The influence of different fever definitions on the rate of fever in neutropenia diagnosed in children with cancer.

backgroundThe temperature limit defining fever (TLDF) is based on scarce evidence. This study aimed to determine the rate of fever in neutropenia (FN) episodes additionally diagnosed by lower versus standard TLDF.

methodsIn a single center using a high TLDF (39.0°C tympanic temperature, LimitStandard), pediatric patients treated with chemotherapy for cancer were observed prospectively. Results of all temperature measurements and CBCs were recorded. The application of lower TLDFs (LimitLow; range, 37.5°C to 38.9°C) versus LimitStandard was simulated in silicon, resulting in three types of FN: simultaneous FN, diagnosed at both limits within 1 hour; earlier FN, diagnosed >1hour earlier at LimitLow; and additional FN, not diagnosed at LimitStandard.

resultsIn 39 patients, 8896 temperature measurements and 1873 CBCs were recorded during 289 months of chemotherapy. Virtually applying LimitStandard resulted in 34 FN diagnoses. The predefined relevantly (≥15%) increased FN rate was reached at LimitLow 38.4°C, with total 44 FN, 23 simultaneous, 11 earlier, and 10 additional (Poisson rate ratioAdditional/Standard, 0.29; 95% lower confidence bound, 0.16). Virtually applying 37.5°C as LimitLow led to earlier FN diagnosis (median, 4.5 hours; 95% CI, 1.0 to 20.8), and to 53 additional FN diagnosed. In 51 (96%) of them, spontaneous defervescence without specific therapy was observed in reality.

conclusionLower TLDFs led to many additional FN diagnoses, implying overtreatment because spontaneous defervescence was observed in the vast majority. Lower TLDFs led as well to relevantly earlier diagnosis in a minority of FN episodes. The question if the high TLDF is not only efficacious but as well safe remains open.