Analysis of simple pulmonary function screening tests in preoperative assessment before lung resection for...

Simple pulmonary function screening tests were performed prospectively in a group of patients with carcinoma of the bronchus being evaluated for lung resection. Of the 22 who subsequently had a resection performed, 7 suffered post-operative ventilatory difficulties unrelated to technical surgical complications. By analysis of combinations of variables, it was possible to separate significantly the complication group from the non-complication group. This was not possible using one variable alone. The method of empiric determination of values which produce maximum separation is described and contrasted with the more statistically acceptable technique of discriminant analysis. With the latter it is possible to incorporate information from any number of variables and to achieve greater definition of the two groups. It is proposed that such an analysis might provide the means for more valid scientific assessment of simple pre-operative screening tests than is usually employed.