Did the Heart Protection Study have a formal extension trial?

Not a re-randomized extension. The collaborative group collected long-term mortality data through UK national registries after unblinding. Morbidity data came from hospital records. This observational follow-up extended total observation to approximately 11 years.

Did the cardiovascular benefits of simvastatin persist after the trial ended?

Yes. The separation in vascular event curves established during the 5-year randomized phase was maintained through extended follow-up. However, the curves did not continue to diverge once both groups were taking statins at similar rates.

What is the statin legacy effect, and did HPS show one?

A legacy effect means that early statin treatment produces lasting cardiovascular protection beyond the treatment period. HPS showed benefit persistence but not ongoing amplification. WOSCOPS, by contrast, showed continued divergence over 15 years.

Did long-term follow-up reveal any cancer risk from simvastatin?

No. Neither the in-trial data nor the post-trial follow-up showed any increase in cancer incidence or cancer mortality. The CTT meta-analysis, which included HPS data, confirmed no cancer signal at any site.

Was there a late signal for liver damage from simvastatin 40 mg?

No. Liver enzyme elevations during the trial were similar between groups. Post-trial UK prescribing data and the FDA's own review showed no delayed hepatotoxicity, and routine liver monitoring was eventually removed from the label.

How does the HPS follow-up compare to WOSCOPS 15-year data?

WOSCOPS showed a persistent 27% reduction in coronary death at 15 years despite post-trial treatment equalization. HPS showed benefit maintenance without further divergence. The difference likely reflects faster statin uptake in the HPS placebo group and the higher baseline risk of HPS participants.

Did HPS post-trial data contribute to the statin-diabetes link?

Yes. HPS data were included in the Sattar et al. 2010 meta-analysis that quantified a 9% relative increase in new diabetes diagnoses per 1.0 mmol/L LDL reduction. The absolute risk was small compared to the vascular benefit.

Did the 40 mg simvastatin dose show myopathy risk in long-term follow-up?

In-trial myopathy at 40 mg was very rare (0.05% per year). The higher myopathy risk that prompted the FDA's 2011 simvastatin dose restriction applied to the 80 mg dose, which was studied in SEARCH, not HPS.

What were the main limitations of the HPS post-trial data?

The post-trial phase was observational, not randomized. Morbidity data were less complete than mortality data. No serial lipid measurements were collected after unblinding. Results are specific to simvastatin 40 mg and may not generalize to higher-potency statins.

Heart Protection Study Extension Data and What Happened After the Trial Ended

Q: Should patients stop simvastatin after 5 years based on HPS data?

No. The HPS follow-up shows that stopping statin therapy halts additional benefit accumulation. Current ACC/AHA guidelines recommend indefinite statin therapy for secondary prevention. The absolute event difference from the randomized phase was never closed by later statin initiation in the former placebo group.

By HealthRX Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

At a glance

| Parameter | Detail | |---|---| | Trial | Heart Protection Study (HPS) | | N | 20,536 | | Intervention | Simvastatin 40 mg daily | | Comparator | Matching placebo | | In-trial duration | Mean 5.0 years | | Post-trial follow-up | 11 years total (approximately 6 years beyond randomized phase) | | Primary endpoint | First major vascular event (coronary death, MI, stroke, revascularization) | | Key in-trial result | 24% relative risk reduction in major vascular events (Collins et al., Lancet 2002) | | Key post-trial finding | No additional divergence in vascular event curves after unblinding; no late safety concerns |

Why post-trial data matters for statins

Randomized controlled trials typically run 3 to 6 years. That window captures enough events to prove efficacy, but it cannot answer the questions patients actually ask: Does the benefit last? Does it grow? Does something bad show up at year 8 or 12? The Heart Protection Study enrolled the broadest high-risk population of any statin trial, including patients with diabetes, peripheral arterial disease, and prior stroke, not only coronary heart disease. Its post-trial follow-up therefore addressed durability in a population more diverse than WOSCOPS or 4S ever tested.

What the original trial established

Between 1994 and 1997, HPS randomized 20,536 UK adults aged 40 to 80 with high vascular risk to simvastatin 40 mg or placebo. By the scheduled end in 2001, simvastatin produced a 24% proportional reduction in first major vascular events (p < 0.00001). The benefit was consistent across subgroups that had previously been considered uncertain: women, patients over 70, diabetics without coronary disease, and those with baseline LDL-C below 3.0 mmol/L.

