. Trajectories of systolic blood pressure (BP) over time (N = 10 447).
Medication adherence and therapy intensification across the 3 BP control groups (for a definition of the 3 groups, see the “Dependent or Outcome Variable” subsection of the “Methods” section). PDC indicates proportion of days covered. P<.001 for all comparisons.
Ho PM, Magid DJ, Shetterly SM, Olson KL, Peterson PN, Masoudi FA, Rumsfeld JS. Importance of Therapy Intensification and Medication Nonadherence for Blood Pressure Control in Patients With Coronary Disease. Arch Intern Med. 2008;168(3):271–276. doi:10.1001/archinternmed.2007.72
Despite the importance of blood pressure (BP) control in secondary prevention, a significant proportion of patients with coronary disease have uncontrolled BP.
This retrospective cohort study of patients with coronary disease (N = 10 447) evaluated the impact of medication nonadherence and therapy intensification on reaching target BP goals. Medication adherence was calculated as the proportion of days covered for filled prescriptions of antihypertensive medications. Therapy intensification included dosage increase or increase in number of antihypertensive medications. The primary outcome was uncontrolled systolic BP (SBP) over time, using a latent class model that incorporated longitudinal SBP data and assigned patients to SBP trajectory groups. Multivariable regression evaluated the association between medication nonadherence (ie, proportion of days covered, <0.80) and therapy intensification with SBP control over time, with adjustment for demographics and clinical characteristics.
Three SBP trajectory groups were identified: (1) patients with BP that remained controlled (ie, SBP, ≤140 mm Hg) over time (n = 9114 [87.2%]); (2) patients with high BP that became controlled (n = 779 [7.5%]); and (3) patients with BP that remained high over time (n = 554 [5.3%]). In multivariable analyses, therapy intensification (odds ratio, 1.31; 95% confidence interval, 1.01-1.70) and medication nonadherence (odds ratio, 1.73; 95% confidence interval, 1.34-2.24) were associated with uncontrolled BP compared with high SBP that became controlled over time.
These findings suggest that medication nonadherence can help explain why BP levels remained elevated despite intensification of antihypertensive medications. Successful BP control is seen with a combination of intensification and adherence, suggesting that therapy intensification must be coupled with interventions to enhance medication adherence.
Coronary artery disease (CAD) is common and affects more than 13 million patients in the United States.1 Uncontrolled blood pressure (BP) among patients with CAD is associated with increased risk of recurrent cardiovascular events, including death, nonfatal myocardial infarction, and nonfatal stroke.2 The guidelines of the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure currently recommend a treatment goal of less than 140/90 mm Hg among patients with CAD, although lower BP targets may still be beneficial.3 Despite the importance of BP control, fewer than 50% of patients with CAD in clinical practice have their BP at levels recommended by national guidelines.4,5
Previous studies have focused mainly on patient characteristics associated with uncontrolled BP in the CAD population, such as diabetes or left ventricular hypertrophy.5 Although the recognition of patient factors associated with uncontrolled BP is important, such recognition does not directly support the changes in care delivery necessary to increase the proportion of patients with CAD who have controlled BP. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure outlines the importance of therapy intensification and medication adherence for achieving BP control.6 However, relatively little is known about the significance of these factors in achieving BP control in clinical practice. Moreover, previous studies have not assessed the impact of both medication nonadherence and therapy intensification on the achievement of BP goals among patients with CAD.
Accordingly, the objectives of this study were to assess the association between antihypertensive medication nonadherence and therapy intensification with BP control among patients with known CAD in a large managed care organization. Specifically, we assessed BP control rates based on serial BP measurements over time. Then, we evaluated the association between antihypertensive medication nonadherence and therapy intensification with BP control. The findings of this study may have important implications for evaluating BP management among patients with CAD and designing interventions to improve BP control and patient outcomes.
