Do statins buy healthspan? What STAREE found in 10,000 healthy adults over 70

By Marcus Renshaw · 2026-08-29 · updated 2026-09-16

• Cardiovascular
• Bloodwork
• Longevity

Every argument about statins in healthy people runs into the same wall. The trials measure events, the people taking the drug care about years, and nobody has run the trial that connects the two. Risk calculators tell you your odds of a heart attack. They do not tell you whether avoiding that heart attack leaves you walking, thinking, and living independently for longer. STAREE was built to answer exactly that question, and its results were presented at the European Society of Cardiology congress in Munich on 29 August 2026. I read its second endpoint before its first, because the second one is the question everyone actually has. The answer is unusually clean, and it is not the one either camp wanted.

STAREE gave 9,971 healthy adults over 70 atorvastatin 40 mg or placebo for a median 5.9 years. Major cardiovascular events fell from 8.3% to 6.0%, a 30% relative reduction. Disability free survival, which counts death, dementia and persistent physical disability together, did not differ between the groups. Fewer events did not translate into more good years.

What did STAREE actually test?

STAREE recruited through Australian general practice and enrolled 9,971 community dwelling adults aged 70 and over, mean age about 75, of whom 51.9% were women. Participants took atorvastatin 20 mg daily for four weeks and then escalated to 40 mg, or an identical placebo. Median follow up was 5.9 years. Lead investigator Sophia Zoungas presented the results at ESC 2026, with simultaneous publication in the New England Journal of Medicine.

The entry criteria are the part worth reading twice, because they define who this trial does and does not speak for. Per the published protocol, participants had to be living independently in the community with no history of clinical cardiovascular disease, no diabetes, and no dementia. Excluded were anyone already on a statin or other lipid lowering therapy, anyone with a total cholesterol above 7.5 mmol/L, anyone with moderate or severe kidney or liver disease, anyone scoring below 78 on the modified Mini-Mental State Examination, and anyone with a serious illness carrying an expected five year mortality.

That is a deliberately healthy, statin naive, low risk population. It is close to the group the 2026 dyslipidemia guideline sweeps in almost wholesale, where more than 93% of adults aged 70 to 79 are eligible. That recommendation has never had a clean primary prevention trial behind it. The closest was PROSPER in 2002, which randomized 5,804 adults aged 70 to 82 to pravastatin 40 mg and cut its primary endpoint at a hazard ratio of 0.85, but it enrolled people with established vascular disease alongside people who only had risk factors, so it never separated the two questions. STAREE does.

The trial ran two co-primary endpoints, which is the design decision that makes the whole thing interesting:

• Major cardiovascular events. As reported: cardiovascular death, nonfatal myocardial infarction, stroke, or coronary revascularization, first occurrence.
• Disability free survival. A composite of death from any cause, dementia, or persistent physical disability. The protocol defines persistent physical disability as loss of an activity of daily living, meaning mobility, bathing, transferring, toileting, dressing or feeding, documented at two consecutive visits six months apart. Dementia is either cognitive impairment two or more standard deviations below population norms that interferes with independent living, or a clinical diagnosis from the medical record.

The second endpoint is the unusual one. Most cardiovascular trials stop at events. This one asked whether preventing those events showed up in whether people were still functioning.

What happened to cardiovascular events?

The cardiovascular result is solid and it is what a statin is supposed to do.

A 30% relative reduction on a 2.3 percentage point absolute difference over roughly six years works out to a number needed to treat of about 37. For primary prevention that is a good number. Compare it to the Cochrane primary prevention figure of 18 major events avoided per 1000 people treated for five years, a number needed to treat near 56, and STAREE looks better than the pooled average rather than worse.

The lipid effect behind it was smaller than the dose implies. LDL-C fell by a mean of 48 mg/dL on atorvastatin against 16 mg/dL on placebo, and 40 mg of atorvastatin is a high intensity dose that guidelines expect to cut LDL-C by half. The gap is explained by the adherence curve: 80.2% of the atorvastatin group were still taking assigned therapy at one year, 66.1% at three years, and 55.6% at five. That is ordinary for a preventive drug handed to healthy people for six years, and the placebo group’s own 16 mg/dL fall, whatever drove it, narrows the contrast further. It cuts in a useful direction though, because an intention to treat result from a half adherent group understates what the drug does in someone who actually takes it.

Two honest caveats on that composite. Coronary revascularization is one of its four components, and revascularization is the softest of the four because it is partly a decision rather than purely an event. And the benefit was driven mainly by nonfatal events. There was no difference in survival between the groups.

Why did disability free survival not move?

Because the two endpoints are made of different things, and only one of them responded.

