For roughly thirty years, a low-dose aspirin tablet sat on millions of bathroom counters as a daily ritual. Doctors recommended it. Pharmacies stocked dedicated “low-dose 81 mg” aspirin shelves. Parents reminded their adult children. The advice was so universal it stopped sounding like medical guidance and started sounding like common sense — the cardiovascular equivalent of brushing your teeth.
Then in September 2018, within roughly two weeks of one another, three large randomized controlled trials published results that demolished the foundation of that advice. ASPREE in the healthy elderly. ARRIVE in moderate-risk middle-aged adults. ASCEND in patients with diabetes. Different populations, different research groups, different funders. All three reached the same conclusion: for most adults at average or moderate cardiovascular risk, daily aspirin’s bleeding harms equal or exceed its cardiovascular benefits.
Within a year, the American College of Cardiology and American Heart Association revised their primary prevention guidelines. By 2022 the U.S. Preventive Services Task Force had pulled back its recommendation too. The “take a baby aspirin” advice — the one that had outlived disco, the cassette tape, and most marriages from the 1980s — quietly disappeared from the standard prevention checklist for healthy adults.
The reversal matters less because of aspirin specifically and more because of what its trajectory reveals. Established medical guidance that millions of people built daily habits around can be wrong. Not “wrong” in a slow, contested way that nobody notices, but wrong in a way that three contemporary trials can flip in a single news cycle. For strategists evaluating any preventive intervention — whether in healthcare, public policy, business risk management, or product decisions — the aspirin reversal is a structural lesson about how older evidence can quietly degrade as the world it was measured in changes.
Three decades of confident guidance
The aspirin-for-primary-prevention story begins with secondary prevention, which is a different question entirely. In adults who have already had a heart attack or stroke, aspirin clearly reduces the risk of a second event. That evidence base is large, consistent across trials, and remains intact today. Nobody is challenging it.
The leap from secondary to primary prevention happened in the late 1980s and 1990s through trials like the Physicians’ Health Study (1989) and the British Doctors’ Aspirin Trial. These trials enrolled middle-aged healthy men and reported reductions in first heart attacks. The relative risk reductions looked impressive — often in the 25 to 30 percent range — and the absolute risks of bleeding looked manageable. Meta-analyses combined the trials and confirmed a modest but apparently real benefit.
By the early 2000s, this evidence had been translated into guidance. The American Heart Association recommended aspirin for adults at moderate cardiovascular risk. The U.S. Preventive Services Task Force gave it a strong recommendation for adults aged 50 to 59 with sufficient risk. European guidelines followed similar logic with regional variations. Aspirin became one of the most widely recommended preventive medications in adult medicine.
The recommendations rested on three implicit assumptions. First, that the relative risk reduction observed in older trials would continue to translate into a meaningful absolute risk reduction in current populations. Second, that bleeding risk was predictable and lower than cardiovascular benefit at the population level. Third, that the older trials reflected modern medical reality well enough that extrapolating their results forward made sense.
All three assumptions, it would turn out, were quietly eroding while the guidance stayed in place.
The 2018 reversal: ASPREE, ARRIVE, ASCEND
The reversal did not come from a single trial. It came from three large, well-designed RCTs that published essentially in parallel and that, together, made the older evidence base look thin and dated.
ASPREE (Aspirin in Reducing Events in the Elderly) was the largest contemporary primary prevention trial in healthy older adults. Conducted in Australia and the United States, it enrolled 19,114 community-dwelling adults aged 70 and older (65 and older for Black and Hispanic Americans in the U.S. arm) who had no cardiovascular disease, dementia, or major physical disability at baseline. Participants were randomized to 100 mg daily aspirin or placebo. The primary endpoint was disability-free survival.
The results, published in NEJM in October 2018, showed no benefit. Disability-free survival was effectively identical in the two arms. Cardiovascular event rates were not meaningfully reduced. Major hemorrhage was significantly increased in the aspirin group. A companion paper raised the possibility — controversial and still debated — of increased all-cause mortality in the aspirin group, driven partly by a signal for increased cancer mortality. Whether the cancer signal is causal or chance remains contested, but the lack of cardiovascular benefit is not.
