Comparison of prognostic factors between bacteraemic and non-bacteraemic critically ill immunocompetent patients in community-acquired severe pneumococcal pneumonia: a STREPTOGENE sub-study.

BACKGROUND: The presence of bacteraemia in pneumococcal pneumonia in critically ill patients does not appear to be a strong independent prognostic factor in the existing literature. However, there may be a specific pattern of factors associated with mortality for ICU patients with bacteraemic pneumococcal community-acquired pneumonia (CAP). We aimed to compare the factors associated with mortality, according to the presence of bacteraemia or not on admission, for patients hospitalised in intensive care for severe pneumococcal CAP. METHODS: This was a post hoc analysis of data from the prospective, observational, multicentre STREPTOGENE study in immunocompetent Caucasian adults admitted to intensive care in France between 2008 and 2012 for pneumococcal CAP. Patients were divided into two groups based on initial blood culture (positive vs. negative) for Streptococcus pneumoniae. The primary outcome was hospital mortality, which was compared between the two groups using odds ratios according to predefined variables to search for a prognostic interaction present in bacterial patients but not non-bacteraemic patients. Potential differences in the distribution of serotypes between the two groups were assessed. The prognostic consequences of the presence or not of initial bi-antibiotic therapy were assessed, specifically in bacteraemic patients. RESULTS: Among 614 included patients, 274 had a blood culture positive for S. pneumoniae at admission and 340 did not. The baseline difference between the groups was more frequent leukopaenia (26% vs. 14%, p = 0.0002) and less frequent pre-hospital antibiotic therapy (10% vs. 16.3%, p = 0.024) for the bacteraemic patients. Hospital mortality was not significantly different between the two groups (p = 0.11). We did not observe any prognostic factors specific to the bacteraemic patient population, as the statistical comparison of the odds ratios, as an indication of the association between the predefined prognostic parameters and mortality, showed them to be similar for the two groups. Bacteraemic patients more often had invasive serotypes but less often serotypes associated with high case fatality rates (p = 0.003). The antibiotic regimens were similar for the two groups. There was no difference in mortality for patients in either group given a beta-lactam alone vs. a beta-lactam combined with a macrolide or fluoroquinolone. CONCLUSION: Bacteraemia had no influence on the mortality of immunocompetent Caucasian adults admitted to intensive care for severe pneumococcal CAP, regardless of the profile of the associated prognostic factors.