Antihypertensive medication adherence was associated with improvement in certain short-term utilization measures among employees with high prior medical costs.
: To determine if antihypertensive medication adherence is associated with decreased medical and drug costs, medical service utilization, and work absence days.
: Retrospective database study using medical, pharmacy, sick leave, short-term and long-term disability, and workers’ compensation claims data from multiple large US employers from 2001 to 2008.
: We used medical and pharmacy claims to identify employees with hypertension. The index date was the date of the first hypertensionrelated pharmacy claim. Eligible employees had health plan enrollment 6 months before the index date and at least 12 months after the index date. Employees younger than 45 years were excluded from the study. Regression models estimated the effect of the proportion of days covered (PDC) by hypertension medication on outcomes after the index date, including health benefit costs, medical service utilization, and work absence days, as well as some clinical outcomes calculated separately for high-prior-cost and low-prior-cost employees. High-prior-cost employees were those who accounted for the top 60.0% of total medical costs during the 6 months before the index date. The regression models controlled for demographics, job-related variables, and comorbidities.
: Among low-prior-cost employees, high PDC was associated with increased medical and drug costs and work absence days. Among highprior- cost employees, high PDC was associated with decreased medical and drug costs, fewer work absence days and inpatient hospital days, and increased hypertension-specific medical costs.
: Antihypertensive medication adherence was associated with improvement in some short-term utilization measures among highprior-cost employees, but significant short-term improvement was not seen among low-prior-cost employees.
(Am J Manag Care. 2009;15(12):871-880)
Increased antihypertensive medication adherence was associated with several improved short-term outcomes among the 4.6% of employees with the highest prior medical costs.
Hypertension affected 73.6 million US adults in 2006,1 up from 65 million 6 years earlier.2 One in 3 US adults has hypertension.3 High blood pressure is a precursor to many serious health conditions such as coronary heart disease,4 heart failure,5 and stroke,6 and has an overall death rate of 17.9%.1 Besides its health influences, hypertension has financial effects.7 The direct and indirect costs of hypertension for 2008 were expected to total $69.4 billion, making hypertension the second most costly cardiovascularrelated disorder.1,8
There is strong evidence that prescription drugs provide clinical value for treatment of hypertension.9,10 For example, pharmacologic treatment of hypertension reduces the risk of stroke, coronary events, heart failure, and progression of renal disease.9,11-17 On the other hand, there is evidence that hypertension is a major driver of drug spending in many countries.18 Employers want to know whether antihypertensive medications are cost-effective, if drug treatment reduces overall healthcare costs, and whether hypertension medication adherence reduces patients’ need for expensive cardiovascular services, hospitalizations, and emergency department treatment.
Results have shown that antihypertensive drug therapy is an important factor in containing the costs of managing hypertension and its complications.17 However, the therapeutic and economic benefits of drug treatment are often demonstrated in the controlled settings of clinical trials. These benefits may not be realized in day-to-day practice, wherein the cost-effectiveness of drug therapy may fail because of factors such as inadequate blood pressure control,14,17 discontinuation of or switching between therapies,19 and poor compliance.10,20
Nonadherence to antihypertensive drugs is high,21-26 with possibly substantial economic consequences.26 When conditions are treated suboptimally, symptoms and complications may worsen, leading to increased use of hospital and emergency department services, office visits, and other medical resources.24,27 For some chronic conditions, including hypertension, there is evidence that increased adherence may generate medical savings that more than offset the associated increases in drug costs.22,28-31
Few studies have measured the cost-effectiveness of medication adherence for treatment provided under benefit plans in employed population settings, although some studies21,32-36 have attempted to measure the effect on healthcare costs of drug use reduction following imposition of coverage limits or copayment requirements in Medicaid settings. To our knowledge, a single study18 has evaluated the relationships among antihypertensive medication adherence, medical services utilization, and healthcare costs among a large benefit plan population. The results showed that all-cause medical costs and hospitalization rates were significantly lower for patients with high medication adherence.18
The primary objective of this study was to quantify the relationship between antihypertensive medication adherence and health benefit costs, medical service utilization, work absence days, and certain clinical outcomes such as stroke, myocardial infarction, and cardiovascular procedures. This study adds to the compliance literature by measuring a broader set of outcomes relevant to employers than previously studied, including medical and drug costs and utilization, work absence days because of sick leave, short-term and long-term disability, and workers’ compensation claims, as well as clinical outcomes for high-prior-cost and low-prior-cost employees.
