Major Cardiovascular Events in Hypertensive Patients Randomized to Doxazosin vs Chlorthalidone: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT)

Context Hypertension is associated with a significantly increased risk of morbidity and mortality. Only diuretics and β-blockers have been shown to reduce this risk in long-term clinical trials. Whether newer antihypertensive agents reduce the incidence of cardiovascular disease (CVD) is unknown.

Objective To compare the effect of doxazosin, an α-blocker, with chlorthalidone, a diuretic, on incidence of CVD in patients with hypertension as part of a study of 4 types of antihypertensive drugs: chlorthalidone, doxazosin, amlodipine, and lisinopril.

Design Randomized, double-blind, active-controlled clinical trial, the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial, initiated in February 1994. In January 2000, after an interim analysis, an independent data review committee recommended discontinuing the doxazosin treatment arm based on comparisons with chlorthalidone. Therefore, outcomes data presented herein reflect follow-up through December 1999.

Setting A total of 625 centers in the United States and Canada.

Participants A total of 24,335 patients (aged ≥55 years) with hypertension and at least 1 other coronary heart disease (CHD) risk factor who received either doxazosin or chlorthalidone.

Interventions Participants were randomly assigned to receive chlorthalidone, 12.5 to 25 mg/d (n=15,268), or doxazosin, 2 to 8 mg/d (n=9067), for a planned follow-up of 4 to 8 years.

Main Outcome Measures The primary outcome measure was fatal CHD or nonfatal myocardial infarction (MI), analyzed by intent to treat; secondary outcome measures included all-cause mortality, stroke, and combined CVD (CHD death, nonfatal MI, stroke, angina, coronary revascularization, congestive heart failure [CHF], and peripheral arterial disease); compared by the chlorthalidone group vs the doxazosin group.

Results Median follow-up was 3.3 years. A total of 365 patients in the doxazosin group and 608 in the chlorthalidone group had fatal CHD or nonfatal MI, with no difference in risk between the groups (relative risk [RR], 1.03; 95% confidence interval [CI], 0.90-1.17; P=.71). Total mortality did not differ between the doxazosin and chlorthalidone arms (4-year rates, 9.62% and 9.08%, respectively; RR, 1.03; 95% CI, 0.90-1.15; P=.56.) The doxazosin arm, compared with the chlorthalidone arm, had a higher risk of stroke (RR, 1.19; 95% CI, 1.01-1.40; P=.04) and combined CVD (4-year rates, 25.45% vs 21.76%; RR, 1.25; 95% CI, 1.17-1.33; P<.001). Considered separately, CHF risk was doubled (4-year rates, 8.13% vs 4.45%; RR, 2.04; 95% CI, 1.79-2.32; P<.001); RRs for angina, coronary revascularization, and peripheral arterial disease were 1.16 (P<.001), 1.15 (P=.05), and 1.07 (P=.50), respectively.

Conclusion Our data indicate that compared with doxazosin, chlorthalidone yields essentially equal risk of CHD death/nonfatal MI but significantly reduces the risk of combined CVD events, particularly CHF, in high-risk hypertensive patients.

The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) is a randomized, double-blind, active-controlled trial sponsored by the National Heart, Lung, and Blood Institute (NHLBI). ALLHAT is designed to determine whether the incidence of the primary outcome—a composite of fatal coronary heart disease (CHD) and nonfatal myocardial infarction (MI)—differs between treatment with a diuretic (chlorthalidone) and treatment with each of 3 other types of antihypertensive drugs—a calcium antagonist (amlodipine), an angiotensin-converting enzyme inhibitor (lisinopril), and an α-adrenergic blocker (doxazosin)—in high-risk hypertensive persons aged 55 years or older. Secondary outcomes include all-cause mortality, stroke, and all major cardiovascular disease (CVD) events. In addition, many ALLHAT participants with mild-to-moderate hypercholesterolemia (n=10,337) are also participating in a randomized, open-label trial designed to determine whether in this population, lowering serum low-density lipoprotein cholesterol levels with a 3-hydroxy-3-methylglutaryl coenzyme A (HMG CoA) reductase inhibitor (pravastatin) reduces all-cause mortality compared with a control group receiving usual care. Patient enrollment began in February 1994 and active follow-up is scheduled to end in March 2002.¹

