Strengths and limitations
A major strength of this study is the use of routinely collected prescription- and healthcare information in nationwide registers with high completeness and data validity (32). Denmark has a longstanding tradition for public registers and a universally tax-funded healthcare system, which ensures coverage of the entire Danish population regardless of differences in socioeconomic class or insurance status (32). Due to OCS and other asthma drugs being prescription-only medication, no potential over-the-counter exchanges were neglected.
However, several limitations much be acknowledged. In lack of access to diagnostic data from primary care, we used medical prescription data as proxies for asthma diagnoses. The method of identifying asthma patients from prescribing data has been validated as a reliable method by several European studies (19, 20). We chose a conservative upper age cut-off of 45 years to minimize the presence of COPD patients. We furthermore excluded individuals diagnosed with COPD, cystic fibrosis or diseases commonly treated with OCS in order to increase the probability of the OCS use being attributed to asthma. The study design’s restrictions might make our estimates more conservative, reflected in our finding of a prevalence of active asthma at 3.4%. Also, the mildest cases of asthma, requiring less than two asthma drugs per year, were not identified. Due to the use of prescription data, we might have overestimated the actual OCS use, as a dispensed prescription is not synonymous to the medication being consumed. The use of dispensed prescriptions, however, reduce the risk of misclassification due to primary non-adherence. We were not able to account for possible stockpiling. Data on the underlying indications for the prescribed OCS were not available, and although we sought to describe only the OCS usage in asthma treatment, OCS prescriptions for asthma patients are often prescribed for other conditions (31). We sought to accommodate this by excluding patients with several comorbidities likely treated with OCS. In addition, we did not include OCS prescriptions filled during periods of apparent inactive asthma as defined in the study design, which possibly have led to an underestimation of the total OCS exposure per individual.