Research Population and Cohort Definitions
The patients included in the study are derived from more than 670,000 employees across the United States whose health benefits, salary, and demographic data are maintained in the Human Capital Management Services Research Reference Database at an employee level. The data used in this study are from 2001 to 2008.
The study cohort (employees with hypertension) was defined using health insurance claims data for healthcare services and prescription drug claims. Employees with hypertension were identified using any of the following inclusion criteria: (1) at least 2 medical claims (any kind) for hypertension (primary, secondary, or tertiary International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code of 401.xx), (2) an emergency department visit with a primary diagnosis of hypertension, or (3) an inpatient admission with a primary diagnosis of hypertension. In addition, all employees were required to have a hypertension maintenance medication prescription (antihypertensive, diuretic, b-blocker, angiotensin-converting enzyme inhibitor, calcium channel blocker, or a-agonist or a-blocker). A flowchart summarizing the selection of employees based on the eligibility requirements is given in . The date of the first hypertension prescription among the data was defined as the employee’s index date. The first hypertension medical claim was required to occur any time before 1 year after the index date. The measurements of medical and drug costs, medical service utilization, and work absence days were generally made over a 12-month period beginning at the index date. The measurements of clinical outcomes (stroke, myocardial infarction, and cardiovascular procedures) were made over a 24-month period beginning at the index date. Employees in the study cohort were required to be at least 45 years of age on the index date and to be employed and enrolled in a health insurance plan from 6 months before to at least 12 months (≥24 months for clinical outcomes) after the index date.
Descriptive variables were examined to provide information on demographic characteristics and on certain healthcare utilization and copayment values of the study cohort. In addition, data on subcohorts were included in the analysis of each particular outcome.
Hypertension-Related Drug Use
The study examined the number of employees using each class of hypertension-related medication. The proportion of days covered (PDC) was calculated directly from the days supply information in the claims data for all employees for the 12-month period after the index date as follows: the number of days each employee had a supply of any hypertensionrelated medication available divided by the number of days in the measurement year. Overlap periods (days when the employee possessed >1 type of hypertension-related medication) were not counted twice.
Effect of PDC on Outcomes
Regression techniques were used to estimate the effect of PDC during 12 or 24 months after the index date on the following 9 outcome (dependent) variables: (1) annual (year after the index date) medical and drug costs; (2) annual hypertension-specific (ICD-9-CM code 401.xx) medical costs; (3) the annual number of medical services; (4) the annual number of inpatient hospital days; (5) annual number of work absence days because of sick leave, short-term disability, long-term disability, or workers’ compensation; (6) annual probability of a workers’ compensation claim; (7) 2-year (after the index date) probability of a stroke; (8) 2-year probability of a myocardial infarction; and (9) 2-year probability of a cardiovascular procedure (coronary artery bypass, percutaneous transluminal coronary angioplasty, or coronary stents).
Medical costs included all insurance payments for all medical services as recorded in the health insurance claims data. Drug (pharmacy) costs included all insurance payments for all drug costs as recorded in the health insurance claims data. Hypertension-specific medical costs included all insurance payments for medical claims with a primary diagnosis of hypertension. All costs were inflation adjusted to 2008 US dollars. Measurement of annual work absence days because of sick leave, short-term disability, long-term disability, and workers’ compensation was based on payroll records and on disability carriers’ claims data.
Regression methods were used to measure the association between outcomes and PDC (measured as a continuous variable). Generalized linear models with a gamma distribution and log link were used to estimate the level of continuous (medical and drug costs, medical service utilization, and work absence days) outcomes. Logistic regression analysis was used to predict the likelihood of having any dichotomous outcome (workers’ compensation claim, stroke, myocardial infarction, or cardiovascular procedure). Only employees eligible for a given benefit were included in the regression models used to estimate medical and drug costs or work absence days from that benefit type. In the models for stroke, myocardial infarction, and cardiovascular procedures, employees with these events in the 6 months before the index date and employees with fewer than 24 months of health plan enrollment after the index date were excluded.
