Although the protocol called for a simultaneous sample of serum to be drawn at the time of thoracentesis, this was not always done by the house staff. Serum samples were accepted within 24 hours of the thoracentesis. In the five patients with a discrepancy between Lights criteria and the albumin gradient, a true simultaneous sample was drawn.
Using the albumin gradient, two patients who had exudates were misclassified as transudates (1.5 and 1.3 g/dl). Both patients had positive cytologic findings in the pleural fluid. One had adenocarcinoma of unknown primary with a large right-sided pleural effusion, but also had a history of an anterior myocardial infarction with decreased ejection fraction by cardiac catheterization. The other patient had smallcell carcinoma with obstruction and collapse of the left lower lobe.
The sensitivity and specificity for identifying exudates with Lights criteria are 100 percent and 72 percent, respectively, from the data in our 59 patients. The corresponding sensitivity and specificity for identifying exudates with the albumin gradient are 95 percent and 100 percent. The difference between the sensitivities is clearly not significant, with a p value of 0.5; however, the difference in specificity approaches statistical significance with a p value of 0.06.