Ventilator‐associated pneumonia in a paediatric intensive care unit in a developing country with high HIV prevalence

Abstract

To obtain preliminary prevalence, aetiological and outcome data on South African paediatric patients with ventilator-associated pneumonia (VAP). Non-bronchoscopic bronchoalveolar lavage (BAL) specimens taken between January 2004 and September 2005 were prospectively recorded and related clinical data were retrospectively reviewed. VAP was defined as a new isolate on BAL and a modified Clinical Pulmonary Infection Score > or =5. A total of 230 patients aged 3.9 (2.2-9.1) months (median interquartile range (IQR) ) underwent 309 BALs during 244 paediatric intensive care unit (PICU) admissions. Most patients (84%) were admitted with acute infectious diseases, with a 70% incidence of comorbidity. Thirty-three patients (14.3%) were HIV-exposed but uninfected and 58 (25.2%) were HIV-infected. Of 172 BALs taken > or =48 h after intubation, 63 specimens from 55 patients fulfilled VAP criteria. Acinetobacter baumannii was the most common VAP pathogen, followed by Klebsiella pneumoniae, viruses, yeasts and Staphylococcus aureus. Patients who developed VAP had a higher proportion of comorbid conditions (76% vs. 55%, P= 0.01) and reintubations (39% vs. 12%, P < 0.0001) when compared with non-VAP patients. Median (IQR) length of PICU stay was 12.5 (5-21) days versus 8 (5-14) days (P= 0.03); and the risk adjusted PICU mortality was 1.38 versus 0.79 (P= 0.002) in VAP versus non-VAP patients, respectively. VAP is associated with significant morbidity and mortality and may relate to the high incidence of comorbid conditions in this population. Primary VAP pathogens differ from developed countries.