Antihypertensive Medication Adherence in the Department of Veterans Affairs
Article Outline
Abstract
Purpose
Adherence measures the extent to which patients take medications as prescribed by their health care provider. The control of hypertension is dependent on medication adherence and may vary on the basis of antihypertensive medication class and other factors.
Methods
The Department of Veterans Affairs’ automated pharmacy database captures pharmacy medication use; International Classification of Diseases, 9th Revision, diagnostic codes; and laboratory and patient demographic data on a monthly basis. Hypertensive patients who used thiazide diuretics, beta-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, calcium channel antagonists, and alpha-blockers from July 2002 to December 2003 were studied. The first date of prescription filling for each patient within the date range was the index date from which fill and refill dates were collected for up to 18 months to calculate medication posession ratios and days out of medication ratios. Patients were categorized as adherent if the medication posession ratio was 80% or greater. Logistic regression was used to study the association of medication class, age, gender, ethnicity, Veterans Affairs facility, and co-diagnosis with diabetes, schizophrenia/psychosis, depression, and dementia with medication adherence.
Results
We studied 40,492 hypertensive patients taking at least one antihypertensive drug class. The average age per class ranged from 67.4 to 72.9 years; 96% were male; and 51% were white, 8% were African-American, 4% were Asian-American, and 3% were Hispanic. Unadjusted adherence rates based on the medication posession ratio ranged from 78.3% for thiazide diuretics to 83.6% for angiotensin receptor blockers (P
<.001). The number of medications (either total or antihypertensive) and age were independent predictors of better adherence. Black ethnicity and depression were associated with worse adherence.
Conclusions
Adherence rates with all antihypertensive medications were high. Although there were statistical differences by drug class, these differences were small. Ethnicity and depression identified groups that might benefit from programs to improve adherence.
Keywords: Hypertension, Medication adherence, Antihypertensive drug class, Thiazide diuretics, Beta-blockers, Angiotensin-converting enzyme inhibitors, Angiotensin receptor blockers, Calcium channel antagonists, Alpha-blockers
Hypertension is the most common chronic cardiovascular condition in the United States, affecting more than 50 million Americans.1, 2 The benefits of decreasing blood pressure in hypertensive individuals on the morbidity and mortality associated with cardiovascular and renal disease has been established in many randomized controlled trials,3, 4, 5 including the seminal Department of Veterans Affairs (VA) study published in 1970.6 However, data from the National Health and Nutrition Survey III indicate that only 46% of men and 65% of women with a blood pressure level of 140/90 mm Hg or greater are currently receiving antihypertensive treatment in the United States.7 Of those under treatment, only 50% of men and 58% of women have adequately controlled hypertension. The difficulty in controlling hypertension is likely multifactorial, including patient belief systems dismissing it as a significant medical problem, lack of consistent medical follow-up, provider failure to adjust drug regimens to achieve control, medication side effects and dosing, lack of economic resources to access health care or purchase medications, and lack of patient adherence to medications.8, 9
Adherence, in the medical context, may be defined as the extent to which a patient’s behavior follows the recommendations of his or her health care practitioner.10 In terms of drug therapy, adherence describes the degree of correspondence of the actual dosing history with the prescribed medication regimen. This may be a particular problem in the management of hypertension. Patients with hypertension are often asymptomatic. The disease then lacks a tangible reinforcement factor to foster medication adherence in contrast with what may be seen in other chronic conditions, such as diabetes or asthma, in which symptoms often remind the patient of the need to take medications.
Prior research has found conflicting information on those factors that are associated with better medication adherence.11, 12, 13 In addition, in the area of hypertension, debate has focused on whether there is a difference in side effects among the different classes of antihypertensives that, in turn, may lead to variable patient adherence to prescribed medication regimens. In this report, we examine adherence to antihypertensive medications prescribed in a regional VA health care system. We study whether there is a difference in adherence to medications representing different antihypertensive medication classes as well as factors that are associated with different degrees of antihypertensive medication adherence across all antihypertensive classes.
