Reliably estimating prevalences of atopic children: An epidemiological study in an extensive and representative primary care database

Electronic health records stored in primary care databases might be a valuable source to study the epidemiology of atopic disorders and their impact on health-care systems and costs. However, the prevalence of atopic disorders in such databases varies considerably and needs to be addressed. For this study, all children aged 0-18 years listed in a representative primary care database in the period 2002-2014, with sufficient data quality, were selected. The effects of four different strategies on the prevalences of atopic disorders were examined: (1) the first strategy examined the diagnosis as recorded in the electronic health records, whereas the (2) second used additional requirements (i.e., the patient had at least two relevant consultations and at least two relevant prescriptions). Strategies (3) and (4) assumed the atopic disorders to be chronic based on strategy 1 and 2, respectively. When interested in cases with a higher probability of a clinically relevant disorder, strategy 2 yields a realistic estimation of the prevalence of atopic disorders derived from primary care data. Using this strategy, of the 478,076 included children, 28,946 (6.1%) had eczema, 29,182 (6.1%) had asthma, and 28,064 (5.9%) had allergic rhinitis; only 1251 (0.3%) children had all three atopic disorders. Prevalence rates are highly dependent on the clinical atopic definitions used. The strategy using cases with a higher probability of clinically relevant cases, yields realistic prevalences to establish the impact of atopic disorders on health-care systems. However, studies are needed to solve the problem of identifying atopic disorders that are missed or misclassified