In 9,971 Australians aged 70 and over with no heart disease, diabetes or dementia, atorvastatin cut major cardiovascular events by 30 percent over a median 5.9 years, 297 against 412 on placebo. The trial's second headline measure, survival free of death, dementia or persistent physical disability, did not move: 637 against 676, a gap it could not separate from chance.
Across 9,971 Australians aged 70 and over, none of them with heart disease, diabetes or dementia, a 40 mg atorvastatin tablet taken every day lowered the rate of major cardiovascular events from 15.5 to 10.9 per 1,000 person-years. Person-years are the unit of exposure a trial counts in: 1,000 person-years is a thousand people followed for a year, or five hundred followed for two. In plain counts that is 297 events against 412 over a median of 5.9 years. The hazard ratio, the rate in one group divided by the rate in the other, was 0.70, and its 95 percent confidence interval, the range the data leave plausible, ran from 0.61 to 0.82.
The trial had named two primary endpoints in its protocol before it began, two outcomes it was built to answer rather than merely to record, and the second of them did not move. Death from any cause, dementia or persistent physical disability occurred in 637 participants taking atorvastatin and in 676 taking placebo, 21.6 against 23.0 per 1,000 person-years. The hazard ratio was 0.94, and its interval, 0.84 to 1.05, crosses 1, the value that would mean no difference between the groups. The P value was 0.25, which puts a gap that size well inside the range chance alone produces.
Both numbers describe the same 9,971 people over the same 5.9 years. Fewer of them had heart attacks and strokes. The same share of them died, developed dementia or became persistently disabled. The trial's authors, quoted in the American College of Cardiology's summary of the result, put part of that down to arithmetic: about 80 percent of the deaths in the trial were from causes other than cardiovascular disease, so preventing cardiovascular events left most of the deaths where they were.
How we know
STAREE was a double-blind, randomised, placebo-controlled trial run at general medical practices across Australia. A randomised controlled trial assigns each participant to a group by chance, so that the groups differ in the treatment and not in who chose it, and double-blind means neither the participants nor the clinicians treating them knew which they were taking, the placebo being identical to the drug. Community-dwelling adults at least 70 years of age with no history of cardiovascular disease, diabetes or dementia were assigned in a 1 to 1 ratio: 4,984 to atorvastatin 40 mg once daily and 4,987 to placebo. Their mean age was 74.7 years, with a standard deviation of 4.5, which is the typical distance of one person's age from that average, and 51.9 percent were women.
Both primary endpoints were composites, meaning each one counted several different events together and treated the first of them to happen as the endpoint reached. The first counted death from cardiovascular causes, non-fatal myocardial infarction, which is a heart attack, non-fatal stroke, and coronary revascularisation, the surgery or stenting that restores blood flow to the heart. The second counted death from any cause, dementia and persistent physical disability, and it is what the trial means by disability-free survival. The analyses ran to a hierarchical testing plan, an order of testing fixed before anyone saw the results.
Serious adverse events occurred in 131 participants taking atorvastatin, 2.7 percent of that group, and in 129 taking placebo, also 2.7 percent. Musculoskeletal, hepatobiliary and diabetes-related adverse events were more common in the atorvastatin group. Hepatobiliary means the liver and the bile ducts, and the abstract puts no counts against any of the three. The trial was funded by three grants from Australia's National Health and Medical Research Council, by a Stroke Prevention grant from the Heart Foundation of Australia and by Monash University, and it is registered as NCT02099123.
This piece rests on the paper's abstract, notes and conclusions at NEJM.org, read in full, and on the American College of Cardiology's summary of the trial. The journal's full text is paywalled, so the authors' own discussion of what the trial cannot show is not quoted here, and no figure appears above that the abstract does not state.
Two primary endpoints, two answers
A trial that meets one primary endpoint and misses the other has not failed, and it has not half succeeded. It has asked two questions and got two answers, and the answers differ because the questions do. The cardiovascular composite counts the events a statin acts on. The disability-free survival composite counts death from any cause, which includes every death a statin does not touch, alongside dementia and persistent physical disability.
That is where the 80 percent figure does its work. If four deaths in five in this trial were not cardiovascular, then removing a share of the remaining fifth moves the second composite very little, and the gap of 39 events between 637 and 676 is what very little looks like across nearly 10,000 people and six years. The interval on that hazard ratio, 0.84 to 1.05, is wide enough to hold a modest benefit and a modest harm at once, which is why the honest reading is that the trial did not detect a difference, not that it showed there is none.