Compliance was imperfect by design. About 18% of placebo-allocated participants started open-label statins during the trial, and roughly 15% of simvastatin-allocated participants stopped their drug. This crossover diluted the observed treatment effect, meaning the true biological effect of 5 years of LDL lowering was almost certainly larger than 24%.

Design of the post-trial follow-up

After unblinding in 2001, all surviving participants were eligible for open-label statin therapy through the UK National Health Service. The HPS collaborative group collected mortality and cause-of-death data through national registries (the Office for National Statistics) and hospital episode statistics, extending observation to approximately 11 years from original randomization.

The post-trial phase was observational, not a re-randomized extension. Three critical design features affected interpretation:

Statin uptake equalized rapidly. Within 1 to 2 years of unblinding, a large majority of both original groups were taking statins. The prior LDL difference between arms collapsed.
Registry linkage avoided selection bias. Mortality data came from national death records, not clinic visits, so loss to follow-up for the hard endpoint of death was minimal.
No new consent for extended observation was required for mortality outcomes because registry data were used under existing ethical approvals. Morbidity outcomes (non-fatal MI, stroke) were less completely captured post-trial.

This design closely mirrors the approach used in the WOSCOPS 15-year follow-up (Ford et al., NEJM 2007) and the Statin Therapies for Elevated Lipid Levels meta-analysis frameworks.

Framework for evaluating post-trial statin data

When reading any statin legacy-effect analysis, three questions structure the interpretation:

| Question | What to look for | HPS answer | |---|---|---| | Does the benefit persist after treatment stops? | Event curves stay separated | Yes, curves did not converge | | Does the benefit amplify (legacy effect)? | Curves continue to diverge | No. Divergence plateaued once LDL differences disappeared | | Do late harms emerge? | Excess cancer, cognitive decline, liver disease | No signal in any category |

Post-trial vascular outcomes

The primary finding: the separation in major vascular event curves achieved during the in-trial period was maintained through extended follow-up, but it did not widen. Once both groups had similar statin exposure, event rates ran roughly parallel.

This contrasts with the WOSCOPS legacy data, where West of Scotland investigators reported ongoing divergence 10 years after the pravastatin trial ended (Ford et al., NEJM 2007). Several explanations have been proposed for why HPS did not show the same amplification:

Faster statin uptake in placebo-arm survivors. UK prescribing guidelines changed substantially between 2001 and 2003. Former placebo patients adopted statins faster than WOSCOPS placebo patients did in the late 1990s.
Higher-risk population. HPS participants had more advanced disease at entry. Plaque biology in secondary prevention may respond differently to delayed statin initiation than primary-prevention plaque.
Shorter original treatment window. HPS ran 5 years; WOSCOPS ran 4.9 but had a younger cohort with more remaining life-years during which differential atherosclerosis progression could compound.

Mortality

All-cause mortality showed a modest but non-significant trend favoring original simvastatin allocation through extended follow-up. Vascular mortality remained lower in the simvastatin group. Non-vascular mortality was balanced, addressing the concern raised in older statin meta-analyses about possible excess non-cardiovascular deaths.

| Outcome | In-trial HR (95% CI) | Post-trial direction | |---|---|---| | All-cause mortality | 0.87 (0.81-0.94) | Maintained, not amplified | | Vascular mortality | 0.83 (0.75-0.91) | Maintained | | Non-vascular mortality | 0.93 (0.85-1.03) | No divergence | | Cancer incidence | 0.98 (0.89-1.08) | No late signal |

Data drawn from HPS Collaborative Group, Lancet 2002 for in-trial estimates and from the CTT Collaborators extended dataset (Cholesterol Treatment Trialists' Collaboration, Lancet 2010) for pooled long-term statin safety.

Long-term safety: what the extension ruled out

Cancer

The most persistent public concern about statins has been cancer. HPS was large enough to examine site-specific cancer incidence during the trial and found no excess (Collins et al., Lancet 2002). Post-trial follow-up confirmed this. The CTT individual-patient-data meta-analysis, which included HPS data through extended follow-up, found no increase in cancer incidence or cancer mortality at any site with any duration of statin therapy (CTT Collaborators, Lancet 2010).

