Factors influencing pleural drainage in parapneumonic effusions

The identification of parapneumonic effusions (PPE) requiring pleural drainage is challenging. We aimed to determine the diagnostic accuracy of radiological and pleural fluid findings in discriminating between PPE that need drainage (complicated PPE (CPPE)) and those that could be resolved with antibiotics only (uncomplicated PPE (UPPE)).

A retrospective review of 641 consecutive PPE, of which 393 were categorized as CPPE and 248 as UPPE. Demographics, radiological (size and laterality on a chest radiograph) and pleural fluid parameters (pus, bacterial cultures, biochemistries) were compared among groups. Logistic regression was performed to determine variables useful for predicting chest drainage, and receiver-operating characteristic curves assisted in the selection of the best cutoff values.

According to the likelihood ratios (LR), findings increasing the probability of chest tube usage the most were: effusions occupying ≥1/2 of the hemithorax (LR 13.5), pleural fluid pH ≤7.15 (LR 6.2), pleural fluid glucose ≤40mg/dL (LR 5.6), pus (LR 4.8), positive pleural fluid cultures (LR 3.6), and pleural fluid lactate dehydrogenase >2000U/L (LR 3.4). In the logistic regression analysis only the first two were selected as significant predictors of CPPE. In non-purulent effusions, the effusion's size and pleural fluid pH retained their discriminatory properties, in addition to a pleural fluid C-reactive protein (CRP) level >100mg/L.

Large radiological effusions and a pleural fluid pH ≤7.15 were the best predictors for chest drainage in patients with PPE. In the subgroup of patients with non-purulent effusions, pleural fluid CRP also contributed to CPPE identification.