Predictors of childhood food allergy

Methods This a cross sectional study. To assess which characteristics were associated with food allergy, we conducted simple logistic regressions of each type of atopy with food allergy as the dependent variable. A p<0.05 was considered significant. In the data modeling, we did not assume any hierarchical effects of families and schools on the food allergy status of participating children. Therefore, logistic regression with fixed effects only was used in modeling the data. Stepwise multilogistic regression with forward entry was used to determine a subset of characteristics that predicts best the presence of food allergy within the target population. Data were analyzed using SPSS statistical package (version 18).