Risk index for predicting complications and prognosis in Thai patients with neutropenia and fever

New strategies in the treatment of febrile neutropenic patients have been proposed during the past decade. It is more and more widely accepted that febrile neutropenic patients are a heterogeneous population and they have varying risks for complications and death. However, most of the data have been collected from patients in Western countries. The purpose of the study was to identify types of infection and etiologic organisms in febrile neutropenic patients at Siriraj Hospital, Thailand, and also to develop a prediction model in order to identify patients who are expected to have a favorable outcome or a low-risk subset.

The medical records of chemotherapy-induced neutropenic patients with fever hospitalized at Siriraj Hospital, Thailand, from January 1999 to December 2000 were analyzed. Data included patient characteristics, epidemiological data and the potential factors at the onset of fever for predicting patient outcome. A scoring system for predicting patients with favorable outcome was developed. The scoring system developed from this study was compared with a previously used scoring system.

Of 220 patients with 267 febrile neutropenic episodes, 71.8 per cent had hematologic malignancies and 28.2 per cent had solid tumors. Bacteremia was found in 61 episodes (22.8%) and gram negative bacilli were the most common causative organism in bacteremia (88.6%). Overall mortality was 17.7 per cent. Multivariate analysis revealed that the factors predicting outcome were burden of illness, control of cancer, duration of neutropenia and dehydration. The scoring system developed from this set of data revealed that a score > or = 16 identified patients with a favorable outcome with a specificity of 90.2 per cent, sensitivity of 76.6 per cent and positive predictive value of 85.4 per cent.

The causative organisms of bacterial infections in febrile neutropenic patients in Thailand are still gram negative bacteria. The locally developed risk index has a fair accuracy to identify patients with favorable outcome and may be used to identify patients suitable for less aggressive treatment strategies.