Predicting Complicated Parapneumonic Effusion in Community Acquired Pneumonia: Hospital Based Case-Control Study

To identify predictors of complicated parapneumonic effusion (CPE)/empyema in patients of community acquired pneumonia (CAP) by using clinical and simple laboratory variables like hemoglobin (Hb), serum C-reactive protein (CRP), serum albumin (SA) levels and total leukocyte counts (TLC).

This prospective case-control study was conducted after institutional ethical approval. Subjects between ages of 2-59 mo with World Health Organization (WHO) defined CAP with written, informed parental consent were included. Cases had CAP with CPE/empyema diagnosed by pleurocentesis. Controls had severe CAP without significant pleural collection on chest X-ray (CXR). Patients with congenital and chronic diseases/infections and possible immune deficiency were excluded. Variables with univariate association with case-control status were considered as potential predictors. Final prediction model was developed by Forward Stepwise Logistic Regression (FSLR). Adjusted odd's ratios (Adj OR) were smoothened into nearest whole numbers to develop KGMU-CPE score.

From 2016 to 17, 30 cases (66.6% males, age 38.7 + 14.9 mo) and 118 controls (78% males, age 17.8 + 16.9 mo) were included. In FSLR, predictors of CPE/empyema were ibuprofen intake (adj OR 6.8; 95%CI: 1.07-43.6), infective focus elsewhere (adj OR 28.2; 95%CI: 1.4-563.1), hypoalbuminemia <3.1 g/dL (adj OR 6.9; 95%CI: 1.22-39.3), serum CRP >20 mg/dL (adj OR 59; 95%CI: 1.86-1874.7), Hb <10 g/dL (adj OR 21.1; 95%CI: 2.8-158.1) and TLC >10,000 (adj OR 37; 95%CI: 5.7-239.8) and these six variables formed KGMU-CPE Score with a minimum score of 0 and maximum of 25. KGMU-CPE score area under the ROC curve was 0.97 and cut- off 15.55 had sensitivity of 80% and specificity of 94% for predicting CPE/empyema.

Using simple clinical and laboratory parameters it is possible to predict CAP with CPE/empyema. Use of ibuprofen is to be avoided in CAP as it associated with CPE. KGMU-CPE score had good diagnostic accuracy and needs external validation.