Hepatotoxicity

During the trial, persistent ALT elevations (>3x ULN) occurred at similar rates in both arms. Post-trial prescribing registries in the UK have not identified a delayed hepatotoxicity signal from simvastatin. The FDA simvastatin label removed the recommendation for routine periodic liver enzyme monitoring in 2012, partly informed by long-term data from HPS and other large trials.

Myopathy and rhabdomyolysis

In-trial myopathy (CK >10x ULN with muscle symptoms) occurred in 0.05% per year on simvastatin 40 mg versus 0.01% on placebo. This low incidence at the 40 mg dose was reassuring but could not address the risk at 80 mg, which later trials (SEARCH) showed to be substantially higher. The FDA's 2011 dose restriction on simvastatin 80 mg was informed partly by SEARCH but did not affect the 40 mg dose that HPS studied.

Diabetes onset

HPS predated the widespread recognition of statin-associated new-onset diabetes. Post-hoc analyses of HPS data contributed to the CTT meta-analysis that quantified a roughly 9% relative increase in diabetes diagnosis per 1.0 mmol/L LDL reduction (Sattar et al., Lancet 2010). In absolute terms, this translated to approximately one extra diabetes case per 255 patients treated per year, a figure dwarfed by the vascular event reduction.

The "legacy effect" debate

The concept of a statin legacy effect proposes that early lipid lowering produces atherosclerosis regression or stabilization that continues to prevent events years after the drug is withdrawn. Evidence for this is mixed.

Trials suggesting a legacy effect:

WOSCOPS 15-year follow-up: 27% lower coronary death in pravastatin group despite post-trial treatment equalization (Ford et al., NEJM 2007)
ASCOT-LLA 11-year follow-up: persistent mortality benefit from atorvastatin

Trials showing benefit maintenance but not amplification:

HPS: curves parallel after unblinding
LIPID: 6-year post-trial data showed attenuation of benefit

The most likely reconciliation is that a legacy effect exists but depends on baseline risk, duration of treatment, and speed of post-trial statin adoption in the control group. When the control arm rapidly adopts statins (as happened in HPS), the window for differential plaque biology closes quickly, and curves run parallel rather than continuing to diverge.

Limitations of the HPS follow-up

Several constraints limit how far the post-trial data can be pushed:

Observational phase. After unblinding, treatment was no longer randomized. Confounding by indication (sicker patients more likely to start statins) may have partially masked any residual legacy benefit.
Morbidity data less complete than mortality. Non-fatal events were captured through hospital records, which may undercount events managed in primary care.
No lipid measurements in extended follow-up. Without serial LDL data after unblinding, the speed and completeness of treatment equalization between arms can only be estimated from prescribing records.
Simvastatin 40 mg specifically. Results cannot be automatically extrapolated to higher-potency statins (atorvastatin 80 mg, rosuvastatin 20 mg) that achieve deeper LDL lowering.

Clinical translation

For practicing clinicians, the HPS follow-up data support three positions:

First, statins do not cause late-emerging harm. Twelve years of observation revealed no delayed cancer, liver, or cognitive signal. This is directly relevant when counseling patients who worry about "long-term effects."

Second, stopping a statin does not erase prior benefit, but it does stop accumulating new benefit. The parallel post-trial curves argue against the idea that a finite statin course can replace lifelong treatment. Current ACC/AHA guidelines recommend indefinite statin therapy for secondary prevention, and HPS follow-up data are consistent with that recommendation.

Third, starting a statin earlier is better than starting later, even if both patients eventually take one. The absolute event gap established during the randomized phase was never closed by post-trial statin uptake in the placebo group. Every year of untreated high LDL-C translated into vascular events that could not be recovered.

Frequently asked questions

›

References

Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360(9326):7-22. PubMed
Ford I, Murray H, Packard CJ, et al. Long-term follow-up of the West of Scotland Coronary Prevention Study. N Engl J Med. 2007;357(15):1477-1486. PubMed
Cholesterol Treatment Trialists' (CTT) Collaboration. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet. 2010;376(9753):1670-1681. PubMed
Sattar N, Preiss D, Murray HM, et al. Statins and risk of incident diabetes: a collaborative meta-analysis of randomised statin trials. Lancet. 2010;375(9716):735-742. PubMed
Grundy SM, Stone NJ, Bailey AL, et al. 2018 AHA/ACC Guideline on the Management of Blood Cholesterol. J Am Coll Cardiol. 2019;73(24):e285-e350. PubMed
FDA. Simvastatin (Zocor) prescribing information. FDA Label