ALLHAT-LLT: Questions, Questions, and
More Questions (and Some Answers)
Patrick J. Skerrett, MA,* and Richard C. Pasternak, MD
Address tion and Infection—Thrombolysis in Myocardial
*Harvard Health Publications, Harvard Medical School, Infarction 22 (PROVE IT) trial [3] published in the spring
10 Shattuck Street, Suite 612, Boston, MA 02115, USA. of 2004, results from lipid-lowering trials have been almost
E-mail: pjskerrett@hms.harvard.edu
exuberantly positive. Trials involving 3-hydroxy-3-methyl-
Current Atherosclerosis Reports 2004, 6:375–380
glutaryl coenzyme A reductase inhibitors (ie, statins), in
Current Science Inc. ISSN 1523-3804
Copyright © 2004 by Current Science Inc. particular, have demonstrated clear benefits for diverse
patient populations [4–10]. ALLHAT-LLT, in contrast,
found no benefit for pravastatin versus usual care in older
The Antihypertensive and Lipid-Lowering Treatment to
patients with well-controlled hypertension and moderately
Prevent Heart Attack Trial—Lipid Lowering Trial (ALLHAT-
elevated low-density lipoprotein (LDL) cholesterol.
LLT) compared 40 mg/d of pravastatin with usual care
This finding was not, as some have asserted [11], evi-
among 10,355 men and women aged 55 years or older with
dence that lipid-lowering therapy is ineffective among
stage 1 or 2 hypertension, at least one additional coronary
individuals without known coronary heart disease (CHD).
heart disease risk factor, and low-density lipoprotein (LDL)
Instead, it underscores the difficulty of conducting an
cholesterol levels of 120 to 189 mg/dL. After a mean of 4.8
open-label trial in an era of rapidly changing professional
years of treatment and follow-up, the difference in total
and public understanding of the possible benefits of lipid-
cholesterol between the two arms was 9.6%, whereas in a
lowering therapy. It also focuses on the importance of trial
small, nonrandomized subsample, the LDL cholesterol dif-
design, particularly appropriate sample size calculations
ferential was 16.7%. No differences were observed
that incorporate contemporary realities of adherence.
between the pravastatin and usual-care groups with respect
Finally, the findings of ALLHAT-LLT serve to remind us of
to all-cause mortality, cardiovascular deaths, noncardiovas-
the intricate relationships between the degree of lipid low-
cular deaths, and a composite endpoint of fatal coronary
ering, relative risk reductions, and the absolute risk of the
heart disease plus nonfatal myocardial infarction. Despite
population studied.
these null findings, the results of ALLHAT-LLT are not
In this article, we offer an overview of the ALLHAT-LLT
inconsistent with previous trials because of the very small
trial that puts the results in context with a similar statin
lipid differences in the two arms. This indirectly supports
trial as well as with the wealth of evidence regarding con-
the hypothesis that LDL cholesterol lowering is central to
temporary lipid-lowering therapy.
the cardiovascular benefits associated with statin therapy,
with greater clinical impacts observed when there are
greater differences between treatment and control arms.
ALLHAT-LLT: Design and Results
ALLHAT-LLT underscores the difficulty of conducting an
The ALLHAT-LLT was a component of the larger ALLHAT
open-label trial in an era of rapidly changing professional
study, a randomized, double-blind trial initially designed
and public understanding of the possible benefits of lipid-
to compare the incidence of CHD and nonfatal myocardial
lowering therapy and highlights the substantial gap between
infarction (MI) among 40,000 high-risk hypertensive men
actual care in clinical practice and optimal care based on
and women assigned to either a thiazide diuretic (chlor-
the best knowledge from randomized clinical trials.
thalidone), a calcium channel blocker (amlodipine), an
angiotensin-converting enzyme (ACE) inhibitor (lisino-
pril), or an α-adrenergic blocker (doxazosin) [12]. The goal
Introduction of ALLHAT-LLT was to determine if, compared with usual
In the small but growing universe of large-scale lipid-low- care, pravastatin reduced all-cause mortality in older
ering trials, the lipid-lowering component of the Antihy- patients with moderate hypertension and dyslipidemia
pertensive and Lipid-Lowering Treatment to Prevent Heart and at least one additional risk factor. The trial design
Attack Trial—Lipid Lowering Trial (ALLHAT-LLT) [1••] called for randomizing 20,000 participants in the lipid-
stands as something of an anomaly. From the Lipid lowering component [13]. Interestingly, this sample size
Research Clinics Coronary Primary Prevention Trial pub- estimate was derived from the expected benefit shown in
lished in 1984 [2] to the Pravastatin or Atorvastatin Evalua- the Scandinavian Simvastatin Survival Study (4S) [6], a
, 376 Clinical Trials and Their Interpretations
Table 1. Lipid changes in ALLHAT-LLT by treatment group
Change compared with baseline, %
Parameter Pravastatin Usual care
Total cholesterol -17.2 -7.6
LDL cholesterol* -27.7 -11.0
HDL cholesterol* +3.3 +2.4
Total cholesterol differential 9.6
LDL cholesterol differential 16.7
*Nonrandom samples of 5% for usual-care patients and 10% for pravastatin patients.