Kaiser Permanente of Colorado (KPCO), Aurora, is an integrated, nonprofit managed care organization that provides medical services to more than 415 000 members in the Denver and Boulder, Colorado, metropolitan areas. Patients are initially identified with an algorithm that is applied to automated databases consisting of hospitalization records and outpatient diagnoses. The diagnosis of CAD is validated, based on prior myocardial infarction, percutaneous coronary intervention, or coronary artery bypass graft surgery, by chart review before inclusion in the registry.
The KPCO has an active clinical pharmacy specialist–managed, physician-monitored disease management service called CPCRS that assists primary care providers and cardiologists with implementation and long-term management of evidence-based treatment for patients with CAD.7 Lipid levels and BP are managed by the clinical pharmacy specialists as part of the service. Almost all of the patients (95.3%) included in the current study were enrolled in the CPCRS program at some point during the observation period and had active management of their BP by a clinical pharmacy specialist in addition to usual care.
We conducted a retrospective cohort study of patients in the KPCO CAD registry. Patients were included if they had at least 12 months of continuous follow-up after entry into the registry and a minimum of 4 outpatient BP measurements recorded in the computerized medical record, which was available beginning in July 2000. A total of 10 447 patients met the inclusion criteria for the current study.
The automated pharmacy records at KPCO include all medications dispensed at each outpatient facility. The majority of patients (>98%) have prescription drug coverage for a nominal copayment. The nominal copayment and the location of pharmacies at the same site as the clinic offices serve as incentives for patients to fill their prescriptions within the system.
Medication adherence was calculated as the proportion of days covered (PDC), which was based on the total number of days supplied for each filled antihypertensive medication divided by the observation time interval. The observational time interval began on the day of first dispensed prescription of a medication or patient entry into the CAD registry if they were already receiving the medication. The observational time interval ended on the day of the last dispensed prescription plus the number of days supplied for that prescription. Classes of antihypertensive medications considered included β-blockers, dihydropyridine calcium channel blockers, nondihydropyridine calcium channel blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, and diuretics. For patients who were prescribed multiple medications, a summary PDC measure was calculated based on the averaged PDC for all antihypertensive medications. Patients were classified as nonadherent based on a PDC of less than 0.80, consistent with the literature.8,9
We evaluated therapy intensification based on changes in antihypertensive prescriptions filled during each patient's first and last 6 months of follow-up. We recorded daily dosages for each medication and defined dosage increase as any increase in daily dose in milligrams for any antihypertensive medication. Therapy intensification was defined as a dosage increase for any drug or an increase in the total number of antihypertensive medications between the first and last 6 months of follow-up.10 A single substitution of drug class was not counted as intensification.
The primary outcome variable was pattern of systolic BP (SBP) control. To model this, we used a SAS procedure (TRAJ; SAS Institute Inc, Cary, NC) to identify BP trajectories over time.11- 13 With this procedure, SBP measurements are treated as censored normal data following a polynomial time course and given a discrete latent class assignment. The procedure isolates distinct trajectories of SBP over time (one for each latent class) and fits a mixture model to calculate the probability of membership in each latent class for each patient. The Bayesian information criterion is used to determine the optimal number of trajectories and is analogous to the adjusted R2 in that it balances model complexity and model fit.12
We examined models that incorporated 2 to 8 trajectory groups based on all available BP data (median number of BP measurements per patient, 20; median follow-up, 4.5 years). The model with 6 distinct BP trajectory groups appeared to best fit the data and had the optimum Bayesian information criterion. Next, we calculated the average SBP for each trajectory group according to tertiles of follow-up duration using all available BP measurements within each tertile. During the first tertile of follow-up, the average SBP was less than 140 mm Hg in 4 of the groups and greater than 145 mm Hg in 2 of the groups. Therefore, we combined the 4 groups in which the SBP was less than 140 mm Hg and defined 3 clinically relevant SBP groups as follows: (1) patients with controlled SBP that remained stable and controlled over time based on an SBP less than or equal to 140 mm Hg (normal-normal); (2) patients who started with high SBP that decreased over time (high-normal); and (3) patients who started with high SBP that remained high over time (high-high).
