Study objectives: To determine if the diagnosis of pleural malignancy is dependent on the volume of pleural fluid sampled.
Design and setting: Single-center retrospective chart review.
Patients: Two hundred eighty-two patients who underwent diagnostic thoracentesis between October 1, 1998, and June 30, 1999.
Interventions: Charts were analyzed for volume of fluid, pathologic and clinical diagnoses, and demographics. Patients were classified into quartiles based on the volume of pleural fluid collected. Sensitivity and negative predictive value (NPV) were calculated for each quartile for diagnosis of pleural malignancy by cytology of pleural fluid. Further analyses were done regarding the effect of sex, race, age, smoking history, and personal history of malignancy on diagnosis.
Results: In total, 374 samples from 282 patients were identified (140 men and 142 women). Pleural malignancy within 6 months of initial thoracentesis was diagnosed in 99 patients (35.1%). No differences were detected for sensitivity and NPV for diagnosis of pleural malignancy between any two quartiles (p > 0.05). Samples collected from women had a higher sensitivity for predicting pleural malignancy (p = 0.0011), and those collected from nonsmokers had a slightly higher but not statistically significant sensitivity for predicting pleural malignancy (p = 0.057). Samples collected from subjects with no history of malignancy had a significantly higher NPV than samples collected from subjects with a history of malignancy (p < 0.001). After adjusting for these demographic and medical history factors, the associations of the pleural fluid volume quartiles with sensitivity and NPV did not change.
Conclusion: The sensitivity for diagnosis of pleural malignancy is not dependent on the volume of pleural fluid extracted during thoracentesis.