老年膝关节骨性关节炎患者跌倒风险的列线图预测模型构建及验证-现代医学

老年膝关节骨性关节炎患者跌倒风险的列线图预测模型构建及验证

单位：南京医科大学附属淮安第一医院免疫风湿科, 江苏淮安 223300

分类号：R684.3;R161.7

出版年·卷·期（页码）：2025·53·第八期（1265-1271）

摘要：

目的：构建基于列线图的老年膝关节骨性关节炎(KOA)患者跌倒风险预测模型，并验证该模型的临床应用效能。方法: 回顾性分析2022年1月至2024年1月于本院就诊的205例老年KOA患者数据资料，按照8∶2比例分为训练集(164例)、验证集(41例)，并依据患者过去6个月内是否出现跌倒分为跌倒组、无跌倒组，比较训练集中两组患者人口学资料、基础疾病、衰弱程度、Kellgren-Lawrence分级评分、单腿站立时间、Morse跌倒评估量表(MFS)、膳食多样性，利用单因素、多因素Logistic回归确定跌倒的风险因素，并利用R软件绘制列线图风险预测模型，结合验证集数据进行受试者工作特征(ROC)曲线、校准曲线及临床决策曲线分析。结果：训练集患者在6个月内出现跌倒患者共74例，跌倒发生率为45.12%。经二元Logistic回归确定，合并糖尿病、单腿站立时间短、衰弱程度重、MFS评分＞45分、膳食多样性低为老年KOA患者发生跌倒的独立危险因素。其预测老年KOA患者的跌倒风险的曲线下面积(AUC)为0.94(0.91～0.98)，准确度为0.87(0.80～0.91)，灵敏度为0.81(0.73～0.89)，特异性为0.93(0.88～0.99)，截断值为0.346。经验证集验证，该模型的AUC为0.97(0.92～0.99)，准确度为0.88(0.74～0.96)，灵敏度为0.80(0.64～0.96)，特异性为0.99(0.99～0.99)，截断值为0.346。跌倒发生的列线图的校准曲线斜率接近1，且拟合优度检验结果显示，预测概率与实际概率比较，差异无统计学意义。当高风险阈值＞0.02时，老年KOA患者易发生跌倒风险。结论：老年KOA患者跌倒风险主要与合并糖尿病、单腿站立时间、衰弱程度、膳食多样性及MFS评分有关，利用这些独立危险因素构建的基于列线图的跌倒风险预测模型具有高敏感度、高特异性特点，但其准确性仍需进一步验证。

Objective: To construct a nomogram-based prediction model of fall risk for elderly patients with knee osteoarthritis(KOA) and validate its efficiency in clinical application. Methods: The clinical data of 205 elderly patients with KOA treated in our hospital from January 2022 to January 2024 were retrospectively analyzed. The patients were divided into training set(164 cases), validation set(41 cases) according to the ratio of 8∶2. The patients were subdivided into fall group and non-fall group based on whether they had experienced fall in the past 6 months. The demographic data, underlying disease, degree of weakness, Kellgren-Lawrence grade score, time of standing on one leg， the Morse Fall Scale(MFS) score and dietary diversity were compared between the two groups in the training set. The fall risk factors were determined by using univariate, multivariate Logistic regression analyses. The nomogram risk prediction model was drawn using R software. Combined with the validation set data, the receiver operating characteristic(ROC) curve, calibration curve and clinical decision curve analyses were conducted. Results: A total of 74 patients in the training set had falls within 6 months with an incidence of 45.12%. Binary Logistic regression analysis showed that complication of diabetes mellitus, short one-leg standing time, severe frailty, MFS scores>45, and low dietary diversity were independent risk factors for falls in elderly patients with KOA. The area under curve(AUC), accuracy, sensitivity and specificity for predicting the risk of falls in elderly KOA patients were [0.94(0.91-0.98)], [0.87(0.80-0.91)], [0.81(0.73-0.89)] and [0.93(0.88-0.99)] respectively with a cutoff value of 0.346. The validation set demonstrated an AUC of [0.97(0.92-0.99)], accuracy of [0.88(0.74-0.96)], sensitivity of [0.80(0.64-0.96)], specificity of [0.99(0.99-0.99)] and a cutoff value of 0.346. The slope of the calibration curve in the nomogram of fall risk was close to 1, and the goodness-of-fit test showed a good fitting ability between the predicted probability and the actual probability with no statistically significant difference. Elderly KOA patients were at risk of fall when the high risk threshold was > 0.02. Conclusion: The risk of fall in elderly KOA patients is mainly related to complication of diabetes, one-leg standing time, frailty degree, dietary diversity and MFS scores. The nomogram-based fall risk prediction model constructed based on these independent risk factors is characterized by high sensitivity and high specificity, but its accuracy needs further verification.