Following independent data reviews on January 6 and January 21, 2000, the director of the NHLBI accepted a recommendation to discontinue the doxazosin treatment arm in the blood pressure (BP) component of the trial. While there were essentially no differences in the rates of the primary outcome or all-cause mortality between the 2 treatment groups, there was a statistically significant 25% higher incidence of major CVD events in participants assigned to the doxazosin group compared with those assigned to the chlorthalidone group. In addition, the likelihood of observing a significant difference for the primary outcome by the scheduled end of the trial was very low. It was determined that participants assigned to the doxazosin group should be informed of their BP treatment assignment and that the major clinical findings regarding this treatment and its comparison agent, chlorthalidone, should be reported as soon as possible. Regarding other treatment comparisons, the ALLHAT Data and Safety Monitoring Board (DSMB) emphasized the crucial importance of continuing the rest of the BP and lipid-lowering components of the trial. In this article, we report the findings that formed the basis for the decision to discontinue the doxazosin arm of ALLHAT.

The rationale and design of ALLHAT are presented in detail elsewhere.¹ Briefly, eligible participants for ALLHAT were men and women aged 55 years or older who had systolic BP (SBP) of at least 140 mm Hg and/or diastolic BP (DBP) of at least 90 mm Hg, or took medication for hypertension, and had at least 1 additional risk factor for CHD. These risk factors included previous MI or stroke, left ventricular hypertrophy by electrocardiogram or echocardiogram, history of type 2 diabetes, current cigarette smoking, and low high-density lipoprotein cholesterol level. Details of the inclusion and exclusion criteria have been described previously.¹ From February 1994 to January 1998, 42,448 participants were recruited in 625 centers in the United States, Canada, Puerto Rico, and the US Virgin Islands. All participants signed an informed consent form and all centers received institutional review board approval.

Participants were assigned by a computer-generated randomization schedule to 1 of 4 treatments: chlorthalidone, amlodipine, lisinopril, or doxazosin, in a ratio of 1.7:1:1:1, respectively. Randomization was stratified by center and blocked over time to maintain the ratio. The randomization code was held only by the ALLHAT Clinical Trials Center (CTC). The treatment goal in all 4 arms was a DBP of less than 90 mm Hg and an SBP of less than 140 mm Hg. The therapeutic goal was to achieve BP control with the lowest possible dosage of the randomly assigned study drug. All 4 drugs were encapsulated and identical in appearance so the identity of each agent was masked at each dosage level. Dosages for doxazosin were 2, 4, and 8 mg/d; corresponding dosages for chlorthalidone were 12.5, 12.5, and 25 mg/d, respectively. If participants did not meet the BP goal while taking the maximum tolerated dosage of the initial medication, an open-label Step 2 agent (atenolol, 25-100 mg/d, reserpine, 0.05-0.2 mg/d, or clonidine, 0.1-0.3 mg twice per day), or an open-label Step 3 agent (hydralazine, 25-100 mg twice per day) could be added. After initial titration visits, participants were seen routinely every 3 months during the first year of follow-up and every 4 months thereafter.

The primary end point of the BP component was the composite of fatal CHD and nonfatal MI. Four main protocol-defined secondary outcomes were also monitored, including (1) all-cause mortality, (2) combined CHD (CHD death, nonfatal MI, revascularization procedures, and hospitalized angina), (3) stroke, and (4) combined CVD (CHD death, nonfatal MI, stroke, revascularization procedures, angina [treated in hospital or as outpatient], congestive heart failure [CHF; treated in hospital or as outpatient], and peripheral arterial disease [in-hospital or outpatient revascularization]). The individual components of the combined outcomes were also examined.

