Antihypertensive Drug Therapies and the Risk of Ischemic Stroke FREE

THE PRIMARY purpose of the pharmacological treatment of hypertension is to prevent major cardiovascular complications such as stroke. The 4 most widely used antihypertensive drug classes include diuretics, β-blockers, calcium channel blockers, and angiotensin-converting enzyme (ACE) inhibitors. Randomized clinical trials (RCTs) have not demonstrated major differences between these antihypertensive drug classes with regard to lowering of blood pressure,^1- 4 quality of life,^3,5 or regression of left ventricular mass.^6- 8 Results of recent meta-analyses of antihypertensive drug treatment compared with placebo suggest that low-dose diuretic therapy is effective in reducing the risk of stroke, coronary heart disease, congestive heart failure, and total mortality, whereas β-blockers prevent stroke and congestive heart failure but seem less effective in preventing coronary heart disease,⁹ especially in the elderly.¹⁰ The Systolic Hypertension in Europe Trial has demonstrated that compared with placebo, the calcium channel blocker nitrendipine reduces the incidence of stroke in older adults with isolated systolic hypertension.¹¹

These findings from placebo-controlled trials suggest that β-blockers, diuretics, and calcium channel blockers are more effective than placebo in the primary prevention of stroke. Less consistent are the findings from individual RCTs with separate treatment arms for β-blockers and diuretics.^12- 15 Although findings from the International Prospective Primary Prevention Study in Hypertension trial¹⁴ suggested similar reductions in stroke incidence with β-blockers and diuretics, the Heart Attack Primary Prevention in Hypertension (HAPPHY)¹⁵ and the Medical Research Council trials^12- 13 observed a larger reduction of stroke incidence with diuretic therapy than β-blocker therapy. The Captopril Prevention Project trial¹⁶ reported an increased risk of stroke in subjects receiving captopril compared with β-blockers or diuretics. However, the recently completed Swedish Trial in Old Patients with Hypertension-2 Study¹⁷ demonstrated that in hypertensive patients older than 70 years, calcium antagonists and ACE inhibitors apparently did not differ from conventional therapy (β-blockers or diuretics) with regard to the reduction of the incidence of cardiovascular morbidity and mortality, including stroke. The Captopril Prevention Project and the Swedish Trial in Old Patients with Hypertension-2 trials were not designed to distinguish between β-blockers and diuretics.

To assess the association between antihypertensive drug therapy and incident ischemic stroke, we conducted a population-based case-control study among pharmacologically treated hypertensive patients.

The setting was the Group Health Cooperative of Puget Sound (GHC), a large staff-model health maintenance organization with more than 400 000 members in western Washington State.

Case patients included GHC enrollees, aged 30 to 79 years, who were treated pharmacologically for hypertension and who sustained an incident fatal or nonfatal ischemic stroke between July 1, 1989, and December 31, 1996. Potential ischemic stroke cases were identified from computerized GHC hospital discharge diagnoses, Washington State death files, and the billing records for GHC enrollees who received medical care or services from non-GHC providers. Diagnostic criteria for ischemic stroke were adopted from the Cardiovascular Health Study.¹⁸ These criteria included (1) rapid onset of neurologic deficit or subarachnoid hemorrhage, (2) deficit persisting for longer than 24 hours unless computed tomography or magnetic resonance imaging show evidence of permanent damage, and (3) no underlying brain trauma, tumor, or infection to cause symptoms. Control subjects were obtained from a companion study of risk factors for myocardial infarction at GHC.¹⁹ Controls were a randomly selected sample of these GHC enrollees who were treated pharmacologically for hypertension and were frequency matched to the myocardial infarction cases by sex, age (within decade), and calendar year.

Each subject was assigned an index date. For the cases, the index date was the date of the stroke; for controls, the index date was a randomly selected date within the calendar year for which they had been selected as controls. In most strata based on sex, 10-year age categories, and index year, the control-case ratio was larger than 3:1. We excluded cases and controls (1) who were enrollees for less than 1 year or who had fewer than 4 visits before their index dates, (2) who had had a previous stroke, (3) who had a diagnosis of congestive heart failure, (4) who did not have a diagnosis of hypertension in their medical record, and (5) whose stroke was a complication of a procedure or surgery. A history of stroke was assessed by medical record review. Subjects with a history of congestive heart failure were excluded because of concern about confounding by the indication of congestive heart failure for ACE inhibitors.

Information on demographics, health habits, cardiovascular risk factors, and comorbidities were abstracted from medical records or obtained from a telephone interview of consenting survivors. Abstraction of the information from the medical records was performed by trained research assistants who were aware of case-control status but unaware of the purpose of the study.

