Results of the ALLHAT Trial: Is the Debate About Initial Antihypertensive Drug Therapy Over?



Results of the ALLHAT study have focused attention on the preferred approaches to the management of hypertension. Some of the conclusions of this trial have already been questioned. In this issue of The JCH, Marvin Moser, MD, and Michael A. Weber, MD, two of the senior editors, present their critiques of the ALLHAT results.

Recent publications of the major findings of the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) have captured the interest of the media and physicians worldwide.1 The results have important implications for the treatment of hypertension.

Results of the ALLHAT study on blood pressure (BP) control rates appeared in the last issue of The Journal of Clinical Hypertension and reported that careful attention to target Bps had resulted in goal BP in more than 60% of a population of patients (mean age, 67 years) that included 35% black patients and 36% with diabetes.2 These results were obtained despite protocol limitations on how medications could be combined. The study sent a message to physicians that more patients can be treated to goal Bps; at present only about 30% of hypertensive patients in the United States are at goal levels. If more people are treated to pressures of <140/90 mm Hg, fewer cardiovascular (CV) events will occur.

The final ALLHAT results published last month1 concluded that, in this trial of more than 33,000 relatively high-risk hypertensive patients who were treated over a 5 year period either with chlorthalidone, a diuretic, in doses of 12.5–25 mg/day; amlodipine, a calcium channel blocker (CCB), in doses of 2.5–10 mg/day; or lisinopril, an angiotensin-converting enzyme (ACE) inhibitor, in doses of 10–40 mg/day, patients receiving the diuretic experienced fewer overall CV events than those on the other agents. There were no differences, however, in the primary outcome of fatal or nonfatal coronary heart disease (CHD) events and no mortality difference, but there were some differences in outcome on selected end points. For example, patients on diuretics had a lower incidence of heart failure and strokes than the group randomized to lisinopril. This was especially true in black patients. The risk of hospitalized/fatal heart failure was not, however, statistically significant. The diuretics were more effective in reducing the occurrence of hospitalized/fatal heart failure in the amlodipine group.


There was some difference in achieved BPs in ALL-HAT among the three drugs tested. On average, systolic BPs were 4 mm Hg lower with diuretics than with lisinopril in black subjects and 3 mm Hg lower in patients 65 years or older. Overall, the systolic BP was 2 mm Hg higher in the ACE inhibitor group compared to the diuretic cohort. These results were not unexpected given the demographics of the patients studied. Black subjects and the elderly generally experience a greater decrease in BP on diuretics compared to medications that block the renin-angiotensin-aldosterone system. The systolic BP was 0.8 mm Hg higher and the diastolic BP 0.8 mm Hg lower with amlodipine compared to chlorthalidone (these differences were significant because of the large numbers of patients studied (chlorthalidone: 15,255; amlodipine: 9048; and lisinopril: 9054). Again, these results are similar to those noted in other comparative studies.

In ALLHAT, the second drug that could be used in patients who had not achieved goal BPs could not be a study drug. While patients on a diuretic could receive a β blocker, a logical second drug, subjects who did not reach goal BP on the ACE inhibitor, lisinopril, could not by protocol receive a diuretic or a CCB—logical choices for a second drug. Some patients did receive a combination ACE inhibitor/diuretic, but most received other agents such as a β blocker, reserpine, clonidine, or hydralazine. This sequence of care is not the usual one followed in practice and may have contributed somewhat to the difference in outcome. For example, if a diuretic had been routinely added to the ACE inhibitor, the difference in Bps between groups most probably would have been less; this combination usually reduces or eliminates any differences in response to ACE inhibitors between black and white patients. It is also possible that any difference in outcome, especially regarding the occurrence of heart failure, between the ACE inhibitor and diuretic-based treatment groups would have been minimized or eliminated. In addition, patients on a CCB could not be given either a diuretic or an ACE inhibitor as a second agent. As noted, despite this lack of more logical choices of second-step medications, a large number of patients achieved goal BP.


The ALLHAT results appear to have settled the debate about the benefits or risks of diuretics in the management of hypertension.