The objective of these models was to determine if PDC is associated with any of the aforementioned outcomes, after controlling for the following: age, sex, salary, location of residence, proportion of hypertension-related medication claims filled using mail order versus retail pharmacy, amount of employee copayments for hypertension-related medications, the number of other (nonhypertension) medications, month of the index date, emergency department use during 6 months before the index date, and the Charlson Comorbidity Index (and specific indicators for diabetes mellitus, hyperlipidemia, myocardial infarction, and chronic heart failure). Because medical costs before the index date were found to be strongly associated with most outcomes in the study, these analyses were performed separately for employees with high versus low total medical costs in the 6 months before the index date. The cutoff between high prior cost and low prior cost was determined by identifying the highest-cost employees (who accounted for the top 60.0% of total medical costs before the index date [top 3 “cost quintiles,” or 4.6% of the population]), and separating them from the lowest-cost employees (who accounted for the remaining 40.0% of total medical costs before the index date). To obtain a solid model for each outcome and subgroup, the following approach was used. We initially included in the model all of the aforementioned control variables and then used the automatic “stepwise” variable selection function to maintain in the model only those variables that had significant association with the outcome. As a result, the specific models for different outcomes and subgroups have different sets of control variables.
The study cohort included 6236 employees, all of whom had health plan enrollment from 6 months before their index date to at least 12 months after. The flowchart in Table 1 summarizes how the sample was obtained based on the eligibility requirements and the specific inclusion criteria. Out of 86,396 employees who met the hypertension diagnosis inclusion criterion shown in Table 1, only 7.2% met all other eligibility requirements and were included in the study cohort. Of 6236 employees, 1.3% (n = 84) had an emergency department visit with a primary diagnosis of hypertension, 3.7% (n = 228) had an inpatient admission with a primary diagnosis of hypertension, 0.4% (n = 25) had both an emergency department visit and an inpatient admission for hypertension, and the remaining 94.6% (n = 5899) had at least 2 outpatient medical claims (primary, secondary, and tertiary ICD-9-CM codes) for hypertension.
gives demographic and basic medical service utilization statistics for the study cohort, as well as for the subcohorts used in the analysis of specific outcomes. The mean age of employees in the study cohort and subcohorts was approximately 52 years, the proportion of women varied between 30.1% and 38.7%, and the annual salary varied between $52,000 and $66,000, with the lowest salary in the subcohort with eligibility for the analysis of work absence days. The PDC among the different subcohorts varied between 60.2% and 63.4%. Between 7.5% and 11.8% of employees in the different subcohorts had emergency department visits in the 6 months before the index date. Between 18.1% and 25.2% of employees filled their hypertension medications using mail order versus retail. Copayment per day supply ranged from $0.43 to $0.49, and the number of nonhypertension drugs ranged from 5.8 to 6.2. The Charlson Comorbidity Index ranged from 0.58 to 0.67. Diabetes was present in 23.4% to 24.1% of employees in the various cohorts. Between 3.4% and 4.2% of employees had chronic heart failure, between 0.7% and 2.3% had a myocardial infarction, and between 47.9% and 51.0% had
Hypertension-Related Drug Use Among the employees in the study, the most commonly used medication class was antihypertensives (4508 employees [averaging 6.6 prescription claims per employee]). This was followed by diuretics (1596 employees [averaging 5.0 claims]), β-blockers (1350 employees [5.9 claims]), angiotensin-converting enzyme inhibitors (1252 employees [5.7 claims]), calcium channel blockers (1086 employees [5.7 claims]), and α-agonists or α-blockers (72 employees [4.4 claims]).
As shown in the , 40.3% of the cohort had a PDC between 1% and 60%. Approximately 30.3% had a PDC between 61% and 90%. The remaining 29.4% of the cohort had a PDC above 90%.
Effect of PDC on Outcomes
Table 3 gives the estimated regression coefficients from the 9 pairs of models, and Table 4 summarizes the results of the analyses measuring the short-term effects of PDC on various adjusted medical and drug costs, medical service utilization, work absence days, and clinical outcomes. The results are provided separately for employees with high versus low prior medical costs.