The GHC computerized pharmacy database was used to assess antihypertensive drug use. The pharmacy records contain information about the type, dose, and quantity of drug dispensed; the prescription fill date; and dosing instructions. When dosing instructions were missing from the pharmacy database, we used the instructions available in the medical record. The pharmacy data were searched for antihypertensive drug prescriptions immediately preceding the index date. When a subject who was assumed to be at least 80% compliant received enough pills to last until the index date, that person was classified as a potential current user on that date. This process was repeated to assess use at 30 and 60 days before the index date. A current user of antihypertensive drugs was defined as a user for at least 30 days before and on the index date. This definition excludes recent starters or switchers of antihypertensive drug therapies, whose drug course was first prescribed or changed within 30 days of the index date.

A subject was considered pharmacologically treated for hypertension when a recording of antihypertensive drug use for the indication of hypertension was present in the medical records and when the subject was classified as a current antihypertensive drug user at the index date according to data available in the computerized pharmacy database.

Complete data were uniformly available from the medical records for case-control status and medical conditions such as pharmacological treatment of hypertension, angina, and diabetes. In preliminary analysis of demographic and behavioral risk factor data, such as smoking, physical activity, race, marital status, and educational level, the agreement between medical record and self-reported measures (telephone interview) was good to excellent. Self-reported data, if available, were used for these variables; if not, then data from the medical record were used. Data were missing on smoking (1.1% of subjects), physical activity (8.0%), race (2.4%), educational level (27.0%), total cholesterol level (4.8%), duration of treatment for hypertension (10.5%), and pretreatment blood pressure (30.1%). We used an approximate Bayesian bootstrap method to impute missing values. This multiple imputation method is a modification of the hot-deck method and takes account of the imputation variability.²⁰ In sensitivity analysis, the results using multiple imputed data were similar to those in the analysis limited to subjects with complete data.

All statistical tests were 2-tailed. We used stratification and logistic regression to control for potential confounders of the association between antihypertensive drug therapy and ischemic stroke, and odds ratios to estimate the relative risk (RR). Data were analyzed using commercially available software (SAS, Version 6.12; SAS Institute, Cary, NC).

The association of ischemic stroke with antihypertensive drug therapies was assessed separately for subjects with and without clinical cardiovascular disease (CVD). We defined CVD as possible, probable, or definite diagnoses of angina, claudication, cardiac arrhythmias (including atrial or ventricular arrhythmia) or history of myocardial infarction, transient ischemic attack, coronary angioplasty, coronary bypass surgery, or carotid endarterectomy. First, we compared single-drug users of one of the major antihypertensive drug classes and users of major 2-drug combinations, with single-drug users of benzothiadiazide (thiazide) diuretics as the reference group. Second, we compared antihypertensive drug regimens that did not include a thiazide diuretic with regimens that included a thiazide diuretic, among single-drug users and users of 2 antihypertensive drugs from different drug classes.

During the study period, 611 treated hypertensive patients were hospitalized for or died out of the hospital of a first ischemic stroke. We also identified 3505 population-based controls who were eligible. We excluded 92 cases and 199 controls with congestive heart failure, 113 cases and 453 controls who were not sufficiently compliant with their antihypertensive drug regimens to be classified as current users, and 66 cases and 181 controls who recently started or switched any antihypertensive drug therapy. In total, 231 cases and 715 controls were excluded for 1 or more of these reasons. We included 380 ischemic stroke cases (21 fatal and 359 nonfatal) and 2790 controls who were treated pharmacologically for hypertension.

The clinical characteristics of cases and controls are summarized in Table 1. Compared with controls, cases were older and more likely to be men and had a higher systolic blood pressure at treatment and before treatment. A number of risk factors for ischemic stroke were more common among cases than controls.

Because of concern about confounding by indication, we compared characteristics of controls who were users of thiazide diuretics with controls who used other antihypertensive drugs separately for subjects with and without a history of CVD (Table 2). Among controls with no history of CVD, users of thiazide diuretics were slightly older and had a slightly longer history of treated hypertension, and fewer had diabetes than users of nonthiazides. Other risk factors for stroke were similar for users of thiazide diuretics compared with users of other antihypertensive drugs. Among control subjects with a history of CVD, users of thiazide diuretics were slightly younger, had a lower treated systolic blood pressure, and were less likely to have diabetes, angina, or a history of cardiovascular procedures than users of an antihypertensive drug regimen that did not include a thiazide diuretic.