Methods
The Veterans Integrated Service Network 21 Pharmacy DataMart and associated databases were used. Veterans Integrated Service Network 21 is a geographic grouping of six VA institutions in Northern California, Northern Nevada, and Hawaii. Automated data extraction routines capture pharmacy, International Classification of Diseases, 9th Revision (ICD-9), laboratory, provider, and patient demographic data on a monthly basis. These data were loaded into an SQL Server database system, and a web-browser application (Proclarity, Knosys Inc, Boise, ID) was used to develop and view presorted data cubes. The pharmacy database reflects data that are used to dispense medications at the 6 facilities within Veterans Integrated Service Network 21.
Patients were included if they had a diagnosis of hypertension (ICD-9 codes 401.1-401.9) and a history of medication use from the VA from July 2002 to December 2003. Medication use is defined as having received 1 or more medications in the following drug classes: thiazide diuretics, beta-blockers, angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers, calcium channel antagonists, and alpha-blockers. To be included, 1 of these medications had to be dispensed from July 2002 to October 2002. The index date is the earliest fill date within this 4-month time period. This study looks at the entire treated hypertensive population within Veterans Integrated Service Network 21. Patients with a serum creatinine of 2.5 mg/dL or greater during the time period of analysis (July 2002 to December 2003) were excluded. Patients were also excluded if they had no pharmacy activity for at least 6 months after their index prescription date. This was done by eliminating all records with a maximum date minus minimum prescription date of less than 180 days. Patients receiving potassium-sparing diuretics separate from thiazide diuretics and not in a combination product were excluded, as were patients receiving the calcium channel antagonists bepridil and nimodipine, because these medications are not indicated for hypertension.
All fill and refill dates were collected from July 2002 to December 2003 for each medication within the antihypertensive classes studied. For a specific patient, the first date of prescription filling within the index date time range is the index date for that medication category. From that date, prescription filling dates were collected through December 2003. The last prescription fill date is defined as the finish date for that patient. Medication data elements collected include drug, antihypertensive drug class, fill or refill date, days’ supply, and quantity dispensed. Other data elements for covariate analysis collected include the date of birth, gender, ethnicity, VA facility, number of active medications at the time of the finish date, and ICD-9–based diagnoses of diabetes mellitus, mental illness (schizophrenia and other psychotic ICD-9 codes, as well as depression), Alzheimer’s disease, and dementia. Age in years was estimated by taking the difference in days between October 1, 2003, and the date of birth and dividing by 365. The number of different antihypertensive drug classes was calculated with a cross-tabulation of the main data table using Microsoft Access (Microsoft Corp, Redmond, Wash).
Medication adherence was calculated with 2 variables using definitions from Steiner et al.14 For each patient’s drug class, the sum of days dispensed (except for the prescription dispensed on the finish date) was compared with the number of days elapsed from index date to finish date. This is the medication possession ratio, and in the ideal situation it would approach 100%.15, 16 In addition, the days out of medication was calculated by subtracting for each fill the difference in the number of days between fills to the number of days supplied. The sum of the gap in days between fills and supply for each patient between the index date and the finish date is the out of medication variable, and in the ideal situation it would approach 0. In our calculations, a negative number indicates overcompliance with medication.
The potential effect of other variables was examined using logistic regression analysis. For this purpose, the population was broken down into adherent and nonadherent groups based on adherence rates of greater or less than 80% with a medication posession ratio of 0.8 or greater defining a patient as adherent. Covariates in the regression model included age, gender, ethnicity, antihypertensive medication class, number of antihypertensive medication classes, number of active medications, and presence or absence of diabetes, dementia, or mental illness as defined above. In individuals taking only one antihypertensive, adherence to thiazide diuretics was compared with adherence to each antihypertensive medication class.
Results
We studied 40,492 patients who were taking at least one medication of the 6 antihypertensive drug classes (Table 1). Patients had an average age per group ranging from 67.4 to 72.9 years; 96% were male; and 50.6% were white, 8.2% were African-American, 4.1% were Asian-American, and 3.4% were Hispanic. Ethnicity was not available for analysis for 33.3% of patients. One antihypertensive was taken by 47.9% of patients, 2 antihypertensives were taken by 35% of patients, 3 antihypertensives were taken by 13.7% of patients, and 4 or more antihypertensives were taken by 3.4% of patients, with 1.7 as the average number of antihypertensives taken.
