Risk stratification for atopic dermatitis, particularly in utero or early in life, can be an important step toward researching or implementing targeted prevention interventions. Family history and environmental factors associated with the development of atopic dermatitis abound, yet a method of combining risk factors to determine an aggregate risk is currently lacking.

We performed a retrospective, cross-sectional database study on the Leumit Health Services electronic medical record database (Israel). A clinical score was developed using the odds ratios for predictive factors from a logistic regression model. Infant-mother dyads were included if the infant was born after 2010 and before 2019. The outcome measure was a ICD-9 diagnosis of atopic dermatitis before reaching three years of age (n=7,370) and the healthy arm consisted of a cohort of non-atopic dermatitis children (n=63,852). Raw clinical variables include parental and sibling history of atopic conditions, gender, expected season of birth, maternal medications while pregnant, and environmental factors. All variables were derived from the prenatal period. Score performance was measured against the entire development cohort and two independent external cohorts from different geographies.  

24% of the development cohort and 26%-28% of the external validation cohorts were considered high risk. The negative predictive value in the development cohort was 95% and it was 83%-93% in the external validation cohort, representing a 26%-35% increase in negative predictive value over the standard of care. 

We present the development and validation of a novel scoring system aimed at identifying high-risk infants for atopic dermatitis. The score serves as an effective elimination tool in clinical settings, efficiently screening out low-risk individuals based on its high Negative Predictive Value.