Cow’s milk is one of the most common food allergens among children. Oral food challenge tests determine the threshold dose of allergens, but have not been standardized. To reduce the severe reactions, we developed a practical model of the test. We studied 111 high-risk patients who underwent a first milk oral food challenge on the risk-stratified dose between 2011 and 2017 for predicting the severe reaction risk. Severe reactions were defined as showing > 3 of Sampson’s classification grade. Twenty-eight patients (25%) showed severe reactions without death. Prior to oral food challenge, severe reaction patients experienced milk avoidance (71% vs. 45%, p = 0.02) or bronchial asthma (61% vs. 28%, p = 0.003) more frequently and showed higher milk-specific IgE levels (median 28.3 vs. 7.7 U_A/mL, p < 0.0001) than non-severe reaction patients. Multivariate logistic regression analyses established a formula including severe reaction-associated factors; increased levels of milk-specific IgE (odds ratio 11.61, p = 0.001), milk avoidance (odds ratio 3.88, p = 0.02), and bronchial asthma (odds ratio 3.75, p = 0.02). This model had 86% sensitivity and 56% specificity (cut-off 0.25) for risk. Five patients with < 25% probability developed severe reactions, which started in > 3 grade dyspnea up to 20 mL of challenge.

Conclusion: This model could effectively reduce the severe reaction development on the first milk oral food challenge test according to the individual needs.