Table 1. Baseline Characteristics of Study Patients (N=40,492)
| Average Age (SD) | 68.6(11.2)years |
| Gender (Men/Women) | 39,038/1454(96.4%/3.6%) |
| Race Distribution (%) | |
| White | 20,495(50.6) |
| Unstated | 13,502(33.3) |
| Black | 3330(8.2) |
| Asian | 1644(4.1) |
| Hispanic | 1364(3.4) |
| Other | 157(0.4) |
| Average number antihypertensive drug classes (SD) | 1.73(0.83) |
| Number of antihypertensives (%) | |
| 1 | 19,390(47.9) |
| 2 | 14,178(35.0) |
| 3 | 5563(13.7) |
| ≥4 | 1361(3.4) |
The largest number of patients (48.7%) took an ACE inhibitor, 38.4% took a beta-blocker, 30.4% took a calcium antagonist, 28.5% took a thiazide diuretic, 18.4% took an alpha-blocker, and 8.4% took an angiotensin receptor blocker (Table 2). Unadjusted adherence rates based on medication posession ratio ranged from 78.3% for thiazide diuretics to 83.6% for angiotensin receptor blockers. This difference across antihypertensive classes was statistically significant by chi-square test (P<.001). Similar findings were noted using the out-of-medication parameter. After logistic regression analysis, older age, increasing number of antihypertensive drug classes and total number of medications, and treating facility were predictors of better adherence. Depression and ethnicity were predictors of worse adherence (Figure 1). (African-Americans were less adherent than Asians who were less adherent than whites.) Similar findings were found among diabetic patients, except that treating facility was no longer a statistically significant predictor of adherence (data not shown).
Table 2. Unadjusted Adherence to Antihypertensive Medication Classes
| Drug Class (%) N=40,492 | Average MTOT⁎ (SD) | Adherent by MTOT† (% within drug class) | Average MOUT‡ (SD) | Adherent by MOUT§ (% within drug class) |
|---|---|---|---|---|
| Angiotensin converting enzyme inhibitor (48.7) | 0.940(0.223) | 16,008(81.1) | 0.174(0.168) | 13,423(68.0) |
| Beta-blocker (38.4) | 0.937(0.228) | 12,360(79.5) | 0.183(0.176) | 10,203(65.6) |
| Calcium antagonist (30.4) | 0.955(0.233) | 10,163(82.7) | 0.168(0.165) | 8590(69.9) |
| Thiazide diuretics (28.5) | 0.920(0.220) | 9044(78.3) | 0.180(0.182) | 7715(66.8) |
| Alpha-blocker (18.4) | 0.950(0.269) | 5876(78.8) | 0.190(0.176) | 4825(64.7) |
| Angiotensin receptor blocker (8.4) | 0.950(0.205) | 2843(83.6) | 0.166(0.164) | 2406(70.8) |
⁎MTOT is also known as the Medication Possession Ratio. [Sum of days supply of medication dispensed except for last fill/number of days between first and last fill]. |
†Defined as an MTOT of >0.80. |
‡MOUT is also known as the Medication Gaps Ratio. [Sum of gaps/number of days between first and last fill; a gap is the number of days between the time where a preceding prescription is expected to last based on quantity and days supply dispensed and the date when the refill is actually filled]. |
§Defined as an MOUT of ≤0.20. |
Figure.
Predictors of antihypertensive medication adherence.* *Using logistic regression analysis.
Among patients taking only 1 antihypertensive, older age remained a predictor of better adherence, and ethnicity and depression remained predictors of worse adherence (data not shown). There was a trend toward an association between increasing number of medications and better adherence that did not reach statistical significance (P=.08). After adjustment for variables found to be predictors of better or worse adherence, adherence to thiazide diuretics was less likely when compared with adherence to ACE inhibitors, beta-blockers, calcium antagonists, and angiotensin receptor blockers (Table 3).
