Objectives: To compare the accuracy of Light's criteria for categorizing a pleural effusion as an exudate with several alternative criteria.
Design: Prospective evaluation of patients who underwent a diagnostic thoracocentesis.
Setting: Community teaching hospital in Lleida, Spain.
Patients and methods: Medical records and pleural fluid characteristics of 241 consecutive patients with pleural effusion admitted over a 29-month period were reviewed. Forty eight of these patients were excluded for different reasons. Light's criteria and a different cutoff level for the pleural fluid cholesterol level were applied and their accuracies were calculated.
Results: Of the 193 patients included, 38 (20%) had transudates and 155 (80%) exudates. The accuracy of Light's criteria for identifying exudates was 92% [confidence intervals (CI), 88-96%], with a sensitivity of 97% (CI, 94-100%) and specificity of 71% (CI, 57-85%). A cutoff level of 50 mg dl-1 was selected for pleural cholesterol, which yielded a sensitivity and specificity of 84% (CI, 79-90%), with an accuracy of 84% (CI, 72-96%). Overall, pleural cholesterol misclassified more exudates as transudates than Light's criteria (15 vs. 3.2%, P < 0.001). The combination of pleural cholesterol with lactate dehydroegnase (LDH) or pleural fluid/serum protein ratio revealed a comparable accuracy to that achieved with Light's criteria.
Conclusions: Light's criteria are just as useful as the association of pleural cholesterol and LDH to detect exudates. In the present study, no parameter, including pleural cholesterol, was superior to Light's criteria.