An inverse relationship between asthma prevalence and medication dispensation trend: a 12-year spatial analysis of electronic health record data in Alberta, Canada

To the Editor:

Asthma is a chronic inflammatory disease of the airways that affects >300 million people worldwide [1]. As of 2019, 2.95 million Canadians were diagnosed with asthma [2]. While there are standard recommendations and guidelines for the diagnosis and treatment of asthma, patients living with this disease often experience poor health-related quality of life [3] and continue to experience exacerbations [4]. Apart from the increasing air pollution, rapid urbanisation, and a growing trend in marijuana and vaping, particularly among the adolescents [5–8], poor patient adherence to asthma medications is a major contributing factor to these outcomes [9–12]. Despite efforts to change this narrative through patient and physician education [13–15], it is well reported that adherence to treatment in asthma is varied, with rates of <50% in children [16] and 30–70% in adults [1, 17].

In Canada, Alberta Health Services data shows an average of 21 000 emergency department admissions for asthma exacerbations each year between 2014 and 2018 [18]. Despite the availability of optimal care facilities in the province of Alberta, such a high number of emergency department visits due to asthma exacerbations led us to seek to understand the current use of asthma medication. In this study, we investigated the yearly prevalence of asthma and the average dispensation rate of 10 common classes of asthma medications in Alberta, Canada, from 2008–2009 to 2019–2020.

In this retrospective study, we obtained medication dispensation data between the 2008–2009 and 2019–2020 fiscal years from the Pharmaceutical Information Network, a platform within the provincial Electronic Health Record of Alberta that provides information about active and previous medications of patients of all ages [18]. To identify patients with physician-diagnosed asthma in a health administrative database, we used a previously validated definition of asthma [19]. Briefly, asthma was defined based on the International Classification of Diseases (ICD) 10th edition codes for asthma (J45) if there was at least one hospitalisation discharge record within the Discharge Abstract Database in a fiscal year or two physician billing records within the past 2 years with a diagnosis of asthma-like symptoms (ICD-9-CM code 493). Asthma incidence was identified using the same case definition above but with the additional criteria of being an Alberta resident for 5 years before the diagnosis of asthma. We obtained the total number of dispensations for the following drug categories for patients with asthma: inhaled corticosteroids (ICS), long-acting β₂-adrenergic receptor agonists (LABA), ICS/LABA, short-acting β₂-adrenergic receptor agonists (SABA), antibiotics, long- and short-acting muscarinic receptor antagonist (LAMA and SAMA, respectively), oral corticosteroids (OCS), leukotriene receptor antagonists (LTRA), and biologics.

Descriptive statistics are presented as n (%). We used Spearman's rank-order correlation to analyse the association between asthma prevalence and prescription dispensation for each of the drug categories, for all patients and separately for adults and paediatric patients. The coefficients were plotted as heatmaps and a p-value <0.05 was considered statistically significant. For representation, we denoted each fiscal year as a single year; for example, the fiscal year 2008–2009 was denoted as 2009. The study was approved by the ethics committee of the University of Alberta (identifier Pro00097627_AME1). As we used anonymised summary data from the provincial health records, signed informed consent was not required.

Between 2009 and 2020, the prevalence of asthma in Alberta increased from 231 949 to 464 040. This is an ∼1.5 times rise in population-adjusted asthma prevalence (from 6.46% in 2009 to 9.70% in 2020). In contrast, between 2009 and 2020, the prevalence-adjusted average rate of the prescription dispensation of ICS/LABA (0.67% versus 0.73%), ICS (0.37% versus 0.34%), SABA (0.98% versus 0.99%), antibiotics (0.44% versus 0.46%), LAMA (0.12% versus 0.15%) and LTRA (0.23% versus 0.21%) remained low and the signal was relatively stable. There was a marginal increase in the dispensation of OCS during the period (0.14% versus 0.20%). The dispensation of biologic medications for asthma increased by five times (0.006% versus 0.03%). Upon stratification based on age, we observed that there was a decline in ICS/LABA dispensation in the paediatric patient population (0.21% versus 0.17%) versus the adult population (0.89% versus 0.88%). While in adults, there was a decline of ICS (0.36% versus 30%), and SABA (1.18% versus 1.12%), we observed an increase of dispensation of ICS (0.39% versus 0.51%) and SABA (0.55% versus 0.73%) in the paediatric population. The trend in dispensation between 2009 and 2020 for antibiotics, LAMA, LTRA, OCS and biologics remained consistent in these two populations (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Asthma and asthma medication in Alberta, Canada. a) The trend in asthma and dispensation of asthma medications in adult and paediatric patients between 2009 and 2020, b) Correlation heatmap demonstrating asthma incidence and medication dispensation in all asthmatic patients, and separately for adult and paediatric patients between 2009 and 2020. Numbers indicate the correlation coefficient values (Spearman's ρ). ICS: inhaled corticosteroids; LABA: long-acting β₂-adrenergic receptor agonists; SABA: short-acting β₂-adrenergic receptor agonists; LAMA: long-acting muscarinic receptor antagonists; OCS: oral corticosteroids; SAMA: short-acting muscarinic receptor antagonists; LTRA: leukotriene receptor antagonists.

