Predicting Infectious ComplicatioNs in Children with Cancer: an external validation study

Area of Science:

• Pediatric Oncology
• Infectious Diseases
• Clinical Decision Rules

Background:

• Fever and neutropenia (FN) in children with cancer poses a significant risk of serious infection.
• Accurate prediction of microbiologically documented infection (MDI) is crucial for appropriate management.
• The Predicting Infectious ComplicatioNs in Children with Cancer (PICNICC) clinical decision rule (CDR) was developed to aid in this prediction.

Purpose of the Study:

• To validate the PICNICC CDR for predicting MDI in Australian children with cancer experiencing FN.
• To assess the cost-effectiveness of current FN management strategies.

Main Methods:

• Retrospective collection of data from 650 FN episodes, including demographics, outcomes, and costs.
• Assessment of the PICNICC CDR's discrimination, calibration, sensitivity, and specificity.
• Comparison of the CDR's performance in the validation cohort versus its derivation dataset.

Main Results:

• The PICNICC CDR performed poorly upon initial presentation without recalibration (AUC-ROC 0.638).
• Recalibration and application at day 2, considering known MDI proportions, improved sensitivity to 87.7%.
• Predictive ability for bacteraemia was similar to the derivation study, especially after recalibration (AUC-ROC 0.71).
• Length of stay significantly contributes to FN treatment costs (AUD 2183/day).

Conclusions:

• The PICNICC CDR requires local recalibration for optimal performance in predicting MDI in this cohort.
• The rule shows promise for identifying low-risk patients after a short observation period, potentially reducing healthcare costs.
• Implementing a low-risk pathway using the PICNICC CDR after inpatient observation may be safe and cost-effective.