WO2004056360A1 - Methods and dosage forms for reducing heart attacks in a hypertensive individual with a diuretic or a diuretic and an ace inhibitor combination - Google Patents

Abstract

The present invention relates to methods and dosage forms for reducing and/or preventing the incidence of cardiovascular disease, including heart attacks, in individuals who are at risk, such as those individuals suffering from hypertension. The treatments and dosage forms of the present invention concern the administration of a diuretic, such as a thiazide diuretic like Thalitone®, either alone or in combination with an ACE-inhibitor, such as ramipril or ramiprilat like Altace®, to reduce and/or prevent the incidence of cardiovascular disease, namely, heart attacks or failure, in individuals who suffer from hypertension or are at risk.

Description

METHODS AND DOSAGE FORMS FOR REDUCING HEART ATTACKS IN A HYPERTENSIVE INDIVIDUAL WITH A DIURETIC OR A DIURETIC AND AN ACE INHIBITOR COMBINATION

This application claims the benefit of U.S. Provisional Patent Application Serial No. 60/433,845, which was filed on December 16, 2002.

FIELD OF THE INVENTION

The present invention relates to methods and dosage forms for reducing and/ or preventing the incidence of cardiovascular disease, including heart attacks, in individuals who are at risk, such as those individuals suffering from hypertension. The treatments and dosage forms of the present invention concern the administration of a diuretic, such as a thiazide diuretic like Thalitone®, either alone or in combination with an ACE-inhibitor, such as ramipril or ramiprilat like Altace®, to reduce and/or prevent the incidence of cardiovascular disease, namely, heart attacks or failure, in individuals who suffer from hypertension or are at risk.

SUMMARY OF THE INVENTION

It has now been discovered that the incidence of cardiovascular disease, including heart attacks, in individuals, especially those who are at risks for hypertension, can be reduced, if not ekminated by administering to such individuals and effective amount of a diuretic, such as a thiazide diuretic like chlorthalidone, either alone or in combination with an angiotensin-converting enzyme inhibitor ("ACE-Inhibitor"), such as ramipril or ramiprilat. When the diuretic is co-administered with the ACE-inhibitor, they may be administered in a single dosage form, such as a tablet, liquid, caplet or capsule, or in separate individual dosage forms. A preferred diuretic for use in accordance with the present invention is thalidone, such as Thalitone®, administered as a single daily dose in an amount of 15mg, 30mg, 45mg, 50mg, 60mg or more. A preferred ACE- inhibitor for use in accordance with the present invention is ramipril, such as Altace®, administered as a single daily dose in an amount of 1.25mg, 2.5mg, 5mg, lOmg or more. Quite unexpectedly, it has now been discovered that the therapeutic effect of a diuretic in reducing or eliminating cardiovascular disease, including heart failure, in individuals at risk is much greater as compared to such individuals who are treated with a beta-adrenergic blocker, like propanolol and atenolol, a calcium- channel blocker, an ACE-inhibitor, such as Lisinopril, a vasodialator, such as hydralazine, a central agonists, such as oral clonidine or methyldopa, reserpine or an alphai-blocker, like prazosin and doxazosin. Even more unexpectedly, the therapeutic effect of the combination of a diuretic, such as a thizaide, like Thalitone®, and an ACE-inhibitor, such as Altace®, is far superior to the individual therapeutic effect achieved in such individuals when treated with only a diuretic or a beta-adrenergic blocker, like propanolol and atenolol, a calcium-channel blocker, an ACE-inhibitor, such as Lisinopril, a vasodialator, such as hydralazine, a central agonists, such as oral clonidine or methyldopa, reserpine or an alphai-blocker, like prazosin and doxazosin.

While the present invention has been described in the context of preferred embodiments and examples, it wiU be readily apparent to those skilled in the art that other modifications and variations can be made therein without departing from the spirit or scope of the present invention. Accordingly, it is not intended that the present invention be limited to the specifics of the foregoing description of the preferred embodiments and examples, but rather as being limited only by the scope of the invention as defined in the claims appended hereto.

Overview

This protocol describes a practice-based, randomized, clinical trial of antihypertensive pharmacologic treatment and, in a specific subset, cholesterol-lowering, in 40,000 high-risk hypertensive patients, including at least 55% African- Americans (self-described "black"). The purpose of the antihypertensive trial component is to determine whether the combined incidence of fatal coronary heart disease (CHD) and non-fatal myocardial infarction differs between diuretic (chlorthalidone) treatment and three alternative antihypertensive pharmacologic treatments — a calcium antagonist (amlodipine). an ACE inhibitor (lisinopril), and an alpha adrenergic blocker - (doxazosin)*. Because of the established benefit of antihypertensive treatment in reduction of stroke, total morbidity and mortality from cardiovascular diseases, and all-causes mortality, the antihypertensive trial component will not include a placebo or no-treatment control group. The purpose of the cholesterol-lowering trial component is to determine whether lowering serum cholesterol in moderately hypercholesterolemic men and women aged 55 years and older with the 3-hydroxymethylglutaryl coenzyme A (HMG CoA) reductase inhibitor pravastatin will reduce aD-cause mortality as compared to a control group receiving "usual care".

Secondary objectives of both trial components are to compare the effects of their respective treatment regimens on cardiovascular mortality, major morbidity, health costs, and health-related quality of life. Additional secondary objectives of the antihypertensive trial are to compare the effects of alternative treatments on all-cause mortality and on major hypertension- related morbidity such as incidence and regression of left ventricular hypertrophy and progressive renal dysfunction. Also the effect of the antihypertensive regimens on the aforementioned primary and secondary outcomes will be assess in key subgroups [over age 65, women, African-Americans, type LC diabetics]. Also additional secondary objectives of the lipid- lowering trial are to assess the long-term safety of HMG CoA reductase inhibitors in men and women aged 55 years and above (particularly with regard to mortality from non-cardiovascular causes), the effect of lipid-lowering on cancer incidence and mortality, and the effect of lipid lowering on the combined incidence of fatal CHD and non-fatal myocardial infarction, especially in key subgroups [over age 65, women, African- Americans, type II diabetics]. Also, because this component of tire trial will not be blinded, the incidence of myocardial infarction based on centrally coded changes in the biennial study ECG will be looked at as an end point. The mean duration of the trial is expected to be 6.0 years, ranging from 4.2 years (for the last patient entered) to 8 years (for the first patient entered).

To maximize statistical power for the primary hypotheses of the antihypertensive trial, i.e., the comparison of each alternative drug regimen to diuretic, 1.7 times as many patients will be-assigned to its diuretic arm as to each of its other three arms (Table Ll). It is anticipated that half of ALLHAT participants will be randomized to both trial components and that half will be randomized only to the antihypertensive trial component.

* On January 24. 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000:283:1967-1975. Table Ll: Design of ALLHAT

Because of the prohibitive cost of incorporating so many participants in a traditional clinical trial structure employing independently funded clinics, this trial will adopt an organizational model using patients recruited through approximately 600 primary care and hypertension practices, clinics, and health centers, each contributing an average of 67 patients. Department of Veterans Affairs (VA) hypertension clinics are expected to comprise approximately 70 of these clinical sites and to contribute about 20% of the study patients. Forms will be kept to a minimum, and few clinical procedures not performed for routine patient care will be required.

* On January 24, 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000;283: 1967-1975. II. Background and Rationale

A. Antihypertensive Trial

1) Hypertension and Coronary Heart Disease (CHD)

An estimated 50 million people in the U.S. have elevated blood pressure (systolic blood pressure, SBP, > 140 mmHg and or diastolic blood pressure, DBP, > 90 m Hg) or are taking antihypertensive medication. In 1983, 23 million people were taking antihypertensive drugs [1]. The ambulatory care cost of treating hypertension in the U.S. was estimated at $7.5 billion for 1986 [2]. These costs in large part are determined by the costs of the agents used, which, given the number of patients treated, have substantial economic implications. All other factors remaining constant, the incremental cost of treating 25 million patients with a drug costing $100 per patient-year of therapy compared to one costing $500 per patient-year is $10 billion.

Despite the known etiologic relationship of hypertension to CHD, large-scale randomized clinical trials in mild to moderate hypertension (DBP 90-114 mmHg) in largely -middle-aged subjects failed to demonstrate conclusively that antihypertensive drug treatment reduces the occurrence of CHD death or non-fatal myocardial infarction. The pooled results of nine such trials, employing primarily thiazide-like diuretics and involving over 43,000 subjects, suggest a 9% benefit, with 95% confidence limits consistent with a 19% benefit or a 1% adverse outcome [3]. This observed treatment effect compares with a maximum predicted effect on CHD of approximately 23% for an equivalent BP difference, as derived from epidemiologic data. In contrast, the observed beneficial effect on stroke in these trials, 36%, is almost exactly that which would be predicted from epidemiologic data [4]. A more recent overview [5] of 14 trials in participants with all levels of hypertension estimated a somewhat larger effect (14% benefit, 95%o confidence interval, 4-22%). While there is reason to suspect that this may be an over-estimate of the benefit, these overviews do not include the strongly positive results of recent trials in the elderly, especially the Systolic Hypertension, in the Elderly Program (SHEP), in which diuretic-based treatment reduced stroke incidence by 36%> and major CHD events by 27% (95% confidence interval, 4-43%)[6].

One explanation for the failure of previous trials to demonstrate the expected degree of CHD reduction is that adverse effects of study drugs, particularly diuretics, may have offset the potential benefit of blood pressure reduction. These adverse effects include diuretic-induced hypokalemia, hypomagnesemia, hyperuricemia, hyperlipidemia, hyperglycemia, impaired insulin sensitivity, and probably increased ventricular ectopic activity [6,7]. The diuretic-induced increase in total cholesterol has been estimated to be approximately 4%, and in LDL cholesterol as much as 10%> [7], though these effects may be attenuated with long-term treatment [5]. Such increases in blood lipids would be sufficient, if sustained, to offset a substantial portion of the CHD benefit from blood pressure reduction.

2) New Classes of Antihypertensive Agents In the early 1980's two new classes of antihypertensive agents, the calcium antagonists and angiotensin-converting enzyme (ACE) inhibitors, were developed and licensed for use in chronic antihypertensive therapy. These agents tend to cost more than older agents such as diuretics and beta-blockers, and evidence that might justify their use despite the increased cost (such as greater efficacy or fewer side effects) is limited [2]. The fourth Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC) recommended beta-blockers, calcium antagonists, ACE-inhibitors and diuretics as equally acceptable first-line therapy [9]. All four classes of drugs have been found to control diastolic blood pressure as single agents in 50% or more of patients with mild hypertension. The fifth JNC report [10] reconsidered the choice of initial therapy and recommended diuretics and beta-blockers as preferred agents. In addition, JNC-V added the alpha adrenergic blockers, as well as drugs with combined alpha and beta blocking actions, as alternative first-line treatments.

Of these drag classes, only beta-blockers have been compared directly to diuretics in large-scale, long-term clinical trials in hypertension. Three such trials completed in Europe in 1985-86 showed approximate equivalence of effects on morbidity and mortality in diuretic- and beta-blocker-based regimens. Pooled analysis of these trials yields a 6% (95% confidence interval, -10% to +22%) lower CHD mortality from beta- blockers [11]. These data are in contrast to the recent Medical Research Council (MRC) Trial in the Elderly, in which patients treated with a thiazide diuretic had significantly lower rates of CHD compared to beta-blocker (atenolol) treatment or placebo, both by about 45%.

Other data have shown that calcium-blockers may inhibit development of atherosclerotic lesions in rabbit models, but trial data on morbidity and mortality are conflicting. One trial of diltiazem in post-MI patients was interpreted as showing benefit in patients without low ejection fraction, but an overview pooling all post-MI trials with calcium-blockers reported a 6% (-4% to +18%>) increase in mortality [12]. An update of this overview including three additional trials in patients with angina pectoris or myocardial infarction suggests more favorable results with agents that slow the heart rate compared to dihydropyridine calcium blockers [13]. Other trials in hypertensives have shown a decrease in left ventricular mass with calcium-blocker treatment [14].

Among ACE-inhibitors, at least three of the seven licensed drugs in this class have been reported to reduce left ventricular hypertrophy [15]. ACE inhibitors reduce mortality in both severe and less severe heart failure [16], and reduce morbidity, including CHD, in asymptomatic left ventricular dysfunction [17]. With regard to effects on atherosclerosis, Chobanian and colleagues have reported prevention of coronary lesions in the Watanabe rabbit model [18], perhaps due to effects on cellular proliferation in the vessel wall.

Several alpha-blockers have been shown to have moderate favorable effects on the lipid profile, particularly on LDL cholesterol [19,20]. A few studies have also found improvements in insulin resistance, an observation that may be especially relevant to patients with type JJ diabetes mellitus [21,22]. There is also some evidence that these agents may reduce left ventricular hypertrophy and platelet aggregability and stimulate tissue plasminogen activator [23-27].

Only two long-term randomized trials have compared representatives of all of these drug classes: the one-year trial conducted by the VA Cooperative Study Group on Antihypertensive Agents [28], and the 4.4-year Treatment of Mild Hypertension Study (TOMHS) [19]. While these trials have reported some differences in BP control, side effects, quality of life, biochemical effects, and target-organ changes, these differences did not present a pattern that consistently favored some drugs and not others.

Data from a large variety of studies in humans and animal models thus suggest that newer drugs may be superior, equivalent or inferior to standard drugs in the treatment of hypertension. A report from the British Hypertension Society stated: "We thus conclude that beta-blockers or diuretics are equally acceptable first-line treatments.... Unfortunately... large-scale trials have not used newer antihypertensives such as calcium antagonists and angiotensin-converting enzyme inhibitors. There are therefore no comparable data for these widely used drugs, and we urgently need large-scale comparative trials to assess the role of these agents [29]." U.S. investigators have arrived at similar conclusions [30-32].

3) Importance of Comparing Antihypertensive Agents in African- Americans

Hypertension is considerably more common among African-Americans than Caucasians, and its sequelae are more frequent and severe. Prevalence of hypertension in the second National Health and Nutrition Examination Survey (NHANES II) was 51% in African-Americans aged 25-74 years compared to 40% in Caucasians [33]. Incidence of end-stage renal disease secondary to hypertension is nearly eight-fold higher in African- American than Caucasian hypertensives [34]. Risks of left ventricular hypertrophy, stroke and stroke death have all been reported to be greater among African- American hypertensives. Suggested explanations for these differences have included higher prevalence of co-existing illnesses such as diabetes among African- Americans, and decreased access to medical care.

Given that treatment effectively reduces hypertension-related cardiovascular morbidity and mortality, populations with decreased access to care might be expected to suffer disproportionately high rates of these complications. In such groups, particularly those of lower socioeconomic status (SES), cost of drug therapy may become the overriding consideration in selection and maintenance of treatment. The current trend toward use of more expensive agents is thus more likely to become a barrier to treatment in lower SES persons, who are disproportionately represented by African- Americans and who also bear a greater burden of hypertension-related diseases. If cheaper drugs such as diuretics are equally effective in preventing the complications of hypertension as are other available agents, low SES African- Americans are those most likely to benefit from this information. If cheaper drugs are less effective, low SES African- Americans are those most likely to suffer the consequences, since they will tend to receive cheaper drugs or none at all. In either situation, and given the relative lack of clinical trials information in African-Americans, they should be heavily represented in the current trial so that the results will be directly applicable to them. For these reasons, the population for this trial will be at least 55% African-American.

B. Cholesterol-Lowering Trial

1) Cholesterol and Coronary Heart Disease

Circulating levels of cholesterol, specifically cholesterol associated with the low- density lipoprotein (LDL) fraction, have been established by observational epidemiologic studies, by metabolic, pathologic, and genetic studies in humans and selected animal models, and by randomized clinical trials, to be a major etiologic factor in coronary heart disease (CHD) [35]. The clinical trials that have demonstrated a reduction in CHD incidence from lowering LDL-cholesterol levels have been conducted primarily in middle-aged men with hypercholesterolemia or established CHD [36-41]. Experimental evidence for the efficacy of cholesterol lowering in older men is confined to the analysis of small subgroups of clinical trials, and is lacking for women of any age. This paucity of clinical trial data led the National Cholesterol Education Program's (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults to allow considerable room for physician judgment regarding: the elderly in their 1987 guidelines [42].

The uncertainty in application of the NCEP guidelines to older men and to women is a matter of considerable consequence to the public health. While myocardial infarctions in middle-age are responsible for tragic losses of life and productivity, more than 80% of all CHD deaths m the U.S. occur after age 65 [43]. Women comprise more than 60% of the population above age 65, and the majority of CHD events in this age group are in women. In 1989, CHD was responsible for nearly 400,000 deaths and $14.4 billion in direct health care costs annually in this age group. This annual toll can be expected to increase over the next four decades as the aging of the "baby boom" generation and gains in longevity shift the age distribution of the U.S. population upward.

While the evidence relating elevated cholesterol levels to CHD is strong and consistent, clinical trials [44,45] have not demonstrated that cholesterol-lowering reduces total mortality. Moreover, observational studies [46] show a U-shaped relationship of cholesterol and mortality, with higher mortality rates for persons with cholesterol levels less than 160 mg/dl and greater than 240 mg/dl compared to those in the range 160-240 mg/dl. Deaths from hemorrhagic stroke, some cancers, and respiratory and digestive diseases contribute to the excess mortality at the lowest cholesterol levels [46]. In addition, reductions in CHD mortality from moderate (drug-induced) cholesterol lowering in randomized primary prevention trials in hypercholesterolemic patients have been offset by increases in trauma-related mortality [44]. These trials were not designed with adequate statistical power to address total mortality within the planned treatment period, though some have observed favorable post-trial trends [45,47]. Larger and more powerful clinical trials are needed to address this issue.

2) Cholesterol in Older Persons and African-Americans While observational studies suggest that clinical trial results may be applicable to men with lower cholesterol levels and to women, the prognostic power of cholesterol levels in these studies diminishes with advancing age [48-51]. Pooled analyses of 21 male and 13 female cohorts suggest that the decrease in CHD risk associated with lower blood cholesterol levels above age 65, though statistically significant at least in men, is perhaps 60% of that below age 65, with considerable variation among studies [50]. However, this apparent diminution in relative risk associated with aging may be more than offset by the concomitant increase in absolute incidence of CHD deaths and myocardial infarctions [51]. Thus, the potential public health value of cholesterol lowering in seniors cannot be ignored. The unanswered question is the extent to which these "attributable" CHD events are actually preventable when cholesterol-lowering treatment is initiated at a relatively advanced age.

Although LDL-cholesterol levels are slightly lower and HDL-cholesterol levels _. are slightly higher in African-Americans than non-minority populations, hypercholesterblernia is still a substantial problem among African- Americans, particularly in women with obesity and diabetes. Approximately 23% of African- American men and women aged 25-74 years had high-risk, serum cholesterol levels in 1976-80 [33], in whom reduction of cholesterol levels would be expected to reduce CHD mortality and morbidity. Prior cholesterol-lowering trials have included only a handful of African- American subjects. The assumption that cholesterol lowering will produce similar reductions in CHD in African-Americans and Caucasians is unproven.

3) Use of HMG CoA Reductase Inhibitors in Older Men and Women

Many previous clinical trials of cholesterol-lowering in middle-aged men have been weakened by the limited efficacy and acceptability of the drugs employed. However, lovastatin, the first of the HMG CoA reductase inhibitors, has been used increasingly widely since its approval by the FDA in September, 1987, and has been efficacious and well-tolerated by patients of all ages [52-55]. Two small angiographic trials have demonstrated a beneficial effect of lovastatin on coronary atherosclerosis [56,57]. Two additional HMG CoA reductase inhibitors, pravastatin and simvastatin, were approved by the FDA in late 1991 ; both have been used for several years in other countries. The FDA is currently considering approval of a fourth HMG CoA reductase inhibitor, fluvastatin.

Few serious side effects of these drugs have been observed to date. The absence of adverse lenticular changes in the 1990 report of the EXCEL study of lovastatin [55] prompted the FDA to remove its requirement for annual slit lamp exams from the product label. Significant but reversible asymptomatic elevations in serum transaminase levels have been seen in 1-2% of patients who have received these drugs. Myositis, in rare cases progressing to rhabdomyolysis and renal failure, is the most serious reported side effect of the HMG CoA reductase inhibitors and is potentiated by concomitant use of other potentially myotoxic drugs (immunosuppressive drugs, fibrates, etc.) and by impaired renal function. However, the incidence of myositis is quite low in the absence of these factors and does not significantly exceed placebo rates at low doses. Although HMG CoA reductase inhibitors may potentiate the effects of anticoagulants, clinically significant interactions with the antihypertensive agents to be used in ALLHAT have not been reported.

Overall, because of their efficacy, ease of adrrrinistration, low toxicity, and compatibility with most other drugs, the HMG CoA reductase inhibitors appear to be well-suited for use in older men and women. They also offer the opportunity to extend our knowledge of the benefits and safety of cholesterol lowering to cholesterol levels and degrees of reduction not previously addressed by large clinical trials.

A two-year pilot study for a trial of cholesterol lowering in seniors (Cholesterol Reduction in Seniors Program, CRISP) was initiated in July, 1990, at five clinical centers. Although it used a conventional trials model, with funded clinical sites instead of the office-based recruitment model planned for the current trial, the pilot study demonstrated the feasibility of recruiting older persons into a trial of cholesterol lowering [58]. A total of 431 men and women aged 65 and above were recruited into the pilot, surpassing the goal of 400 subjects within ten months. The pilot study cohort, which is 72% female with 25% minorities, was followed through June, 1992, to compare the compliance, safety, and efficacy of two alternative dosage regimens (20 mg and 40 mg, daily) of lovastatin versus placebo. Both dosages were well-tolerated, with 85-90% compliance after one year of treatment. The mean LDL reduction (28%) obtained with the 40 mg dosage only slightly exceeded, that obtained with the 20 mg dosage (24%). Small increases in HDL cholesterol (7% and 9%) and decreases in triglycerides (4% and 10%) were also observed (respectively) for the 20 mg and 40 mg dosages.

I. Hypotheses and Study Power

A. Antihypertensive Trial Component

The primary hypotheses of this trial component are that the combined incidence of fatal CHD and nonfatal myocardial infarction will be lower in hypertensive patients receiving (1) a calcium antagonist (amlodipine), (2) an ACE inhibitor (lisinopril), or (3) an alpha adrenergic ►^■ blocker (doxazosin)* as first-line therapy than in those in whom a similar degree of blood pressure control is achieved using a thiazide-like diuretic (chlorthalidone) as first-line therapy. These hypotheses will be tested in a population of men and women aged 55 years and older, all with at least one additional CHD risk factor besides hypertension, of whom at least 55% will be African-American. The statistical power to test these hypotheses is approximately 82.5%, based on the following assumptions:

1) Sample size of 40,000 (approximately 22,000 men and 18,000 women), allocated among four treatment groups as shown in Table 1.1.

2) Six-year incidence of CHD events of 7.8% in the diuretic group. This rate, 1.35% per year, is based on the experience of Framingham and HDFP (adjusted downward by 33- 50% for temporal trends and by 25% for the healthy volunteer effect) and is similar to the rate observed more recently in SHEP.

3) A 20% reduction in CHD event rate (before adjustment for non-compliance and losses to follow-up, which combine to produce a 16.3% effective reduction) in each of the three non-diuretic treatment arms compared to the diuretic arm.

4) Using a time-dependent Markov model, rates of crossover between each of the other study drugs and chlorthalidone and/or non-study medication are assumed to be 2.75%o in each of the first three years and 6% over the last three years of follow-up (based on TOMHS). The probability of crossing over at least once during the study is assumed to be approximately 24%. It is assumed (conservatively) that the CHD risk associated with the non-study drugs is the same as for chlorthalidone.

5) CHD status will be undeterminable at the end of the study for 16.8% of patients (8.6% of person-years) due to competing risks (non-CHD death) or loss to follow-up, based on data from Framingham and HDFP (see Appendix I).

6) A 25% reduction in CHD event rates (before adjustment for non-compliance and losses to follow-up) among the 10,000 patients randomized to the active treatment arm of the cholesterol-lowering trial component.

7) A type I error = 0.05 (two-sided). This corresponds to a critical Z-score of 2.37 after adjustment for multiple comparisons.

Power estimates ranged from 77 to 86% for more pessimistic or optimistic assumptions of crossover (#4) and loss (#5) rates. Additional details regarding these calculations may be found in Appendix I.

* On January 24, 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000;283:1967-1975. ^■ Secondary hypotheses pertaining to the effect of the following end points in patients *" randomized to receive amlodipine, lisinopril, or doxazosin* (relative to those receiving chlorthalidone) will also be assessed: (1) all-cause mortality, (2) combined coronary heart^' disease (CHD + revascularization procedures + hospitalized angina), (3) stroke, (4) combined cardiovascular disease (CHD + stroke 4- revascularization procedures + angina [hospitalized or treated] + CHF [hospitalized or treated] + peripheral arterial disease [hospitalized or outpatient revascularization procedure]), (5) left ventricular hypertrophy by ECG, (6) renal disease including the reciprocal slope of serum creatinine and end-stage renal disease (dialysis or transplant), (7) health-related quality of life, (8) fatal and non-fatal cancer by type, and (9) gastrointestinal bleeding.

B. Cholesterol-Lowering Trial Component: (See Protocol Addendum 1)

The primary hypothesis of this trial component is that mortality from all causes ^'will be lower in the-subset of hypertensive patients described above with LDL cholesterol levels between 120 and 189 mg/dl (between 100 and 129 mg/dl for those with known CHD) who are randomized to receive pravastatin plus a cholesterol-lowering diet than in those randomized to receive usual care. The statistical power to test this hypothesis is approximately 80%, based on the following assumptions:

1) Sample size of 20,000 (approximately 11,000 men and 9,000 women) allocated equally between pravastatin and usual care groups.

2) Six-year total mortality of 13.2% (2.35% per year) in the usual care group (based on data from Frarningham, HDFP and SHEP (see Appendix I). Based on mortality data in a high-risk subgroup of SHEP participants selected for comparability to ALLHAT patients, it is estimated that 40% of deaths will be due to CHD, 16% due to other cardiovascular causes, and 44% due to noncardiovascular causes.

3) A 12.5%o reduction in mortality in the pravastatin treatment arm. This estimate is based on the assumption that CHD mortality is reduced by 25%, that mortality from other cardiovascular diseases is reduced by 15%, and that other causes of mortality are unaffected. If it is assumed that full compliance to the drug regimen would reduce LDL cholesterol levels by 30% and that the mean LDL cholesterol level is 155 mg/dl at entry (Range: 120-189 mg/dl), a 25% reduction in CHD corresponds to a logistic regression coefficient of 0.0062. For comparison, the logistic coefficient relating total cholesterol to CHD mortality in 356,222 MRFIT (male) screenees was 0.0118 for the full age range (35-57 years) and 0.0086 for the 36,704 men in the oldest age group (55-57 years).

^' Extrapolation from the MRFIT data suggests that the coefficient relating total cholesterol to CHD in ALLHAT might fall between 0.005 and 0.006.

* On January 24; 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000;283: 1967-1975. 4) A "dropout" rate (from pravastatin treatment to no treatment) of 5% in Year 1, and 2.5% in all subsequent years, and a "drop-in" rate (from no treatment to pravastatin or a similar drug) of 2% per year. These rates are projected to yield 15.3% of pravastatin patients off treatment and 10.6% of usual care patients on treatment at the end of 6 years. Note that the power of the cholesterol-lowering trial, which (unlike the antihypertensive trial) focuses on the generic effects of cholesterol-lowering, rather than the effects of a specific drug or drug class, is not diminished by "crossovers" from pravastatin to other regimens that produce equivalent lipid changes.

5) No losses to follow-up.

6) A 10% reduction in mortality rate in each of the three non-diuretic treatment arms of the anti-hypertensive trial component.

7) A type I error α = 0.05 (two-sided), corresponding to a critical Z-score of 1.96.

Power was also calculated for a range of annual mortality rates (2.2 to 2.5%) and for a somewhat more pessimistic set of assumptions regarding compliance (leading to 17.8% and 12.9%> prevalence of drop-outs and drop-ins, respectively, at the end of six years). Statistical power estimates ranged from 75 to 82% under these assumptions (see Appendix I).

Secondary hypotheses pertaining to reduction of the following end points in patients randomized to receive pravastatin (relative to those receiving usual care) will also be assessed: (1) the combined incidence of CHD death and nonfatal myocardial infarction, especially in certain subgroups like African-Americans, patients over age 65 (the original CRISP hypothesis), type II diabetics, and women, (2) changes in the biennial study ECG indicative of myocardial infarction, (3) cause-specific mortality, (4) total and site-specific cancer incidence, and (5) health-related quality of life.

The power of this study to address the effect of pravastatin on the combined incidence of CHD death and nonfatal myocardial infarction is estimated to be 97% overall and close to 80% in any subgroup containing 10,000 patients with risk characteristics similar to the overall cohort (see Appendix 1). However, the objectivity of the clinical diagnosis of nonfatal myocardial infarction is potentially compromised by the fact that the treating physicians will not be blinded as to whether their patients have received pravastatin or usual care. To guard against bias, the incidence of nonfatal myocardial infarction will also be assessed by changes in the biennial study ECG, the evaluation of which will be performed by coders who are unaware of the patient's treatment assignment. A pravastatin-usual care difference in clinical events will be given credence only if confirmed by a qualitatively similar difference in the ECG end point. TV. Eligibility and Exclusions

A. Antihypertensive Trial

1. Age/sex: Men and women aged 55 years and older.

2. Medical history: History of documented hypertension.

3. Seated blood pressure:

Eligibility is based on the patient's current treatment status and on the average of two seated blood pressure measurements at each of two visits (Table IV.1):

(a) Patients whose blood pressure has been controlled (the majority of blood pressure measurements < 160/100 mmHg) with one or two antihypertensive drags for at least two months are eligible. Patients who are taking three or more antihypertensive drugs at subtherapeutic doses or in ineffective combinations, and who are felt likely to be controllable on the ALLHAT protocol, can enter the trial at the discretion of the principal investigator or his/her designee.

(b) For untreated patients, a diagnosis of hypertension must first be established. (The JNC V criteria for diagnosing hypertension are included in the Manual of Operations.) Patients who are untreated or who have been treated for less than two months must meet the minimal as well as the maximal blood pressure criteria shown in Table IV.l . To qualify for entry, the lower SBP or DBP limit and both upper limits must be met on two occasions at least one day apart. Patients who do not meet the blood pressure entry criteria at Visit 1 or Visit 2 may be re-evaluated for blood pressure eligibility at a later time.

Table IV.l. Blood Pressure Eligibility Criteria

4. At least one of the following: a. One or more of the following manifestations of atherosclerotic cardiovascular disease: (i) Old or age-indeterminate myocardial infarction or stroke (>6 months),

(ii) History of revascularization procedure (ever),

(iii) Documented atherosclerotic cardiovascular disease. This includes, but is not limited to: coronary, penpheral vascular, aortic, or carotid stenosis as documented by angiography, Doppler studies, or diminished ankle-arm index; ischernic heart disease as documented by electrocardiography (e.g., ST-T wave changes), echocardiography, or radionuclide imaging; history of intermittent claudication; or history of transient ischernic attack. (A more complete list of manifestations of atherosclerotic cardiovascular disease is provided in the Manual of Operations.) b. Type II diabetes mellitus [plasma glucose >140 mg/dl (fasting) or 200 mg/dl (non- fasting) and/or on insulin or oral hypoglycemic agent] (in the past 2 years), c. HDL-cholesterol <35 mg/dl (on any 2 determinations within past 5 years) ά. Left ventricular hypertrophy (one of the following) on any ECG within the past 2 years:

• R amplitude in V₅ or V_<5 > 26 mm.

• R amplitude in V₅ or Vg plus S amplitude in V] > 35 mm.

• R amplitude in aVL > 12 mm.

9 R amplitude in Lead I > 15 mm.

• R amplitude m Leads JJ or HI, or aVF > 20 mm.

9 R amplitude in Lead I plus S amplitude in Lead III > 25 mm. β R amplitude in aVL plus S amplitude in V₃ > 28 mm for men or > 22 mm for women β Computerized ECG machine documented LVH

For visual LVH reading, QRS amplitudes are measured in the second to last complete normal beat of the lead.

e. Left ventricular hypertrophy on any echocardiogram (within the past 2 years) based on 25 mm or more combined wall (ventricular septum plus posterior wall) thickness. f. Current cigarette smoking (any cigarettes smoked in past 30 days). 5. Exclusions: a. Symptomatic MI or stroke within past six months. b. Hospitalized or treated symptomatic congestive heart failure and or ejection fraction < 35%o, if known. c. Angina pectoris within past 6 months. (This implies actual chest pain. If patient has history of angina and no chest pain, even if he/she is on antianginal drugs, then this exclusion does not apply. See exclusion 5e.) d. Known renal insufficiency (serum creatinine > 2 mg/dl). e. Patients requiring diuretics, calcium antagonists, ACE inhibitors, or alpha adrenergic blockers for reasons other than hypertension. (If a patient is on a calcium-channel blocker for angina, he/she may be switched to a beta-b locker for this indication if it is deemed safe to do so. This exclusion criterion would not then be applicable.) f. Patients requiring more than two antihypertensive drugs to achieve satisfactory blood pressure control (SBP < 160 mmHg and DBP < 100 mmHg). Patients who are taking three or more antihypertensive drugs at subtherapeutic doses or in ineffective combinations, and who are felt likely to be controllable on the ALLHAT protocol, can enter the trial at the discretion of the principal investigator or his/her designee. See JN A 3(a). g. Sensitivity or contraindications to any of first-line study medications. h. Factors suggesting a low likelihood of compliance with protocol,- such as dementia, history of alcohol or drug abuse within past six months, plans to move or travel extensively, or history of unreliability in keeping appointments or taking prescribed drugs. i. Diseases, such as non-curable malignancy, likely to lead to non-cardiovascular death over the course of the study. j. Blood pressure over 180 mmHg systolic or over 110 mmHg diastolic on two separate readings during step-down of antihypertensive medications. k. Current participation in another clinical trial

B. Cholesterol-Lowering Trial

1. Eligible and enrolled in antihypertensive trial.

2. Fasting LDL Cholesterol: 120 to 189 mg/dl (100 to 129 mg/dl for patients with known CHD). These outpoints, which correspond roughly to the 30th and 90th percentiles in men and the 25th and 85th percentiles in women in the ALLHAT age range, are projected to include approximately 60% (24,000) of ALLHAT participants from whom 20,000 would remain afterrefusals and other exclusions.

3. Fasting triglyceride level below 350 mg/dl.

4. Additional Exclusions: a. Current use of prescribed lipid-lowering agents or large doses (> 500 g/day) of non- prescription niacin . Eligible patients must be off lipid-lowering drugs at least two months and off probucol for more than one year at the time of Visit 2. b. Contraindications to HMG CoA reductase inhibitors (e.g., significant liver disease, ongoing immunosuppressive therapy, known allergy or intolerance to the study drug). c. Known untreated secondary cause of hyperhpidemia (e.g., hypothyroidism, nephrotic syndrome). d. ALT > 2.0 x upper limit of normal.

V. Recruitment

Recruitment for ALLHAT will rely primarily on chart review to identify patients who are potentially eligible for the antihypertensive or both trial components. Data needed to make the definitive determination of eligibility for the antihypertensive trial component will be obtained in a series of pre-randomization visits, which will take place over a period generally not exceeding two months. The number and frequency of those visits will depend on the complexity of the patient's pre-study regimen, the blood pressure response to step-down from that regimen, and the patient's suitability for and interest in the cholesterol-lowering trial component. Because only patients who have been randomized to the antihypertensive trial component will be considered for randomization to the cholesterol-lowering trial component, randomization to the latter will not take place until the first post-randomization (4 week) visit for the antihypertensive trial. The steps leading from identification of these various categories of potential ALLHAT candidates to randomization in one or both study components are described below.

Chart Review:

At each clinical site, patients who might potentially be suitable for the antihypertensive component of ALLHAT and the subset of such patients who might also be eligible for the ^' cholesterol-lowering component of ALLHAT will be identified by chart review. Information on blood pressure and antihypertensive treatment, LDL (or total if LDL is unavailable) cholesterol levels and cholesterol-lowering diet and/or drugs, other relevant medical history, and a sense of the patient's reliability and compliance with previously prescribed treatments should be reviewed for conformity with the study eligibility requirements (Chapter IV) before his/her initial study visit.

A. Antihypertensive Trial Component

Visit 1: Preliminary Determination of Eligibility and Interest

The objective of Visit 1 is to assess eligibility for and interest in ALLHAT and to begin withdrawing the patient from any existing antihypertensive medications. It is anticipated that _. the majority of treated hypertensives will have been identified by chart review, and that much of the pertinent information (age, risk factor status, number of antihypertensive drugs, etc.) will already be known. The investigator will be required to complete a one-page questionnaire to document that all the preliminary inclusion and exclusion criteria (Chapter TV) have been met. Any documentation not attainable by chart review or not available within the past two years (ECG to assess the presence of left ventricular hypertrophy, fasting glucose level for diabetes, total cholesterol (TC) level to assess lipid eligibility), or within the last five years for the two determinations of HDL, will be considered part of the patient's routine medical management and will not be specifically reimbursed by the study.

Visit 1 will consist primarily of obtaining the first entry blood pressure, answering the patient's questions about the study, and obtaining the patient's informed consent to begin the step-down if necessary from pre-study antihypertensive drugs. Recommendations for antihypertensive drag withdrawal are included below. If a patient's antihypertensive medications can be safely switched without tapering, the participant may move directly to Visit 2. Visits 1 and 2 should be separated by at least one day, but need not be consecutive visits.

For untreated patients, a diagnosis of hypertension according to JNC V criteria needs to be established. Once this is done, those patients whose SBP > 140 rnmHg and/or DBP > 90 mniHg and whose SBP and DBP do not exceed 180 and 110 mmHg, respectively (see Table IV.l), may also be considered for entry into ALLHAT. For new patients found to have elevated blood pressure at their initial visit to the clinical site, this initial visit may serve as ALLHAT Visit 1 provided that any additional evaluations needed to determine study eligibility are performed (at no cost to the study). The subsequent course of such patients is simplified by elimination of the need for a step-down from a pre-study drug regimen.

Step-down visits:

Not all patients on antihypertensive medications will require step-down visits. In general, the following categories of antihypertensive drugs can usually be stopped without tapering the dose:

1. Diuretics: The patient should be informed to contact the physician if he/she develops marked edema and/or significant increase in dyspnea (shortness of breath, either at night or on exertion).

2. Reserpine.

3. Angiotensin converting enzyme (ACE) inhibitors.

4. Calcium channel blockers.

5. • Vasodilators (e.g., hydralazine).

6. Alphaj -blockers (e.g., prazosin).

The following categories need to be tapered if the patient is taking more than the usual starting doses:

1. Central agonists (e.g., oral clonidine or methyldopa) - taper over 1-2 weeks

. 2: Beta-adrenergic blockers (e.g., propranolol or atenolol) - taper over 2 weeks

Post-myocardial infarction patients receiving beta-blockers for prophylaxis need not discontinue therapy. There may be occasional circumstances where, in the physician's judgment, closer monitoring or a longer period of withdrawal is preferred. Extra care should be taken in tapering antihypertensive drugs in those patients with cardiovascular disease.

Patients whose blood pressure exceeds 180 mmHg systolic or 110 mmHg diastolic should return within a few days for a repeat blood pressure measurement. If the blood pressure is still above 180/110, the patient should not be randomized into the antihypertensive trial. Visit 2: Randomization:

. Patients who have met all ALLHAT eligibility criteria and in the judgment of the investigator can safely discontinue all prior antihypertensive drugs and be randomized to one of the four ALLHAT treatment arms shall, after giving their informed consent, be entered into the study at Visit 2. This visit will generally take place between 1 day and 12 weeks after Visit 1, depending on the length of time required to step down from pre-study medications. Patients initially taking no drugs or well-controlled on one drug may be randomized soon after Visit 1, while other patients may require a longer step-down process (generally less than three months) before they can complete Visit 2. More prolonged step-downs are discouraged (though not prohibited), since many patients who cannot quickly be withdrawn from their pre-study regimens may. also be more difficult to maintain on a simple regimen during the trial.

