A new prognostic model for chemotherapy-induced febrile neutropenia

The objective of this study was to develop and validate a new prognostic model for febrile neutropenia (FN). This study comprised 1001 episodes of FN: 718 for the derivation set and 283 for the validation set. Multivariate logistic regression analysis was performed with unfavorable outcome as the primary endpoint and bacteremia as the secondary endpoint. In the derivation set, risk factors for adverse outcomes comprised age a parts per thousand yen60 years (2 points), procalcitonin a parts per thousand yen0.5 ng/mL (5 points), ECOG performance score a parts per thousand yen2 (2 points), oral mucositis grade a parts per thousand yen3 (3 points), systolic blood pressure < 90 mmHg (3 points), and respiratory rate a parts per thousand yen24 breaths/min (3 points). The model stratified patients into three severity classes, with adverse event rates of 6.0 % in class I (score a parts per thousand currency sign2), 27.3 % in class II (score 3-8), and 67.9 % in class III (score a parts per thousand yen9). Bacteremia was present in 1.1, 11.5, and 29.8 % of patients in class I, II, and III, respectively. The outcomes of the validation set were similar in each risk class. When the derivation and validation sets were integrated, unfavorable outcomes occurred in 5.9 % of the low-risk group classified by the new prognostic model and in 12.2 % classified by the Multinational Association for Supportive Care in Cancer (MASCC) risk index. With the new prognostic model, we can classify patients with FN into three classes of increasing adverse outcomes and bacteremia. Early discharge would be possible for class I patients, short-term observation could safely manage class II patients, and inpatient admission is warranted for class III patients.