Health outcomes of medication adherence and persistence among elderly ACE/ARB users

Medication-taking behavior is often described by measures of persistence and adherence. Most prior studies have not distinguished patients who use medications regularly and then discontinue treatment from patients who continued to use medications, but use them nonadherently. This study examined the differential impact of adherence and persistence on health outcomes among elderly using angiotensin-converting enzyme (ACE) inhibitors or angiotensin-II receptor antagonists (ARBs). The prescription refill patterns of 19,892 ACE/ARB users in 2003 were assessed using data from Pennsylvania's Pharmaceutical Assistance Contract for the Elderly (PACE). Adherence was evaluated using the Proportion of Days covered (PDC), computed as the percentage of days for which any ARB/ACE medication was available. Multivariate logistic regression was used to examine 2004 hospitalization risk across adherence/persistence categories. Cox regression was employed to examine associations between adherence/persistence and 2004-2005 mortality. Covariates included demographics, Charlson comorbidity, average PDC level of other chronic therapeutic classes, and the overall length of time ACE/ARB medications had been used. Compared to users who were both persistent and adherent (PDC≥80%), nonpersistent-adherent users were significantly more likely to be hospitalized (OR=1.40, P<0.0001). Odds ratios for nonpersistence-nonadherence and persistence-nonadherence were 1.18 (P=0.09) and 1.10 (P=0.05), respectively. For mortality, nonpersistent-adherent users (HR=1.44, P<0.0001) and nonpersistent-nonadherent users (HR=1.28, P=0.03) were significantly more likely to die than persistent-adherent users. The hazard ratio for persistent-nonadherent users was 1.09 (P=0.22). These results suggest that the consequences of ARB/ACE nonpersistence may be more severe than nonadherence, and highlight the importance of interventions aimed at improving cardiovascular medication persistence.