296 - Agreement Between Parent-Reported and Clinician-Documented Food Allergies in Young Children

Publication Number: 2286.296

Daniel Reiss, Boston Children's Hospital, Boston, MA, United States; Sheryl L. Rifas-Shiman, Harvard Medical School and HPHCI, Boston, MA, United States; Emily Oken, Harvard Pilgrim Health Care Institute, Boston, MA, United States; Jenifer R. Lightdale, Harvard Medical School, Boston, MA, United States; Karen Switkowski, Stanford University School of Medicine, Boston, MA, United States

Background: Much currently available epidemiologic data on childhood food allergies rely on parent reports, including those used in large nationally representative cohorts to track overall and allergen-specific prevalence. Few studies, however, have compared parent reports with clinical records.

Objective: To assess the statistical agreement between parent-reported and clinician-documented food allergy data in early childhood within Project Viva, a well-established, longitudinal, pre-birth research cohort unselected for allergy risk.

Design/Methods: We reviewed parent reports of any food allergies for children on Project Viva questionnaires administered at ages 2, 3, and 4 years , and categorized responses for the 10 most common food allergens. We matched these subjects with medical records and abstracted the same 10 food allergens through age 5 years to allow for potential lag in documentation. We assessed agreement between parent-reported and clinician-documented food allergies using Cohen's kappa statistic. IRB approval was granted for all study procedures.

Results: Of the 2128 children enrolled in Project Viva, 584 had both questionnaire data on food allergens and corresponding medical records . Parents reported a total of 150 food allergies in 84 children, and clinicians documented 132 in 63 children (Table 1). The most prevalent allergens in both groups were peanut (n=36 for parent-reported and 35 for clinician-documented), egg (n=27 and 24), and milk (n=27 and 17). Overall agreement between parent report and clinician documentation for the 10 most commonly reported allergens in the cohort was 98.2%, corresponding to a kappa of 0.62 , indicating substantial agreement. Agreement at the level of individual allergens (Table 2) ranged from fair (k=0.25 for soy and k=0.28 for strawberry) to substantial (k=0.81 for peanut).

Conclusion(s): Parent reports and clinician documentation of food allergens in our cohort showed substantial overall agreement, although there was variation at the level of individual allergens. These findings suggest that parent-reported questionnaire data regarding food allergies are likely reliable in Project Viva. Our findings may be generalizable to other epidemiologic studies of food allergy relying on parent report, although comparison between parent report and medical records is ideally something that other studies will pursue , when possible. Further research is also needed to compare parent-reported and clinician-documented allergies with rates of confirmatory allergy testing.

Table 1
Food allergy by allergen and reporter.

Table 2
Agreement of clinician and parent by specific allergen.