目的 评估床旁膈肌超声在预测胸腔镜肺叶切除术后肺部并发症（PPCs）中的应用价值。 方法 收集河南省人民医院择期行胸腔镜肺叶切除术患者，记录一般资料、围术期相关指标；术前，术后1、3、5 d床旁超声检测膈肌活动度（DE1d、DE3d、DE5d）。根据是否发生PPCs将患者分为两组（PPCs组和非PPCs组）。将单变量分析中有统计学差异的参数纳入多变量logistic回归分析中，筛选PPCs的独立影响因素。绘制受试者工作曲线（ROC），并通过曲线下面积（AUC）评估膈肌超声预测PPCs的效能。 结果 共纳入患者934例，其中537例发生PPCs（57.5%）。单因素分析显示：与非PPCs组比较，PPCs组糖尿病患者占比、术后24 h活动VAS评分增加；胸管拔除时间、单肺通气时间、术后住院时间长；PPCs组患者术后DE1d、3d手术侧、健侧；DE5d手术侧DE值低（P < 0.05）。多因素logistic回归分析显示：单肺通气时间延长、糖尿病、术后（DE1d、3d）手术侧、健侧和DE5d手术侧DE值降低是发生PPCs的独立危险因素。ROC曲线显示：术后DE1d、3d的手术侧、健侧预测PPCs的效能较高（AUC分别为0.797、0.821、0.933、0.929）；术后DE5d手术侧、健测预测PPCs效能较弱（AUC分别为0.703、0.512）。 结论 术后相应时间DE降低、单肺通气时间延长、糖尿病是胸腔镜肺叶切除术后发生PPCs的独立危险因素，术后3 d应用床旁超声评价DE对PPCs具有较高的预测价值。

Objective To assess the predictive value of bedside diaphragmatic ultrasound in predicting pulmonary complications （PPCs） after thoracoscopic lobectomy. Methods The patients who had undergone elective thoracoscopic lobectomy in Henan Provincial People's Hospital were collected. General information and perioperative indicators were recorded. Diaphragmatic motility was detected by bedside ultrasound preoperatively and on postoperative days 1， 3， and 5. The patients were divided into two groups （PPCs group and non?PPCs group）. Statistically different parameters in univariate analysis were included in multivariate logistic regression analysis to screen independent influencing factors of PPCs. Receiver operating curve （ROC） was drawn， and the performance of diaphragm ultrasound for predicting PPCs was evaluated by the area under the curve （AUC）. Results 949 patients were included in this study. PPCs occurred in 537 patients （57.5%）. Univariate analysis showed that as compared with the non?PPCs group， the proportion of diabetic patients and postoperative VAS score in the PPCs group increased， time to chest tube removal and one?lung ventilation， and postoperative hospital stay were longer. PPCs group had a lower DE value on the operative side and healthy side on postoperative days 1 and 3 and on the operative side alone on day 5 （P < 0.05）. Multivariate logistic regression analysis showed that prolonged one?lung ventilation time， diabetes mellitus， DE values on the operative side and unoperative side on postoperative days 1 and 3， and decreased DE value on the operative side on day 5 were independent risk factors for PPCs. The ROC curve shows that the efficacy of DE for predicting PPCs on the operative and healthy sides on postoperative days 1 and 3 was relatively high （AUC of 0.797， 0.821， 0.933， and 0.929； respectively）. The efficacy of DE for predicting PPCs on the operative side was poor on postoperative day 5 （AUC of 0.703 and 0.512， respectively）. Conclusions A decrease in postoperative DE， prolonged one?lung ventilation time， and diabetes are independent risk factors for PPCs after thoracoscopic lobectomy. Bedside ultrasound evaluation of DE has a higher value in predicting PPCs three days after surgery.