Clinical predictors of bacteremia in immunocompetent adult patients hospitalized for community-acquired pneumonia

Background: The clinical usefulness of blood cultures in the management of patients hospitalized with community-acquired pneumonia (CAP) is controversial. Aim: To determine clinical predictors of bacteremia in a cohort of adult patients hospitalized for community-acquired pneumonia. Material and Methods: A prospective cohort of 605 immunocompetent adult patients aged 16 to 101 years (54% male) hospitalized for CAP was studied. The clinical and laboratory variables measured at admission were associated with the risk of bacteremia by univariate and multivariate analysis using logistic regression models. Results: Seventy seven percent of patients had comorbidities, median hospital stay was 9 days, 7.6% died in hospital and 10.7% at 30 days. The yield of the blood cultures was 12.6% (S. pneumoniae in 69 patients, H. influenzae in 3, Gram negative bacteria in three and S. aureus in one). These results modified the initial antimicrobial treatment in one case (0.2%). In a multivariate analysis, clinical and laboratory variables associated with increased risk of bacteremia were low diastolic blood pressure (Odds ratio (OR): 1.85, 95% confidence intervals (CI) 1.02 to 3.36, p < 0.05), leukocytosis >= 15,000/mm(3) (OR: 2.18, 95% CI 1.22 to 3.88, p < 0.009), serum urea nitrogen >= 30 mg/dL (OR: 2.23, 95% CI 1.22 to 4.05, p < 0.009) and serum C-reactive protein >= 30 mg/dL (OR: 2.20, 95% CI 1.22 to 3.97, p < 0.01). Antimicrobial use before hospital admission significantly decreased the blood culture yield (OR: 0.14, 95% CI 0.04 to 0.46, p < 0.002). Conclusions: Blood cultures do not contribute significantly to the initial management of patients hospitalized for community-acquired pneumonia. The main clinical predictors of bacteremia were antibiotic use, hypotension, renal dysfunction and systemic inflammation.