Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Predictors of Uncontrolled Hypertension in Ambulatory Patients Eric L. Knight, Rhonda L. Bohn, Philip S. Wang, Robert J. Glynn, Helen Mogun, Jerry Avorn Abstract—Hypertension remains poorly controlled in the United States. Improvement of its management will require an understanding of the patient characteristics and treatment factors associated with uncontrolled hypertension. We studied antihypertensive medication use, comorbidity, and blood pressure measurements for 525 hypertensive patients in 3 different healthcare systems over a 1-year period. We concomitantly conducted comprehensive patient interviews covering demographic factors, knowledge of hypertension and its treatment, and medication side effects. Ordinal logistic regression was used to identify factors associated with poor blood pressure control. Mean age of the patients was 65⫾11 years. Mean systolic blood pressure (SBP) was 143⫾15 mm Hg; and mean diastolic blood pressure (DBP), 80⫾9 mm Hg. Only 39% (203/525) of patients had mean blood pressure ⬍140/90 mm Hg during the study period; about half (257/525) had stage 1 hypertension (mean SBP 140 to 159 mm Hg and/or mean DBP 90 to 99 mm Hg), and 12% (65/525) had stage 2 or greater hypertension (SBP ⬎160 mm Hgand/or DBP ⬎100 mm Hg). Multivariate analysis revealed several independent predictors of poor control: older age, multi-drug regimens, lack of knowledge by the patient of their target SBP, and a report of antihypertensive drug side effects. Patients with angina had a higher likelihood of adequate blood pressure control. Fewer than 40% of the treated patients studied had a mean blood pressure ⱕ140/90 mm Hg, and specific patient-related factors appear to predict poor control. Some of these may be amenable to modification. Further identification of patients at risk of poor control can lead to targeted interventions to improve management. (Hypertension. 2001;38:809-814.) Key Words: hypertension 䡲 epidemiology 䡲 education H factors, including compliance behavior, the presence of comorbid illness, and patient knowledge, as well as patterns of treatment and systems of care. Much of the literature on blood pressure control is focused on antihypertensive medication adherence, and good adherence has been associated with good blood pressure control.7 However, blood pressure is also determined by other factors, including age, severity of disease, health habits, and intensity of care. The current study was designed to test several issues related to blood pressure control that have not been extensively examined in prior studies. First, we sought to assess the role of patient-specific factors in blood pressure control, as well as to explore the relationship between age and blood pressure control.8 –11 We also wanted to measure the relationship between patient knowledge of target blood pressure goals and blood pressure control, as lack of such knowledge has been associated with poor compliance with medications and clinic visits.12 Second, we wished to examine the impact of different healthcare systems on blood pressure control, an especially important issue at a time when systems of care are in such rapid flux. Systems that emphasize regular, ongoing primary ypertension is the second most common reason for an outpatient physician visit in the United States, accounting for approximately 30 million visits a year.1 However, in a recent national study, only 29% of people with hypertension had their blood pressure controlled ⬍140/90 mm Hg.2 A study of older men at 5 Department of Veterans Affairs sites found even worse blood pressure control, with fewer than 25% of patients having a blood pressure ⬍140/90 mm Hg, despite an average of ⬎6 hypertension-related visits per year.3 Similar results were found in a study of unionized healthcare workers with hypertension.4 The reasons for this epidemic of uncontrolled hypertension are unclear, and several explanations have been considered. In a study of innercity patients with uncontrolled hypertension, lack of a primary care physician and noncompliance were both associated with poor blood pressure control.5 There is also some evidence that intensity of medical therapy is related to blood pressure control; however, this is difficult to measure accurately. The cost of antihypertensive medication has also been linked to poor blood pressure control.6 In addition to the underlying pathophysiology, blood pressure control is potentially