← Longevity & Supplement Guides
Where the risk factor came from
On the morning of 12 April 1945, in a cottage at Warm Springs, Georgia, the most closely watched body in the world stopped working. Franklin Roosevelt was sitting for a portrait. He said his head hurt. Fifteen minutes later his cardiologist, Howard Bruenn, put a cuff on his arm and recorded a pressure above 300 over 190.
Bruenn had been taking that measurement for more than a year. In March 1944 he had found the president breathless and blue-lipped, with an enlarged heart and a pressure of 186 over 108. He started digitalis. He kept notes, and twenty-five years later he published them in the Annals of Internal Medicine, which is why we can be precise about a death that was, at the time, described to the public in far vaguer terms.
Here is the part worth sitting with. Nothing about Roosevelt’s numbers was hidden from medicine. The cuff had been in clinical use for decades. The readings were taken, written down, discussed by serious men. What did not exist — what nobody in 1945 had — was a framework that turned a number measured in a man who felt fine into a reason to do anything at all.
That framework has a name now. We call it a risk factor, and we live inside it so completely that it is hard to see as an invention.
What doctors believed, and why it was not stupid
For most of the first half of the twentieth century, a high pressure in an older person was widely read as essential — not in the sense of severe, but in the sense of necessary. The idea was that arteries stiffen with age, and the body raises the pressure to keep pushing blood through them. Take the pressure away and you might starve the brain and the kidneys of blood.
Paul Dudley White, then the most famous cardiologist in America, wrote in his 1931 textbook Heart Disease that treating hypertension was a difficult and nearly hopeless task given what was then known — and that, for all anyone knew, the pressure might be an important compensatory mechanism.
Read that in context and it is not obscurantism. It is an honest account of a profession with a good instrument and no safe tools. The era’s interventions ran to strict rice diets and surgically cutting sympathetic nerves. When your options are that blunt, caution about intervening is a defensible position, not a failure of nerve.
The quote everyone gets wrong
There is a sentence from 1931 that gets wheeled out whenever anyone wants to show how foolish the past was. It belongs to the Liverpool physician John Hay, and in the form that circulates it holds that the real danger of a high blood pressure is that somebody will find it and then try to do something about it. It is offered as proof that the profession of the day was against intervening at all.
The sentence is real. The version in circulation is not. Hay prefaced it — the passage begins “There is some truth in the saying that…” — which turns a flat declaration into a wry acknowledgement of a professional joke. Elias and Goodell, writing in the Journal of Clinical Hypertension in 2019, call this contextomy: quotation by amputation.
Read the rest of Hay’s lecture and he argues the opposite of his reputation. He wanted blood pressure measured in young adults at the first opportunity. He wanted family history taken seriously, exercise used as therapy, and raised readings investigated in every patient. White, for his part, abandoned the compensatory theory entirely; by 1971 he was writing about what sustained high pressure does to the heart over time.
The past was not stupid. It was under-equipped. Those are different conditions, and confusing them is how each generation flatters itself.
A president dies; his vice-president signs
In 1948, Harry Truman — who had been Roosevelt’s vice-president, and who had watched what an untreated pressure did to him — signed the National Heart Act. It created the National Heart Institute and seeded a twenty-year epidemiological study of heart disease with $500,000.
The study needed a town. Framingham, Massachusetts was chosen for reasons that are almost comically practical: local physicians were enthusiastic, the town had already sat through a tuberculosis study in the 1910s and 1920s and knew what being studied felt like, it had only two hospitals (later one), which made long-term follow-up tractable, and it was close enough to Boston for the consultants. Paul Dudley White — the same White — lobbied for it.
The design fell to a young Public Health Service physician, Gilcin Meadors, whose stated mission was to study heart disease in an ordinary, unselected population and find the factors that preceded it. He and a nurse, Nell McKeever, went door to door, largely to persuade people that the federal government did not want anything from them.
Timeline
From a number to a reason to act
1912
American life insurers publish mortality data linking build and blood pressure to early death.
1931
White’s textbook calls treatment nearly hopeless; Hay’s lecture urges measurement anyway.
1948
The National Heart Act is signed; the Framingham study enrols its first residents.
1961
Kannel and colleagues publish “Factors of Risk” — measurements in well people, tracked forward.
1967
The VA trial is stopped early: acting on the number changed what happened to people.
1998
The Framingham Risk Score turns the findings into an equation any clinician can run.
2015
SPRINT tests a lower target head to head, and finds benefit alongside real harms.
2024
The PREVENT equations replace the old score, add kidney measures, and drop race as an input.
Dates are publication or enactment dates from the sources listed at the foot of this article. The shaded band marks the interval between blood pressure being tied to mortality in insurance data and the first randomised evidence that acting on it changed outcomes.
The sample that was not quite a sample
Framingham is remembered as the study that watched an ordinary American town. The recruitment was messier than that, and the mess is instructive.
Investigators wrote to a random sample of two in every three local families with members aged 30 to 59. Of roughly 6,500 people contacted, about 4,500 agreed. Then some 700 residents who had not been sampled volunteered and were enrolled anyway, bringing the Original Cohort to 5,209 — 2,336 men and 2,873 women.
