Clinical Investigations Respiration 2005;72:357–364 DOI: 10.1159/000086248 Pleural Fluid Parameters Identifying Complicated Parapneumonic Effusions D. Jiménez Castro a,b G. Dı´az Nuevo c A. Sueiro a,b A. Muriel a E. Pérez-Rodrı´guez a,b R.W. Light d a Respiratory Department, Hospital Ramo ´ n y Cajal and b Department of Medicine, Alcala ´ de Henares University, and c Zarzuela Clinic, Madrid, Spain; d Pulmonary Division, Saint Thomas Hospital and Vanderbilt University, Nashville, Tenn., USA Received: June 1, 2004 Accepted after revision: November 25, 2004 David Jiménez Castro Respiratory Department, Hospital Ramo ´ n y Cajal ES–28034 Madrid (Spain) Tel. +34 91 336 83 14, Fax +34 91 675 33 29 E-Mail djc_69_98@yahoo.com ABC Fax + 41 61 306 12 34 E-Mail karger@karger.ch www.karger.com © 2005 S. Karger AG, Basel 0025–7931/05/0724–0357 Accessible online at: www.karger.com/res Key Words Drainage W Parapneumonic effusions W Analysis W Likelihood ratio W pH Abstract Background: Controversy exists regarding the clinical utility of pleural fluid parameters as prognosticators of complicated parapneumonic effusions that require drainage. Objectives: The purpose of this prospective study is to further assess the utility of these parameters in the management of a larger series of parapneumonic effusions and to determine appropriate binary decision thresholds. Methods: We studied 238 consecutive pa- tients with parapneumonic effusions who underwent diagnostic thoracentesis. Results: We found that pleural fluid pH had the highest diagnostic accuracy (area under the curve, AUC: 0.928; 95% confidence interval, CI: 0.894–0.963) compared with pleural fluid glucose (AUC: 0.835; 95% CI: 0.773–0.897), LDH (AUC: 0.824; 95% CI: 0.761–0.887) or pleural fluid volume (AUC: 0.706; 95% CI: 0.634–0.777). The optimal binary decision threshold for pleural fluid pH identifying complicated effusions requir- ing drainage was 7.15. Binary, multilevel and continuous likelihood ratios (LRs) for pH were calculated to estimate the likelihood of complication of the pleural effusion. Val- ues for the LRs were compared for each of the three strat- egies, and relative clinical and statistical significances were assessed. Binary LRs provided significantly less information than continuous strategies. Conclusion: The pH has the highest diagnostic accuracy for identifying complicated parapneumonic pleural effusions. The bina- ry decision threshold determining the need for chest drainage is 7.15 in our patient series. We recommend continuous LRs to estimate the post-test probability of the complication as they provide the most information compared with binary LRs. Our results do not support the use of pleural fluid LDH as independent predictor of complicated parapneumonic effusions. Copyright © 2005 S. Karger AG, Basel Introduction Despite the advent of potent antibiotics, bacterial pneumonia still results in morbidity and mortality. The annual incidence of bacterial pneumonia is estimated at 1.8–8 cases per 1,000, with approximately 20% requiring hospitalization [1]. As many as 57% of inpatients with bacterial pneumonia have an accompanying pleural effu- For editorial comment see p. 345