Differentiating tuberculous from malignant pleural effusions: a scoring model.

Abstract

BACKGROUND: Patients with tuberculous or malignant pleural effusions frequently have similar clinical manifestations and pleural fluid profile. The aim of our study was to derive a simple clinical score for differential diagnosis of these two clinical entities. MATERIAL/METHODS: Our retrospective study involved 106 patients with tuberculous pleurisy and 286 with malignant effusions, seen during a 9-year period. Clinical and laboratory variables with (model 1) and without (model 2) the addition of pleural adenosine deaminase entered into a multivariate analysis to calculate a scoring system (range 0 to 10) for the detection of tuberculous effusions. RESULTS: In model 1, four variables predicted a tuberculous etiology: adenosine deaminase > or = 40 U/L (5 points), age <35 years (2), temperature > or 37.8 degrees C (2), and pleural fluid red blood cell count < 5 x 10(9)/L (1). In addition to the last three items, model 2 identified other predictive parameters: no history of malignancy (3), pleural protein > or = 50 g/L (1), and pleural fluid to serum lactate dehydrogenase ratio > or 2.2 (1). Summated scores of > or 5 in model 1 and > or 6 in model 2 yielded measures of sensitivity (95% and 97%), and specificity (94% and 91%) for discriminating tuberculous from malignant effusions, respectively. The area under the ROC curve for models 1 and 2 was 0.987 and 0.982, respectively. CONCLUSIONS: The combination of clinical data and pleural fluid chemistry profile into a score-based model can facilitate differential diagnosis between tuberculous and malignant effusions.