脑卒中后疲劳列线图预测模型的构建及评估

Construction and Verification of the Nomogram Prediction Model of Post-stroke Fatigue

目的:分析脑卒中后疲劳(post-stroke fatigue,PSF)的危险因素,构建并评估列线图预测模型。方法:选取2020年4月—2021年4月福建医科大学附属协和医院收治的233例脑卒中患者。收集所有患者的基线资料,并根据是否出现PSF将所有患者分为PSF组和非PSF组。统计分组情况,对PSF进行单因素及多因素分析,并构建及验证PSF列线图模型。结果:233例脑卒中患者中发生PSF 108例,发生率高达46.4%(108/233),纳入PSF组,125例未发生PSF,纳入非PSF组。单因素分析表明,PSF组年龄>65岁、糖尿病、脑卒中后疼痛及抑郁状态占比均高于非PSF组(P<0.05)。年龄大、脑卒中后疼痛、糖尿病及抑郁状态均是PSF的独立危险因素(P<0.05)。列线图预测模型ROC曲线下面积为0.711(95%CI:0.645,0.778),绘制的校准曲线显示预测值和实际值趋于一致。H-L检验结果表明模型的区分度和校准度良好(χ^(2)=9.178,P=0.328)。结论:基于高危因素构建的列线图预测模型具有较好的预测效能,为识别PSF的高风险患者及制定干预措施提供借鉴。

Objective: To analyze the risk factors of post-stroke fatigue(PSF), and to establish and evaluate nomogram prediction model. Method: A total of 233 stroke patients admitted to the Fujian Medical University Union Hospital from April 2020 to April 2021were selected. The baseline data of all patients were collected, and all patients were divided into PSF group and non PSF group according to the presence or absence of PSF. The grouping condition was counted, the PSF was analyzed by single factor and multiple factor, and the PSF nomogram model was constructed and verified. Result: Among 233 stroke patients, 108 had PSF, with an incidence rate of 46.4%(108/233). They were included in the PSF group, while 125 had no PSF, and were included in the non PSF group. Univariate analysis showed that the proportions of age >65 years old, diabetes, post-stroke pain and depression state in the PSF group were higher than those in the non PSF group(P<0.05). Older age, post-stroke pain, diabetes and depression state were independent risk factors for PSF(P<0.05).The area under curve of the nomogram prediction model ROC was 0.711(95%CI: 0.645, 0.778), and the plotted calibration curve showed that the predicted value was consistent with the actual value. The H-L test result was showed that the model had good discrimination and calibration( χ^(2)=9.178, P=0.328). Conclusion: The nomogram prediction model constructed based on high-risk factors has good prediction efficiency, which can provide reference for identifying PSF high-risk patients and formulating intervention measures.