JAMA DECEMBER 18, 2002

Major outcomes in moderately hypercholesterolemic, hypertensive patients randomized to pravastatin vs usual care: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT-LLT)

ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group

Bottom Line

In older adults with hypertension and moderate hypercholesterolemia, pravastatin did not significantly reduce all-cause mortality compared to usual care, a null result largely attributed to high rates of off-protocol statin use in the control group that blunted the differential in LDL-C reduction.

Key Findings

1. At a mean follow-up of 4.8 years, all-cause mortality (the primary outcome) was not significantly different between the pravastatin and usual care groups (RR 0.99; 95% CI, 0.89-1.11; P=0.88), with 6-year mortality rates of 14.9% vs. 15.3% respectively.

2. Coronary heart disease (CHD) event rates showed no significant difference between the intervention and control groups.

3. By year 4, total cholesterol levels were reduced by 17% in the pravastatin group versus 8% in the usual care group.

4. In a random sample assessing LDL-C, reductions were 28% with pravastatin versus 11% with usual care, representing a very narrow biological separation between the arms.

5. A massive 'drop-in' effect occurred in the usual care arm: 32% of usual care participants with CHD and 29% of those without CHD started taking non-study lipid-lowering drugs during the trial.

Study Design

Design

Randomized Controlled Trial

Open-Label

Sample

10,355

Patients

Duration

4.8 yr

Median

Setting

Multicenter, North America

Population Ambulatory adults aged ≥55 years with hypertension, at least 1 additional CHD risk factor, moderate hypercholesterolemia (LDL-C 120-189 mg/dL; or 100-129 mg/dL if known CHD), and triglycerides <350 mg/dL.

Intervention Pravastatin 40 mg daily

Comparator Usual care (standard medical therapy managed by primary care physicians, which increasingly included off-protocol lipid-lowering drugs)

Outcome All-cause mortality

Study Limitations

• The trial utilized an open-label (nonblinded) design, increasing the risk of differential co-interventions.

• Unprecedented 'drop-in' rates to statin therapy in the usual care control group severely compromised the trial's statistical power.

• Adherence to assigned therapy in the pravastatin arm declined over time, with approximately one-third of individuals nonadherent by year 5.

• The resulting narrow separation in total cholesterol and LDL-C reduction between the two arms precluded a reliable assessment of the cardiovascular benefits of statin therapy.

Clinical Significance

While ALLHAT-LLT formally failed to demonstrate a mortality benefit for primary and secondary prevention with pravastatin, the trial is not viewed as evidence against the lipid hypothesis. Rather, it highlights a crucial methodological pitfall in pragmatic trial design: when the standard of care rapidly evolves, 'usual care' control arms can adopt the intervention, diluting the treatment effect. The trial underscores that cardiovascular benefits in statin trials depend strictly on achieving a meaningful, sustained gradient in LDL-C reduction.

Historical Context

ALLHAT-LLT was designed in the early 1990s as a lipid-lowering subset of the massive ALLHAT hypertension trial, a time when statin therapy was not universally prescribed for moderate hypercholesterolemia. However, during the trial's follow-up (1994-2002), multiple landmark studies (such as 4S, WOSCOPS, and CARE) published definitive evidence of the survival benefits of statins. These paradigm-shifting publications prompted community physicians to appropriately prescribe statins to many ALLHAT-LLT patients randomized to the 'usual care' control arm, fundamentally compromising the study's ability to detect a difference between the groups.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

Pravastatin is an HMG-CoA reductase inhibitor. How does inhibiting this enzyme lead to a reduction in circulating LDL-C levels?

Key Response

Inhibiting HMG-CoA reductase decreases hepatic cholesterol synthesis. This triggers a compensatory upregulation of hepatic LDL receptors, which pull more LDL cholesterol from the blood, thereby lowering serum LDL-C levels.

Resident

If a patient from the ALLHAT-LLT demographic (hypertensive, older, moderate hyperlipidemia) presented today, how would modern ASCVD risk-guided management differ from the usual care approach used during the trial?

