Survival following Staphylococcus aureus bloodstream infection: A prospective multinational cohort study assessing the impact of place of care.
Nambiar K., Seifert H., Rieg S., Kern WV., Scarborough M., Gordon NC., Kim HB., Song K-H., Tilley R., Gott H., Liao CH., Edgeworth J., Nsutebu E., López-Cortés LE., Morata L., Walker AS., Thwaites G., Llewelyn MJ., Kaasch AJ., International Staphylococcus aureus collaboration (ISAC) study group (with linked authorship to members in the Acknowledgements) and the ESCMID Study Group for Bloodstream Infections and Sepsis (ESGBIS) None.
BACKGROUND: Staphylococcus aureus bloodstream infection (SAB) is a common, life-threatening infection with a high mortality. Survival can be improved by implementing quality of care bundles in hospitals. We previously observed marked differences in mortality between hospitals and now assessed whether mortality could serve as a valid and easy to implement quality of care outcome measure. METHODS: We conducted a prospective observational study between January 2013 and April 2015 on consecutive, adult patients with SAB from 11 tertiary care centers in Germany, South Korea, Spain, Taiwan, and the United Kingdom. Factors associated with mortality at 90 days were analyzed by Cox proportional hazards regression and flexible parametric models. RESULTS: 1851 patients with a median age of 66 years (64% male) were analyzed. Crude 90-day mortality differed significantly between hospitals (range 23-39%). Significant variation between centers was observed for methicillin-resistant S. aureus, community-acquisition, infective foci, as well as measures of comorbidities, and severity of disease. In multivariable analysis, factors independently associated with mortality at 90 days were age, nosocomial acquisition, unknown infective focus, pneumonia, Charlson comorbidity index, SOFA score, and study center. The risk of death varied over time differently for each infective focus. Crude mortality differed markedly from adjusted mortality. DISCUSSION: We observed significant differences in adjusted mortality between hospitals, suggesting differences in quality of care. However, mortality is strongly influenced by patient mix and thus, crude mortality is not a suitable quality indicator.