The occurrence of study end points was assessed at each follow-up visit and reported to the ALLHAT CTC. Because ALLHAT was a large, simple trial, study outcomes were to be counted based on clinic investigator report.¹ However, for deaths and hospitalizations, copies of death certificates and hospital discharge summaries were requested. These documents were used to support the clinician-assigned diagnoses of MI, stroke, CHF, angina, revascularization procedures, peripheral arterial disease, cancer, unintentional injury, or attempted suicide. All documents were verified for completion and appropriateness by the CTC. For a random (10%) subset of fatal CHD and strokes and hospitalized nonfatal MIs and strokes, more detailed information was routinely requested by the CTC to validate the procedure of using clinician diagnoses with accompanying documentation. For this subset, in-hospital electrocardiograms and enzyme levels (for CHD) and neurologists' reports and computed tomography and/or magnetic resonance imaging reports (for strokes) were evaluated by the Endpoints Subcommittee to determine whether the events met study criteria. Heart failure was not part of the original validation process. However, a one-time sample of heart failure hospitalizations was also reviewed. The agreement rate between the subcommittee and the clinic investigators was 92% (72/78) for the primary end point and 83% (20/24) for CHF hospitalizations and was similar in both the chlorthalidone and doxazosin groups.

Given the achieved sample size of 24,335 in the doxazosin and chlorthalidone groups combined, ALLHAT had about 83% power to detect a 16% reduction in risk of the primary outcome at a 2-sided α level of .0178, accounting for treatment crossovers, losses to follow-up, and multiple comparisons.¹ Data were analyzed according to participants' randomized treatment assignments regardless of their subsequent medication adherence. Cumulative event rates were calculated using the Kaplan-Meier procedure. Therefore, an individual's duration in the study began at baseline and ended at the date of last known follow-up. Rates are presented only through 4 years because less than 5% of the outcomes occurred after this time. The log-rank test was used to evaluate differences between cumulative event curves and to obtain 2-sided P values. Relative risks (RRs) (hazard ratios) and 95% confidence intervals (CIs) were calculated using a proportional hazards model.² Both the log-rank test and the proportional hazards model incorporated the participants' entire trial experience. Relative risks and P values should not be overinterpreted because the comparison between doxazosin and chlorthalidone was terminated earlier than scheduled.³

The DSMB met semiannually to review the accumulating data and to monitor the trial for either superiority or inferiority of the 3 agents compared with chlorthalidone. A specific charge of the DSMB was to evaluate the unblinded data for emergence of clinically important treatment differences that might warrant alteration of the protocol or early termination of 1 or more arms or the entire trial. The Lan-DeMets version of the O'Brien-Fleming group sequential boundaries was used to assess treatment group differences, and conditional power was used to assess futility.^4,5

Table 1 presents baseline characteristics for the chlorthalidone and doxazosin treatment groups in ALLHAT. Participants had a mean age of 67 years; 47% were women; 49% were white non-Hispanic, 35%, black, and 16%, Hispanic; 36% were diabetic. The large sample size resulted in virtually identical distributions of baseline characteristics in the 2 treatment groups. Figure 1 shows the number of patients randomized and followed up to the time of the interim analysis on which this report is based.

At the time of the present analysis, 501 (3.2%) of the chlorthalidone group and 338 (3.7%) of the doxazosin group were lost to follow-up. The median length of follow-up for all participants was 3.3 years. The number of patients eligible for follow-up visits in the chlorthalidone and doxazosin groups were 15,013 and 8924, respectively, at 1 year, decreasing to 6826 and 4030 at 3 years and 3229 and 1904 at 4 years. Among patients who were randomly assigned to the chlorthalidone group and were seen in the clinic or contacted by telephone within 2 months of the annual scheduled visits, 88% (11,207/12,667) were taking chlorthalidone or another diuretic at 1 year, decreasing to 87% (5178/6108) at 3 years and 86% (2596/3027; 83% taking a diuretic alone and 3% taking a diuretic plus an α-blocker) at 4 years. Only 1% were taking an α-blocker without a diuretic at 4 years. Among patients assigned to the doxazosin group, 82% (5936/7269) were taking doxazosin or another α-blocker at 1 year, decreasing to 76% (2674/3512) at 3 years and 75% (1292/1730; 62% taking an α-blocker alone and 13% taking an α-blocker plus a diuretic) at 4 years. About 10% were taking a diuretic and not an α-blocker at 4 years. The most common reasons stated for discontinuing medication at 4 years in the chlorthalidone (n=604) and doxazosin (n=467) groups were symptomatic adverse effects (20% and 19%, respectively) or an unspecified refusal (30% and 29%, respectively); discontinuations were 5% and 1%, respectively, for an abnormal laboratory value.

