External Validation Of Asthma Prediction Tool In The French Elfe Cohort

Many of the proposed predictive scores for risk stratification of asthma have limited utility in preschool children due to inclusion of invasive procedures. Recently, the Leicester cohort (UK) developed a non-invasive prediction tool and we aimed to perform an external validation in our cohort. Participation at year 1 and 5 acted as the inclusion criteria. Information about 10 predictors (age1) and outcomes (asthma -age5) was obtained from parental responses to annual questionnaires. The total scores (0-15) were used for risk categorization (low, medium, and high).Logistic regression was used to calculate the odds ratio for developing asthma in each category. The tool’s performance was assessed by Brier scores and the areas under the receiver operating characteristic curves (ROC). We had complete information on 9,998 children. 71% were classified at low risk, 26.3% medium risk and 2.5% at high risk of developing asthma at age 5. In comparison to the low risk group, children in the medium risk category were 3 times more likely to develop asthma (p=0.000) whereas 14 times more likely in the high risk category (p=0.000). At age 5, 16.1% children were asthmatic and 84% non-asthmatic. In comparison to the development cohort, calibration was better (Brier score: 0.16 vs 0.12) whereas discrimination was expectedly lower (AUC 0.74 vs 0.65) yet sufficient. The tool displayed potential for early risk stratification of asthma in preschool children.