Purpose: The aim of this study is to validate medication proxies for the identification of children diagnosed with atopic disorders that can be applied in various types of epidemiological research.
Methods: Records of 7439 children, aged between 0 and 10 years, in the period 2001 until 2010, were retrieved from the Registration Network Groningen database, a general practitioners database in the north-eastern part of the Netherlands. The sensitivity and positive predictive value (PPV) of 22 medication proxies for the identification of children diagnosed with atopic disorders (asthma, atopic dermatitis, and allergic rhinitis) were computed using the registered diagnoses as gold standards. In addition, different capture periods (1 year, half year, and length of study period) for the detection of prescriptions were tested for all the medication proxies.
Results: The highest PPV (0.84, 95 % CI 0.81-0.87) in combination with a sufficient sensitivity value (0.54, 95 % CI 0.50-0.57) for the identification of children diagnosed with asthma was yielded for the medication proxy, ≥2 prescriptions for anti-asthma medication within 1 year, including 1 inhaled steroid. PPV and sensitivity were even higher in the age group 6-10 years. The proxies designed for the identification of children diagnosed with atopic dermatitis and allergic rhinitis yielded only high PPVs (≥0.75) in combination with low sensitivity values (≤0.22). Altering the capture period for the detection of prescriptions to half a year or the length of the study period only affected sensitivity values.
Conclusion: Children diagnosed with asthma can be identified reliably with a range of medication proxies. The use of prescription data for the identification of children diagnosed with atopic dermatitis and allergic rhinitis is questionable.
Keywords: Asthma; Atopic diseases; Eczema; Hay fever; Prescriptions; Sensitivity; Specificity.