Virtually every adult who suffered a first heart attack, stroke, or heart failure episode had at least one detectable risk factor above optimal levels years before the event struck. That is the central finding of a study spanning more than 9.3 million people in South Korea and roughly 6,803 adults in the United States, published in the Journal of the American College of Cardiology in September 2025. The result, showing that more than 99 percent of cardiovascular events were preceded by a measurable warning sign, reframes heart disease not as a sudden catastrophe but as a slow accumulation of signals that clinicians and patients can act on earlier.
Why the 99 percent finding changes the prevention timeline
The study drew on two independent cohorts tracked for years. South Korea’s National Health Insurance Service cohort followed approximately 9,341,100 participants from 2009 through 2022. The U.S. arm relied on the Multi-Ethnic Study of Atherosclerosis, an NHLBI-sponsored cohort of roughly 6,803 adults aged 45 to 84 whose baseline exams ran from July 2000 to July 2002 and who were followed through 2019. In both populations, the prevalence of at least one nonoptimal traditional risk factor before a first cardiovascular event exceeded 99 percent.
The researchers defined “nonoptimal” at thresholds well below the levels that typically trigger medication. Blood pressure of 120/80 or higher, total cholesterol of 200 mg/dL or higher, fasting glucose of 100 mg/dL or higher, or current smoking each counted. These cutoffs sit below the clinical treatment lines most doctors use, which means a large share of people who eventually had events were living with risk factors that standard care would not yet flag as urgent. Philip Greenland of Northwestern University put it plainly: virtually everyone had a risk factor that could have been addressed.
That distinction between nonoptimal and clinically elevated matters because it determines when intervention begins. If the first three years of recorded nonoptimal values already capture the vast majority of future event victims, then waiting for readings to cross higher clinical thresholds, such as blood pressure of 140/90, cholesterol of 240, or glucose of 126, delays action on people already at risk. The study’s secondary analysis used those higher thresholds and still found overwhelming prevalence of at least one elevated factor, but the nonoptimal framing pushes the detection window earlier and wider.
Two cohorts, two continents, one consistent signal
The strength of this finding rests on its replication across strikingly different health systems and populations. The Korean cohort, drawn from a national insurance screening program with repeated biennial health checks, gave researchers access to serial measurements on millions of people over 13 years. The MESA cohort, though far smaller, offered deep phenotyping: its original protocol specified detailed sampling, imaging, and laboratory measurements across multiple U.S. ethnic groups over nearly two decades. That both datasets converged on the same result, more than 99 percent prevalence of at least one nonoptimal risk factor before a cardiovascular event, makes it hard to dismiss as an artifact of one country’s screening habits or population genetics.
The peer-reviewed paper focused on four traditional risk factors: blood pressure, total cholesterol, fasting glucose or diabetes status, and smoking. It did not incorporate newer biomarkers, genetic risk scores, or environmental exposures. An accompanying editorial by Neha J. Pagidipati of the Duke Clinical Research Institute framed the result as a call to reconsider how aggressively clinicians act on readings that fall between optimal and clinically elevated.
What the data cannot yet answer about early risk
Several gaps remain. The published findings report the prevalence of nonoptimal risk factors before events but do not release participant-level event counts or detailed risk-factor distributions for the MESA arm. The NHLBI data repository makes MESA data available to qualified researchers, but independent re-analysis testing whether the first three years of nonoptimal readings alone can predict later events with high specificity has not yet been published. That kind of temporal analysis would clarify whether early, modest elevations carry predictive power strong enough to justify shifting treatment guidelines, or whether the 99 percent figure simply reflects how common it is for middle-aged and older adults to have at least one reading above optimal.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