determined by multiple other Received October 19, 2000; first decision November 13, 2000; revision accepted March 21, 2001. From the Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Mass. Correspondence to Jerry Avorn, MD, Chief, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital, 221 Longwood Ave, Boston, MA 02115. E-mail [email protected] © 2001 American Heart Association, Inc. Hypertension is available at http://www.hypertensionaha.org 809 810 Hypertension October 2001 care and have blood pressure management guidelines in place should, in theory, have more success at controlling hypertension than other forms of healthcare delivery. However, the literature is unclear as to which systems of care provide superior care for patients with hypertension.13 Third, we sought to explore the relationship between comorbidity and blood pressure control. Patients with substantial comorbidity, especially when blood pressure is a contributing factor, might be more likely to have better blood pressure control because the impact of hypertension would be more apparent to them compared with those with asymptomatic hypertension and no comorbidity. For example, we have previously shown that patients with cardiac disease comply better with antihypertensive regimens, which would be expected to lead to better blood pressure control.14 Significant cardiac disease may also result in lower blood pressures secondary to cardiac dysfunction. Fourth, we wanted to explore whether perceived adverse drug effects would influence blood pressure control, because patients with side effects may be less likely to take their medications as directed.15 This hypothesis is indirectly supported by studies showing poorer compliance with hypertension therapy in patients prescribed thiazides compared with calcium channel blockers.14,16 In addition, patients who report adverse effects may have other characteristics (eg, dissatisfaction with the healthcare system) that may influence their blood pressure control. Methods The study was conducted at outpatient clinics in 3 sites: a not-forprofit managed care organization, a large tertiary care hospital, and a Veterans Affairs Medical Center (VAMC). The managed care organization provided comprehensive drug coverage with a $5 to $10 copayment. At the hospital clinic, most patients received drug coverage through Medicaid or free care. At the VAMC site, all patients received free prescription coverage. Patients were identified on the basis of antihypertensive medication use and a diagnosis of hypertension. Letters were sent to prospective subjects, inviting them to participate in the study and enabling them to opt out. All traceable patient identifiers were transformed to anonymous coded study numbers to protect confidentiality. Institutional review boards at all sites approved this study. At the managed care site, 49% of patients (250/515) agreed to participate, of whom 96% (240) had complete survey and chart data. At the tertiary care site, 40% (63/157) agreed to participate, of whom 70% (44) had complete survey and chart data. At the VAMC, 38% (247/658) agreed to participate, of whom 97% (241) had complete survey and chart data. Study participants were quite similar to nonparticipants for most characteristics, although participants at the hospital clinic were more likely to be taking ACE inhibitors than were nonparticipating patients, and participants at the VAMC site were more likely to be taking -blockers or ACE inhibitors than were those who declined to participate. We recorded information from all healthcare encounters during 1 calendar year, including date, systolic (SBP) and diastolic (DBP) blood pressures, symptoms, diagnoses, and medication use. The survey obtained data on age, race, gender, educational level, marital status, education by the physician, use of medication reminders, health habits, knowledge of target blood pressure, and side effects. For Spanish-speaking patients, the survey instrument was translated into Spanish, back-translated for accuracy, and administered by a medical interpreter. For patients taking combination drugs, compliance was counted in terms of each constituent drug class. Blood pressure categories were based on standard criteria17 (in mm Hg): (1) SBP ⬍140 and DBP ⬍90, (2) SBP ⱖ140 and ⬍160, and/or DBP ⱖ90 and ⬍100 (stage 1 hypertension), and (3) SBP ⱖ160 or DBP ⱖ100 (stage 2 or greater hypertension). A minimum of 2 blood pressure readings were required to classify a