THE UNIVERSITY OF BRITISH COLUMBIA
BACKGROUND: Much of the evidence on the early use of inhaled controllers after severe asthma exacerbations is about their short-term benefit, leaving a gap in evidence on their longer-term outcomes. METHODS: We used administrative health data from British Columbia, Canada (2001-2012) to evaluate readmission rate (primary outcome), adherence to controller medications, and use of reliever medications associated with different inhaled controller treatments as an add-on to systemic corticosteroids (SCS) over one-year following discharge from an asthma-related admission in individuals 12-55 years of age. Exposure was assessed in the 60 days after discharge, and categorized as monotherapy with SCS (SCS-only) versus SCS plus inhaled controller therapy (SCS + inhaler); the latter was further divided into SCS + inhaled corticosteroid (SCS + ICS) and SCS + ICS and long-acting beta agonists (SCS + ICS/LABA). Propensity score-adjusted regression models were used to estimate relative rates (RR) of outcomes across exposure groups. RESULTS: The final cohort included 2,272 post-discharge periods (43.0% SCS-only, 26.9% SCS + ICS, and 30.1% SCS + ICS/LABA). Readmission rate was significantly lower in the SCS + inhaler versus SCS-only (RR = 0.74 [95%CI 0.59-0.93]), but similar between SCS + ICS and SCS + ICS/LABA (RR = 0.78 [95%CI 0.59-1.04]). Long-term adherence, defined as medication possession ratio, to controller medications was 83% higher in SCS + inhaler than SCS-only, and 64% higher in SCS + ICS/LABA than in SCS + ICS. The use of reliever medications was similar across exposure groups. CONCLUSION: Early initiation of inhaled controllers after discharge from an asthma-related hospitalization was associated with significantly better long-term adherence to controller medications as well as reduced rate of readmissions. Combination therapy with ICS/LABA seems to be at least as effective as mono-therapy with ICS in reducing the risk of readmission, with the added benefit of better long-term adherence.