Baseline demographic factors and comorbidities were compared across the 3 BP control groups using the χ2 test for categorical variables and analysis of variance for continuous variables. We also compared the percentage of patients with antihypertensive medication dosage increases or addition of medications and nonadherence across the 3 groups using χ2 tests. In our primary multivariable logistic regression analyses, we assessed the independent association between medication nonadherence and therapy intensification with uncontrolled BP (high-high group) using patients in the high-normal group as the referent group. These multivariable models included patient demographics and cardiac and noncardiac comorbidity variables.
To further assess the robustness of our findings, we performed a series of secondary analyses. First, we used patients in the normal-normal BP group as the referent group. Second, we assessed the association between medication nonadherence and therapy intensification among patients in the high-normal group compared with those in normal-normal group. Third, we restricted the cohort to patients with a hypertension diagnosis. Fourth, we performed sensitivity analysis for patients whose SBP levels were around 140 ± 2 (mean ± SD) mm Hg by (1) excluding them, (2) including them in the normal-normal group, and (3) including them in the high-high group. The results of these sensitivity analyses were consistent with the primary analysis and are not further reported. Fifth, we redefined our outcome variable of SBP control using the following method: (1) taking the average of first 2 SBP measurements for each patient within the first tertile of observation; (2) taking the average of last 2 SBP measurements for each patient within the last tertile of observation; and (3) categorizing patients into SBP control groups based on differences in the BP averages between the 2 time periods. Then, we repeated the multivariable analysis. Sixth, we used the “max-scaled” R2-square estimates to assess the relative contribution of nonadherence to the explained variance for uncontrolled BP over time.14
Finally, randomized controlled studies have found that even adherence to placebo is associated with improved outcomes, suggesting a “healthy adherer” effect whereby adherence to drug therapy may be a surrogate marker for overall healthy behavior.15 Because statin (3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitor) medications were prescribed for most patients (approximately 85%), we evaluated for this healthy adherer effect using statin adherence (PDC, ≥0.80) as a surrogate. First, we assessed the correlation between antihypertensive and statin medication adherence among patients receiving both medications (n = 8834). Next, we evaluated the association between statin adherence and BP control using the same multivariable modeling approach as the primary analyses. Finally, we added statin adherence to the multivariable models that included antihypertensive medication adherence and therapy intensification to determine whether statin adherence remained associated with BP control.
The study was approved by the institutional review board of KPCO. All analyses were performed using SAS statistical software, version 9.1 (SAS Institute Inc).
Patients in the normal-normal group had an average SBP level of 126.7 ± 11.3 mm Hg, which remained stable over time, and they made up the majority of the cohort (87.2%) (Figure 1). Patients in the high-normal group (7.5%) had an initial mean SBP level of 146.7 ± 8.1 mm Hg, which decreased to 128.1 ± 9.0 mm Hg at the end of the observation period. Patients in the high-high group (5.3%) had an initial SBP level of 154.4 ± 11.2 mm Hg, which remained high at the end of follow-up (152.5 ± 12.0 mm Hg).
Baseline patient characteristics based on BP control groups are outlined in the Table. Patients in the high-high and the high-normal groups were older and more likely to be female and to have comorbidities (eg, hypertension, diabetes, and cerebrovascular disease) than patients in the normal-normal group. Patients in the high-high and the high-normal groups were also more likely to be nonadherent to antihypertensive medications and to have therapy intensification (Figure 2). Of nonadherent patients in the high-high group, 67.2% received therapy intensification, while 60.1% of nonadherent patients in the the high-normal group received therapy intensification.
