Asthma-Related Health Outcomes Associated with Short-Acting β₂-Agonist Inhaler Use: An Observational UK Study as Part of the SABINA Global Program

The Clinical Practice Research Datalink (CPRD-GOLD) currently encompasses primary care data for over 40 million patients across the UK [22], and is representative of the population with respect to age, sex, and body mass index (BMI) [23]. Hospital admissions and outpatient consultation data were obtained from the Hospital Episode Statistics (HES) database. HES only covers hospitals in England; therefore, only around 60% of CPRD practices have patient-level linkage to HES, the Office for National Statistics mortality data, and the Index of Multiple Deprivation (IMD) socioeconomic data.

We conducted a retrospective, longitudinal, open-cohort study; asthma patients were identified using validated diagnostic codes [24]. Patients had to have an asthma diagnostic code within 3 years of index date and could either be a new diagnosis (first code occurred in the baseline year, which was 12 months prior to index date) or prevalent disease (≥ 2 validated asthma codes or asthma medication prescriptions > 1 year apart). Index date was set as the latest date of any of the following: asthma diagnosis, 12th birthday, start of study period (April 1, 2007), 1 year after the general practitioner (GP) practice began recording research quality data (CPRD quality control), or 1 year after their continuous CPRD practice registration date (Fig. S1). Patients were censored at the earliest date of death, the end of study period (December 31, 2017), the last CPRD data collection date, or the date transferred out of a CPRD practice. Patients managed on the highest treatment step [continuous or frequent oral corticosteroid (OCS) use] were excluded as these patients are often managed in both primary and secondary care, and their CPRD records may not be complete. Moreover, these patients are unique and regularly use OCS as part of their management; they also constitute approximately < 1% of the cohort. All patients who met the eligibility criteria were included in the prescription analysis. Only patients with CPRD–HES-linked records contributed to the analysis of asthma outcomes (as the outcome measurements required hospital data).

Exacerbations were defined as previously described [1, 25, 26], as asthma worsening that necessitated a short course of OCS (GP-managed exacerbation), an Accident and Emergency (A&E) department visit for asthma, a hospital admission or death secondary to asthma. A course of OCS was defined as a prescription [not during an annual asthma review or on days treated for other steroid responsive diseases (see Appendix for list of other steroid responsive diseases)] of a total amount ≤ 300 mg. Primary care consultations for asthma were identified in CPRD as the date of a consultation that occurred on the same day as an asthma diagnostic code. Hospital consultations were identified in HES as the date of a respiratory outpatient clinic. Of note, mortality events in this asthma cohort were too few to be analyzed separately.

The following variables were recorded in the baseline year (12 months before the study entry date): ‘SABA inhaler use’ (number of canisters prescribed), ‘ICS use’ (maximum ICS dose prescribed), proportion of days covered (PDC) by ICS prescriptions, number of exacerbations, and treatment step. SABA inhaler use was categorized as ‘low use’ (0–2 SABA canisters/year) or ‘high use’ (≥ 3 SABA canisters/year). Guideline recommendations for mild asthma in the UK are such that ≥ 3 puffs per week is considered inadequate control [27]. In the UK, SABA canisters contain 100 or 200 puffs and some patients keep a ‘spare’ SABA inhaler; therefore, a cut-off of ≥ 3 canisters was set. SABA inhaler use was also used as a categorical variable (canisters per baseline year: 0–1, 2, 3–6, 7–12, and ≥ 13). PDC was used as a measure of adherence [28] and was calculated as the proportion of the baseline year that had days covered by ICS prescriptions. Treatment step was assigned based on correlation between prescriptions recorded and treatment proposed in the 2016 Scottish Intercollegiate Guidelines Network/British Thoracic Society (SIGN/BTS) guidelines [27]; patients were thus classified into steps 1–5 according to their highest ICS prescription dose in the baseline year [step 1 = no ICS, step 2 = low-dose ICS (400–799 µg, beclomethasone dipropionate equivalent), step 3 = low-dose ICS + long-acting β₂-agonist, step 4 = medium-dose ICS (800–1599 µg) ± additional therapies, step 5 = high-dose ICS (> 1600 µg) ± additional therapies] [25]. A history of atopy, gastroesophageal reflux, smoking, pneumonia, anxiety, and depression was recorded using appropriate diagnostic codes (available upon request). Chronic obstructive pulmonary disease (COPD) was identified by validated diagnostic codes, a smoking history, and age > 35 years, while non-smokers with COPD were identified using validated diagnostic codes and age > 35 years [29, 30].

