Asthma symptoms are non-specific during infancy, making the identification of different subgroups among preschool children with early respiratory manifestations an important challenge. We previously used a clustering approach to identify bronchial obstructive phenotypes in 1-yr-old infants from the Pollution and Asthma Risk: an Infant Study (PARIS) birth cohort. In the present study, we examined whether these phenotypes were stable at 3 yr and studied their comorbidity and risk factors. Partitioning around medoids (PAM) method was applied at 1 and 3 yr of age to cluster children according to wheezing, dry night cough, dyspnoea with sleep disturbance and breathlessness. The resulting groups were used to derive phenotypes in 2084 children during their first 3 yr of life. Analysis of associated comorbidity and risk factors was conducted using multinomial logistic regression. PAM groups were similarly defined at both ages so that two respiratory phenotypes were identified between birth and 3 yr: cough phenotype (CP) and dyspnoea phenotype (DP) including 14.1% and 30.7% of children, respectively. CP infants experienced more often allergic features than DP, dominated by respiratory infections. Parental history of allergy, potential allergen exposure and psychosocial factors were associated with CP. Day care centre attendance was more frequent in DP as well as exposure to domestic chemical pollution, suggesting a greater vulnerability to pathogens. Finally, dry night cough and dyspnoea disturbing the sleep appear to be markers of two respiratory profiles potentially allergic and infectious before 3 yr old.
© 2012 John Wiley & Sons A/S.