In the group with low prior cost, there was a significant positive association between PDC and 3 of 9 measured outcomes, including annual medical and drug costs, annual hypertension- specific medical costs, and the annual number of work absence days. Specifically, a 10% increase in PDC was associated with 1.7% to 7.4% increases in these outcomes. In the group with high prior cost, PDC had varying effects. As PDC increased in this population, annual medical and drug costs, the annual number of inpatient hospital days, and the annual number of work absence days significantly decreased. For every 10% increase in PDC, the group with high prior cost could expect 5.1% less annual medical and drug costs, 8.2% fewer annual inpatient hospital days, and 12.5% fewer annual work absence days. However, annual hypertensionspecific medical costs increased concomitantly with higher PDC among patients with high prior cost.
Among the control variables, the most influential were the following: (1) the number of nonhypertension drugs, which had statistically significant positive associations with outcomes variables in 10 of 18 models (Table 3); (2) the Charlson Comorbidity Index (significant positive associations in 9 of 18 models); (3) chronic heart failure (significant positive associations in 7 of 12 models); (4) age (significant positive associations in 8 of 18 models; in the models for the subcohorts with low prior cost, the association of outcomes with age was positive, and in the models for the subcohorts with high prior cost [except for the model for myocardial infarction], the association of outcomes with age was negative); (5) myocardial infarction (significant positive associations in 5 of 12 models); (6) sex (significant positive associations in 6 of 18 models; in the models for the subcohorts with low prior cost, the association of outcomes with female sex was negative, and in the models for the subcohorts with high prior cost, the association of outcomes with female sex was positive);
and (7) diabetes (significant negative associations in 6 of 18 models).
This study measured the association of antihypertensive medication adherence with health benefit costs, medical service utilization, and work absence days. Among employees who were diagnosed as having hypertension and had low costs before the index date, PDC was associated with small but significant increases in healthcare costs and work absence days. Better adherence to hypertension-related medications was significantly associated with higher medical and drug costs, ypertension-specific medical costs, and the number of work absence days during the year following the index date. The effect of better adherence on other measured outcomes was not significant in this cohort with low prior cost, however.
Among employees with high prior cost, better adherence was significantly associated with lower healthcare costs and with fewer inpatient hospital days and work absence days. However, better adherence in this cohort was associated with significantly higher hypertension-specific medical costs. Although the effect of better adherence on other outcomes was not significant in this cohort, all other outcomes except for 2-year probability of a cardiovascular procedure tended to decrease as adherence increased.
The pattern of association of adherence with cost and hospitalization outcomes in the group with high prior cost is consistent with the results of an earlier study18 that demonstrated reductions in overall healthcare costs and hospitalization rates with increased drug utilization. The pattern of association observed in the group with low prior cost is a new finding. The earlier study did not divide the cohort by prior cost or health status but analyzed the total population with hypertension. In the present study, prior cost had a statistically significant effect on most outcomes measured and may be a proxy for severity of illness before diagnosis or treatment. Therefore, the observed difference in patterns of association of antihypertensive medication adherence with most outcomes between cohorts with low versus high prior costs is not surprising.
The association between PDC and significant health events did not show the expected pattern based on the literature’s indications that pharmacologic treatment of hypertension reduces the risk of stroke, coronary events, and heart failure. 9,11-17 A 2-year time frame may be inadequate to measure the effect of better hypertension management. Also, because the measurement approach assessed adverse health events in much the same period as medication adherence, the positive association between PDC and event likelihood could, in some cases, reflect a sudden increase in compliance following a frightening medical problem. Furthermore, this study did not permit the analysis of clinical variables such as blood pressure readings.
This study adds to the literature on the association between medication adherence and medical and drug costs, health service utilization, and clinical outcomes by measuring more outcomes than prior studies. These include the number of work absence days, the number of inpatient hospital days, and the number of medical services, as well as probabilities of workers’ compensation claims, stroke, myocardial infarction, and cardiovascular procedures.
The study cohort and subcohorts were defined based on retrospective medical and drug claims data; therefore, causal relationships between adherence and outcomes cannot be presumed in this study. Also, there are some inherent drawbacks to the use of medical claims data when measuring medical service utilization and medical and drug costs such as the possibility that ICD-9-CM codes on medical claims do not accurately reflect a patient’s diagnosis. Finally, the lack of information about an employee’s duration of hypertension or severity is a limitation of retrospective analyses. There is evidence that the severity of hypertension is an influential factor when evaluating the effect of PDC on outcomes.37 Unfortunately, the data did not include reliable indicators of blood pressure readings to control for hypertension severity when measuring the effect of PDC.
