Relationship Between Medication Adherence and Post -- MI Outcomes
Relationship Between Medication Adherence and Post -- MI Outcomes
The proportion of patients in the full and usual coverage groups who were fully and nonadherent to the study medications is presented in online Appendix A (top panel). Table I shows the baseline characteristics of all patients randomized to the full and usual coverage arms, stratified by study medication and achieved adherence. Age, gender, comorbidity scores, and copayment were similar across groups. Nonadherent patients in the full coverage cohort were slightly less likely to have used cardiovascular medications before their MI, less likely to have undergone revascularization during their index MI admission, and more likely to have had comorbid conditions than adherent patients. Full coverage patients who were and were not adherent to their prescribed statin had an average adherence (measured by PDC) of 92% and 48%, respectively, as compared with 54% among controls (Table I). Similar differences were observed for ACEI/ARBs and β-blockers.
(Enlarge Image)
Supplementary Figure 1.
Proportion of patients in the full and usual coverage groups according to level of adherence achieved stratified by study medication. The top panel classifies patients as fully or nonadherent based on a proportion of days covered (PDC) of 80%. The bottom panel classifies patients as full, partially, and nonadherent based on PDC of ≥80%, 60% to 79%, and <60%, respectively.
As previously reported, eliminating copayments did not significantly reduce the trial's primary outcome, first major vascular event, or revascularization (hazard ratio [HR] 0.93, 95% CI 0.82–1.04, P = .21), in the trial population. However, when stratified by achieved adherence, patients with ≥80% adherence to each of the study medications were significant less likely than controls to experience a major vascular event or undergo revascularization ( Figure 1). By contrast, nonadherent patients in the full coverage cohort had events rates comparable with those among controls (P < .01 for all interaction terms between adherent and nonadherent patients). For example, full coverage patients who were adherent had a 24% lower hazard of event-free survival than controls (HR 0.76, 95% CI 0.63–0.92, P < .01) whereas nonadherent full coverage patients had no such advantage (HR 1.01, 95% CI 0.86–1.19, P = .90). Similar significant reductions were seen in the other 2 targeted drug classes. The same was true for patients who were adherent to all 2 group combinations and to all 3 of the study medications when considered together. The results were virtually identical in models that adjusted for markers of health-seeking behavior (Table II).
(Enlarge Image)
Figure 1.
Adjusted HR (95% CI) of first major vascular event or revascularization by adherence level achieved among patients in the full coverage study arm compared with patients randomized to the usual coverage. Each point represents the HR in the adherence stratum of the full coverage group compared to usual coverage. Also shown are event rates per 100 person-years. Data adjusted for age, gender, and comorbidity score as well as the cluster and block randomized study design. The P values represent the results of a likelihood ratio test for the null hypothesis that the impact of full coverage for adherent and nonadherent subjects did not differ relative to usual care.
After categorizing adherence into 3 groups, the proportion of patients in the full and usual coverage groups who were fully, partially, and nonadherent to the study medications is presented in online Appendix A (bottom panel). Partially adherent patients in the full coverage cohort (PDC 60%-79%) had no significant reduction in clinical outcomes for any of the drugs evaluated compared with controls (Figure 2), despite achieving greater medication adherence (online Appendix B). For example, full coverage patients who were partially adherent to β-blockers had an average adherence of 71% compared with 49% among the entire control group but had an equivalent risk of major vascular events or revascularization than controls (HR 1.01, 95% CI 0.79–1.29, P = .96, interaction P-value between adherent and nonadherent patients <.01). Reclassifying partially adherent patients as those who achieved a PDC between 40% and 79% yielded very similar results (Table III).
(Enlarge Image)
Figure 2.
Cumulative incidence of major vascular event or revascularization among full coverage patients stratified by level of adherence to each of the 3 study medications in the 6 months after randomization. Full, partial, and nonadherence were defined based on PDC of ≥80%, 60% to 79%, and <60%, respectively.
To evaluate whether the impact of adherence was similar on other outcomes, we repeated our analyses using rates of major vascular events, a prespecified secondary outcome, and found very similar results (online Appendix C). Patients in the full coverage group who were fully adherent to the study medications had a 25% to 30% lower hazard of this outcome compared with controls, whereas partially and nonadherent patients derived no benefit from therapy.