ARRIVE (Aspirin to Reduce Risk of Initial Vascular Events) was a 12,546-patient trial of moderate-risk middle-aged adults conducted across seven countries. Men aged 55 and older and women aged 60 and older with multiple cardiovascular risk factors but no diabetes and no prior cardiovascular event were randomized to 100 mg enteric-coated aspirin daily or placebo. The trial ran for a median of five years.
The results, published in The Lancet in late August 2018, showed no significant difference in the composite cardiovascular endpoint between aspirin and placebo. Gastrointestinal bleeding events were roughly doubled in the aspirin arm. Importantly, observed event rates in the placebo group were substantially lower than the trial had been powered to expect — the “moderate risk” population recruited turned out to be lower-risk in practice than older risk models had predicted.
ASCEND (A Study of Cardiovascular Events in Diabetes) was a 15,480-patient trial in adults with diabetes but no established cardiovascular disease, conducted in the U.K. Participants were randomized to 100 mg daily aspirin or placebo and followed for a mean of 7.4 years. Diabetes is itself a major cardiovascular risk factor, so this population should have been one of the strongest cases for aspirin’s benefit.
The results, published in NEJM alongside ASPREE, showed a small absolute reduction in serious vascular events of around 1 percent over the trial period. That reduction was almost exactly offset by an equivalent absolute increase in major bleeding events. The net clinical benefit, in other words, was essentially zero. The trial authors concluded that “the absolute benefits were largely counterbalanced by the bleeding hazard.”
Taken together, the three trials covered the three populations that had previously been considered the strongest candidates for aspirin’s primary prevention benefit: healthy elderly, moderate-risk middle-aged adults, and diabetics. In all three, the answer converged on the same disappointing conclusion. Modest reductions in cardiovascular events, when present at all, were offset by roughly equivalent increases in major bleeding.
The 2019 ACC/AHA guideline reversal
Major cardiology guideline bodies revise recommendations on a slow cadence, often waiting years for evidence to consolidate before changing direction. The aspirin reversal moved unusually fast.
The 2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease was published in March 2019, roughly six months after the ASPREE, ARRIVE, and ASCEND papers. The guideline downgraded aspirin from a routine recommendation to a much narrower indication.
The new language was that low-dose aspirin “may be considered” for primary prevention in select adults aged 40 to 70 at higher cardiovascular risk who were not at increased bleeding risk. For adults over 70, the guideline explicitly recommended against aspirin for primary prevention. For adults at increased bleeding risk regardless of age, aspirin was also not recommended. The change moved aspirin from a broad public-health intervention to a narrow case-by-case decision in a small minority of patients.
The U.S. Preventive Services Task Force followed in 2022 with its own pullback. The 2022 USPSTF Final Recommendation Statement narrowed its endorsement substantially, ultimately settling on language that aspirin’s net benefit for primary prevention is small at best in adults aged 40 to 59 at higher cardiovascular risk, and that initiation of aspirin for primary prevention in adults 60 and older has no net benefit. For most adults, in other words, USPSTF stopped recommending it.
Within a few years, the guidance that millions of healthy adults had built a daily habit around had been narrowed almost to the point of disappearance.
Why the old trials didn’t catch this
The interesting question is not “why did the new trials show no benefit?” but rather “why did the old trials show a benefit that didn’t reproduce?” The trials from the 1980s and 1990s were not poorly conducted. The Physicians’ Health Study, in particular, was a high-quality randomized trial. The answer, in the cardiovascular medicine community’s current consensus, has three parts.
Background cardiovascular risk fell substantially. Between roughly 1990 and 2015, the rate of first cardiovascular events in adults at any given level of measured risk fell substantially across most developed economies. Statins moved from a specialty drug to a routine prescription. Blood pressure control improved. Smoking rates declined. Cardiovascular event rates per thousand person-years at any given baseline risk profile were materially lower in 2015 than they had been in 1990. Aspirin’s relative risk reduction may have stayed roughly the same, but applied to a smaller baseline rate, the absolute benefit shrank.