Among patients with available medication information at 1 year, 27% of those assigned to chlorthalidone were taking a Step 2 or 3 drug (atenolol, 18%; reserpine, 3%; clonidine, 7%; and hydralazine, 4%), whereas 32% of those assigned to doxazosin were receiving step-up therapy (atenolol, 20%; reserpine, 3%; clonidine, 9%; and hydralazine, 5%). At 3 years, the corresponding percentages were 37% (atenolol, 25%; reserpine, 4%; clonidine, 9%; and hydralazine, 8%) and 44% (atenolol, 29%; reserpine, 5%; clonidine, 13%; and hydralazine, 10%), respectively, and at 4 years, 40% (atenolol, 26%; reserpine, 5%; clonidine, 10%; and hydralazine, 9%) and 47% (atenolol, 31%; reserpine, 6%; clonidine, 14%; and hydralazine, 12%), respectively. Patients could be taking more than 1 step-up drug.

At 1 year, 40% of participants assigned to chlorthalidone and still taking their blinded medication were taking the study maximum dosage of 25 mg/d, whereas 45% of those assigned to doxazosin and still taking the blinded medication were taking the study maximum dosage of 8 mg/d. The corresponding percentages were 53% and 58% at 3 years and 57% and 61% at 4 years, respectively.

Mean (SD) seated BP at the initial visit was 145/83 mm Hg in both groups, with 90% of patients reporting antihypertensive drug treatment (Table 1). Among patients returning for follow-up visits, the mean (SD) BP at 1 year was 137/79 (15/9) mm Hg in the chlorthalidone group and 140/79 (17/10) mm Hg in the doxazosin group; at 2 years, the corresponding BPs were 136/78 (16/9) and 138/78 (17/10) mm Hg; and at 4 years, 135/76 (16/10) and 137/76 mm Hg (Figure 2). At the initial visit, the proportion of participants at or below the BP goal (<140/90 mm Hg) was 27% in both treatment groups; at 2 years, it was 61% in the chlorthalidone group and 54% in the doxazosin group; and at 4 years, it was 64% and 58%, respectively.

Mean total serum cholesterol levels at baseline were 216 mg/dL (5.58 mmol/L) in the chlorthalidone group and 215 mg/dL (5.56 mmol/L) in the doxazosin group. At 2 years, the respective mean levels were 204 and 196 mg/dL (5.27 and 5.06 mmol/L) and at 4 years, 196 and 188 mg/dL (5.06 and 4.86 mmol/L). By 4 years, about 39% (1740/4502) of both groups combined reported taking lipid-lowering drugs, largely HMG CoA reductase inhibitors. Mean serum potassium levels at baseline were 4.3 mmol/L in the chlorthalidone group and 4.4 mmol/L in the doxazosin group and at 4 years were 4.0 and 4.4 mmol/L, respectively. Mean serum glucose levels at baseline were 124 mg/dL (6.8 mmol/L) in the chlorthalidone group and 123 mg/dL (6.8 mmol/L) in the doxazosin group and at 4 years were 124 and 118 mg/dL (6.8 and 6.6 mmol/L), respectively.

Among all combined CVD events that were deaths and/or hospitalizations, the proportions with documentation at the time of the present analysis (ie, a death certificate or a hospital discharge summary) were 92.2% and 92.6% in the chlorthalidone and doxazosin groups, respectively. Outcomes data that formed the basis for the decision to terminate the doxazosin arm are presented in Table 2.

