Skin health survey on atopic dermatitis among Japanese children: The Tohoku Medical Megabank Project Birth and...

Precise skin phenotypic data are indispensable in accurately diagnosing atopic dermatitis (AD). Therefore, this study examined the interobserver concordance for AD and non-AD diagnoses between two dermatologists. AD prevalence determined by the self-reported physician diagnoses and the diagnoses determined from the United Kingdom (UK) diagnostic criteria were compared with the diagnoses made by the two dermatologists, using data from a skin health survey.

This study included 1,638 children that participated in the skin health survey, which was part of the Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study. AD was assessed using dermatologist assessments, self-reported physician diagnoses, and the UK diagnostic criteria. The concordance for diagnoses was evaluated using kappa. The sensitivity and specificity of the self-reported physician diagnoses and the UK diagnostic criteria were calculated by comparing them with the two dermatologists' diagnoses.

Among the 1,638 children, 393 (24.0 %), 194 (11.9 %), and 597 (37.2 %) were diagnosed with AD by the two dermatologists, physicians, and the UK diagnostic criteria, respectively. The kappa (95 % CI) of the interobserver concordance for AD or non-AD diagnoses between the two dermatologists was 0.78 (0.75-0.81). The sensitivity and specificity of the self-reported physician diagnoses were 26.7 % and 94.1 %, respectively. The sensitivity and specificity of the UK diagnostic criteria were 85.0 % and 82.4 %, respectively.

Interobserver concordance for AD or non-AD diagnoses between the two dermatologists was substantial. Self-reported physician diagnoses exhibited low sensitivity that potentially indicated underdiagnosis of AD, whereas the UK diagnostic criteria might overdiagnose AD.