Enterococcal bloodstream infection. Design and validation of a mortality prediction rule

Summary Background To develop a prediction rule to describe the risk of death as a result of enterococcal bloodstream infection. Methods A prediction rule was developed by analysing data collected from 122 patients diagnosed with enterococcal BSI admitted to the Clinica Universidad de Navarra (Pamplona, Spain); and validated by confirming its accuracy with the data of an external population (Hospital del Mar, Barcelona). Results According to this model, independent significant predictors for the risk of death were being diabetic, have received appropriate treatment, severe prognosis of the underlying diseases, have renal failure, received solid organ transplant, malignancy, source of the bloodstream infection and be immunosuppressed. The prediction rule showed a very good calibration (Hosmer–Lemeshow statistic, P = 0.93) and discrimination for both training and testing sets (area under ROC curve = 0.84 and 0.83 respectively). Conclusions The predictive rule was able to predict risk of death as a result of enterococcal bloodstream infection as well as to identify patients, who being below the threshold value, will have a low risk of death with a negative predictive value of 96%.