Bleeding risk did not fall in parallel. Aspirin’s bleeding harm is mechanistically direct: it inhibits platelet aggregation, which makes any bleed worse. That harm did not benefit from the same background improvements that reduced cardiovascular risk. As a result, the ratio of bleeding harm to cardiovascular benefit shifted unfavorably over the decades, even though the underlying biology of aspirin had not changed.
The older trials were better at detecting heart attacks than at detecting bleeds. Endpoint adjudication standards have tightened considerably. Modern trials capture more bleeding events because they look harder, define endpoints more precisely, and follow patients in more structured ways. Older trials likely underestimated bleeding harm by today’s standards, which made the harm-benefit calculation look more favorable than it actually was even at the time.
The combination of these three effects means the older trials did not give a “wrong” answer at the time, exactly. They gave an answer that reflected the specific population, baseline risk environment, and measurement standards of their era. As all three drifted, the answer drifted with them. By the time ASPREE, ARRIVE, and ASCEND ran, the underlying conditions had changed enough that the older trials’ conclusion no longer applied.
This is a structurally important point that gets lost in lay coverage of medical reversals. The reversal is not necessarily a story of older trials being fraudulent or incompetent. It can be a story of trials answering a real question about their specific time and place, with answers that quietly become inapplicable as conditions change.
The medical-reversal pattern this fits into
The aspirin reversal is not isolated. The last two decades of cardiovascular and broader preventive medicine have produced a series of similar reversals, where decades-old guidance based on early trials gets revised or eliminated when contemporary evidence accumulates. The pattern is consistent enough to be worth naming explicitly.
Hormone replacement therapy was widely recommended for postmenopausal women through the 1990s on the basis of observational studies and small trials suggesting cardiovascular benefit. The Women’s Health Initiative trial reversed the guidance in 2002, showing increased rather than decreased cardiovascular risk in the population studied. Beta-carotene supplementation was widely recommended as a cancer-preventive antioxidant through the 1980s on the basis of observational data; the CARET trial in the 1990s found it actually increased lung cancer risk in smokers and was stopped early. The saturated fat hypothesis underlying decades of dietary guidance has been substantially revised as long-term randomized evidence has accumulated.
The reversals share structural features. Each began with early evidence that pointed in a particular direction, was translated into broad public health guidance, persisted for years or decades as the guidance hardened into common knowledge, and was eventually reversed by larger and better-designed contemporary trials. The reversals tended to come not from a single decisive trial but from the convergence of multiple trials within a short window, as happened in 2018 with aspirin.
In each case, the original guidance was supported by what looked, at the time, like reasonable evidence. The reversals were not driven by anyone discovering that the original researchers had been sloppy or fraudulent. They were driven by the discovery that early effect sizes were smaller than they appeared, did not generalize as far as believed, or operated against a baseline that had since changed.
What this means for evaluating prevention claims
For strategists, executives, and decision-makers who evaluate “preventive intervention” claims — whether the intervention is medical, regulatory, organizational, or product-level — the aspirin reversal carries several structural lessons.
Established guidance based on older trials is not the same as established truth. The aspirin recommendation was as institutionalized as preventive medicine guidance gets. AHA, USPSTF, and the equivalent bodies in most developed economies recommended it. Insurance plans covered it. Public health campaigns promoted it. None of that institutional weight prevented the underlying evidence from being thin enough to crumble when better-designed trials ran. Institutional adoption is not evidence. It is a signal of past consensus, which is a different thing.
Small absolute benefits do not survive changing baselines. If your prevention intervention shows a 20 to 30 percent relative risk reduction in trials run against a high-baseline-risk population, the absolute benefit may be material at trial time but trivial against a population whose baseline risk has fallen by half. The aspirin literature is a textbook case. The biology did not change. The relative effect did not change much. The baseline did, and the calculation flipped. Any preventive intervention whose case rests on modest relative risk reduction is vulnerable to this kind of erosion.