We assessed the incremental impact of adherence to each additional class of post-MI secondary prevention by categorizing patients in the full coverage arm who filled prescriptions for all 3 study medications based upon the number of these classes to which they were fully adherent. The benefit of therapy increased linearly with the number of medications to which patients were adherent (online Appendix D, P < .001 for linear trend). However, only full coverage patients who were adherent to all 3 classes had outcome rates significantly different than controls (HR 0.65, 95% CI 0.48–0.86).
(Enlarge Image)
Supplementary Figure 2.
Adjusted hazard ratio (95% CI) of first major vascular event or revascularization for patients who filled all 3 study medications by the number of medications to which they were adherent during the 6 months after randomization. Each point represents the hazard ratio in the specific adherence stratum of the full coverage group compared to usual coverage. Data are adjusted for age, gender, and comorbidity score as well as the cluster and block randomized study design. The P values represent the results of a test for trend.
The results of our analysis identifying the optimal cut point to distinguish adherent from nonadherent patients are presented in online Appendix E. Although adherence in each quintile differed from drug to drug, across drug classes, adherence was associated with significant reductions in clinical outcomes beginning in those quintiles corresponding to a mean PDC of 70% and 80%.
(Enlarge Image)
Supplementary Figure 3.
Adjusted hazard ratio (95% CI) of first major vascular event or revascularization by quintile of adherence level achieved. Each point represents the hazard ratio in the specific adherence quintile of the full coverage group compared to usual coverage. Also shown is the mean level of adherence achieved in that stratum. Data are adjusted for age, gender, and comorbidity score as well as the cluster and block randomized study design.
Finally, although our primary analyses evaluated all outcomes occurring after randomization, we repeated these analyses excluding events during the 6 months after randomization. We found a similar pattern, except that ACEI/ARB adherence was no longer associated with a reduction in major vascular event or revascularization (online Appendix F).
Results
The proportion of patients in the full and usual coverage groups who were fully and nonadherent to the study medications is presented in online Appendix A (top panel). Table I shows the baseline characteristics of all patients randomized to the full and usual coverage arms, stratified by study medication and achieved adherence. Age, gender, comorbidity scores, and copayment were similar across groups. Nonadherent patients in the full coverage cohort were slightly less likely to have used cardiovascular medications before their MI, less likely to have undergone revascularization during their index MI admission, and more likely to have had comorbid conditions than adherent patients. Full coverage patients who were and were not adherent to their prescribed statin had an average adherence (measured by PDC) of 92% and 48%, respectively, as compared with 54% among controls (Table I). Similar differences were observed for ACEI/ARBs and β-blockers.
(Enlarge Image)
Supplementary Figure 1.
Proportion of patients in the full and usual coverage groups according to level of adherence achieved stratified by study medication. The top panel classifies patients as fully or nonadherent based on a proportion of days covered (PDC) of 80%. The bottom panel classifies patients as full, partially, and nonadherent based on PDC of ≥80%, 60% to 79%, and <60%, respectively.
Impact of Achieving Full Adherence on Clinical Outcomes
As previously reported, eliminating copayments did not significantly reduce the trial's primary outcome, first major vascular event, or revascularization (hazard ratio [HR] 0.93, 95% CI 0.82–1.04, P = .21), in the trial population. However, when stratified by achieved adherence, patients with ≥80% adherence to each of the study medications were significant less likely than controls to experience a major vascular event or undergo revascularization ( Figure 1). By contrast, nonadherent patients in the full coverage cohort had events rates comparable with those among controls (P < .01 for all interaction terms between adherent and nonadherent patients). For example, full coverage patients who were adherent had a 24% lower hazard of event-free survival than controls (HR 0.76, 95% CI 0.63–0.92, P < .01) whereas nonadherent full coverage patients had no such advantage (HR 1.01, 95% CI 0.86–1.19, P = .90). Similar significant reductions were seen in the other 2 targeted drug classes. The same was true for patients who were adherent to all 2 group combinations and to all 3 of the study medications when considered together. The results were virtually identical in models that adjusted for markers of health-seeking behavior (Table II).