The investigator will telephone the Clinical Trials Center regarding each patient who meets all eligibility requirements at Visit 2, including a signed consent form. The Clinical Trials Center will review the eligibility and exclusion criteria and will assign that patient a study identification number and a bottle number corresponding to (1) chlorthalidone, (2) amlodipine, (3) lisinopril, or (4) doxazosin*. The treatment assignment will be masked from both the practitioner and patient. A resting ECG, serum glucose, serum potassium and creatinine, fasting lipid profile and ALT should be obtained at this visit for all patients who are randomized. Each randomized patient will be issued an appropriate supply of his/her starting dose of the assigned study drug and will be instructed to return for the first dosage titration (Visit 3) four weeks later (see Section VI).

All randomized patients will be given appropriate hygienic advice (sodium and alcohol restriction, smoking cessation, exercise, caloric restriction if overweight) with reinforcement as needed during the trial.

B. Cholesterol-Lowering Trial Component

Visit 1: Preliminary Determination of Eligibility and Interest

Patients who have satisfied all Visit 1 eligibility requirements for the antihypertensive trial component (see above) and/or have consented to begin step-down from pre-study antihypertensive drugs should also be informed of the cholesterol-lowering trial component of ALLHAT. Those who indicate their possible interest in this component and have not been treated with lipid lowering drugs (including 500 mg or more per day of over-the counter niacin) during the two months preceding Visit 1 shall be considered as potential candidates for this trial. Patients who have taken probucol within one year preceding Visit 1 are also ineligible for this component of ALLHAT.

* On January 24, 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000;283:1967-1975. Visit 2: Fasting LDL Determination (Randomization to Antihypertensive Trial Component)

A fasting lipid battery (total cholesterol, triglycerides, HDL-cholesterol, calculated LDL- cholesterol) and serum ALT will be obtained^'Tor patients who: are eligible and randomized into the antihypertensive trial component of ALLHAT. Patients who have not fasted at least 9 hours (12 hours optimum) should have their blood draw rescheduled within a week of Visit 2. If rescheduling for fasting lipids is required, this will be considered part of Visit 2.

Visit 3 or 4: Randomization

Patients with fasting LDL-C between 120 and 189 mg/dl (between 100 and 129 mg/dl for patients with known CHD) and fasting TG < 350 mg/dl at Visit 2 will be informed by telephone of their eligibility for the cholesterol-lowering trial component and told to come in fasting for Visit 3. If they sign the Informed Consent to participate in this ALLHAT component at Visit 3, the investigator will phone the Clinical Trials Center, review the eligibility and exclusion criteria for the lipid-lowering component, and receive a random assignment for the patient to either pravastatin or usual care. Each patient randomized to receive pravastatin will be issued an appropriate supply of 20 mg tablets and instructed to take two each evening. Patients assigned to usual care will not be prescribed any lipid-lowering medication by ALLHAT. Patients assigned to usual care as well as those assigned to pravastatin will be advised to follow the NCEP Step I diet (<30% of calories from fat, <10% of calories from saturated fat, <300 mg cholesterol per day). A fasting lipoprotein profile will be obtained at this visit as a baseline for each randomized participant in this trial component.

Maintenance of Racial Composition of ALLHAT:

Before their practices are selected as clinical sites for ALLHAT, potential study investigators will be asked to indicate the approximate proportion of African-American patients they expect to recruit into the study. Clinical sites will be selected to produce an overall study population of at least 55% African- Americans and will be monitored by the Clinical Trials Center throughout the study to assure that their performance matches their expectations. If the overall proportion of African- Americans appears to be falling significantly short of 55 >, the Steering Committee may implement remedial measures such as temporarily freezing recruitment of non- African- American patients at some or ail existing clinical sites or adding new clinical sites to correct the shortfall.

Table V.l: Schematic Summary of Entry of Patients into Two ALLHAT Components:

* Separate reimbursements are not provided for these visits. At Visit 3 (1 month), reimbursement is provided if the patient is randomized to the lipid-lowering component. For visits past 1 month, reimbursement is not provided for visits other than those at 3, 6, 9 and 12 months during the first year, and every 4 months thereafter.

**Total cholesterol, triglyceπ^'de, and HDL cholesterol levels. LDL calculated by Friedewald formula.

Post-randomization visits are shaded. Antihypertensive Intervention

The blood pressure goal in all four arms* will be <90 mrnHg diastolic and <140 mmHg systolic.¹ The number and dose of study drugs prescribed in pursuit of these goals will be influenced by patient tolerance and clinical judgment, particularly in use of greater than two- drug regimens. With rare exceptions, treatment should be intensified for patients with BP levels > 160 mm Hg systolic and/or > 100 mm Hg diastolic, even if low doses of drugs from the same classes as the blinded Step 1 drugs must be added.

The therapeutic goal is to achieve blood pressure control on the lowest possible dosage of the first-line drug. The addition of second- line (open label) drugs should be reserved for those in whom the maximal dosage level of the first-line drug is insufficient.

Each of the four* first-line drugs will be administered once daily in the morning. The following dosage levels will be available for each drug:

Table VI.l. First-Line (Blinded) Antihypertensive Drugs

Sources of the four Step 1 agents are: chlorthalidone: Ogden Bioservices, Inc., R c viϊle, Maryland; amlodipine: Pfizer, Inc., New York, New York; and lisinopril: Zeneca Pharmaceuticals Group, Wilmington, Delaware; and doxazosin*: Pfizer, Inc., New York, New York.

The identity of the drug will be masked at each dosage level , but the identity of the dosage level will not be masked. The initial dosage level will be used only during the first week after randomization to rninimize the potential side effects of doxazosin*. (For the other three drugs, the initial dose and Step 1 dosages are identical.) The Step 1 dosage level should be initiated at the end of the week. A clinic visit is not required.

All patients will be re-evaluated at least at 1 month for dose titration if needed (Visit 3) and at 3 months (Visit 4). Study medication will be initiated at the initial dose and patients should typically return at one-month intervals for any necessary increase in dosage until both the

* On January 24, 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000;283: 1967-1975.

^ese goals follow the Fifth Joint National Committee recommendations [10]. The JNC set the systolic goal lower than that used in SHEP, which was between 140 and 159 mm Hg with a mean attamed value of 142 mm Hg, because of the known strong epidemiologic relationship of systolic pressure with CVD mortality [59]. LX. Outcome Measurements

Occurrences of study endpoints will be documented by a checklist completed by the study physician at each follow-up visit and supplemented by interim reporting as needed. These diagnoses will be supported by copies of death certificates, discharge summaries and face sheets as described below. The following outcome measures will be obtained and tabulated over the course of the stud}':

A. Death (documented by death certificate).

The underlying cause of death will be classified by the physician-investigator at the clinical site as due to (1) Coronary Heart Disease, (2) Other Cardiovascular Disease, (3) Neoplastic Disease, (4) Other Medical Causes, or (5) Non-Medical Causes. A National ^' Death Index (NDI) Search will be performed near the end of the study to identify and document deaths that may have occurred among patients who are lost to follow-up. Because of the time lag inherent in the NDI, a private tracing service will also be utilized for selected participants. Physicians will also be asked to report cause of death on the study endpoint form.

B. Cardiovascular End Points

1) Myocardial infarction (documented by hospital discharge summary or face sheet or by biennial study ECG), including suspected myocardial infarction with thrombolytic therapy.

2) Stroke (documented by hospital discharge summary or face sheet).

3) Congestive heart failure a) Hospitalized or procedure (documented by hospital discharge summary or face sheet) b) Not hospitalized but treated (documented by check box on end point questionnaire)

4) Angina pectoris a) Hospitalized or procedure (i) with or (ii) without a revascularization procedure (documented by hospital discharge summary or face sheet) b) Not hospitalized but treated (documented by check box on end point questionnaire) 5) Peripheral arterial disease a) Hospitalized or procedure (i) with or (ii) without a revascularization procedure (documented by hospital discharge summary or face sheet) or outpatient revascularization procedure (documented by procedure sheet) b) Treated medically as outpatient (documented by check box on end point questionnaire)

6) Left ventricular hypertrophy (documented by biennial study ECG)

The LVH inclusion criteria are based on specific ECG criteria as listed in IV.A.4.d and will be interpreted at the clinical site. ECGs will be re-read centrally to assign Minnesota Codes. The outcome criteria for LVH are based on the Minnesota Code. The Minnesota Coding Center will use Codes 3-1 or 3-3 to identify prevalent LVH. These amplitude criteria sets are generally considered "probable ECG-LVH", but when combined with any 4-3 or more severe 4-code, or 5-3 or more severe 5-code, it is considered "definite ECG- LVH".

Minnesota Code 3-1 : R amplitude > 26 mm in either V₅ or V₆ or R amplitude > 20 in any of leads I, II, III, aVF, or R amplitude > 12 mm in lead aVL.

Minnesota Code 3-3: R amplitude in V₅ or Vg plus S amplitude in V] > 35 mm or R amplitude > 15 mm but < 20 mm in Lead I.

The Coding Center will also document incident ECG-LVH and progression/regression of ECG-LVH using serial ECG comparison.

C, Other End Points

1) Decreased renal function (documented by reciprocal slope of serum creatinine level versus time — continuous measure)

2) End stage renal disease (initiation of chronic dialysis, kidney transplant) a) Hospitalized or procedure (documented by hospital discharge summary or face sheet) b) Treated as outpatient (documented by check box on end point questionnaire)

3) Cancer — Site and Type a) Hospitalized or procedure (documented by hospital discharge summary or face sheet) b) Treated as outpatient (documented by check box on end point questionnaire)

4) Nonfatal accidents and attempted suicides a) Hospitalized or procedure (documented by hospital discharge summary or face sheet) b) Treated as outpatient (documented by check box on end point questionnaire)

5) Gastrointestinal bleeding a) Assessed through data from the Health Care Finance Administration and the Department of Veterans Affairs

6) Quality of life — A generic categorical measure of health status will be used to assess health related quality of life.

7) Medical care utilization— Utilization data will be collected by interview. Costs will be assigned to each unit of utilization (hospitalization, office visits, procedures, etc.) based on its DRG. In addition, one question will be asked to ascertain quality of life on a continuous scale in order to determine quality-adjusted life years. For a 10% sample of patients over age 65, these interview data will be cross-checked versus Medicare records.

The study investigators will be required to complete and submit to the Clinical Trials Center a short end points questionnaire for each occurrence of a study endpoint identified at or between regular visits. For each end point involving a death or hospitalization, the investigator will also obtain and submit a copy of the deatli certificate or hospital discharge summary or face sheet upon which the diagnosis is based. For a random (10%) subset of hospitalized (fatal and nonfatal) myocardial infarctions and strokes, the Clinical Trials Center will request the more detailed information as described in Appendix I so that the in-hospital ECGs and enzyme levels (for myocardial infarctions), and neurologists' reports and CT and/or M I reports (for strokes) can be evaluated by the study end points committee and the accuracy of the discharge diagnoses (versus the definitions in Appendix I) can be assessed.

^•v

Λ.. Study Organization

Overview:

ALLHAT will employ an organizational structure that differs markedly from the usual NHLBI-supported clinical trial. The trial will be performed by a large number (600) of practicing physician-investigators who will be compensated on a per capita basis for each, patient seen according to a fixed payment schedule. Approximately 20% of study patients are expected to be recraited by Department of Veterans Affairs (VA) hypertension clinics. The Clinical Trials Center, in addition to its conventional data handling and monitoring responsibilities, will be responsible for identifying and paying these physician-investigators, hiring regional coordinators to monitor recruitment and compliance, and for awarding and supervising subcontracts for a central laboratory and an ECG coding center. A Steering Committee will be selected for their expertise in the relevant subject areas. A detailed description of the nature and role of the study components is given below.

Program Office

The Program Office, located in the NHLBI, Division of Epidemiology and Clinical Applications (DECA) and Division of Heart and Vascular Diseases (DHVD), will award and monitor the contract that provides funding for the study, set up the agreements to fund the VA clinics, and hold the IND for the study. The Director, NHLBI, will appoint the Data and Safety Monitoring Board (DSMB) and the Chair and Vice-chair of the Steering Committee. With the concurrence of the Director, NHLBI, the Program Office will appoint the Steering Committee and any other committees deemed necessary to advise the NHLBI on issues pertaining to the progress or results of the study.

Clinical Trials Center:

The Clinical Trials Center will have primary responsibility for identifying suitable medical practices to participate in ALLHAT, paying them according to a fixed fee schedule for each patient randomized and study form completed, and for editing, storing, and analyzing data generated by the study. Its investigators and staff will have a central role in designing the data collection system and in monitoring data quality. Specific Clinical Trials Center responsibilities include:

1) Developing, preparing, and distributing the study protocol, data forms, and Manual of Operations and Procedures.

2) Appointing and paying practicing physicians to provide clinical sites for conduct of the study.

3) Appointing and paying regional coordinators (see below).

4) Obtaining Institutional Review Board (OPRR) approval for uncovered practices.

5) Maintaining files of annual financial disclosure statements for the Steering Committee members to identify potential conflicts of interest. 6) Subcontracting for a central laboratory and an ECG coding center to provide timely and standardized measurements needed by the study (see Chapter VJJI).

7) Subcontracting for a Drug Distribution Center to receive, bottle, label, and distribute study medications to the clinic sites.

8) Monitoring the performance of study components and providing timely summary reports to the Program Office and to the Steering Committee.

9) During recruitment, monitoring the proportion of African- Americans at each clinical site and recommending appropriate corrective action if the overall proportion for the study as a whole appears to be falling significantly short of the 55% target.

10) Providing detailed and up-to-date statistical reports of study progress to the Data and Safety Monitoring Board (DSMB) at their semi-annual meetings (see below).

11) Maintaining a referral network for study participants who move to a new geographic region and are unable to continue to see their original study physician.

12) Providing logistical support (as needed) and minutes for study meetings.

13) Coordinating and supervising end point verification activities.

14) Initiating searches through the National Death Index to establish the vital status of patients who are lost to follow-up at intervals recommended by the DSMB and Steering Committee (see below).

15) Preparing study manuscripts in collaboration with the Steering Committee.

Clinical Sites:

These will consist of 600 separate medical practices, designated by the Clinical Trials Center to conduct the study. It is expected that some practices (particularly the VA clinics, HMOs, and large group practices) will provide larger numbers of patients and that some practices may contribute fewer than 100 patients. The proportion of African- Americans is also expected to vary among clinics, but will be monitored closely to ensure that the target of 55% overall is met (see Chapter V).

Each clinical site is expected to be under the supervision of a physician identified as responsible for the conduct of ALLHAT. However, study forms may be completed by a physician's assistant or nurse practitioner or other designated qualified personnel, consistent with the internal organization of that medical practice. At each site, one support staff member must he designated as chiefly responsible for protocol implementation; this person will participate in central training and annual meetings. Payment for each patient randomized to each ALLHAT component and for each study visit completed will be made by the Clinical Trials Center upon receipt of the relevant completed, correct and signed study form. Regional Coordinators:

Regional coordinators will be physicians with expertise in hypertension and cholesterol lowering treatment, who will handle routine protocol questions for approximately 50 clinical sites apiece. Under direction of the Clinical Trials Center, they will assist in solving problems related to quality control, protocol adherence, recruitment and retention for the sites assigned to them. Physician coordinators will be supported by a nursing coordinator and may opt to participate as clinical sites as well. All participating VA hypertension clinics will be supervised by a single coordinator.

Drag Distribution Center

A Drug Distribution Center will be established by the Clinical Trials Center to (1) receive, package and distribute all pharmaceuticals required for the two ALLHAT components, (2) implement a system of masking so that the four first-line antihypertensive agents cannot be distringuished from each other by the study investigators or their patients (the second-line antihypertensive drugs will not be masked), and (3) provide appropriate supplies of all study medications to the clinical sites on a timely basis.

Steering Committee:

The Steering Committee will be appointed by the NHLBI to provide expert advice on the study protocol and on all subsequent decisions pertaining to the design and conduct of the study that do not require access to blinded data, and the eventual analysis and publication of the study results. Its voting members will be the NHLBI Project Officer, the principal investigator of the Clinical Trials Center, the Regional Coordinators, and 7-9 experts selected for their expertise and experience in the treatment of hypertension and/or hypercholesterolemia and in key clinical trials issues such as recruitment and adherence. Each Steering Committee member will be required to submit an annual financial disclosure statement to the Chnical Trials Center and to divest themselves of any stock holdings or retainer-type consultant positions in pharmaceutical and other companies that have a direct financial interest in the outcome of the study. The Steering Committee will meet once per year (more frequently during protocol development).

An Executive Committee will be instituted to oversee trial operations between Steering Committee meetings. Composition of the Executive Committee will include the Chair and Vice- Chair of the Steering Committee and representatives of the Program Office, the Clinical Trials Center, and the Department of Veterans Affairs. Reporting to the Executive Committee will be the following subcommittees: Eligibility and Medical Care, Operations, Publications and Ancillary Studies, Scientific and Educational Program, and Endpoints. Each of the subcommittees will have representation from the Program Office, Clinical Trials Center, and Steering Committee to oversee aspects of the trial that require frequent attention and/or special expertise, such as recruitment, adherence, quality control, blood pressure and lipid intervention, laboratory methods, endpoint verification, ancillary studies, publications, and the annual program for the investigators' meetings.

Protocol Review Committee: The Protocol Review Committee will be responsible for advising the NHLBI regarding the initial approval of the study protocol. Its members and chair will be appointed by the Director, NHLBI, and will consist of at least seven experts who are not otherwise affiliated with the study. It will meet in Bethesda when the study protocol has been completed. The meeting will be attended by the principal investigator (and designated staff) of the Clinical Trials Center and the Chair and Vice-Chair of the Steering Committee who will make presentations and answer questions regarding the protocol, and by Program Office staff.

Data and Safety Monitoring Board (DSMB :

The DSMB will be responsible for monitoring all aspects of the study, including those that require access to blinded data. The DSMB and its chair will be appointed by the Director, NHLBI, and will consist of at least seven experts who are not otherwise affiliated with the study. It is likely that the roster of the DSMB members may be largely or even entirely derived from the Protocol Review Committee, which will complete its mission as the DSMB is formed.

The DSMB will meet at least semi-annually. The principal investigator of the Clinical Trials Center and designated Clinical Trials Center staff will attend these meetings (but will not have a vote) and will be responsible for preparing and presenting up-to-date statistical reports on the progress of the study. These reports will include data on recruitment, randomization, adherence, blood pressure levels, plasma lipoproteins, adverse drug responses, and study end points, as well as statistical tests and special analyses requested by the DSMB. The Project Director (who will serve as the DSMB's Executive Secretary), Project Officer and designated NHLBI staff and the Chair and Vice-chair of the Steering Committee will also participate in these meetings m ex officio capacities.

During the active recruitment phase, the DSMB will monitor the progress of recruitment (particularly of African-American patients) and the random allocation of participants to the various treatment arms and may recommend modifications in (or termination of) one or both study components if the study design goals are not being met. The DSMB will recommend when to end the active recruitment phase of the study. The approval of the DSMB will also be required for any significant changes in the protocol recommended by the Steering Committee during the course of the study. All votes will be decided by a simple majority.

At any time during the study, the DSMB may recommend discontinuation of any of the treatment arms of either study component on any of the following grounds:

1) compelling evidence from this or another study of an adverse effect of the study treatment(s) that is sufficient to override any potential benefit on CHD and preclude its further use in the target population;

2) compelling evidence from this or another study of a significant beneficial effect of the study treatment(s), such that its continued denial to the other study groups is ethically untenable; 3) a very low probability of successfully addressing the study hypotheses within a feasible time frame, because of inadequate recruitment, compliance, drag response, event rate, etc.

The DSMB may convene an Executive Session at any time. DSMB members, the Project Director and the Project Officer will attend these sessions.

The Director, NHLBI will make the final decision on whether or not to accept the DSMB's recommendation to discontinue any component of the study.

XL Data Management

Report distribution

At least five types of reports will be generated:

1) Recruitment reports: These are expected to be generated at least weekly, by clinic and region and for the antihypertensive trial and the lipid-lowering trial. These will be distributed to the Project Office, Steering Committee and Regional Coordinators.

2) Other routine monitoring reports include data on visit and medication adherence, quality control and study endpoint documentation. These will be generated at least monthly by clinic and region, and will be distributed to the Project Office, Steering Committee and Regional Coordinators.

3) Reports for clinic use include randomization verification reports, visit schedules and reminders, endpoint documentation reports, and limited cross-forms edits. These will be generated no more often than monthly. Visit schedules are generated as participants are randomized and include all visit windows and expected special procedures for the duration of the study. Reports and appropriate listings will be sent to the clinics, and summary reports will be sent to the Project Office, Steering Committee and Regional Coordinators.

4) Steering Committee reports will be generated for annual meetings and will be similar to routine recruitment and monitoring reports.

5) Data and Safety Monitoring Board reports will include recruitment and monitoring data by treatment group for both the antihypertensive trial and the lipid-lowering trial. They will also include summary reports of data from the central laboratory and biennial and event ECG data, as well as study endpoints by treatment group.

Quality Control

All clinical sites will be required to attend one of three regional training sessions. These training sessions will include orientation to the study protocol, blood pressure measurement training and certification, orientation to the ECG procedures, and training in completion and transfer of study forms.

Periodic refresher training will be held in conjunction with regularly scheduled Steering Committee meetings. These refresher sessions may include a review of correct blood pressure measurement procedures or any problem that may be identified through review of routine monitoring activities.

All forms will be reviewed for completeness and accuracy at the Clinical Trials Center prior to data entry. Any problems identified will be resolved by telephone or facsimile transmission with the clinical site. Study forms will then be double data entered. Limited cross- forms edits will be performed to identify missing forms and procedures. Unblinding

In some special circumstances (e.g., a medical emergency), a patient's assigned treatment group may be revealed. The Regional Coordinators will be the first line of advice in the decision of whether to break the blind or not. If the Regional Coordinator cannot be reached, the investigator should try to contact any of the other Regional Coordinators, or Dr. Davis or Dr. Goff at the Clinical Trials Center, or Dr. Payne or Dr. Cutler at NHLBI to discuss the relevant medical issues. If medically appropriate, an effort will be made to maintain the blinding of the patient and the clinical center investigator. If the regional coordinator agrees, or if the investigator is insistent, the investigator should contact the Clmical Trials Center to determine the unblinded treatment assignment. If there is an emergency and the Clinical Trials Center cannot be contacted, the investigator will reveal the unblinded treatment assignment by contacting a central unblinding facility.

When breaking the blind is determined to be necessary, the circumstances will be documented on a 1-page form by the clinic investigator. The form will be forwarded to the Clinical Trials Center and data entered onto the masterfile as a permanent part of the patient's study record.

Stopping Study Medications

If an investigator believes it is necessary to withdraw a patient from study treatment because of an adverse effect, other symptoms, of physician's judgment, it may not be necessary to break the blind on the patient. The Regional Coordinators will be the first line of advice in the decision of whether to withdraw the patient from study treatment. If the Regional Coordinator agrees, or if the investigator is insistent, the investigator should contact the Clinical Trials Center to inform them that this action is being taken and the reasons for it.

Reimburs ements

The clinical sites will be reimbursed on a capitation basis for each randomization to the antihypertensive trial, randomization to the lipid-lowering trial, each protocol-required follow-up visit, and each completed study endpoint. These payments will be made monthly from the Clinical Trials Center. In order for a reimbursement to be authorized, a study form must be received at. the Clinical Trials Center, all questions regarding that form must be resolved, and the form must be entered onto the study database. In the case of study endpoints, reimbursements will be in several parts made separately for the study form itself, the death certificate or discharge summary, and for additional documentation for the 10% sample for verification. Data Analysis

The primary endpoint of the antihypertensive component of ALLHAT is fatal plus nonfatal CHD. The primary response variable is time from randomization to development of this event. The log rank test [60] will be used to compare each of the non-diuretic treatment groups to the diuretic one. For the secondary endpoints of all-cause mortality, stroke, combined coronary (CHD + revascularization procedures + hospitalized angina) and cardiovascular (CHD + stroke + revascularization procedures + angina [hospitalized or treated] + CHF [hospitalized or treated] + peripheral arterial disease [hospitalized or outpatient revascularization procedure]) outcomes, and end-stage renal disease, the log rank test will also be used. The log rank test will also be used to test if there are treatment differences in the following subgroups for the outcome of fatal and non-fatal CHD — 1) men and women, 2) > 65 years and < 65 years, 3) African Americans, and 4) diabetics and non-diabetics.^' For the outcomes of LVH by ECG, and health-related quality of life, comparison of proportions will be used to see if there are differences in the treatment groups. For the outcome of renal disease, the inverse of the slope of creatinine will be calculated for each participant. The weighted average of the participants' inverses in each treatment group will be calculated and these averages will be compared across groups using the longitudinal models of Laird^' and Ware [61].

The primary endpoint of the lipid-lowering component of ALLHAT is all-cause mortality. The primary response variable is time from randomization to death. The log rank test will be used to compare the group assigned to lipid-lowering^'therapy to the group assigned to no treatment. For the secondary endpoints of fatal and nonfatal CHD, fatal and nonfatal cancer and cause-specific mortality, the log rank test will also be used. The log rank test will also be used to test if there are treatment differences in the following subgroups for the outcome of fatal and non-fatal CHD - 1) men and women, 2) > 65 years and < 65 years, 3) African Americans, and 4) diabetics and 'non-diabetics. For the outcomes of MI by ECG, and health-related quality of life, comparison of proportions will be used to see if there are differences in the treatment groups.

Interim Monitoring and Analysis (See Protocol Addendum 2)

Interim monitoring will focus on patient intake - overall and within clinical center, center adherence to protocol, baseline comparability of treatment groups, sample size assumptions with regard to event rates, crossover rates, competing risk and lost to follow-up, adverse effects data, and effect of treatment on the primary and secondary study outcomes. Interim analyses will coincide with the meetings of the Data and Safety Monitoring Board (DSMB).

We recommend the DSMB use stochastic curtailment for monitoring treatment differences in both the hypertension and the lipid-lowering studies [62,63]. By this method, termination in favor of the alternative hypothesis- (H_a) may be considered if the conditional probability of rejecting the null hypothesis (H₀) at the scheduled end of the study given the current data and assuming H₀ is true, is greater than or equal to some pre-specified value γ₀. Alternatively, termination in favor of the null hypothesis would be considered if the conditional probability of rejecting Ho under the design specified alternative hypothesis is less than some pre-specified value γ_a. With this procedure, the .type I error is inflated slightly above α (depending on the number of looks at the data, the timing of the looks, and the value of γ₀) and the type II error is slightly inflated above β (again, depending on the number of looks at the data, the timing of the looks, and the value of γ_a). The choice of the γ's will be determined by the DSMB.

These monitoring procedures are suggested as guides for the complex and subjective decisions the DSMB must make when considering to continue or to terminate randomization and/or follow-up at each of its meetings.

Hypertension Trial

In this trial, we have three* comparisons of interest — diuretic compared to angiotensin converting enzyme inhibitor, diuretic compared to calcium channel blocker, and diuretic compared to alpha blocker*. Each comparison would have its own monitoring guideline under the Dunnett procedure with α=0.019. Figure 1 shows the 80% stochastic curtailment boundaries for each of the comparisons. The looks will depend on the information time (the number of recorded CHD events divided by the expected number of CHD events) at the calendar time of the Data and Safety Monitoring Board meetings.

Hypertension trial

CP = Conditional power

Figure 1. Monitoring boundaries for the hypertension trial - ACE (or CCB or alpha-blocker*) vs. diuretic

* On January 24, 2000. a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please refer to JAMA.2000;283:1967-1975. We would also wish to use a rule for stopping because of lack of power to show an effect. Here we also propose to use stochastic curtailment or conditional power. We would consider stopping if the conditional power under the proposed alternative hypothesis is less than 10%. Figure 1 also displays this conditional power boundary.

Total mortality will also be monitored in the hypertension tπal.

Lipid Lowering Trial

For this trial we would propose similar procedures as to that of the hypertension trial with the exception that there is only one comparison (all-cause mortality) with an α=0.05. Figure 2 depicts the monitoring boundaries.

Lipid-lowering trial

.* 6

INFORMATION TIME

CP = Conditional power

Figure 2. Monitoring boundaries for the lipid-lowering trial.

XII. Vanguard Phase

The initial six months of the study will comprise a vanguard phase for the full-scale trial. Twenty practices will be selected to carry out this vanguard phase with the goal of randomizing six hundred (of the full complement of 40,000) patients. Objectives of the vanguard phase include:

1. To determine the feasibility of recruitment and follow-up of out-patient hypertensive subjects in office-based practices and hypertension clinics;

2. To determine the proportion of antihypertensive trial subjects eligible and willing to participate in the cholesterol-lowering trial;

3. To optimize strategies for recruitment of at least 55% African- American participants;

4. To develop methods for maximizing adherence to antihypertensive and cholesterol- lowering medication regimens in out-patient hypertensive subjects;

5. To optimize strategies for retention of out-patient hypertensive subjects in office-based practices;

6. To develop methods for standardized endpoint ascertainment in office-based practices and hypertension clinics; and

7. To assess and optimize the effectiveness of various operational strategies, such as drug distribution, Institutional Review Board approval, training, and activities involving the Regional Coordinators.

If the vanguard phase establishes the basic feasibility of this protocol, any necessary modifications will be made and additional practices will be recruited as needed to meet the recruitment goals of the study.

XIII. References

1. Hypertension prevalence and the status of awareness, treatment, and control in the United States: Final report of the subcommittee on definition and prevalence of the 1984 Joint National Committee. Hypertension 1985; 7:457-468.

2. StasonWB. Cost and quality trade-offs in the treatment of hypertension. Hypertension 1989; 13 (SupplI): I-145-I-148.

3. MacMahon SW, Cutler JA, Furberg CD, et al. The effects of drug treatment for hypertension on morbidity and mortality from cardiovascular disease: A review of randomized controlled trials. Prog Cardiovasc Dis 1986; XXLX (3, Suppl 1):99-119.

4. MacMahon S, Cutler JA, Stamler J. Antihypertensive drug treatment: Potential, expected, and observed effects on vascular disease. Hypertension 1990; 335:827-838.

5. 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 randomized drug trials in their epidemiological context. Lancet 1990; 335:827-838.

6. SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension: Final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA 1991;265:3255-3264.

7. Cohen ID, Neaton JD, Prineas RJ, Daniels KA for the MRFIT Research Group. Diuretics, serum potassium, and ventricular arrhythmias in the Multiple Risk Factor Intervention Trial. Am J Cardiol 1987; 60:548-554.

8. Ferrari P, Rosman J, Weidman P. Antihypertensive agents, serum lipoproteins, and glucose metabolism. Am J Cardiol 1991; 67:26B-35B.

9. 1988 Report of the Joint National Committee on Detection, Evaluation and Treatment of High Blood Pressure. Arch Int Med 1988; 148:1023-1038.

10. ^' National High Blood Pressure Education Program. The Fifth Report of the Joint National

Committee on Detection, Evaluation and Treatment of High Blood Pressure. Arch Int MedT 993 ; 153:154-183.

11. Furberg CD, Cutler JA. Diuretic agents versus beta-blockers: Comparisons of effects on mortality, stroke, and coronary events. Hypertension 1989; 13(Suppl I): 1-57— 1-61.

12. Held PH, Yusuf S. Furberg CD. Calcium channel blockers in acute myocardial infarction and unstable angina: An overview. Brit Med J 1989; 299:1187-1192.

13. Yusuf S, Held P, Furberg C. Update of effects of calcium antagonists in myocardial infarction or angina in light of the second Danish Verapamil Infarction Trial (DAVIT-H) and other recent studies. Am J Cardiol 1991: 67:1295-1297.

14. Messerli FH, Oren S, Grossman E. Left ventricular hypertrophy and antihypertensive therapy. Drugs 1988; 35(Suppl 5): 27-33.

15. Yusuf S, Wittes J, Friedman L. Overview of results of randomized clinical trials in heart disease. I. Treatments following myocardial infarction. JAMA 1988; 260:2088-2093. 16. The SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. N En l JMed 1991; 325:293-302,

17. The SOLVD Investigators. Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. N Engl J Med 1992; 327:685-691.

18. Chobanian AV, Haudenschild CC, Nickerson C, et al. Anti-atherogenic effect of captopril in the Watanabe heritable hyperlipidemic rabbit. Hypertension 1990; 15:327-331.

19. Neaton JD, Grimm RH Jr, Prineas RJ, et al. for the Treatment of Mild Hypertension Study Research Group. Treatment of Mild Hypertension Study: final results. JAMA 1993;270:713- 724.

20. Pool JL. Effects of doxazosin on serum lipids: A review of the clinical data and molecular basis for altered lipid metabolism. Am Heart J 1991;121:251-259.

21. Lithell H. Effects of antihypertensive drugs on insulin, glucose and lipid metabolism. Diabetes Care 1991:14:203-209.

22. Sowers J et al. The effects of doxazosin on blood pressure, glucose tolerance and insulin metabolism in patients with mild-to-moderate hypertension and type JJ diabetes mellitus. Presented at the Inter-American Society of Hypertension, April 1993.

23. Grimm et al. Alpha blockade and diuretic treatment of hypertension: A double-blind randomized comparison of doxazosin and hydrochlorothiazide. Presented at the American Society of Hypertension, May 1992.

24. DeQuattro et al. Neurogenic factors in primary hypertension: An alpha-receptor blocker - doxazosin - lowers stress and resting blood pressure. Presented at the American Heart Association 65th Scientific Sessions, November 1992.

25. Hernandez RH et al. Evidence of an antiplatelet aggregation in normotensive subjects and patients with hypertension: An in-vitro study. Am Heart J 1991;121:389-394.

26. Hernandez RH et al. Evidence of an antiplatelet aggregation action of doxazosin in patients with hypertension: An ex-vivo study. Am Heart J 1991;121:395-401.

27. Jansson JH. Effects of doxazosin and atentolol on the fibrinolytic system in patients with hypertension and elevated serum cholesterol. Eur J Clin Pharmacol 1991;40:321-326.

28. Materson BJ, Reda DJ, Cushman WC, et al. for the VA Cooperative Study Group on Antihypertensive Agents. Single-drag therapy for hypertension in men: a comparison of six antihypertensive agents with placebo. N Engl J Med 1993;328:914-921.

29. Report of the British Hypertension Society Working Party. Treating mild hypertension: Agreement from the large trials. BrMed J 1989; 298:694-698.

30. Zusman RM. Alternatives to traditional antihypertensive therapy. Hypertension 1986; 8:837- 842.

31. O'Kelly BF, Massie BM, Tubau JF, Szlachcic J. Coronary morbidity and mortality, pre-existing silent coronary artery disease, and mild hypertension. Annals Intern Med 1989; 110:1017-1026. 32. National Heart, Lung, and Blood Institute Report of the Task Force on Research in Hypertension. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, May, 1991.

33. National Center for Health Statistics. Health. United States. 1988. DHHS Publication No, (PHS) 89-1232. Public Health Service. Washington, D.C. U.S. Government Printing Office, March, 1989, p.99.

34. Whittle JC, Whelton PK, Seidler AJ, Klag MJ. Does racial variation in risk factors explain black- hite differences in the incidence of hypertensive end-stage renal disease? Arch Intern Med 1991 ; 151: 1359-1364.

35. NTH Consensus Conference. Lowering blood cholesterol to prevent heart disease. JAMA 1985; 253:2080-2086.

36. Lipid Research Clinics Investigators. The Lipid Research Clinics Coronary Primary Prevention Trial.Results. I. Reduction in incidence of coronary heart disease. JAMA 1984; 251 :351-364.

37. Frick MH, Elo O, Heinonen O, et al. Helsinki Heart Study: Primary-prevention trial with gemfibrozil in middle-aged men with dyslipide ia. Safety of treatment, changes in risk factors, and incidence of coronary heart disease. N Engl J Med 1987; 317:1237-1245.

38. Holme I. An analysis of randomized trials evaluating the effect of cholesterol reduction on total mortality and coronary heart disease incidence. Circulation 1990; 82:1916-1924.

39. Ravnskov U. Cholesterol lowering trials in coronary heart disease: frequency of citation and outcome. Br Med J 1992; 305:15-19.

40. Rossouw JE, Lewis B, Rifkind BR. The value of lowering cholesterol after myocardial infarction. N Engl J Med 1990; 323:1112-1119.

41. Buchwald H, Varco RL, Mills JP, et al. Effect of partial ileal bypass surgery on morbidity and mortality from coronary heart disease in patients with hypercholesterolemia. Report of the Program on the Surgical Control of Hypercholesterolemia (POSCH). N Engl J Med 1990: 323:946-955.

42. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Summary of the second report of the Nation Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel II). JAMA 1993;269:3015-3023.

43. Feinlieb M, Gillum RF. Coronary heart disease in the elderly: The magnitude of the problem in the United States. In Wenger NK, Furberg CD, Pitt E (eds): Coronary Heart Disease in the Elderly. New York. Elsevier. 1986. pp. 29-59.

44. Muldoon MF, Manuck SB, Mathews KA. Lowering cholesterol concentrations and mortality: A quantitative review of primary prevention trials. Br Med J 1990; 301 :309-314.

45. Canner PL, Berge KG, Wenger NK, et al for the Coronary Drug Project Research Group. Fifteen year mortality in Coronary Drag Project patients: Long term benefit with niacin. J Am Coll Cardiol 1986; 8:1245-1255.

46. Jacobs D, Blackburn H, Higgins M, et al. Report of the conference on low blood ^• cholesterol :mortality associations. Circulation 1992; 86:1046-1060. 47. The Lipid Research Clinics Investigators. The Lipid Research Clinics Coronary Primary Prevention Trial: Results of Six Years of Post-Trial Follow-Up. Arch Int Med 1992; 152:1399- 1410.

48. Anderson KM, Castelli WP, Levy D. Cholesterol and mortality: 30 years of follow-up from the Framingham Study. J Am Med Assoc 1987; 257:2176-2180.

49. Stamler J, Wentworth D, Neaton JD, for the MRFIT Research Group. Is relationship between serum cholesterol and risk of premature death from coronary heart disease continuous and graded? Findings in 356,222 primary screenees of the multiple risk factor intervention trial (MRFIT). JAMA 1987; 256:2823-2828.

50. Manolio TA, Pearson TA, Wenger NK, Barrett-Connor E, Payne GH, Harlan, WH. Cholesterol and heart disease in older persons and women: review of an NHLBI workshop. Ann Epidemiol 1992; 2: 161-176

51. Gordon DJ, Rif ind BM. Cholesterol-lowering and the older patient. Am J Cardiol 1989; 63.-48H-52H.

52. Hoeg JM, Brewer HB. 3-hydroxy-3-methyglutaryl coenzyme A (HMG CoA) reductase inhibitors in the treatment of hypercholesterolemia. J Am Med Assoc 1987; 258:3532-3536.

53. Gordon DJ, Rifkind BM. 3-hydroxy-3-methyglutaryl coenzyme A (HMG CoA) reductase inhibitors: a new class of cholesterol-lowering agents. Ann Intern Med 1987; 107:759-761.

54. Grundy SM. HMG-CoA reductase inhibitors for treatment of hypercholesterolemia. N Engl J Med 1988; 319:24-33.

55. Bradford RH, Shear CL, Chremos AN, Dujovne C, Downton M, Franklin FA, Gould AL, Hesney M, Higgins J, Hurley DP, Langendorfer A, Nash DT, Pool XL, Schnaper H. Expanded Clinical Evaluation of Lovastatin (EXCEL) results. I. Efficacy in modifying plasma lipoproteins and adverse event profile in 8245 patients with moderate hypercholesterolemia. Arch Intern Med 1991; 151:43-49.

