Different fever definitions and the rate of fever and neutropenia diagnosed in children with cancer: a retrospective two-center cohort study

Background. The definition of fever, and thus fever and neutropenia (FN), varies between different pediatric oncology centers. Higher temperature limit should reduce FN rates, but may increase rates of FN with complications by delaying therapy. This study determined if different fever definitions are associated with different FN rates. Procedure. Two pediatric oncology centers had used three fever definitions in 2004–2011: ear temperature >=38.5°C persisting >=2 hours (low definition); axillary temperature >=38.5°C >=2 hours or >=39.0°C once (middle); and ear temperature >=39.0°C once (high). Clinical information was retrospectively extracted from charts. FN rates were compared using mixed Poisson regression. Results. In 521 pediatric patients with cancer, 783 FN were recorded during 6,009 months cumulative chemotherapy exposure time (501 years; rate, 0.13/month [95% CI, 0.12–0.14]), 124 of them with bacteremia (16%; 0.021/month [0.017–0.025]). In univariate analysis, the high versus low fever definition was associated with a lower FN rate (0.10/month [0.08–0.11] vs. 0.15/month [0.13–0.16]; rate ratio, 0.66 [0.45–0.97]; P ¼ 0.036), the middle definition was intermediate (0.13/month [0.11–0.15]). This difference was not confirmed in multivariate analysis (rate ratio, 0.94 [0.67–1.33]; P ¼ 0.74). The high versus low definition was not associated with an increased rate of FN with bacteremia (multivariate rate ratio, 1.39 [0.53–3.62]; P ¼ 0.50). Conclusion. A higher fever definition was not associated with a lower FN rate, nor with an increased rate of FN with bacteremia. These may be false negative findings due to methodological limitations. These questions, with their potential impact on health-related quality of life, and on costs, need to be assessed in prospective studies