, China.
Sci Rep. 2025 May 1;15(1):15283. doi: 10.1038/s41598-025-00291-0.
Handgrip strength is a key indicator of overall health in older adults, and its decline is linked to various adverse health outcomes. Despite numerous studies on factors influencing handgrip strength, few attempts have integrated multiple factors into a practical clinical tool. This study aims to develop and validate a nomogram based on a logistic regression model to predict the risk of low handgrip strength in older adults. Using data from the China Health and Retirement Longitudinal Study (CHARLS), 1138 participants were included. Firth-adjusted logistic regression identified predictors of low handgrip strength, with variable selection based on the Bayesian Information Criterion (BIC). Model performance was assessed using calibration curves, ROC curves, and decision curve analysis (DCA). Internal validation was performed with 10-fold cross-validation and bootstrapping, determining the optimal risk threshold. Key predictors identified included age, chronic disease history, marital status, lifestyle, education, BMI, activities of daily living, and glycated hemoglobin. The simplified model exhibited good discriminatory ability (AUC = 0.78) and calibration performance. The optimal threshold (0.40) yielded sensitivity of 72.5% and specificity of 69.8%. Decision curve analysis confirmed significant net benefit within the clinically relevant threshold range. The nomogram provides a practical tool for identifying at-risk individuals and guiding intervention, integrating modifiable and non-modifiable factors for personalized risk assessment and early intervention.
握力是老年人整体健康的关键指标，其下降与各种不良健康结果相关。尽管有大量关于影响握力因素的研究，但很少有人尝试将多种因素整合到一个实用的临床工具中。本研究旨在开发并验证一种基于逻辑回归模型的列线图，以预测老年人握力低下的风险。利用中国健康与养老追踪调查（CHARLS）的数据，纳入了1138名参与者。费思调整逻辑回归确定了握力低下的预测因素，并基于贝叶斯信息准则（BIC）进行变量选择。使用校准曲线、ROC曲线和决策曲线分析（DCA）评估模型性能。通过10倍交叉验证和自抽样进行内部验证，确定最佳风险阈值。确定的关键预测因素包括年龄、慢性病病史、婚姻状况、生活方式、教育程度、体重指数、日常生活活动和糖化血红蛋白。简化模型表现出良好的区分能力（AUC = 0.78）和校准性能。最佳阈值（0.40）产生的灵敏度为72.5%，特异度为69.8%。决策曲线分析证实了在临床相关阈值范围内显著的净效益。该列线图为识别高危个体和指导干预提供了一个实用工具，整合了可改变和不可改变的因素用于个性化风险评估和早期干预。