Table 3. Adherence to Antihypertensive Class Compared with Thiazide Diuretics in Patients Taking One Antihypertensive
| Odds Ratio⁎ | P Value | (95% Confidence Interval) | |
|---|---|---|---|
| Thiazide user (n = 2039) versus: Angiotensin converting enzyme inhibitors (n=6402) | 0.692 | <0.001 | (0.615-0.779) |
| Beta-blockers (n=3890) | 0.830 | <0.01 | (0.734-0.938) |
| Calcium antagonist (n=2789) | 0.647 | <0.001 | (0.565-0.741) |
| Alpha-blocker (n=1480) | 0.987 | 0.870 | (0.839-1.160) |
| Angiotensin receptor blocker (n=972) | 0.649 | <0.001 | (0.534-0.789) |
⁎Using logistic regression analysis. Odds ratios are adjusted for gender, race, comorbidities and count of all medications. |
Discussion
For thiazide diuretics, calcium antagonists, beta-blockers, ACE inhibitors, and angiotensin receptor blockers, approximately 80% of medication prescribed appeared to be taken. This is consistent with findings from the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial in which more than 80% of participants were taking the thiazide diuretic, calcium antagonist, or ACE inhibitor at 1 year.4 Unadjusted adherence rates based on medication posession ratio ranged from 78.3% for thiazide diuretics to 83.6% for angiotensin receptor blockers, which were statistically significant differences across antihypertensive classes. In multivariate analysis, however, adherence rates among the antihypertensive classes were not significantly different for all patients in our study. However, when the analysis was limited to patients taking only one antihypertensive medication and when adjusted for other variables, patients taking thiazide diuretics were found to be less adherent than those taking ACE inhibitors, beta-blockers, calcium antagonists, or angiotensin receptor blockers.
Other studies of differences in adherence by antihypertensive class found similar results. In a study using medical claims data, ACE inhibitors and calcium antagonists had higher rates of compliance than beta-blockers and thiazide diuretics, although overall rates of adherence were low, varying from 35% for ACE inhibitors and calcium antagonists to 29% for beta-blockers and 15% for thiazide diuretics.17 In a Canadian population, Caro et al18 found that 89% of patients taking ACE inhibitors were adherent over 6 months, compared with 86% taking calcium antagonists, 85% taking beta-blockers, and 80% taking diuretics. A study of 21,723 patients from the Merck-Medco database using the refill analysis approach found similar rankings of adherence by antihypertensive drug class.19 In addition, a relatively small angiotensin receptor blockers group was the most adherent. It seems that there are small, but perhaps important, differences in adherence by antihypertensive medication class. Quality of life measures have been related to discontinuation rates. The calcium antagonist nifedipine was associated with both more symptomatic complaints and higher discontinuation rates than the ACE inhibitor cilazapril and the beta-blocker atenolol.20 Another trial compared the beta-blocker bisoprolol with nifedipine on quality of life and withdrawal from the study.21 Again, nifedipine was associated with both more adverse effects and a greater rate of withdrawal. It is interesting to note that our rates and the Canadian adherence rates are higher overall than those reported in other studies. This may reflect the lower cost of medications for VA and Canadian populations.
Our results found a positive association between antihypertensive medication adherence and older age, as well as with the number of cardiovascular and total medications. Blacks and patients with depression were found to be less adherent. These results were found for the entire cohort, as well as when diabetic patients or those patients taking only one antihypertensive were analyzed separately. These results are similar to those found in a study of persistence of statin therapy in which nonwhite race, lower income, older age, depression, and dementia were independent predictors of poor long-term persistence.12 In another study, major depression was associated with poor adherence to a regimen of prophylactic aspirin after a diagnosis of coronary artery disease.22 Shalansky and Levy13 found that taking fewer medications was associated with lower adherence to chronic cardiovascular regimens. Eagle et al23 found that patients with myocardial infarction and hypertension were more likely to take beta-blockers than patients with only one of these conditions. These results may identify patient groups who would profit from more intensive efforts to promote adherence.