In multivariable analyses, medication nonadherence (odds ratio [OR], 1.73; 95% confidence interval [CI], 1.34-2.24) and therapy intensification (OR, 1.31; 95% CI, 1.01-1.70) were significantly associated with having uncontrolled BP (high-high group) compared with having high SBP that became controlled over time (high-normal group). Other significant factors in the multivariable models included female sex (OR, 1.39; 95% CI, 1.09-1.77), history of atrial fibrillation (OR, 0.65; 95% CI, 0.49-0.85), coronary artery bypass graft surgery (OR, 0.77; 95% CI, 0.61-0.98), and depression (OR, 0.74; 95% CI, 0.57-0.95). Similarly, patients with uncontrolled BP were more likely to be nonadherent to antihypertensive medications (OR, 1.50; 95% CI, 1.23-1.81) and to have therapy intensification (OR, 2.86; 95% CI, 2.35-3.49) than patients with normal BP over time (normal-normal group). The findings were consistent when the analysis was restricted to patients with a diagnosis of hypertension. These findings suggest that medication nonadherence may explain why BP levels remained elevated despite intensification of treatment with antihypertensive medications.
In contrast, patients in the high-normal group were more likely to have therapy intensification than patients in the normal-normal group (OR, 2.21; 95% CI, 1.88-2.60); however, there was no association between medication nonadherence and being in the high-normal group compared with the normal-normal group (OR 0.91; 95% CI 0.77-1.09). This suggests that patients in the high-normal group received therapy intensification and were taking antihypertensive medications as prescribed, which resulted in the decrease in BP levels over time.
In sensitivity analysis using the average of the first and last two BP measurements to define outcomes, the association between antihypertensive medication nonadherence (OR,1.23; 95% CI, 1.03-1.48) and therapy intensification (OR, 1.12; 95% CI, 0.95-1.33) in patients with uncontrolled BP compared with patients with high BP that decreased over time remained consistent. Next, we assessed the relative contribution of nonadherence to the explained variance for uncontrolled BP over time, and it accounted for approximately 23% of the variance.
Finally, we assessed for the presence of a healthy adherer effect using statin medication adherence as a surrogate marker. Adherence to antihypertensive medications and to statin medications was modestly correlated (r = 0.35; P < .001). In multivariable analyses, statin adherence was associated with BP control (OR, 0.53; 95% CI, 0.44-0.65). The findings were consistent even after adjustment for antihypertensive medication adherence and therapy intensification (OR, 0.54; 95% CI, 0.43-0.66).
In a large managed care organization cohort of patients with CAD, we found that approximately 12% of patients with CAD had uncontrolled BP, but this number decreased to approximately 5% over a median follow-up period of 4.6 years. Patients with uncontrolled BP were more likely to have intensification of antihypertensive therapy and to be nonadherent to antihypertensive treatment compared with patients with high BP that became controlled over time, suggesting that medication nonadherence may be an explanation for the continuously elevated BP levels despite titration of antihypertensive medications. The findings were consistent regardless of how BP control was defined.
The successful interaction of multiple factors, including patient, provider, and health care system, is often required to achieve the BP levels that are recommended by national guidelines.16 Previous studies have usually focused on either the patient or the provider in isolation and found that medication nonadherence or therapeutic inertia accounted for a small variation for why BP remained uncontrolled.17,18 In contrast to previous studies, we evaluated the impact of both patient and provider factors on BP control. In a setting with disease management, antihypertensive medication therapy was being intensified for patients with high BP. However, patients with uncontrolled BP were more likely to be nonadherent than patients with high BP that became controlled, suggesting that therapy intensification must be coupled with efforts to improve medication adherence to achieve desired BP levels. Our findings suggest that both therapy intensification and medication adherence are important components for achieving recommended BP goals and need to be evaluated concurrently to identify gaps in hypertension care.
The findings of our study have several potential clinical implications. First, in contrast to previous methods of assessing BP control, which usually involved a limited number of BP measurements, we used all available measurements and identified trajectories of BP control over time. This method may provide a better overview of BP control than previous methods based on single cross-sectional assessments, because BP control rates will vary depending on when the cross-section is measured. Second, our findings highlight the clinical importance of medication nonadherence and suggest that nonadherence should be routinely assessed in clinical practice, especially in patients who do not seem to be responding appropriately to therapy intensification. Finally, the assessment of either therapy intensification or medication adherence in isolation provides a limited view of why BP is not well controlled. Our findings demonstrate that both are important components for achieving BP control on a population basis. Future quality improvement efforts should incorporate both of these components as part of multifaceted interventions to improve BP control.