(Enlarge Image)
Figure 1.
Adjusted HR (95% CI) of first major vascular event or revascularization by adherence level achieved among patients in the full coverage study arm compared with patients randomized to the usual coverage. Each point represents the HR in the adherence stratum of the full coverage group compared to usual coverage. Also shown are event rates per 100 person-years. Data adjusted for age, gender, and comorbidity score as well as the cluster and block randomized study design. The P values represent the results of a likelihood ratio test for the null hypothesis that the impact of full coverage for adherent and nonadherent subjects did not differ relative to usual care.
Impact of Lower Levels of Adherence on Clinical Outcomes
After categorizing adherence into 3 groups, the proportion of patients in the full and usual coverage groups who were fully, partially, and nonadherent to the study medications is presented in online Appendix A (bottom panel). Partially adherent patients in the full coverage cohort (PDC 60%-79%) had no significant reduction in clinical outcomes for any of the drugs evaluated compared with controls (Figure 2), despite achieving greater medication adherence (online Appendix B). For example, full coverage patients who were partially adherent to β-blockers had an average adherence of 71% compared with 49% among the entire control group but had an equivalent risk of major vascular events or revascularization than controls (HR 1.01, 95% CI 0.79–1.29, P = .96, interaction P-value between adherent and nonadherent patients <.01). Reclassifying partially adherent patients as those who achieved a PDC between 40% and 79% yielded very similar results (Table III).
(Enlarge Image)
Figure 2.
Cumulative incidence of major vascular event or revascularization among full coverage patients stratified by level of adherence to each of the 3 study medications in the 6 months after randomization. Full, partial, and nonadherence were defined based on PDC of ≥80%, 60% to 79%, and <60%, respectively.
Sensitivity Analyses
To evaluate whether the impact of adherence was similar on other outcomes, we repeated our analyses using rates of major vascular events, a prespecified secondary outcome, and found very similar results (online Appendix C). Patients in the full coverage group who were fully adherent to the study medications had a 25% to 30% lower hazard of this outcome compared with controls, whereas partially and nonadherent patients derived no benefit from therapy.
We assessed the incremental impact of adherence to each additional class of post-MI secondary prevention by categorizing patients in the full coverage arm who filled prescriptions for all 3 study medications based upon the number of these classes to which they were fully adherent. The benefit of therapy increased linearly with the number of medications to which patients were adherent (online Appendix D, P < .001 for linear trend). However, only full coverage patients who were adherent to all 3 classes had outcome rates significantly different than controls (HR 0.65, 95% CI 0.48–0.86).
(Enlarge Image)
Supplementary Figure 2.
Adjusted hazard ratio (95% CI) of first major vascular event or revascularization for patients who filled all 3 study medications by the number of medications to which they were adherent during the 6 months after randomization. Each point represents the hazard ratio in the specific adherence stratum of the full coverage group compared to usual coverage. Data are adjusted for age, gender, and comorbidity score as well as the cluster and block randomized study design. The P values represent the results of a test for trend.
The results of our analysis identifying the optimal cut point to distinguish adherent from nonadherent patients are presented in online Appendix E. Although adherence in each quintile differed from drug to drug, across drug classes, adherence was associated with significant reductions in clinical outcomes beginning in those quintiles corresponding to a mean PDC of 70% and 80%.
(Enlarge Image)
Supplementary Figure 3.
Adjusted hazard ratio (95% CI) of first major vascular event or revascularization by quintile of adherence level achieved. Each point represents the hazard ratio in the specific adherence quintile of the full coverage group compared to usual coverage. Also shown is the mean level of adherence achieved in that stratum. Data are adjusted for age, gender, and comorbidity score as well as the cluster and block randomized study design.
Finally, although our primary analyses evaluated all outcomes occurring after randomization, we repeated these analyses excluding events during the 6 months after randomization. We found a similar pattern, except that ACEI/ARB adherence was no longer associated with a reduction in major vascular event or revascularization (online Appendix F).
Source...