Developing a Prediction Model for Determination of Peanut Allergy Status in the Learning Early About Peanut Allergy (LEAP) Studies.

Background: The Learning Early About Peanut Allergy (LEAP) study team developed a protocol-specific algorithm using dietary history, peanut-specific IgE, and skin prick test (SPT) to determine peanut allergy status if the oral food challenge (OFC) could not be administered or did not provide a determinant result.
Objective: To investigate how well the algorithm determined allergy status in LEAP; to develop a new prediction model to determine peanut allergy status when OFC results are not available in LEAP Trio, a follow-up study of LEAP participants and their families; and to compare the new prediction model with the algorithm.
Methods: The algorithm was developed for the LEAP protocol before the analysis of the primary outcome. Subsequently, a prediction model was developed using logistic regression.
Results: Using the protocol-specified algorithm, 73% (453/617) of allergy determinations matched the OFC, 0.6% (4/617) were mismatched, and 26% (160/617) participants were nonevaluable. The prediction model included SPT, peanut-specific IgE, Ara h 1, Ara h 2, and Ara h 3. The model inaccurately predicted 1 of 266 participants as allergic who were not allergic by OFC and 8 of 57 participants as not allergic who were allergic by OFC. The overall error rate was 9 of 323 (2.8%) with an area under the curve of 0.99. The prediction model additionally performed well in an external validation cohort.
Conclusion: The prediction model performed with high sensitivity and accuracy, eliminated the problem of nonevaluable outcomes, and can be used to estimate peanut allergy status in the LEAP Trio study when OFC is not available.
(Copyright © 2023 The Authors. Published by Elsevier Inc. All rights reserved.)