Adherence may depend on the perceived health benefits of the medication and the ability to reach goals such as a blood pressure or serum cholesterol levels.24, 25 Failure to refill antihypertensives on time corresponds to patients not taking medication on a regular basis, resulting in loss of blood pressure control.14 Failure to follow treatment regimens has been associated with poorer treatment outcomes.26 It has been estimated that 5% to 15% of hospitalizations among the elderly are the result of poor adherence to prescribed medications.27 Of all medication-related US hospital admissions, 33% to 69% are the result of poor medication adherence, with a resultant cost of approximately $100 billion per year.10 Poor adherence might also result in unintended consequences. Patients who fail to take their antihypertensives as prescribed, leading to worse blood pressure control, might have additional medications added to their antihypertensive regimen. If all medications are then taken as prescribed, hypotension and other untoward consequences might result.
Different approaches to the prescribing of medication might improve patient adherence. Changes in dosing schedules have been tried. The combination of an ACE inhibitor and a thiazide diuretic in a single pill resulted in better adherence than when the medications were given separately.28 Although combination therapy might simplify dosing schedules, it is usually at increased costs. Changes in antihypertensive medication, as sometimes dictated by formulary changes, may have deleterious effects on adherence. Changes in therapy might also be dictated by failure to reach treatment goals, medication side effects, or medication cost. Patients who switched statins for unknown reasons were less likely by 18.9% to be highly compliant with statin use as defined by the probability of having a medication possession ratio of 0.8 or more.29
Our study has limitations. As in other observational studies, we are unable to distinguish the effect of treatment from those factors that influence the selection of the treatment. We used pharmacy databases that accurately describe the dispensing of medications. We do not have any information on whether dispensed medications were actually taken by patients. Another limitation of our study is that we are unable to determine how many patients are taking antihypertensives for hypertension or for coexisting illnesses in which these medications are also used. Although we cannot determine the extent of this use, it is unlikely that the presence of other diagnoses greatly influences our results. The prevalence of hypertension far exceeds that of the other disorders for which some of these drugs may be used. Further, over the time span of our study, it is unlikely that there have been any major shifts in treatment patterns such that higher proportions of patients would be receiving these medications for illnesses other than hypertension.
Strategies to improve blood pressure control and decisions concerning the best pharmacologic treatment for these patients will have major morbidity and mortality consequences. It will also have important implications for the cost of medical care. What can be done to improve medication adherence? The provider and the health delivery system, as well as the patient, have responsibility to improve adherence. In a prior study, patients with better adherence to medications for chronic cardiovascular conditions had physicians who reported more job satisfaction, saw more patients, and more often provided patients with definite future appointments or telephone contact than did physicians of patients with poor adherence.30 In a Danish study of primary care disease management for diabetes, regular follow-up visits resulted in substantially lower glycohemoglobin levels.31 The current trend in VA and other health systems toward advanced access,32 in which patients initiate most health visits, might lead to worse medication adherence. Recent reviews have suggested that patient and caretaker education, improved dosing schedules, clinical reminders, special pill containers, support groups, rewards for adherence, and telephone-based computer systems for monitoring and educating patients improve adherence.10, 33, 34 In the case of hypertension, self-monitoring of blood pressure can also enhance adherence.35, 36 Although it is necessary to improve adherence in all patients, on the basis of our results and the results of other investigators, it might be useful to target specific patient subgroups with interventions that would improve adherence to medications. Ultimately, adherence to medication is a key factor in determining the success of various therapeutic plans, and plans to improve adherence will be a key factor in improving patient outcomes.