56. Brown G, Albers JJ, Fisher LD, Schaefer SM, Lin JT, Kaplan C, Zhao XQ, Bisson BD, Fitzpatrick VF, Dodge HT. Regression of coronary artery disease as a result of intensive lipid lowering therapy in men with high levels of apolipoprotein B. N Engl J Med 1990; 323:1289- 1298.

57. Kane JP, Malloy MJ, Ports TA, Phillips NR, Diehl JC, Havel RJ. Regression of coronary artery disease during treatment of familial hypercholesterolemia with combined drug regimens. JAMA 1990; 284:3007-3012.

58. LaRosa J, Applegate W, Grouse J, Hunninghake D, Grimm R, Knopp R, Eckfeldt J, Davis C, Gordon D. Cholesterol lowering in the elderly - Results of the Cholesterol Reduction in Seniors Program (CRISP) Pilot Study. Arch Intern Med 1994;154:529-539.

59. Stamler J, Stamler R, Neaton J. Blood pressure, systolic and diastolic, and cardiovascular risks. U.S. population data. Arch Intern Med 1993;153:598-615.

60. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemofher Rep. 1966;50:163-170. 61. Laird NM, Ware JH. Random effects models for longitudinal data. Biometrics 1982; 38:963- 974.

62. Lan KKG and Wittes J. The B-value: A tool for monitoring data. Biometrics 1988; 44:579- 585.

63. Davis BR and Hardy RJ. Upper bounds for type I and type II error rates in conditional power calculations. Communications in Statistics 1990; A19(10):3571-3584.

Appendix I: Sample Size Calculations

Power calculations for ALLHAT were done separately for the antihypertensive (AH) and lipid lowering (LL) components. For the AH component power was based on two-tailed comparisons of the diuretic arm to each of the other arms using an overall Type I error rate of .05. This was accomplished using a Dunnett type adjustment for multiple comparisons. For the LL component there was only one comparison, hence no adjustment was needed.

To estimate power for ALLHAT, we had to specify expected event rates, treatment effects, and rates of crossovers and losses to follow-up from competing risks or other reasons.

Antihypertensive Component

We conservatively estimated the CHD rate in the diuretic arm to be about 1.35% a year (a 6 year rate of approximately 7.8%). This was based on exponential regression models applied to data from Framingham 12, Framingham 16, and HDFP. We included from the Framingham data all hypertensives aged 45-75, excluding those with recent MI (within 2 years). The variables included in the model were age, sex, and whether or not the patient was at high risk (defined as meeting ALLHAT entry criteria). Rates were adjusted to a mean age of 67 and a 55% prevalence of males. A similar analysis was done on the HDFP data, which included those 50 years old and older. The stepped and referred care cohorts of HDFP were analyzed separately. The following reductions were applied for secular trends and the healthy volunteer effect.

Table 1: Adjustment Factors

For example, the estimated yearly event rate from the exponential regression of Framingham 12 was multiplied by (l-l/2)(l-l/4)=3/8. The estimated event rates based on the exponential regressions adjusted for secular trends and the healthy volunteer effect were as follows: Table 2: Estimated. Yearly CHD Rates From Framingham and HDFP

We felt that a 1.35% yearly CHD rate was reasonable. SHEP rates were somewhat higher. SHEP rates automatically incorporate a healthy volunteer effect, and since it was a recent trial, its rates would need less of a secular trend adjustment. We decided to consider a range of event rates from 1.05% a year to 1.65% a year.

Crossover rates were estimated from TOMHS. We fit a time-dependent Markov model to the data. A model assuming a 2.75% chance of crossing over to another medication during each of the first three years, and 6% for each of the last three years appears to fit well (see Figure 1). Under this model, about 24%> of all patients will cross over to another medication at least once in 6 years, and about 21% of all patients will be crossed over to another medication at the end of 6 years. A patient in the diuretic arm who crosses over to another active antihypertensive medication is assumed to have a reduced event rate even though some of the other antihypertensives may confer no benefit. A patient in an active antihypertensive arm who crosses over to another antihypertensive medication is assumed to have an increased event rate consistent with the diuretic arm even though that patient may have crossed over to another antihypertensive medication that is as beneficial as that to which he/she was assigned. A' patient may cross over and then cross back. We assume that because the physician will have leeway to select among a wide variety of second line antihypertensive medication, there will be a negligible percentage of patients who are taking no medication whatsoever. This assumption is somewhat anti-conservative, but we feel that it is offset by the conservative assumptions alluded to above. We also considered two other rates with 22%> and 26% probability of crossing over at least once, respectively. These correspond to approximately 20% and 22.5% of patients on another medication at the end of 6 years, respectively (see Figures 2 and 3). The latter rate ^' appears to be quite conservative. The three rates of 22%, 24%, and 26% for at least one crossover will henceforth be referred to as crossover rates 1, 2, and 3, respectively.

Loss from competing risks was estimated to be approximately 8% over 6 years. Tlis was composed of other cardiovascular mortality (2.6% over 6 years) and non-cardiovascular mortality (5.4% over 6 years). These rates were calculated in a manner similar to the way we computed event rates for the primary endpoint. The same healthy volunteer and secular trend adjustments were applied. We added about 1.5% per year for losses to follow-up, yielding a total loss rate of approximately 16.8% over 6 years. We considered two other loss rates of 16.8%±5%. The three loss rates of 11.8%, 16.8%, and 21.8% will henceforth be referred to as loss rates 1, 2, and 3, respectively.

To compute power for the AH component we also had to consider the LL component. It appears from HDFP and Framingham that the patients who qualify for the LL component are at approximately the same risk as those who do not. We therefore made this assumption. Before considering the benefit of antihypertensive medication, we reduced the event rate in the LL active arm by 25% to account for beneficial effects of cholesterol lowering. Note that this is conservative in that it assumes that all of the LL active patients will stay on the drag and receive its full benefit. We assumed that 20,000 of the 40,000 ALLHAT patients would be in the LL component.

We assumed a 20% reduction in event rate in an active antihypertensive arm. We then computed power based on:

1) The optimal allocation of patients to the diuretic and treatment arms, namely the ratio of the number of patients in the diuretic arm to the number in each other AH arm should be:

(total # arms - l)^1/2= -Js (see Table 3).

2) An adjustment for comparisons of each treatment to the diuretic (adjusted critical value of approximately c=2.37).

Table 3: Approximate Allocation of Patients

We used a computer program which estimates trial event rates based on yearly rates of events, crossovers, and losses. We ran this program separately for patients in the LL active and LL placebo arms. Based on these values and the allocation of patients specified in Table 3, we obtained trial event rates for the diuretic arm and the other AH arms. For example, using a

* On January 24, 2000, a recommendation to discontinue the doxazosin arm of the antihypertensive trial was accepted by NHLBI. Please referto JAMA.2000;283:1967-1975. yearly event rate of approximately .0135, crossover rate 2, and loss 2, we estimate that the trial event rates would be as follows:

Table 4: Estimated 6 Year CHD Rates

We estimate the overall event rate in the diuretic arm and another AH arm to be: po=C0710)(.75)+(.0537)(.25)= 0667, p,=(.0595)(.75)+(.0449)(.25)= 0559.

An arcsin transformation was used for the test statistic.

In Formula (1), no and n i are the sample sizes in the diuretic and another AH arm, respectively, andpo and pi are the observed proportion of events in those arms. For the optimal allocation of patients, n₀= 40,OOθ[V3/(3 + V3^~)j = 14,641 and m= 40,OOθ[l/(3 + V3^*)J = 8,453. We will reject if the Z statistic exceeds the adjusted critical value c=2.37. If we denote the cumulative normal distribution function by Φ(x), power can be shown to be:

Recall that for our example using a yearly event rate of .0135, crossover rate 2, and loss 2, we found that po=.0667 and pι=.0559. Substituting these values and no=l4,641 and m=8,453 into Formula (2), we get power =.824. Lipid Lowering Component

The primary endpoint for the lipid lowering component is total mortality. Wε assume that the vital status of all participants can be ascertained from the National Death Index, hence there will be no loss to follow-up.

Based on previous experience with HMG CoA reductase inhibitors, we feel that compliance will be quite good, with the bulk of noncompliance occurring early in the trial. We also estimate that given the cost of LL agents and the relatively modest lipid levels of the patients, there will not be many LL placebo patients taking active LL medication. We assume that each year about 2% of all LL placebo patients will take active LL medication with a benefit similar to that of Pravastatin (2% dropin per year). This means that about 11.4% of all LL placebo patients will take active medication at least once in 6 years. We further assume that 5%> of the LL active patients will stop taking their medication at some point in the first year, and that in each of the remaining years about 2.5% of LL active patients will stop taking their medication ( dropout rate of 5% in the first year and 2.5% each year thereafter). This corresponds to approximately 16.3% of LL active patients stopping their medication at least once in 6 years. Under the above assumptions, approximately 10.6% of all LL placebo patients will be taking active medication at the end of 6 years, and about 15.3% of the LL active patients will ^"be off their medication at the end of 6 years. We also considered a more pessimistic set of assumptions, namely a yearly dropin rate of 2.5% and a dropout rate of 6% during the first year and 3% a year thereafter. The first set of assumptions will be referred to as dropin/dropout 1, and the more pessimistic set of assumptions will be referred to as dropin/dropout 2.

We estimated the mortality rate to be approximately 2.35% per year. This estimate was based on separate incidence rates for CHD mortality (between 4.5% and 5% over 6 years), other cardiovascular mortality (about 2.6% over 6 years), and non-cardiovascular mortality (about 5.4% over 6 years). These separate rates were estimated using the parametric regression methods we used for the primary endpoint.. The same secular trend and healthy volunteer adjustments were made as were made for the AH component. This gave an estimated mortality rate between 2.25% and 230% per year. We increased this to 2.35%/year based on a somewhat higher mortality rate observed in SHEP. We considered a range of yearly mortality rates between 2.20% and 2.50%.

It is difficult to estimate a reasonable percent reduction in mortality from LL medication. A 14% reduction would occur if there were a 30% reduction in CHD death, a 15% reduction in other cardiovascular mortality, and no reduction in non-cardiovascular mortality. This seems a little optimistic. We felt that a 12.5% reduction was reasonable, so we considered three different percent reductions, 11%, 12.5%, and 14%o.

We reduced the mortality rate in the three AH arms other than the diuretic by 10%.

Six year mortality rates were computed for the LL placebo and LL active arms in a manner similar to the calculations for the AH component. Separate rates were computed for the diuretic arm and other AH arms, and a combined rate was obtained using the allocations shown in Table 3. The 6 year rates in the LL component of ALLHAT are shown in Table 5 below. Table 5: Estimated 6 Year Mortality Rates

The 6 year rates in the LL placebo and active arms are therefore:

p₀= 1-73/(3 + V3)|.13253)+ [3/(3 + V3)](.11998) = .12457

pi = [V3/(3 + 73 )J(.11903) + [3/(3 + 73 )|.10769) = .11184

Power was again computed using the arcsin transformation. The only differences between this calculation and the AH calculation are that the sample sizes n₀ and m in the LL placebo and active arms are equal, and there is no adjustment for multiple comparisons. Thus we can use Formula 2 with no=nι=T 0,000 and c=1.96.

Results

Power for the AH component under different assumptions is depicted in Figures 4-6. There is not a great difference in power under the different assumptions. Looking at crossover rate 2 and loss 2, we see that the power under the anticipated diuretic rate of 1.35% per year is 82.4%, as we calculated above.

Power for the LL total mortality component is depicted in Figure 7. We see that the power is almost exactly 80%o under dropin/dropout rate 1 and a 12.5% reduction in mortality from LL treatment. It drops to 68.6% if there is only an 11 % reduction in mortality (other assumptions as before). On the other hand, the power will be 88.1% if there is a 14% reduction in mortality from LL treatment (other assumptions as before). Under the more pessimistic dropin dropout rate 2, the power is 76.9% under a 2.35%/yr mortality rate and a 12.5% reduction in mortality from LL treatment (not shown in the graph).

Appendix II: DetaOed Definitions of Coronary Death, Nonfatal Acute Myocardial Infarction and Stroke

Coronary Death:

Death consistent with coronary heart disease (CHD) as underlying cause on death certificate, plus any one of the following:

1) Pre-terminal hospitalization with myocardial infarction,

2) Previous angina or myocardial infarction and no known potentially lethal non-coronary disease process

3) Death within 24 hours of symptoms of CHD or death within 24 hours without symptoms but with no known potentially lethal non-coronary disease process (includes instantaneous death and unwitnessed death)

4) Death resulting from a procedure related to coronary artery disease such as coronary bypass grafting (CABG) or percutaneous transluminal coronary angioplasty (PTCA).

Note: Deaths due to a non-coronary underlying cause in which the terminal event was an MI shall be ascribed to the underlying cause — not to CHD.

Coronary death will be subclassified as:

1) Definite fatal MI: no known non-atherosclerotic cause and definite MI within 4 weeks of death,

2) Definite fatal CHD: no known non-atherosclerotic cause and one or both of the following: chest pain within 72 hours of death or a history of chronic ischemic heart disease in the absence of valvular heart disease or non- ischemic cardiomyopathy, or

3) Possible fatal CHD: no known non-atherosclerotic cause, and death certificate consistent with underlying cause.

In patients without a known potentially lethal non-coronary disease process, coronary death will also be classified as rapid or non-rapid, based on whether death did or did not occur within 24 hours after the onset of symptoms (or the time at which the patient was last seen without symptoms).

- Nonfatal Acute Myocardial Infarction

The identification and coding of definite clinical acute myocardial infarction (MI) should be based on meeting at least two of the following three generally accepted criteria, consistent with the World Health Organization criteria:

1. Symptoms (such as chest pain) compatible with an acute MI of at least 20 minutes duration.

2. ECG changes compatible with an acute MI, such as new persistent ST segment elevation of > 0.1 mV or new pathologic Q waves (QRS>0.04 sec), each in two contiguous leads.

A serum biochemical marker compatible with an acute MI, such as:

X Total CK at least twice the upper limits of normal with an MB fraction of >5%> of total CK, or

X Troponin >2 times the upper limit of normal, or

X LDH1/LDH2 ratio >1.

Stroke

Rapid onset of persistent neurological deficit attributable to an obstruction or rupture of the arterial system, including stroke occurring during surgery, that is not known to be secondary to brain trauma, tumor, infection, or other non-ischemic cause. The deficit must last more than 24 hours unless death supervenes or there is a demonstrable lesion compatible with acute stroke on CT orMRI scan.

A. Non-fatal stroke (either of the following):

1) Unequivocal objective findings of a localizing neurological deficit (with or without recent onset of severe headache or loss of consciousness), AND duration longer than 24 hours AND absence of other disease process causing neurological deficit such as neoplasm, subdural hematoma, cerebral angiography, or metabolic disorder, and/or

2) Diagnosis of stroke based on abnormality demonstrated by CT or MRI consistent with current neurological symptoms or signs, or positive lumbar puncture (for sub arachnoid hemorrhage).

B. Fatal stroke: Death certificate listing stroke as consistent with, underlying, or immediate cause of death, plus any one or more of the following:

1) Preterminal hospitalization with stroke as defined above,

2) Previous stroke and no known potentially lethal non-cerebrovascular disease process, and/or

3) Stroke diagnosed as cause of death at post-mortem examination.

Appendix III: Electrocardiographic Criteria for Silent Myocardial Infarction, Ischemia, Left Ventricular Hypertrophy, and Bundle Branch Block

The ALLHAT study records electrocardiographic information from baseline and biennial clinic visits and from a sample of acute hospitalizations that are designated as quality control events. Clinic ECGs are evaluated for prevalent and interim events including "silent" myocardial infarction.

Prevalent and Interim ECG Findings at Clinic Visits

A determination that an ALLHAT participant has prevalent MI, ischemia, LVH or bundle branch block can be made using Minnesota Code criteria. Interim MI, ischemia, LVH or bundle branch block can be made using the criteria shown for simultaneous comparison of ECGs.

Prevalent Baseline ECG Findings

1. Prevalent MI Baseline ECG coded: a) any 1 -1 -x code OR b) any 1-2-x PLUS 4-1-x, or 4-2, or 5-1, or 5-2

2. Prevalent Ischemia Baseline ECG coded: a) any 4-2 through 4-1-x b) any 5-2 through 5-1

3 Prevalent LVH

Baseline ECG coded: a) 3-1 or 3-3 (soft LVH) b) 3-1 or 3-3 PLUS any 4-3 through 4-1-x, or 5-3 through 5-1 (hard

LVH)

4. Prevalent bundle branch block Baseline ECG coded: a) 74-1 b) 7-2-1 c) 7-4

Interim ECG Events

An evolving ECG pattern between the baseline visit and an ECG from a later visit confirmed by simultaneous ECG comparison documents the interim event.

a. Interim Ml Any EDI through ED7 b. Interim ischemic event Any EVl -EV9 pattern c. Interim LVH or progression / regression of LVH E-LVH 1 through E-LVH 4 d. Interim bundle branch block . EBBB 1 through EBBB 3

Attachment 1 Simultaneous Comparison of ECGs

Simultaneous ECG Comparison Explanations:

• A code 1-2-6 is considered no Q-code for the purposes of serial comparison.

• An Equivocal Q-code is a 1-2-8 or any 1-3-x code.

• A Diagnostic Q-code is any 1-1-x or any 1-2-x except 1-2-6 or 1-2-8. « The designation of "ED" means evolving diagnostic Q-code pattern. β All ED patterns are confirmed as significant increase by serial comparison.

• An EDI through ED7 cannot be assigned if a 7-1-1 code is present.

• ^• An ED2 through ED7 cannot be assigned if a 7-2-1 or 7-4 code is present.

• The designation "EV" means evolving ST-T wave pattern.

• All EV patterns are confirmed as significant increases or decreases by serial comparison. β An EVl through EV9 cannot be assigned if a 7-1-1, 7-2-1, or 7-4 code is present.

Significant Serial Change Patterns

Definite Q-wave MI (evolving diagnostic pattern)

EDI. No Q-code (or a 1-2-6) in baseline ECG followed by a record with a Diagnostic

Q-code (Minnesota Code 1-1-1 through 1-2-5 or 1-2-7), confirmed as a significant increase.

OR

A 1-2-8 or any 1-3-X code in baseline ECG followed by a record with any 1-1-X code, confirmed as a significant increase.

ED2a. An Equivocal Q-code (1-2-8 or any 1-3-x code) and no major ST depression in baseline ECG followed by a record with a Diagnostic Q-code (1-1 -1 to 1 -2-5 or 1-2-7) PLUS a major ST depression (4-1-X or 4-2), confirmed as a significant increase. ED2b. An Equivocal Q-code (1-2-8 or any 1-3-x code) with pre-existing major ST depression (4 1-X or 4-2) in baseline ECG followed by a record with a Diagnostic Q-code (1-1 - 1 to 1-2-5 or 1-2-7) PLUS more severe ST depression (4-1-X), confirmed by a significant increase.

ED3a. An Equivocal Q-code (1-2-8 or any 1-3-x code) and no major T-wave inversion in baseline ECG followed by a record with a Diagnostic Q-code (1-1-1 to 1-2-5 or 1- 2-7) PLUS a major T-wave inversion (5-1 or 5-2), confirmed as a significant increase.

ED3b. An Equivocal Q-code (1-2-8 or any 1-3-x code) with pre-existing major T-wave inversion (5-1 or 5-2) in baseline ECG followed by a record with a Diagnostic Q-code (1-1 - 1 to 1 -2-5 or 1 -2-7) PLUS more severe T-wave inversion (5-1 or 5-2), confirmed as a significant increase.

ED4a. An Equivocal Q-code (1-2-8 or any 1-3-x code) and no ST elevation in baseline ECG followed by a record with a Diagnostic Q-code (1-1 -1 to 1 -2-5 or 1 -2-7) PLUS an ST segment elevation (9-2), confirmed as a significant increase.

ED4b. An Equivocal Q-code (1-2-8 or any 1-3-x code) with pre-existing ST elevation (9-2) in the baseline ECG followed by a record with a Diagnostic Q-code (1-1 -1 to 1 -2-5 or 1 -2-7) PLUS more severe ST elevation (9-2), confirmed as a significant increase.

ED5. No Q-code (or a 1-2-6) and neither 4-1-X nor 4-2 in baseline ECG followed by a record with an Equivocal Q-code (1-2-8 or any 1-3-x code) PLUS 4-1-X or 4-2, confirmed as a significant increase.

ED5b. No Q-code (or a 1 -2-6) with pre-existing major ST depression (4-1 -x to 4-2) in baseline ECG followed by a record with an Equivocal Q-code (1-2-8 or any 1-3-x code) PLUS more severe ST depression (4- 1-X), confirmed as a significant increase.

ED6a. No Q-code (or a 1 -2-6) and neither 5-1 nor 5-2 in baseline ECG followed by a record with an Equivocal Q-code (1-2-8 or any 1 -3-x code) PLUS a 5-1 or 5-2, confirmed as a significant increase.

ED6b. No Q-code (or a 1 -2-6) with pre-existing major T-wave inversion (5-1 or 5-2) in baseline ECG followed by a record with an Equivocal Q-code (1-2-8 or any 1-3-x code) - PLUS more severe T-wave inversion (5-1 or 5-2), confirmed as a significant increase.

ED7a. No Q-code (or a 1 -2-6) and no 9-2 in baseline ECG followed by a record with an Equivocal Q-code (1-2-8 or any 1-3-x code) PLUS a 9-2, confirmed as a significant increase.

ED7b. No Q-code (or a 1 -2-6) with pre-existing ST elevation (9-2) in the baseline ECG followed by a record with an Equivocal Q-code (1-2-8 or any 1-3-x code) PLUS more severe ST elevation (9-2), confirmed as a significant increase.

Diagnostic ECG:

Dl. An ECG record with a Diagnostic Q-code (1-1-1 to 1-2-5 or 1-2-7). D2. An ECG record with ST segment elevation (9-2) PLUS T wave inversion (5-1 or 5-2).

Evolving ST-T Pattern:

^• This diagnosis cannot be assigned if a 7-1-1, 7-2-1, or 7-4 is present. For hospitalized participants, the EV patterns can occur from either increases or decreases in the severity of the code.

EVl . Either 4-0 (no 4-code), 4-4 or 4-3 in baseline ECG followed by a record with 4-1-

1, 4-1-2, or 4-2, confirmed as a significant increase; OR for hospital ECGs only, 4-1-1, 4-1-2, or 4-2 in the earliest hospital ECG followed by an event record with a 4-0, 4-4, or 4-3, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG or follow-up ECG but no significant increase in Q codes found.

EV2. Either 4-2 or 4-1-2 in baseline ECG followed by a record with 4-1 -1, confirmed as a significant increase, or a 4-2 in baseline followed by a record with 4-1 -2, confirmed as a significant increase; OR for hospital ECGs only, 4-1-1 in the earliest hospital ECG followed by an event record with a 4-1-2 or 4-2, confirmed as a significant decrease, OR 4-1-2 in the baseline ECG followed by a 4-2 in the follόw-up ECG, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s). present in baseline ECG or follow-up ECG but no ignificant increase in Q codes found.

EV3. Either 5-0, 5-4 or 5-3 in baseline ECG followed by a record with 5-2 or 5-1 , confirmed as a significant increase; OR for hospital ECGs only, 5-1 or 5-2 in the earliest hospital ECG followed by an event record with a 5-0, 5-4 or 5-3, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG Or follow-up ECG but no significant increase in Q codes found.

EV4. Code 5-2 in baseline ECG followed by a record with 5-1, confirmed as a significant increase; OR for hospital ECGs only, 5-1 in the earliest hospital ECG followed by an event record with a 5-2, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG or follow-up ECG but no significant increase in Q codes found. EV5. Code 9-0 in baseline ECG followed by a record with 9-2, confirmed as a significant increase; OR for hospital ECGs only, 9-2 in the earliest hospital ECG followed by an event record with a 9-0, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG or follow-up ECG but no significant increase in Q codes found.

EV6. Code 4-1-1 in baseline ECG followed by a record with 4-1-1, confirmed as a significant increase; OR for hospital ECGs only, 4-1-1 in the earliest hospital ECG followed by an event record with a 4-1-1, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG or follow-up ECG but no significant increase in Q codes found.

EV7. Code 5-1 in baseline ECG followed by a record with 5-1, confirmed as a significant increase; OR for hospital ECGs only. 5-1 in the earliest hospital ECG followed by an event record with a 5-1, confirmed as a significant decrease.

Pins

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG or follo -up ECG but no significant increase in Q codes found.

EV8. Code 5-2 in baseline ECG followed by a record with 5-2, confirmed as a significant increase; OR for hospital ECGs only, 5-2 in the earliest hospital ECG followed by an event record with a 5-2, confirmed as a significant decrease.

Plus

Either no Q-code in both the baseline ECG and the follow-up ECG OR Q-code(s) present in baseline ECG or follow-up ECG but no significant increase in Q codes found.

EV9a. No Q-code (or a 1-2-6) and pre-existing major ST depression, T wave inversion, or ST elevation (4-2 / 4-1-x, 5-2 / 5-1, or 9-2) in baseline ECG followed by an Equivocal Q-code (1-2-8 or any 1-3-x code) and less severe or absent ST depression, T wave inversion, or ST elevation (4-0 /4-4 / 4-3 / 4-2 / 4-1-x, 5-0 / 5- 3 / 5-2 / 5-1, or 9-0 / 9-2), with Q-code confirmed as a significant increase, and ST or T wave changes confirmed as a significant decrease.

EV9b. An Equivocal Q-code (1-2-8 ^"or any 1-3-x code) and pre-existing major ST depression, T wave inversion, or ST elevation (4-2 / 4-1-x, 5-2 / 5-1, or 9-2) in baseline ECG followed by a Diagnostic Q-code (1-1-1 to 1-2-5 or 1-2-7) and less severe or absent ST depression, T wave inversion, or ST elevation (4-0 / 4-4 /4-3 / 4-2 / 4- 1-x, 5-0 / 5-3 / 5-2 / 5-1, or 9-0 / 9-2), with Q-code confirmed as a significant increase, and ST or T wave changes confirmed as a significant decrease. Evolving Bundle Branch Block

E-BBB 1. No 7-1-1 in reference followed by an ECG with 7-1-1 with the QRS duration increased by > 0.02 sec, confirmed as a significant increase.

E-BBB 2. No 7-2-1 in reference followed by an ECG with 7-2-1 with the QRS duration increased by > 0.02 sec, confirmed as significant increase.

E-BBB 3. No 7-4 in reference followed by an ECG with 7-4 with the QRS duration increased by > 0.02 sec, confirmed as a significant increase.

Evolving LVH

E-LVH 1. No 3-1 in the reference ECG followed by an ECG with a 3-1, confirmed as a sigificant increase.

E-LVH 2. No 3-3 in the reference ECG followed by an ECG with a 3-3, confirmed as a significant increase.

E-LVH 3. 3-1 in the reference ECG followed by an ECG with 3-1 or no 3 code confirmed as a significant decrease.

E-LVH 4. 3-3 in the reference ECG followed by an ECG with 3-3 or no 3 code confirmed as a significant decrease.

Equivocal ECG Pattern:

EL An ECG record with a 1-2-8; or an ECG record with a 1-3-x in the absence of 7-2-1 or 7- 4.

E2. An ECG record with ST-segment depression (4-1-x, 4-2, or 4-3).

E3. An ECG record with T-wave inversion (5-1, 5-2, or 5-3).

E4. An ECG record with ST-segment elevation code 9-2.

Other ECG Pattern:

All other ECG findings, including normal.

Uncodable ECG Pattern:

UI. Technical errors coded 9-8-1 by Minnesota Code. a. Three or more missing leads. b. Muscle tremor artifact that produces possible false initial R's. c. Other technical errors making Q-wave measurement impossible, such as extreme lack of centering, or marked clipping. d. Other conditions defined as "uncodable" by the Minnesota Code. Absent ECG:

A. No ECG available for coding.

Appendix TS⁷: Quality Control Evaluation of Hospitalized Myocardial Infarctions (MI)

The following definitions for acute hospitalized MI are not intended for the classification on the AL04 by site investigators, but rather for quality control assessment review by the Endpoints Subcommittee.

The criteria presented are based on the CCSP Pilot Study, the Minnesota Heart Survey, and other surveillance studies as incorporated into the ARIC study. The combinations of pain, ECG and enzyme categories required for each diagnosis below are approximately the same as those contained in the above-mentioned documents.

Definite Hospitalized MI

Must meet one or more of the following criteria:

1. Evolving diagnostic ECG pattern (EDI - ED7) (Appendix HI, Attachment 1) OR

2. Diagnostic ECG pattern (Dl or D2) and abnormal enzymes (Appendix HI, Attachment 1) OR

3. Cardiac pain (defined below) and abnormal enzymes AND a. Evolving ST-T pattern EV1-EV9 (Appendix HI, Attachment 1) OR b. Equivocal ECG pattern El through E4 (Appendix JJI, Attachment 1)

Probable in-hospital MI

Must meet one or more of the following criteria in the absence of sufficient evidence for Definite Hospitalized MI:

1. Cardiac pain and abnormal enzymes or

2. Cardiac pain and equivocal enzymes and a. Evolving ST-T pattern or b. Diagnostic ECG pattern or 3. a. Abnormal enzymes and b. Evolving ST-T pattern

Suspect in-hospital MI

Must meet one or more of the following criteria in the absence of sufficient evidence for Definite or Probable in-Hospital MI.

1. Abnormal enzymes or

2. Cardiac pain and incomplete enzymes and a. Diagnostic ECG pattern or b. Evolving ST-T pattern or

3. Cardiac pain and equivocal enzymes or

Equivocal enzymes and a. Diagnostic ECG pattern or b. Evolving ST-T pattern or c Equivocal ST-T pattern

The definitions of specific elements of chest pain, enzymes and ECGs which contribute to the final diagnosis of definite, probable, suspect, or no MI are- provided below. Definition of Cardiac Pain

Pain having both of the following characteristics:

1. It occurs anywhere in the anterior chest, left arm, or jaw. and

2. Absence of a definite non-cardiac cause of pain.

Abnormal Cardiac Enzymes

Enzymes are classed as abnormal if any enzyme values recorded meet any of the following criteria:

La. CK-MB is "present": ( if laboratory uses the criterion of "present" or "absent" without reporting a more specific value) or the CK-MB is greater than or equal to 10% of the total CK value, and b. There is no known non-ischemic cause (cardiac surgery, severe muscle trauma, rhabdomyolysis) for the elevated enzyme value. or

2.a. The ratio LDHi : LDH₂ >1. and b. There is no evidence of hemolytic disease. or

3. . Total CK and LDH are both at least twice the upper limits of normal. (These increases do not have to occur on the same day). and b. There is no known non-ischemic cause (cardiac surgery, severe muscle trauma, rhabdomol sis) for the elevated enzyme value and no evidence of hemolytic disease.

Equivocal Cardiac Enzymes

Enzymes are classed as "equivocal" if the criteria for abnormal enzymes are not met and if: 1. Either total CK or total LDH are at least twice the upper limits of normal. or

2. Both total CK and total LDH are between the upper limits of normal and twice the upper limits of normal. (These increases do not have to occur on the same day.) or

3. CK-MB - 5-9% of total CK or is "weakly present."

A summary of the enzyme diagnostic criteria, as related to total CK and LDH is given in the following algorithm, Table 1.

Table 1. Algorithm for Total CK and LDH Enzyme Diagnostic Criteria

Table 2 Proposed ALLHAT Diagnostic Criteria for In-Hospital MI

Cardiac

Pain ECG Findings Enzymes Diagnosis

Present Evolving Diagnostic Abnormal Definite MI

Pattern (ED1-ED7) Equivocal Definite MI

Incomplete Definite Ml

Normal Definite MI

Diag. ECG Pattern Abnormal Definite MI

Equivocal Probable MI

Incomplete Suspect MI

Normal No MI

Evolving ST-T Abnormal Definite MI

(EVl - EV9) Equivocal Probable MI

Incomplete Suspect MI

Normal No MI

Equivocal ECG Abnormal Definite MI

Pattern Equivocal Probable MI

Incomplete No MI

Normal No MI

Absent, Uncodable, Abnormal Probable MI or other Equivocal Suspect MI

Incomplete No MI

Normal No MI

Not Evolving Diagnostic Abnormal Definite MI

Present, Pattern (ED1-ED7) Equivocal Definite MI

Unknown Incomplete Definite Ml or Missing Normal Definite MI

Diag. ECG Pattern Abnormal Definite MI

Equivocal Suspect MI

Incomplete No MI

Normal No MI Evolving Abnormal Probable MI ST-T Pattern Equivocal Suspect MI (EV1-EV9) Incomplete No MI Normal No MI

Equivocal ECG Abnormal Suspect MI

Pattern Equivocal Suspect MI

Incomplete No MI

Normal No MI

Absent, Uncodable Abnormal Suspect MI or other Equivocal No MI

Incomplete No MI

Normal No MI

PROTOCOL ADDENDUM 1 Revised Power for Lipid-Lowering Portion of ALLHAT

Power for the lipid-lowering portion of ALLHAT was revised based on a total sample size of 10,000 and subgroups of size 4,000. The dropout rate was taken as the more pessimistic of the two scenarios in the protocol-6% in the first year and 3% per year thereafter. The dropin was also taken as the more pessimistic of the two protocol scenarios, namely 2.5% per year. The loss for total mortality was taken to be 0, while for the CHD it was as specified in the protocol for the antihypertensive component (slightly over 3% per year). The event rates shown are after accounting for benefits of antihypertensive therapy. The power results are shown in Tables 1 and 2. A more pessimistic loss of 5% per year was also considered. This reduced power by about 2 percentage points.

Table 1: Power for total mortality

Loss assumed to be 0, dropout=6% in first year, 3% per year thereafter, dropin=2.5% per year.

Table 2: Power for CHD

Same dropout and dropin as above. Power assuming the loss rate in the protocol is given.

PROTOCOL ADDENDUM 2 Monitoring Boundaries

When data are examined repeatedly in a trial, the type 1 error rate can be increased dramatically if no account is taken of these multiple "looks" (Armitage, Mc.Pherson and Rowe, 1969; Proschan, Follmann, and Waclawiw, 1992). Boundaries have been constructed in two- armed trials that either eliminate or greatly ameliorate this type 1 error rate inflation. These are called group-sequential rather than sequential boundaries because interim analyses are conducted after groups of data accrue rather than after every new observation. The most popular of these require extremely strong evidence to declare significance very early in the trial, and rouglly the same degree of evidence at the end of the trial that would be required had there been no interim looks. This was not the case for the first group-sequential procedure proposed (Pocock, 1977). For this reason. Pocock himself recommends against using his procedure (personal communication).

The O'Brien-Fleming (1979) boundary has the above properties. The drawback is that it requires the number of looks to be specified in advance, and these looks must occur after equal increments of information. In the context of this trial, information time t is estimated by the ratio of the number of events observed thus far to the number expected by the end of the trial. Thus t=0 is the beginning and t=T is the end of the study. Suppose we planned eight looks at the data after equal increments of information (numbers of events). The boundary is given in Figure 1 for the comparison of a given arm with the diuretic. A multiple comparison adjustment has been made for the three comparisons with the diuretic.

Lan and DeMets (1983) proposed a "spending function" approach which allows the data to be examined after different amounts of information, and does not require the number of looks to be specified in advance. The spending function α*(t) represents the cumulative amount of type 1 error probability that is spent by information time t, with α*(l)=α. The amount of type 1 error probability spent is inextricably linked to information time. This contrasts with the method of Slud and Wei (1982), which spends a fixed amount of type 1 error probability at each look at the data, regardless of the-information time. Serious inflation of the type 1 error rate can occur with the Slud and Wei procedure, but not with the Lan-DeMets procedure (Proschan, Follmann, and Waclawiw, 1992). One of the spending functions suggested by Lan and DeMets (see the equation in the Appendix) has boundaries very similar to the O'Brien-Fleming boundary if the looks happen to occur after equal increments of information (see Figure 1). The advantage of the Lan-DeMets approach is that it can be used even when the looks are not equally spaced. In this case the boundaries would change somewhat

Another extremely useful monitoring tool is conditional power (Lan, Simon, and Halperin, 1982 or Lan and Wittes, 1988). The conditional probability of obtaining a statistically significant result at the end of the trial is computed under different hypothesized treatment effects. Unlike the O'Brien-Fleming or Lan-DeMets boundaries, conditional power is usually used to justify terminating a trial which has no realistic chance of producing a statistically significant result. The trial is stopped if the conditional power is very low even assuming a large treatment benefit for the remainder of the trial. Stochastic curtailment refers to stopping a trial because the conditional power crosses a pre-specified threshold value. For example, one could agree to stop the trial if the conditional power assuming the pre-specified alternative hypothesis is less than or equal to .10. If one uses such a rule, the chance of a type 2 error (accepting the null hypothesis when it is false) is greater than it would be without stochastic curtailment. This is because one could accept the null hypothesis at the end of the study or at an interim point. The degree of type 2 error rate inflation is quite small. Lan, Simon, and Halperin (1982) showed that it is fairly small even if one monitors the trial continuously. Davis and Hardy (1990) showed that in the more realistic situation in which a trial is monitored 5 to 10 times, the inflation is much smaller.

An important issue that comes up in group sequential monitoring is that of information time. We mentioned above that information time is estimated using the number of events observed thus far divided by the number expected by the end of the trial. But the number in what arms? In two armed monitoring one could either use the^' total number of events in both arms or the number of events in the control arm. The advantage of using both arms is that it provides a larger sample size to estimate information time. A disadvantage is that in order to estimate the number of events to expect by the end of the trial, one has to project not only a control group event rate, but a treatment effect as well. We have four antihypertensive arms in ALLHAT, hence three treatment effects to specify. It is recommended that we use the diuretic arm events to determine information time.

References

Aπnitage, P., Mc.Pherson, C.K., and Rowe, B.C. (1969). Repeated significance tests on accumulating data. Journal of the Royal Statistical Society A. 132, 235-244. Lan, K.K.G. and DeMets, D.L. (1983). Discrete sequential boundaries for clinical trials. Biometrika 70, 659-663.

Lan, K.K.G., Simon, R., and Halperin, M. (1982). Stochastically curtailed tests in long term clinical trials. Communications in Statistics-Sequential Analysis 1(3), 207-219.

Lan, K.K.G. and Wittes, J. (1988). The B-value: a tool for monitoring data. Biometrics AA, 579- 585.^'

O'Brien, P.C. and Fleming, T.R. (1979). A multiple testing procedure for clinical trials. Biometrics 35, 549-556.

Pocock, S.J. (1977). Group sequential methods in the design and analysis of clinical trials. Biometrika 64, 191-199.

Proschan, M.A., Follmann, D.A., and Waclawiw, M.A. (1992). Effects of assumption violations on type 1 error rate in group sequential monitoring. Biometrics 48, 1131-1 143.

Slud, E. and Wei, L.J. (1982). Two sample repeated significance tests based on the modified Wilcoxon statistic. Journal of the American Statistical Association 77, 862-868.

APPENDIX

Recommended Spending Function

The Lan-DeMets spending function we recommend is where α is the two-sided type 1 error rate for a given comparison with diuretic, Φ is the standard normal distribution function, /and Z_{α 4} is its 100(l-α/4)th percentile. When we adjust for multiple comparisons with the diuretic, the value of Zα_/ becomes approximately 2.64. PROPOSED STOPPING RULES

1) Use Lan-DeMets version of O'Brien-Fleming for harm/benefit.

2) The boundaries will be symmetric.

3) Information time will be calculated as proportion of expected events in diuretic arm.

4) Take first look at about 10% of information time and then annually at DSMB meetings.

5) Pay special attention at beginning to results that cross the Haybittle-Peto boundary of Z=4.0 for the antihypertensive component, Z=3.0 for the lipid-lowering component.

