ABSTRACT
BACKGROUND: Pneumonia is responsible for 50% of antibiotics prescribed in ICUs. Treatment failure, ie, absence of improvement or clinical deterioration under antibiotic therapy, presents a dilemma to physicians. BAL is an invasive method validated for etiologic diagnosis in pneumonia. STUDY OBJECTIVE: To evaluate in ICU patients the impact of BAL in the etiologic diagnosis, treatment, and outcome of pneumonia with treatment failure. DESIGN: Prospective clinical study. SETTING: Nonsurgical, medical ICU of a university hospital in Brazil. PATIENTS AND PARTICIPANTS: Sixty-two episodes of pneumonia treated for at least 72 h without clinical improvement in 53 patients hospitalized for diverse clinical emergencies. Mean duration of hospitalization was 14.2 days. Mean duration of previous antibiotic therapy was 11.4 days. INTERVENTIONS: Bronchoscopy and BAL were performed in each episode. BAL fluid was cultivated for aerobic and anaerobic bacteria; the cutoff considered positive was 10(4) cfu/mL; 10(3) cfu/mL was also analyzed if under treatment. Pneumocystis carinii, fungi, Legionella spp, and Mycobacterium spp were also researched. MEASUREMENTS AND RESULTS: Fifty-eight of 62 BAL were performed under antibiotics. The results showed positivity in 45 of 62 (72.6%); 42 of the 45 positive episodes (93.3%) had > 10(4) cfu/mL. The three cases with between 10(3) and 10(4) cfu/mL were considered positive and were treated according to BAL cultures. The main agents were Acinetobacter baumannii (37.1%), Pseudomonas aeruginosa (17.7%), and methicillin-resistant Staphylococcus aureus (MRSA; 16.1%); 46.7% of the episodes (21 of 45) were polymicrobial. BAL results directed a change of therapy in 34 episodes (54.8%). Overall mortality was 43.5%. There was no difference in mortality among positives, negatives, and patients who changed therapy guided by BAL culture. CONCLUSIONS: (1) BAL fluid examination was positive in 45 of 62 episodes (72.6%), with 58 of 62 BAL performed under antibiotics. This suggests that BAL may be a sensitive diagnostic method for treatment failures of clinically diagnosed pneumonias, even if performed under antibiotics; (2) the main pathogens in our study were A baumannii, P aeruginosa, and MRSA, and approximately 45% of infections were polymicrobial; (3) BAL culture results directed a change of therapy in 75.6% of positive episodes (34 of 45) and in 54.8% of all episodes of treatment failure (34 of 62); and (4) there was no difference in mortality among positives, negatives, and patients who changed therapy guided by BAL culture.