study subject. The outcome variable was blood pressure control defined in the 3 ordered categories noted above; we used a proportional odds model to evaluate potential determinants of blood pressure control.18 Univariate associations were tested by placing each variable into a separate proportional odds model. For categorical variables, referent variables were created. The proportional odds assumption was not violated for any variables on univariate analysis at a P value ⱕ0.05. Variables that were associated with the above-defined outcomes at a P value ⬍0.2 were entered into the multivariate model. Variables with a P value ⱕ0.05 were kept in the final multivariate model, as were gender, race, and site of care. We also examined interaction terms for potential entry into the model. All analyses were performed using SAS software. Results A total of 525 subjects completed the survey and had data available for a 1-year medical record review. Their mean age was 65⫾11 years, 64% were male, and 89% were white. Table 1 presents basic demographic information of the study population and other descriptive statistics by site. The managed care population was remarkable for the number of patients using only 1 antihypertensive medication during the study period (49%, compared with ⬍30% for the other sites, P⬍0.001). The VAMC site was notable for its expected preponderance of males and for the number of patients using multiple antihypertensive medications during the study period. The hospital clinic population was relatively young and ethnically diverse (30% white, versus ⬎90% white for the other sites, P⬍0.001). A mean of 7.4⫾5.0 SBP and DBP recordings was obtained for each subject. The mean SBP for the study population was 143⫾15 mm Hg , and the mean DBP was 80⫾9 mm Hg. Based on the mean BP during the study year, 203 patients were normotensive, 257 had stage I hypertension, and 65 had stage 2 or higher blood pressure. Table 2 presents the mean SBPs and DBPs in each category. On univariate analysis, we found the following characteristics to be associated with poor blood pressure control (Table 3): older age, female gender, calcium channel blocker therapy, the use of ⱖ2 antihypertensive medications during the study period, site of care, presence of arthritis, being widowed, lack of knowledge of appropriate target SBP, and experience of an adverse effect of an antihypertensive medication. The presence of angina and a history of myocardial infarction were both associated with better blood pressure control. With construction of a multivariable model as described above, we found the following factors to be significantly and independently associated with poor blood pressure control after adjusting for all other significant predictors: age ⱖ65, the use of ⱖ2 antihypertensive medications during the study period, lack of knowledge of appropriate target SBP, and the experience of a specific side effect attributed to an antihypertensive medication (see Table 4). Treated hypertensive patients age ⱖ65 were 2.5 times more likely to have higher blood pressure than patients ⬍55 years, patients on ⱖ4 hypertensive medications were almost 5 times as likely to have higher blood pressure than patients on 1 antihypertensive medication, patients who were not aware that their target Knight et al TABLE 1. Predictors of Uncontrolled Hypertension 811 Characteristics of Patients in the 3 Study Sites Variable Managed Care (n⫽240) VAMC (n⫽241) Hospital Clinic (n⫽44) 66.5⫾12.0 65.4⫾10.3 59.1⫾11.7 Mean age, y⫾SD Gender, % female 62 4 Mean SBP, mm Hg⫾SD 144⫾14.5 Mean DBP, mm Hg⫾SD 80 141.5⫾14.8 146.9⫾14.0 81.3⫾7.7 78.6⫾8.7 84.9⫾10.1 97 92 30 32 Race, % white Antihypertensive drug class, % Thiazide 38 27 -Blocker 43 59 75 Calcium channel blocker 29 53 34 ACE inhibitor 37 62 50 6 17 7 Other No. of antihypertensive medication classes during the study period, % 0 5 5 14 1 49 23 16 2 35 33 36 3 8 28 27 4 or 5 3 11 7 Angina, % 9 24 9 History of myocardial infarction, % 3 30 14 11 Congestive heart failure, % 5 14 History of stroke, % 6 14 9 Diabetes, % 20 37 32 Lack of knowledge of goal SBP, % 48 43 66 Reported side effects attributed to antihypertensive medication, % 21 45 48 SBP should be ⱕ140 mm Hg were 1.5 times more likely to be in a higher blood pressure category, and patients who attributed a side effect to a specific antihypertensive medication were twice as likely to be in a higher blood pressure category. There was a trend for managed care patients to have worse