The Vitality algorithm: Classifying infant IgE-mediated food allergy in research studies when participant oral food challenges are not performed

Background: Food allergy algorithms provide a systematic approach to assigning food allergy outcomes in research studies when oral food challenges (OFC) are not performed. Currently no such algorithms, validated for infants, exist. We aimed to develop and validate a research algorithm for classifying infant food allergies. Methods: An algorithm incorporating skin prick test (SPT), allergen ingestion, and reaction data, was developed and validated by comparing peanut, cashew, egg and cow’s milk algorithm outcomes to OFC outcomes, in infants from the Vitality (n=285) (NCT02112734), EarlyNuts (n=111) (ACTRN12618001990213), and HealthNuts (n=191) studies. Results: Validation included 740 OFCs from 587 infants (191 SPT-negative [0mm] and 396 SPT-positive [≥ 1mm] infants; first-OFC median age: 14.4 months). The algorithm assigned all SPT-negative infants as tolerant, achieving 99.5% agreement with OFCs (peanut 100% [n=135/135]; egg 98.2% [n=55/56]). The algorithm assigned an allergic or tolerant outcome to 57% (82/144) of peanut SPT-positive, 28% (30/108) cashew SPT-positive, 76% (n=196/258) egg SPT-positive, and 88% (n=35/39) milk SPT-positive, infants. Agreement with OFC was 81% (n=66/82) for peanut, 83% (n=25/30) for cashew, 90% (n=176/196) for egg and 97% (n=34/35) for milk. Accuracy and areas under the curve (AUCs) were fair-to-excellent: peanut (AUC=0.79 [95% confidence interval [CI]: 0.70-0.90]; sensitivity=88% [95%CI: 75-96]; specificity=70% [95%CI: 51-84]), cashew (AUC=0.83 [95%CI: 0.69-0.97]; sensitivity=83% [95%CI: 59-96]; specificity=83% [95%CI: 52-98]); egg (AUC=0.85 [95%CI: 0.77-0.93]; sensitivity=92% [95%CI: 86-95]; specificity=79% [95%CI: 59-92]); milk (AUC=0.98 [95%CI: 0.94-1]; sensitivity=96% [95%CI: 80-100]; specificity=100% [95%CI: 69-100]). Conclusion: Using routinely collected data, this validated infant-focused food allergy algorithm performed well in classifying peanut, cashew, egg and cow’s milk allergies, offering applicability in research settings.