Severity of illness scores: Are they useful in febrile neutropenic adult patients in hematology wards? A prospective multicenter study

Objective: To assess the prognostic value of two severity of illness scores, commonly used for critically ill patients, Simplified Acute Physiology Score (SAPS II) and Organ Dysfunctions and Infection (ODIN), in predicting mortality in febrile neutropenic patients in hematology wards. Design: A 2-month prospective multicenter study. Setting: Thirty-six hematologic and/or stem cell transplant units in France. Patients: All adult patients with a first febrile neutropenic episode (polymorphonuclear cells <500/mm³) were included. Interventions: SAPS II was calculated on day 1 of fever, and ODIN on days 1 and 8. The end point was the mortality rate on day 28. Measurements and Main Results: Twenty-eight (6.6%) of the 421 patients included died before day 28. The mortality rate predicted by SAPS II was 23.8%, indicating a poor calibration. The SAPS II score at day 1 was greater in nonsurvivors than in survivors (44 ± 11 vs. 38 ± 7, p < .0001), as was the number of patients with one or more organ failures at day 1 (14 vs. 2%, p < .0001), and day 8 (42 vs. 3%, p < .0001). The pattern of change in the scores over the first 8 days differed significantly between survivors and nonsurvivors. In multivariate analysis, only ODIN on day 1 and day 8, and spontaneous neutropenia were independent predictors for death. Conclusions: SAPS II and ODIN scores are inaccurate for predicting individual outcome of febrile neutropenic patients in hematology wards. Serial measurements of these scores during the first week of hospitalization could be more accurate than a single measurement. Besides severity scores and organ failures, the type of neutropenia is at least as important in assessing the prognosis.