Objectives: To determine the relative usefulness of different criteria for the separation of pleural transudates from exudates.
Design: Prospective evaluation of patients referred for thoracentesis.
Setting: Community teaching hospital.
Patients: Three hundred fifty-one consecutive patients with pleural effusions referred for thoracentesis. Fifty-four of these patients were excluded from the analysis.
Measurements: We recorded clinical characteristics and final diagnosis and measured pleural fluid and serum levels of total protein, lactate dehydrogenase, and cholesterol. All patients included were followed up until final diagnosis. MEAN RESULTS: Forty-four (15 percent) pleural effusions were transudates and 253 (85 percent) were exudates. The criteria of Light et al, with a sensitivity of 98 percent and a specificity of 77 percent for exudates, showed the best accuracy (95.2 percent). Moreover, when the cutoff used for the criteria of Light et al was modified according to our own laboratory results, specificity rose to 93 percent with almost a similar accuracy (94 percent). Protein pleural fluid/serum ratio > 0.5 and pleural fluid cholesterol > 60 mg/dl showed equal specificity (91 percent), but the former had better sensitivity for exudates (88 percent vs 81 percent).
Conclusions: When the proportion of exudates included is 85 percent or more, as in the present series, the criteria of Light et al remain the method that offers the highest accuracy for segregating transudates from exudates.