A comparison between criteria for diagnosing atopic eczema in infants

Epidemiological studies have shown different estimates of the frequency of atopic eczema (AE) in children. This may be explained by several factors including variations in the definition of AE, study design, age of study group, and the possibility of a changed perception of atopic diseases. The role of IgE sensitization in AE is a matter of debate.

To determine the prevalence and cumulative incidence of AE in a group of unselected infants followed prospectively from birth to 18 months of age using different diagnostic criteria; to evaluate the agreement between criteria; and to describe the association between atopic heredity and postnatal sensitization, respectively, and the development of AE according to the different diagnostic criteria.

During a 1-year period a consecutive series of 1095 newborns and their parents were approached at the maternity ward at the Odense University Hospital, Denmark and a cohort of 562 newborns was established. Infants were examined and followed prospectively from birth and at 3, 6, 9, 12 and 18 months of age. AE was diagnosed using four different criteria, the Hanifin and Rajka criteria, the Schultz-Larsen criteria, the Danish Allergy Research Centre (DARC) criteria developed for this study and doctor-diagnosed visible eczema with typical morphology and atopic distribution. Additionally, the U.K. diagnostic criteria based on a questionnaire were used at 1 year of age. Agreement between the four criteria was analysed at each time point and over time, and agreement between the four criteria and the U.K. questionnaire criteria was analysed.

The cumulative 1-year prevalence of AE using the Hanifin and Rajka criteria was 9.8% (95% confidence interval, CI 7-13%), for the Schultz-Larsen criteria it was 7.5% (95% CI 5-10%), for the DARC criteria 8.2% (95% CI 6-11%), for visible eczema 12.2% (95% CI 9-16%) and for the U.K. criteria 7.5% (95% CI 5-10%). The pairwise agreement between criteria showed good agreement, with rates varying between 93% and 97% and kappa scores between 0.6 and 0.8. Agreement analysis of diagnoses between the four criteria demonstrated that cumulative incidences showed better agreement than point prevalence values.

Agreement between different criteria for diagnosing AE was acceptable, but the mild cases constituted a diagnostic problem, although they were in the minority. Repeated examinations gave better agreement between diagnostic criteria than just one examination. Atopic heredity was less predictive for AE than sensitization to common food and inhalant allergens in early childhood.