Abstract
Background: Predicting in early life which children will develop asthma is difficult.
Objective: We aimed to develop a prediction model for asthma integrating socio-demographic, growth, lifestyle and dietary factors, clinical conditions, allergy and lung function in critical developmental periods from prepregnancy to adolescence.
Methods: We used individual participant data of 134,563 children from the EU Child Cohort Network to predict current asthma during childhood (median 7y), and in adolescence (median 13y) by candidate predictors in consecutive periods from prepregnancy to infancy, preschool age, and school age. We performed 1-stage fixed-effect meta-analyses on the DataSHIELD platform using backward logistic regression and area under the receiver operating curve modeling and external leave-one-cohort-out cross-validation.
Results: Early life factors predicted childhood asthma well (range AUC (95%CI): 0.63 (0.62, 0.65) to 0.74 (0.68, 0.80)). Prominent predictors were parental asthma, eczema or allergy, higher maternal prepregnancy body mass index, child’s sex, lower gestational age and birth weight, and higher infant weight gain (pregnancy to infancy), wheezing and smoke exposure (preschool age), and wheezing and obesity (school age) (p-values <0.025). Findings remained consistent for adolescent asthma and across North and South European countries, maternal education levels and sex.
Conclusions: This large-scale validated prediction model effectively predicts asthma based on early life factors from prepregnancy onwards across different socio-demographic backgrounds and sex, indicating its potential for guiding personalized early intervention strategies.
Objective: We aimed to develop a prediction model for asthma integrating socio-demographic, growth, lifestyle and dietary factors, clinical conditions, allergy and lung function in critical developmental periods from prepregnancy to adolescence.
Methods: We used individual participant data of 134,563 children from the EU Child Cohort Network to predict current asthma during childhood (median 7y), and in adolescence (median 13y) by candidate predictors in consecutive periods from prepregnancy to infancy, preschool age, and school age. We performed 1-stage fixed-effect meta-analyses on the DataSHIELD platform using backward logistic regression and area under the receiver operating curve modeling and external leave-one-cohort-out cross-validation.
Results: Early life factors predicted childhood asthma well (range AUC (95%CI): 0.63 (0.62, 0.65) to 0.74 (0.68, 0.80)). Prominent predictors were parental asthma, eczema or allergy, higher maternal prepregnancy body mass index, child’s sex, lower gestational age and birth weight, and higher infant weight gain (pregnancy to infancy), wheezing and smoke exposure (preschool age), and wheezing and obesity (school age) (p-values <0.025). Findings remained consistent for adolescent asthma and across North and South European countries, maternal education levels and sex.
Conclusions: This large-scale validated prediction model effectively predicts asthma based on early life factors from prepregnancy onwards across different socio-demographic backgrounds and sex, indicating its potential for guiding personalized early intervention strategies.
| Originalsprog | Engelsk |
|---|---|
| Artikelnummer | PA1343 |
| Tidsskrift | The European Respiratory Journal |
| Vol/bind | 64 |
| Udgave nummer | Suppl. 68 |
| Antal sider | 1 |
| ISSN | 0903-1936 |
| DOI | |
| Status | Udgivet - 2024 |