ICU重症患者出现获得性衰弱的预测模型与验证

Prediction Model and Validation of Acquired Weakness in Critically Ill Patients in ICU

目的探讨重症监护室(ICU)内重症患者出现获得性衰弱的独立影响因素,构建预测模型并实施预测效果的验证。方法选择2017年1月至2022年12月期间睢县中医院ICU接诊治疗的患者368人的数据实施回顾性分析。采集发生ICU获得性衰弱患者的资料,进行单因素分析、多因素分析以及预测模型构建和预测效能分析。结果纳入本次调查的368名患者中,有189名患者判定为出现ICU获得性衰弱,发生率为51.36%。单因素分析结果显示,出现和未出现ICU获得性衰弱患者的入驻ICU时间长度,急性生理与慢性健康评分(APACHEⅡ)评分,使用神经肌肉阻滞剂情况,血乳酸最高值的数据差异有统计学意义(P<0.05)。多因素分析结果显示,入驻ICU时间长度,APACHEⅡ评分,使用神经肌肉阻滞剂情况,血乳酸最高值是患者发生ICU获得性衰弱的独立影响因素,差异有统计学意义(P<0.05)。依据多因素分析所筛选出来的变量构建列线图风险模型。C-index为0.713。利用Bootstrap自抽样法进行内部验证,重复自抽样1000次,获得校准曲线,平均绝对误差为0.043。利用logistic回归模型的独立影响因素以及P值预测概率对患者发生ICU获得性衰弱的情况进行受试者特征曲线(ROC)曲线的预测,约登指数分别为19.97%、32.92%、15.11%、37.30%、47.20%。结论ICU患者具有较高的获得性衰弱发生风险,若干因素均是该种病变出现的影响因素。利用这些影响因素构建的预测模型具有良好的预测效能,另外需要在工作中对这些影响因素进行监控,以便早期发现和干预。

Objective To explore the independent influencing factors of acquired weakness in critically ill patients in the ICU,construct a predictive model,and validate the predictive effect.Methods A retrospective analysis was conducted on the data of 368 patients treated in Sui County Traditional Chinese Medicine Hospital ICU from January 2017 to December 2022.The data of patients with acquired frailty in ICU were collected for univariate analysis,multi-factor analysis,prediction model construction and prediction efficacy analysis.Results Among the 368 patients included in this survey,189 patients were diagnosed with ICU acquired weakness,with an incidence rate of 51.36%.The results of univariate analysis showed that there were statistically significant differences in the length of ICU admission time,The Acute Physical and Chronic Health score(APACHEⅡ)score,use of neuromuscular blockers,and highest blood lactate levels between patients with and without ICU acquired weakness(P<0.05).The results of multivariate analysis showed that the length of time spent in the ICU,APACHEⅡscore,use of neuromuscular blockers,and highest blood lactate levels were independent influencing factors for the occurrence of ICU acquired weakness in patients(P<0.05).Construct a column chart risk model based on the variables selected through multiple factor analysis.The C-index is 0.713.Using Bootstrap self sampling method for internal validation,repeat self sampling 1000 times to obtain a calibration curve with an average absolute error of 0.043.The independent influencing factors and P-value prediction probability of logistic regression model were used to predict the ROC curve of patients with ICU acquired weakness.The youden indices were 19.97%,32.92%,15.11%,37.30%,and 47.20%,respectively.Conclusion ICU patients have a high risk of acquired weakness,and several factors are influencing the occurrence of this disease.The prediction model constructed using these influencing factors has good predictive performance,and it is also necessary to monitor these influencing factors in work for early detection and intervention.