control than VAMC patients, but there was no evidence of any difference between the hospital clinic site and managed care organization (however, small numbers at the hospital site makes it difficult to assess this comparison). Angina was associated with relatively better blood pressure control, with these patients having only a third the risk of poor control compared with patients without angina. Antihypertensive drug class had no impact on blood pressure control. (Table 3). Discussion Only 39% of the 525 hypertensive patients we studied had a mean blood pressure ⬍140/90 mm Hg, despite ongoing TABLE 2. Mean SBP and DBP by Hypertension Category Blood Pressure Category No. of Patients Mean SBP⫾SD Mean DBP⫾SD Controlled hypertension 203 129⫾8 76⫾7 Stage 1 hypertension 261 148⫾6 82⫾8 61 170⫾7 85⫾11 Stage 2 hypertension or greater medical treatment. Older age was associated with poorer blood pressure control relative to that seen in younger treated patients. SBP increases with age, but DBP tends to level off at age 55 to 60.19 This increase in SBP is a strong predictor of cardiovascular disease with advancing age.20 The clinical benefits of treating systolic hypertension in older patients have now been demonstrated in several randomized, placebocontrolled trials, yet the implications of these trials may not be fully accepted in routine practice.8 –10 Even the oldest old, age ⬎80, appear to benefit from antihypertensive therapy.21 It is possible that despite the proven benefit of antihypertensive therapy in this age group, older patients are still treated less aggressively compared with younger patients. Lack of knowledge of appropriate SBP was also found to be a risk factor for poor blood pressure control. Patients who indicated in the interview that they were not aware that their target SBP should be ⱕ140 mm Hg were significantly more likely to have a higher mean blood pressure documented in their medical records. Such patients may have been less likely to take their medication, adopt healthy lifestyle changes, or see their physician if their blood pressure was outside the ideal range. However, we did not observe a relation between patient-reported education by their physician and blood pressure control. One explanation for this may be that physicians are more likely to educate patients whose blood pressure is 812 Hypertension October 2001 TABLE 3. Univariate Predictors of Poor Blood Pressure Control Variable Age ⬍55 y 55–64 y 65–74 y ⱖ75 y White Female gender Site Managed care VAMC Hospital clinic Antihypertensive drug class Thiazide diuretic -Blocker Calcium channel blocker ACE inhibitor Other Total No. of antihypertensive drugs 0 1 2 3 4 or 5 Alcohol abuse Angina History of myocardial infarction Congestive heart failure History of stroke Diabetes Arthritis Highest educational level Grade school or junior high school High school or vocational school College Postgraduate Marital status Single Married Widowed Divorced or separated Provider explained hypertension (referent⫽yes) Somewhat No Education about hypertension (referent⫽yes) Somewhat No Used a reminder to take blood pressure pills Exercise to lower blood pressure Weight loss to lower blood pressure Lack of knowledge of appropriate SBP Lack of knowledge of appropriate DBP Side effect attributed to antihypertensive medication Number (Total⫽525) Odds of Poor Control 95% Confidence Interval 98 111 205 111 468 194 referent 1.12 2.05 2.33 0.69 1.64 0.65–1.93 1.27–3.32 1.38–3.94 0.41–1.16 1.17–2.31 240 241 44 referent 0.71 1.86 0.50–1.00 1.03–3.37 171 280 212 261 59 referent 0.97 1.44 1.29 1.56 0.69–1.37 1.02–2.03 0.92–1.81 0.92–2.62 30 181 179 99 36 55 83 85 51 52 143 97 0.87 referent 1.63 1.83 3.31 0.69 0.44 0.55 0.62 1.40 0.86 1.64 77 284 137 27 0.93 referent 0.73 1.03 0.53–1.62 43 324 88 70 1.59 0.86–2.94 1.73 0.78 1.09–2.74 0.45–1.35 83 72 1.00 0.68 0.58–1.73 0.45–1.05 48 56 482 371 207 246 148 180 1.12 1.44 1.40 0.75 0.81 1.61 0.97 2.06 0.63–1.99 0.84–2.46 0.77–2.56 0.52–1.08 0.58–1.13 1.16–2.24 0.68–1.40 1.41–3.01 0.41–1.83 1.09–2.42 1.14–2.94 1.65–6.65 0.40–1.17 0.28–0.70 0.35–0.87 0.36–1.09 0.81–2.43 0.60–1.24 1.08–2.51 0.49–1.09 0.48–2.21 Knight et al TABLE 4. Multivariate Model of Poor Blood Pressure Control Variable Odds of 95% Confidence Poor Control Interval Age group* 55– 64 y 1.26 65–74 y 2.50 0.71–2.24 1.49–4.19 ⱖ75 y 2.56 1.45–4.52 VAMC 0.63 0.40–1.01 Hospital 0.94 0.42–2.11 0 0.90 0.41–1.98 2 1.91 1.25–2.91 3 2.53 1.50–4.28 4 or 5 Site** No. of