ABSTRACT
Microbiological sampling is an indispensable targeted antibiotic therapy for critically ill patients. Invasive respiratory sampling by bronchoalveolar lavage (BAL) can be performed to obtain samples from the lower respiratory tract. It is debated as to whether blood markers of infection can predict the outcome of BAL in a medical intensive care unit (ICU). Retrospectively, all ICU patients undergoing BAL from 2009-2018 were included. A total of 468 BAL samples from 276 patients (average age 60 years, SAPS2 47, ICU-mortality 41.7%) were analyzed. At the time of BAL, 94.4% patients were mechanically ventilated, 92.9% had suspected pneumonia, 96.2% were on antibiotic therapy and 36.3% were immunocompromised. Relevant bacteria were cultured in 114/468 (24.4%) cases of BAL. Patients with relevant bacteria in the culture had a higher ICU mortality rate (45.6 vs. 40.4%, p = 0.33) and were significantly less likely to be on a steroid (36 vs. 52%, p < 0.01) or antimycotic (14.9 vs. 34.2%, p < 0.01), while procalcitonin (PCT), C-reactive protein (CRP), and white blood cell (WBC) counts were similar. The area under the receiver operating curve (AUC) values for positive culture and PCT, CRP and WBC counts were low (0.53, 0.54 and 0.51, respectively). In immunocompromised patients, AUC values were higher (0.65, 0.57 and 0.61, respectively). Therefore, microbiological cultures by BAL revealed relevant bacteria in 24.4% of samples. Our data, therefore, might suggest that indication for BAL should not be based on blood markers of infection.