Medication adherence was associated with improvement in some short-term utilization measures among high-prior-cost employees. However, significant short-term improvement was not seen among low-prior-cost employees.
Author Affiliations: From Human Capital Management Services (WDL, KM, AKM, NLK), Cheyenne, WY; and Daiichi Sankyo, Inc (JP), Superior, CO.
Funding Source: Funding for this study was provided by Daiichi Sankyo, Inc.
Author Disclosures: Dr Lynch reports serving on an advisory board for Daiichi Sankyo, Inc, and receiving lecture fees from that company. Dr Pesa is an employee of Daiichi Sankyo, Inc, the funder of the study, and reports owning stock in the company. The other authors (KM, AKM, NLK) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (WDL, AKM, NLK); acquisition of data (NLK); analysis and interpretation of data (WDL, KM, AKM, JP, NLK); drafting of the manuscript (WDL, KM, JP, NLK); critical revision of the manuscript for important intellectual content (JP, NLK); statistical analysis (KM, AKM, NLK); obtaining funding (JP); administrative, technical, or logistic support (JP); and supervision (NLK).
Address correspondence to: Nathan L. Kleinman, PhD, Human Capital Management Services, 1800 Carey Ave, Ste 300, Cheyenne, WY 82001. E-mail: firstname.lastname@example.org.
1. American Heart Association. Heart Disease & Stroke Statistics: 2009 Update At-a-Glance. Dallas, TX: American Heart Association; 2009. http://www.americanheart.org/downloadableheart/1240250946756LS-1982%20Heart%20and%20Stroke%20Update.042009.pdf. Accessed July 27, 2009.
2. Fields LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, Sorlie P. The burden of adult hypertension in the United States 1999 to 2000: a rising tide. Hypertension. 2004;44(4):398-404.
3. Greenland P, Knoll MD, Stamler J, et al. Major risk factors as antecedents of fatal and nonfatal coronary heart disease events. JAMA. 2003;290(7):891-897.
4. Levy D, Larson MG, Vasan RS, Kannel WB, Ho KK. The progression from hypertension to congestive heart failure. JAMA. 1996;275(20):1557-1562.
5. MacMahon S, Rodgers A. Blood pressure, antihypertensive treatment and stroke risk. J Hypertens Suppl. 1994;12(10):S5-S14.
6. Hodgson TA, Cai L. Medical care expenditures for hypertension, its complications, and its comorbidities. Med Care. 2001;39(6):599-615.
7. American Heart Association. High blood pressure statistics. http://www.americanheart.org/presenter.jhtml?identifier=4621. Accessed October 10, 2008.
8. Eaddy MT, Shah M, Lunacsek O, Stanford RH. The burden of illness of hypertension and comorbid diabetes. Curr Med Res Opin. 2008;24(9):2501-2507.
9. National Heart, Lung, and Blood Institute. The Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Bethesda, MD: National Institutes of Health; 1997. NIH publication 98-4080 ed. http://www.ncbi.nlm.nih.gov/books/bv.fcgi?rid=hbp. Accessed October 10, 2008.
10. Taira DA, Wong KS, Frech-Tamas F, Chung RS. Copayment level and compliance with antihypertensive medication: analysis and policy implications for managed care. Am J Manag Care. 2006;12(11):678-683.
11. Collins R, Peto R, MacMahon S, et al. Blood pressure, stroke, and coronary heart disease, part 2: short-term reductions in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet. 1990;335(8693):827-838.
12. Dahlöf B, Lindholm LH, Hansson L, Scherstén B, Ekbom T, Wester PO. Morbidity and mortality in the Swedish Trial in Old Patients With Hypertension (STOP-Hypertension). Lancet. 1991;338(8778):1281-1285.
13. Ambrosioni E. Pharmacoeconomics of hypertension management: the place of combination therapy. Pharmacoeconomics. 2001;19(4):337-347.
