From the Journals

Preoperative variables can predict prolonged air leak


Key clinical point: Preoperative variables can be evaluated to determine patient risk for prolonged air leak (PAL) in lung resection for cancer.

Major finding: A nomogram demonstrated 76% discriminatory accuracy in predicting PAL after lung resection.

Data source: Analysis of 2,522 pulmonary resections performed at eight hospitals within the University of Pittsburgh health system from January 2009 to June 2014.

Disclosures: The researchers had no conflicts of interest to disclose.

View on the News

Nomogram may predict PAL

The authors of this study “have performed a rigorous set of analyses to create this model,” Chi-Fu Jeffrey Yang, MD, of Duke University, Durham, N.C., noted in his invited commentary (J Thorac Cardiovasc Surg. 2017 March;53[3]:700-1). “The strengths of this study include its sound statistical analysis and study design,” Dr. Yang wrote. He gave the authors credit for using bootstrapping to internally validate the model.

However, Dr. Yang said that the database used by the researchers did not account for “numerous important variables,” including presence of pleural adhesions and emphysema status. The analysis also grouped lobectomy and segmentectomy together, and did not consider intraoperative variables such as sealant use, or postoperative management.

Dr. Chi-Fu Jeffrey Yang
Dr. Chi-Fu Jeffrey Yang

While Dr. Yang commended the study authors for developing a “reliable nomogram,” getting it implemented in the clinic is another hurdle. “It is commonly cited that it takes approximately 17 years for research evidence to translate into daily practice,” he said. To shorten that time line, he suggested the authors take a cue from various tech groups: Develop an app that surgeons can use.

Dr. Yang had no relevant financial relationships to disclose.



Prolonged air leak is a well-known complication after lung cancer surgery that can worsen patient outcomes and drive up costs, and while international authors have developed tools to calculate the risk of PAL, their use has been limited in the United States for various reasons. Researchers at the University of Pittsburgh have reported on a predictive model that uses easy-to-obtain patient factors, such as forced expiratory volume and smoking history, to help surgeons identify patients at greatest risk for complications and implement preventative measures.

Adam Attaar and his coauthors reported that their nomogram had an accuracy rate of 76%, with a 95% confidence interval, for predicting PAL after surgery (J Thorac Cardiovasc Surg. 2017 March;153[3]:690-9). “Using readily available candidate variables, our nomogram predicts increasing risk of prolonged air leak with good discriminatory ability,” noted Mr. Attaar, a student at University of Pittsburgh, and his coauthors.

Previously published reports put the incidence of PAL complications at 6%-18%, they noted. In the University of Pittsburgh series of 2,317 patients who had pulmonary resection for lung cancer or nodules from January 2009 to June 2014, the incidence was 8.6%.

In this series, patients with PAL were more likely to be older, men, and smokers, and to have a lower body mass index, peripheral vascular disease, chronic obstructive pulmonary disease, a history of steroid use, a high Zubrod score and lower forced expiratory volume.“They were less likely to have diabetes or to be hospitalized before surgery,” the researchers said. Surgical factors that characterized patients with PAL were resection for primary lung cancer rather than benign or metastatic tumors; lobectomy/segmentectomy or bilobectomy rather than wedge resection; a right-sided resection; thoracotomy; and a surgeon with higher annual caseloads.

Not all those factors made it into the nomogram, however. The nomogram scores each of these 10 variables to calculate the risk of PAL, in order of their weighting: lower forced expiratory volume, procedure type, BMI, right-sided thoracotomy, preoperative hospitalization, annual surgeon caseload, wedge resection by thoracotomy, reoperation, smoking history, and Zubrod score. A second nomogram drops out surgeon volume to make it more generalizable to other institutions.

In explaining higher surgeon volume as a risk factor for PAL, the researchers said that high-volume surgeons may be operating on patients with variables not accounted for in the Society of Thoracic Surgeons General Thoracic Surgery Database. “These unmeasured variables … could reveal modifiable technical factors to reduce the incidence of PAL and require further study,” the researchers said.

Fast-track discharge has gained acceptance in recent years as a way to spare patients a prolonged hospital stay and cut costs, but in this series the median hospital stay for patients with PAL was 10 days vs. 4 days for non-PAL patients (P less than 0.001).

“An accurate and generalizable PAL risk stratification tool could facilitate surgical decision making and patient-specific care” and aid in the design of trials to evaluate air-leak reduction methods such as sealants, buttressed staple lines, and pneumoperitoneum the researchers wrote.

Going forward, further development of the model would involve a multicenter study and inclusion of risk factors not accounted for in the thoracic surgery database, they noted.

The researchers had no relevant financial relationships to disclose.

Next Article:

   Comments ()