Key Response

Today, usual care for an older patient with hypertension and other risk factors would almost certainly include a statin based on pooled cohort equations indicating a 10-year ASCVD risk greater than 7.5 percent. This shift in standard practice is exactly what contaminated the ALLHAT-LLT control group as guidelines evolved.

Fellow

The ALLHAT-LLT trial failed to show a mortality benefit with pravastatin. Why is this trial considered a lesson in achieved LDL difference rather than evidence that statins are ineffective in this population?

Key Response

The trial is a classic example of drop-in bias. By year six, 32 percent of the usual care group had initiated statins, resulting in only a 16.7 percent difference in LDL-C between arms. This gradient was too small to power a detectable difference in clinical outcomes based on known event reduction per mmol/L of LDL lowering.

Attending

How does the phenomenon of cross-contamination observed in ALLHAT-LLT change the way you teach the interpretation of negative pragmatic clinical trials to trainees?

Key Response

A negative pragmatic trial comparing an intervention to usual care does not mean the intervention lacks efficacy if usual care rapidly adopts the intervention. Attendings must teach trainees to check the separation of biological markers (like the achieved LDL-C delta) before declaring a therapy ineffective based solely on an intention-to-treat p-value.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

ALLHAT-LLT suffered from significant drop-in and drop-out rates. How can trialists design future pragmatic open-label studies to account for temporal shifts in standard of care that threaten statistical power and intention-to-treat analyses?

Key Response

Trialists can utilize adaptive designs, event-driven sample size re-estimations based on blinded pooled event rates or biomarker separation, and pre-specify causal inference methods like marginal structural models or instrumental variable analyses to complement the primary intention-to-treat analysis.

Journal Editor

As an editor evaluating a manuscript with massive dilution of the intervention effect due to control group contamination, what specific statistical reporting and sensitivity analyses would you mandate before considering publication?

Key Response

An editor should mandate detailed reporting of the timing and types of off-protocol therapies, an assessment of the achieved biomarker gradient versus the assumptions in the initial power calculation, and per-protocol or treatment-received analyses to explore the true causal effect alongside the ITT results.

Guideline Committee

Given that the intention-to-treat analysis of ALLHAT-LLT was null for all-cause mortality, how do guideline committees synthesize this data alongside positive trials to maintain strong Class I recommendations for statin therapy in primary prevention?

Key Response

Committees rely on meta-analyses, such as those by the CTT Collaboration, which plot event reduction against absolute LDL-C reduction. Because ALLHAT-LLT had minimal LDL-C separation, its null result perfectly aligns with the regression line showing that smaller LDL-C reductions yield smaller clinical benefits, reinforcing the core guideline principle that lower is better.

Clinical Landscape

Noteworthy Related Trials

1995

WOSCOPS Trial

n = 6,595 · NEJM

Tested

Pravastatin 40 mg daily

Population

Men aged 45 to 64 years with moderate hypercholesterolemia and no history of MI

Comparator

Placebo

Endpoint

Non-fatal MI or death from CHD

Key result: Pravastatin reduced the risk of the primary endpoint by 31 percent and improved overall mortality.

2003

ASCOT-LLA Trial

n = 10,305 · Lancet

Tested

Atorvastatin 10 mg daily

Population

Hypertensive patients with moderate cholesterol and at least 3 other CV risk factors

Comparator

Placebo

Endpoint

Non-fatal MI and fatal CHD

Key result: Atorvastatin significantly reduced the primary endpoint by 36 percent compared to placebo, leading to early termination of the trial.

2016

HOPE-3 Trial

n = 12,705 · NEJM

Tested

Rosuvastatin 10 mg daily

Population

Intermediate-risk individuals without a history of cardiovascular disease

Comparator

Placebo

Endpoint

Composite of cardiovascular death, non-fatal MI, or non-fatal stroke

Key result: Rosuvastatin significantly lowered the risk of major cardiovascular events by 24 percent compared to placebo.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis