Abstract:

Objective To construct a risk prediction model for cognitive frailty in elderly patients with diabetic retinopathy（DR），providing references for the prevention of cognitive frailty in elderly DR patients. Methods 320 elderly DR patients from November 2024 to April 2025 were selected in a tertiary A ophthalmology hospital in Nanchang by convenient sampling method. The survey was conducted using general data questionnaires，the frailty risk screening tool for elder inpatients，the Mini-Mental State Examination Scale，the Mini Nutritional Assessment Scale，the Short version of Center for Epidemiologic Studies Depression Scale，and the Visual Function Related Quality of Life Scale-25. The risk factors for cognitive frailty in elderly DR patients were explored using univariate analysis，Lasso regression，and binary Logistic regression analysis. A nomogram prediction model was constructed，and the Hosmer-Lemeshow test，receiver operator characteristic curve，and calibration curve were applied to evaluate the model’s fitting performance. Internal validation was conducted through the enhanced Bootstrap method with 1，000 re-samplings. Results A total of 314 valid questionnaires were collected，and the incidence of cognitive frailty in elderly DR patients was 27.39%. Age，the number of chronic diseases，the degree of visual impairment，the quality of life related to visual function，depression，and nutritional status are the influencing factors of cognitive frailty in elderly DR patients（P<0.05）. Hosmer-Lemeshow test showed that χ²=9.530，P=0.300；the receiver operator characteristic curve was 0.936；the sensitivity was 0.919；the specificity was 0.820；the calibration curve had a Brier score of 0.091 and a mean prediction error of 0.021；the internal validation results of Bootstrap method showed that the specificity was 0.908 and the sensitivity was 0.672. Conclusion The risk predictive model of cognitive frailty in elderly DR patients has good predictive power，enabling clinical staff to identify the elderly DR patients with high risk of cognitive weakness and to take preventive measures.

Key words: Diabetic Retinopathy, Cognitive Frailty, Predictive Model, Ophthalmic Nursing

摘要：

目的 构建老年糖尿病视网膜病变（diabetic retinopathy，DR）患者认知衰弱风险预测模型，为老年DR患者认知衰弱的预防提供参考。 方法 便利选取2024年11月—2025年4月于南昌市某三级甲等眼科专科医院就诊的320例老年DR患者为调查对象。采用一般资料调查问卷、老年住院患者衰弱风险筛查工具、简易智能精神状态检查量表、微型营养评定量表、简版流调中心抑郁量表、视功能相关生存质量量表-25进行调查。运用单因素分析、Lasso回归分析、二元Logistic回归分析探讨老年DR患者认知衰弱的风险因素，绘制老年DR患者认知衰弱列线图预测模型，通过Hosmer-Lemeshow检验、受试者操作特征曲线下面积和校准曲线检验模型的预测效果及校准度，并通过Bootstrap法重复抽样1 000次进行内部验证。 结果 回收有效问卷314份，老年DR患者认知衰弱发生率为27.39%。年龄、慢性病数量、视觉损伤程度、视功能相关生存质量、抑郁、营养状况是老年DR患者认知衰弱的影响因素（P<0.05）。Hosmer-Lemeshow检验结果显示，χ²=9.530，P=0.300；受试者操作特征曲线下面积为0.936，灵敏度为0.919，特异度为0.820；校准曲线Brier评分为0.091，平均预测误差为0.021；Bootstrap法内部验证结果显示，特异度为0.908，灵敏度为0.672。 结论 老年DR患者认知衰弱风险预测模型具有较好的预测效能，可为医护人员早期识别高危患者提供参考。

关键词: 糖尿病视网膜病变, 认知衰弱, 预测模型, 眼科护理学