Reproducibility of food challenge to cow’s milk: Systematic review with individual participant data meta-analysis

Abstract

Cow's milk (CM) is an increasingly common cause of severe allergic reactions, but there is uncertainty with respect to severity of reactions at low-level CM exposure, as well as the reproducibility of reaction thresholds. We undertook an individual participant data (IPD) meta-analysis of studies reporting double-blind, placebo-controlled food challenges in CM to determine the rate of anaphylaxis to low-level exposures and the reproducibility of reaction thresholds. We performed a systematic review and IPD meta-analysis of studies reporting relevant data. Authors were contacted to provide additional data and/or clarification as needed. Risk of bias was assessed using the National Institute for Clinical Excellence methodologic checklists. Thirty-four studies were included, representing data from over 1000 participants. The cumulative ED₀₁ and ED₀₅ (cumulative doses causing objective symptoms in 1% and 5% of the at-risk allergic population) were 0.3 (95% confidence interval [CI], 0.2-0.5) and 2.9 (95% CI, 1.6-5.4) mg, respectively. At meta-analysis, 4.8% (95% CI, 2.0-10.9) and 4.8% (95% CI, 0.7-27.1) of individuals reacting to ≤5 mg and ≤0.5 mg of CM protein had anaphylaxis (minimal heterogeneity, I²= 0%). Then 110 individuals underwent repeat double-blind, placebo-controlled food challenges; the intraindividual variation in reaction threshold was limited to a ½-log change in 80% (95% CI, 65-89) of participants. Two individuals initially tolerated 5 mg CM protein but then reacted to this dose at a subsequent challenge, although neither had anaphylaxis. About 5% of CM-allergic individuals reacting to ED₀₁ or ED₀₅ exposure might have anaphylaxis to that dose. This equates to 5 and 24 anaphylaxis events per 10,000 patients exposed to an ED₀₁ or ED₀₅ dose, respectively, in the broader CM-allergic population. Most of these anaphylactic reactions would be mild and respond to a single dose of epinephrine.