Gilbert Sebastien, Maghera Sonam, Seely Andrew J, Maziak Donna E, Shamji Farid M, Sundaresan Sudhir R, Villeneuve Patrick J
Division of Thoracic Surgery, University of Ottawa, Ottawa, Ontario, Canada; The Ottawa Hospital Research Institute, Ottawa, Ontario, Canada.
Faculty of Medicine, University of Ottawa, Ottawa, Ontario, Canada.
Ann Thorac Surg. 2016 Nov;102(5):1674-1679. doi: 10.1016/j.athoracsur.2016.05.035. Epub 2016 Jul 22.
Predictive models of prolonged air leak have relied on information not always available preoperatively (eg, extent of resection, pleural adhesions). Our objective was to construct a model to identify patients at increased risk of prolonged air leak using preoperative factors exclusively.
From 2012 to 2014, data on consecutive patients undergoing pulmonary resection were collected prospectively. Prolonged air leak was defined as lasting longer than 7 days and requiring hospitalization. Factors associated with the primary outcome (p < 0.2) were included in a multivariate model. Regression coefficients were used to develop a weighted risk score for prolonged air leak.
Of 225 patients, 8% (18/225) experienced a prolonged air leak. Male gender (p = 0.08), smoking history (p = 0.03), body mass index (BMI) 25 or below (p < 0.01), Medical Research Council (MRC) dyspnea score above 1 (p = 0.06), and diffusion capacity for carbon monoxide below 80% (Dlco) (p = 0.01) were selected for inclusion in the final model. Weighted scores were male gender (1 point), BMI 25 or below (0.5 point), smoker (2 points), Dlco% below 80% (2 points), and MRC dyspnea score above 1 (1 point). The area under the receiver operating characteristic curve was 0.8 (95% confidence interval [CI] = 0.7 to 0.9]. An air leak score above 4 points offered the best combination of sensitivity (83% [95% CI = 58 to 96]) and specificity (65% [95% CI = 58 to 71]).
A subgroup of lung resection patients at higher risk for a prolonged air leak can be effectively identified with the use of widely available, preoperative factors. The proposed scoring system is simple, is clinically relevant to the informed consent, and allows preoperative patient selection for interventions to reduce the risk of prolonged air leak.
持续性漏气的预测模型所依赖的信息在术前并非总是可得（例如，切除范围、胸膜粘连情况）。我们的目标是构建一个仅使用术前因素来识别持续性漏气风险增加患者的模型。
2012年至2014年，前瞻性收集了连续接受肺切除术患者的数据。持续性漏气定义为持续超过7天且需要住院治疗。将与主要结局相关（p<0.2）的因素纳入多变量模型。使用回归系数来制定持续性漏气的加权风险评分。
225例患者中，8%（18/225）发生了持续性漏气。男性（p = 0.08）、吸烟史（p = 0.03）、体重指数（BMI）25或更低（p < 0.01）、医学研究委员会（MRC）呼吸困难评分高于1（p = 0.06）以及一氧化碳弥散量低于80%（Dlco）（p = 0.01）被选入最终模型。加权评分如下：男性（1分）、BMI 25或更低（0.5分）、吸烟者（2分）、Dlco%低于80%（2分）以及MRC呼吸困难评分高于1（1分）。受试者工作特征曲线下面积为0.8（95%置信区间[CI]=0.7至0.9）。漏气评分高于4分提供了最佳的敏感性（83%[95%CI = 58至96]）和特异性组合（65%[95%CI = 58至71]）。
通过使用广泛可得的术前因素，可以有效识别出持续性漏气风险较高的肺切除患者亚组。所提出的评分系统简单，在知情同意方面具有临床相关性，并允许术前对患者进行选择以采取干预措施降低持续性漏气的风险。
