[Agreement between type of lung resection planned and resection subsequently performed on lung cancer patients]

To assess agreement between planned lung resections and the type subsequently performed on a series of patients, to assess whether tumor location (central or peripheral) affected the degree of discrepancy, and, in the case of unscheduled pneumonectomies, to examine why the planned resection had to be extended.

Prospective, observational clinical study of 199 patients scheduled for lung cancer surgery. Tumors were preoperatively classified as central or peripheral, and the type of operation planned--lobectomy (or bilobectomy) or pneumonectomy--was compared with the operation finally performed. Rates of agreement and Wilks' lambda statistic were calculated.

Twenty unscheduled pneumonectomies were performed. Agreement between planned and performed operations was found in 86.9% of cases (76.9% in central tumors and 95.4% in peripheral tumors). Wilks' lambda statistic was 0.38 (0.42 for central tumors and 0.17 for peripheral tumors). Seven unscheduled pneumonectomies were performed due to hilar node involvement.

The resections performed differed from the resections initially planned in 13% of the bronchial carcinoma operations, in most cases because the planned lobectomy had to be converted to pneumonectomy, a situation which occurred more often with central tumors and was more often due to direct invasion of anatomic structures rather than hilar spread.