Disability free survival counts all cause death, dementia, and persistent physical disability. Statins moved none of those three. There was no difference in dementia, which is itself a useful finding given years of post-marketing worry about statins and memory, and no difference in all cause survival. What the drug moved was nonfatal heart attacks and strokes and the revascularizations that follow them. Those are real events worth avoiding, and over six years in this population they were not frequent enough, or disabling enough, to shift a composite dominated by dying and by losing cognitive or physical function for reasons that have nothing to do with atherosclerosis.

Zoungas put it plainly in the presentation: the lower incidence of major, nonfatal cardiovascular events with atorvastatin “did not result in a corresponding improvement in disability-free survival.”

It is worth being precise about what that is and is not. It is not a null result on the drug. The cardiovascular endpoint was tested first and it was clearly positive, which is what the trial’s closed testing procedure required before disability free survival was tested at all. It is a null result on the further claim, the one the longevity world actually makes, that pushing a cardiovascular risk marker in the right direction buys you functional years. Over six years, in healthy people in their mid seventies, it did not.

What did it cost?

The harms were consistent and small in absolute terms, and they land on exactly the markers a self-tracker already has on a panel.

Serious adverse events were identical between arms, which is the headline safety read and it is reassuring. The muscle signal is real but modest, a 2.6 percentage point difference against a placebo arm in which 29.4% reported musculoskeletal events anyway. The two that deserve attention are the liver signal, which is small in absolute terms but more than triples against placebo, and the diabetes signal, which reproduces the new onset diabetes effect statins have carried for over a decade.

What does this change if you are not 70?

Directly, not much, and pretending otherwise would be the whole failure mode of reading trial results as a self-experimenter. STAREE studied people over 70 with no diabetes, no cardiovascular disease, no dementia, and no prior statin exposure. If you are 45 and running a stack, the trial excluded your age group, and if your total cholesterol is high or you already take a lipid lowering drug it excluded you outright. The absolute risk arithmetic that makes the number needed to treat 37 at age 75 does not survive being moved to age 45, where six year event rates are a fraction of that.

Indirectly it changes something that matters more, and it is the reason this piece exists rather than a summary of the press coverage.

The implicit model behind most longevity protocols is a chain: move the biomarker, prevent the event, buy the year. STAREE ran that chain end to end in a real randomized trial with a hard functional endpoint, and it broke at the last link. LDL-C moved. Events moved. Function did not. That does not mean the chain never holds, and six years is a short window for an effect that plausibly needs decades to compound, which is a genuine and probably correct defence of statins here. But it does mean the last link is an assumption, not a result, and it is an assumption almost nobody states out loud when they justify an intervention by pointing at a moved marker.

That is the test I have started applying to anything I run: not whether it moves a marker, but whether anybody has checked that moving the marker changes how the years actually go. Very little passes it. It is the same structural problem as an epigenetic clock reading two years younger, or rapamycin’s mouse lifespan data failing to reproduce in human endpoints. A surrogate that responds is not the same as an outcome that improves, and an outcome that improves is not the same as a life that goes better. STAREE is the rare case where somebody funded the trial that checks all three, which is why the null half of it is more informative than the positive half.

How do you track the tradeoff on your own panel?

The asymmetry in STAREE is the practical problem with any preventive drug. The benefit is invisible: you cannot feel a heart attack that did not happen. The harms are measurable, and three of the four categories in that table show up on ordinary bloodwork. That asymmetry is why people quit statins for reasons the data does not support, and it is fixable by measuring rather than guessing.

If a statin is on the table, the markers worth having are ALT and AST for the hepatobiliary signal, and HbA1c plus fasting glucose for the diabetes signal. Both should have a pre-treatment baseline, because a single value three months in tells you nothing without the number it moved from. Creatine kinase is worth running only if muscle symptoms actually appear, since a resting CK in an untrained person and a CK two days after heavy resistance work are different measurements, and treating the second as a drug signal is a common and expensive mistake.

On timing, both of those markers have a settling period, so testing too early just buys noise. Retest cadence follows the marker, not the calendar: liver enzymes respond within weeks, HbA1c reflects roughly the previous three months and cannot be read sooner than that. And a normal result is not the same as an unchanged one, which is the difference between a reference range and your own baseline. An ALT that climbs from 18 to 38 is still inside the range and is still the drug doing something.

The other half is knowing what you started from. STAREE screened on total cholesterol, enrolled only people not currently on lipid lowering therapy, and did not use ApoB. If you are making this decision decades earlier than the trial population did, ApoB and Lp(a) tell you more about your actual particle burden than the total cholesterol figure STAREE screened on.

If you are in the age band this trial actually studied, the useful next move is to run your numbers through PREVENT and then read the 6.0% against 8.3% as what a statin plausibly buys you over six years, alongside a liver and glucose baseline before starting rather than after. If you are twenty years younger, the useful move is smaller and stranger: notice how rarely any intervention you take has been tested against an endpoint that measures whether you are still functioning, and how much of the case for it rests on the link STAREE just failed to demonstrate.

