OPTIMIZING SELECTION OF PATIENTS FOR MAJOR LUNG RESECTION

Objectives: It is not known whether a normal diffusing capacity for carbon monoxide permits safe lung resection in patients with marginal spirometric values, or whether normal spirometric values negate the adverse effects of a low diffusing capacity. The purposes of this study were (1) to determine the best predictors of morbidity and mortality and (2) to assess whether interactions exist between diffusing capacity and spirometry that help estimate outcome after major lung resection. Design: A retrospective analysis of 376 patients who underwent lung resection was performed. Three hundred three had lung cancer and 73 had other disease. Two hundred eighty-four underwent lobectomyl bilobectomy and 92 had pneumonectomy, We assessed the relationship of 23 preoperative variables to 18 postoperative events classified into categories as pulmonary or cardiac complications, overall morbidity, and operative mortality. Results: The best single predictor of complications was the percent predicted postoperative diffusing capacity. The incidences of pulmonary and cardiac complications, morbidity, and mortality were inversely related to predicted postoperative diffusing capacity percent (p < 0.004 for each). Multivariate logistic regression analyses identified only predicted postoperative diffusing capacity percent and age as significant independent predictors of pulmonary complications, morbidity, and death, and these with prior myocardial infarction predicted cardiac complications. There were no interactions between percent predicted postoperative forced expiratory volume in 1 second and predicted postoperative diffusing capacity percent in estimating risks of complications. Conclusion: Predicted postoperative diffusing capacity percent is the strongest single predictor of risk of complications and mortality after lung resection. There is little interrelationship of predicted postoperative diffusing capacity percent and predicted postoperative forced expiratory volume in 1 second, indicating that these values should be assessed independently in estimating operative risk.