Harm-benefit ratios are time-dependent. Aspirin’s bleeding risk did not improve with the background improvements in cardiovascular medicine. When you evaluate a preventive intervention with both benefits and harms, you should explicitly ask whether the benefits are likely to scale down faster than the harms over time. If yes, the intervention’s net case may degrade silently while the guidance stays in place.
Reversals tend to converge. ASPREE, ARRIVE, and ASCEND published within roughly two weeks of one another. The Women’s Health Initiative results landed in a similarly compressed window. When evidence is shifting underneath an established recommendation, the shift often becomes visible all at once rather than gradually. If you wait for a single decisive trial, you may already be late. The earlier signal is usually that multiple groups have started running larger, contemporary trials on the same question — which itself indicates that the experts closest to the evidence are not as confident in the existing guidance as the public-facing recommendations imply.
Be skeptical of “the relative risk reduction was 30 percent” framings. Relative risk reductions are the dominant unit of preventive medicine marketing precisely because they sound bigger than absolute risk reductions do. Always translate to absolute terms before deciding whether an intervention is worth its harms and costs. Aspirin’s relative effect made it look impressive; its absolute effect, especially against modern baselines, did not.
The broader principle is that the aspirin story is not a story about aspirin. It is a story about the structure of evidence in preventive interventions. Old evidence is fragile in ways that institutional adoption can disguise. Multiple contemporary trials beat a single old trial, even when the old trial was high quality. And benefits and harms can move in different directions over time, silently reshaping the case for an intervention long after the original guidance has hardened.
Strategist takeaway
If you are responsible for any kind of preventive program — risk management, public health, product safety, organizational policy — the aspirin reversal suggests a few operating rules.
Treat older guidance as a hypothesis to be re-tested rather than a fact to be defended. When the conditions underlying the original evidence change substantially — baseline risk, measurement standards, the available alternatives — the guidance probably needs revisiting. Build a habit of asking, periodically, “if the original trials were run today in current conditions, would they still produce the same answer?”
Pay closer attention to absolute risk reductions than relative ones. The aspirin reversal’s mechanics were almost entirely in the gap between the two. A relative risk reduction that has not changed becomes an absolute risk reduction that has shrunk when the baseline drops, and the harm side of the ledger does not necessarily shrink with it.
Watch for clusters of contemporary trials investigating an established recommendation. When three large groups simultaneously decide that an established intervention deserves a modern, well-powered re-test, the academic and clinical community is signaling reduced confidence in the existing guidance, regardless of what public-facing recommendations still say.
Build harm-benefit reviews into the prevention program’s calendar. Aspirin’s downside did not appear suddenly. It was always there, just outweighed by a benefit that gradually faded. Periodic re-evaluation of the harm side, against current rather than historical benefit estimates, would have surfaced the reversal earlier.
The cost of getting this wrong is not abstract. Millions of adults took daily aspirin for years or decades on the strength of guidance that turned out to be net-neutral or net-negative for them. That is a real harm, distributed quietly across a population, and it accumulated because the institutions responsible for the guidance defaulted to inertia rather than periodic re-evaluation. The structural fix is to build the re-evaluation into the system rather than waiting for outside trials to force it.