FAQ

What did the STAREE trial find?

STAREE randomized 9,971 healthy adults aged 70 and over to atorvastatin 40 mg or placebo for a median 5.9 years. Major cardiovascular events fell from 8.3% to 6.0%, a hazard ratio of 0.70, which is a 30% relative reduction. The co-primary endpoint of disability free survival, a composite of death, dementia and persistent physical disability, showed no difference between groups.

Should healthy people over 70 take a statin?

STAREE supports the cardiovascular case: about 37 people treated for six years to prevent one major cardiovascular event, with serious adverse events no more common than on placebo. It does not support a healthspan case, because disability free survival did not improve. Whether that trade is worth making is a genuine judgment call and one to make with a clinician who knows your history.

Do statins cause dementia or memory problems?

STAREE found no difference in dementia between the atorvastatin and placebo groups over roughly six years in nearly 10,000 adults over 70. Dementia was a prespecified component of a co-primary endpoint rather than an afterthought, so this is among the better quality evidence available on the question, and it points against the concern.

Why did fewer heart attacks not mean more disability free years?

The two endpoints measure different things. Disability free survival is driven by all cause death, dementia and loss of activities of daily living, none of which the statin moved. The drug moved nonfatal heart attacks, strokes and revascularizations. Over six years in this population those events were not frequent enough to shift a composite dominated by other causes of decline.

What side effects showed up in STAREE?

Musculoskeletal events in 32.0% on atorvastatin against 29.4% on placebo, diabetes related events in 4.0% against 2.7%, and hepatobiliary events in 3.3% against 0.9%. Serious adverse events occurred in 2.7% of both groups, so the difference sits in mild and moderate events rather than severe ones.

Does STAREE apply to me if I am under 70?

Not directly. The trial enrolled only adults aged 70 and over, and excluded anyone with diabetes, existing cardiovascular disease, dementia, or prior lipid lowering therapy. Absolute benefit tracks baseline risk, so the number needed to treat of 37 does not carry to a lower risk 45 year old. The structural lesson about surrogate markers and functional outcomes does carry.

Which blood markers should I watch on a statin?

ALT and AST for the liver signal, and HbA1c plus fasting glucose for the new onset diabetes signal, each with a pre-treatment baseline so you can see movement rather than just a value inside a reference range. Creatine kinase is worth checking only if muscle symptoms appear, and it should not be drawn shortly after heavy resistance training.

How large is a 30% risk reduction in real terms?

In STAREE it was the difference between 8.3% and 6.0% over about six years, so a 2.3 percentage point absolute reduction and roughly 37 people treated to prevent one event. A relative reduction always sounds larger than the absolute one, and the absolute figure shrinks as baseline risk falls, which is why the same drug looks very different at age 75 and at age 45.

Sources

• Atorvastatin, Cardiovascular Events, and Disability-free Survival in Older Adults, New England Journal of Medicine 2026: the trial report itself, published simultaneously with the ESC presentation. Both hazard ratios, both confidence intervals, the event counts, the event rates per 1000 person-years and the serious adverse event rates are taken from this abstract. The full text sits behind a paywall at NEJM
• STAREE presentation summary, American College of Cardiology: event counts for both co-primary endpoints, adverse event rates, and the presenter’s stated conclusion
• STAREE: Statins Cut MACE Risk by 30% in Older Adults With No CVD History, TCTMD: population, dosing schedule, follow up duration, and the adverse event breakdown
• STAREE trial protocol, Trials 2023: the co-primary endpoint definitions, how persistent physical disability and dementia were adjudicated, the inclusion and exclusion criteria, and the closed testing procedure
• ESC 2026 congress coverage, Healio: the number needed to treat of 37 over six years, the roughly 30% LDL-C reduction, and the dementia and survival findings
• STAREE Trial: Atorvastatin Cuts Major CV Events by 30% in Older Adults, Medscape: the mean LDL-C reductions of 48 mg/dL against 16 mg/dL and the adherence figures at one, three and five years. Registration required
• Pravastatin in elderly individuals at risk of vascular disease (PROSPER), The Lancet 2002: the closest prior trial in this age group, 5,804 adults aged 70 to 82, mixing primary and secondary prevention
• Implications of the 2026 Dyslipidemia Guideline for Primary Prevention Statin Therapy, JAMA 2026: more than 93% of adults aged 70 to 79 statin eligible under the 2026 guideline
• Statins in persons at low risk of cardiovascular disease, American Family Physician 2017: the established new onset diabetes signal, at a number needed to harm of 204 over five years
• Statins for the primary prevention of cardiovascular disease, Cochrane: the pooled primary prevention benchmark of 18 major events avoided per 1000 people treated over five years