Why it matters
The question facing a healthy person in their seventies is not the one a cardiovascular endpoint describes. It is about the years ahead and what will be possible in them. The paper opens by saying that the effectiveness and safety of statins for preventing cardiovascular events and for extending disability-free survival in older adults remain uncertain, and STAREE now puts a separate number against each half of that sentence.
Both numbers are real and they point in different directions, which is the kind of result that gets simplified one way or the other. Fewer heart attacks and strokes in the treated group is not nothing: it is 297 events against 412. The same share of people reaching death, dementia or persistent disability is not nothing either. What STAREE leaves open is whether a longer trial, or one in people old enough that cardiovascular disease is the likeliest thing to kill them, would close the distance between its two answers.
The trial answers for one population, one drug and one dose. Everyone in it was aged 70 or over, living in the community, and free of cardiovascular disease, diabetes and dementia at entry, so it says nothing about people who already have any of those and nothing about anyone younger. It was run at general medical practices across Australia, and whether the same result would appear in a population carrying a different burden of disease is untested here. The two primary endpoints disagree, and the disability-free survival interval, 0.84 to 1.05, is wide: it is consistent with a small benefit and with a small harm, so this is a trial that failed to detect a difference rather than one that established there is none. Musculoskeletal, hepatobiliary and diabetes-related adverse events were more common on atorvastatin, and the abstract gives no counts for them, so the size of that difference cannot be read here. The median follow-up was 5.9 years in people whose mean age was 74.7, a short window for an endpoint that includes dementia. And this piece rests on the paper's abstract, notes and conclusions plus a conference summary, because the full text is paywalled, so the authors' own account of the trial's limitations is not represented in it.
Whether taking a statin changes what happens to any individual. The trial reports rates in two groups of roughly 5,000 people, not outcomes for a person. It does not establish that atorvastatin has no effect on disability-free survival, only that this trial did not detect one at this size over this length of time. It does not compare atorvastatin against any other statin, or 40 mg against any other dose, because one drug at one dose was tested. It does not address adults aged 70 and over who already have cardiovascular disease, diabetes or dementia, all of whom were excluded at entry, nor anyone already taking a statin for a reason this trial did not test. And it does not report what the trial found on any endpoint beyond the two primary ones, because the abstract does not carry them.
Did the statin work or not?
It depends which of the trial's two primary endpoints you ask about, and that is the finding rather than a dodge. Major cardiovascular events fell from 15.5 to 10.9 per 1,000 person-years, a hazard ratio of 0.70 with a confidence interval of 0.61 to 0.82. Survival free of death, dementia or persistent physical disability did not differ, at a hazard ratio of 0.94, interval 0.84 to 1.05, with a P value of 0.25.
What does a hazard ratio of 0.70 mean here?
It means major cardiovascular events occurred in the atorvastatin group at 0.70 times the rate in the placebo group over the trial, which is where the 30 percent lower rate comes from. The 95 percent confidence interval of 0.61 to 0.82 is the range the trial's data leave plausible for that figure, and it sits entirely below 1, the value that would mean no difference. It compares rates between two groups of about 5,000 people and is not a probability for any individual.
Why did fewer heart attacks not show up as longer disability-free survival?
The trial's authors, quoted in the American College of Cardiology's summary, point to the causes of death: about 80 percent of the deaths in the trial were not cardiovascular. The second endpoint counted death from any cause alongside dementia and persistent physical disability, so most of what it counted lay outside what a statin acts on. The gap it recorded was 637 events against 676.
Who was not in this trial?
Anyone under 70, and anyone aged 70 or over who already had cardiovascular disease, diabetes or dementia, since all three were exclusions at entry. Participants were community-dwelling and recruited through general medical practices across Australia, so people in residential care are not represented. The trial also tested one drug at one dose, atorvastatin 40 mg once daily, against an identical placebo.
This is research reporting, not medical advice. We describe what studies found. We never tell you what to do. Talk to a doctor before changing anything about your health. Read our full position.
The evidence, read properly, every three days.
We read the paper so you do not have to. Free, and no supplement ads, ever.
Powered by Buttondown