Objectively measured, but not self-reported, medication adherence independently predicts event-free survival in...

Medication nonadherence is presumed to be related to poor clinical outcomes, yet this relationship rarely has been tested using objective adherence measures in patients with heart failure. Which objective indicators of medication adherence predict clinical outcomes are unknown. The study objective was to determine which indicators of medication adherence are predictors of event-free survival.

Patients (N = 134) with heart failure (69% were male, aged 61 +/- 11 years, 61% with New York Heart Association class III/IV heart disease) were enrolled in this 6-month longitudinal study. Adherence was measured using two measures: 1) an objective measure, the Medication Event Monitoring System (MEMS); and 2) self-reported adherence (Medical Outcomes Studies Specific Adherence Scale). Three indicators of adherence were assessed by MEMS: 1) dose-count, percentage of prescribed doses taken; 2) dose-days, percentage of days correct number of doses taken; and 3) dose-time, percentage of doses taken on schedule. Events (emergency department visits, rehospitalization, and mortality) were obtained by patient/family interview and hospital databases.

In Cox regression, two of the three MEMS indicators, dose-count and dose-day, predicted event-free survival before and after controlling for age, gender, ejection fraction, New York Heart Association class, angiotensin-converting enzyme inhibitor use, and beta-blocker use (P = .004, P = .008, and P = .224, respectively). Self-report adherence did not predict outcomes (P = .402).

Dose-count and dose-day predicted event-free survival. Neither dose-time nor self-reported adherence predicted outcomes. Health care providers should assess specific behaviors related to medication taking rather than a global patient self-assessment of patient adherence.