The validity of register data to identify children with atopic dermatitis, asthma or allergic rhinoconjunctivitis

BackgroundThe incidence of atopic dermatitis, wheezing, asthma and allergic rhinoconjunctivitis has been increasing. Register-based studies are essential for research in subpopulations with specific diseases and facilitate epidemiological studies to identify causes and evaluate interventions. Algorithms have been developed to identify children with atopic dermatitis, asthma or allergic rhinoconjunctivitis using register information on disease-specific dispensed prescribed medication and hospital contacts, but the validity of the algorithms has not been evaluated. This study validated the algorithms vs gold standard deep telephone interviews with the caretaker about physician-diagnosed atopic dermatitis, wheezing, asthma or allergic rhinoconjunctivitis in the child. MethodsThe algorithms defined each of the three atopic diseases using register-based information on disease-specific hospital contacts and/or filled prescriptions of disease-specific medication. Confirmative answers to questions about physician-diagnosed atopic disease were used as the gold standard for the comparison with the algorithms, resulting in sensitivities and specificities and 95% confidence intervals. The interviews with the caretaker of the included 454 Danish children born 1997-2003 were carried out May-September 2015; the mean age of the children at the time of the interview being 15.2years (standard deviation 1.3years). ResultsFor the algorithm capturing children with atopic dermatitis, the sensitivity was 74.1% (95% confidence interval: 66.9%-80.2%) and the specificity 73.0% (67.3%-78.0%). For the algorithm capturing children with asthma, both the sensitivity of 84.1% (78.0%-88.8%) and the specificity of 81.6% (76.5%-85.8%) were high compared with physician-diagnosed asthmatic bronchitis (recurrent wheezing). The sensitivity remained high when capturing physician-diagnosed asthma: 83.3% (74.3%-89.6%); however, the specificity declined to 66.0% (60.9%-70.8%). For allergic rhinoconjunctivitis, the sensitivity was 84.4% (78.0-89.2) and the specificity 81.6% (75.0-84.4). ConclusionThe algorithms are valid and valuable tools to identify children with atopic dermatitis, wheezing, asthma or allergic rhinoconjunctivitis on a population level using register data.