Introduction:
Hospitalization for acute myocardial infarction (AMI) affects medication adherence in prevalent statin users. Our objective was to estimate the association between changes in statin adherence and all-cause mortality after AMI discharge.
Hypothesis:
Patients who are adherent both pre- and post-AMI have the lowest risk of all-cause mortality.
Methods:
Medicare administrative claims were used to identify AMI hospitalizations in 2008-2010. Patients were ≥66 years old, continuously enrolled ≥360 days pre-AMI with a statin prescription claim, discharged to home/self-care, and survived ≥180 days post-AMI with continuous enrollment. Statin adherence was measured in the 180 days pre- and post-AMI hospitalization using proportion of days covered and categorized as severely nonadherent, moderately nonadherent, and adherent. The exposure was categorical change in statin adherence from pre- to post-AMI (9 categories, see Figure); adherent/adherent was the reference group. Patients were followed for all-cause mortality from 180 days post-discharge for up to 18 months. A multivariable Cox proportional hazards model estimated hazard ratios (HRs).
Results:
Of 101,011 eligible patients, 15% decreased, 20% increased, and 64% did not change statin adherence categories. Compared to patients who were adherent pre- and post-AMI, the adjusted HR (95% confidence intervals [CIs]) for patients who
increased
from severely nonadherent to adherent was 0.93 (95% CI: 0.85-1.02); other increases in adherence had similar HRs (see Figure). Compared to patients who were adherent pre- and post-AMI, the adjusted HR for patients who
decreased
from adherent to severely nonadherent was 1.22 (95% CI: 1.13-1.33); other decreases in adherence had similar HRs.
Conclusions:
Although patients with decreased statin adherence had the worst mortality outcomes, those with increased adherence had similar or better outcomes than continuously adherent patients, showing that, even after an AMI, it is not too late to improve statin adherence.
Opening the black box of the group‐based trajectory modeling process to analyze medication adherence patterns: An example using real‐world statin adherence data