参考文献：

[1] 李明, 贾丙申, 李君, 等.SMSC-EXO来源lncRNA-SNHG14对OA大鼠软骨细胞损伤及炎性痛的调节作用和机制研究[J].东南大学学报(医学版), 2023, 42(1):22-31.
[2] 赵欣宇, 余安乐, 宁铭慧, 等.老年膝骨关节炎病人跌倒危险因素及干预进展[J].全科护理, 2024, 22(9):1669-1673.
[3] 高海鹏.老年膝关节炎患者跌倒发生的影响因素[J].中国老年学杂志, 2023, 43(3):591-594.
[4] 寇文凯, 叶素妮, 陈雪芮, 等.基于多状态马尔科夫模型的跌倒状态纵向转移及其风险因素关联[J].四川大学学报(医学版), 2025, 56(1):230-238.
[5] 王聪智, 王佳智, 张林, 等.基于logistic回归构建住院老年患者衰弱预测模型及验证[J].齐齐哈尔医学院学报, 2023, 44(20):1922-1927.
[6] 叶浩翊, 吴小燕, 韩雯, 等.列线图预测DWI阴性急性缺血性脑卒中的可行性研究[J].临床放射学杂志, 2024, 43(12):2047-2052.
[7] 中华医学会骨科学分会关节外科学组.骨关节炎诊疗指南(2018年版)[J].中华骨科杂志, 2018, 38(12):705-715.
[8] 韩君, 王君俏, 谢博钦, 等.Fried衰弱表型和FRAIL量表及埃德蒙顿衰弱评估量表在社区高龄老年人衰弱筛查中一致性和适用性的比较研究[J].中国全科医学, 2021, 24(21):2669-2675.
[9] 刘鑫, 谭旭仪, 邝高艳, 等.膝骨关节炎中医证型、K-L分级与IL-1β、IL-6、TNF-α、VAS评分的相关性[J].中医学报, 2022, 37(12):2681-2687.
[10] 姜丽萍, 陆皓, 汉瑞娟, 等.老年膝关节骨性关节炎患者跌倒风险临床预测规则的构建[J].中国医科大学学报, 2020, 49(6):541-545.
[11] 周君桂.中文版Morse跌倒评估量表用于住院老年患者跌倒风险评估的初步研究[D].广州:南方医科大学, 2010.
[12] ZHENG J, ZHOU R, LI F, et al.Association between dietary diversity and cognitive impairment among the oldest-old:findings from a nationwide cohort study[J].Clin Nutr, 2021, 40(4):1452-1462.
[13] SWINKELS A, ALLAIN T J.Physical performance tests, self-reported outcomes, and accidental falls before and after total knee arthroplasty:an exploratory study[J].Physiother Theory Pract, 2013, 29(6):432-442.
[14] 张英, 薛升奎, 李超, 等.老年膝关节骨性关节炎患者跌倒效能评估及影响因素探究[J].老年医学与保健, 2024, 30(4):1140-1144.
[15] 韩清波, 钟媛, 张巧玲, 等.老年住院患者复合跌倒风险评分表的构建及应用研究[J].护理学杂志, 2024, 39(1):56-59, 71.
[16] 余娟, 陈敏, 刘璐.Morse跌倒评分联合预见性护理在恶性血液病住院放化疗患者中的应用[J].淮海医药, 2023, 41(4):416-419.
[17] 牛建梅, 赵茜, 梁景仪, 等.骨科患者围手术期的营养风险筛查与自主疼痛评估和跌倒风险的关系[J].宁夏医学杂志, 2023, 45(6):517-520, 476.
[18] 蒯春玲, 王晓宇, 舒宁波, 等.老年糖尿病视网膜病变患者跌倒风险感知现状及影响因素分析[J].生命科学仪器, 2024, 22(4):164-166.
[19] 缪婕, 计莹, 沈琳辉, 等.老年男性2型糖尿病患者血糖波动与住院跌倒风险的相关性[J].上海护理, 2024, 24(7):41-44.
[20] 徐云佳, 舒碧芸, 郑永韬, 等.基于竞争风险模型分析超高龄人群跌倒致长期卧床的危险因素及预测模型研究[J].中国全科医学, 2024, 27(18):2192-2197，2211.
[21] 程文昌, 霍少川, 刘付懿斐, 等.等速肌力训练对老年人膝关节肌力和跌倒风险的影响[J].中国医药导报, 2022, 19(19):107-110.
[22] 荆海燕, 梁碧娟, 陈玉兰, 等.老年衰弱前期患者跌倒恐惧及其影响因素现状调查研究[J].中国医药导报, 2024, 21(29):119-124.
[23] 周志庆, 刘欢, 陈梦奇, 等.老年住院患者跌倒影响因素及风险列线图模型的构建[J].济宁医学院学报, 2024, 47(04):295-299.
[24] 李九红, 陈婷, 黄伶智, 等.3种跌倒风险评估工具对养老机构中老年人跌倒风险的预测价值比较[J].护理研究, 2024, 38(13):2296-2301.
[25] 张林英, 梅群超, 张炜琦, 等.抗阻训练联合高乳蛋白营养在老年2型糖尿病伴肌少症病人中的应用研究[J].实用老年医学, 2023, 37(10):995-999.
[26] 邬馨益, 徐林燕, 陈卓琦, 等.老年人运动认知功能减退综合征风险预测模型的构建与验证[J].护理学报, 2024, 31(10):73-78.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】

提示：您还未登录，请登录！点此登录