Prescription analysis described the number and proportion of patients in each SABA category in the baseline year by treatment step. Outcomes analysis addressed the association between SABA inhaler use and exacerbations, primary care consultations, and hospital consultations. The mean number of exacerbations during the first year of follow-up and the incidence rate of exacerbations during total study follow-up were estimated according to asthma severity designated by treatment step (grouped as BTS steps 1/2 for mild asthma and steps 3–5 for moderate-to-severe asthma) and SABA inhaler use. To estimate the association between SABA inhaler use and exacerbations, the Cox proportional hazard regression model was used with Andersen-Gill extension to account for repeat outcomes. To estimate the association between SABA inhaler use and primary care consultations, a mixed-effects negative binomial regression model was used; random effects accounted for clustering by patient. To estimate the association between SABA inhaler use and hospital consultations, zero-inflated negative binomial regression, with age as the zero-inflated variable, was used. All regression models used complete case analysis, were stratified by BTS step, and adjusted for age, gender, socioeconomic status (defined using the 2015 Index of Multiple Deprivation [31]), BMI, smoking history, exacerbation history, ICS use, PDC, and comorbidities. Sensitivity analyses were conducted including SABA inhaler use as a categorical variable, excluding patients with a COPD co-diagnosis and excluding patients who were not prescribed a SABA inhaler in the baseline year, and for BTS step 2 patients using ICS. All analyses were performed using STATA 14.2.

The protocol for this research was approved by the Independent Scientific Advisory Committee (ISAC) for the Medicines and Healthcare products Regulatory Agency (MHRA) Database Research (protocol number 18_080R) and the approved protocol was made available to the journal and reviewers during peer review. Generic ethical approval for observational research using CPRD with approval from ISAC was granted by a Health Research Authority (HRA) Research Ethics Committee (East Midlands–Derby, REC Reference No. 05/MRE04/87). Linked pseudonymized data were provided for this study by CPRD. Data were linked by National Health Service (NHS) Digital, the statutory trusted third party for linking data, using identifiable data held only by NHS Digital. Select practices consent to this process at a practice level, with individual patients having the right to opt-out.

Of the 709,268 asthma patients identified in the CPRD from April 1, 2007 to December 31, 2017, a total of 574,913 patients were eligible for the prescription analysis (Fig. S1). Of the 134,355 patients excluded, 26,627 belonged to BTS step 6 or were not following BTS guidelines. In step 1, 12% of the patients were high SABA users (Fig. 1). However, the proportion of high SABA users increased with increasing treatment step (step 2 = 44%, step 3 = 54%, step 4 = 57%); over two-thirds of the patients in step 5 were high SABA users (68%). Of note, across BTS steps, a proportion of asthma patients did not receive any SABA inhaler prescriptions in the baseline year (step 1 = 56%, step 2 = 9.6%, step 3 = 14%, step 4 = 13%, step 5 = 9.8%) (Fig. S2). Of those in step 1 not prescribed a SABA inhaler in the baseline year (n = 151,760), 91% in the year before or after the baseline year had either been prescribed asthma medication (n = 105,741) or had attended a GP asthma consultation (n = 31,710). Thus, only 7.3% (n = 14,309) of patients in BTS step 1 did not have any asthma treatment in the year before or after the baseline year.

