Food Allergies and Eating Disorders: An Exploration of Screening Prevalence and Associated Factors Among Adults

Abstract

Background/Objectives: Food allergy (FA) affects up to 10% of adults and requires ongoing dietary vigilance that may increase vulnerability to eating disorders (EDs), yet evidence in adults remains limited. This study compared ED screening and history between adults with and without food allergy (FA), and explored factors associated with positive screening among those with FA. Methods: This cross-sectional study recruited 371 adults (134 with FA, 237 without) via an online survey. Positive screening required meeting thresholds on both the Eating Attitudes Test-8 (EAT-8) and the Sick, Control, One, Fat, Food (SCOFF); ED history and treatment were assessed via self-report. Group differences were examined using chi-square tests and, for the primary comparison, age- and sex-adjusted logistic regression. Among those with FA, correlations identified candidate predictors of positive screening, which were entered into a further logistic regression. Results: FA status was not associated with positive screening for an ED (37.3% vs. 37.1%; p = 0.972), including after adjusting for age and sex. FA was significantly associated with a greater lifetime history of EDs (36.4% vs. 20.7%; p = 0.001) and ED treatment (17.9% vs. 10.1%; p = 0.032). Among those with FA, positive screening was independently predicted by ED history and shorter allergy duration in a multivariable model (Nagelkerke R2 = 0.14); however, allergy duration and age were highly correlated, so this effect could not be shown to be independent of age. Conclusions: FA alone was not statistically associated with current positive screening but was linked to lifetime history and treatment, suggesting that this relationship may be more evident earlier in the FA experience. ED history and shorter allergy duration were associated with positive screening among those with FA, though the duration effect may partly reflect age; sex was linked to screening only at the bivariate level, given too few male positive cases to model. Findings are preliminary, given the FA-unvalidated screening tools used, warranting longitudinal validated research.