For many years, physicians had argued that the use of diuretics had not reduced the occurrence of CHD events to as great a degree as predicted by epidemiologic data. They reasoned that the metabolic effects of these agents theoretically might actually increase the risk of CV disease and that diuretics might be “out-classed as initial therapy for hypertension.”3 Statements in the literature describing these metabolic changes included: “The adverse effects of diuretics on uric acid metabolism, serum potassium, plasma cholesterol, and triglycerides may contribute to increases in the incidence of CHD, angina pectoris, myocardial infarction, and congestive heart failure…”4 or an even more extreme comment, “Diuretics elevate blood sugar, cause overt diabetes, induce diabetic ketoacidosis, elevate total cholesterol and LDL-C [low-density lipoprotein cholesterol], and reduce HDL-C [high-density lipoprotein cholesterol], cause deterioration of renal function and worsening of left ventricular hypertrophy. They are contraindicated in patients with hyperglycemia, hyperlipidemia, and coronary heart disease.”5 These statements were widely quoted in the literature. Yet, there were abundant data from carefully conducted hypertensive treatment trials demonstrating the reduction of CHD events when diuretics were used.6

These reports appeared to negate the “metabolic abnormalities” argument.7–11 In the clinical trials where diuretics were used as initial therapy or in combination with β blockers there was: 1) a reduction, not an increase, in CHD events; 2) cholesterol levels were not significantly increased; 3) there was an increase of only approximately 0.6% in new onset diabetes and the minimal blood glucose level increases were probably of limited clinical importance; and 4) the progression of renal disease and heart failure was decreased, not increased, with diuretic use.

Prospective studies refuted the argument that hypokalemia, a not uncommon finding with high-dose diuretics, had resulted in serious ventricular arrhythmias12—a finding that had been reported by one or two researchers whose studies were poorly controlled or included only carefully selected patients.13,14

The so-called short fall in reduction of CHD events (16% compared with a predicted 20%–25%) with the degree of BP lowering achieved in the diuretic and/or β blocker-based trials (−12/−5 mm Hg treated compared to control) probably resulted from the fact that the clinical trials were of 3–5 years duration and the epidemiologic studies were longer than 10 years.15 There is some evidence that reducing BP to this degree over longer periods of time improves outcome.16 There is also evidence from diuretic-based treatment trials17 in the elderly that CHD events can be reduced to levels close to those estimated by the longer-term observational studies.

Despite the increasing number of hypertension treatment trials that demonstrated benefit with diuretics and the repeated recommendations of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure for their use as initial therapy,18,19 the use of these agents decreased in the 1980s and 1990s. They were off the promotional radar screen; they were off patent; newer drugs promised more exciting results.20,21 With a decrease in the use of diuretics, it was to be anticipated that there would be increased numbers of resistant hypertensives. In fact, many physicians who worked as specialists in hypertension noticed an increase in their referrals of “resistant hypertension.” In several large clinics where resistant hypertension was defined as Bps >160/100 mm Hg on at least 2–3 medications, antihypertensive drug resistance disappeared in almost 50% of patients when diuretic dosage was increased or diuretics were added to the treatment regimen. The message was clear—diuretics are an important component of hypertension management.

The Systolic Hypertension in the Elderly Program (SHEP) study21 in 1991 described dramatic decreases in strokes and CV events in the elderly on the regimen based on diuretics.17 Following publication of this trial, an editorial in The Lancet on November 23, 1991, attempted closure to the “metabolic abnormalities negate blood pressure lowering benefits” argument. “Results of the SHEP study should drive a stake through the heart of some fondly held hypotheses…Significant reduction of 31% in nonfatal and fatal MIs in subjects with abnormal ECGs argues against a thiazideinduced risk for coronary death.”

In 1993 the role of diuretics was again defined based on available data in a book chapter titled “Diuretics Should Continue To Be Preferred Initial Therapy in Hypertension.”22“There is increasing evidence that diuretics as initial monotherapy or in combination with other antihypertensive drugs are highly effective in not only reducing blood pressure but in decreasing morbidity and mortality in hypertensive patients…Diuretics should continue to be considered as a preferred medication in the management of hypertension.”