6) Use conditional power under the protocol specified alternative hypothesis for futility.

rtensive and Lipid-Lowering Prevent Heart Attack Trial

The University of Texas School of Public Health

Coordinating Center for Clinical Trials

ssB&ssssssass&sfssi&aβsasam

The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) is a practice- based clinical trial sponsored by the National Heart, Lung, and Blood Institute (NHLBI). The trial is being conducted in approximately 600 office-based practices and general medical and specialty clinics throughout the U.S.A., Puerto Rico, the Virgin Islands, and in Canada.

A total of 42.418 patients were enrolled between February 14, 1994 and January 31 , 1998, a large percentage of whom are African-American, A vanguard phase was conducted in the first half of 1994; the full-scale trial began in the fall of 1994 and will continue for eight years, until March 31 , 2002.

The study has two components:

• An antihypertensive component, to determine whether newer antihypertensive agents, such as ACE inhibitors, calcium blockers, and alpha blockers, reduce incidence of coronary heart disease (CHD) in high-risk hypertensives when compared to diuretics.

• A lipid-lowering component, to determine whether reduction of serum cholesterol with pravastatin, an HMG-CoA reductase inhibitor, reduces total mortality in moderately hypercholesterolemic older hypertensives.

Because the efficacy of lipid-lowering can be tested in a subset of patients targeted for study in the antihypertensive component of ALLHAT, the two trials have been combined for approximately the cost of conducting either one alone.

Because African-Americans suffer disproportionately from hypertension. and its sequelae, a large percentage of participants will be African-American.

Background

Antihypertensive Component in the early 1980's, three new classes of antihypertensive agents, the calcium antagonists, angiotensin- converting enzyme (ACE) inhibitors, and alpha-adrenergic blockers, were developed and licensed for use in chronic antihypertensive therapy. Some of these agents can cost up to 30 times as much as older therapies such as diuretics and beta blockers. They are also believed to have fewer side effects and may have ancillary properties (in addition to their blood-pressure lowering effects) that could reduce morbidity and mortality from coronary heart disease (CHD).

Comparison of these agents in such studies as the TOMHS {Arch Intern Med 1991; 151:1413-1423) and the VA Monotherapy Trial (N Engl J Med 1993; 328:914-921 ), however, has failed to show major differences in side effects or blood pressure lowering. Efficacy of newer agents in preventing CHD has not been evaluated in large-scale clinical trials. Yet, despite their increased cost and lack of proven superiority over older agents, their use (particularly use of calcium antagonists and ACE inhibitors) has increased dramatically in the past five to ten years.

Lipid-Lowering Component

Elevated cholesterol is known to be a major CHD risk factor, but trials demonstrating a reduction in CHD from cholesterol lowering have not demonstrated a net reduction in all-cause mortality. Since these trials have been conducted primarily in middle-aged men, the extrapolation of their findings to older men and women has also been questioned. The introduction of HMG-CoA reductase inhibitors in the late 1980's provided a powerful new lipid-lowering therapy that is well-tolerated and has few adverse effects, A recent NHLBI study conclusively demonstrated the feasibility and efficacy of lowering cholesterol with lovastatin in older adults (Arch int Med 1994; 154:529-539).

ALLHAT Study Design

The rationale and design for ALLHAT has been published (Am J Hypertension 1996; 9:342-360). In ALLHAT, hypertensive patients High-risk are randomly assigned to receive one of four drugs in a double- blind design. No patient receives placebo, and a limited choice of second step agents are provided for patients not controlled on first- line medication. Patients are followed every 3 months for the first year and every 4 months thereafter for an average of 6 years of follow-up.

Approximately 10,000 of the patients in the antihypertensive component were also randomized to diet plus lipid-lowering or diet plus usual care in an unblinded design. After randomization, frequency and content of follow-up visits for the two trial components are identical.

ALLHAT Study Candidates

• Antihypertensive Component o Age 55 years or older o Known hypertensive with BP less than or equal 160/100 mmHg on treatment, or BP greater than or equal 140/90 mmHg and less than or equal 180/110 without treatment . o At least one of the following:

1. Left ventricular hypertrophy on ECG or echocardiogram

2. Known atherosclerotic CVD

3. Type II diabetes mellitus

4. HDL cholesterol <35 mg/dl

5. Current cigarette smoker • Major Exclusions: o Recent MI or stroke

Q Known congestive heart failure or angina pectoris o Need for any study drug for reasons other than hypertension o Need for more than two antihypertensive drugs to control BP Serious systemic disease Elevated serum creatinine (2 mg/dl or greater) Lipid-Lowering Component

° Eligible and randomized in antihypertensive component o LDL cholesterol 120-189 mg/dl (100-129 for patients with CHD)

Major Exclusions: o Current use of lipid-lowering medications o Contraindications to HMG-CoA reductase inhibitors o Known untreated secondary cause of hyperlipidernia o ALT > 2 times upper li it of normal

Revised February 20, 2002

THE UNIVERSITY OF TEXAS-HOU HEALTH SCIENCE CENTER

News Article fVfonday Morning

Monday, April 17, 2000

Study Shows Blood Pressure Drug Lowers Risk of Cardiovascular Disease

ALLHAT researchers want to "Our study results proved that

By Jackie Preston know if cheaper drugs, such as at present, chlorthalidone diuretics, are better in should be the first line of

Office of Public Affairs combating high blood pressure defense in treating older adults and its adverse side effects, with high blood pressure at risk including heart attack and for heart disease," Davis said. stroke.

Chlorthalidone, a common diuretic used to treat people The School of Public Health -with high blood pressure, The study followed 24,335 was awarded a $103.2 million dramatically decreases the risk patients age 55 and older with contract from the NHLBI to of cardiovascular disease high blood pressure and at least coordinate ALLHAT in 1993, compared to doxazosin, an one of several cardiovascular making it the largest contract alpha-blocker, say researchers disease risk factors, such as ever awarded to UT-Houston. at UT-Houston. diabetes and a history of stroke. . The patients were randomly More than 4.6 million assigned to receive one of four

The findings were reported by Americans suffer from heart high blood pressure drugs, School of Public Health failure, a leading cause of including chlorthalidone and researchers as part of the disability and death in the U.S. doxazosin. Antihypertensive and Lipid- and the most common hospital Lowering Treatment to Prevent discharge diagnosis among Heart Attack Trial (ALLHAT), "The doxazosin group had a 25 people age 65 and older. which is an eight-year, multi- percent increased risk for center study funded by the cardiovascular disease and a More than 50 million Americans National Heart, Lung and Blood doubled risk for heart failure have high blood pressure, or Institute (NHLBI). The study compared with the hypertension. The disease is 50 results will be published in the chlorthalidone group," Davis percent more prevalent in April 19 issue of the Journal of said. African-Americans than in the American Medical whites. Association. The NHLBI halted the doxazosin part of the study

"We showed that study early based on the findings. participants who took chlorthalidone cut their risk of Davis said the study may lead heart failure in half, compared to to better therapies for older those who took doxazosin," said patients with high blood Barry R. Davis, M.D., Ph.D., pressure. professor of biometry at the School of Public Health and deputy director of the school's Coordinating Center for Clinical Trials.

April 20, 2000 Relationship of Antihypertensive Treatment Regimens and Change Blood Pressure to Risk for Heart Failure in Hypertensive Patients Randomly Assigned to Doxazosin or Chlorthalidone: Further Analyses from the Antihypertensive and Lipid-Lowering treatment to prevent Heart Attack Trial

Barry R. Davis, MD, PhD; Jeffrey A. Cutler, MD; Curt D. Furberg, MD, PhD; Jackson T. Wright Jr., MD, PhD; Michael A. Farber, MD; James V. Felicεtta, MD; and John D. Stokes, MD, for the ALLHAT Collaborative Research Group*

Background: The Antihypertensive and Lipid-Lowering treatλ'feasure.me.nts: Data on blood pressure, medication, and inciment to prevent Heart Attack Trial reported that treatment initident heart failure (treated outside hospital, hospitalized, or fatal) ated with doxazosin compared with chlorthalidone doubled the from February 1994 through December 1999. risk for heart failure in high-risk hypertensive patients (relative risk, 2.04 [95% σ_r 1.79 to 2.32]). Patients assigned to doxazosin Results: After the treatment groups were categorized as having therapy had a mean in-trial systolic/diastolic blood pressure 3/0 no exposure to open-label medications (monotherapy) or exposure mm Hg higher than that in patients assigned to chlorthalidone. to open-label therapy, the relative risk for heart failure with doxSixty-eight percent (6167 of 9061) of the former patients and azosin versus chlorthaiidone was 3.10 (CI, 2.51 to 3.82) and 1.42 59% (9081 of 15256) of the latter patients were given additional (CI, 1.20 to 1.59), respectively. After adjustment for follow-up medications to achieve a target blood pressure of less than systolic diastolic blood pressure, the overall relative risk was 2.00 140/90 mm Hg. (Ci, 1.72 to 2.32).

Objective: To ascertain the influence of open-iabe! antihyperConclusion: In high-risk patients with hypertension, the higher tensive drugs and subsequent blood pressure on relative risk for risk for heart failure while taking doxazosin compared with heart failure. chlorthalidone is attenuated but not eliminated by adding other

Design: Randomized, double-blind, active-controlled clinical trial. antihypertensive drugs. The small observed difference in systolic blood pressure does not explain this increased risk.

Setting: 623 sites in the United States and Canada.

Patients: Hypertensive patients 55 years of age or older with at least one additional risk factor for cardiovascular disease. Ann Intern Med 2002; 137,313-320. mvWJmnals.Drg

For author affiliations, see end of text

Iαtervention: Chlorthalidone (12.5 to 25 mg/d) or doxazosin (2 " For a complete list of members of the ALLHAT Collaborative Research Group, to 8 mg/d) for a planned foilow-up of 4 to 8 years. see JAMA. 2O0D;283 1973-5

"he Antihypertensive an Lipid-Lowering treatment to 2.32]; P < 0.001), was significantly increased compared prevent Heart Attack Trial (ALLHAT), a randomized, with the chlorthalidone arm. Equally important, a low double-blind, rnulticenter clinical trial including 42 418 probability existed that doxazosin would show benefit over participants at 623 clinical sites, is designed to determine chlorthalidone for τhe. primary end point by the scheduled whether treatment begun with a calcium-channel blocker study end, given the lack of difference (relative risk, 1.02 (amlodipine), an angiotensin-converting enzyme inhibitor [95% CI, 0.90 to 1.17]; P > 0.2] at that time. These (lisinopril), or an α-adrenergic blocker (doxazosin) comintention- to- treat analyses compared patients assigned to pared to treatment with a diuretic (chlorthalidone) reduces chlorthalidone with patients assigned to doxazosin. the incidence of fatal coronary heart disease or nonfatal In this article, we analyze how treatment changes may myocardial infarction in high-risk patients with hypertenhave affected the comparison of doxazosin with chlorthalision (I). Secondary end points are all-cause mortality, done in terms of heart failure. Our major objectives are to stroke, and other cardiovascular events. A lipid-lowering ascertain to what extent the relative risk with doxazosin trial in a subset of 10 336 participants is designed to deversus chlorthalidone depends on 1) whether the assigned termine whether decreasing cholesterol levels with a hy- drugs were used as monotherapy or in combination with droxymethyl glutamyl coenzyme A reductase inhibitor other agents and 2) the difference in decreases in systolic (pravastatin) compared with usual care reduces all-cause and diastolic blood pressure. mortality in older, moderately hypercholesterolemic patients. In January 2000, the doxazosin arm of the trial was METHODS discontinued. (2) because ^•major cardiovascular disease (relStudy Design ative risk, 1.25 [95% CI, 1.17 to 1.33]; P < 0.001), espeThe rationale and design- of ^'ALLHAT are described in cially heart failure (relative risk,. 2.04 [95% CI, 1.79 to detail elsewhere (1). In brief, eligible participants were men

®;2002 American College of Physicians-American Society oftnrema] Medicine E-313 ART! CLE I Multidrug Therapy -And Changes in Blood Pressure in ALLHAT

ALLHAT defined symptomatic heart failure as clear-cut-

Context signs or symptoms of left or right ventricular dysfunction

Previously published results of this randomized, double- chat cannot be attributed to other causes. A patient had to blind trial showed that hypertensive patients treated with have at least one symptom (paroxysmal nocturnal dyspnea, doxazosin rather than chlorthalidone more often develdyspnea at rest, New York Heart Association class III dysoped heart failure. Did differences in blood pressure conpnea or other symptoms [on less than ordinary exertion], trol or treatment with other drugs explain these findings? or orthopnea) and one sign (rales, ankle edema, tachycar¬

Contribution dia, cardiomegaly or characteristic pulmonary pattern on chest radiography, S₃ gallop, or jugular venous distenrion)

This analysis shows that the slightly worse blood pressure (3). Symptoms and signs were determined by the clinic control achieved with doxazosin did not explain higher investigatot through patient history, chart review, or conrisks for heart failure. Risks with doxazosin versus chlorthalidone were highest among patients who received sultation with the treating physician, but such data were monotherapy (relative risk, 3.10) and lowest among panot collected centrally. tients who received stepped care with additional drugs A one-time sample of 24 hospitalized or fatal heart (relative risk, 1.42). failure events was reviewed in a blinded manner- by the ALLHAT Endpoints Subcommittee; 20 of 24 (83%) were

Implications deemed to have complete data for a definitive diagnosis. Of

Increased risks for heart failure with doxazosin versus the 20 cases, the agreement rate between the subcommittee chlorthalidone are attenuated but not eliminated when and the clinic investigators was 90% (18 of 20) and similar additional antihypertensive drugs are used. in both treatment groups. Details on the validity of diagnosis of heart failure will be provided in a future report.

-The Editors

Measurement and Treatment of Blood Pressure

The ALLHAT protocol specified a sτepped-care treatand women 55 years of age or older who had systolic or ment program for hypertension. Trained observers using diastolic hypertension (≥ 140/90 mm Hg or hypertension standardized techniques measured blood pressures during controlled with medication) plus at least one additional the trial (1, 2). All blood pressures were calculated as the risk factor for coronary heart disease events. The risk facaverage of two measurements obtained with a 30-second tors included previous (>6 months) myocardial infarction interval between them. The blood pressure goal in all four or stroke, left ventricular hypertrophy on electrocardiαgra- study arms was less than 140/90 mm Hg. This level was to phy or echocardiography, history of type 2 diabetes, curbe achieved with the lowest possible dose of blinded first- rent cigarette smoking, and a low high-density lipoprotein line drug, with addition of second- and third-line open- cholesterol level. Persons with a history of hospitalized or^' label therapy as needed after reaching the maximal dose of treated symptomatic heart failure or a known ejection fracfirst-line drug. tion less than 0.35 were excluded. Chlorthalidone and doxazosin were to be taken once

. Onless the drug regimen required tapering for safety, daily in the morning. By design, doses were selected to participants^' who had been taking antihypertensive medicaachieve equivalent blood pressure control in the treatment tions continued ro do so until the day of randomization, at groups. The first, second, and third dosage levels were 12.5 which point they stopped taking all previous medications. mg/d, 12.5 mg/d (sham titration), and 25 mg d, respecOn the day after randomization, treatment with the study tively, for chlorthalidone and 2 mg/d, 4 mg/d, and 8 mg/d drug was initiated. for doxazosin. Double-blinded 1-m and 12.5-mg doses of

Enrollment occurred from February 1994 through doxazosin and chlorthalidone, respectively, were used for^' January 1 8. The original reported number of 42 448 die first week to minimize the frequency of postural hypoparticipants and 625 sites changed because 30 patients tension associated with doxazosin. The identity of drugs with poor documentation of informed consent were exwas masked at each dosage level, but dosage level was not. cluded (2). Participants were assigned by a compuτer-gen- After randomization, patients were seen for dose titraerated randomization schedule in a ratio of 1.7:1 to receive tion as needed at 1 and 3 months per the protocol, but chlorthalidone or doxazosin. Randomization was stratified they could return more often until target blood pressure by center and blocked over time to maintain the ratio. All was reached. Subsequently, required visits occurred at participants gave written- informed consent.- and all centers 3-mσnth intervals during the first year and at 4-month obtained institutional review board approval. intervals thereafter. Open-label second-step drugs were

Ascertainment of Outcomes added as needed and tolerated. These agents were reserpine

At each clinic visit, occurrence of study end points was (0.05 to 0.2 mg/d), clonidine (0.1 to 0.3 mg twice daily), assessed by the clinical investigator. A hospital discharge and atenolol (25 to 100 mg/d). The choice of second-step summary was required for each hospitalized study outdrug was at the discretion of the treating clinician-investicome, and a death certificate was required for each death. gator. The third-step agent, was hydralazine (25 to 100 g

£.-314 3 September 2002 Annals of Inreπiai Medicine Volume 137 ^■ Number 5 (Part i) ^' ivwwjnnals.org- Mulπdnig herapy and Changes mBlood Pressure in ALLHAT I ART-ΪCLE

Figure. Participants who underwent randomization and were followed in the monotherapy and open-label therapy analyses, by study treatment group. ^

Values in parendieses are percentages of total participants. * Patients disconunued from therapy with randomly assigned drug. ACE anζiotensm- converting enzvme; HF = heart failure. twice daily). Details of these therapies are provided elsestrongly discouraged to avoid dilution of treatment comwhere (I). parisons. When clinical conditions other than uncontrolled

Investigators could choose to prescribe open-label anhypertension (for example, angina or heart failure) were tihypertensive drugs other than those provided by the presenr, drugs from any study class could be used. If a step study. However, use o£ the drug classes under study — thi1 drug was specifically required for blood pressure control azide diuretics, calcium antagonists, angiotensin-convert- while blinded drug treatment was continued, the step 1 ing enzyme inhibitors, and α-adrenergic blockers — as drug could be added as open-label therapy but the dose www.annals.orj 3 September 2002 Annals of internal Medicme Volume 137 • Number 5 (Parr JE-315- ARTICLE | Multidru Therapy and Changes in Blood Pressure m ALLHAT

Table 1. Baseline Characteristics of Participants

Characteristic No Exposure IO Open- _^abel Drugs (Monotherapy) Exposure to Ope n-Label Therapy

ChlorthaUdαne Group Doxazosin Group Chlorthalidone Group Doxazosin Group

Particpants, n 6175 2894 8969 5967

Age. y" 668 ± 79 66 8 ± 8 0 67 0 = 7 5 66 8 = 7 6

Ethnicity, %

White, non-Hispanic 37 5 37 8 53 8 50 5

Black non-Hispanic 33 4 283 30 9 35 1

White Hispanic 18 8 22 5 8 3 7 9

Black hispanie 5 1 6 0 2 0 2 2

Other 3 9 5 5 4 9 4 3

Women, % 51 3 507 44 1 44 4

Level of education, y 1050 ± 4 - 104 + 4 5 11 3 = 4 0 11 3 ± 3 7

Current cigarette smoking, % 24 5 23 4 20 2 20 9

Receiving antihypertensive treatment before 85 5 8 9 93 3 92 8 study entry, %

Atherosclerotic cardiovascular disease, % 39 9 42 1 48 8 47 3

ST-T wave, % 9 9 8 5 102 10 5

Type 2 diabetes, % 35 4 35 0 36 2 35 1

Low high-density lipoprotein cholesterol level, % 11 9 11 5 12 7 12 2

Left ventπcular hypertrophy, %

On elertrocardiography 15 0 14 3 164 16 7

On echocardiography 3 5 3 2 4 6 4 6

Blood pressure, mm Hg*

Systolic 143 9 ± 15 3 144 4 ± 15 4 147 8 = 15 7 147 2 ± 15 8

Diastolic 83 8 ± 9 8 83 9 £ 9 9 84 1 = 10 3 83 9 ± 10 1

Serum potassium level, mmoM.* 4 1 ± 07 4 4 i 0 8 4 3 = 0.7 4 3 ± 0 6

Fasting serum glucose level, mmob'L (mg/dL)" 6 3 1 3 4 (124 ± 61) 6 8 = 3 3 (126 = 59) 6 9 = 3 1 (124 ± 56) 6 8 ± 3 0 (126 = 54)

Serum creatinine concentration, μmcl L 77 8 ± 23 3 (0 8 = 03) 77 8 £ 23 3 (0 8 ± 03) 77 8 = 23 6 (0 8 ± 03) 77 8 ± 22 9 (0 8 ± 0 3)

(mg/dL)*

Serum cholesterol level, mmol/l (mgJdV

Total cholesteiol 5 6 = 1 1 (216 = 42) 5 6 - 1 ¹ (216 ± 42) 5 6 = 1 1 (216 ± 42) 5 6 ± 1 1 (216 ± 42)

Low density lipoprotein cholesterol 3 5 i i 0 (135 . ± 37) 3 5 + 1 0 (135 = 37) 3 5 = 1 0 (135 ± 37) 3 5 ± 0 9 (135 ± 35)

High-αens y lipoprotein cholesterol 1 2 = 0 4 ⁽46 ± 15) 1 2 1 0 4 (46 = 5) 1 2 = 04 (46 ± 15) 1 2 ± 0 1 (46 = 15)

Fasting tπglycεπαes 2 0 - 1 5 (177 - r 133) 1 9 ± 1 9 (168 ± 168! 2 0 = 1 5 (177 - : 133) 9 = 1 3 (168 = 115)

¹ Values Vith the sign Me the rneun =. SD

could not exceed one half the maximum recommended by randomized" analyses was the interval from randomization the Fifth Joint National Committee (4). In most such to first diagnosis of heart failure for patients who had an cases, unblinding the patient or the investigator to treatoutcome, or from randomization to study end or loss to ment assignment was not necessary. (Overall, <1% of follow-up for those without an outcome. In this study, drug identities were provided to either participant or invesoutcome rates were compared by treatment group for partigator.) Any of nine possible open-label drug types — diticipants with no exposure to open-label medication, and uretic, calcium-channel blocker, angiotensin-converπng results of this comparison were contrasted with rates in the enzyme inhibitor, α-adrenergic blocker, atenolol, reserrespective treatment groups with such exposure. Similar pine, clonidine, hydralazine, and other antihypertensive comparisons were done for patients with exposure and drug (which could include nonatenolol β-blockers, althose with no exposure to antihypertensive heart failure though drug name was not logged) — could be recorded at medications. Dose— exposure analyses were done to comeach visit. Because diuretics, angiotensin-convert g enpare the influence of the two doses of chlorthalidone with zyme inhibitors, and /3-blockers are beneficial in preventthat of the three doses of doxazosin (six possible compariing or treating heart failure, we call these antihypertensive sons) on the observed relative risk. heart failure drugs. Data on antihypertensive drug use beCumulative event fates were calculated by using the fore study enrollment were collected, but not by drug class. Kaplan-Meier procedure (5). Relative risks (hazard ratios)

Statistical Analysis with 95% CIs and two-sided P values were calculated by

Data were analyzed for use of additional drugs accordusing a proportional hazards model (5). The drug-exposure ing to participants' randomized treatment assignments arid and no-drag-exposure analyses were also performed by usregardless of continuation of blinded study treatment. The ing Cox regression with a time-dependent covariate repreoutcomes "all heart failure" (treated outside hospital, hossenting exposure and an interaction term representing the pitalized, or fatal) and "hospitalized or fatal heart failure" product of this term and drug assignment. Cox regression were examined by treatment group for the entire cohort: analyses limited to the duration of receipt of the various The time to event for the previously published "as- doses of doxazosin and chlorthalidone w«re also per-

E-316|3 September 2002| Λnπ-ιboπnremal edιone| Volume 137 • Number 5 (Part 1) www.aifflals.on: Multidrug Therapy and Changes in Blood Pressure in ALLHAT | AR-T-ICLE

formed. Because these time-dependent analyses are subject lection, analysis, and interpretation of the data plus the to statistical and epidemiologic biases, the P values, relative decision to submit the manuscript for publication. ~^" risks, and confidence intervals for them should be interpreted with caution (6).

To account for differences in systolic blood pressure RESULTS between the randomized treatment groups as previously

The results presented here are restricted to the same reported, two types of analyses were performed. First, pardata set as the previously published analysis (1). The meticipants were stratified by their blood pressure at 9 to 12 dian duration of follow-up for all participants was 3.3 months after randomization into categories of systolic years. Five hundred (3.3%) patients in the chlorthalidone blood pressure (≥ l4θ mm Hg or < 140 mm Hg), diastolic group and 338 (3.7%) patients in the doxazosin group blood pressure (≥90 mm Hg or <90 mm Hg), and comwere lost to follow-up. During the trial, abour 59% of binations of these blood pressures (eight groupings). Mean patients in the chlorthalidone group and 68% of patients differences in systolic and diastolic blood pressure between in the doxazosin group were stepped- up to additional the treatment groups at this time and event rates, relative drugs. risks, and 95% CIs beyond 1 year were calculated. Second,

The Figure shows the number of patients who were Cox regression analyses of the entire cohort that included randomly assigned and followed to the time of each analfixed covariates of baseline systolic and diastolic blood presysis. It also shows the number of participants who reached sure and time-dependent covariates of follow-up systolic the maximal dose level of blinded step 1 drug and those and diastolic blood pressure were performed. who were stepped up to any open-label drug or any anti¬

Follow-up blood pressures were obtained at clinic vishypertensive heart failure open-label drug before heart failits. Missed visits resulted in missing blood pressures. Reaure occurred. The relative risk for receiving additional sons for missing visits were recorded and included loss to medications before reported onset of heart failure (doxazofollow-up, refusal to return, and intercurrent illness. Analsin versus chlorthalidone) was 1.31 (95% CI, 1.27 to 1.35) yses were done by using all available information (which (P < 0.001). Of patients who never started taking antihydecreased sample sizes owing to missing values) and by pertensive heart failure medication, about one fourth assigning follow-up blood pressures for the missing obserstarted therapy with a calcium antagonist; these patients vations (a measurement not captured in the last 6 months) had few heart failure events (7 and 2 in the chlorthalidone according to time period and treatment group by using the and doxazosin groups, respectively) . - •jnethod of multiple imputation (7, 8).

Table 1 shows baseline characteristics of the chlortha¬

Role of the Funding Source lidone and doxazosin groups, stratified by no exposure to

The National Heart, Lung, and Blood Institute sponopen-label medication and exposure to such drugs. In each sored the study and was involved in all aspects .other than stratum, the distribution of characteristics by treatment direct operations of the study centers. This included colgroup was very similar, with a few exceptions. In the

Table 2. Event Rates and Relative Risks for Al! Heart Failure and Hospitalized or Fatal Heart Failure

Heart Failure and Medication Category Events per 100 Persons Patients with Outcomes Relative Risk P Value at 4 Years* (95% C!)

Chlorthalidone Doxazosin Chlorthalidone Doxazosin

, Group Group Group Group r

All heart failure

As randomly assigned 4.46 ± 0.26 8.14 ± 0.43 420 491 2.04 (1.79-2.32) <0.001

No exposure !o open-label therapy 2.64 ± 0.31 6.63 ± 0.79 144 227 3.10 (2.51-3.82) 0.001

Exposure to open-label therapy 6.93 ± 0.59 8.75 ± 0.72 276 264 1.42 (1.20-1.69) <0.001

Exposure to antihypertensive heart failure open-label therapy 6.90 ± 0.69 9.81 = 0.95 207 214 ^■ 1.45- (1.20-1 .76) < 0.001

No exposure to antihypertensive heart failure open-label therapy 6.10 ± 1.22 7.43- ± 1.62 69 50 1.25 (0.87-1 .80) 0.2

Hospitalized or fatal heart failure As randomly assigned 3.53 ± 0.23 5.77 ± 0.37 327 346 1.83 (1.58-2.13) <0.001 No exposure to open-label therapy 2.03 = 0.28 3.69 £ 0.57 109 139 2.52 (1.96-3.24) <0.001 Exposure to open-label therapy 5.64 ± 0.54 ' 6.77 £ 0.62 218 207 1.39 (1.15-1 .68) <0.001 Exposure to antihypertensive heart failure . open-label therapy ^• 5.36 ± 0.64 7.83 £ 0.87 172 172 1.38 (1.12-1.71) <0.001 No exposure to antihypertensive heart failure open-label therapy 4.01 ± 0.90 4.61 ± 1.22 46 35 1.31 (0.85-2.04) >0.2

* Data are presented as the mean ≤ SE. hvm.₃niials.orj 3 September 20021 Annals of internal Medicine I Volume 137 ^■ Number 5 (Part 1) E-317 ARTICLE | Multidrug Therapy and Changes in Blood Pressure inALLHAT

Tabl 3. Relative Risk for Heart Failure, by Dose of Doxazosin Compared with Chlorthalidone, in Participants Not Exposed to Open-Label Therapy

Doxazosin Dosage vs. All Heart Failure Hospitalized or Fatal Heart Failure

Chlorthalidone Dosage

Relative Risk (95% CI) P Value Relative Risk (95% CI) P Value

2 mg/d vs. 12.5 mg/d 2.81 .(2.24-3.53) <0.001 2.27 (1.73-2.98) ^' <C001

4 mg/d vs. 12.5 mg/d 3.51 (2.78-4.43) ■C0.001 2.84 (2.15-3.76) < 0.001

8 mg/d vs. 12.5 mg/d 3.25 G.28-4.64) ^■C0.001 2.81 (1.84-4.30) <0.001

2 mg/d vs. 25 mg/d 2.43 (1.73-3.42) <0.001 1.84 (1.22-2.77) 0.004

4 mg/d vs. 25 mg/d 3.04 (2.10-4.39) <0.001 2.30 (1.48-3.58) <0.001

8 mg/d vs. 25 mg/d 2.81 (2.21-3.53) <0.001 2.27 (1.73-2.98) 0.043

monotherapy stratum, the doxazosin arm included more the dose of doxazosin was increased from 2 mg to 4 mg but black patients than white patients compared with the not when it was increased from 4 mg to 8 mg. For a fixed chlorthalidone arm, whereas in the open-label therapy stradose of doxazosin, relative risks were decreased when the tum, the chlorthalidone arm included more black patients dose of chlorthalidone was increased from 12.5 to 25 mg. than white patients. In addition, in the monotherapy straHowever, all relative risks were substantial and consistent tum, the doxazosin arm included more Hispanic patients with the overall results. and more persons with atherosclerotic cardiovascular disThe results of the two outcome comparisons for the ease but. fewer persons with ST-T wave abnormalities. -entire cohort shown in Table 2 are not controlled for About 30% of participants who had been taking antihyfollow-up blood pressure. Throughout the trial, systolic pertensive therapy before study entry had a blood pressure blood pressure was 2 to 3 mm Hg higher in the doxazosin less than 140/90 mm Hg. Overall, the mean blood presgroup than the chlorthalidone group, but diastolic blood sures for persons receiving previous therapy (90.2%) and pressure did not differ substantially between the groups those not receiving previous therapy were 145/83 mm Hg (2). The proportions of missing blood pressure measureand 156/89 mm Hg, respectively. ments -were 7% at 1 month, 16% at 6 months, 16% at 1

Table 2 shows results of Cox regression analyses comyear, 20% at 3 years, and 16% at 5 years. paring the treatment groups with respect to heart failure Table 4 shows event rates and relative risks of the outcomes for all patients as randomly assigned, those with treatment groups for heart failure outcomes beyond 1 year, no exposure to open-label medication, and those with exwith stratification to control for blood pressure at ] year. posure to such medication. The 4-year event rates are The stratum with the lower systolic or diastolic blood preshigher among participants exposed to open-label medicasure had a higher relative risk and a smaller difference in tion than in those not exposed. For both outcomes, the mean systolic blood pressure for doxazosin versus chlortharisk with doxazosin treatment compared with chlorthalilidone. In the combination groupings, the stratum with the done treatment is substantial, in patients without and with best control (blood pressure < 140/90 mm Hg) had the exposure to open-label drugs. The risk ratios in the monogreatest relative risk and only a small difference in mean therapy groups (3.1 for all cases of heart failure and 2.5 for systolic blood pressure. After adjustment for baseline and hospitalized or fatal heart failure) are greater than those in follow-up blood pressures, the results were essentially unthe as-randomized groups. The risk ratios for participants changed for all heart failure (relative risk. 2.04 [CI, 1.79 to with exposure to open-label medication are smaller (about 2.32] vs. 2.00. [CI, 1.76 to 2.28]; P < 0.001) and hospi1.4) but still statistically^, significant [P < 0.001). talized or fatal heart failure (relative risk, 1.83 [CI, 1.58 to

Cox regression with a time-dependent covariate repre2.13] vs. 1.80 [CI, 1.54 to 2.09]; P < 0.001). senting exposure to open-label therapy (or antihypertensive heart failure open-label therapy) plus the interaction term of exposure and treatment assignment yielded nearly idenDISCUSSION tical results. When race, Hispanic ethnicity, presence of We found that doxazosin is less effective than atherosclerotic cardiovascular disease, and presence of chlorthalidone in preventing major cardiovascular events, ST-T wave abnormalities were added as covariates, the especially heart failure. Because this was an active control findings were essentially unchanged. trial, we cannot determine whether chlorthalidone was

Table 3 shows the relative risk for all heart failure and beneficial, doxazosin was harmful, or both. Chlorthalidone hospitalized or fatal heart failure in the treatment groups, has been shown- to prevent and treat heart failure (9), according to dose levels of doxazosin and chlorthalidone. whereas doxazosin has not been shown to do either. In the Tliis analysis is limited to duration of follow-up with no Systolic Hypertension in the Elderly Program, risk for exposure to open-label therapy. The increased risk was apheart failure was reduced by 50% with use of chlorthaliparent at all dose levels of doxazosin. At a fixed dose of done compared with placebo (9). chlorthalidone, an increase in relative risk was noted when In ALLHAT, not all participants continued to take

E-31813 September 2002 Annaii of Inrernal Medicine Volume 137 • Number- 5^' (Part 1) www:anπals.oπj Multidrug Therapy and Changes in Blood Pressure in ALLHAT ! ARTICLE

their assigned treatment, and reasons for stepping up to cluding antihypertensive heart failure therapy, diminished open-label medication might have differed between the but did not eliminate the relative risks of both outcomesrin randomized groups These reasons may haΛ e had a differthe two treatment groups The incidence of these outcomes ential effect on subsequent risk for heart failure A major in each treatment group was higher after exposure to other finding of the current study is that higher risk for heart drugs than before such exposure Perhaps participants with failure \\ ith doxazosin compared with chlorthalidone was hypertension that w s harder to control were at greater risk attenuated but not eliminated bv the addition of other for heart failure antihypertensive drugs Analyses by dose that were restricted to participants

The current analyses are subject to indication and ditaking monotherapy are subject to the same potential biagnostic bias (10) Indication bias happens when the invesases described above For a fixed dose of chlorthalidone, tigator gives additional medication on the basis of signs the relative risk tended to increase with increasing doses of and symptoms, for example to control blood pressure or doxazosin, whereas for a fixed dose of doxazosin, the rela- reduce perceived side effects Diagnostic bias can occur nve risk decreased with the increased dose of chlorthaliwhen the investigator or patient is influenced by knowldone Increasing doses of doxazosin relative to increasing edge about treatment Participants in the doxazosin group doses of chlorthalidone were associated with shorter durawere significantly more likely (relative risk, 1 31) than tion of treatment with the medication If doxazosin had no those in the chlorthalidone group to receive other drugs, effect on prevenπon of heart failure, comparison of paand administration of other drugs tended to occur earlier tients taking a fixed dose of chlorthalidone with patients in the doxazosin group Since time to heart failure was taking increasing doses of doxazosin might include an inshorter in the doxazosin group than the chlorthalidone creasing proportion of persons at greater risk for heart failgroup, a greater proportion (46%) of all heart failure events ure owing to harder-to-control blood pressure This effect occuired befoie use of open-label therap in ihe doxazosin would tend to increase die relative risk group compared wth the chlorthalidone group (34%), reFinally, the mean systolic blood pressure was 2 to 3 sulting in an exaggeration of the relative risk Diagnostic mm Hg lower in the chlorthalidone group than the doxbias should not have occurred because open-label therapy azosin group throughout the trial However, the relative had not yet been given Indication bias is possible because risks changed little after adjustment for differences in blood pressures and side erfects may

influenced why blood piessure throughout the trial On the basis of results certain participants were stepped up to additional medicafrom the Framingham Heart Study (11), this observed diftion but otheis were not ference in svstohc blood pressure would be associated with

On the basis of this reasoning, participants with expoat most a 2% reduction m relative risk, a finding consistent sure to open-label medication should have had an attenuwith ours (relative risk, 2 00 after ad_justment [reduced ated relative risk This effect was seen for all cases of heart from 2 04]) In addition, stratification of participants by failure and hospitalized or fatal heart failure In addition, blood pressure at 1 year showed that lower blood pressures xhe presence of other drugs could have furtner influenced weie associated with greater relative increases in risk It the differences because of blood-pressure-related and unappears that in ALLHAT, the degree of difference in, blood related effects Diagnostic and indication bias could have pressure does not explain the increased risk for heart failure played some role, since participants and investigators knew in the doxazosin group compared with the chlorthalidone which open-label drugs were being taken and why The group Possible limitations of our analyses include meapostexposure results indicate that open-label therapy, insurement error in blood pressure and possible loss of bal-

Tablε 4 Event Rates and Relative Risk for Heart Failure after 1 Year of Therapy with Doxazosin or Chlorthalidone, and Differences in Blood Pressure 9 to 12 Months after Randomization

Blood Pressure Rate of Heart Failure Relative Risk with Use of Difference in Systolic Diastolic Blood

Chlorthalidone Group Doxazosin Group (95% CI) and Chlorthalidone Group mm Hg mm Hg

Systolic

≥140 1 48 1 93 - 30 (1 05-1 60) 1 6/0 O

<140 096 1 51 1 58 (1 18-2 13) 04/-1 2

Diastolic

≥90 1 54 1 "»9 1 17 (0 8S-1 53) 43/03

<90 1 09 1 76 1 61 (1 29-2 00) 24/-06

Systolic/diastolic

≥140/≥9O 1 53 1 79 1 17 (0 88-1 55) 23/0 0

≥140/<9O 1 41 2 10 1 49 (1 08-2 05) 1 2/-0 E

<140/≥90 1 61 1 9 1 11 (025-3 98) 02 -O 1

^•^140/<90 092 1 50 1 63 (1 20-2 05) 05/-1 1 wwΛr als org 3 September 2002 Annais of Internal Mem-roe Volume T3T • Number 5 (Pare ]) E-31S ARTICLE | ulndrug Therapy and Changes in 31ood Pressure in ALLHAT

ance of known and unknown characteristics since we used Dr or Medicine, Medical Cenpostrandomization variables ter

* Dr Wright Case Western Reserve University School of VIediαne

In patients at very low πsk for heart failure, such as 10900 Euclid Avenue, Cleveland, OH 44106 young persons with uncomplicated hypertension or those

Dr Farber Pitman Internal λϊedicme Assocπtes, 410 North Broadwrv, without other risk factors for caidiovascular disease, the Pitman J findings of ALLHAT neither support nor refute a strategy Dr Felicetta Girl T Hayden Veterans Affairs Medical Centc 650 Easr of initial therapy with α-adrenergic blockers, allowing adIndian School Ro.d, Phoenix AZ 8501:2 dition to or replacement of other antihypertensive drug Dr S ol.es 2323 Cutlev. Roaα, Palm Harbor, FI 34683 classes if blood pressure is not λvell controlled In older men with benign prostatic hvperrrophv m whom an Author Contributions Conception and design B Davis, J A Cuder, C D Furberg J T Wright α-adrenergic blocker seems like the best treatment for the

Analysis and inte^rpretation of the data B R Davis, JA Cuder, C D uroparhv, coexisting hypertension should be treated with Furberg, I T Wnght another antihypertensive drug as well However, our anal- Drafting of the artic'e B R Davis, JA Cutler J T Wright, MA Farses provide no guidance as to selection of a second drug ber class Crincal revision of the article for mportanr intellectual content B R.

