Abstract
BACKGROUND: The independent influence of blood culture testing and bloodstream infection (BSI) on hospital mortality is unclear. METHODS: We included all adults treated in non-psychiatric services at our hospital between 2004 and 2011. We identified all blood cultures and their results to determine the independent association of blood culture testing and BSI on death in hospital using proportional hazards modeling that adjusted for important covariates. RESULTS: Of 297 070 hospitalizations, 48 423 had negative blood cultures and 5274 had BSI. 12 529 (4.2%) died in hospital. Compared to those without blood cultures, culture-negative patients and those with BSI were sicker. Culture-negative patients had a significantly increased risk of death in hospital (adjusted hazard ratio [HR] ranging between 3.1 and 4.4 depending on admission urgency, extent of comorbidities, and whether the blood culture was taken in the intensive care unit). Patients with BSI had a significantly increased risk of death (adj-HR ranging between 3.8 and 24.3] that was significantly higher when BSI was: diagnosed within the first hospital day; polymicrobial; in patients who were exposed to immunosuppressants or were neutropenic; or due to Clostridial and Candidal organisms. Death risk in culture negative and bloodstream infection patients decreased significantly with time. CONCLUSIONS: Risk of death in hospital is independently increased both in patients with negative blood cultures and further in those with bloodstream infection. Death risk associated with bloodstream infections varied by the patient's immune status and the causative microorganism.