Feasibility and Concurrent Validity of the Family-Assisted Severe Illness Therapy Tool Compared With Pediatric Early Warning Systems in Pediatric Oncology

PURPOSE Children with cancer in low- and middle-income countries face high preventable in-hospital mortality, often because of delayed recognition of clinical deterioration. Pediatric Early Warning Systems (PEWS) enable early detection but require frequent clinician assessments that are difficult to sustain in resource-constrained settings. The Family-Assisted Severe Illness Therapy (FASTER) tool trains caregivers to perform simple bedside assessments as an alternative screening approach. We evaluated the concurrent validity, feasibility, and acceptability of caregiver-recorded FASTER scores versus clinician-recorded PEWS scores among pediatric oncology inpatients at the Uganda Cancer Institute. METHODS In this prospective, concurrent validity study, 50 caregiver-patient pairs were enrolled. After structured training, caregivers performed FASTER assessments every 6 hours over 48 hours while a study nurse independently recorded PEWS scores every 8 hours, yielding 321 concurrent paired observations for analysis. Concurrent validity was assessed using Spearman correlation and kappa agreement statistics. Diagnostic performance for PEWS-defined deterioration (PEWS ≥3) was evaluated using sensitivity, specificity, predictive values, and AUROC. Feasibility and acceptability were assessed through adherence monitoring and structured exit interviews. RESULTS FASTER scores correlated moderately to strongly with PEWS scores (ρ = 0.65 [95% CI, 0.58 to 0.71]; P < .001), although categorical agreement was poor (unweighted κ = 0.07). For PEWS-defined deterioration, FASTER showed high sensitivity (96.7%) and modest specificity (47.5%); AUROC was 0.89 (95% CI, 0.84 to 0.94), reflecting yellow-threshold discrimination given only two PEWS-red events. The positive predictive value was 29.7%, and the negative predictive value was 98.4%. Caregiver adherence was high (98.3%), and acceptability was strong (100% useful, 78% easy/very easy), although lower education predicted greater difficulty ( P = .014). CONCLUSION FASTER demonstrated concurrent validity with PEWS and high feasibility and acceptability; its predictive validity and impact on clinical outcomes require further study.