Randomized controlled clinical trials have found that even adherence to treatment with placebo is associated with improved patient outcomes, suggesting the presence of a healthy adherer effect.15,19- 22 Patients who are adherent to treatment with medications may be more likely to follow lifestyle recommendations and to practice other healthy behaviors that are associated with improved outcomes.23,24 In our study, adherence to statin therapy was modestly correlated with adherence to antihypertensive medications and associated with controlled BP even after adjustment for antihypertensive medication adherence and therapy intensification. These findings suggest that patients who are adherent to treatment with antihypertensive medications may be more adherent in general, particularly regarding medications and other healthy behaviors. Future studies should prospectively determine whether patients who are adherent to treatment with antihypertensive medications also practice other healthy behaviors associated with BP reduction (eg, exercising daily). Also, our findings highlight the importance of nonadherence as a risk marker for adverse outcomes and suggest the need for studies to evaluate the strategy of identifying nonadherent patients and to develop interventions to improve adherence and outcomes.
Several potential limitations of this study should be recognized. First, our findings should be interpreted in the context of an active disease management program in which clinical pharmacy specialists routinely manage antihypertensive medications in addition to the usual care provided to patients. Our findings should be replicated in settings in which disease management programs for BP management have not been implemented. Second, we found an association between statin adherence and BP control and suggest that statin adherence may be a marker for healthy adherer effect. A recent meta-analysis demonstrated a 4−mm Hg change (95% CI, −5.8 to −2.2 mm Hg) in SBP among patients receiving statins compared with placebo; however, in our analysis, we were comparing different levels of adherence among all patients who were prescribed a statin and not a statin vs a placebo.25 Therefore, the expected difference in SBP between highly adherent and less adherent statin users would be smaller than the meta-analysis findings. Furthermore, antihypertensive and statin medication adherence were correlated, suggesting that there is a general tendency to take medications, which we labeled the healthy adherereffect. Third, we considered a comprehensive list of commonly used classes of antihypertensive medications but did not include α-blockers, central-acting agents, and direct vasodilators. Finally, we found an association between nonadherence and therapy intensification with uncontrolled BP over time. Future studies should address the directionality of these associations as well as the impact of the number of medications and, more importantly, dosing complexity on medication nonadherence.
The results of our study suggest that both therapy intensification and medication adherence are important factors for BP control in community cohorts. The 2 factors, which, to our knowledge, have not been not previously elucidated together in nontrial settings, may help explain the continued “gaps” in BP control in clinical practice despite the widespread recognition of the importance of BP control. Studies are urgently needed to evaluate interventions that target nonadherent patients and to determine if such interventions can improve patient outcomes and eliminate gaps in care.
Correspondence: P. Michael Ho, MD, PhD, Denver Veterans Affairs Medical Center, 1055 Clermont St (111B), Denver, CO 80220 (Michael.firstname.lastname@example.org).
Accepted for Publication: June 14, 2007.
Author Contributions: Dr Ho had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Ho, Olson, and Rumsfeld. Acquisition of data: Magid and Shetterly. Analysis and interpretation of data: Ho, Magid, Shetterly, Olson, Peterson, Masoudi, and Rumsfeld. Drafting of the manuscript: Ho. Critical revision of the manuscript for important intellectual content: Ho, Magid, Olson, Peterson, Masoudi, and Rumsfeld. Statistical analysis: Magid and Shetterly. Obtained funding: Ho. Administrative, technical, and material support: Magid. Study supervision: Ho and Rumsfeld.
Financial Disclosure: None reported.
Funding/Support: This study was funded in part by an award from the American Heart Association (0535086N). Dr Ho is supported by a Veterans Affairs Research & Development Career Development Award (05-026-2), and Dr Peterson is supported by an award from the American Heart Association (0670017N).
Role of the Sponsor: The funding sources did not have involvement in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, and approval of the manuscript for this article.