Infection Probability Score, APACHE II and KARNOFSKY scoring systems as predictors of bloodstream infection onset in hematology-oncology patients

Background Bloodstream Infections (BSIs) in neutropenic patients often cause considerable morbidity and mortality. Therefore, the surveillance and early identification of patients at high risk for developing BSIs might be useful for the development of preventive measures. The aim of the current study was to assess the predictive power of three scoring systems: Infection Probability Score (IPS), APACHE II and KARNOFSKY score for the onset of Bloodstream Infections in hematology-oncology patients. Methods A total of 102 patients who were hospitalized for more than 48 hours in a hematology-oncology department in Athens, Greece between April 1st and October 31st 2007 were included in the study. Data were collected by using an anonymous standardized recording form. Source materials included medical records, temperature charts, information from nursing and medical staff, and results on microbiological testing. Patients were followed daily until hospital discharge or death. Results Among the 102 patients, Bloodstream Infections occurred in 17 (16.6%) patients. The incidence density of Bloodstream Infections was 7.74 per 1,000 patient-days or 21.99 per 1,000 patient-days at risk. The patients who developed a Bloodstream Infection were mainly females (p = 0.004), with twofold time mean length of hospital stay (p < 0.001), with fourfold time mean length of neutropenia (p < 0.001), with neutropenia < 500 (p < 0.001), suffered mainly from acute myeloid leukemia (p < 0.001), had been exposed to antibiotics (p = 0.045) and chemotherapy (p = 0.023), had a surgery (p = 0.048) and a Hickman catheter (p = 0.025) as compared to the patients without Bloodstream Infection. The best cut-off value of IPS for the prediction of a Bloodstream Infection was 10 with a sensitivity of 75% and specificity of 70.9% Conclusion Between the three different prognostic scoring systems, Infection Probability Score had the best sensitivity in predicting Bloodstream Infections.