Safety and predictors of adverse events during oral immunotherapy for milk allergy: severity of reaction at oral challenge, specific IgE and prick test

Summary Background Strict avoidance is the only accepted management for cow's milk ( CM ) allergy. CM oral immunotherapy ( CM ‐ OIT ) is under investigation. Objectives To evaluate long‐term safety of CM ‐ OIT . To identify clinical/immunological predictors of adverse events. Methods Prospective longitudinal epidemiological intervention study. CM ‐allergic children aged 5–18 underwent a Spanish‐approved CM ‐ OIT protocol without premedication. Clinical data, skin prick test ( SPT ) and specific IgE ( sIgE ) at baseline and 1 year after OIT were registered. All dose‐related reactions, treatments needed and cofactors involved were recorded. Through survival analysis, we studied the cumulative probability of reactions resolution over time and clinical/immunological risk factors of reactions persistence. Results 81 children were recruited. Mean follow‐up was 25 months. 95% of children suffered reactions, 91% of which affected a single organ. Reactions were heterogeneously distributed: (a) 60 children (75%) had occasional symptoms which ceased over time. 86% of them reached complete desensitization (200 mL). (b) 20 children (25%) suffered frequent (78% of total reactions), more severe and unpredictable reactions, which persisted during follow‐up or led to withdrawal (6 cases). Reactions persistence was associated with a higher frequency and severity. Kaplan–Meier estimate revealed a cumulative probability of reactions resolution of 25% at 3 months (95% CI: 1.9–4.1) and 50% (95% CI: 6.1–9.9) at 8 months based on all patients. Cox proportional hazards multivariate regression model identified 3 variables (CM‐ sIgE ≥ 50 KU L −1 , CM‐SPT ≥ 9 mm and Sampson's severity grades 2, 3 and 4 at baseline food challenge) as independent risk factors of reactions persistence. The combination of 2 or 3 of these factors involved hazard ratios to develop persistent reactions of 2.26 (95% CI: 1.14–4.46; P  = 0.019) and 6.06 (95% CI: 2.7–13.7; P  < 0.001), respectively. Clinical implications CM ‐ OIT was insufficiently safe in 25% of children. The above‐mentioned clinical and immunological parameters would help clinicians to identify highly reactive patients before CM ‐ OIT . In them, individualized schedules and premedication should be considered.