Infection Probability Score: A predictor of Clostridium difficile-associated disease onset in patients with haematological malignancy

Purpose: to assess the predictive power of three systems: Infection Probability Score, APACHE II and KARNOFSKY score for the onset of Clostridium difficile-associated disease (CDAD) in hematology-oncology patients. Methods and sample: A retrospective pilot surveillance survey was conducted in the hematology unit of a general hospital in Greece. Data were collected by using an anonymous standardised case-record form. The sample consisted of 102 hospitalized patients. Results: The majority of the patients (33.3%) suffered from acute myeloid leukemia. The cumulative incidence of CDAD was 10.8% and the incidence rate of C difficile associated diarrhea was 5 per 1000 patient-days (14.2 per 1000 patient-days at risk). Patients with CDAD had twofold higher time of mean length of hospital stay compared with patients without CDAD (38.82 +/- 23.88 vs 19.45 +/- 14.56 days). Additionally patients with CDAD had received a greater number of different antibiotics compared to those without CDAD (5.18 +/- 1.99 vs 2.54 +/- 2.13), suffered from diabetes, from non Hodgkin's lymphoma, had a statistically significant higher duration of neutropenia >= 3 days and had received antifungal treatment. The best cutoff value of IPS for the prediction of CDAD was 13 with a sensitivity of 45.5% and a specificity of 82.4%. Conclusions: IPS is an early diagnostic test for CDAD detection. (C) 2010 Elsevier Ltd. All rights reserved.