Modeling the Impact of Increased Adherence to Asthma Therapy

Background: Nonadherence to medications occurs in up to 70% of patients with asthma. The effect of improving adherence is not well quantified. We developed a mathematical model with which to assess the population-level effects of improving medication prescribing and adherence for asthma. Methods: A mathematical model, calibrated to clinical trial data from the U.S. NHLBI-funded SOCS trial and validated using data from the NHLBI SLIC trial, was used to model the effects of increased prescribing and adherence to asthma controllers. The simulated population consisted of 4,930 individuals with asthma, derived from a sample the National Asthma Survey. Main outcomes were controller use, reliever use, unscheduled doctor visits, emergency department (ED) visits, and hospitalizations. Results: For the calibration, simulated outcomes agreed closely with SOCS trial outcomes, with treatment failure hazard ratios [95% confidence interval] of 0.92 [0.58–1.26], 0.97 [0.49–1.45], and 1.01 [0–1.87] for simulation vs. trial in the in placebo, salmeterol, and triamcinolone arms, respectively. For validation, simulated outcomes predicted mid- and end-point treatment failure rates, hazard ratios 1.21 [0.08–2.34] and 0.83 [0.60–1.07], respectively, for patients treated with salmeterol/triamcinolone during the first half of the SLIC study and salmeterol monotherapy during the second half. The model performed less well for patients treated with salmeterol/triamcinolone during the entire study duration, with mid- and end-point hazard ratios 0.83 [0.00–2.12] and 0.37 [0.10–0.65], respectively. Simulation of optimal adherence and prescribing indicated that closing adherence and prescription gaps could prevent as many as nine million unscheduled doctor visits, four million emergency department visits, and one million asthma-related hospitalizations each year in the U.S. Conclusions: Improvements in medication adherence and prescribing could have a substantial impact on asthma morbidity and healthcare utilization.