antihypertensive drugs during the study period*** 4.70 2.22–9.95 Angina 0.33 0.20–0.56 Lack of knowledge of appropriate SBP 1.55 1.09–2.20 Attributed a specific side effect to a specific antihypertensive medication 2.06 1.41–3.01 The odds ratio represents the likelihood of being in a higher blood pressure category. The model is adjusted for gender and race. *Referent group: age⬍55; **referent group: managed care site; ***referent group: 1 antihypertensive drug during the study period. out of control. In some controlled trials, educational interventions have been shown to be effective in improving blood pressure control. In 1 study,22 patients in the intervention group showed significant reductions in both SBP and DBP compared with those of controls. Educational interventions have also been shown to improve compliance with blood pressure medication.23 However, not all trials of patient education have improved compliance or blood pressure control.24 The same authors did find improvement in both compliance and blood pressure control when patients were taught to measure their own blood pressures and chart it, along with their pill-taking schedule.25 The important difference between these 2 approaches may be the relevance of the educational message to the patient’s specific blood pressure levels (as opposed to general concepts about hypertension and risk). To assess the influence of the system of care, we controlled for multiple factors which differed across the 3 sites studied, including age, gender, race, antihypertensive medication class, and number of antihypertensive medications used. After adjusting for these and other factors, we found no differences between the study sites in adequacy of blood pressure control, though there was a trend for managed care patients to have worse control than VAMC patients. Thus, these findings provide no evidence that patients in managed care systems have better blood pressure control than patients in the VA system or patients in an urban, academic practice. Other studies have found similar results. For example, in the Medical Outcomes Study there was an equivalent lowering of baseline SBP after 2 years in managed care patients versus fee-for-service patients.13 In an earlier study, patients in a managed care organization had slightly better blood pressure Predictors of Uncontrolled Hypertension 813 control than patients receiving care in a fee-for-service setting.26 However, this study was limited in that it excluded patients with significant comorbidities such as ischemic heart disease, diabetes, and chronic lung disease. In the present study, which included patients with these common conditions, none of the 3 systems of care performed admirably well in managing hypertension. The impact of comorbidity on hypertension control has not been examined extensively in large studies. We found angina to be associated with better blood pressure control. This may be due to improved compliance in these patients, more aggressive treatment, or the direct effect of cardiovascular disease.14,27,28 There was also a trend for patients with congestive heart failure or a history of myocardial infarction to have better blood pressure control. Although the study sample included a large number of diabetic patients, diabetes was not associated with better control despite published recommendations suggesting that these patients should be particularly well controlled, with a target of ⬍130/85 mm Hg.29 We also found a relationship between the number of antihypertensive drugs used during the study period and blood pressure control, although this measure has some intrinsic limitations. For example, a patient who was switched from thiazide monotherapy to -blocker monotherapy after 1 month would have the same number of medications used during the study year as a patient who was maintained on both a thiazide and a -blocker throughout the study period. The observed relationship between increased number of antihypertensive drugs and poorer blood pressure control could be explained by several factors. First, patients whose blood pressure is more difficult to control are likely to be treated with multiple drugs. Thus, this measurement may be a consequence of poor blood pressure control. Second, patients who frequently are switched from 1 drug class to another may also have poorer blood pressure control, because they might have a greater percentage of days uncovered. Third, frequent switching could also be an effort to