Among 127 cases and 1566 controls free of CVD, the adjusted risk of ischemic stroke was 2.03- to 2.79-fold higher among users of single-drug therapy with β-blockers, calcium channel blockers, or ACE inhibitors than among users of a thiazide diuretic alone (Table 3). The use of β-blockers, calcium channel blockers, or ACE inhibitors in combination with a thiazide diuretic was not significantly associated with an increased risk of ischemic stroke compared with the use of a thiazide diuretic alone. The use of any 2 antihypertensive drugs not including a thiazide diuretic was associated with a 2.48-fold increase in the risk of ischemic stroke compared with the use of a thiazide diuretic alone. Among 186 cases and 912 controls with a history of CVD, the use of calcium channel blockers in combination with a thiazide diuretic or a nonthiazide 2-drug combination were each associated with an increased risk of ischemic stroke, compared with users of thiazide diuretics alone (Table 3). Use of other single antihypertensive drugs or 2-drug combinations, compared with the use of thiazide diuretics alone, was not significantly associated with an increased risk of ischemic stroke.

The second analysis was conducted among 348 cases and 2608 controls who used 1 or 2 antihypertensive drugs (Table 4). Among 142 cases and 1652 controls free of CVD, subjects who did not use a thiazide diuretic had, after adjustment for potential confounding factors, an increased risk of ischemic stroke compared with subjects who used a thiazide diuretic (RR, 1.85; 95% confidence interval [CI], 1.26-2.71). Among subjects with a history of clinical CVD who did not use a thiazide diuretic, the increased adjusted risk of ischemic stroke compared with users of thiazide diuretics was not statistically significant (RR, 1.25; 95% CI, 0.87-1.80).

Additional adjustment for educational level, marital status, duration of treated hypertension, treated diastolic blood pressure, pretreatment diastolic blood pressure, body mass index (calculated as weight in kilograms divided by the square of height in meters, physical activity, glucose and potassium levels, and use of alcohol had trivial effects on the findings.

Stratified analysis revealed no effect modification of the association between ischemic stroke and use of thiazide diuretics in subgroups based on sex, median age (70 years), presence of diabetes, smoking status, median pretreatment blood pressure (diastolic, 100 mm Hg; systolic, 170 mm Hg), median treated blood pressure (diastolic, 84 mm Hg; systolic, 152 mm Hg), median total serum cholesterol level (6.16 mmol/L [238 mg/dL]), and median serum potassium level (4.1 mmol/L). The results were virtually the same for the use of thiazide diuretics below the daily modal dose and above or at the daily modal dose (25 mg for hydrochlorothiazide, 25 mg for chlorthalidone, and 5 mg for metolazone) and for the use of thiazide diuretics alone or in combination with potassium-sparing agents.

In this population-based case-control study among pharmacologically treated hypertensive patients without clinically recognized CVD, the use of antihypertensive drug regimens that did not include a thiazide diuretic was associated with an 85% increased risk of ischemic stroke compared with the use of an antihypertensive drug regimen that included a thiazide diuretic. Even among users of 2 antihypertensive drugs, the use of nonthiazide antihypertensive drug regimens was associated with a higher risk of ischemic stroke (40%). This association persisted after adjustment for many potential confounding factors, and was consistent across a variety of subgroups. Among subjects with clinically manifest CVD, this association was less pronounced.

The International Prospective Primary Prevention Study in Hypertension trial, in which β-blocker therapy was compared with non–β-blocker (mostly diuretic) therapy, showed no significant difference in stroke incidence between the treatment groups (RR, 0.97; 95% CI, 0.64-1.47). However, this trial may not allow a valid comparison of stroke risk because 67% of the patients allocated to β-blocker therapy also received a diuretic. The HAPPHY trial found a nonsignificantly higher risk of stroke for β-blockers compared with diuretics (RR, 1.29; 95% CI, 0.84-1.83). Findings from the Medical Research Council trials clearly suggest a higher risk of stroke with β-blocker therapy than with diuretic therapy in middle-aged subjects (RR, 2.28; 95% CI, 1.31-3.96),¹² and a nonsignificantly higher risk in older subjects (RR, 1.23; 95% CI, 0.86-1.79).¹³ Results from the Captopril Prevention Project trial showed an increased risk of stroke with captopril therapy compared with β-blocker or diuretic therapy (RR, 1.25; 95% CI, 1.01-1.55). This increased risk of stroke may have been due to a failure of randomization.²¹ In a clinical trial designed to assess the effects of antihypertensive therapy on carotid atherosclerosis,²² subjects randomized to receive the calcium channel blocker isradipine had a higher rate of stroke events than those randomized to receive hydrochlorothiazide (RR, 2.00; 95% CI, 0.50-7.93). In the Swedish Trial in Old Patients with Hypertension-2 trial, subjects allocated to receive ACE inhibitors and calcium antagonists, respectively, had a similar risk of stroke compared with subjects allocated to receive β-blocker or diuretic therapy (respective RRs, 0.90 [95% CI, 0.74-1.08] and 0.88 [95% CI, 0.73-1.06]). These findings from RCTs tend to favor thiazide diuretics over other antihypertensive drug therapies for reduction of stroke risk and are consistent with the findings from our study. Additional support for a particular benefit of thiazide diuretics comes from the recent interim analysis of the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial.²³ Compared with patients who received chlorthalidone, patients treated with the α-blocker doxazosin mesylate had an increased risk of stroke (RR, 1.19; 95% CI, 1.01-1.40).

