Epidemiological measures of childhood asthma: cross-sectional and longitudinal consistency.

Background: Defining childhood asthma varies considerably, and the extent of agreement between various measures is not clearly understood in the absence of a recognized 'gold standard'. We compared different definitions of childhood asthma, identified characteristics that might have influenced their accuracy and an acquisition of an 'asthma' label in wheezy and treated children.

Methods: Using a prospective, population-based birth cohort of 623 children followed up to the age of 14 years the concordance between parental opinion, doctor's diagnosis reported by the parent and asthma's diagnosis in general practice (GP) was analysed using latent class analysis (LCA).

Results: At the age of eight, 'ever asthma' prevalences ranged from 15.5% (parental opinion) to 21.5% (GP record). 35% of children by the age of eight years had at least one reported label of asthma, reflecting both cross sectional and longitudinal inconsistencies. By the age of 14 years, 16% of children were inconsistently defined as 'ever asthmatic' by their parents. The prevalence of 'ever asthma' estimated by LCA was 19.3%, indicating a parental report of a doctor's diagnosis to be the most sensitive and specific definition. The likelihood of being labelled with asthma was higher in those with a parental or sibling history of asthma, but not determined by socio-demographic characteristics.

Conclusions: Although the estimates of prevalence were similar for parental reports and GP records, agreement between the three sources was less than expected. Parental report of a doctor's diagnosis of asthma is sensitive, specific, longitudinally consistent and not subject to large socio-economic bias.