Personalized Cutoffs for the Diagnosis of Neutropenic Fever Based on Patients' Baseline Body Temperature: A Retrospective Pilot Study.

Background The management of neutropenic fever patients remains challenging. Patients' individual baseline body temperature may provide diagnostic and prognostic value. Methods This study is a retrospective analysis of 92 adults admitted for neutropenic fever to model the length of stay (LOS) and the ability to find a definitive diagnosis using the deviation of patients' temperature on admission from their outpatient baseline, acuity on admission, neutropenia level and persistence, fever persistence, and patients' age. Results Patients' average baseline body temperature was 36.7°C+/-0.3°C - the body temperature had to be over four standard deviations above the baseline to reach the gold standard fever threshold of 38.0°C. Their average fever on admission was 38.1°C. Fever etiologies were identified in 48%, and all constituted infections. Multiple regression modelling demonstrated that a longer LOS of >3 days was predicted by larger deviation from baseline body temperature at admission and by fever persistence at 72 hours post-admission, after correcting for the persistence of severe neutropenia (absolute neutrophil count <500) at 72 hours, age, neutropenia level, and need for intensive care unit admission. A similar model could not predict the ability to identify a fever-explaining diagnosis. Conclusions This pilot project provides support for the use of patients' individual baseline body temperature rather than a pre-established universal fever cutoff in the diagnosis of neutropenic fever. Using a personalized cutoff is expected to avoid missing cases. Further, deviation from patients' baseline body temperature at admission could serve as a predictor for the hospital LOS, which can serve as a potential tool for hospital bed management.