HDL—high-density lipoprotein; LDL—low-density lipoprotein.
randomized trial of patients at considerably higher risk ferential was 16.7% (a decline of 27.7% in the pravastatin
and with much higher baseline cholesterol levels than in group and 11.0% in the usual-care group).
ALLHAT-LLT. All-cause mortality was similar for the two groups (Table
The ALLHAT-LLT study was conducted in 513 prima- 2), as were cardiovascular disease (CVD) deaths, non-CVD
rily community-based North American clinical centers. deaths, and the composite endpoint of fatal CHD and nonfa-
Between March 1994 and May 1998, 10,355 participants tal MI, as well as stroke, heart failure, and cancer.
were recruited. Entry criteria included age of 55 years or Why didn't ALLHAT-LLT “work?” The negative results
older, stage 1 or 2 hypertension, at least one additional from the trial can be traced to several practical issues as
CHD risk factor, LDL cholesterol of 120 to 189 mg/dL well as more fundamental factors. Although the reduction
(100 to 129 mg/dL if known CHD), and triglycerides less in total cholesterol in the ALLHAT-LLT group assigned to
than 350 mg/dL. A total of 5170 participants were ran- pravastatin was comparable with that seen in other statin
domized in a nonblinded fashion to open-label pravasta- trials, the 9.6% difference between the pravastatin and
tin (40 mg/d) and 5185 were assigned to usual care. usual-care groups was markedly lower than the mean 23%
Patients assigned to statins were offered the drug after difference for eight previous trials [1••]. Poor adherence in
receiving counseling about the National Cholesterol Edu- the pravastatin group and high crossover rates in the usual-
cation Program (NCEP) Third Adult Treatment Panel care group contributed to the modest difference in total
(ATP III) guidelines, whereas the others were assigned to cholesterol. It is possible that the high adoption of lipid-
their usual source of medical care after receiving counsel- lowering therapy, especially statin therapy, was attributable
ing about the ATP III guidelines. to the widely publicized results from several landmark sta-
Baseline characteristics were similar across the groups. tin trials [4–8] reported between the beginning of recruit-
Mean total cholesterol was 224 mg/dL, LDL cholesterol ment in 1994 and the end of follow-up in 2002. Also,
was 146 mg/dL, high-density lipoprotein (HDL) choles- because the trial was nonblinded, these crossovers were
terol was 48 mg/dL, and triglycerides were 152 mg/dL. probably not random, but rather were likely due to lower-
Mean age was 66 years; 49% of participants were women, risk patients dropping off therapy and higher-risk patients
38% were black, and 23% were Hispanic. Mean duration seeking more aggressive therapy.
of follow-up was 4.8 (± 1.3) years. Total cholesterol was In all likelihood, expanding use of lipid-lowering ther-
measured every other year in all participants; LDL choles- apy made it difficult to find patients who were not already
terol levels were measured in a random sample of 10% of being treated with statins and whose physicians were will-
pravastatin participants and 5% of usual-care participants. ing to allow them to be randomized to usual care. Instead
The primary outcome was all-cause mortality. Several of recruiting the intended 20,000 patients, the trial was
secondary outcomes were also evaluated, including a com- able to enroll only half that amount. Furthermore, as more
posite of nonfatal MI or fatal CHD, cause-specific mortal- aggressive lipid lowering became the standard of care, the
ity, and total and site-specific cancers. patients available for recruitment into the usual-care arm
Adherence to the assigned treatment declined through- had lower cardiovascular risk. Although the investigators
out the trial. Approximately 70% of participants reported originally calculated an 85.5% power to detect a 14%
taking 80% of their assigned pravastatin. In the usual-care reduction in mortality with a sample size of 20,000 partici-
group, crossover subjects reached 17% by the fourth year. pants, they stated in the final report that the trial of 10,000
Levels of total cholesterol and LDL cholesterol declined participants ultimately had 84% power to detect a 20%
significantly more in the pravastatin group than in the mortality reduction with an LDL cholesterol differential of
usual-care group (Table 1). At year 4, the difference in total 35%. The investigators did not elaborate on how they
cholesterol was 9.6% (a decline of 17.2% in the pravastatin moved from the original calculations to the final ones.
group and 7.6% in the usual-care group), whereas in a Clearly, the changing nature of the population required a
small, nonrandomized subsample, the LDL cholesterol dif- larger reduction in mortality than originally planned.