In summary, doxazosin recipients were more hkely Davis, J A. Cutler, C D Furberg, J T Wright, MA Father, J V Felicthan chlorthalidone recipients to be given additional drugs etta, J D Stokes.

Final approval of the artide B R Davis, JA Cutlet, C D Furberg, J T A disproportionately higher rate of heart failure occurred Wright, J V Felicetta m the doxazosin group than the chlorthalidone group beProvision of stud's materials or patients B R. Davis, M A Farber, J D fore exposure to additional medication The data suggest a Stokes dose— response relationship for doxazosin compared with Statistical expertise B R Davis chlorthalidone Stepping up to any drug, even a hearr failObtaining ot ftuiding- B R.

JΛ Cuder ure prevention drug, decreased but did not eliminate the Administrate, e technical or logistic support B R Davis, J A Cutler relative πsk for heart failure events Differences m blood C D Furberg Collection and assembly of data B R

J V Felicetta, J D Stokes pressure during receipt of study treatment appeared to ac count for very little of the observed results The principal finding remains that treatment with doxazosin compared References with chlorthalidone carries an excess πsk for heart failure in ¹ DaΛ is BR Cutler JA Gordon DJ Furberg CD, Wright JT Jr, Cushman high-risk patients with hypertension, regardless of the dose ^"v/C et a! Raπon?k and design for rf Anπhvputcnsrvt. and Lipid Loveπng Treatment to P went Heart Attack Tπal (ALLHAT) ALLHAT Research used or addition of other drugs Group im I Hypercens 1996,9342 60 JPMID 8-22 31

2 Major cardiovascular n hvpertensrve patients randomized lo doxazosin

From the tinvsersiw of Texas School of Public Health, Houston, Texas

vs chlorrl alidone. the ^•muhvoeπensve - ά lipid lowering treatment to prevent the National Heart, Lung, and Blood Institute, Bcthcsda, λ'laryknα, heart attack tnal (ALLHAT) ALLHAT Collaborative Research Croup JAMA. Wake Forest University Scnool or Medicine Wm ton-Salεm, orth 2000 283 1967 "5 _tPMlD ) 0789664] Carolina Case Western Reserve Umversm ichool of εdiunc

3 PiHer LP, Davis BR, Cutler JA, et al Validation of heart Mure events in land, Ohio, Pieman Internal Medicine Associates, Pitman, New Jersey, ALLHAT partiαpjnts assigned ro doxazosin fAb trrct] Am J Hypertens 2001, and Carl T Harden Vetcans Affairs Medical Center, Phoenix, Aruom 14 18LA.

4 The fifth report of the Joint Nauonal Committee on Detecuon, Evduanon,

Grant Support By contract NO-HC-35130 from the National Heart, and Treatment of Hgh Blood Pressure (JNC V) Arch Intern Med. 1593,153 Lung, and Blood Institute Smdy medications were supplied by Pfizer, 154-83 rPMID 8422206] Inc. (amlodipine and doxazosin), AstraZeneca (atenolol and Jisinooπl), Klein JP, Modschberger ML

Techniques for Censored and and Bristol Myers Squibb (prayasτaαn) Pfizer, In , provided financial Truncated Data New York. Spπnger-Verlag; 1 97 support to ie National Heart, Lung, and Blood Institute 6 Jenmson C, Tumbull BW Group Sequciπal Methods with Applicaπσns to Clinical Tnals Boca Raton, FL Chapman & Hall/CRC Pr; 2000

Potential Financial Conflicts of Interest: Consultancies B Da' is, 7 Rubin DB, Scheπker N MUIUDIC imputation for interval εsumanon >τom JT Wright Jr J V Felicetta, Honorana. C D Furbet , J Wright Jr , simple ranoom samples with lgnonble nonresponse Journal of the Ameπcan J V Feliceαa, Stock Ownership M Farber Grants Received C D Statisαcal Assocauon 1986,81 366-74 Furberg, J T Wright Jr , J V Fehcerta. S Schafer JL Vlulπplc imputaπon a n er Stat Method¹; Med Res 1999,83-

15 [PMID 103478571

Requests for Single Reprints Barry R Da' is, MD, PhD, the U ver- 9 Kostis JB, Daws BR, Cuder J, Gπrom RH Jr, Bergε KG, Cohen JD, et al. sity of Texas School of Public Health, 1200 Herman Prεssler Street, Prevention of heaπ hulure by antihvperterisive drug treatment m older persons Houston TX 77030, e-mail, bdavιs@sph uth rrnc sdu with jsolateα systole hypertension. SHEP Cooperative Research Group JAMA 1997278 212-6 {PMID 92186671

Current author addresses _*n author contributions are available at www 10 Peduza P, Wittes J, Detre K, Holfbrd T Analyas as-randomized and the annals org problem of non-adherence an example from the Veterans Affairs Randomized Tπal of Coronary artery Bypass Surgery Star Med. 1993,12 1185-95 [PMID

Current Author Addresses: Dr Davis The University of Texas School 8210821] of Public Health, 1200 Herman Pressler Screet, Houston, TX 7⁷030 11 K2nneIWB. DAgostmo RB, Sιlberdιarz H BelangerAJ, Wilson PW, levy Dr Cutler The National, Heart, Lung, and Blood Institute, 6701 Rock- D Profile for estimating nsk of hearr f ilure Arch Intern Med 19°9,159 1197- ledge Dn-ve, Bethesda, MD 20892 204 [PMID 10371227]

£-320(3 Strembr- 2002 Anna oHnrenial McJic-nrlVolume 137 • Number 5 (Psπ 1} www aπnals-oπr Baseline Characteristics of Participants in the Antihypertensive and Lipid Lowering Treatment to Prevent

Heart Attack Trial (ALLHAT)

Richard H. Grimm, Jr, Karen L. Margolis, Vasilios Papademetriou, William C. Cushman,

Charles E. Ford, Judy Bettencourt, Michael H. Alderman, Jan N. Basile, Henry . Black,

Vincent DeQuattro, John Eckfeldt, C. Morton Hawkins, H. Mitchell Perry, Jr, Michael Proschan, for the ALLHAT Collaborative Research Group

Abstract — Di tiretics and β-blockers have been shown to reduce the risk of cardiovascular morbidity and mortality in peopl e with hypertension in long-term clmical trials. No study has compared newer more costly antihypertensive agents (calcium antagonists. ACE inhibitors, and α-adrenergic blockers) with diuretics for reducing the incidence of cardiovascular disease in an ethnically diverse group of middle-aged and elderly hypertensive patients. The study is a randomized, double-blind, active-controlled clinical trial designed to determine whether the. incidence of the primary outcome, fatal coronary heart disease or nonfatal myocardial infarction, differs between treatment initiation with a diuretic versus each of 3 other antihypertensive drugs. Men and women aged ≥55 years with at least 1 other cardiovascular disease risk factor were randomly assigned to chlorthalidone (12.5 to 25 mg/d), amlodipine (2.5 to 10 mg'd), lisinopril (10 to 40 mg'dj, or doxazosin (2 to 8 mg/d) for planned follow-up of 4 to 8 years. This report describes the baseline characteristics of the Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) participants. A total of 42 448 participants were randomized from 625 sites in the United States, Canada, Puerto Rico, and the US Virgin Islands. The mean age was 67 years, with 35% aged ≥70 years. Among those randomized, 36% were black, 19% were Hispanic, and 47% were women. The sample includes a high proportion of people with diabetes (36%), patients with existing cardiovascular disease (47%). and smokers (22%). There were no important differences between the randomized treatment groups at baseline. ALLHAT will add greatly to our understanding of the management of hypertension by providing an answer to the following question: are newer antihypertensive agents similar, superior, or inferior to traditional treatment with diuretics? {Hypertension. 2001;37:19-27.)

Key Words: hypertension, essential s antihypertensive agents B diuretics a clmical trials m lipids

. ver 40 million people in the United States have elevated stroke, the reduction for coronary heart disease (CHD) events 'blood pressure (BP), ie, systolic BP (SBP) was 10% to 15%, which was less than expected on the basis ≥ 140 mm Hg and/or diastolic BP (DBP) ≥90 m Hg, or of epidemiological data.² they are taking antihypertensive medication.¹ Hypertension Subsequently, in the Systolic Hypertension in the Elderly affects half of white American men and women aged 60 to 74 Program (SHEP) trial, low-dose thiazide diuretic treatment years and over two thirds of black men and women in this age was shown to reduce CHD death and nonfatal myocardial group. Large-scale randomized clinical trials conducted in the infarction (MI) by 27% (95% CI 6% to 43%).³ Olher trials in 1970s and 1980s in largely middle-aged subjects with stage 1 older persons with diastolic and systolic hypertension reand 2 hypertension (DBP 90 to 114 mm Hg) demonstrated ported similar results.⁴-⁵ One possible explanation- for the that antihypertensive drug treatment reduced the rate of failure of earlier trials to demonstrate the expected degree of stroke by 40%. However, in contrast to the findings for CHD reduction is that adverse effects of study drugs, partic-

Received May 2, 2000; first decision May 18, 2000; revision accepted June 30, 2000.

From the Berman Center for Outcomes and Clinical Research an Hennepin County Medical Center (R.H.G., .L.M.). Minneapolis. Minn; Veterans Affairs Medical Center (V.P.), Washington, DC: Memphis Veterans Affairs Medical Center (W.C.C.), Memphis, Tenn; University of Texas-Houston (C.E.F., J.B.. CM.H.), School of Public Healtli, Houston; Albert Einstein College of Medicine (M.H.A.), Bronx, NY; Veterans Affairs Medical Center (J.N.B.), Charleston, SC; Rush-Presbyrerian-St. Luke's Medical Center (H.R.B.), Chicago, 111: Los Angeles County/University of Southern Califo nia Medical Center and White Memorial Medical Center (V.D.), Los Angeles: University of Minnesota Hospital and Clinic (J.E.), Minneapolis: Veterans Affairs Medical Center (H.M.P.), St. Louis, Mo: and the National Heart. Lung, and Blood Instituie (M.P.), Division of Epidemiology and Clinical Applications, Bethesda, Md.

Correspondence to Richard H. Grimm, Jr, MD, PhD, Director, Bεrman Center for Outcomes and Clinical Research, Hennepm County Medical Center-S65B, 701 Park Ave South, Minneapolis, MN 55415.

© 2001 American Heart Association, Inc.

Hypertension is available at htrpt/Λv iv.hypertensionaha.org- 20 Hypertension January 2001

uiarly high-dose thiazide diuretics, may have offset the the study was designed to recruit high proportions of groups potential benefit of BP reduction. These adverse effects heavily burdened by hypertension-related morbidity: the elmclude diuretic-induced hypokalemia, hypornagnesernia. hy- derly, women, African Americans, and people with type 2 peruricemia, hyperhpidemia, hyperglycemia, impaired insudiabetes. In addition to the antihypertensive trial, the trial also lin sensitivity, and, possibly, increased ventricular ectopic randomized a subset of participants to a lipid-lowering trial activity.^5-8 However, these side effects are minimal at curdesigned to compare iota! mortality in patients with mild to rently recommended doses (eg, 12.5 to 25 mg chlorthalimoderate hypercholesterolemia randomized to pravastatin done), and a recent meta-analysis underlined the particular versus usual care. The baseline characieristics of this group CHD beneSt for regimens based on low-dose diuretics.'⁵ will be described in a separate report. -The present report

In the late 1970s and the 1980s, newer and costlier describes in detail the baseline characteristics of the particiantihypertensive agents, such as calcium antagonists, ACE pants in the antihypertensive component of ALLHAT. inhibitors, and α-adrenergic blockers, were introduced for use as antihypertensive agents. However, evidence that might Methods justify their use in preference to the older classes of drugs is The double-blind, randomized, active-controlled design of ALLHAT limited and conflicting. Only a few studies have examined has been described in derail previously.²⁴ The participants in ALLdifferent antihypertensive agents in parallel group trials. The HAT are high-risk hypertensive paiients recraited at 625 clinical sites in the United States, Canada, Puerto Rico, and the US Virgin ACE inhibitor captopril has been compared with diuretics Islands. Recruitment took place between February 14, 1994, and and/or β-blockers in 2 large trials, the Captopril Prevention January 31. 1998. The trial ϊs scheduled to eD in March 2002, after Project (CAPPP)¹⁰ and the UK Prospective Diabetes Study a mean follow-up of 6 years. However, in February 2000. the (UKDPS).¹' Neither study showed an overall advantage for doxazosin arm of the trial was discontinued because of the 4-year captopril in the prevention of the primary cardiovascular end cumulative congestive heart failure rate of doxazosin compared with chlorthalidone (8.1% versus 4.5%. respectively). Additionally, there point. The results of some observational studies and clinical was a 25% higher risk of combined cardiovascular disease outcomes trials have raised questions about the efficacy of calcium in the doxazosin group (relative risk 1.25. 95% CI 1.17 to 1.33).²⁵ antagonists, particularly the short-acting dϊbydropyridiυes, BP eligibility criteria were hased on the patient's current antihyfor preventing cardiovascular events in hypertensive patients pertensive treatment status and on the average of 2 seated BP with heart disease or diabetes ^l2-^1; however, other data measurements at each of 2 visits. For untreated patients (or those treated for <2 months), the BP inclusion criteria at both visits were suggest that the commonly prescribed calcium antagonists are SBP of at least 140 ram Pig or DBP of at least 90 mm Hg. At both safe and effective for preventing cardiovascular morbidity in visits. SBP ≤ i 80 mm Hg and DBP ≤ i 10 mm Hg were required. For these groups of patients.¹⁶~¹ those who had been on treatment with 1 to 2 drugs for ≥2 months,

Four randomized trials have compared representatives of the criteria at visit 1 were SBP ≤ l 60 mm Hg and DBP ≥3 drug classes. The 1-year trial conducted by the Depart≤ 100 mm Hg. and the criteria at visit 2 were SBP ≤ 180 m H and DBP ≤l 10 mm Hg. The higher readings at visit 2 allowed for partial ment of Veterans' Affairs Cooperative Study Group on withdrawal of antihypertensive medication. Patients who were taking Antihypertensive Agents.²⁰ the 4-year HAhfE study,-¹ and the therapeutic doses of >2 anUbypertensive drugs were not eligible. In 4.4-year Treatment of Mild Hypertension Study (TOMHS)²² addition to meeting the BP eligibility criteria, patients had to be at reported some differences in BP control, side effects, quality least 55 years old and have at least 3 additional risk factor for of life, biochemical effects, and target-organ changes. Howcardiovascular morbidity. These additional inclusion criteria are listed in Table 1 , along with the exclusion criteria for the trial. The ever, these differences did not present a pattern that consisinitial design included recruitment goals of 45% women and 55% tently favored one class of drugs over others. These trials did black participants. not have cardiovascular end points as the primary outcome After giving their informed consent, participants were randomly for comparisons of drug classes. The recently completed assigned to receive 1 of 4 double-blinded step 1 treatments given Swedish Trial in Old Patients with Hypertension (STOP-2)²³ once daily: chlorthalidone (12.5 mg for the first and second titration and 25 mg for the third), amlodipine (2.5, 5, or 10 mg), lisinopril (10. compared ACE inhibitors, calcium antagonists, and diuretics 20, or 40 mg), or doxazosin (2, 4. or 8 mg). Each of the study and/or β-blockers in 6614 older patients with hypertension. In medications is identical in appearance at all dosages. Randomization that study, BP reduction and fatal and nonfatal cardiovascular was blocked and stratified by clinical center. Allocation of particievents were similar among the 3 groups. However, STOP-2 pants into each arm was in the ratio of 1.7:1:1:1, with the largest did not include blacks, persons with stage 1 hypertension, or number assigned to chlorthalidone to maximize statistical poweτ for the comparison of the diuretic arm to each of the 3 nondiurctic arms. ' anyone aged <70 years. Furthermore, the results of this trial The initial doxazosin dose was 1 mg for 1 week, followed by 2 mg need confirmation in a larger trial with a broader population for 1 month and monthly titrations thereafter to achieve a BP goal of ■of hypertensive persons. SBP < 140 and DBP <90 mm Hg. Chlorthalidone, amlodipine, and

Thus, more data are needed to permit an assessment of lisinopril were titrated similarly, beginning with the lowest dose, whether the newer classes of drugs are superior, equivalent, except that there was no dosage change after 1 week. Additionally, open-label medications were provided to treat participants who did or inferior to diuretics for lowering the rates of hypertensive not attain satisfactory BP control on the maximum tolerated dose of cardiovascular complications. In particular, the Antihypertenbunded step 1 medication. Three open-label medications were sive and Lipid Lowering Treatment to Prevent Heart Attack available in step 2, and 1 was available in step 3. The step 2 Trial (ALLHAT) study was designed to deteimine whether medications provided include reserpine (0.05 to 0.2 mg daily), the combined incidence of fatal CHD and nonfatal MI differs clonidine (0.1 to 0.3 mg twice daily), and atenolol (25 to 100 mg once daily). The step 3 medication is hydralazine (25 to 100 mg between diuretic treatment and 3 alternative antihypertensive , twice daily). All participants were given standard advice on lifestyle pharmacological treatments. To generalize the results of factors (sodium, alcohol, physical activity, and caloric intake), with ALLHAT to a broad population of people with hypertension. reinforcement as needed during the study. Grimm ct al Baseline Characteristics of Participants in ALLHAT 21

TABLE 1. Additional inclusion and Exclusion Criteria

Additional inclusion criteria (any 1 of those listed) Old (>6 moi or age-iπdeterrniπate Ml or stroke documented by Hospital diagnosis Q waves on ECG

Akiπesis or dyskmesis on echocardiogram or veπtr,culogram Brain infarct on CT or MRI History of revascuranzation procedure including Angioplasty (coronary or peripheral vascular) Bypass surgery (coronary, peπpneral vascular, carotid, vertebrobasilar) Aortic aneurysm repair

Other revascularization (atherectomy, stent placement) Other documented atherosclerotic cardiovascular disease, including History of angina pectons

History of intermittent claudiεation gangrene or ischemic ulcers History or transient ischemic attack

Coronary, peπpheral vascular, or carotid stenosis of ≥50% documented by angiography or Doppler studies Ischemic heart disease documented by Stress thallium (reversible or fixed ischemia) Dipyndamole thallium {reversible or fixed ischemia) Exercise testing (ST depression ≥l mm for ≥1 mm) Stress echocardiogram (reversible wall motion abnormality) Hotter monitoring (ST depression ≥1 mm for ≥1 m ) Ankle-arm index <0 9

Abdominal aortic aneurysm detected by uitrasoπography CT, or x-ray Carotid or femoral bruits Major ST depression or T-wave inversion on ECG within past 2 y J-pomt αepression at 0 5 mm and rollowmg ST segment flat or o nswing in any ot leads I, II aVL or V,-V₆ T-wave inverted at least 1 mm in any of leads 1, II, aVL aVF, or j-v₆ Type 2 diabetes mellitus Fasting plasma glucose ≥7 8 mmol/L within past 2 y fJomastiπg plasma glucose ≥11 1 mmol/L within past 2 y Taking insulin or oral hypogfycemic agent Currer>t cigarette smoking

HDL cholesterol <090 mmol/L on 2 occasions within past 5 y Left ventricular hypertrophy on ECG or echocardiogram within past 2 y R amplitude m V₅ or V₆ >2B mm R amplttuαe in V₅ or V_B plus S amplitude in V, >35 mm R amplitude in aVL >12 mm R arnplrtuoe in lead l >15 mm

R amnlituoe in leads II or ill or a VF >2D mm, R amplituoε in lead I plus S amplitude in lead III >25 mm R amDlitude in aVL plus S amplitude in V₃ >28 mm τor men or >22 mm for women Computerized ECG machine documented left ventricular hypertrophy Combined wall thickness of ≥25 mm on echocardiogram Exclusion criteria Symptomatic Ml or stroke within past 6 o Symptomatic congestive heart failure Left ventricular ejection fraction <35%, if known Symptomatic angina pectons within the past 6 mo Known renal insufficiency (serum creaanine ≥180 μmol/L)

Requirement tor thiazide-like diuretics, ACE inhibitors, calcium antagonists, or α-adrenergic blockers tor reasons other than high BP Sensitivity or contraindication to any of step 1 medications

Requirement tor >2 antihypertensive drugs to achieve satisfactory BP ≤160/100 mm Hg or BP >1807110 mm Hg Low likelihood of compliance with protocol (eg, dementia, substance abuse) Diseases likely to lead to πoπcardiovascular death during course of study Current participation in another clinical trial 22 Hypertension January 2001

TABLE 2. Frequency Distribution of ALLHAT Participants by Race, Ethnicity, Gender, Age at Trial Entry, History of Diabetes at Trial Entry, and Preexisting ASCVD

Gender Age

Race Total Hispanic Male* Female* 55-59 y 60-69 y 70-79 y 80 ^J- y Diabetes ASCVD

White n 25 292 (596) 5314 14 698 10 592 4321 11 445 7790 1736 8 476 13 278

Black, π 15 094 (35 6) 1406 6 852 8 241 3298 7 008 3 817 971 6 023 5790

Asian/Pacific Islander, n 481 (1 1) 11 275 206 97 242 130 12 173 243

American Indian/ 78 (02) 5 57 21 21 37 20 0 32 39

Alaskan native, n

Other, n 1 503 (35) 1364 698 805 352 723 376 52 593 555

Total, n 42 448 8100 22 580 19 865 8089 19 455 12133 2771 15 297 19 905

Total, % 1000 19 1 53 2 46 8 19 1 45 8 28 6 6 5 36 0 46 9

Values in parentheses are percentages

"Cell may not sum to total because of missing gender data on 2 white participants and 1 black participant

At baseline, participants had blood diawn for the measurement of Participants classifying themselves as "other" were the third serum potassium, fastmg glucose creat ine total cholesterol, HDL largest group, wilb 1503 (or 3- 5%) of the total randomized, cholesterol, tπglyceπdes, and ala ne amiπotransferase LDL chomost of whom also descπbed themselves as being ofHispamc lesterol was estimated by the Fnedewald formula When possible, participants who bad consumed food or beverages within tne past 8 oπgm There were 8100 (19 1%) Ilisnamc participants, of hours were askεd to return later for a fastmg blood draw An ECG these, 65 6% designated tbemseh es as "λvhue Hispanic ', was performed if there was no existing ECG ith the past year All 17 4%, as ' black Hispanic^"' 0 2%, as either "Asian Hispanic^"' ECGs weie then tentrally read with the use of Minnesota Code oi "'American Indian Hispanic", and 16 8%, as ' other Hiscriteria At baseline the clinical center study coordmatoi was instructed to complete a questionnaire listing the inclusion criteria panic "' Men constituted 53 2% of the ALLHAT participants (Table 1) checking al) conditions that were known and documcnred The largest age category for participants was aged 60 to 69 to apply to the participant In addition, the questionnane included years (⁴5 8%) The second laigest age subgioup w as aged 70 items about race, ethnicity gender years of education current to 79 years (28 6%), and there were 19 1% aged 55 lo 59 estiogen use cur-ent regular aspiπn use cigarette smok g (past or years and 6 5% ageα ≤S0 yeais At baseline, a large propoi- current), and the presence of CHD The presence of CHD was defined as a history of MI (including silent MI), primary cardiac tion of ALLHAT participants had diabetes melhtus (36 0%) arrest coronary revasculaπzation angma angiographically defined and/or eλ idence of ASCVD (46 9%) coionary stenosis > 50% or reversible coronary pcrfusion defect on Table 3 shows baseline characteristics by randomized nomnvasne cardiac tesnng Height and weight were measured at treatment gioup By design, the chlorthalidone group was the baseline largest with 15 268 participants (36 0%), and each of the 3

For the baseline data in the present study, a participant was considered to have diabetes if type 2 diabetes melhtus was checked other drug groups badjust over 9000 participants (21 3%) In on the list of inclusion criteria (Table 3) The criteria for the this table, and all subsequent tables, the non-Hispanic white diagnosis of diabetes were the criteria of the American Diabetes category excludes the 5314 (12 5%) participants who deAssociation at the inception of the smdy and these were not changed scπbed themselves as "white Hispamcs '^" The participants when the American Diabetes Association lowered the fastmg glucose have been combined with the "other" category The black criterion m 1997^2ή A participant was considered to have atherosclerotic disease (ASCVD) according to the following category includes both Hispanic and non-Htspamc blacks defϊniUoii if CHD as defined above was listed as present on the The characteristics of Hispanic ALLHAT participants will be baseline questionnaire or if the inclusion cπteua checklist noted the descπbed in more detail m a separate publication Three presence of old or age-mdetennmate WI or stroke, history of significant, but small, differences ( <0 05) in the randomrevascularization procedure, or othei ASCVD (see Table 1 for ized treatment groups were noted There was a small differdefinitions)

Follow-up procedures, smdy εnα points and ascertainment of ence in serum potassium between the hsmopπl and chlorthaevents have been descπbed previously ^{2 25} The primary smdy end lidone groups, which is likely due to the drawing of fastmg point is the combined incidence of fatal CHD and nonfatal MI The blood after randomization in some participants Compaied study sample size was calculated to have 80% power to detect a 16% with the chlorthalidone group, the participants randomized to difference in the primary end point between the dmrenc and each of the other 3 dreg groups, after accounting for treatment crossovers, amlodipine were slightly less likely to have a history of CHD losses to follow-up, and multiple comparisons and had a slightly lower creatmine level

Baseline characteristics by geuder and race are provided in

Results Table 4 The mean age of the participants λvas 67 years, and

Table 2 provides the frequency distribution of ALLHAT both male and female white participants were older than the randomized participants by race, ethnicity, gender, age at black participants This difference resulted pπmaπly from a entry, history of diabetes, and preexisting ASCVD. Of the larger proportion of blacks compared with whites in the 55 to -42 448 participants randomized in ALLHAT. 25 292 (59 6%) 59 age category (29 1% versus 15 8%, respectively) and a were white, and 15 09⁴ (35 6%) were black Asians and larger proportion of whites compared with blacks the 70 to American Indians made up 1 1% and 0.2%, respectively 79 age category (32 0% versus 25.9%, respectively). A Grimm et al Baseline Characteristics of Participants ALLHAT

TABLE 3. Baseline Characteristics fay Antihypertensive Treatment Group

Antihypertensive Agent

Baseline Charactenstics Chlorthalidone Amiodipiπe Lisinopril Doxazosin All

Samole size 15 268 9053 906O 9067 42 448

Female, % 470 47 3 463 46 ^ 468

Race %

White ran Hispanic 47 2 47 6 470 46 4 471

Black 252 35 5 355 36 3 35 6

Other¹ 176 169 175 17.2 174

On BP treatment, % 902 90 3 902 90 2 90 2

Smoking status %

Current 21 9 21 9 21 9 21 21 8

Past 40 5 40 O 403 401 403

History of diabetes, % 36 3 368 355 355 360

History of CHD, % 260 24 5Φ 252 262 257

Aspmπ use % 360 365 364 365 363

LVH on ECG,t % 5 2 52 5 i 5 0 5 2

Age, y 669+77 66 9=77 66 9±77 66 8=77 669+77

Education,! y 11 0+40 11 0+39 11 0±4 1 11 0+4 0 11 0+4 0

Visit 1 BP

SBP mm Hg 144 8±13 8 144 8=14 0 145 0+14 1 144 8⁺139 144 8=14 0

DBP mm dg 83 3±99 83 2=10 0 83 4+9 9 83 5=9 7 834i9 9

Pulse bpm 737±107 736± ι07 73 5+108 735 + 107 736=107

BMI gmι^? 297+6 1 29 8=6 1 29 8+6 1 297=5 9 298+6 1

Potassium t mmol/L 4 3i±0 69 4 35=0 70 437=0 72$ 4 36 = 070 4 35=070

Fasting glucose | mmol/L 685+324 683=3 16 582=311 6 80=312 683+3 17

Creatinine, μ.mol L 77 83+2365 77 06=22 13§ 77 83+2289 77 83=21 36 77 83±22 89

Total cholesterol mmol/L 560+1 13 5 61 = 1 14 5 58=1 10 5 57=1 10 559=1 12 cLDL cholesterol t mmol/L 3 52=.0 97 3 51 =0 97 3 52+0 95 3 51 =094 3 52=0 96

HDL cholesterol mmol/L 1 21i038 1 22+0 38 1 21 + 0 38 1 21=037 1 21=0 38

Fasting triglycerides, mmol/L 1 95+1 48 1 95=1 53 1 95-1 58 1 92=1 53 1 95⁺1 52

Values are mean=SD, unless indicated otherwise LVH indicates left ventricular hypertrophy, BMI body mass inαex, and cLDL, calctilateo LDL

"Other races include 5314 noπblack Hispamcs, 78 American Indians/Alaskan natives, and 481 Asians/Pacific Islanders Black HisDanics were included in black category

+ECGs were available for only 38 955 participants, education was provided by 39538 participants, serum ootassium, creatmine, and cholesterol levels were available for 40126 participants, lasting glucose was available lor 31 255 participants, cLDL cholesterol was available for 3749B paπicipants, HDL cholesterol was measured m 40099 participants, tπglyceπdes were measured m 31 304 participants

%P<001, §P<005 (for each characteristic, 2-propoπιon or 2 sample mean difference tests were perrormed comparing each of the 3 πonoiuretic groups with chlorthalidone, respectively) smaller proportion of female participants compared with male males were more hkεly to be current smokers, and blades participants was white (38 5% versus 54 6%, respectively), general w ere more likely to have a history of diabetes, had a and a larger proportion was black (^l 5% versus 30 4%, higher resting pulse and had a higher glucose level Black respectively) women had higher body mass index, and blacks overall,

Most of the participants (90 2%) were receiving antihypercompared with whites were much more likely to have central tensive treatment at baseline Compared with other groups, laboratory readings of left ventricular hypertrophy on ECG blacks who bad received arrπhypertensive treatment for ≥2 (8 6% versus 3 3%, respectively) Whites were more hkely months had slightly higher baseline DBP and also had a lower than blacks to nse aspirin (48% versus 25%, respectively) and BP control rate (SBP <140 and DBP <90 mmHg) at estrogen (284% versus 11 4% respectively, among women) baseline compared with whites (26 9% versus 29 8%. respecFurthermore, whites were almost twice as likely as blacks to tively). The best BP control was observed m white males have a history of CHD (33% versus 17%, respectively) (30 9%) BP control levels weie 28 1% in white females Baseline lipoprotein levels diffeied by race and gender 27 6% in black males, and 26 3% in black females Black subgroup Blad s had higher total cholesterol, LDL choles- 24 Hypertension Januarj 2001

TABLE 4 Baseline Characteristics of ALLHAT Participants Stratified by Gender and Race

Al! Male Female -"

Characteristic All Whπe (NH) Black All White (NH)* Black* All White (NH)* Black^*

Sample size n (%) 42448±100019978±471 15094=356 22580±532 12337±546 6852 ±303 19865+468 7639+385 8241 ±415

Age y 669±77 675=75 663+78 667+73 671=70 661=74 671±82 680=81 664±82

55-59 % 191 158 218 180 153 208 203 167 228

60-69 % 458 459 464 479 483 483 435 422 449

70-79 % 286 319 253 292 320 259 279 317 248

≥80 % 65 63 64 49 44 51 84 95 76

Education, y 110+40 123=32 101±38 115+40 126=34 101=40 104=39 119±28 101=36

Entry treatment status

On medications -≥Z 868 872 869 860 869 854 878 876 881 mo %

SBP, mm Hg 143±13 143=13 143⁺13 142+13 142±13 142=13 143+13 143±13 ¹43=13

DBP mmHg 82 + 10 81 ±9 83+10 82+9 81 ±9 84±9 -82+10 80+10 83=10

On medications <2 34 28 40 35 28 44 32 27 37 mo, %

SBP, mm Hg 158±13 158=13 160⁺13 158±13 159=13 160±13 158+14 158+13 160=13

DBP mmHg 89±10 88=9 91=10 90+10 88=9 93±10 89+10 87±10 90±10

Uπt-eated % 98 100 91 104 102 102 91 97 82

SBP mmHg 159+12 158=11 159+13 158±12 158=11 158=13 160±12 159±12 160=13

DBP m Hg 91+9 89=9 92r9 91 = 9 89±9 93=9 90+9 89+9 92±9

SBP/DBP 274 298 269 286 309 276 261 281 263 %

Cigarette smoker % 218 210 251 240 208 317 193 215 197

Eysmαker % 403 474 345 525 580 447 264 302 260

Never smoked % 379 316 404 235 2¹3 236 543 483 543

Historv of diabetes 360 315 399 340 312 361 384 321 431

%

History of CHD, % 256 331 174 311 387 207 193 243 147

Treatment with 363 480 253 435 541 302 282 382 213 aspirin %

Treatment vit 180 284 114 estrogen, %

ECG LVH, % 52 33 86 50 31 90 54 36 82

Pulse, bom 736±107 729=110 747=105 725±111 718=112 738=109 748±101 745±104 754=101

Weight kg 831+181 852=181 845=182 878 + 167 902+164 876±173 777+180 772=179 818±186

BMI kg/m² 298+61 296±58 305=67 291+52 293=50 289±54 304+69 300±68 316=71

Serum aπalytes

Potassium mrπol/L 44+07 44=06 43=08 44+07 44=06 43=07 43+07 43±06 42+08

Fasting 683±317 658=277 704=353 682±311 660±272 678±331 695+336 656+285 729±371 glucose mmol'L

Creat ine, μmol/L 7783=2289 7706=1984 8240 ±2594 8240±2289 8317=1908 9156+2671 6943=19846791 ±1755 7325=2136

Total 559±112 558=110 563=116 540±105 535=101 537+108 586+115 594=114 585=118 cholesterol, mmol/L cLDL 352=096 348=091 360=103 341+090 338=087 345±096 365±100 364+096 371 = 107 cholesterol mmol/L

HDL 121 =.038 113=035 135+041 112+033 103±029 123=037 134±039 129±039 143=041 cholesterol, mmolL

Fasting 195=: 152 220=170 146+102 198+153 216=166 149=107 194=152 228=175 150±103 triglycerides, mmol/L

Values are mean=SD, unless indicated otherwise NH indicates non-Hispanic Black Hispanics were included in black category *Cell may not sum to toral because of missing gender data on 2 white participants and 1 black participant Grimm et al Baseline Characteristics of Participants m ALLHAT 25

TABLE 5 Entrance Criteria by Gender and Race

All Male Female -

Characteristic All White (NH) Black All* White (NH;^* Black^* All* White (NH)^* Black^*

Sa pie size n 42448 19978 15 094 22 580 12 337 6 852 19 865 7 639 8 241

History of Ml'stroke % 23 1 27 8 196 27 8 31 23 9 17 9 21 5 16 2

Revasculanzation procedure % 129 20 8 54 178 259 72 74 126 40

Other ASCVD % 24 0 27 5 197 232 267 18 3 24 9 28 7 21 0

Ischemic ST T ave changes % 10 4 92 131 98 87 12 9 11 1 10 1 13 2

Type 2 diabetes mellrtus % 36 0 31 5 399 34 0 31 2 361 38 4 32 1 43 1

Current smoking, % 19 9 19 1 22 3 21 4 183 27 5 18 2 20 2 18 0

HDL cholesterol <35 mg/dL % 11 7 164 6 6 152 201 8 5 77 10 5 4 9

LVH by echocardiogram % 4 1 41 4 1 38 35 3 7 4 5 52 4 4

LVH on ECG, % 165 98 2 2 168 98 27 6 16 2 9 8 21 3

Black Hispanics were included m black category

'Cell may not sum to total because of missing gender data on 2 white participants and 1 black participant terol, HDL cholesterol, and markedly lower tπglyceπde years, 625 centers in the United States, Canada Puerto Rico, levels than did whites Women had higher total cholesterol and the US Viigm Islands enrolled ^Λ2 4^8 high-nsk patients LDL cholesterol, and HDL cholesterol, whereas men had High nsk was determined on the basis of the presence of higher tnglycendes hypertension, age ≥55 years and at least I additional

Table 5 lists the proportion oi participants with each of the cardiovascular πsk factor πsk factor criteria that quabfϊed participants for entrv m the Patients in ALLHAT were randomly assigned to 1 of ^Λ tπal Compared with black participants white participants treatment groups chlorthalidone, amlodipine hsmopπl or were more likely to be entered into the study on the basis of doxazosin The treatment groups were balanced at baseline, history of MI and/or stroke, revascularization procedures or with no clinically important differences any of the recorded other ASCVD Blacks more often qualified for the studv for vanables The studv by design included a high percentage of ischemic ECG changes, diabetes, smoking, and left ventπc African Amencans (35 6%) and nearlv equal proportions bv ulεir hypertrophy on ECG Men were more hkelγ to be gender (46 8% w omen) It also included a large cohort of entered into the studv because of previous MI, stroke, or Hispanic participants (1 1%) At entry, 35 3 % of the patients revasculanzation whereas women were more likely to be were aged >70 vears, 36% had diabetes 46 9% had a history entered on the basis oi a history of diabetes Low HDL of ASCVD

half of whom had CUD) and 22% were cholesterol defined as <35 mg dL was a frequent inclusion current smokers Because of the large number of patients criterion foi white men (20 2%) but much less so m blacks em oiled in ALLHAT, it will be possible to examine the effect and white women of the ALLHAT treatments m these subgroups Basehne characteristics ior high-risk subgroups are given Blacks and lhspanics have been underrepresented most Table 6 People with diabetes melhtus made up 15 294 previous tnais In hypertension tπals that measured cardio(36 0%) of the ALLHAT participants As expected, diabetic vascular events blad s have represented a significant subparticipants were more overweight compared with other group only trials with diuretic-based active therapy arms groups, w ith a mean body mass index of 31 1 /m' Blacks are known to have more frequent and more severe Diabetics also had higher fastmg glucose (9 51 mmol/Ll hypertension that develops at an earlier age ' This finding is higher tnglycendes (191 1 mg dL), and lower HDL cholesreflected in the ALLHAT group at baseline the black terol (44 9 mg/dL) compaied with other subgroups There parπcipants are younger and have higher DBP Blacks also was a lower proportion of diabetes among participants with suffer more cardiovascular complications and have higher baseline cai diovascular disease and vice versa In some cases, πsk for end-stage renal disease ⁶+⁷ At baseline m the present after eligibility was established the documentation of addistudy, compared with whites, blacks were less likely to be tional entry criteria may have been incomplete The group enrolled on the basis of a history of previous MI or stroke, aged >70 years had somewhat higher baseline svstohc and revasculanzation procedures or ASCVD, and blacks were lower diastolic pressures compared with the otαer high-risk more hkεly to be enrolled on the basis of diabetes smoking, groups and this group was more likely lo have left ventricular or left ventricular hypertrophy on ECG ALLHAT will hypertrophy In both the >70-vear-old and diabetic groups provide an excellent opportunity to study and evaluate the women made up balf of the group Approximately 60% of the effect the type of therapy and BP control on cardiovascular ASCVD and currently smoking subgroups were men events m blacks

ALLHAT also may be able to address some questions

Discussion raised from other tπals In CAPPP patients with hypertension

ALLHAT is the largest randomized double-blind trial ever were randomized an open-label fashion to either convenconducted panents with hypertension Over a period of 4 tional therapy (αiurehcs or β-blockers) or captopnl ¹⁰ Al- 26. Hypertension January 2001

TABLE 6. Characteristics of Cardiovascular Risk Factor Subgroups

Risk Factor Subgroups

Baseline Characteristics Diabetes ASCVD Smokers ≥70 y All

Sample size, n (%) 15 297+36.0 19 905+46.9 9272=21.8 14 904+35.1 42 448

Female, % 49.9 42.4 41.5 49.1 46.8

Race, %

White, non-Hispanic 41.2 55.5 45.3 51.3 47.1

Black 39.4 29.1 40.9 32.1 35.6

Other^* 19.4 15.4 13.8 16.6 17.4

History of diabetes, % 100.0 24.5 20.0 34.1 36.0

History of CHD, % ^• 18.8 54.4 17.4 30.1 25.6

ASCVD, % 31.9+0.4 100.0 33.4 ±0.5 56.7+0.4 6.9±0.2

Current smoking, % 12.1 15.6 100.0 13.2 21.8

LVH on ECG.t % 8.9 11.4 13.0 17.9 16.5

Age, y 66.5=7.4 68.2 ±7.8 . 64.1 + 6.8 75.5+4.7 66.9=7.7

Visit 1 BP

SBP, mm Hg 144.7+13.5 144.4±14.0 145.2+14.6 145.9±13.7 144.8±14.0

DBP, mm Hg 82.1 +9.9 82.5±10.0 84.7+10.0 80.9±10.0 83.4=9.9

BMI, kg/m² 31.1 +6.2 29.2+5.8 28.3+5.9 28.3+5.4 29.8±6.1

Potassium,! mmol/L 4.4+0.7 4.4⁺0.7 4.4+0.7 4.4 ±0.7 4.4+0.7

Fasting glucose.! mmol/L 9.39+3.87 6.33+2.73 6.12+2.68 6.53 ±2.81 6.83⁺3.17

Creatinine, μmot/L 7S.3+24.42 80.12+22.13 77.83⁺22.89 81.64±23.65 77.83+22.89

Jotal cholesterol, mmol/L 5.57+1.18 5.59+1.12 5.66+1.12 5.53±1.09 5.59+1.12 cLDL cholesterol,! mmol/L 3.47+0.98 3.52+0.95 3.49±0.96 3.48+0.94 3.52=0.96

HDL cholesterol,! mmol/L 1.16=0.35 1.2+ 0.37 1.23 ±0.40 1.24 ±0.39 1.21 ±0.38

Fasting triglycerides, mmol/L 2.16+1.85 1.94=1 ,44 1.90⁺1.58 1.79+1.22 1.95=1.52

Values are meaπ±SD, unless indicated otherwise.

