老年人粪或尿失禁发生风险的预测模型研究

Predictive model of fecal or urinary incontinence among older adults in China

目的构建老年人粪/尿失禁发生风险的预测模型并进行验证。方法采用2018年进行的第7次中国健康长寿纵向调查的横断面调查数据, 以问卷中"您是否能控制大小便"为主要效应指标, 利用受试者工作特征曲线(ROC)寻找以小腿围预测老年人粪/尿失禁的最佳截点, 并通过单因素Logistic回归探讨老年人粪/尿失禁的相关影响因素。采用随机抽样的方法抽取70%的调查数据作为训练集, 剩余30%的调查数据作为测试集, 在训练集中用多因素Logistic回归分析方法建立一个囊括所有预测因子的预测模型, 绘制列线图, 并利用测试集对模型的预测效能进行验证。结果多因素Logistic回归分析结果显示, 年龄、较小的小腿围(男性<28.5 cm、女性<26.5 cm)、无法连续步行1 km、无法提起5 kg重物、无法连续做3次蹲起、日常活动能力受限和有泌尿系统、神经系统和脑血管疾病史是老年人粪/尿失禁的危险因素(均P<0.05), 女性、较好的社会经济状况以及正常的体质指数(BMI)是粪/尿失禁的保护因素。使用上述12个因素构建中国老年人粪/尿失禁状态的Logistic回归预测模型, 该模型一致性指数(C-index)值为0.907, 表明该模型有较好的预测能力。训练集和测试集预测模型校正曲线的吻合度均较好, 整体样本、训练集、测试集ROC曲线下面积(AUC)分别为0.906(95%CI:0.896~0.917)、0.907(95%CI:0.894~0.921)、0.910(95%CI:0.892~0.928), 表明该模型预测能力较高, 区分度良好。结论年龄、性别、小腿围、连续步行1 km、提起5 kg重物、连续做3次蹲起、日常活动能力、泌尿系统、神经系统和脑血管疾病史、社会经济状况和BMI是老年人粪/尿失禁的影响因素, 基于上述指标构建的列线图对老年人粪/尿失禁有较好的预测效能。

Objective To construct and validate a predictive model of fecal/urinary incontinence among older adults in China.Methods Data was obtained from the Seventh Chinese Longitudinal Healthy Longevity Survey in 2018.In the questionnaire,"Are you able to control your bowel and urine",was regarded as the main effect indicator.Receiver operating curves(ROC)were used to find the best cut-off values of calf circumference for predicting fecal/urinary incontinence,and univariate Logistic model method was used to explore the potential factors associated with fecal/urinary incontinence among community-living older adults in China.A random sampling method was used to extract 70% of the survey data as the training set,and the remaining 30% of the survey data as the test set.A multivariate Logistic regression analysis was conducted in the training set to build a prediction model that encompassed all predictors,and a nomogram was plotted.Results Logistic regression analysis showed that age,small calf circumference(male<28.5 cm,female<26.5 cm),inability to walk 1 km continuously,inability to lift 5 kg items,inability to do three consecutive squats,limited daily activities,and a history of urinary system disorders,nervous system disorders,and cerebrovascular disorders were all risk factors for fecal/urinary incontinence for older adults in China.Female,better socioeconomic status,and normal body mass index were protective factors for fecal/urinary incontinence.The Logistic regression model for predicting fecal/urinary incontinence among Chinese older adults was constructed using the above twelve factors.The consistency index(Cindex)value of the model was 0.907,indicating that the model had good predictive ability.The area under the ROC curve(AUC)of the overall sample,training set and test set were 0.906(95%CI:0.896-0.917),0.907(95%CI:0.894-0.921)and 0.910(95%CI:0.892-0.928),respectively,indicating that the model had high prediction ability and good discrimination.Conclusions Age,sex,calf circumference,ability to walk 1 km continuously,ability to lift 5 kg items,ability to do three consecutive squats,daily activities,history of urinary system disorders,nervous system disorders and cerebrovascular disorders,socioeconomic status,and body mass index were independent predictors for fecal/urinary incontinence among older adults in China.The nomogram based on the above indicators has a good predictive effect on fecal/urinary incontinence for older adults.