In the 1990s many physicians had concluded that diuretics might be acceptable therapy in uncomplicated cases of hypertension but that other agents should be preferred in patients with comorbid conditions. The argument had persisted that diuretics might not be effective or may not be one of the medications of choice in patients with left ventricular hypertrophy or in diabetics. But data had indicated that the use of these agents resulted in regression of left ventricular hypertrophy.23,24 Other data indicated that their use in diabetics reduced CV events.25 The ALLHAT results in high-risk elderly subjects indicate that diuretics are just as effective in preventing CV events in diabetic subjects as in nondiabetics. The use of these medications should not be limited to low-risk patients.

The Swedish Trial in Old Patients with Hypertension-2 (STOP-2) study26 in 1999 reported no difference in outcome among diuretics, ACE inhibitors, and CCBs in elderly patients.

Thus, it should not have been a major surprise when the ALLHAT study reported that diuretics had proved to be as effective, or even more so, in reducing some clinical events than other agents tested.


Cost should not be the major reason for selecting one medication or procedure over another. However, if two treatments are equally effective, then the less expensive option should be chosen. This is certainly true in the management of hypertension.27,28 The ALLHAT investigators conclude that in view of their findings, cost should enter into the equation of choice and diuretics, which are considerably less costly than most other antihypertensive agents, should be considered as the first-step drug in treatment unless there are contraindications to their use.


It is important to remember that there are many reasons to use medications in addition to a diuretic in the management of hypertension. First, fewer than 50% of subjects will achieve goal BP on diuretics as monotherapy. More than 50% of black and elderly patients do achieve goal BP levels, but a majority of hypertensive patients, especially those with diabetes or evidence of renal disease, will not respond to diuretic monotherapy with normotensive levels. On the other hand, clinical trial results, including ALLHAT, were not achieved with the use of just one medication; more than 50% of subjects on any medication as monotherapy did not achieve goal BP levels. The recent trials describing benefits from treatment with ACE inhibitors or angiotensin receptor blockers (ARBs) were not trials of monotherapy.29–34 In most of these studies more than 50%–60% of patients were receiving diuretics in addition to study drugs. Improvement in outcomes, whether from heart or renal failure, usually resulted from the use of two or more medications with different mechanisms of action. There are clearly advantages to using more than one medication in the management of hypertension. If two or more agents are used, one of them should logically be a diuretic.35

There are some physicians who, based on the ALLHAT results, have begun to advise patients who are well controlled on other agents, such as ACE inhibitors or CCBs, to change their medication. Although it is clear that a diuretic should be part of a treatment regimen in hypertensive individuals, this may not be the best advice. If Bps are at levels >140/90 mm Hg, a small dose of a diuretic should be added, but if Bps are well controlled and the patient is feeling well, the decision to change therapy should be left to the physician.

Data with the ACE inhibitors or ARBs from other trials cannot be ignored. These medications (usually with a diuretic) are highly effective in slowing down progression of renal disease and, importantly, preventing new onset diabetes. There is also other evidence for the use of drugs other than diuretics in the management of hypertensive patients. For example, β blockers are effective in lowering BP; they are the drugs of choice in postmyocardial infarction patients and in angina, etc.

Thus, there is little doubt that ACE inhibitors, CCBs, β blockers, and ARBs will and should continue to be used as therapy for hypertension. ALLHAT did not report that these drugs were ineffective; only that they may not be as, or more effective, in preventing CV events (especially heart failure) than a diuretic in an elderly population of relatively higher-risk patients. As noted, the demographics of the study population, as well as the drug choices for second-step therapy, may have influenced the recorded differences between the ACE inhibitor and chlorthalidone groups.


Physicians involved in the ALLHAT trial and the National Heart, Lung, and Blood Institute should be congratulated on obtaining data that are of great importance for the management of hypertension. Publication of this trial has also put the management of hypertension back on the radar screen of the media.

There are no walkathons or telethons for hypertension; it is not a disease that is frequently highlighted on television or in newspapers. An important study like ALLHAT will help to alert the public to the problems associated with the management of hypertension and should help to improve outcome. Widespread media coverage will help to gain the attention of the literally millions of hypertensive patients who are either not on treatment or who are inadequately treated.

The ALLHAT study is a landmark trial confirming previous data and clarifying some of the myths and misconceptions that have been promulgated through the years.36 It should not be taken as just another study comparing different drugs; the results should be reviewed carefully and incorporated into the management of all patients with hypertension.