Clustering of adherence variability metrics and clinical outcomes in asthma

Background: Poor adherence with asthma preventer medication contributes to worse symptom control and increased exacerbation risk. Adherence is often expressed as the mean proportion of prescribed puffs taken over a period. New metrics may capture individual variability patterns linked with distinct clinical outcomes. Aims and Objectives: With electronically-recorded medication data from a 6-month cluster randomised trial (Foster JM et al, 2014. JACI 134:1260-68), we examined novel adherence variability metrics, and their association with symptom control (Asthma Control Test [ACT] score) and exacerbation risk. Methods: Adherence metrics were calculated from months 0-2. They included time adherence area under curve (T-AUC), reflecting gaps in adherence over time, and standard deviation of the %prescribed puffs/day taken (SD-PT). Dominant metrics from factor analysis were used for hierarchical clustering. We compared outcomes over months 2-6, and exacerbation risk over the whole study period. Results: Two factors explained u003e94% of total variance, primarily driven by T-AUC and SD-PT. Two main patient clusters were identified: Cluster 1 (n=61) had better T-AUC than Cluster 2, and had less decline in ACT over months 2-6 (median (range) 1 (-16, 11) vs -2 (-9, 4); p=0.002). Cluster 2 (n=21) trended towards shorter time to first exacerbation (Mann-Whitney p=0.053) and greater exacerbation risk (Cox coefficient (95% CI): 2.25 (0.8-6.3); p=0.126). Conclusions: Novel metrics showed that low time adherence was associated with greater risk of decline in symptom control and a trend to higher exacerbation risk. Adherence patterns may exhibit ‘memory’ relevant to future clinical status, warranting validation in a larger dataset.