帕金森病患者脑深部电刺激术后肺部并发症列线图预测模型的构建及其预测价值

Nomogram model construction for predicting pulmonary complications after deep brain stimulation and its prodictive value in patients with Parkinson's disease

目的分析帕金森病患者脑深部电刺激术后发生肺部并发症的影响因素,构建预测术后肺部并发症发生风险的列线图模型。方法中国科技大学附属第一医院神经外科自2015年3月至2019年12月采用脑深部电刺激术(DBS)治疗帕金森病患者272例,其中术后发生肺部并发症56例(肺部并发症组),无肺部并发症216例(无肺部并发症组)。回顾性比较肺部并发症组和无肺部并发症组患者的临床资料,多因素Logistic回归分析确定帕金森病患者术后发生肺部并发症的影响因素,根据多因素Logistic回归分析结果构建预测帕金森病患者术后肺部并发症发生风险的列线图模型,采用受试者工作特征(ROC)曲线分析列线图模型的预测效能。结果与无肺部并发症组比较,肺部并发症组患者中术前肺部疾病史、术前白蛋白<35 g/L、术前1 s用力呼气量(FEV1)/用力肺活量(FVC)<60%、手术时间≥180 min、年龄≥70岁者所占比例较高,术前血红蛋白含量、帕金森病睡眠量表(PDSS)评分、统一帕金森病评定量表运动部分(UPDRSⅢ)评分较低,术后住院时间较长,差异均有统计学意义(P<0.05)。多因素Logistic回归分析结果显示术前肺部疾病史(OR=4.230,95%CI:2.035~8.207,P=0.002)、术前白蛋白<35 g/L(OR=6.159,95%CI:2.570~5.091,P=0.014)、术前FEV1/FVC<60%(OR=31.771,95%CI:6.702~66.412,P=0.000)、手术时间≥180 min(OR=3.550,95%CI:2.261~10.065,P=0.009)、年龄≥70岁(OR=3.714,95%CI:1.451~4.827,P=0.001),PDSS评分(OR=1.017,95%CI:1.351~13.880,P=0.043)为帕金森病患者术后发生肺部并发症的独立危险因素。基于上述因素建立预测帕金森病患者术后肺部并发症发生风险的列线图模型,ROC曲线显示列线图模型预测患者术后肺部并发症发生的曲线下面积(AUC)为0.841(95%CI:0.774~0.904,P=0.000),灵敏度为84.03%,特异度为75.06%。结论有肺部疾病史、术前白蛋白<35 g/L、术前FEV1/FVC<60%、手术时间≥180 min、年龄≥70岁及PDSS评分较低的帕金森病患者DBS后易发生肺部并发症。基于上述变量构建的列线图模型预测术后肺部并发症发生风险的效能较高。

Objective To explore the influencing factors for pulmonary complications of patients with Parkinson's disease(PD)after deep brain stimulation(DBS),and to construct a nomogram model for predicting pulmonary complications after DBS.Methods Two hundred and seventy-two patients with PD accepted DBS,admitted to our hospital from March 2015 to December 2019,were chosen in our study;they were divided into pulmonary complication group(n=56)and non-pulmonary complication group(n=216).The clinical data of patients from the two groups were compared retrospectively.Multivariate Logistic regression was used to analyze the risk factors for pulmonary complications of patients with PD after DBS,and a nomogram model was established to predict the risk of pulmonary complications;receiver operating characteristic(ROC)curve was used to analyze the prediction performance of the model.Results As compared with non-pulmonary complication group,the pulmonary complication group had significantly higher percentages of patients with history of pulmonary disease,preoperative albumin<35 g/L,preoperative forced expiratory volume in 1 s(FEV1)/forced vital capacity(FVC)<60%,operation time≥180 min,and age≥70 years,and significantly lower Parkinson's disease sleep scale(PDSS)scores,unified Parkinson's disease rating scaleⅢ(UPDRSⅢ)scores,and preoperative albumin level,and statistically longer postoperative hospital stays(P<0.05).Multivariate Logistic regression analysis results showed that history of pulmonary disease(OR=4.230,95%CI:2.035-8.207),preoperative albumin<35 g/L(OR=6.159,95%CI:2.570-5.091),preoperative FEV1/FVC<60%(OR=31.771,95%CI:6.702-66.412),operation time≥180 min(OR=3.550,95%CI:2.261-10.065),age≥70 years(OR=3.714,95%CI:1.451-4.827),and PDSS scores(OR=1.017,95%CI:1.351-13.880)were the independent risk factors for pulmonary complications of patients with PD after DBS.Nomogram model established by using the above indicators showed that area under the curve for predicting pulmonary complications of patients with PD after DBS was 0.841(95%CI:0.774-0.904,P=0.000),with sensitivity of 84.03%and specificity of 75.06%.Conclusion PD patients with history of pulmonary disease,preoperative albumin<35 g/L,preoperative FEV1/FVC<60%,operation time≥180 min,age≥70 years,and low PDSS scores are trend to have pulmonary complications after DBS;the nomogram model based on the above variables is highly effective in predicting the risk of postoperative pulmonary complications.