In the correlation analysis between asthma prevalence and medication dispensation, we observed that with increasing asthma prevalence, the average dispensation of ICS (ρ= −0.63, p=0.03), LABA (ρ= −0.76, p=0.006) and SAMA (ρ= −0.78, p=0.004) reduced, while LAMA (ρ=0.73, p=0.01), OCS (ρ=0.97, p<0.001) and biologics (ρ=0.99, p<0.001) increased significantly. Upon age stratification, we found that in addition to ICS, LABA and SAMA, the dispensation of SABA and LTRA also reduced significantly with increasing asthma prevalence (ρ= −0.80 and −0.86, respectively; both p<0.01) in the adult patient population, while OCS and biologics increased. In the paediatric patient population, only ICS/LABA reduced significantly (ρ=−0.92, p<0.001) with increasing asthma prevalence while the dispensation of LAMA (ρ=0.69, p=0.02), OCS (ρ=0.89, p<0.001) and biologics (ρ=0.74, p=0.008) increased significantly (figure 1).

Our study reports, for the first time, the trend in asthma prevalence and dispensation of asthma medication over 12 years (2008–2009 to 2019–2020) using the largest public administrative database for health information in Canada, with the medical records of nearly 4.5 million Albertans over the past 20 years. Our observation of an increasing prevalence of asthma in the province of Alberta was similar to the trend in other provinces and across Canada [7, 19, 20]. Our findings are also similar to a recent USA-based study on the trend of asthma prevalence that showed a closely similar trend in asthma prevalence 10 years apart (7.7% in 2009–2010 and 9.3% in 2018–2019) [21]. We observed that despite the increasing asthma prevalence, dispensation for the majority of the 10 common classes of asthma medications was either inversely correlated with asthma prevalence (ICS, LABA and SAMA) or remained unchanged (ICS/LABA, SABA, antibiotics, SAMA and LTRA). There was also a clear distinction in medication dispensation trends between the adult and paediatric patient populations.

While current guidelines recommend ICS or ICS/LABA as the controller medications for effective asthma management [1, 8], our observation of declining dispensation of ICS and ICS/LABA (in the paediatric population) in the past decade is a strong indicator of poor alignment of guideline-based recommendations. Although LAMA showed a positive correlation with ICS/LABA dispensation, it is not clear from our data why LAMA dispensation showed a positive correlation with asthma (in the adult population), particularly when the guidelines do not recommend LAMA as a monotherapy, and should only be used as add-on therapy with ICS/LABA [1] and has no indication in the paediatric population as per Health Canada recommendations. We also observed a significant negative association between asthma prevalence and the dispensation of ICS and SABA. Globally, the use of SABA is recommended as rescue/reliever therapy if taken with ICS [1]. However, our results of average SABA dispensation do not reciprocate with the results of the recently published SABA Use in Asthma study, where the investigators reported a mean±sd of 3.9±4.4 canisters per year in Alberta [22]. This difference could be due to the cross-sectional nature of the study and a different patient population that was part of the inclusion criteria of that study. We also observed the inappropriate use of other medications (for example, LAMA use in children). The increasing trend of the use of biologics in asthma could be driven by the newly marketed and approved drugs for the management of severe asthma. This trend is quite similar to the US scenario [23], and several factors such as higher income, private insurance and access to a specialist are likely to contribute to the higher use of biologics.

Although our data demonstrate an overwhelming low rate of asthma medication dispensation over a prolonged period, the reasons for this cannot be concluded from our study. However, we believe that these findings could be driven by both physician- and patient-related factors. One plausible physician-related factor for the low dispensation rates is inappropriate billing for asthma. For example, in 2019–2020, 42% of patients who met our inclusion criteria for asthma had no prescriptions filled for asthma-related medications [22]. In addition, we cannot comment on the administrative data as to what the physician prescribed to the patients. Another possible explanation for the low dispensation rates is that the physician prescribed the appropriate medications but the patient did not have the medication dispensed at the pharmacy. Nevertheless, this is also a limitation of using administrative data. However, patient-related factors that are likely to affect compliance include but are not limited to, affordability, lack of asthma education, understanding of the roles of the medications and poor perception of disease severity [17, 24]. The striking nature of our data warrants further understanding as to what factors could have led to such poor dispensation rates.

Our study has some limitations to consider. Firstly, in administrative databases, the diagnosis of asthma could be over- or under-represented due to a lack of information regarding confirmation of the diagnosis with the use of spirometry. Secondly, the database receives information from different sources, such as clinics, pharmacies and diagnostic centres located all over the province, and the possibility of errors cannot be ruled out. Thirdly, heterogeneity among the databases is common due to different algorithms, sources or management of the records, and extraction of appropriate data can be tedious due to different data sources and algorithms used for the purpose.

In summary, using 12-year administrative data, we observed that despite higher asthma prevalence, there was a low dispensation of asthma medications in Alberta, Canada. Our findings have significant clinical and public health implications. Further work is needed to understand the drivers of this low dispensation and its clinical impact.