Introduction: Despite the availability of a number of life-prolonging therapies, patients with heart failure (HF) still experience high rates of death and hospitalization. Medication noncompliance is a determinant of such events.

Purpose: To understand the relationship of medication compliance and the temporal affect of compliance in HF patients and to further explore whether patients on multiple medications are at greater risk of such events. Our objectives were threefold: 1) to understand the temporal relationships between medication compliance for HF medications and the risk of an event; 2) to determine how the use of multiple HF medications contributed to the number of events, and 3) to determine whether medication compliance with a HF medication correlates with medication compliance with other drugs, within an individual.

Methods: We followed, for one year, 134 patients who were part of a RCT who were admitted to hospital with HF. We documented death and readmission in these patients. We constructed a longitudinal model (GEE) on determinants of compliance for each drug class. We categorized compliance between drugs, with a patient as having positive correlation ( > 0.30), low correlation (between > -0.3 and < 0.3) and negatively correlated ( - < 0.3), and low correlation ( < 0.3). We studied the impact of correlated compliance across different classes of drugs on HF events using an odds ratio to compare drugs that were positively correlated to those with low correlation, and negatively correlated to those with low correlation.

Results: In this prospective longitudinal cohort we found that previous compliance was a significant positively associated determinant of future compliance for ACE-Inhibitors and diuretics. Patients using beta-blockers, male gender, lower left ventricular fraction (LVEF) and increased knowledge about HF had significantly higher compliance. Among digoxin users, patients with co-morbid disease had higher compliance. Patients on multiple medications (digoxin and beta-blocker) had an odds ratio (OR) = 3.0; p = 0.005 of having an event and patients on beta-blockers and digoxin had an OR = 2.48; p = 0.081.

Conclusion: Knowing that previous compliance can predict future compliance can help us target patients early before they have an event. Early intervention in patients on multiple medications may be warranted due to the increased risk of having an event.