Setting definitions of childhood asthma in epidemiologic studies

Abstract

The definitions of childhood asthma differ among epidemiological studies. We aimed to compare the diagnostic accuracy and agreement of previous asthma definitions and to evaluate questionnaire-driven asthma definition patterns. Data on 808 participants aged 7-12years from 20 schools were analyzed. Asthma definitions based on symptoms assessed by questionnaire, parent-reported medical asthma diagnosis, current use of asthma medication, airway reversibility after bronchodilatation, and increased (≥35ppb) levels of exhaled nitric oxide (eNO) were compared with a gold standard (medical diagnosis of asthma with asthma symptoms in the past 12months and/or airway reversibility). Concordance was assessed by Cohen's kappa. Data-driven analysis was applied to the questionnaire, and six scores were determined "asthma," "rhinitis," "cough," "non-specific respiratory symptoms," "treated asthma," and "uncontrolled asthma." Tree decisions were built using these scores. Definitions of asthma based on respiratory symptoms, airway reversibility, and parent-reported medical diagnosis had a low sensitivity but high specificity. Agreement between reported and objective measures was poor. Parent-reported medical asthma diagnosis, but not reversibility or eNO, was able to predict questionnaire-driven symptom patterns. Decision trees indicated that those with recent non-specific respiratory symptoms had a higher probability of positive bronchodilation. A standardized operational definition of asthma should include a composite score based on reported asthma medical diagnosis, symptoms, and lung function.