No significant difference was observed between the 2 treatment groups for the primary outcome or all-cause mortality (Table 2; Figure 3). At the time of the decision to terminate the doxazosin vs chlorthalidone comparison, about 61% of the total expected number of CHD events had occurred in the chlorthalidone group. If the protocol-specified alternative hypothesis (16% reduction) was assumed for the remainder of the trial, there was only a 1% likelihood of finding a significant beneficial effect of doxazosin at the scheduled end of the trial.

Combined CHD (RR, 1.10; 95% CI, 1.00-1.12) and stroke (RR, 1.19; 95% CI, 1.01-1.40) were increased in the doxazosin group compared with the chlorthalidone group. Moreover, there was a 25% higher risk of combined CVD outcomes in the doxazosin group (RR, 1.25; 95% CI, 1.17-1.33) (Table 2 and Figure 3). This included a doubled risk of CHF (RR, 2.04; 95% CI, 1.79-2.32) (Table 2 and Figure 3). The event curves separated largely during the first year but continued to diverge thereafter. The result for fatal and nonfatal CHF with hospitalization (RR, 1.83; 95% CI, 1.58-2.13) was similar to that for all CHF. The differential effect of treatment on CVD outcomes and CHF was consistently observed among the following predefined subgroups: patients younger than 65 years vs those aged 65 years or older; black persons vs nonblack persons; and patients with vs without diabetes (Table 3). Increased risk also was observed in the doxazosin group compared with the chlorthalidone group for the other major components of the combined CVD outcomes, including coronary revascularization (RR, 1.15; 95% CI, 1.00-1.32) and angina (RR, 1.16; 95% CI, 1.05-1.27). Doxazosin was associated with a nonsignificant increase in peripheral arterial disease (RR, 1.07; 95% CI, 0.88-1.30).

The decision to discontinue the doxazosin arm of the antihypertensive trial component was based on several factors. Foremost was a significantly higher incidence of combined CVD events and, in particular, CHF events, for the doxazosin group compared with the chlorthalidone group. In addition, with essentially equal rates in the 2 treatment groups for the primary CHD outcome and total mortality, a beneficial effect of doxazosin at the scheduled trial termination was highly unlikely based on conditional power calculations. There were also negative trends for stroke and for combined CHD, particularly 2 of its components, coronary revascularizations and angina.

To our knowledge, no other large-scale trial comparing diuretic to α-blocker therapy has been conducted. Earlier trials were too small to detect a doubling in CHF risk. A double-blind randomized trial of the effect of vasodilator therapy (added to standard pharmacotherapy) on mortality in chronic CHF has compared placebo, isosorbide-dinitrate hydralazine, and prazosin.⁶ There was a mortality benefit for isosorbide-dinitrate hydralazine compared with placebo but not for prazosin compared with placebo. Drugs were discontinued by 22% of the participants in the isosorbide-dinitrate hydralazine and placebo groups and by 27% of those in the prazosin group. Discontinuations due to worsening of heart failure were 8.5% in the prazosin group, 7.5% in the isosorbide-dinitrate hydralazine group, and 5.5% in the placebo group.

It is difficult to judge whether in ALLHAT the CHF rate with doxazosin is the same as, less than, or more than would be expected without antihypertensive drug treatment. In the Systolic Hypertension in the Elderly Program (SHEP), whose participants were about 5 years older on average than those in ALLHAT, 4.4% of the placebo group and 2.3% of the diuretic group experienced heart failure during 4.5 years of follow-up, and the incidence of heart failure increased with age.⁷ In ALLHAT, the 4-year cumulative CHF incidence was 4.5% in the chlorthalidone group and 8.1% in the doxazosin group. The ALLHAT participants had more risk factors (other than hypertension) than their SHEP counterparts did. The criteria for CHF diagnosis in ALLHAT were adapted from SHEP.⁷ In the recently reported Swedish Trial in Old Patients with Hypertension–2 Study (STOP-2), about 10% of participants experienced heart failure during 6 years of follow-up.⁸ Participants in STOP-2 were about 9 years older on average and had more severe hypertension compared with those in ALLHAT.