"Other races include 5314 nonblack Hispanics, 78 American Indians/Alaskan natives, and 481 Asians/Pacific Islanders. Black Hispanics were included in black category.

!ECGs were available for only 38 955 participants; serum potassium, creatinine, and cholesterol levels were available for 40 126 participants; fasting glucose was available for 31 255 participants: cLDL cholesterol was available for 37498 participants; HDL cholesterol was measured in 40099 participants; and triglycerides were measured in 31 304 participants.

though there was no difference in cardiovascular mortality or HOPE study design leaves many questions unansweted. fatal and nonfatal Ml between the 2 groups, fatal and nonfatal Thus, ALLHAT still remains uniquely positioned to provide strokes were more frequent in the group treated with captoan answer to the primary question: are newer antihypertenpril. This finding was partially attributed to the fact that the sive agents superior, similar, or inferior to traditional therapy captopril group had higher baseline SBP and baseline DBP, with diuretics? which remained slightly higher than values in the convenALLHAT is an ongoing study examining the highest tional therapy group throughout the study. No such baseline priority hypertension treatment question at the turn of the BP differences exist in ALLHAT. century. The results of ALLHAT will significantly add to our

The recently completed Heart Outcomes Prevention Evalunderstanding of the management of hypertension and will uation (HOPE) study compared the effect of the ACE also contribute to the formulation of future management inhibitor ramipril with placebo in older high-risk patients guidelines. with preserved left ventricular function.²³ That study demonstrated that compared with placebo, ramipril significantly- Acknowledgment reduced the rates of death, MI, stroke, revascularization This study was supported by a contract with the National Heart, Lung, and Blood Institute. procedures, heart failure, and other cardiovascular complications. Less than half the patients, in HOPE had hypertension. References In hypertensive paπicipants, study medication was added to 1. Bun V , Whelton P, Roccella EJ. Brown C, Cutler JA. Higgins M, Horari their existing hypertensive therapy, resulting in lower SBP MJ, Labarthe D. Prevalence of hypertension in the US adult population. and DBP in rarnipril-treated patients. Although this small Hypertension. 1995;25:305-313.

2. Collins R, Peto , MacMahon S, Hebert P. Fiebach N, Eberlein K, amount of BP lowering (2 to 3 mm Hg) would be expected to Godwin J, Qϊzilbash N, Taylor i. Hennekens C. Blood pressure, stroke. account for well under half of the benefit, of ramipril, the and coronary heart disease; 2: short-term reductions in blood pressure: Grimm et al Baseline Charactci istics of Participants

oven lew of ranoomised drug tπals m 'heir epidemiological context amlodrome on moroidny and.mor'ali'V m severe chronic heart failure Lancet 1 90,~35 S27-S3S V Engl J V,ed 1996335 1 107- 11 1 J SHEP Coopcra ε Research Group Pre> ention of stroke bv anfthyper- I" < derman MH, Coiien H, Roque R Maαhaven S Eπect of long-acting tensnc drug treatment in older persons -with isolated svstolic hy perand short-acimg calcium antagonists on cardiovascular outcomes in tension JAMA 1991 265 3255-3264 Hypertensive patients Lance! 1997,349 594-598 Dahlof B. Lmdhoim LH, Hansson L, Schersten B, Ekbom T, Wester ? 18 Siaessen J Fagard R, Lutgarde T Cehs H Arabidze G, Birkenhager W, Morbidity and mortality in trie Swedisn Tπal in Old Patients with HyperBulpitt C Randomised douDie-bhnd companson of placebo and active tension fSTOP-Hypeπension) Lant-tt 1991.33R 1281-1285 treatment for older patient? v, ιth isolated systolic hypertension lcncet \1RC Working Parry Medical Research Council tnal of 'reitment of 1997,350 "*5~-764 hvperteniion in older adults principal lesults BW 1992 304 405-412 19 Tuomilehlo J, Rastenvte D Birkenhager WH, Thys L Antikamen R Joint ^"National Commit'ee on Preventi n Detection. Evaluation and B-αlp CI Flet her AE, Forene F, Goldhaber A Palahm P, Sam C, Treatment o High Blood Pressure The sixth report of ihe Jomt National Fagard R Effects oi calcium ch-mnei blockade in older patients with

Rationale and Design for the Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT)

Barry R. Davis, Jeffrey A. Cutler, David }. Gordon, Curt D. Furberg, Jackson T. Wright, Jr., William C. Cushman, Richard H. Grimm, John LaRosa, Paul K. Whelton, H. Mitchell Perry, Michael H. Alderman, Charles E. Ford, Suzanne Oparil, Charles Francis, Michael Proschan, Sara Pressel, Henry R. Black, and C. Morton Hawkins, for the ALLHAT Research Group

Are newer types of antihypertensive agents, hypercholesterolemic patients (a subset of the which are currently more costly to purchase on 40,000) with a 3-hydroxymethylglutaryl coenzyme average, as good or better than diuretics in A (HMG CoA) ϊeductase inhibitor, pravastatin, reducing coronary heart disease incidence and will reduce all-cause mortality compared to a progression? Will lowering LDL cholesterol in control group receiving "usual care." moderately hypercholesterolemic older ALLHAT's main eligibility criteria are: 1) age 55 individuals reduce the incidence of cardiovascular or older; 2) systolic or diastolic hypertension; and disease and total mortality? 3) one or more additional risk factors for heart,

These important medical practice and public attack ( eg, evidence of atherosclerotic disease or health questions are to be addressed by the type II diabetes). For the lipid-lowering trial, Antihypertensive and Lipid Lowering Treatment participants must have an LDL cholesterol of 120 to Prevent Heart Attack Trial (ALLHAT), a to 189 mg/dL (100 to 129 mg/dL for those with randomized, double-blind trial in 40,000 high-risk known CHD) and a txiglyceride level below 350 hypertensive patients. ALLHAT is designed to mg/dL. The mean duration of treatment and determine whether the combined incidence of follow-up is planned to be 6 years. Further fatal coronary heart disease (CHD) and nonfatal features of the rationale, design, objectives, myocardial infarction differs between persons treatment program, and study organization of randomized to diuretic (chlorthalidone) treatment ALLHAT are described in this article. Am and each of three alternative treatments' — a J Hypertens 1996;9:342-360 calcium antagonist (amlodipine), an angiotensin converting enzyme inhibitor (lisinopril), and an oi-adrεnergic blocker (doxazosin) . ALLHAT also KEY ORDS: Hypertension, hypercholesterolemia, contains a randomized, open-label, lipid-lowering pharmacologic therapy, clinical trial, ALLHAT trial designed to determine whether lowering LDL trial, chlorthalidone, amlodipine, doxazosin, cholesterol in 20,000 moderately lisinopril, economics.

Received January 9, 1995. Accepted September 28, 1995. School of -Medicine, Winston-Salem, North Carolina (CDF); Divi¬

From the Coordinating- Center for Clinical Trials, University of sion of Hypertension, Case Western University School of Medicine, Texas School of Public Health, Houston, Texas (BRD, CEF, SP, Cleveland, Ohio (JTW); Memphis Veterans' Administration. MediCMH); Division of Epidemiology and Clinical Applications, Nacal Center, Memphis, Tennessee (WCC); Department of Internal tional Heart, Lung, and Blood Institute, Bethesda, Maryland (JAC, Medicine, University of Minnesota Medical School, Minneapolis, DJG, MP); Department- of. Public Health Sciences, Bowman Gray Minnesota ( RHG) ; f ulane Medical Center, New Orleans, Louisiana AJB-ΛPRIL 1995-VOL. 9, NO. 4. PART 1 A TIHyPERTENSlVE AND LIPID-LOWERING CLINICAL TRIAL 343-

million people in the US demonstrate the expected degree of CHD reduction- blood pressure (systolic is that adverse effects of study drugs, particularly di[SBP] ≥ 140 mm Hg or uretics, may have offset the potential benefit of blood

pressure [DBP] ≥ 90 mm pressure reduction. These adverse effects include diHg) or are taking antihypertensive medication.^5,2 Hyuretic-induced hypokalemia, hypomagnesemia, hyp- pertension is considerably more common among eruricemia, hyperhpidemia, hyperglycemia, impaired blacks than among whites, and its sequelae are more insulin sensitivity, and probably increased ventricular frequent and severe in the former. Suggested explanaectopic activity.¹'^10,11 However, these side effects are tions for these increased rates of complications among minimal at currently recommended doses. blacks ha e included a higher prevalence of coexisting In the late 1970s and in the 1980s, new types of illnesses, such as diabetes mellitus, and less effective antihypertensive agents — calcium antagonists, angio- treatment and control due in part to decreased access tensin converting enzyme (ACE) inhibitors, and α- to medical care. The variation in. cost of treating hyadrenergic blockers — ere developed and approved pertension is, in large part, determined by the cost of for use in chronic antihypertensive therapy. These the antihypertensive agents used.³ Given the number agents are currently more costly to purchase on averof patients treated (23 million in 1988-91), drug age. However, evidence that might justify their use in choice has substantial economic implications.^2,3 All preference to the older classes of drugs is limited. other factors remaining constant, the incremental Only two large long-term randomized trials have yearly cost of treating 25 million patients with a drug compared representatives of all of these drug classes: costing $100 per patient compared to one costing $500 the 1-year trial conducted by the Department of Veterper patient is $10 billion. ans' Affairs Cooperative Study Group on Antihyper¬

Despite the known etiologic relationship of hypertensive Agents,¹² and the 4.4-year Treatment of Mild tension to coronary heart disease (CHD), results of Hypertension Study (TOMHS) .¹³ While these trials ^' large-scale randomized clinical trials in mild to modhave reported some differences in blood pressure conerate hypertension (DBP 90 to 114 mm Hg) in largely trol, side effects, quality of life, biochemical effects, middle-aged subjects have generally failed to demonand target-organ changes, these differences did not strate that antihypertensive drug treatment reduces present a pattern that consistently favored one class the rate of CHD death or nonfatal myocardial inof drugs over others. Also, these trials did not have farction.⁴ Overviews of all hypertension trials have clinical endpoints as the primary outcome for comparshown that antihypertensive treatment does lead to a isons of drug classes. reduction in CHD event rates.³ However, the reducOther relevant data come both from animal experition is. less than expected based upon epidemiological ments and clinical trials in patients with heart disease. data." Also, the cited overviews did not take into ac-_. Calcium channel blockers inhibit development of athcount the strongly positive results of the recent Sys- erosclerotic lesions in rabbit models, but chnical trial toIic^'Hypertension in the El erly Program (SHEP) , in data on morbidity and mortality are conflicting. In a which diuretic-based treatment reduced major CHD trial of dilriazem in post-myocardial infarction (MI) events by 27% (95% confidence interval, 4 to 43%)/ patients, post-hoc analyses suggested a detrimental efOther trials in older persons with diastolic /systolic fect in patients with a low ejection fraction, but benefit hypertension reported similar results.^8,9 One possible in patients without a low ejection fraction. An ^'overexplanation given for the failure of previous trials to view of all post-MI trials with calcium channel blockers reported a 6% (95% confidence interval, -4% to +18%) increase in mortality. ⁴ An update of this overview that

(JL); Welch Center for Prevention, Epidemiology and Clinical Research, Johns Hopkins Medical Institutions, Baltimore, Maryland included three additional trials in patients with angina (PKW); St. Louis Veterans' Administration Medical Center, St. pectoris or myocardial infarction suggested unfavorLouis, Missouri (HMP); Department of Epidemiology and Social able results, particularly with dihydropyridine calcium Medicine, Albert Einstein College of Medicine, Bronx, New York (MHA); Division of Medicine, University of Alabama at Birchannel blockers.¹⁵ The increased mortality with the mingham, Birmingham, Alabama (SO); Columbia University, New- short-acting formulations of ni εdipine and nicardipine • York, New York (CF); and Rush-Presbyterian-St. Luke's Medical occurred primarily in patients with a recent MI. This Center, Chicago, Illinois (HRB) .

A list of participating centers and investigators appears at the outcome might be different with a long-acting dihydro- end of this article. pyridine, such as amlodipine.

■This research was supported by Health and Human Services con-^' Angiotensin converting enzyme (ACE) inhibitors tract number (NOX-HC-35130) from the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of reduce mortality in both severe and less severe heart Health and Human Services, Bethesda, Maryland. See additional failure, ^16-1S and reduce morbidity, including CHD, in statement in the acknowledgements. asymptomatic left ventricular dysfunction.'⁹ Improve¬

Address correspondence and reprint requests to Baπv R. Davis/ MD, PhD, University of Texas School of Public Health.1200 Herman ments in insulin resistance have been reported with Pressler Street, Houston TX 77030. ACE inhibitors; an observation that may be especially 344 DAVIS ET AL AJH-ΛPRIL 1995-VOL. 9, MO. 4, PART 1

relevant to patients with type II diabetes mellitus.^20,21 terol Reduction in Seniors Program (CRISP), demonFurthermore, Chobanian and colleagues have restrated the feasibility of recruitment and retention of ported prevention of coronary lesions in the Watathe elderly in a placebo-controlled trial of a reductase nabe rabbit model with captopril treatment,²² perhaps inhibitor.⁴² due to effects on cellular proliferation in the vessel The one exception noted above is the Scandinavian wall. Antiatherosclerotic effects of ACE inhibitors Simvastatin Survival Study (4S) .⁴¹ This study ranhave not been demonstrated in humans. domized 4444 men and women from six Scandinavian

The α-blockers have been shown to have modercountries with documented CHD and cholesterol levately favorable effects on lipid profile, particularly on els between 212 and 309 mg / dL to treatment with HDL cholesterol, LDL cholesterol, and the LDL /HDL simvastatin or placebo. The primary endpoint was toratio.^13,23 Improvements in insulin resistance also have tal mortality with a median follow-up of 5.4 years. been reported with α-blockers.²⁴ There is some eviThe results of the trial showed the following: 1 ) a dence that these agents may reduce platelet aggrega- reduction of 35% in the mean LDL cholesterol in the bility and stimulate tissue plasminogen activator.^25-27 simvastatin group; 2) 30% fewer deaths in the simvasThese data from existing studies in humans and tatin group compared to the placebo group; 3) a 42% animal models do not permit a determination as to decrease in CHD mortality without offsetting trends whether newer drugs are superior, equivalent, or infein other causes of death; 4) nonfatal CHD endpoints rior to older drugs in the treatment of hypertension similarly and significantly reduced; 5) all-cause morand the prevention of its cardiovascular complicatality reduced in older (60 to 70 years) and younger tions. Given the clinical and public health importance patients; and 6) CHD rate reductions for both men of this issue, results of large-scale comparative trials and women. are urgently needed to assess the role of newer versus The 4S trial results were not yet reported when the older antihypertensive agents in cardiovascular dispresent study was designed. When the results were ease prevention. announced, the Steering Committee considered the

Experimental evidence for the efficacy of cholesquestion of whether the cholesterol trial should conterol-lowering m reducing the incidence of CHD has tinue. The decision to continue was based on two reabeen derived almost entirely from studies in white, sons. First, the 4S trial and Antihypertensive and middle-aged men, and is lacking for women, minoriLipid Lowering Treatment to Prevent Heart Attack ties, and the elderly. Observational epidemiological Trial (ALLHAT) deal with very different study popuevidence further suggests that the relationship of cholations. Indeed, only 3% of the patients randomized lesterol levels and CHD is less strong at older ages to the ALLHAT cholesterol trial as of November 1994 (although the attributable risk associated with choleswould have satisfied the 4S trial entry criteria of prior terol remains high in the elderly because of their high CHD, age below 70 years, and total cholesterol greater absolute event rates),²⁸ and is less compelling for than 212 mg / dL. Second, since the ALLHAT control women and minorities than for men.²⁹ Despite the group is assigned to receive "usual care" rather than reductions in CHD in metaanalyses of these trials, a placebo, study physicians are not prevented from reductions in CHD mortality have been offset by inprescribing cholesterol-lowering drugs to any patient creases in other causes of death.^30-40 While the net who is thought to need them. The only change made change in total mortality tended to be favorable in was to exclude patients with CHD and LDL above high risk populations,^34,40 such as men with prior 129 mg/dL, the patients thought most likely to be myocardial infarction,^30,32 it was not favorable in most advised to take cholesterol-lowering drugs. Preprimary prevention trials.³¹ However, the cholesterol- viously, this upper limit had been 159 mg/dL. Allowering trials published to date, with one excepthough some physicians may also decide to prescribe tion," have not been designed with sufficient statistithese drugs for other ALLHAT "usual care-" patients, cal power to address the impact on total mortality the Steering Committee believed that the benefit of despite the reductions in CHD incidence. Also, the cholesterol lowering in primary prevention and in interpretation of these analyses has been clouded by secondary prevention at LDL levels below 130 g/ the limited duration of treatment or degree of cholesdL was still uncertain, even after the 4S trial, and that terol-lowering in most of these trials, as well as the most usual care patients in these categories would not possible toxicity of some of the older cholesterol-lowreceive such drugs. ering drugs.^30,35"40 The advent of 3-hy roxymethyI- glutaryl coenzyme A (HMG CoA) reductase inhibitors, which can lower LDL cholesterol levels by 25% OBJECTIVES AND DESIGN or more with few apparent side effects, has facilitated the development of larger more powerful trials to adALLHAT, sponsored by the National, Heart, Lung, dress this issue. A recent 1-year pilot study, Cholesand Blood Institute (NHLBI) in conjunction with the AJH-APR1L 1995-VOL. 9, NO. 4. PART 1 ANTIHYPERTENSIVE AND LIPID-LOWERING CLINICAL TRIAL 345-

Department of Veterans' Affairs, is a practice-based, LDL cholesterol levels between 120 and 189 mg/ dL randomized, clinical trial in 40,000 high-risk hyper(between 1O0 and 129 mg/dL for those with known tensive patients 55 years and older, of whom about CHD) who are randomized to receive pravastatin plus 45% will be women and at least 55% will be black. a cholesterol-lowering diet (National Cholesterol EduALLHAT has two components. The antihypertensive cation Program [NCEP] Step I diet⁴⁴) than in those component is a randomized, double-blind trial derandomized to receive diet plus usual care. The ratiosigned to determine whether the combined incidence nale for the sample size is presented in the Appendix. of fatal coronary heart disease (CHD ) and nonfatal Secondary hypotheses for this component are listed in myocardial infarction differs between diuretic (chlorTable 2. thalidone) treatment and three alternative antihyperOne of the main reasons for choosing all-cause mortensive pharmacologic treatments — a calcium antagotality as the primary endpoint was that this trial is unnist (amlodipine), an ACE inhibitor (lisinopril), and blinded. Further, the assessment of myocardial inan α-adrenergic blocker (doxazosin). The lipid-lowfarction for this component will rely on the routine cenering component is a randomized, open-label trial detrally coded, electrocardiogram (ECG), rather than the signed to determine whether lowering serum cholespotentially biased assessment of the study physicians. terol in 20,000 moderately hypercholesterolemic men Although other secondary endpoints will be examined, and women aged 55 years and older (a subset of the these will be regarded as "soft data" that will at best 40,000 from the antihypertensive trial) with an HMG confirm or supplement the primary endpoint. CoA reductase inhibitor (pravastatin) will reduce all- While a blinded study would certainly have been cause mortality as compared to a control group receivpreferable in many ways, other factors did not make it ing "usual care." feasible within the overall context of ALLHAT. Compli¬

Hypotheses and Study Power The primary hypotheance and cross-over rates will be monitored as the trial ses of the antihypertensive trial component are that the progresses, and the study will not be continued if the combined incidence of fatal CHD and nonfatal myocaractual data indicate inadequate power. dial infarction (first or recurrent) will be lower in hypertensive patients randomized to 1 ) a calcium antagonist (amlodipine), 2) an ACE inhibitor (lisinopril), or ENROLLMENT AND FOLLOW-UP 3) an α-adrenergic blocker (doxazosin) as first-line PROCEDURES therapy than in those randomized to a thiazide-like diuretic (chlorthalidone) as first-line therapy. Thus the Recruitment and Baseline Visits Recruitment for statistical design must account for three primary comALLHAT will rely on a variety of methods, particularly parisons. chart review within the participating chnical site to

To maximize statistical power for the antihypertenidentify patients who are potentially eligible for the sive trial, 1.7 times as many patients will be assigned trial components. The visit schedule and procedures for to its diuretic arm as to each of its other three arms ALLHAT participants are delineated in Table 3. Data (Table I ).⁴³ The rationale for the sample size is preneeded to make the definitive determination of eligibilsented in the Appendix. One of the assumptions is that ity for the antihypertensive trial component will be obhalf of the ALLHAT participants will be randomized tained in two prerandomization visits, which will gento both trial components and that half will be randomerally take place !■ day to 2 months apart. The objective ized to the antihypertensive trial component only. Secof Visit 1 is to assess ehgibiUty for and interest in ondary hypotheses for this component are listed in ALLHAT and to begin withdrawing patients from β~ Table 2. blockers and central α-agonists if needed. It is antici¬

The primary hypothesis of the cholesterol-lowering pated that many treated hypertensive patients will have trial component is that mortality from all causes will been identified by chart review, and that much of the be lower in the subset of hypertensive patients with pertinent information (eg, age, risk factor status, and

TABLE 1. ANTICIPATED SAMPLE SIZE OF ALLHAT TREATMENT GROUPS

Antihypertensive Trial (4 A ms)

Cholesterol-Lowering Trial (2 Anns) Chlorthalidone Amlodipine Lisinopril Doxazosin Total

Pravastatin 3,655 2,115 2,115 2,115 10,000 Usual Care 3,655 2,115 2,115 2,115 10,000 Not Eligible 7,310 4,230 4,230 4,230 20,000 Total 14,620 8,460 8,460 8,460 40,000 346 DAVIS ET AL AΓH-APKΠ ms-voL. 9, NO. 4, PART I

TABLE 2. SECONDARY HYPOTHESES FOR THE ALLHAT TRIAL COMPONENTS

Antihypertensive Trial — The following endpoints (or their incidence) will be reduced in patients randomized to receive amlodipine, lisinopril, or doxazosin relative to those receiving chlorthalidone:

1. All-cause mortality

2. Combined coronary heart disease (CHD or revascularization procedures or hospitalized angina)

3. Stroke

4. Combined cardiovascular disease (CHD or stroke or coronary revascularization procedures or angina [hospitalized or medically treated] or CHF [hospitalized or medically treated] or peripheral arterial disease [hospitalized or outpatient revascularization procedure]),

5. Left ventricular hypertrophy by ECG

6. Renal disease a. Slope and reciprocal of serum creatinine b. End-stage rena] disease (initiation of chronic renal dialysis or kidney transplant)

7. Health-related quality of life

8. Major costs of medical care

Lipid-Lowering Trial — The following endpoints (or their incidence) will be reduced in patients randomized to receive pravastatin relative to those receiving usual care:

1. The combined incidence of CHD death and nonfatal myocardial infarction, especially in certain subgroups, eg, blacks, patients over age 65 (the original Cholesterol Reduction in Seniors Program [CRISP] hypothesis³⁸), type II diabetics, and women

2. Changes in the biennial study ECG indicative of myocardial infarction

3. Cause-specific mortality (eg, cancer, trauma)

4. Total and site-specific cancer incidence

5. Health-related quality of life

6. Major costs of medical care

CHD, coronary hrørf disease; CHF, co gesfroe heart failure; ECG, electrocardiogram.

number of antihypertensive drugs) will already be can enter the study at Visit 2. This visit will generally known. Additional interim visits may be needed for take place between 1 day and 8 weeks after Visit 1, ^* patients on 3-blockers or central α-agonists in order to depending on the length of time required to step down step down the medication. Only patients who have from prestudy medications or determine hypertension been randomized to the antihypertensive trial compostatus. Patients initially taking no drugs or well-connent will be considered for randomization to the cholestrolled on one drug may be randomized soon after Visit terol-lowering trial component, and randomization to 1, while other patients may require a longer step down the latter will not take place before the first postran- process (generally less than 2 months) before they can domization visit for the antihypertensive trial, usually complete Visit 2. More prolonged step downs are dis4 weeks later. couraged (though not prohibited), since many patients

Blood pressure eligibility criteria for the antihyperwho cannot quickly be withdrawn^'from their prestudy tensive trial, listed in Table 4, are based on the patient's regimens may also be more difficult to maintain on a current treatment status and on the average of two simple regimen during the trial. All randomized paseated blood pressure measurements at each of two tients will be given appropriate hygienic advice (sovisits. For untreated patients, the criteria used are the dium and alcohol reduction, exercise, caloric restriction current JNC V definitions of diastolic and systolic hyif overweight) with reinforcement as needed during the pertension (stage TU).¹ For treated patients, the criteria trial. are a reasonable degree of blood pressure control, ie, Patients who have satisfied all Visit 1 eligibility re≤ 160 mm Hg systolic and ≤ 100 mm Hg diastolic at quirements for the antihypertensive trial component or visit 1, and ≤ 180 mm Hg systolic and ≤ 110 mm Hg have consented to begin a step down fro prestudy diastolic at visit 2 (when medication may have been ^■antihypertensive drugs will also be informed of the chopartially withdrawn) . Additional inclusion and exclulesterol-lowering trial component of ALLHAT. Those sion criteria for the antihypertensive and lipid-lowering who indicate interest and have not been treated with trials are presented in Table 5. lipid-lowering drugs during the 2 months preceding Visit 1 are considered as potential candidates for this

Randomization Patients who meet the ALLHAT elitrial. gibility criteria, can safely discontinue prior antihyperA fasting- lipid battery (total cholesterol, triglycerides, tensive drugs and be randomized to one of the four HDL cholesterol, calculated LDL cholesterol⁴⁴ ) and serum ALLHAT treatment arms, and give informed consent alarime- hcmsarninase (ALT, formerly SGPT) will be ob- AfH-APRJL 1995-VOL 9, NO i. PART 1 ANTTHYPERTENSrVH AND IPID-LOWERING CLINICAL TRIAL 347-

TABLE 3. ALLHAT PATIENT VISIT SCHEDULE

Purpose

Visit # Months From Visit 2 Antihypertensive Trial Cholesterol-Lowering Trial

-6.0 to 1 day Identify potential participant

I -2.0 to 1 day Assess eligibility and interest.

Λ a, b, c As needed Step down from prestudy antihypertensive drugs if on β- blockers or central α- agoπists Randorriization, Fasting LP Profile*, ALT laboratory diet/ lifestyle counselling Routine data collection, Randomization, fasting LP profile, dosage titration if NCEP, step 1 diet neededt Routine data collection, Dosage titration if needed, ALT, TCt dosage titration if neededt

5, 6, ^: 6, 9, 12 (more often if Routine data collection, Routine data collection, dosage nee ed)t dosage titration if titration if neededt neededt

8, 9, 10, . . Every 4 months Routine data collectiont Routine data collectiont

* Total cholesterol, tngh/ceride, HDL, and LDL cholesterol levels. LDL calculated fey the Fπedewnld formula t Pastnmdomizanon msxis.

ALT, Alamne awmotrmφrase; NCEP, Nnhorwl Cftoiesierol Education Program", TC, Total cholesterol.

tained at Visit 2. Patients who indicated interest and who visits; full lipid profiles will be done in a randomly chosen had an LDL cholesterol between 120 and 189 mg/dL 5% cohort. (between 100 and 129 mg/ dL for patients with known CHD) and fasting triglycerides ≤ 350 mg/dL at this visit .will be informed by telephone of their eligibility for the TREATMENT PROGRAM cholesterol-lowering trial component and told to come in fasting for Visit 3. If they are eligible to partϊtipate in Antihypertensive Intervention The blood pressure this ALLHAT component at Visit 3, the investigator will goal in all four arms is <90 mm Hg diastolic and <140 phone the Qrnical Trials Center and receive a random mm Hg systolic. The therapeutic goal is to achieve assignment for the patient to either pravastatin or usual blood pressure control on the lowest possible dosage care. (Patients also can be randomized into this trial at of the first-line drug. The number and dose of study Visit 4.) Each patient randomized to receive pravastatin drugs prescribed in pursuit of these goals will be inwill be issued an appropriate supply of 20 mg tablets and fluenced by patient tolerance and clinical judgment, instructed to take two each evening. Patients assigned to particularly in use of greater than two-drug regimens. usual care as well as those assigned to pravastatin will The dosage levels available for each drug are listed in be advised to follow the NCEP Step I diet (<30% of Table 6. calories from fat, <10% of calories from saturated fat, The identity of the drug will be masked at each <300 mg cholesterol/day). Study physiάans retain the dosage level. The initial dosage level will be used only option to reduce the dosage in patients who cannot tolerduring the first week after randomization to minimize ate the full dose. Lipid profiles will be performed on all potential first dose hypotension with doxazosin. For ALLHAT patients at Visit 2 and on aU patients in the the other three drugs, the initial dose and Step 1 doscholesterol trial at Visit 3. In the pravastatin group, cholesages are identical. Also, in order to allow three dose terol levels will also be measured at Visit 4 and all annual levels for the other agents with maintenance of the isits; full lipid profiles will be done in a randomly chosen blind, doses 1 and 2 of chlorthalidone are both 12.5 10% cohort In the usual care group, cholesterol levels mg. Patients will typically return at 1-month intervals will measured at the second, fourth,, and sixth annual for any necessary increase in dosa-ge until both the 348 DAVIS ET AL ΛJH-ΛPRJL 1995-VOL. S, NO. 4, PART 1

TABLE 4. ALLHAT BLOOD PRESSURE ELIGIBILITY CRΠΈRIA

Lower Limit* Upper Limitt (mm Hg) (mm Hg)

Status at Visit 1 and Visit 2 SBP DBP SBP DBP

Taking 1-2 drugs for hypertension for at least 2 months 1601; lOOi, 180§ πo§ Taking drugs for < 2 months or currently untreated 140 90 180 110

* SBP or DBP Joiner limit must be met at Visit j and Visή 2. t SBP and DBP upper limit must be met at visit 1 and Visit 2. $ Visit 1 only. % Visit 2 only.

systolic and diastolic goal pressures are reached. If randomization into the antihypertensive component of the initial dose of the blinded drug is not tolerated, it ALLHAT. will be discontinued. A rechallenge with the medication can be attempted later, but patients will be treated with open-label drugs as needed to provide DETERMINATION OF OUTCOMES adequate blood pressure control and will continue to

Endpoint Ascertainment Occurrences of study end- be followed in the study. points will be documented by a checklist completed

Tor patients in any of the four treatment arms who at each follow-up visit and supplemented by interi are unable to attain satisfactory blood pressure control reporting as needed. These diagnoses will be supported on the maximum tolerable dosage of their first-line by ^'copies of death certificates and hospital discharge drug, a choice of second- and third-line drugs are prosummaries. The outcomes that will be obtained and vided in open-label form for use in addition to (not tabulated over the course of the study are listed in Table substitution for) the first-line drug (Table 6), unless 7. The underlying cause of death will be classified by the first-line drug is not tolerated. The choice of secthe hysidan-investigator at the clmical site. A National ond-line drug(s) is at the discretion of the treating Death Index (NDI) search will be performed near the study investigator. Since the study investigators are end of the study to identify and document deaths that blinded to the identity of the first-line drug to which may have occurred among patients who are lost to foleach patient is assigned, it is likely that the frequency low-up. Because of the time lag inherent in the NDI, a of use of each of the second-line drugs will be similar private tracing service will also be used for selected among the four treatment arms. Although in special cases, investigators may choose to prescribe second- participants.

The study investigators will be required to complete line antihypertensivedrugs other than those provided and submit to the Clinical Trials Center a short end- by the study, thiazide diuretics, calcium antagonists, points questionnaire for each occurrence of a study end- ACE inhibitors, and α-adrenergic blockers are point identified at or between regular visits. For each avoided unless maximum tolerated doses of a three- endpoint involving a death or hospitalization, the indrug regimen have been tried and are unsuccessful vestigator will also obtain and submit a copy of the in controlling blood pressure. death certificate or hospital discharge summary upon which the diagnosis was based. For a^' random (10%)

Cholesterol-Lowering Intervention The cholessubset of hospitalized (fatal and nonfatal) myocardial terol-lowering component of ALLHAT will employ a infarctions and strokes, the Clinical Trials Center will randomized comparison of an HMG CoA reductase inrequest more detailed information. For this subset, in- hibitor (pravastatin) plus diet versus usual care plus hospital ECGs and enzyme levels (for myocardial indiet in a subset of patients participating in the antihyfarctions), and neurologists' reports and computed topertensive component of the study. The dosage of pramography (CT) or magnetic resonance imaging (MET) vastatin will be 40 mg, taken in the evening. All particireports (for strokes) will be evaluated by the study . pants in this ALLHAT component will receive instrucEndpoints Committee and the-accuracy of the discharge tion in the Step 1 diet recommended by the National diagnoses assessed. Cholesterol Education Program⁴³ upon randomization into the study. Randomization into this trial component Data Analyses The primary endpoint of the antihywill take place at least 4 weeks and up to 90 days after pertensive component of. ALLHAT is combined fatal ΛJH-APRIL 1995-VOL. 9, NO. 4, PART I ANTIHYFERTENSIVE AND LTPID-LOWERING CLINICAL TRIAL 349-

TABLE 5. MAJOR ALLHAT INCLUSION AND EXCLUSION CRITERIA

-Antihypertensive Trial

1. Inclusion a) One or more manifestations of atherosclerotic cardiovascular disease: I) old (>6 months) or age-indeterrninate myocardial infarction or stroke; 2) history of revascularization procedure; or 3) documented atherosclerotic cardiovascular disease h) Type II diabetes melhtus [plasma glucose >I 0 mg/dL (fasting) or 200 mg/dL (norvfastjng) or on insulin or oral hypoglycemics] c) HDL cholesterol <35 mg/dL (at ≥2 determinations within past 5 years) d) Left ventricular hypertrophy on electrocardiogram or echocardiogram e) ST-T wave electrocardiogram changes indicative of ischemia f) Current cigarette smoking

2. Exclusion a) Symptomatic myocardial infarction or stroke within the past 6 months b) Symptomatic congestive heart failure or ejection fraction <35%, if known c) Angina pectoris within the past 6 months d) Serum creatinine ≥ mg/dL e) Requirement for thiazide-like diuretics, calcium antagonists, angiotensin converting enzyme inhibitors, or a-blockers for reasons other than hypertension f) Requirement for more than two antihypertensive drugs to achieve satisfactory blood pressure control g) Sensitivity or contraindications to any of the first-line study medications h) Factors suggesting a low likelihood of compliance with the protocol, eg, plans to move or travel extensively i) Diseases likely to lead to noncardiovascular death over the course of the study j) Blood pressure > 180 mm Hg systolic or >1_Q mm Hg diastolic on two separate readings during screening or step-down

Lipid-Lowεring Trial

1. Inclusion a) Enrollment in the antihypertensive tπal b) An LDL cholesterol of 120-189 mg/dL (100-129 mg/dL for patients with known congestive heart disease) with a friglyceride level ≤350 mg/dL

2. Exdusion a) Current use of prescribed lipid-lowering agents or large doses (≥5D0 mg/day) of nonprescription niacin b) Contraindications to hepatic hydroxymethyl glutaryl coenzyme A (HMG CoA) reductase inhibitors (eg, significant liver disease, ongoing immunosuppressive therapy, known allergy or intolerance to the study drug) c) Known untreated secondary cause of hyperlipidemia (eg, hypothyroidism, nephrotic syndrome) d) Alanine aminotransferase (ALT) >2.0 X upper limit of normal

CHD and nonfatal MI. The primary response variable The primary endpoint of the lipid-lowering compois time from randomization to development of this nent of ALLHAT is all-cause mortality. The primary event. The log-rank test⁴⁶ will be used to compare response variable is time from randomization to each of the nondiuretic treatment groups to the dideath. The log-rank test will be used to compare the uretic group. For the secondary endpoints of all-cause group assigned to pravastatin plus diet to the group mortality, stroke, combined coronary and cardiovasassigned to usual care plus diet. For the secondary cular outcomes, and end-stage renal disease, the log- endpoints of combined fatal CHD and nonfatal MI, rank test will also be used. For the outcomes of left fatal and nonfatal cancer, and cause-specific mortalventricular hypertrophy (LVH) by ECG and health- ity, the log-rank test will also be used. In addition, related quality of life, comparison of proportions will the log-rank test will be used to compare treatments be used to see if there are differences in the treatment within each of the following subgroups for the outgroups. For the outcome of renal disease, the reciprocome of combined fatal CHD and nonfatal MI: men, cal of a participant's creatinine values at basehne, 3 women, ≥ 65 years, < 65 years, blacks, nonblacks, months, and years 2, 4, and 6 will be obtained. Using diabetics, and nondiabetics. For the outcomes of MI treatment group as a fixed effect and time as a ranby ECG and health-related quality of life, comparison dom effect, a treatment by time interaction effect will of proportions will be used to see if there are differbe estimated using the longitudinal models of Laird ences in the treatment groups. and Ware.⁴⁷ Interim monitoring will focus on patient intake 350 ^' DAVIS ET AL A H-APKTL 1995-VOL 9, NO. 4, PART 1

TABLE 6. ALLHAT FIRST- (BLTNDED), SECOND-, AND THIRD-LINE (OPEN LABEL) ANTIHYPERTENSIVE DRUGS*

Initial

Doset Dose If Dose 2t Dose 3t

Step 1 Agent

Chlorthalidone 12.5 12.5 12.5 25

Amlodipine 2.5 2.5 5 10

Lisinopril 10 10 20 40

Doxazosin 1 2 4 8

Step 2 and Step 3 Agents

Reserpine 0.05 daily or 0 1 daily 0.2 daily 0.1 every 2 days

Clonidine (oral) 0.1 twice daily 0.2 twice daily 0.3 twice daily

Atenolol 25 daily 50 daily 100 daily

Hydralazine (third-line) 25 twice daily 50 twice daily 100 twice daily

* Sources of the four first-line agents are: chlortimlidone: Ogden Bioservkes, Inc., Rockvillε, Maryland; amlodipine: Pfizer, Inc., New York, N w York; lisinopril: Zeneca Pharmaceuticals Group, Wilmington, Delaware; and doxazosin: Pfizer, Inc., Nm York, Nno York. t All doses in milligrams.

overall and within each clinical center; center adherfollow-up; adverse effects data; and effects of treatence to protocol; baseline comparability of treatment ment on the primary and secondary study outcomes. groups; sample size assumptions with regard to event Interim analyses will coincide with the meetings of rates, cross-over rates, competing risk, and loss to the Data and Safety Monitoring Board ( DSMB) . Stochastic curtailment will be used for monitoring treatment differences in both the antihypertensive and the

TABLE 7. ALLHAT OUTCOMES lipid-lowering studies.⁴⁸'*⁹

1. Death

ORGANIZATIONAL STRUCTURE a. Definite myocardial infarction b. Definite coronary heart disease ALLHAT has an organizational structure that differs c. Possible coronaiy heart disease markedly from the usual NHLBI-supported clinical d. Stroke e. Congestive heart failure trial. This so-called "large, simple trial" model, implef. Other cardiovascular disease mented previously in the International Study of Infarct g. Cancer Survival (ISIS) trials coordinated by Oxford University h. Accident, suicide, or homicide investigators⁵⁰ and first used by NHLBI in the Digitalis i. Other noncardiovaεcular cause Investigative Group trial/¹ is appropriate when the folj. Unknown cause lowing conditions apply: D a very large sample size

2. Myocardial infarction is needed, 2) a streamlined protocol is possible, 3) the

3. Stroke targeted conditions are commonly encountered in clin¬

4. Angina (hospitalized or treated) ical practice, and 4) there is widespread interest in the

5. Congestive heart failure (hospitalized or treated) study question among clinicians.

6. Peripheral arterial disease (hospitalized or treated)

The trial will be performed by a large number (400 ^' 7. New cancer diagnosis (hospitalized or treated)

8. Accident oτ attempted suicide (hospitalized or to 500) of practicing physician-investigators who will treated) be compensated on a per capita basis for each patient

9. Left ventricular hypertrophy φiennial study, seen according to a fixed payment schedule. Approxielectrocardiogram) mately 15% to 20% of study patients are expected to be

10. Renal function recruited by Department of Veterans' Affairs (DVA) a. Slope of the reciprocal of serum creatinine level hypertension clinics. The Clinical Trials Center, in adversus time dition to its conventional data handling and monitor' b. End-stage renal disease (initiation of chronic renal ing responsibilities, will be responsible for identifying dialysis or kidney transplant) and paying these physician-investigators, for en¬

11. Quality of life listing regional physician and nurse study coordina1Z Medical care use tors to monitor recruitment and compliance, and for AJH-APR1L 1995-VOl. 9, NO- i, PART 1 ANTIHYPERTENSIVE AND LIPID-LOWERING CLINICAL TRIAL 351

awarding and supervising subcontracts for a drug disommended by the Steering Committee during the tribution center, a central laboratory, and an ECG codcourse of the study. ing center. A Steering Committee of experts in the At any time during the study, the DSMB may recrelevant subject areas has also been appointed bv ommend discontinuation of any of the treatment arms NHLBI. of either study component on any of the following

Practitioners will be reimbursed a fixed fee for each grounds: patient randomized- to each component of the trial

1 ) Compelling evidence from this or another trial of and for each subsequent study visit completed. This a significant adverse effect of the study treatment (s) fee is expected to cover the costs of the data collection that is sufficient to override any potential benefit re(step down and titration visits, questionnaires, blood garding CHD and preclude its further use in the target drawing, ECG recording) specified above. The fee population; does not include the cost of required laboratory work

2) Compelling evidence from this or another trial of and ECG coding, which will be performed by central a significant beneficial effect of a study treatment, facilities and paid for directly by the Clinical Trials such that its continued denial to the other study Center, or the costs of documenting study endpoints, groups is ethically untenable; or for which there will be separate reimbursement. 3) A very low probability of successfully addressing

The Clinical Trials Center will have overall responthe study hypotheses within a feasible time frame, besibility for training and quality control. All clinical cause of inadequate recruitment, compliance, drug resites will be required to attend a training session. The sponse, event rate, or other key performance criteria. training session will include orientation to the study protocol, blood pressure measurement training and The Director of the NHLBI will make the final decicertification, orientation to ECG and laboratory procesion on whether or not to accept the DSMB's recomdures, and training in recruitment and retention of mendation to discontinue any component of the participants, as well as completion and transfer of study. study forms. Periodic refresher training will be held in conjunction with regularly scheduled Study Investigators' meetings. These refresher sessions may inCONCLUSIONS clude a review of correct blood pressure measurement procedures or any problem that may be identified Hypertension is a frequent health problem in Amerithrough review of routine monitoring activities. cans, especially among older individuals and blacks.