eliminate side effects, though the identification of a specific side effect was not highly correlated with the number of antihypertensive medications during the study period. Fourth, poor control may also be a reason to switch medication class. Another potentially important observation was the relationship between poor blood pressure control and adverse events attributed to antihypertensive medications. We have shown previously that patients initially prescribed ACE inhibitors or calcium channel blockers for hypertension were more likely to adhere to therapy than patients prescribed thiazides, possibly because the former medications may have fewer side effects.14 Patients who report adverse effects may also differ in terms of unmeasured factors which are also associated with uncontrolled hypertension. In the present study, no specific medications were associated with poorer versus better blood pressure control on multivariate analysis. Several limitations of this study deserve mention. The study design was observational, so each patient was managed at the discretion of his or her physician. Therefore, based on this study, it is difficult to impute causality with certainty for 814 Hypertension October 2001 any of the factors associated with poor blood pressure control. In addition, despite the comparability of respondents and nonrespondents in terms of basic demographic characteristics and medication use, it is possible that the absence of data from patients who declined to participate could have affected the findings and their ability to be generalized. This study provides a framework for identifying hypertensive patients who are at high risk of poor control, and many of the factors identified may be amenable to improvement. Older patients can be targeted for greater attention to blood pressure control, particularly in light of the evidence for improvement in clinical outcomes with hypertension therapy in this population. Patients with poor knowledge of the goal of their hypertension therapy should be informed about their target blood pressure, to enable them to participate more fully in their own management. Finally, clinicians should discuss with patients the potential adverse effects of their hypertension therapy, seek out the presence of such symptoms in routine follow-up history-taking, and when they occur, modify therapy. These data provide further evidence that poor blood pressure control is common in a variety of health care settings, and that patients at particular risk of poor control can be identified. Targeted interventions to improve management in such patients could make a substantial difference in stemming the epidemic of poorly controlled hypertension in the United States. Acknowledgments This work was funded by a research grant from the Outcomes Research Department of the Bristol-Myers Squibb Pharmaceutical Research Institute. Dr Knight was supported in part by grants from The Medical Foundation and the Claude Pepper Older Americans Independence Center (OAIC) at the Harvard Medical School Division on Aging. The authors would like to thank Sherry Hawley, Laura Van Houten, and Asra Warsi for their extensive time and effort performing the surveys and data collection and organizing the data for these analyses. References 1. National Ambulatory Medical Care Survey: 1997 Summary. Atlanta, Ga: Centers for Disease Control and Prevention; 1999. Advance data, No. 305. 2. Mulrow PJ. Detection and control of hypertension in the population: the United States experience. Am J Hypertens. 1998;11(pt 1):744 –746. 3. Berlowitz DR, Ash AS, Hickey EC, Friedman RH, Glickman M, Kader B, Moskowitz MA. Inadequate management of blood pressure in a hypertensive population. N Engl J Med. 1998;339:1957–1963. 4. Stockwell DH, Madhaven S, Cohen H, Gibson G, Alerman MH. The determinants of hypertension awareness, treatment, and control in an insured population. Am J Public Health. 1994;84:1768 –1774. 5. Shea S, Misra D, Ehrlich MH, Field L, Francis CK. Predisposing factors for severe, uncontrolled hypertension in an inner-city minority population. N Engl J Med. 1992;327:776 –781. 6. Ahluwalia JS, McNagny SE, Rask KJ. Correlates of controlled hypertension in indigent, inner-city hypertensive patients. J Gen Intern Med. 1997;12:7–14. 7. Gonzalez-Fernandez RA, Rivera M, Torres D, Quiles J, Jackson A. Usefulness of a systemic hypertension in-hospital educational program. Am J Cardiol. 1990;65:1384 –1386. 