A total of 336,412 patients were eligible for the outcomes cohort (Fig. S1); 222,135 (66%) were receiving BTS steps 1/2 treatment, and 114,277 (34%) received steps 3–5 treatment (Table 1). Approximately 60% were female across BTS steps; the majority of patients were aged 18–55 years, 59% in the BTS steps 1/2 cohort and 51% in the BTS steps 3–5 cohort. Among all asthma patients, only 61,428 (18%) had ≥ 75% of days covered with ICS prescriptions, while 118,237 (53%) had no ICS prescriptions. High SABA use occurred among 26% and 57% of BTS steps 1/2 and 3–5 asthma patients, respectively. The proportions of asthma patients at BTS steps 1/2 and 3–5 who experienced ≥ 1 exacerbation during the baseline year were 7% and 24%, respectively.

The mean number of exacerbations during the first year of follow-up was 1.7–2.2 times higher among high SABA users than among low SABA users across treatment steps (Fig. 2). In all patients, the rate of exacerbations was almost double in high SABA users compared with low SABA users [BTS steps 1/2: low SABA use = 1.66 exacerbations per 10 person-years, 95% confidence interval (CI) 1.62–1.69 and high SABA use = 2.96 exacerbations per 10 person-years, 95% CI 2.87–3.05; BTS steps 3–5: low SABA use = 4.15 exacerbations per 10 person-years, 95% CI 4.04–4.26 and high SABA use = 7.83 exacerbations per 10 person-years, 95% CI 7.70–7.96; Table 2]. The increase in exacerbation rates in high SABA users compared with low SABA users remained regardless of where the exacerbation was treated (by a GP, in an A&E department, or in a hospital; Table 3).

Patients at BTS steps 1/2 with high SABA use had a 20% increased rate of exacerbations compared with low SABA users, even after adjusting for ICS use, age, gender, BMI, socioeconomic status, smoking, recent exacerbations and history of pneumonia, atopy, reflux, COPD, anxiety, and depression [adjusted hazard ratio (HR) = 1.20, 95% CI 1.16–1.24]. Patients at BTS steps 3–5 with high SABA use had a 24% increased rate of exacerbations (adjusted HR = 1.24, 95% CI 1.20–1.28) (Table 4).

For BTS steps 1/2 patients, a significant dose–response effect of increased exacerbation risk was found with increasing SABA inhaler use once patients used ≥ 3 SABA canisters/year [SABA canisters/year, HR (95% CI) 2 = 1.00 (0.96–1.05); 3–6 = 1.14 (1.09–1.19); 7–12 = 1.26 (1.18–1.33); ≥ 13 = 1.52 (1.37–1.68); Table 5]. In BTS steps 3–5 patients, the dose–response effect occurred at ≥ 2 SABA canisters/year [SABA canisters/year, HR (95% CI) 2 = 1.09 (1.04–1.14); 3–6 = 1.19 (1.14–1.23); 7–12 = 1.37 (1.31–1.43); ≥ 13 = 1.55 (1.47–1.64); Table 5].

Estimates were not significantly affected by the exclusion of patients with COPD or zero-SABA inhaler users in the sensitivity analyses (Tables S1–S3). Similar results were seen among patients at BTS steps 1/2 who were prescribed ICS (Table S4). Notably, exacerbation risk was strongly associated with the highest adherence tertile in this subset of patients.

The adjusted incidence rate ratio (IRR) for primary care consultations comparing high SABA users with low SABA users was 1.24 (95% CI 1.23–1.26) among BTS step 1/2 patients and 1.13 (95% CI 1.11–1.15) among patients at BTS steps 3–5 (Table 6). For hospital consultations, the adjusted IRR was 1.19 (95% CI 1.12–1.27) among BTS steps 1/2 patients and 1.19 (95% CI 1.13–1.26) among BTS steps 3–5 patients, respectively (Table 7).