The Clinical Trials Center's responsibilities with reIt is associated with a significantly increased risk of gard to quality control include: 1 ) reviewing ail forms morbidity and mortality. Only diuretics and 3-block- for completeness and accuracy prior to data entry; 2) ers have been shown to reduce this risk in long-term resolving problems by telephone or facsimile transclinical trials among hypertensive individuals. mission with clinical sites; 3) providing double data Whether newer more costly antihypertensive agents entry of forms; 4 ) cross-forms editing to identify missconfer increased benefit or not in terms of reduced ing forms and procedures; 5) monitoring the perforincidence of cardiovascular disease is unknown. mance of study components and providin ^' timely Also unknown is the potential benefit of treating summary reports to the Program Office and to the moderately hypercholesterolemic older men and women with an HMG CoA-reductase inhibitor in Steering Committee; and 6) providing detailed and terms of reduced incidence of not only coronary heart up-to-date statistical reports of study progress to the disease but also total mortality. The results of DSMB at their meetings. A.LLHAT are expected to be available by the year

The DSMB will be responsible for monitoring all 2002, and should help resolve these issues of major aspects of the study, including those that require acimportance to medical practice and public health. cess to blinded data. The DSMB and its chair were appointed by the Director of NHLBI; they are experts who are not otherwise affiliated with the study. During the active recruitment phase, the DSMB will moniACKNOWLEDGEMENTS tor the progress of recruitment (particularly of black This study is supported by contract with the National, patients) and the random allocation of participants Heart, Lung, and Blood Institute. The ALLHAT investigato the various treatment arms and may recommend tors acknowledge contributions of study medications supmodifications in (or termination of) one or both study plied by Pfizer, Inc. (amlodipine and doxasozin), Zeneca (atenolol and lisinopril), and Bristol-Myers Squibb components if the study design goals are not being (pravastatin), and the financial support provided by rriet. The approval of the DSMB will also be required Pfizer to NHLBI. The authors would li e to acknowledge for any other significant changes in the protocol recthe special contribution to the planning and Vanguard 352 DAVIS ET AL AJH-APRJL 1995-VOL. 9. NO. 4, PART 1

phases of ALLHAT of Dr. Ten^" Manolio, former NHLBI 15. Yusuf S, Held P, Furberg C: Update of effects of calproject officer. cium antagonists in myocardial infarction or angina in light of the second Danish Verapami] Infarction Trial (DA TT-II) and other recent studies. Am ] Cardiol

REFERENCES 1991;67:1295-1297.

^•1. The Fifth Report of the Joint National Committee on 16. The SOLVD Investigators: Effect of enalapril on surDetection, Evaluation, and Treatment of High Blood vival in patients with reduced left ventricular ejection Pressure (JNC V). Arch Intern Med 1993;153:154-183. fractions and congestive heart failure. N Engl J Med 1991;325:293-302.

2. Burt VL, Whelton P, Rocella EJ, et al: Prevalence of hypertension in the U.S. adult population: results from 17. Pfeffer MA, Braunwald E, Moye LA, et al: The effect the Third National Health and Nutrition Examination of captopril on morbidity and mortality in patients survey, 1988-1991. Hypertension 1995;25:305-313. with left ventricular dysfunction following myocardial infarction. N Engl } Med 327:678-684, 1992.

3. ^• Manolio TA, Cutler JA, Furberg CD, et al: Trends in the pharmacologic management of hypertension in the 18. The Acute Infarction Ramipril Efficacy (A1RE ) Study United States. Arch Intern Med 1995';15:829-837. Investigators: Effects of ramipril on mortality and morbidity of survivors of acute myocardial infarction with

4. MacMahon SW, Cutler JA, Furberg CD, et al: The efchnical evidence of heart failure. Lancet 1 93;342:821- fects of drug treatment for hypertension on morbidity 828. and mortality from cardiovascular disease: a review of randomized controlled trials. Prog Cardiovasc Dis 19. The SOLVD Investigators: Effect of enalapril on mortal1986;29(3, suppl 1):99-119. ity and the development of heart failure in asymptomatic patients with reduced left ventricular ejection frac¬

5. Collins R, Peto R, MacMahon S, et al: Blood piesstire, tions. N Engl J Med 1992;327:685-691. stroke, and coronary heart disease. Part 2. Short-term reductions in blood pressure: overview of randomized 20. Tollare T, Lithell H, Berne C: A comparison of the efdrug trials in their epidemiological context. Lancet fects of hydrochlorothiazide and captopril on glucose 1990;335:827-838. and lipid metabolism in patients with hypertension. N Engl J Med 1989;321:868-873.

6. MacMahon S, Peto K, Cutler ], et al: Blood pressure, stroke, and coronary heart disease. Part 1. Prolonged 21. Paolisso G, Gamhardella A, Verza M, et al: ACE inhibidifferences in blood pressure: prospective, observation improves insulin-sensitivity in aged insulin-resistional studies corrected for the regression dilution bias. tant hypertensive patients. J Hum Hypertens 1992; Lancet 1990;335:765-774. 6:175-179.

7. SHEP Cooperative Research Group: Prevention of 22. Chobanian AV, Haudenschild CC, Nickerson C, et al: stroke by antihypertensive drug treatment in older perAnti-atherogenic effect of captopril in the Watanabe sons with isolated systolic hypertension: final results heritable hyperlipidemic rabbit. Hypertension 1990; of the Systolic Hypertension in the Elderly Program 15:327-331. (SHEP). JAMA 1991;265:3255-3264. 23. Pool JL: Effects of doxazosin on serum lipids: A review

8. Dahlof B, Lindhohn LH, Hansson L, et al: Morbidity of the chnical data and molecular basis for altered lipid and mortality in the Swedish Trial in Old Patients with metabolism. Am Heart J 1991; 121:251 -259. Hypertension (STOP-Hypertension). Lancet 1991; 24. Lithell H: Effects of antihypertensive drugs on insulin, 338:1281-1285. glucose and lipid metabolism. Diabetes Care 1991;

9. MRC Working Party: Medical Research Council of 14:203-209. treatment of hypertension in older adults: prin pal re25. Hernandez RH, Carvajal AR, Guerrero PJ, et al: Evisults. Br Med J 1992;304:405-412. dence of an antiplatelet aggregation in normotensive

10. Cohen JD, Neaton TD, Prineas RJ, Daniels KA, for the subjects and patients with hypertension: an in-vitro MRFIT Research Group: Diuretics, serum potassium, study. Am Heart J 1991;121:389-394. and ventricular arrhythmias in the Multiple Risk Factor 26. Hernandez RH, Guerrero PJ, Carvajal AR, et al: EviIntervention Trial. Am J Cardiol 1987;60:548-554. dence of an antiplatelet aggregation action of doxa¬

11. Ferrari P,- Rosman J, Weid ari P: Antihypertensive zosin in patients with hypertension: an ex-vivo study. agents, serum lipoproteins, and glucose metabolism. Am Heart ] 1991;121:395-401. Am J Cardiol 1991;67:26B-35B. 27. Jansson JH: Effects of doxazosin and atenolol on the

12.. Materson BJ, Reda DJ, Cushman WC, et al, for the DVA fibrinolytic system in patients with hypertension and Cooperative Study Group on Antihypertensive Agents: elevated serum cholesterol. Eur J Clin' Pharmacol Single-drug therapy for hypertension in men: a com1991;40:321 -326. parison of six antihypertensive agents with placebo. N 28. Gordon DJ, Rifkind BM: 3-Hydroxy-3-methyglutaryl Eng] J Med 1993;328:914-921. coenzyme A ^'(HMG CoA) Teductase inhibitors: a new

13. Neaton JD, Gήmm RH Jr, Prineas RJ, et al, for the Treatclass of cholesterol-lowering agents. Ann Intern Med ment of Mild Hypertension Study Research Group: 1987;107:759-761. Treatment of Mild Hypertension Study: final results. 29. Jacobs D, Blackburn H, Higgins M, et al: Report of the . JAMA 1993;270:713-724. conference on low blood cholesterol: mortality associa¬

14. Held PH, Yusuf S, Furberg .CD: Calcium channel blocktions. Circulation 1 92;86:1046-1060. ers in acute myocardial infarction and unstable angina: 30. Holme I: An analysis of randomized trials evaluating an overview. Br Med J 1989;299:1187-1192. the effect of cholesterol reduction on total mortality AJH-APRJL 1995-VOL 9, NO. 4, PART I ANTIHYPERTENSIVE AND LIPID-LOWERING CLINICAL TRIAL 353

and coronary heart disease incidence. Circulation rank order statistics arising in its consideration. Cancer 1990;82:1916- 1924. Chemother Rep 1966;50:163-170.

31. Muldoon MF, Manuck SB, Mathews KA: Lowering 47. Laird NM, Ware JH: Random effects models for longicholesterol concεntrations.and mortality: a.quantitative tudinal data. Biometrics 1982;38:963-974. review of primary prevention trials. Br Med J 1990; 48. Lan KKG, Wittes J: The B-value: a tool for monitoring 301:309-314. data. Biometrics 1988;44:579-585.

32. Roussow JE, Lewis B, Rifkind BM: The value of low49. Davis BR, Hardy RJ: Data monitoring in clinical trials: ering cholesterol after myocardial infarction. N Engl J the case for stochastic curtailment. J Clin Epidemiol Med 1990;323:1112-1119. 1994;47:1033-1042.

33. Ravnskov U: Cholesterol lowering trials in coronary- 50. Sleight P: Trials and tribulations: the ISΪS experience. heart disease: frequency of citation and outcome. Br Aust NZ J Med 1992;22(5, suppl) :583-586. Med J 1992;306:1367-1373.

51. Williford WO, Collins JF, Garg R, et al: Design and

34. Davey Smith G, Song F, Sheldon TA: Cholesterol lowimplementation. of the NHLBI and the VA Cooperative ering and mortality, the importance of considering iniStudies Program Collaborative Study trial to evaluate tial level of risk. Br Med J 1993;306:1367-1373. the effect of digitalis on mortality in heart failure. Con¬

35. Holme I: Relation of coronary heart disease incidence trolled Clin Trials 1992; 13:386. and total mortality to plasma cholesterol reduction in ^•52. Vital and Health Statistics. Serum Lipids of Adults 20- randomised trials: use of meta-analysis. Br Heart J 74 Years: United States, 1976-80, Series II: Data from 1993;69 (suppl) -.S42-S47. the National Health Survey, No. 242, DHHS Publica¬

36. Thompson SG. Controversies in meta-analysis: the case tion No. (PHS) 93-1692. US Public Health Service, Hy- of the trials of cholesterol reduction. Stat Meth Med attsville, Maryland, 1993. Res 1993;2:1 3-192.

37. Law MR, Wald NJ, Thompson SG: By how much and APPENDIX how quickly does reduction in serum cholesterol lower risk of ischemic heart disease? Br Med J 1994; 308:367- Considerations for Sample Size The statistical 373. power to test the primary hypothesis of the antihyper¬

38. Law MR, Thompson SG, Wald NJ. Assessing possible tensive trial is approximately 82.5%, based on the folhazards of reducing serum cholesterol. BMJ 2994;308: lowing assumptions; 1 ) sample size of 40,000 (approxi373-379. mately 22,000 men and 18,000 women; 2) 6 year inci¬

39. Gould AL, Roussow JE, Santanellσ NC, et al: Cholesdence of CHD events of 7.8% in the diuretic group terol reduction yields clinical ^"benefit; a new look at old (derived from the Framingham Study, the Hypertendata. Circulation (In press). sion Detection and Follow-up Program [HDFP], and

40. Gordon DJ: Cholesterol lowering and total mortality, in the Systolic Hypertension in the Elderly Program ^' Rifkind BM (ed): Contemporary Issues in Cholesterol [SHEP] [personal communication]); 3) a 16.3% reducLowering: Clmical and Population Aspects. Marcel De- kker, New York, 1994. tion in the CHD event rate after adjustment for non- compliance and losses to follow-up in each of the three

41. Scandanavian Simvastatin Survival Study Group: Randomised trial of cholesterol lowering ih 4444 patients noήdiuretic treatment arms compared to the diuretic with coronary heart disease: the Scandanavian Simvaarm; 4) rates of cross-over between each of the other statin Survival Study (4S). Lancet^"l994;344:13S3-1389. study drugs and chlorthalidone and nonstudy medica¬

42. LaRosa J, Applegate W, Crouse J, et al: Cholesterol tion of 2.75% in each of the first 3 years and 6% over lowering in the elderl — Results of the Cholesterol Rethe last 3 years of follow-up (rates derived from the duction in Seniors Program (CRISP) Pilot Study. Arch TOMHS [personal communication] ) — ielding a cu¬

^■ Intern Med 1994:154;529-539. mulative 24% rate of patients crossing over to another

43. Dunnett CW. A multiple comparisons procedure for " medication at least once in 6 years; 5) CHD status uncomparing several treatments with a control. J Am Slat determinable at the end of the study for 16.8% of pa¬

' Assoc 1955;60:573-583. ^• tients due to competing risks (non-CHD death) or loss

44. Expert Panel on Detection, Evaluation, and Treatment to follow-up; 6) a 25% reduction in CHD event rates of High Blood .Cholesterol in Adults: Summary of the (before adjustment for noncomplianc^'e and losses to second report of the Nation Cholesterol Education Profollow-up) among the 10,000 patients randomized to gram (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults the active treatment arm of the cholesterol-lowering (Adult Treatment Panel II). JAMA 1993;269:3015- trial component; and 7) a type I error of 0.05 (two- 3023. sided ), corresponding to a critical Z-score of 2.37 after

45. Friedewald WT, Levy RI, Fredrickson DS: Estimation adjustment for multiple comparisons using the of the concentration of low-density lipoprotein cholesDunnett procedure.⁴³ terol in plasma without use of the preparative ultracen- The original ALLHAT protocol used an age criterion trifuge. Clin Chem 1972;18:499-502. of 60 or greater and did not include current cigarette

46. Mantel N. Evaluation of survival data and two new smoking as a risk factor. Lowering the entry age de- 354 DAVIS ET AL AJH-APRIL 1995-VOL 9, NO 4. PART I

creased the CHD event rate, but the addition of the trial have lower LDL cholesterol levels than are typismoking risk factor resulted in the CHD event rate cally treated in ordinary practice. Many patients ϊw estimate remaining at 1.35% /year. More pessimistic the US who clearly need cholesterol-lowering drugs or optimistic assumptions were also considered. These are not being treated despite far higher LDL levels. include event rates of 1.05% /year to 1.65% /year and Given the cost of lipid-lowering agents and the relacross-over rates of 22% to 26% with loss rates of 11.8% tively modest lipid levels of the patients, not many to 21.8%. Power estimates ranged from 77% to 86% patients assigned to no medication are expected to be under these assumptions. taking active lipid-lowering medication. 4) ALLHAT

Based on National Health and Nutrition Examinaphysicians are advised not to randomize patients who tion Survey (NHANES) II⁵² data for ages 65 to 74 years, are already receiving cholesterol-lowering drugs or in which the LDL cholesterol cutpoints for ALLHAT who in their opinion should receive these drugs as patients without CHD corresponded to the 25th and part of their "usual care." Thus, potential cross-overs 86th percentile (me ) and to the 14th and 76th percen- to active treatment are for the most part not being tile (women), just over 60% of patients in the ALLHAT randomized in the first place. 5) Following the publistudy will be LDL-eligible for the cholesterol-lowering cation of the 4S trial results, the protocol was trial. It was assumed that about 80% of LDL-eligible amended to exclude patients with established CHD patients (or 50% of all ALLHAT patients) would par- and LDL cholesterol above 130 mg / dL from the cho- ticipate in the cholesterol-lowering trial. Slightly lower lesterol-lowering trial. Also, the 4S study had a drop- estimates (slightly under 60%) were later obtained in in rate of 13% and a drop-out rate of 10% over the the more recent ( 1988-91 ) NHANES III data (National course of 5.4 years. Center for Health Statistics, personal communication), In the original ALLHAT protocol with an age critereflecting a general downward temporal trend in LDL rion of 60 years or greater and not including current cholesterol levels as well as the incorporation of more cigarette smoking as a risk factor, we estimated a data from blacks and from persons aged 75 to 84 years. 2.35% / year mortality rate and an unadjusted treatThe statistical power to test the primary hypothesis ment difference of 12.5%. With the protocol modificaof the cholesterol-lowering trial is approximately tions and with the results of the 4S trial study (ad85 5%, based on the following assumptions: 1 ) sample justed 30% treatment difference), the new assumpsize of 20,000 (approximately 11,000 men and 9,000 tions were felt to be reasonable women) allocated equally between pravastatin and More pessimistic or optimistic assumptions were usual care groups; 2) 6 year total mortality of 12.2% also considered. These include 1 ) mortality rates of (2.15% /year) in the usual care group (derived from 2.15% /year to 2.50%/year; 2) drop-out rate of 17.8% Framingham, HDFP and SHEP [personal communiand drop-in rate of 12.9%; and 3) reductions in mortalcation]); 3) a 14% reduction in mortality in the praity of 12.5% to 14%. Power estimates ranged from 76 vastatin treatment arm before adjustment for dropto 90% under these assumptions. outs and drop-ins; 4) a "dropout" rate (from pravastatin treatment to no treatment) of 5% in year 1, and ALLHAT RESEARCH GROUP INVESTIGATORS 2.5% in all subsequent years, and a "drop-in" rate (from no treatment to pravastatin or a similar. drug) Investigators at the chnical centers and coordination and service centers of the ALLHAT Research Group are listed of 2% per year — -cumulative rate of 15.3% of pravabelow. statin patients off treatment and 10.6% of usual care Clmical Centers: VAMC Allen Park, Allen Park, Ml: patients on treatment at the end of 6 years; 5) no James R. Sowers, MD (PI); Martin Berman, DO; Carol E. losses to mortality follow-up; 6) a 10% reduction in Bodrie, RN; Elizabeth Jones; Kathy M. Rice. Wade Park mortality rate in each of the three nondiuretic treatVAMC, Cleveland, OH: Eleni 1. Pelecanos, MD, MPH (PI); Dorothy Herd, CNP. Memphis VAMC, Memphis, TN: Gale ment arms of the antihypertensive trial component; H. Rutan, MD, MPH (PI); Roger G. Smith, MD; Anita W. and 7) a type I error a - 0.05 (two-sided), correMcKnight, BSN, RN. Miami VAMC, Miami, PL: Barry J. sponding to a critical Z-score of 1.96. Materson, MD, MBA (PI); Gustavo Godoy, MD; Richard

The drop-in and drop-out rates were derived from Preston, MD; Mary Healy Smith, ARNP-CS, BSN, MS. several assumptions. 1 ) Based on previous experience Manhattan VAMC, New York, NY: Lois Anne Katz, MD (PI); Mary Keary, RN. Dallas County Health Department, with HMG CoA reductase inhibitors, compliance was Selma, AL: Sherri Nichols, RN (PI); Joan Stewart, RN. The expected to be quite good, with the bulk of noncom- Wellness Plan, Detroit, MI: Marc Keshishian, MD (PI); Richpliance occurring early in the trial. 2) In most cases ard Miller, MD; Marlene Vaughn, RN- RUSH Prudential the study physician is the patient's primary care phyHMO, Van B ren Office, Chicago, IL: Vance LauderdaIe,MD sician and thus, there is less concern about outside (PI); Terese Bertucci, RN, MS. RUSH Prudential HMO, Colεman Office, Chicago, IL: Howard Martin, MD (PI); physicians changing patients' medicines than in a Sandy Gibson, MD; Marie Bosley, RN; Donna Lee Patmore conventional university-based trial. 3) The paterson, RN. University of South Carolina, Columbia, SO- G- tients being considered for the cholesterol-lowering Paul Eleazer, MD (PI); Carlton A. Hornung, PhD (PI); AJH-APRIL 1995-VOL 9, NO. 4, PART 1 ANTIHYPERTENSIVE AND LIPID-LOWERTNG CLINICAL TRIAL 355

Davinder Lally, MD (PI); Mya D. Kline. University of South. H. Kaesemeyer, MD (PI); Pamela M. Boitεr, PN. Hamilton Carolina Medical School, Columbia, SC: Lisa Befhea, MD; Health Center, Harrisburg, PA: David K. Kelley, MD (PI); Sharm Steadman, PharmD; Mya D. Kline. People's Health Kixn Myers, PA. Lehigh Valley Hospital, Allentown, PA: NelCenters, SL Louis, MO: Kelly D. Gage, MD (PI); Betty Kerr, son Kopyt, DO (PI); Yehia Y. Mishrikϊ, MD; Mary Anne MA, PNP; Elaine Feehan, RN; Clara Scott. UT-HHSC/MS, Gergϊts, RN, MS, CCRC. Palomar Medical Group, Escondido, Houston . TX: Carlos Herrera, MD (PI); Tina Cormier, ReCA: Emrnet W. Lee, MD (Pϊ); Elaine A. Murphy, CRC. search Assistant. West Alabama Health Services, Inc., Eutazυ, Nøriri Hills Family Practice, North Richlanά Hills, TX: Leo L. AL: Sandral Hullett, MD (PJ); Glenn Hughes, PhD; Al tenberg, MD (PI); Carol Dvorak, RN, BSN. Primary Care Voncelia Hall, LPN. West Alabama Health Services — GreensInternal Medicine, Detroit, MF.Anita Moncrease, MD (PI); boro Center #2, Greensboro, AL: Sandral Hullett, MD (PI) ; Ann Edwards. The Jamaica Hospital, Jamaica, NY: Robert I. Glenn Hughes, PhD; Voncelia Hall, LPN. WAHS— LivingMendelson, MD (PI); Kenneth Nordlicht, MD; Robert 1. ston Health Center, Livingston, A Sandral Hullett, MD Mendelson, MD. Bowdoin Street Health Center, Boston, MA: (PI); Glenn -Hughes, PhD; Voncelia Hall, LPN. Worksite Joseph A. Ingelfinger, MD (PI); Alice Chan. Ohio State Treatment, New York, NY: Karen M. Johnston, MD (PI) ; University General Internal Medicine, Columbus, OH: Robert Karen Fuller, BSN. Sloreiυorkers' Local #3, New York, NY: A. Murden, MD, FACP (PI); Lori McCollough, BA. Mt. Kaϊen M. Johnston, MD (PI ) ; Karen Fuller, BSN. University Smai Outpatient Clinic, Hartford, CT: Ellen Olarsh Nestler, of V School of Medicine, Las Vegas, NV: Stephen Newmark, MD (PI); Nellie Medina, RN. Wake Forest University Family MD (PI); Gail Vranesh, RN. West Gastroenterology Group, Physicians, Winston- Salem, NC: Kevin A. Pearce, MD, MPH Ingϊewooά, CA: Timothy C. Simmons, MD (PI); Fred Glet- (PI); John Summerson, MS. John Peter Smith Hospital, Fort ten, MD; Donald Henderson, MD; Adebambo Ojuri, MD; Worth, TX: Stephen E. Piwinski, MD (PI); Julia Meaders, Bisrat Yirgou. Newark Community Health Center, Newark, RN. Harlem Hospital Center, New YorSc, NY: Velvie Anne NJ: Jonathan N. Tobin, PhD (PI) ; Anita Vaughn, MD (PI); Pogue, MD (PI); Donna Dowie, MD. Trinity Diagnostic AsPrabal Ray, MD; Judith Quiles, RN. Bedford-Stuyvesani sociates, Carrollton, TX: Henry A. Punzi, MD, FACTP (PI); Family Health Center, Brooklyn, NY: M. Monica Sweeney, Lola Hussey, RN, CCRN. John D. Stokes, MD, PA, Palm MD (PI); Jonathan N. Tobin, PhD (PI); Maria Jaime, RPA. Harbor, FLr, John D. Stokes, MD (PI) ; Catherine Rousseau, Lyndon B. Johnson Health Complex, Inc., Brooklyn, NY: JonaMedical Assistant. Medical Parameters, Martinez, GA: Robthan N. Tobin, PhD (PI ); Marie Therese DeCastro, MD; ert B. Rhoades, MD (PI); Shirley K. Wiley, MA. Gould MedSheila Hood, RN. Osteapathic Health Care Center. Philadelical Group, Modesto, CA: George Chao, MD (PI); Carl R. phia, PA: George D. Vermeire, DO (PI); Shirley Combs. Sufϊt, MD; Ruth Bright, RN, CCRC; Dora Darieεn Johnson, PCOM Cambria Street Healthcare Center, Philadelphia, PA: RN, CRC. North Pacific Internal Medicine Group, Portland, Patrice Taylor, DO ( PI) . Jackson-Hinds Camp Hlih Care CenOR; Frank D. McBarron, MD (PI) ; Kathy L. Root Richard ter, Jackson, MS. Evelyn R. Walker, MD (PI); Adonna fowler, MD, PC, Mesa, AZ: Richard F. Fowler, MD (PI); James, RN. UCLA — King j Drew Medical Center, Los Dianne Woods, RN. University of Illinois College of Medicine, Angeles, CA: Harry J. Ward, MD (PI) ; Benjamin Scott; Sed- Peoria, IL: James F. Graumlich, MD (PI); Steven Belknap, onia V. Prets. Mobile Medical and Diagnostic Center, Mobile, MD; Susan Cole, MD; Nancy L. Smith, RN, MS. Carver AL: Raymond L. Bell, MD (PI); Beverly K. Johnson, RN. Family Health Center, Peoria, IL: James F. Graumlich, MD Oflfcftwrsf Corrcrrαmify Health Center, Decatur, GA: Robert (PI); James Barnett, MD; Nancy L. Smith, RN, MS. The Kimbrough, MD (PI); Robert J. Anderson, PharMD; Betsy Cleveland Clinic Foundation, Cleveland, OH: Donald G. Vidt, Redmond, RD. Kaiser Permanente of Georgia, Tucker, GA: MD (PI); David Litaker, MD; Fran Yanik, Nurse CoordinaJoshua Barzilay, MD (PL) ; Jeanne Jordan, RN, BSN. Candler tor. Sin i Hospital of Baltimore, Baltimore, MD: Michael D. Medical Group, Savannah, GA: Paul S. Bradley, MD (PI) ; Freedman, MD (FIX, Ralph Weber, MD; Karyn Heagy, RN. Ray R. Maddox, PharmD. Forest Hill Family Practice, RichClinic for Digestive and Nutritional Disorders, Tallahassee, F L.^¬ mond, VA: Benjamin F. Zambrana, PhD, MD (PI); Kay Joseph Lee Webster, Sr., MD (PI); Thelma J. Caldwell, RN. Zambrana. AHEC Family Practice Center, Pine Bluff, AR: AHEC of South Arkansas Family Practice Clinic, El Dorado, Herbert F. Fendley, MD (PI); Julian Eddie Maples, Jr., RN, AR: Mark K. Buck, MD(PI) ; Renεe McCafferty, MS. ALLRRT. Cardiovascular Physicians, Ltd., Milwaukee, WI: Burton HAT Clinical Center 067 A, Danville, IL: Bisrat Hailemeskel, J. Friedman, MD (PI); Barbara Ullsperger. Albert Einstein PharmD (PI); T. Halloran, MD; R. Sadiq, MD; Lee Ding, Hospital, Bronx, NY. William H. Frishman, MD (PI) ; SuRPH. Danville Carle Clinic, Danville, LL: Bisrat Hailemeskel, zanne E. Furϊa, RN, BSN. The Medical Research Center, Inc., PharmD (PI); T. Halloran, MD; Lee Ding, RPH. Community Washington, DC:; Bruce N. Garrett, MD (PI); Patricia SulliHealthcare Clinic, Houston, TX: Admerle Hall-Hoskins, DO van, RN, BS. Outreach Health Services, Shubula, MS: James (PI) . Mercy Diagnostic and Treatment Center, Chicago, IL: C. Graham, MD ( PI) ; Cathy Duvall, LPN. University Health Charles G. Tεrzian, MD, MPH (PI); Char Piotrowski, RN. Associates, Charleston, WV: Stephen R. Grubb, MD, FACP ALLHAT Clinical Center 071 A, Hinesville, GA: R. Glenn Car(PI) ; Vickie Cruz. Oschnεr Clinic — Perkins, Baton Rouge, ter, MD (PI); LariAnn S. Thomas, MSN, RNCS. University LA: Alok K. Gupta, MD (PI); Helen Penfield, RN. Mid of Oklahoma College of ^' Medicine-Tulsa, Tulsa, OK: Martina J. Delta Family Practice Clinic, Cleveland, MS: Nate Brown, Jelley, MD, MSPH (PI); Tracy Lawson, RN. Bronx Neγhrol- MD (PI); Evelyn Peggy, FNP-C; Sarah Cooks. Wesϊview ogy Hypertension PC, Bronx, NY: Mario A. Henriquez, MD, Clinic, West St. Paul, MN: Marvin A- Heuer, MD (PI); SuFACP (PI); Ana Silvia Henriquez, MPH. Wilson Health san Hassing; Julie Uhlenkott. Cook County Hospital, ChiCenter, Rochester, NY: Frederick J. Zu ihe, MD (PI); Robert cago, IL: Arthur Hoffman, MD, MPH (PI); Judy Statsinger. L. VonBuskirk, BSN. East Albany Medical Center, Albany, Ridgεviεw Research Center, Chanhassen, MN: James E. Carra- " GA: Phillip R. Poulos, MD (PI); Ron Malcolm, PA; Madebie, MPE, MD, CCFP (PI) ; Karen M. Long, RN, MSN, line (Madge) Holm, BSN. Brookdale Hospital Medical Center, CFNP; Yancy Schafer, LPN. UCLA School of Medicine, Los Brooklyn, NY: Samuel Spitalewitz, MD (PI); Alba Correa, Angeles. CA: Ka Kit P. Hui, MD, FACP (PI); Jagoda Pasic, RN. Clεarwater Cardiovascular Consultants, Largo, FL: Jorge MD, PhD; Elaine Pang. University of Mississippi Medical P. Navas, MD (PI); Kay Livingston, RN. Baylor College- of Center, Jackson, MS: Daniel W. Jones, MD (PI); Catherine Medicine, Houston, TX: Carlos Vallbona, MD (PI>; Glori Adair, RN. Augusta Hypertension, PC, Augusta, GA: Wayne Chauca, MD; Valory Pavlik, PhD. Loma Linda Faculty Medi- 356 DAVIS ET AL AJH-APRIL 1995-VOL 9, NO 4, PART 1

cal Offices, Loma Lmda, CA Gregory Wise, MD (PI), Demse Gaurang Shah, MD, Eric Bowes VAMC Minneapolis, MinJackson, MD, Leslie Albert, RN United Hospital, Grand neapolis, MN Jordan Koltzman, MD (PI), Demse FmleVj Forks, ND Richard J Gray, MD (PI), Kelly Hagen, RN RN VAMC Murfreesboro, Murfreesboro, TN Rimda Gupta, ALLHAT Chnical Center 084 A, Pme Bluff, AR Martha Ann MD (PI), Regma Cassidy, PharmD VAMC Nashville, NashFlowers, MD (PI), Stacye D McLemore UT Southwestern ville, TN Ghodrat A Sιamι, MD, PhD (PI), alter Wilkms, Medical Center, Hypertension Division, Dallas, TX Norman RN V4MC Northγort Medical Services, Northport, NY JoKaplan, MD (PI) , Patsy Hargrave ALLHAT Chnical Center anne Holland, MD (PI), Christine SpiUer, RN VAMC 086A, Gahanna OH Albert M Salomon, DO (PI), Marjoπe Phoenix, Phoenix, AZ James V Felicetta, MD (PI), Karen A Considme Pulsifer Medical Associates PC, Rochester, NY Bowen, RN VAMC Richmond, Richmond, V Pramod Mo- David Dobrzynski, MD (PI) , Joyce Hackemer, RPA-C The hanty, MD (PI), Edie Earlev, RN VAMC Salisbury, SalisMed f Peds Managed Care Chmc, WBAMC El Paso, TX , Anbury, NC Ronald D Smith MD (PI), Valena Shipp, RN drew C Quint, MD (PI) , Elisabeth Marcus Valley Medical VAMC-3 San Francisco, San Francisco, CA Barry Massie, Group, Bakersfield, CA Carlos A Alvarez, MD (PI), Lauπe MD (PI) , Elaine Der, RNP VAMC-2 San Francisco, San Mitchell, RN South Carolina Heart Center, Columbia, SC Francisco, CA Marvm Siperstem, MD (PI), Edward Jai, William Lawrence Schoolmeester, MD (PI), Jacqueline PharmD VAMC San Juan, San Juan, PR Jose Luis Cian- Sheπod, BSN 77ιe Mt Sinai Medical Center, Cleveland, OH chini, MD (PI), Jean DaMore, RN VAMC Columbia, SC, Joseph P Frolkis, MD, PhD (PI), Robert L Hayrue, MD, Columbia, SC Alberto Saenz, MD (PI), Delia C Hart, RN, PhD, Pamela Suhan, RN Beaverton Medical Clime, Bever- BSN, Wendy Nicholson, RN, BSN VAMC North Chicago, ton, OR Scott W Falley, MD (PI) Mayo Clinic, Di sion of North Chicago, IL Sant P Singh, MD (PI), Jen-Chieh Hypertension, Rochester, MN Daniel J Wilson, MD (PI), Cheng, MD, Laura Kirkham, RN VAMC LeaOemoorth, Leav- Vincent J Canzanello, MD, Lois Kle Cardiology Foundaenwσrth, KS Donald L Courtney, MD (PI), Mane Cook. tion of Lank au, Wynnewood, PA James F Burke, MD (PI), VΑMC Shreveport, Shreveport, LA. Arthur Chausmer, MD, Heather L Horton, MD, Bettye Briggs, BSN Dekalb M.edιcal PhD (PI), Ann Leitz, RN, MSN VAMC Pittsburgh, PittsSpecialty Center, Decatur, G A Charles A Gilbert, MD (PI), burgh, PA Melissa McNeil, MD (PI), Catherine Kelley, Elaine Furka, RN, CCRN VAMC Augusta, Augusta, GA PharmD VAMC Tampa, Tampa, FL Raquel Rosen, MD Thomas J Hartney, MD (PI) , Nanci McPhail, JviD, Thomas (PI), Nancy Roϊbiecki, LPN Camden Internal Medicine AsJ Hartney, MD VAMC Baltimore, Baltimore, MD Bruce P sociates, PA Camden, SC John B DuBose, III, MD (PI), Hamilton, MD ( PI), Mary Dangleis, RN, CRNP VAMC Dale S Barwick, MSN, RN, CS Raleigh Internal Medicine Kansas City Kansas City, MO Thomas B Wiegmaπn, MD Associates, PA, Raleigh, NC Christopher M Perkins, MD (PI) VAMC Bronx, Bronx, NY Cbve Rosendorff, MD, PhD (PI), James O'Rourke, MD, Lynda E Seal, RN Morehouse (PI), Steven A A tlas, MD, PhD, Chve Rosendorff, MD, School of Medicine, Atlanta, GA Edward F Pamsh, III, MD PhD VAMC Brooklyn bl, Brooklvn, NY Donald F Kreuz, (PI), Tim Bnscoe, PharmD Morehouse Medical Associates, MD (PI ), Stacy Robinson, RN VAMC Buffalo, Buffalo, NY Atlanta, G A Elizabeth Ofiii MD (PI), Victoria M Calhoun, James Lohr, MD (PI), Peggy Gughuzza, RN VAMC East BSN, Marion Chancellor, RN E O Family Health Center, Orange East Orange, Nj Sithipom Sastrasmh, MD ( PI), Inc Miami FL Fati a A Zafar, MD (PI) , Patπcia A Sea- Vicky Johnson VAMC Houston Houston, TX Gabriel B brooks, ARNPC, DNSc Family Practice Group, Pocatello, ID Habib, MD (PI) Florence Chodk ewicz VAMC IndianapoMichael S Baker, MD (PI), Tπsh Berglund PA-C New lis, Indianapolis IN J Howard Pratt, MD (PI), Jerrlyn Britain General Hospital, New Britain, CT Sandra Raff, MD Jones, MSN VAMC Louisville, Louisville, KY Vasti L (PI), James Bernene, MD, Lawrence Koch, MD, Anthony Broadstone, MD (PI ) , Eloise Campbell, RN VAMC Mil- Lachman, MD, Elizabeth Solano, MD, Diane Bernene, RN, waukee Milwaukee WI Mahendr S Kochar, MD (PI), GloLmda Ciarcia, RN ALLHAT Clinical Center 14JB, ria Kotecki, LPN VAMC New Orleans, New Orleans, LA Unionv le, CT Sandra Raff, MD (PI), James Bernene, MD, Vecihi Batuman, MD (PI), Patricia Willhoit, RN, NP John Lawson, MD, Diane Bernene, RN, Lmda Ciarcia, RN VAMC Kansas City, MO, Kansas City, MO Santosh Sharma, ALLHAT Clinical Center 148 A, Shreveport, LA Byron Andra MD (PI), Cheryl Holt, RN VAMC St Louis, St Louis, MO M Jackson, MD (PI), Van Cleary, MD, Byron Andra M David W Moskowitz, MD (PI), Sharon Carmody VAMC Jackson, MD Marshfield Clinic, Marshfield, WI Richard Washington, DC, Washington, DC Vasilios Fapademetxiou, Dart, MD (PI), Dawn David Lakeland Center— Marshfield MD (PI), Barbara Gregory, RN VAMC Alexandria, AlexanClinic, Mmocqua, WI Richard Dart, MD (PI), Lmda Powdria, LA Terσme M Sampson, MD (PI), Everet Witzel, ers, MD, Jane Carl MediQuest Research Group, Ocala, FL MD, Janet Schmitt, PharmD VAMC Battle Creek, Battle Robert L Feldman, MD (PI), Lynn Craggs, BS, Veta Page, Creek, MI David Hallegua, MD (PI) , Michael J UnRN, Mary StandJey, RN, BS, Shan Strickland, LPN Black- derwood, MA V-4MC Bay Pines, Bay Pines, FL Ramon Lostone Cardiology Associates, Pawtucker, RL Kenneth A La- pez, MD (PI), Debbie Williams, RN VAMC Brooklyn #2, Bresh, MD (PI), Elizabeth Burns Chinatown Service Center, Brooklyn, NY Wilba L Green, MD (PI), Amal Farag, Los Angeles, CA Diana J Lee, MD (PI), Jin Sm Khoo, MD; MD, Estehta Anteola, RN VAMC Charleston, Charleston, Praphaphone Inεixiengmay, MPH Koryo Health FoundaSC Jan N Basile, MD (PI) VAMC Dαyfcw, Dayton, OH tion, Los Angeles, CA Diana J Lee, MD (PI) , Sang Lee, MD, Mohammed Saklaven, MD (PI), Anil Kumar Mandal, MD, Praphaphone Insixiengmay, MPH TH E Clmic, bic , Los Helen J Neff, RN^" VAMC Dublin, Dublin, GA Avmash C Angeles, CA , Diana J Lee, MD (PI) , Marilyn Z Norwood, Pradhan, MD (PI); Dianne K Harrison, RN. VAMC Lake RNP, Praphaphone Insixiengmay, MPH Asian Pacific City, Lake City, FL Gmsh Bhaskar, MD (PI), Ut Van Tran, Health Care Venture, Inc , Los Angeles, CA Diana J Lee, MD MD, Gloria Duren VAMC Jackson, JαΛson, MS C Andrew (PI), Praphaphone Insixiengmay, MPH. The Mary Imogene Brown, MD (PI), Ardell Hmton, MS VAMC North Little Bassett Hospital, Coopersiown, NY Anne N. Nafziger, MD, Rock, North Little Rock, AR William J Carter, MD (PI), MHS (PI), Roberta L Steere, RN. 2199 Wαrfcsite HypertenMiriam Rose Oakum, MD; Kathπne Holland, RNP VAMC sion Program, New York, NY Geoffrey Gibson, PhD (PI); Loma Lmda, Loma Linda, CA Paul G St John Hammond, Celia Shmukler, MD, Norma Martmez, RN. University of MD (PI), Jan Scott, RN VAMC Long Beαcft, Long Beach, South Dakota, Sioux Falls, SD Angelina Trupl.o, MD (PI), CA Michael A Weber, MD (PI), Dea na G Cheung, MD, Lmda Williams, RN University Physicians — Rapid City, AJH-APRIL 3995-VOL 9 NO 4 ^DΛRT 1 AN rrHYPERTENSIVE AND LIPID-LOWERING CLINICAL TRIAL 357