8. Amery A, Birkenhager W, Brixko P, Bulpitt C, Clement D, Deruyttere M, De Schaepdryver A, Dollery C, Fagard R, Forette F, et al. Mortality and morbidity results from the European Working Party on High Blood Pressure in the Elderly trial. Lancet. 1985;8442:1349 –1354. 9. Prevention of stroke by antihypertensive drug treatment in older patients with isolated systolic hypertension: final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA. 1991;265:3255–3264. 10. Staessen JA, Fagard R, Thijs L, Celis H, Arabidze GG, Birkenhager WH, Bulpitt CJ, de Leeuw PW, Dollery CT, Fletcher AE, Forette F, Leonetti G, Nachev C, O’Brien ET, Rosenfeld J, Rodicio JL, Tuomilehto J, Zanchetti A. Randomized double-blind comparison of placebo and active treatment for older patients with isolated systolic hypertension: the Systolic Hypertension in Europe (Syst-EUR) Trial Investigators. Lancet. 1997;350:757–764. 11. Burt VL, Cutler JA, Higgins M, Horan MJ, Labarthe D, Whelton P, Brown C, Roccella EJ. Trends in the prevalence awareness, treatment, and control of hypertension in the Adult US Population: data from the health examination surveys, 1960 to 1991. Hypertension. 1995;26:60 – 69. 12. Balazovjech I, Hnilica P Jr. Compliance with antihypertensive treatment in consultation rooms for hypertensive patients. J Hum Hypertens. 1993; 7:581–583. 13. Greenfield S, Rogers W, Mangotich M, Carney M, Tarlov AR. Outcomes of patients with hypertension and non–insulin-dependent diabetes mellitus treated by different systems and specialties. JAMA. 1995;274: 1436 –1444. 14. Monane M, Bohn RL, Gurwitz JH, Glynn RJ, Levin R, Avorn J. The effects of initial drug choice and comorbidity on antihypertensive therapy compliance: results from a population-based study in the elderly. Am J Hypertens. 1997;10:697–704. 15. Shaw E, Anderson JG, Maloney M, Jay SJ, Fagan D. Factors associated with noncompliance of patients taking antihypertensive medications. Hosp Pharm. 1995;30:201–207. 16. Bailey JE, Lee MD, Somes GW, Graham RL. Risk factors for antihypertensive medication refill failure by patients under Medicaid managed care. Clin Ther. 1996;18:1252–1262. 17. The sixth report of the Joint National Committee on prevention, detection, evaluation, and treatment of high blood pressure. Arch Intern Med. 1997;157:2413–2444. 18. McCullagh P. Regression models for ordinal data. J Roy Statist Soc Ser B. 1980;42:109 –142. 19. Kannel WB, Gordon T. Evaluation of cardiovascular risk in the elderly: the Framingham study. Bull NY Acad Med. 1978;54:573–591. 20. Glynn RJ, Field TS, Rosner B, Hebert PR, Taylor JO, Hennekens CH. Evidence for a positive linear relation between blood pressure and mortality in elderly people. Lancet. 1995;345:825– 829. 21. Gueyffier F, Bulpitt C, Boissel JP, Schron E, Ekbom T, Fagard R, Casiglia E, Kerlikowske K, Coope J. Antihypertensive drugs in very old people: a subgroup meta-analysis of randomised controlled trials: INDANA Group. Lancet 1999;353:793–796. 22. Gonzalez-Fernandez RA, Rivera M, Torres D, Quiles J, Jackson A. Usefulness of a systemic hypertension in-hospital educational program. Am J Cardiol. 1990;65:1384 –1386. 23. Roter DL, Hall JA, Mersica R, Nordstrom B, Cretin D, Svarstad B. Effectiveness of interventions to improve patient compliance: a metaanalysis. Med Care. 1998;36:1138 –1161. 24. Sackett DL, Haynes RB, Gibson ES, Hackett BC, Taylor DW, Roberts RS, Johnson AL. Randomised clinical trial of strategies for improving medication compliance in priomary hypertension. Lancet. 1975;1: 1205–1207. 25. Haynes RB, Sackett DL, Gibson ES, Taylor DW, Hackett BC, Roberts RS, Johnson AL. Improvement of medication compliance in uncontrolled hypertension. Lancet. 1976;1:1265–1268. 26. Udvarhelyi IS, Jennison K, Phillips RS, Epstein AM. Comparison of the quality of ambulatory care fee-for-service and prepaid patients. Ann Int Med. 1991;115:394 – 400. 27. Di Bari M, Salti F, Nardi M, Pahor M, De Fusco C, Tonon E, Ungar A, Pini R, Masotti G, Marchionni N. Undertreatment of hypertension in community-dwelling older adults: a drug-utilization study in Dicomano, Italy. J Hypertens. 1999;17:1633–1640. 28. Witteman JC, Grobbee De, Valkenburg HA, van Hemert AM, Stijen T, Burger H, Hoffman A. J-shaped relation between change in diastolic blood pressure and progression of aortic atherosclerosis. Lancet. 1994; 343:504 –507. 29. The fifth report of the Joint National committee on detection, evaluation, and treatment of high blood pressure (JNC V). Arch Intern Med. 1993; 153:154 –183.