References
- National Heart, Lung, and Blood Institute Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure; National High Blood Pressure Education Program Coordinating Committee (The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure). JAMA. 2003;289:2560–2572
- The burden of adult hypertension in the United States 1999 to 2000: a rising tide. Hypertension. 2004;44:398–404
- . Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension (Final results of the Systolic Hypertension in the Elderly Program). JAMA. 1991;265:3255–3264
- . The antihypertensive and lipid-lowering treatment to prevent heart attack trail. JAMA. 2002;288:2981–2997
- . Effects of different blood-pressure-lowering regimens on major cardiovascular events: results of prospectively-designed overview of randomized trials. Lancet. 2003;362:1527–1535
- . Effect of treatment on morbidity in hypertension (II. Results in patients with diastolic blood pressure averaging 90 through 114 mm Hg). JAMA. 1970;213:1143–1152
- Prevalence of hypertension in the US adult population: results from the Third National Health and Nutrition Examination Survey, 1988-1991. Hypertension. 1995;25:305–313
- . Controlling blood pressure in 50% of all hypertensive patients: an achievable goal in the healthy people 2010 report?. J Invest Med. 2003;51:373–385
- . Trends in antihypertensive drug therapy of ambulatory patients by US office-based physicians. Hypertension. 2000;36:600–603
- . Adherence to medication. N Engl J Med. 2005;353:487–497
- . Predictors of medication adherence in the elderly. Clin Ther. 1998;40:764
- Long-term persistence in use of statin therapy in elderly patients. JAMA. 2002;288:455–461
- . Effect of number of medications on cardiovascular therapy adherence. Ann Pharmacother. 2002;36:1532–1539
- . A general method of compliance assessment using centralized pharmacy records (Description and validation). Med Care. 1988;26:814–823
- . Estimating medication persistency using administrative claims data. Am J Manag Care. 2005;11:449–457
- . Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf. 2006;8:565–574
- . Variations in compliance among hypertensive patients by drug class: implications for health care costs. Clin Ther. 1997;19:1446–1457
- Effect of initial drug choice on persistence with antihypertensive therapy (The importance of actual practice data). Can Med Assoc J. 1999;160:41–46
- . Continuation of initial antihypertensive medication after 1 year of therapy. Clin Ther. 1998;20:671–681
- Quality of life with three antihypertensive treatments-cilazapril, atenolol, nifedipine. Hypertension. 1992;19:499–507
- . Bisoprolol and nifedipine retard in elderly hypertensive patients: effect on quality of life. J Hum Hypertens. 2000;14:205–212
- Major depression and medication adherence in elderly patients with coronary artery disease. Health Psychol. 1995;14:88–90
- Adherence to evidence based therapies after discharge for acute syndromes (An ongoing, prospective, observational study). Am J Med. 2004;117:73–81
- Determinants of discontinuation of new courses of antihypertensive medications. J Clin Epidemiol. 2002;55:728–735
- . Predictors of long-term persistence on statins in a subsidized clinical population. Value Health. 2000;3:417–426
- . Patient adherence and medical treatment outcomes. Med Care. 2002;40:794–811
- . Drug therapy of the aged: the problem of compliance and the roles of physicians and pharmacists. J Am Geriatr Soc. 1984;32:301–307
- . A retrospective study of persistence with single-pill combination therapy vs. concurrent two-pill therapy in patients with hypertension. Manag Care. 2000;9(Suppl):S2–S6
- . The effect of switching on compliance and persistence: the case of statin treatment. Am J Manag Care. 2005;11:670–674
- Physicians’ characteristics influence patients’ adherence to medical treatment: results from the medical outcome study. Health Psychol. 1993;12:93–102
- Randomized controlled trial of structured personal care of type 2 diabetes mellitus. BMJ. 2001;323:970–975
- . Improving timely access to primary care: case studies of the advanced access model. JAMA. 2003;289:1042–1046
- . Interventions to enhance patient adherence to medication prescriptions. JAMA. 2002;288:2868–2879
- . Patient adherence: the next frontier in quality improvement. Am J Med. 2004;117:130–132
- . Comparing compliance patterns between randomized treatments. Control Clin Trials. 1997;18:187–203
- Adherence to pharmacologic management of hypertension. Can J Public Health. 1998;89:I16–I18
The views expressed in the article do not necessarily represent the views of the Department of Veterans Affairs or of the United States Government.
PII: S0002-9343(06)00776-5
doi:10.1016/j.amjmed.2006.06.028
© 2007 Elsevier Inc. All rights reserved.