Sources
- McNeil, J. J., Wolfe, R., Woods, R. L., et al. (2018). Effect of aspirin on cardiovascular events and bleeding in the healthy elderly. New England Journal of Medicine, 379(16), 1509-1518. DOI: 10.1056/NEJMoa1805819 (ASPREE)
- Gaziano, J. M., Brotons, C., Coppolecchia, R., et al. (2018). Use of aspirin to reduce risk of initial vascular events in patients at moderate risk of cardiovascular disease (ARRIVE): a randomised, double-blind, placebo-controlled trial. The Lancet, 392(10152), 1036-1046. DOI: 10.1016/S0140-6736(18)31924-X
- ASCEND Study Collaborative Group (2018). Effects of aspirin for primary prevention in persons with diabetes mellitus. New England Journal of Medicine, 379(16), 1529-1539. DOI: 10.1056/NEJMoa1804988
- Arnett, D. K., Blumenthal, R. S., Albert, M. A., et al. (2019). 2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease. Circulation, 140(11), e596-e646. DOI: 10.1161/CIR.0000000000000678
- U.S. Preventive Services Task Force (2022). Aspirin Use to Prevent Cardiovascular Disease: USPSTF Recommendation Statement. JAMA, 327(16), 1577-1584. DOI: 10.1001/jama.2022.4983
Related Articles
- Hormone Replacement Therapy: The Women’s Health Initiative Reversal
- Beta-Carotene And The CARET Trial: When Antioxidants Caused Cancer
- The Saturated Fat Diet-Heart Hypothesis: A Fifty-Year Reversal
- PREDIMED Mediterranean Diet Retraction And Republication
- SSRI Antidepressants And Publication Bias In Clinical Trials
Frequently Asked Questions
Should adults stop taking daily aspirin entirely?
Not necessarily, and the decision should be made with a physician. The 2018 reversal applies to primary prevention — adults who have not had a heart attack or stroke. For adults who have already had a cardiovascular event (secondary prevention), aspirin’s benefit remains well-supported and the guidance has not changed. The reversal also does not mean adults already taking aspirin should stop abruptly without consulting their doctor, since stopping can carry its own short-term risks. The reversal means the case for starting aspirin in healthy adults at average or moderate cardiovascular risk has collapsed.
Did the older trials lie or were they wrong?
Neither, in the simple sense. The older trials were high-quality and accurately measured what they measured in the populations they enrolled at the time they ran. The shift came from background cardiovascular risk falling substantially over the following decades, alongside tighter modern measurement of bleeding harms. The same relative effect applied to a smaller baseline produces a smaller absolute benefit, and the harm-benefit calculation flipped. This is a structural lesson about how older evidence ages, not an indictment of the original researchers.
Why did three trials publish almost simultaneously in 2018?
Each trial had been running for years independently. The timing was not coordinated in any formal sense — ASPREE, ARRIVE, and ASCEND were funded and managed by separate groups. But the fact that three large groups simultaneously decided to re-test aspirin for primary prevention was itself a signal that the academic community had become uncertain about the existing guidance. The trials launched roughly together, reached endpoint around the same time, and published within a few weeks of one another. The convergence amplified the reversal’s impact.
Does this mean we should stop trusting cardiovascular guidelines?
The opposite, actually. The aspirin reversal is an example of the guideline system working — eventually. New evidence accumulated, contemporary trials ran, and the major bodies revised their recommendations within roughly a year. The system has lag, and the lag matters, but it does correct. The stronger lesson is that any specific recommendation should be held with a probability rather than as a certainty, and the probability should shift when contemporary evidence shifts. Trusting “the guideline system” is reasonable; treating any specific recommendation as permanent truth is not.
Are there other long-standing medical recommendations likely to be reversed?
Probably yes, though predicting which ones is difficult. The general pattern — early trials in higher-risk populations producing modest effects that get translated into broad guidance, and then quietly eroding as baseline risk falls — applies to several other preventive interventions in current cardiovascular and metabolic medicine. The structural answer is to pay attention when you see multiple contemporary RCTs being launched on questions that “already have an answer.” That is usually the academic community’s leading indicator that the existing answer is weaker than it looks.
What is the practical takeaway for non-medical strategists?
Treat any preventive intervention claim — in any domain — with the same skepticism that the aspirin story now warrants. Ask what the absolute (not relative) effect size is. Ask how old the supporting evidence is and whether the conditions it was measured under still hold. Ask whether the harm side of the ledger has been re-evaluated recently, or whether it is being assumed to stay constant. If a recommendation has been institutionally stable for decades without being re-tested under current conditions, the probability that it is wrong is higher than the institutional weight suggests.