14. Elliott WJ. The economic impact of hypertension. J Clin Hypertens (Greenwich). 2003;5(3 suppl 2):3-13.
15. Psaty BM, Lumley T, Furberg CD, et al. Health outcomes associated with various antihypertensive therapies used as first-line agents: a
network meta-analysis. JAMA. 2003;289(19):2534-2544.
16. MacMahon S, Neal B, Rodgers A, Chalmers J. The PROGRESS trial three years later: time for more action, less distraction [published correction appears in BMJ. 2004;329(7474):1073]. BMJ. 2004;329(7472):970-971.
17. Degli Esposti L, Valpiani G. Pharmacoeconomic burden of undertreating hypertension. Pharmacoeconomics. 2004;22(14):907-928.
18. Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care. 2005;43(6):521-530.
19. Ambrosioni E. Pharmacoeconomic challenges in disease management of hypertension. J Hypertens Suppl. 2001;19(3):S33-S40.
20. Mar J, Rodríguez-Artalejo F. Which is more important for the efficiency of hypertension treatment: hypertension stage, type of drug or therapeutic compliance? J Hypertens. 2001;19(1):149-155.
21. McCombs JS, Nichol MB, Newman CM, Sclar DA. The costs of interrupting antihypertensive drug therapy in a Medicaid population. Med Care. 1994;32(3):214-226.
22. Rizzo JA, Simons WR. Variations in compliance among hypertensive patients by drug class: implications for health care costs. Clin Ther. 1997;19(6):1446-1457, 1424-1425.
23. Nichol MB, Venturini F, Sung JC. A critical evaluation of the methodology of the literature on medication compliance. Ann Pharmacother. 1999;33(5):531-540.
24. Schlenk EA, Burke LE, Rand C. Behavioral strategies to improve medication-taking compliance. In: Burke LE, Ockene IS, eds. Compliance in Healthcare and Research. Armonk, NY: Futura Publishing Co; 2001:57-70.
25. Jokisalo E, Enlund H, Halonen P, Takala J, Kumpusalo E. Factors related to poor control of blood pressure with antihypertensive drug therapy. Blood Press. 2003;12(1):49-55.
26. Brunenberg DE, Wetzels GE, Nelemans PJ, et al. Cost effectiveness of an adherence-improving programme in hypertensive patients. Pharmacoeconomics. 2007;25(3):239-251.
27. Miller NH. Compliance with treatment regimens in chronic asymptomatic diseases. Am J Med. 1997;102(2A):43-49.
28. Groban MD, Evans RM, Edgren B, et al. Clinical benefits and cost reduction associated with a comprehensive asthma management programme at a managed care organisation. Dis Manag Health Outcomes. 1998;4:93-100.
29. Thompson D, Hylan TR, McMullen W, Romeis ME, Buesching D, Oster G. Predictors of a medical-offset effect among patients receiving antidepressant therapy. Am J Psychiatry. 1998;155(6):824-827.
30. McCulloch D. Managing diabetes for improved health and economic outcomes. Am J Manag Care. 2000;6(21 suppl):S1089-S1095.
31. Balkrishnan R, Christensen DB, Bowton DL. Self-reported health status, prophylactic medication use, and healthcare costs in older adults with asthma. J Am Geriatr Soc. 2002;50(5):924-929.
32. Soumerai SB, Avorn J, Ross-Degnan D, Gortmaker S. Payment restrictions for prescription drugs under Medicaid: effects on therapy, cost, and equity. N Engl J Med. 1987;317(9):550-556.
33. Soumerai SB, Lipton HL. Computer-based drug-utilization review: risk, benefit, or boondoggle? N Engl J Med. 1995;332(24):1641-1645.
34. Stuart B, Zacker C. Who bears the burden of Medicaid drug copayment policies? Health Aff (Millwood). 1999;18(2):201-212.
35. Tamblyn R, Laprise R, Hanley JA, et al. Adverse events associated with prescription drug cost-sharing among poor and elderly persons. JAMA. 2001;285(4):421-429.
36. Heisler M, Langa KM, Eby EL, Fendrick AM, Kabeto MU, Piette JD. The health effects of restricting prescription medication use because of cost. Med Care. 2004;42(7):626-634.
37. Chobanian AV, Bakris GL, Black HR, et al. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report [published correction appears in JAMA. 2003;290(2):197]. JAMA. 2003;289(19):2560-2572.