RESUMEN
OBJECTIVES: We evaluated the respective influence of the causative pathogen and infection site on hospital mortality from severe sepsis related to community-, hospital-, and intensive care unit-acquired infections. DESIGN: We used a prospective observational cohort 10-yr database. We built a subdistribution hazards model with corrections for competing risks and adjustment for potential confounders including early appropriate antimicrobial therapy. SETTING: Twelve intensive care units. PATIENTS: We included 4,006 first episodes of acquisition-site-specific severe sepsis in 3,588 patients. INTEVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We included 1562 community-acquired, 1432 hospital-acquired, and 1012 intensive care unit-acquired episodes of severe sepsis. After adjustment, we found no independent associations of the causative organism, multidrug resistance of the causative organism, infection site, or presence of bacteremia with mortality. Early appropriate antimicrobial therapy was consistently associated with better survival in the community-acquired (0.64 [0.51-0.8], p = .0001), hospital-acquired (0.72 [0.58-0.88], p = .0011), and intensive care unit-acquired (0.79 [0.64-0.97], p = .0272) groups. CONCLUSION: The infectious process may not exert as strong a prognostic effect when severity, organ dysfunction and, above all, appropriateness of early antimicrobials are taken into account. Our findings emphasize the importance of developing valid recommendations for early antimicrobial therapy.