One possible explanation for these findings may lie in the fact that systolic blood pressure is more strongly associated with the occurrence of stroke than diastolic blood pressure,²⁴ and that thiazide diuretics may be more effective in lowering systolic blood pressure than other antihypertensive drugs, whereas the effect of thiazide diuretics on diastolic blood pressure is similar to that of other major antihypertensive drug classes.^{3- 4,12- 13,16,22- 23} Systolic blood pressure during treatment was also slightly lower among users of thiazide diuretics in this study. Adjustment for systolic blood pressure during treatment had little effect on our results. For example, among those receiving monotherapy, the risk of ischemic stroke associated with not using compared with using a thiazide diuretic decreased from 2.42 to 2.30. However, we were not able to test this hypothesis adequately, because many of the current users of thiazide diuretics were not using a thiazide diuretic at the time when their blood pressure during treatment was recorded.

A biological mechanism independent of blood pressure cannot be excluded, as suggested by a post hoc analysis of the Swedish Trial in Old Patients with Hypertension trial in which two thirds of the actively treated patients received a β-blocker and a thiazide diuretic.²⁵ After matching on achieved blood pressure and controlling for initial blood pressure, the subjects receiving active treatment had a 42% decreased risk of stroke compared with those receiving placebo (RR, 0.58; 95% CI, 0.35-0.98), suggesting a non–blood pressure mediated benefit of antihypertensive therapy with β-blockers or diuretics beyond that achieved by merely lowering blood pressure. Moreover, recently it was demonstrated that diuretics may have an additional therapeutic advantage by restoring nocturnal blood pressure decline in patients with sodium-sensitive hypertension.²⁶

The strengths of this observational study are the use of population-based case-control subjects, the completeness of case identification, the comparable ascertainment of potential confounding factors, and the use of pharmacy records to assess antihypertensive drug use in a comparable and unbiased fashion for cases and controls. We used restriction, stratification, and multivariate adjustment to minimize the influence of confounding.

An important limitation of this observational study is that antihypertensive drug treatment was not randomly assigned. Physicians and patients selected antihypertensive drug therapies, and this may have introduced bias. Despite adjustment for potential confounding factors, residual confounding due to incomplete or inaccurate measurement of covariates or unmeasured confounders cannot be excluded.

The preferred design to compare these antihypertensive drug therapies in terms of their risk or benefit with regard to cardiovascular outcomes is the controlled RCT. However, when clinical trial results are lacking or conflicting, well-designed observational studies can complement them. Furthermore, the highly selected nature of participants of RCTs and the strict, protocol-driven conditions under which trials are conducted may limit the generalizability of findings from RCTs to general practice. The common use of a large number of alternative antihypertensive drugs makes evaluation of these therapies in an observational setting feasible. The high degree of similarity in several important clinical characteristics between users of thiazide diuretics and users of other antihypertensive drugs among subjects without CVD suggests minimal confounding by those characteristics that were measurable.

This study suggests a particular benefit of thiazide diuretics in reducing the risk of ischemic stroke. Although the mechanism is not clear, the findings are consistent with those of previous RCTs. Ongoing large-scale clinical trials should help clarify this issue.²⁷ The Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure currently recommends diuretics and β-blockers.²⁸ In the absence of additional clinical trial evidence, the results of our study support the use of thiazide diuretics as first-line antihypertensive agents.

Accepted for publication July 10, 2000.

The research reported in this article was supported in part by grants HL40628, HL43201, and HL60739 from the National Heart, Lung, and Blood Institute, Bethesda, Md. Dr Psaty is a Merck/SER Clinical Epidemiology Fellow, sponsored by the Merck Co Foundation, Rahway, NJ, and the Society for Epidemiologic Research, Baltimore, Md. Dr Kaplan was a Howard Hughes Medical Institute Predoctoral Fellow when this work was conducted.

Corresponding author and reprints: Olaf H. Klungel, PharmD, PhD, Department of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute of Pharmaceutical Sciences, Sorbonnelann 16, 3584 CA Utrecht, the Netherlands (e-mail: o.h.klungel@pharm.uu.nl)