Temperature patterns in suspected pediatric acute appendicitis

BACKGROUND: Temperature is routinely considered in children with suspected acute appendicitis (AA), but classic appendicitis scores usually reduce it to a binary fever item. This study aimed to evaluate whether continuous temperature measurements contain diagnostic or severity-related information that may be obscured by conventional fever thresholds. METHODS: We performed a secondary exploratory analysis of the BIDIAP 2 (BIomarkers for the DIagnosis of Appendicitis in Pediatrics 2) cohort, a prospective multicenter diagnostic cohort of children evaluated for suspected AA. Two continuous temperature variables were assessed: maximum temperature reported at home before Emergency Department (ED) presentation and temperature measured at ED admission. Single-predictor logistic models evaluated temperature as a linear term, a quadratic term, and a restricted cubic spline. Apparent area under the receiver operating characteristic curve (AUC) estimates with 95% confidence intervals (CIs), likelihood-ratio tests, bootstrap optimism correction, and literature-based fever thresholds were examined. RESULTS: For AA versus non-surgical abdominal pain (NSAP), maximum home temperature showed no apparent discrimination when modeled linearly (AUC, 0.494; 95% CI, 0.433-0.554), whereas quadratic and spline models yielded higher apparent AUCs of 0.617, with the highest estimated probability of AA in the low-grade fever range. ED temperature showed a similar but weaker pattern, with AUC increasing from 0.591 in the linear model to 0.613 with quadratic modeling and 0.620 with spline modeling. Among children with AA, temperature behaved more like a severity-related marker: maximum home temperature showed moderate apparent discrimination for complicated AA (AUC, 0.657), with no relevant gain from non-linear modeling. Bootstrap optimism correction modestly reduced the AUCs of non-linear models while preserving the qualitative pattern. CONCLUSIONS: Temperature showed outcome-dependent patterns in suspected pediatric acute appendicitis. The main finding was a non-linear association centered around the low-grade fever range for AA versus NSAP, contrasting with a more consistent association between higher temperature and complicated disease among children with confirmed AA. These findings support interpreting temperature according to the clinical outcome being evaluated, preserving it as a continuous clinical variable, and examining its functional form before reducing it to conventional fever thresholds in diagnostic scores or prediction models.