In ALLHAT, loss to follow-up and documentation of events were similar in the doxazosin and chlorthalidone groups. However, at 4 years, 86% of those assigned to chlorthalidone were still taking a diuretic, whereas 75% of those assigned to doxazosin were still taking an α-blocker. This lack of full adherence may have resulted in an underestimation of the true difference in CVD rates between the 2 treatments.

In this study, mean SBP in the doxazosin group was about 2 to 3 mm Hg higher than in the chlorthalidone group; mean DBP was the same. How much of the differences in CVD end points might be accounted for by the 3–mm Hg difference in SBP? In SHEP, a 12–mm Hg lowering of SBP with a chlorthalidone-based regimen produced a 49% reduction in CHF incidence, while in the Systolic Hypertension in Europe Trial, a 10–mm Hg difference between the nitrendipine and placebo groups was associated with a (nonsignificant) 29% lower CHF rate.^7,9 These data suggest that a 3–mm Hg higher SBP could explain a 10% to 20% increase in CHF, but not a doubling of the risk. Similar calculations for stroke based on a meta-analysis of all diuretic/β-blocker–based treatment trials¹⁰ and for angina based on data from the Hypertension Detection and Follow-up Program¹¹ suggest that 3 mm Hg could account for a 15% to 20% increase in stroke risk and about a 12% increase in angina risk. Thus, the observed BP differential may explain much of the stroke and angina differences observed between chlorthalidone and doxazosin in ALLHAT.

Heart failure affects nearly 4.6 million people in the United States, is a major cause of morbidity and mortality, and is the most common hospital discharge diagnosis among patients older than 65 years.¹² In the Framingham Heart Study, 90% of heart failure cases were preceded by hypertension.¹³ In hypertensive patients especially, left ventricular hypertrophy is a common precursor of heart failure.¹⁴ It is now recognized that both α₁-adrenergic and β-adrenergic activation are related to cardiac hypertrophy, and that α-adrenergic receptors share common intracellular signaling pathways with other hypertrophic growth factors such as endothelin and angiotensin II.^15,16 The largest and longest previous trial testing a diuretic and an α-blocker, the Treatment of Mild Hypertension Study, reported that left ventricular mass was reduced in the chlorthalidone arm compared with placebo, while the change in the doxazosin arm was indistinguishable from placebo.¹⁷ In 2 smaller, shorter trials without a placebo comparison, trends toward greater left ventricular mass reductions in the diuretic vs α-blocker group were seen.^18,19 α-Blockers increase plasma volume²⁰ and may increase plasma norepinephrine levels.²¹ Whether these 2 changes may have contributed to the increased incidence of heart failure cannot be determined from the present study. Also, it is unknown whether α-adrenergic responsiveness is increased or decreased in hypertensive or heart failure patients.²² Thus, currently available mechanistic evidence neither supports nor refutes the biological plausibility of the ALLHAT CHF findings with regard to doxazosin.

In the "Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure," recommendations on initial drug choices for treating hypertension include α-blockers as a possible consideration, especially if the comorbid conditions of dyslipidemia and/or prostatism are present.²³ In this comparative trial, doxazosin was less effective than chlorthalidone in lowering SBP, leading to increased use of Step 2 and Step 3 drugs. Also, more people discontinued doxazosin than chlorthalidone. Thus, ALLHAT results demonstrate that chlorthalidone is superior to doxazosin as a first-line antihypertensive drug in a diverse group of older hypertensive patients with other CVD risk factors. However, the use of doxazosin as part of a multidrug regimen for treating hypertension alone or hypertension with symptoms of benign prostatic hypertrophy was not tested in this trial. Since ALLHAT was not a placebo-controlled trial but rather an active-controlled one, the study does not allow an assessment of whether doxazosin is better than placebo.

ALLHAT is continuing for participants in the remaining antihypertensive treatment arms and those in the lipid-lowering trial component. The trial is likely to provide important information about other pharmacological treatments of hypertension and the utility of lipid-lowering therapy in older, moderately hypercholesterolemic persons with hypertension.