Ravd City, SD Angelina Tru_jillo, MD (PI), C,ndy Need- FACC, Cathy Faughn, CMA Trover Clinic — Madisonville, ham Ogdcn Resea^rch Foundation, Ogden, UT C Basil WilMadιsonvιlle, KY Abi V Rayner, MD (PI) , Connie R Tyler, liams, MD (PI), Cynthia L Slot, CRC Southern Drug ReKN Trover Clinic — Earl gton, Earlmgton KY Abi V search, Inc , Birmingham, AL Barry K McLean, MD (PI), Rayner, MD (PI), Connie R Tyler, RN Wyman Park MediMehnda L Wamwπght, Gail Wmgo Northwestern Medical cal Associates, Baltimore MD Naomi R Feldman, MD (PI), Faculty Foundation, Chicago IL Martin J Arron, MD (PI) Loretta Vogtman LPN Good Samaritan Hospital Inc , BaltiEnckson Medical Ch e, Park Rapids, MN Vern E Erickson, more, MD Duncan Salmon, MD, FACC (PI) BealthSpan MD (PI), Karen Bensen Centre Cardiovascular de Caguas, Health Systems Corporation — United Family Health Center, Caguas PR Pedro ] Colon, MD, FACC (PI) , Nivea I VazMinneapolis, MN Katherme F Guthπe, MD (PI) ALLHAT quez, BBA Woodland Avenue Health Center, Philadelphia, Clmical Center 203A Silver Spring, MD Ravi Passi, MD PA Paul D Donnan MD (PI) , Michelle Fiolkowski, PA- (PI ), Anita Passi Kenneth H Williams and Associates, BaltiQ~~ Columbia Medical Plan, Columbia, MD George Groman, more, MD Kenneth H Williams, MD (PI) , Barbara Socha, MD (PI), Michael Kelemen, MD, Eileen Bπghtwell, BS MD, Kenneth H Williams, MD Deaconess Central Hospital, University of Arkansas Medical Center, Little Rock, AR SteSi Louis, MO Madan Chilappa, MD (PI), Dannielle Cor- phanie L Lawhorn, MD (PI), Carol Davison, RN Southfaoy, PharmD, Cheryl Miller, PharmD Bellevue Hospital west Medical Associates, Las Vega':, NV James W Snyder, Center- — Ambulatory Care Practice, New York, NY Richard MD (PI), Barbara Adelman, RN, Janice Christopher, LPN I Levm, MD, FACP, FACC (PI) New York Downtown Has- II, CCRC Cardiovascular Institute of the South, Lake Charles, pital, New York, NY Ira C Schulman, MD (PI), William J LA Jose A Silva, MD (PI), Nairn Mah oud Adh, MD, Cole, MD, Kelly Ann Mele, RN Bellevue Hospital Center- Luis Felipe Tami, MD, FACC, FCCP, FACA, Catherine Met Plan, New York, NY Stuart A Diekerman, MD (PI), Mary Tami, RN, BSN Cardiology Clinic of Muskogee, inc , Andre Neusy, MD, KeilyAnn Mele, RN Valley Medical CenMuskogee, OK Yee Se C Ong, MD (PI), Viola Chnsty, RN ter, Ketienng, OH Meenakshi Patel, MD (PI), Jill Blum, East Side Internists, Inc , Providence R Richard J Ruggieπ, RN Rmgrose Clinic, Inc , Guthrie OK Robert E Rrrtgrose, MD (PI), Susan Ruggieri Sigurds fanners, MD PC, HanMD ( PI), Laura Parham Altits Medical and Surgical Clinic, cock, Ml Sigurds Janners, MD (PI), Candacε E Koski Jan- Alius, OK Joe Leverett, MD (PI), Sharon Van Pelt, LPN ners, RN, BSN University of Kansas Medical Center, Kansas Cardiology Associates of Johnstown, Johnstown, PA Charles G /, KS Brian Friedman, MD (PI), Lmda Gerrond, MD, J Osc ald, MD, FACC (PI), Amanda Bonng, EMT LanDavid B Wilson, MD, Leigh Ann Price, BS Cardiology Ascaster Heart Foundation, Lancaster, PA Seih J Woritev, MD, sociates, Por< Charlotte, FL Louis D Rosenfϊeld, MD (PI), FACC (PI), Linda Burkhardt, RN Family Medicine, Brent Terra McDonald Christ Hospital Cardiovascular Research Clark, MD Pittsburgh, P A Brent Clark, MD (PI) Sant Ram Center Cincinnati OH Robert Tolrzis MD (PI), Lmda Medical Associates West Grooe PA Deepak Sant Ram, MD Martin, RN, BSN, MBA Truman Medical Center UMKC ( PI) Family Medical Center lohnstown, PA Michael Ta- School of Medicine Kansas City MO Nathaniel Winer, MD tarko, Sr , MD (PI) , Jeanne Spencer, MD Loretta agy, (PI ) Carol J Tudor Atlantic Cardiology Associates Exeter, LPN Vnico Med_'cal Center Conemaugh, PA Michael Ta- NH Mark I Jacobs, MD (PI), Jennifer M Doane, RN tarko, S , MD (PI), Patty Allbaugh, RN Ebandμeff Medical U 5 / 'Montgomery Internal Medicine Residency Program, Center Nanty Glo, PA Michael Tatarko, Sr , MD (PI), Jean Montgomery, AL Stephen 5 Brady, DO (PI), Theresa L Marie Koh, MD Associates in Diagnostic Internal Medicine Dormmey Howard S Ellison, MD PC, Conyers GA HowPittsburgh, PA Peter P Tanzer, MD (PI), Roberta A ard S Ellison, MD (PI) Jamie D Ellison Maimomdes PriMuelier, RN Cardiology Outpatient Clini — University of mary Care Brooklyn, NY Andrew Conti MD (PI) Talum Pittsburgh Pittsburgh, PA Galal M Ziadv, MD (PI ), DeboFamily Health Chmc Gary, IN David E Ross, MD (PI), rah Dongilli, BSN Excelsior Medical Clinic, Su ier, SC JoHenedma W Macabahtaw, MD, Barbara Johnson, RN, seph C Williams, MD (PI), Beverly Hill, Medical AssisNNP Medical Diagnostics Center Indianapolis, IN John tant Memphis Medical Specialists, Inc , Memphis, TN HowHoward Pratt, MD (PI), Jerrlyn Jones, MSN The Health ard W Marker, MD (PI) University of Texas Health Center Associates, Baltimore, MD Boris Kerzner, MD (PI), Susan at Tyler Tyler, TX David R Shafer, MD (PI), Barbara E Childs, BS, BSN, MS Winona Memorial Hospital, IndiaHiltscher, RN Consultants m Medicine, Inc , PS, Belhngham, napolis IN Jack H Hall, MD (PI) , Dottie Fausset, MBA, WA Grant E Deger, MD (PI) University of Arkansas for RN, Pam Lmden, RN Faculty-Resident Group Practice, MiMedical Sciences Diabetes Clinic, Little Rock, AR Vivian A chael Reese Hospital, Chicago IL Kevin E Hunt, MD (PI), Fonseca, MD, MRCP (PI), Janet Hmson, RN Howard UniRosemary Dawkmε, RN Mancoψa Medical Center, Phoenix, versity Hypertension and Lipid Chmc, Washington, DC Tam- AZ Wilham Dachman, MD (PI), Maria Dachman, RN rat M Retta,MD, PhD (PI), Oteho S Randall, D; Shichen University of California Davis Medical Center (UCDMC), Xu, MD, TamratM Retta, MD, PhD Internal Medicine AssoSac mento, CA Charles Whitcomb MD, FACC (PI), Jean- ciates, Washington, DC Jerry M Earll, MD (PI) Hypertenme Robertson Colorado Family Medicine, PC, Denver, CO sion Center, Medical Center of Delaware, Wilmington, DE Constanfane Tsamasfyros, MD (PI), Laura Clayton Mon- William E Miller, MD (PI), Matthew Burday, DO Palm iefiore Medical Center, Bronx, NY Janet U Gorkin, MD (PI), Beach Center for Clinical Investigation, West Palm Beach, FL Mark Menegns, MD, Laurie Posner, MD, Lmda Solomon Lee A Fischer, MD (PI), Holly Hadley, MD, Pearhe SinCardiovascular Medical Specialists, PA, Hollywood, FL Jonagleton Uchen A Okoronkwo, II, MD, PC, Oakland, CA than R Jaffe, MD (PI), Lisa Nitzberg, MD Intermedic Uchertna A Okoronkwo, II, MD (PI) , Jerryann Dunmore Health Center, Port Charlotte, FL Terence P Connelly, MD L5L7 Clinics, New Orleans, LA Henry Rothschild, MD, PhD (PI) , Demse Eckstein, BSN Domi-Med, Inc , Chamblee, GA ( PI ) LSU Hypertension Research Ch c, New Orleans, LA M Donald J Weidler, MD, PhD (PI), Mayra Baez. Morehouse Eileen Cook, MD (PI), Nancy Bark, BS Acadian Cardiology, Family Pracnce Center, Atlanta, GA Donald J Weidler, MD, Lafayette, LA Vernon A, Valentino, MD (PI), Trac Quails, PhD (PI) Lutheran Geriatric Care Services, Port Wayne, IN RN, BSN Androεcoggin Cardiology Associates, Auburn, ME Jerald L Andrew, MD (PI), Chπsty Cobbum, MA, Beth Robert J Weiss, MD, FACC, FACP (PI), Carol Ridley, RN Molnar, RN, NP University of Louisville Hypertension SecHahnemann University Hospital, Philadelphia, PA Steven P tion, Louisville, KY Gurdarshan S Th d, MD, MS, FACP, Kutalek, MD, FACC (PI), Christina Ann Baessler, RN 358 DAVIS ET AL AjH- APRIL 1995-VOL 9 J Q 4 PART 1

Chapel Hill Internal Medi ine, Chapel Hill NC Paula F (PI) , Nancy P Wettermark Truman Medical Center-tart, Miller, MD (PI) United Hospitals Medical Center Newark, Kansas City, MO Diane M Harper, MD (PI ), Roberta 1= NJ Aloysius B Cuyjet, MD, FACC (PI), Thelma Allen O'Kelly, BA Ong Medical Cenier, Oxon Hill, MD Stephen Shch, MS, RNC, CS, CDE Family Practice Center, Columbus, T Ong, MD, MPH (PI), Debbie Clements, Clinical SuperGA Michael F Walsh, MD (PI) , S Troy Smith, PharmD visor St ]ohn Family Practice, St Petersburg, FL Hugo A. MSU-KCMS Internal Medicine, Kalamazoo, MI Anne Cava- St John, MD (PI ) , Teresa St John, BSN Internal Medicine nagh MD (PI), Kathv S Church, BSN, CEN , Danny M Center of Akron, Akron, OH Joseph .A Fmocchio, MD (PI), A naerson MD Inc Sonora CA Danny M Anderson, MD Irene Cnenowith, MD, Norma Durbm, RN, BSN UMDN]- (PI) , Judith Joy Boggess, MD, Lmda B Johnson, LVN St Neτv Jersey Medical Scnool Newark, NJ Maya P Raghuwan- Thomab Medical Group, Nashville, TN Mark C Houston, shi, MD ( PI), Lester Muhammad Beaumont Internal MediMD (PI) , Laurie Hays, RN Heart Center of Salt Lake, Salt cine Λssoc fes PA, Beaumont, TX Carlos Arroyo, MD (PI) Lake City, UT J Joseph Perry, MD (PI ) Wendy Schvanev- Judy Freeman VAMC East Orange, East Orange, NJ Suat eldt, RN Central NOT th Alabama Health Services, Inc , Hunts- Akgun, MD (PI ), Eileen Moser, MD, Lmda Condit, RN vύle. AL Ronald M Wyatt, MD t PI), Deborah Degree, RN, VAMC Danville Danville, IL William Marshall, MD (PI), BS Bernard M Sklar, MD, Inc , Alameda, CA Bernard M Richard Jones, PA VAMC Las Vegas, Las Vegas, NV Gopal Sklar, MD (PI ) , Made C Ma tier, MS, FNP Knoxville CardiDas, MD (PI), Carol A King, DrPH VAMC Albany, NY, ology Associates, Alcoa, TN Alan Lee Smuckler, MD ( PI) , Albany, NY James T Higgins, MD (PI) , Robert Gams, Jamie E Etherton, RN Knoxville Medical Grou — Knoxville, PharmD, J B Goss, RPh, Kerry Johnston, RPh, Michael Knoxville, TN Lee R Dιlworth, MD (PI), Jamie E Etherton, Levin, MD, Marc Stern, MD, Julie Hassenfeld, BA VAMC RN Talladega Internal Medicine, PC, Talladega, AL Simon Providence, Providence, RI Satish Sharma, MD (PI); ElizaGebara, MD (PI), Gma Bliss, RN Clinical Hypertension beth Coccio, RN VAMC Balavia, Rochester, NY Thomas Center, Los Angeles, CA Vincent DeQuattro, MD (PI), L P gree, MD (PI), Rebekah Loy, PhD VAMC Marion, MarJulian Hayward, MD, DePmg Lee, MD Brevard Cardiology ion, IL Mohammed Mansuπ, D (PI), Shaheda Mansuπ, Group, Mεrrύt Island, FL Khahd H Sheikh, MD, FACC MD, Melissa Guess, RN VAMC Des Moines, Des Momεs, (PI), Eugene Killeavy, MD, FACC, Tεm Henegen, RN IA Russell Glynn, MD (PI), Beth Hargens, RN VAMC Salt Wαiriiπgfoπ Coi-fnfy Internal Medicine PC, Sandersvύle, GA Lake City, Salt Lake City UT Ra_jat Kaul, MD (PI ), Jeanie Wilham Rawlmgs, MD (PI), Wentzelle Kim Kitchens, MD, O'Donnell, MSN VAMC Fargo, Fargo, NO Kushal Handa, Jean Rawlings Sumner, MD Jessica Heldreth, LPN Family MD BSc (Pl), Twύa Kειm BS VAMC Decatur, Decaiur, GA Practice Center Comprehensive Medical Corporation, IndianapW Virgil Brown, MD (PI), Anh Le, MD, M Marbnez- olis, IN David L Fryman, MD (PI), Neeta O'Mara, Maldonado, MD, Luis Pimentel, MD, Mary Ell en S weeney, PharmD Goel Medical Corporation, Merrύlville IN Haπsh MD Cleveland Chmc Florida, Ft Laudeidah, FL Jerry O A Shah, MD (PI) , Arun Goel, MD, T Nguven, MD, Kan Ciocon, MD, FACP, FACA, AGSF (PI) , Gregory Cohn, Smith, RMA The Heart Chmc PA Kansas City, KS Nalmi MD, Loπ Blanco, RN Smβi Hospital Center for CardiovascuG Premsmgh, MD (PI), Kiπt Masrani, MD Rita Brown, lar Research Detroit, Ml Nicholas Z Keπn, MD (PI) , Kathy PA Kaiser Landover Centei , Lanύover MD John S Golden, Faitel, RN, BSN Diabetes and Metabolism Associ te , APMC, MD (PI), Valerie Walls ALLHAT Clmical Center 256A, New Orleans LA Jonathan K Wise, MD, FACP (PI) , Skye Bridgeioaler, N] Alexander B Kudryk, MD (PI), Fred M N Noble, LPN I B Price, MD, P Λ, Quιncy, FL Ira B Price, Tepper, MD ALLHAT Clmical Center 257 A Mount Holly, MD ( PI), Patricia Walden Family Pi actice Center, Ottumwa, NJ R Bruce Denmston, MD, FACP (PI) , Annette Pagano, IA Robert H Schneider, MD (PI), Veronica Butler, MD, RN University of Washington Seattle, WA Allan J EllsLinda Hoffman, Joyce Horan, LPN Family Practice Resiworth, PharmD (PI ) Denmark Medical Center Denmark, SC dency Florida Hospital, Orlando, FL David G Pocock, MD Monnieque Singleton, MD (PI), Barbara Bomeau, MOT (PI), Karen L Kundmger, RN, BSN Marc S Posner, MD, Family Medical Center, Charleston, SC C Wayne Weart, PA, Baltimore, MD Marc S Posner, MD (PI), Cynd Comp- PharmD, FASP, BCPS (PI) , Teresa Price The Frist Chmc, ton ALLHAT Clinical Center 302A, Portsmouth, VA Dons Nashville, TN Byron Haitas, MD (PI) , Cynthia A Borum, M Rice, MD (PI) , Bonnie Stadtler Deaconess Research InstiBSN Illinois Center for Clinical Trials, Chicago, IL Glen A tute, Billings, MT Stuart I Ruben, MD (PI ) , Connie D Sussman, PhD (PI) , Seth Tannenbaum, MD, Leshe Zun, Hamilton, RN Medicine II Chmc E A Conway Hospital, MD, MBA, FACEP, Kathleen Colombo, BSN North General Monroe, LA Barbara Beard, DO (PI), Tammy V Jones, DO, Hospital, Neio York, NY Myo Maw, MD (PI), Karen Adam- Regena Tnchell Spnngfield Medical Center, Panama City, son, MD Charles F Scott, MDPC, East Point, GA Charles FL Miεal Khan, MD (PI), Sandra Fernandez Cardiology F Scσtt, MD (Pl), Cymanthιa Crowley, MA Memorial MedConsultants, Pensacola. FL Brent D Videau, MD (PI), Elizaical Center, Savannah, GA Lloyd S Goodman, MD (PI) , beth Stock, Tem Wilcox, RN The Boivhng Green Study CenTheodoia L Gongaware, MD, Nasser Mikhail, MD, Donna ter, Bowling Green, OH William E Feeman, Jr , MD (PI ), Tuteri, RN Simon- illiamson Chmc, PC, Birmingham, AL Gwenda Sue Schroeder Comprehensive Adult Risk EvaluaRichard Fuller, MD (PI ), Maria Harper, RN, James Richard tion ( CARE), Oakland, CA General K Hilliard, MD (PI), Kilgore, PA Indianapolis Cardiovascular Research Office, In- Barbara Holmes, MA David A Johvet, MD, PA, Houston, dianavσhs, IN Bτadley A Wemberg, MD (PI), David TX David A Johvet, MD (PI), Monica Bπzuela Temple Qu n, CCRN East Carolina University School of Medicine, Umvasiiy School of Medicine, Philadelphia, PA David S Greenville, NC Mark D Darrow, MD (PI), Elizabeth A Kountz, MD, FACP (PI) , Margot Boigon, MD, Michael JaMahoney, PA Selma Medical Associates, Winchester, VA cobs, PharmD, Nancy Moffatt, MSN, CRNP Health Care Randolph H Renzi, MD (PI), Lmda Stolhngs, RN Naval Plan Inc , West Seneca, NY Brian D Snyder, MD (PI), Kelly Medical Cenier San Diego, San Diego CA Hollace D Chas- Thomas, RN. Eastwick Medical Associates, Philadelphia, PA tain, II, MD (PI ) , Ker Boyce, MD Thomas A McKmght, MD, Donald Fox, MD (PI), Robert A Centrone, DO, Steven A PC, Fremont NE Thomas A McKmght, MD (PI), Jean K Femstem, MD, Harvey A Soifer, DO, Marie T Cipollone, Schaf ersman, CMA ALLHAT Clinical Center 275 A, PhiladelMT Andre K Artis, PC, Gary, IN James E Carter, Jr , MD phia, PA Pasqtiale F Nestico, MD (PI), Amy Signell Mo( PI ) , Marv Hutchmson Wai nor Family Pi actice, Tuscaloosa, bile Diagnostic Center, Mobile, AL Thomas A Kessler, MD AL H Joseph Fritz, MD ( PT) , Carolyn S Buford, LPN AJH-APRIL 1995-VOL 9. NO. 4, PART 2 ANTIHYPERTENSIVE AND LIPID-LOWERING CLINICAL TRIAL 359

ALLHAT Clinical Center 318 A, New York, NY: Mahshid Ar- Bruce J. McGann.. MD; Sandra Drebes, LPN. Pitman Internal fania Assadi, MD, FACP (PI); Cyrus Assadi, MD. Union Medicine Associates — Swedesboro, Swedesboro, NJ: Michael A. Diagnostic Clinic, Si. Louis, MO: Francois R. Charles, MD Farber, MD (PI); Lewis John DeEugenio, MD; Bruce J. (PI); Marv L. Gregory, DA. University of Arizona, Tucson, McGann, MD; Brenda Locantora, LPN. University of Iowa HosAZ: Charles Y. Lui, MSc, MD, FACC/FACA (PI); Sharon pitals, Iowa City, IA: William La ton, MD (PI); Delores Snyder, RN, MS. Family Medicine Clinic, PC, Onawa, IA: Kuebrich, RN. Mercer University, Macon, GA: Paul H. L7A- Curtis A. Mock, MD (PI); Rhonda R. Gibson, RN, BS. Mac- mato, MD (PI); David C. Parish, MD, MPH, FACP; Bobbye Neal Cenier for Clinical Research, Berwyn, IL: Charles 1. Bar- M. Wie an, RN. Pavillion Medical Associates, Baltimore, MD: eis, MD (PI); Robert J. McNall , Jr., RN, BSN; Teresa James Mersey, MD (PI); Nancy Waligorski, RN, CDE. Elliott Flegel, RN, BSN. Gunnar Medical Group, Chicago, IL: N. Schwartz, MD, PC, Oakland,^' CA: Elliott N. Schwartz, MD Charles J. Bareis, MD (PI); Peter P. Mayock, MD; Robert (PI); Patricia EUen Schwartz, RN. CoMedPw, North Andover, J. McNally, Jr., RN, BSN. Tri-County Emergency Medical SerMA: Albert Sobrado, MD (PI); Janet Pellegrino, FNP. VAMC vices Inc, Hariville, OH: Jean A. Lang, DO (PI); Thomas A. Hampton, Hampton, VA: Donald Richardson, MD (PI); Donald Gibbs, DO; Pamela Blankenship, MA, CPT. Herman Rose, Richardson, MD. ALLHAT Clinical Cenier 367A, New Roads, MD, PA, Fort Worth, TX: Herman Rose, MD, FACP (PI); LA: Donald W. Doucet, MD (PI); Madeline L. Doucet. Russell Maxine Pickard. Drs. Samuels and Huddleston APMC, Chal- W. Simpson, MiD and Associates, PC, Denver, CO: Ralph P. Ec- εtte, LA: Brace S. Samuels, MD (PI); Andre G. Smith, cles, DO (PI); Russell W. Simpson, MD; Marilyn M. Wilson. LPN. Athens Internal Medicine, Athens, AL: Nauman Qure- Medical Arts Clinic, Corsicana, TX: Kent E. Rogers, MD (PI); shi, MD (PI); Theresa Tucker, RN. Doctors Diagnostic CenPriscilla A. Hutchins, Medical Assistant. Navarro Primary ter, Minneapolis, MN: David A. Berman, MD (PI); Diane Health Care Clinic, Corsicana, TX: Kent E. Rogers, MD (PI); Crimmins, RN, CCRN. ALLHAT Clinical Center 328 A, BufBillie J. Conger, RN. San Joaquin General Hospital, French Camp, falo, NY: Joseph L. Maddi, MD ( PI); Johanna Canney, RN. CA: Ra esh Dharawat, MD, FACC (PI); All Usman, MD; Summit Cardiology Associates, Inc., Cuyahoga Falls, OH: AlRa esh Dharawat, MD, FACC. The Brooklyn Hospital Center, fred I. Narraway, DO, FACC (PI ); Stephen L. DiBlasi, DO; Brooklyn, NY: Kenneth Ong, MD (PI); Samuel Chan, MD; Karen S. Kutoloski, DO; Mary L. Hughes, RN, PA-C. Walter J. Pierce, MD; Joann.Varvatsas. K iser Permanente, HarBethesda Family Practice, Cincinnati, OH: John G. O'Hand- bor City, CA: Suzanne Barrett, PAC (PI). Suffolk Community ley, MD (PJ); Beth Akridge, RT. Lawrence J. Misko. MD, Inc., Health Center, Suffiok, VA: Vanessa A. Blowe, MD (PI); Cathy Glendo a, CA Neil E. Doherty, III, MD (PI); Vivian Doherty, R. Gordon, RN, BSN. Wayne State University, Soulhfield, MI: RN, MSN. East Bay Cardiology Medical Group, San Pablo, CA. Gary W. Edelson, MD (PI); Mohamed S. Siddique, MD; Ann Gary B. Marcus, MD, FACC, FACP (PI); Brenda Perry, RN, Pawluszka, RN. Patuxent Medical Group, Inc, Annapolis, MD: MS, NP. Falmoulh Cardiology Associates, PC, Falmouth, MA: Louis Kofi Essandoh, MD (PI); Kim Burroughs. Central Plains Thomas Sbarra, MD (PI); Catherine M. Geary, MSN, RN, CS Clinic, Sw x Falls, SD' Barry J. Lankhorst, MD, FACP (PI); UC7 Heart Disease Prevention Progam, Irvine, CA: Nathan D. Pamela Jo Rogers, LPN. Donald R. Steinmulleτ, MD, SC, Lake Wong, PhD (PI); David Abrahamson, MB, BCh, FACC; Julius Forest, IL: Donald R. Stεinmuller, MD (PI); Mary W. M. Gardin, MD, FACC; Jonathan M. Tobis, MD; Richard Will- Steinmuller, BSN. Community Seroice Chmc, LaGrangε, GA: Joner, MD; Sheila Dea in, RN. Neva Valley Cardiology, Napa, CA: seph Minore, MD (PI); William M. McClellan, Jr., MD; Judy Dale R. Stemple, MD, FACC (PI); Dan L le, RN. Eastlanά W. McCuHough, RN. Lipid Clinic I New England Medical Center, Medical Primary Care, Bloomington, IL: Galen F. Weaver, MD, Boston, MA: Ernst J. Schaefer, MD (PI); Leo Seman, MD, PhD; FACP (PI); Karen Nenne, RN. Ben Taub General Hospital, HousPatricia M. Wedge. University of South Florida, Division of Carditon, TX: Horaάo J. Adrogue, MD (PI); Debby S. Verrett. Cor- ology, Tampa, FL; Lofty L. Basta, MD (PI); Ruth Kallai, ARNP; dioυascular Laboratory, Ryder Memorial Hospital, Humacao, PR: Elizabeth A. Ford, RN, BSN. Ferndale Clim , Femdale, MI: MiMaria L. Rios, MD (PI); Marisol Rios. Primary Health Care chael S. Doyle, MD (PI); Joan Segler, RN, BSN. £./. Clinical Practices PC, Macon, GA: William E. May, MD, FACP (PI); Research, Inc. / Grand Valley Cardiology Specialists, Grand Rapids, Caroline D. Martin, BA, MBA. Michigan Medical Specialists PC, MI: J. Robert Grove, MD (PI); Elaine L. Johnson, RN, CCRC. Grand Rapids, Ml: Marian E. Oleszkowicz, MD (PI); Kyle A. Milwaukee Cardiology Associates, S.C, Mihoaukee, WI: Anita M. Rasikas, MD; Barb Frits a, LPN. Lionel B. Katchem, DO, PC, Arnold, DO (PI); Timothy Somrfiεrs, RN. Suburban Heights Ontario, CA: Lionel B. Katchem, DO (PI); Arlene J. (Penny) Medical Center, S.C, Chicago Heights, IL: Jairo B. Cruz, MD Katchem. ALLHAT Clinical Center 346A, New York, IVY: Edu- (PI); Theresa Garzelloπi, RN; Susan Mouritzen, RN. AHEC ardo L. Pignanelli, MD (PI); Minucha Ferreira-Montesino. ^' Family Medical Center, Fort Smith, AR: Jimmy D. AckJin, MD Center for Family Medicine, Greenville, SC: Palmira S. Snape, (PI); Charles C. Marsh, PharmD. MD (PI); Eva G. Darr, PharmD, BCPS- Richmond Area High NHLBI Project Office, Bethesda, MD: Jeffrey Curler, MD, Blood Pressure Center, Richmond, VA: Dean C. Williams, MD Project Director; Gerald H. Payne, MD, Project Officer; David (PI); Brian Rojas, NREMT-P; Linda Camplong, BA, MPH. Gordon, MD, PhD, Assistant Project Officer; Debra Egan, Howard S. Yager, MDPC, Atlanta, GA: Howard 5. Yager, MD MPH, MSc, Assistant Project Officer; Chuke Nwachuku, MA, (PI); Faye Yager. Stevens Cardiology Group, Edmonds, WA: SteAssistant Project Officer; Michael Proschan, PhD, Biostattsti- phen R. Yamall, MD, FACC (PI); Angelika Micketfj, CMA. cian; Kristee M. Camilletti, Contracting Officer. Goshεn Medical Center — Plainview Site, Rose Hill, NC: Francisco ALLHAT Clinical Trials Center, Houston, TX: C. Morton Becerra, MD (PI) . SUNY-HSC at Syracuse, Syracuse, NY: GunHawkins, Sc.D, Principal Investigator; Barry R Davis, MD, nar H. Anderson, Jr., MD (PI); Nancy D. Blakeman, BSN. PhD, Co-Printipal Investigator; Charles E. Ford, PhD, Co- New York Methodist Hospital, Brooklyn, NY: C.V. Ramana Investigator; David Goff, MD, PhD, Co-investigator; Robert Reddy, MB, BS (PI); Thayyullarhil Bharathan, MD; Muthus- J. Hardy, PhD, Co-investigator; Darwin R. Labarthe, MD, wamy Krishnamurthy, MD, FACP; Manohar R. Angirekula, PhD, Co-investigator; Sara Pressel, MS, Project Manager; MD. UHSCOM, Kansas City;MO: Anthony Dekker, DO (PI); J. Christine Mirvillo, Assistant Project Manager; Beverly Lewis Alderman, PhD; MattFurman, BA. UHSCOM-Excelsior Learner, Assistant Project Manager. ALLHAT Drug DistribuSprings, MO, Excelsior Springs, MO: Anthony Dekker, DO (PI); tion Center, Rockville, MD: Mary Mease, RPh, Director; Greg James LaSalle, DO. Pitman Internal Medicine Associates, Pitman, Berezuk, RPh, Co-Director. NJ: Michael A. Farber, MD (PI); Lewis John DeEugenio, MD; ALLHAT Central Laboratory, Minneapolis, MN: John 360 DAVIS ET AL AJH-APRIL 1995-VOL. 9,^' NO. 4. PART 1

Eckfeldt, MD, PhD, Director; Maren Nowicki, BS, Medical John LaRosa, MD; Suzanne Oparil, MD; Gerald H. Payne, Technologist Specialist; Jean Bucksa, Laboratory Manager. MD; H. Mitchell Perry, Jr., MD; Jeffrey L. ProbstSeld, MD;

ALLHAT ECG Reading Center, Minneapolis, MN: Rich- Paul K. Whelton, MD,^'MSc; Jackson T. Wright, Jr., MD, PhD aid S. Crow, MD, Director; Marsha McDonald, MA, Coordi(Vice-Chair) . nator. . ALLHAT Regional Physician and Study Coordinators:

Pfizer, Inc., New York, NY: Margaret M. Cobb, MD, PhD, Michael H. Alderman,- MD; Lillian Carroll, RN; Henry R. Senior Assodate Medical Director; Patricia Walmsley, MD, Black, MD; Tracy Lucente, MPH; Julie Hynes, MS, RD; WilSenior Associate Medical Director; Tony Whitehead, MD, liam C. Cushman, MD; Sandra M. Walsh, MA; Richard H. Assodate Medical Director. Grimm, MD, PhD; Carla Yunis, MD; Leslie Holland; Su¬

Zeneca, Wilmington, DE: Orysia Tresznεwsky, MD, Dizanne Oparil, MD; Angela J. Williard, RN, BSN; Pamela rector, Cardio-Renal Research; Michael D. Dugan, MD, AsRugasa, RN; Jeffrey L. Probstfield, MD; Kim Dammann, sistant Director, Chnical and Medical Affairs. . RN; Rebecca Letterεr, RN; Susan Ross, RN; Paul K. Whel¬

Bristol-Myers Squibb, Princeton, NJ: Mark E. McGovern, ton, MD, MSc; Jeff Williamson, MD, MHS; Gail Louis, RN; MD, Clinical Research Director; Margot Mellies, MD, DirecJackson T. Wright, Jr., MD, PhD; Robert A. Pospisil, RN. tor, Clinical Research-Metabolism. ALLHAT Data and Safety Monitoring Board: William

ALLHAT Steering Committee: Michael H. Alderman, B. Applegate, MD; Julie S. Buring, ScD; Richard Carleton, MD; Henry R. Black, MD; William C. Cushman, MD; MD (Chair) ; Edward Cooper, MD; Keith C. Ferdinand, MD, Charles K. Francis, MD; Curt D. Furberg, MD, PhD (Chair); FACC; Marian Fisher, PhD; Raymond W. Gifford, Jr., MD; Richard H. Grimm, MD, PhD; C. Morton Hawkins, ScD; Sheldon Sheps, MD.

Claims

Having described the invention, what is claimed is:

(1) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic, wherein the therapeutic effect of the diuretic to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only a beta-adrenergic blocker.

(2) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic, wherein the therapeutic effect of the diuretic to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only a calcium-channel blocker.

(3) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic, wherein the therapeutic effect of the diuretic to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only a vasodialator.

(4) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic, wherein the therapeutic effect of the diuretic to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only a central agonist.

(5) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic, wherein the therapeutic effect of the diuretic to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only reserpine.

(6) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic and an ACE-inhibitor, wherein the therapeutic effect of the diuretic and ACE-inhibitor to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only such diuretic.

(7) A method of reducing cardiovascular disease in an individual at risk comprising administering to the individual on a daily basis an effective amount of a diuretic and an ACE-inhibitor, wherein the therapeutic effect of the diuretic and ACE-inhibitor to reduce cardiovascular disease in such an individual is greater than such therapeutic effect in such an individual treated with only a vasodialator, a central agonist, a calcium-channel blocker, an alpha] -blocker, a beta-adrenergic blocker, reserpine.

(8) A method of claim 1, 2, 3, 4, 5, 6 or 7, wherein the diuretic is chlortahlidone.

(9) A method of claim 6 or 7, wherein the diuretic is chlortahlidone.

(10) A method of claim 6 or 7, wherein the ACE-irihibitor is ramipril.

(11) A method of claim 6 or 7, wherein the ACE-inhibitor is ramiprilat.

(12) A method of claim 6 or 7, wherein the diuretic is chlortahlidone and the ACE-inhibitor is ramipril.

(13) A method of claim 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 or 13, wherein the cardiovascular disease is heart failure.

Ill (14) A method of claim 1,^" 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 or 13, wherein the individuals is a hypertensive individual.