People who volunteer for a heart study are not a random draw from anywhere. Neither, for the purposes of the next seventy years, was the town: the cohort was overwhelmingly white and of European descent. This is not a retrospective accusation. It is in the authors’ own words — when Framingham published its risk score in 1998, the paper’s conclusion states that the equations predict risk in a middle-aged white population sample. Cohorts recruited to reflect the town’s actual diversity, Omni 1 and Omni 2, arrived in 1994 and 2003, forty-six and fifty-five years after the study began.
What got simplified away, in other words, was the question of whom the numbers were about. That question has taken most of a century to work its way back in.
1961: the numbers get a name
In July 1961, five authors — Kannel, Dawber, Kagan, Revotskie and Stokes — published six years of follow-up in the Annals of Internal Medicine under the title “Factors of Risk in the Development of Coronary Heart Disease.” What they reported is now so ordinary it barely registers: among people who were well at the outset, blood pressure, serum cholesterol and a particular pattern on the electrocardiogram predicted who would later develop coronary disease.
The finding mattered. The conceptual move mattered more. Before it, you were ill or you were not, and medicine was mostly the business of the former. After it, you could feel entirely well and still be carrying a measurement that said something about a future that had not happened yet.
That is a strange new category of person, and almost everyone reading this belongs to it. It is the same category the history of “normal” body temperature and the history of the calorie both end up in: a measurement invented for one purpose becoming a standing verdict on ordinary people.
Framingham did not invent the risk factor
Here is where the tidy version falls apart. Nearly every retelling — including, until recently, most of the medical literature — credits the 1961 paper with coining the term “risk factor.” A popular variant gives the credit to life insurers instead.
Both are wrong, and we know this because two historians, David Jones and Gerald Oppenheimer, went looking. Writing in Perspectives in Biology and Medicine in 2017, they searched digitised journals, textbooks and newspapers and found “risk factor” already in use in the late nineteenth and early twentieth centuries — in finance, in agriculture, in manufacturing. It appears in the occupational health literature in 1922. It resurfaces across the 1950s in psychiatry, surgery, cardiology and aerospace medicine. And despite the influence of the 1961 paper, the phrase did not take hold in medicine and public health until the mid-1970s.
So what did Framingham actually contribute? Not the phrase. The arithmetic. It took a loose commercial idea — some inputs make a bad outcome likelier — and made it prospective, quantitative and personal: named measurements, in named people, tracked over named years. A word borrowed from industry became a number you could be assigned.
Association is not permission
This is the step most retellings skip, and it is the one that does the real work.
Framingham was observational. It could say that people with higher pressures went on to have more heart disease. It could not say that bringing a pressure down would change anything — for all the data showed, the pressure could have been a marker of some deeper process that would proceed regardless.
The answer came from a trial. In December 1967 the Veterans Administration Cooperative Study Group, led by Edward Freis, published results in JAMA from 143 men with severe diastolic readings, randomly assigned to drug therapy or placebo. It was stopped after about eighteen months. Among the 70 men on placebo there were four cardiovascular deaths and none among the 73 treated; a further seventeen placebo patients developed complications serious enough to pull them out of the trial. A second report in 1970 extended the finding to milder pressures.
Nineteen years after Framingham began, the loop closed. That interval is the honest shape of this kind of knowledge, and it is worth holding on to the next time a fresh association arrives already packaged as a reason to buy something.
From a town in Massachusetts to the app on your phone
In 1998, Wilson and colleagues published the Framingham Risk Score in Circulation: 2,489 men and 2,856 women, aged 30 to 74, followed twelve years, distilled into sex-specific equations using age, diabetes, smoking, blood pressure category and cholesterol category. After adjustment, roughly 28 per cent of coronary events in men and 29 per cent in women were attributable to pressures above high normal.
That algorithm is the direct ancestor of every “your ten-year risk is X per cent” box you have ever been shown. It became the Pooled Cohort Equations, and in 2024 those were succeeded by the American Heart Association’s PREVENT equations, which estimate ten- and thirty-year risk, include heart failure, add kidney function and body-mass index — and remove race as an input.
That last change deserves a moment. Race was never a biological variable in those equations. It was a stand-in for everything the original cohorts had not measured and could not see. Taking it out is a correction to the arithmetic, not a discovery about bodies.
The definition moved, and tens of millions changed category
In 2017, American guidelines redefined high blood pressure downward, to readings at or above 130 over 80. Muntner and colleagues then ran both definitions against the same national survey data and published the result in Circulation in 2018.
One definition, two answers
Share of US adults classified as having hypertension
Bar lengths are the published percentages of US adults (Muntner et al., Circulation 2018, analysing NHANES 2011–2014). The same people, the same readings, two lines drawn in different places.
Three things this story is usually told wrong
“Framingham coined the risk factor.” It did not. The phrase was already circulating in finance, agriculture, manufacturing and occupational health decades earlier, and did not become common in medicine until the mid-1970s.
“Doctors used to refuse to treat high blood pressure.” The famous 1931 line is quoted with its opening clause removed. The same lecture argued for measuring young adults and investigating raised readings in every patient.
“Framingham studied a random sample of a town.” Around 700 of the 5,209 original participants were volunteers who had not been sampled, and the cohort was overwhelmingly white — as the study’s own 1998 risk-score paper states.
Under the older definition, 31.9 per cent of US adults had hypertension. Under the new one, 45.6 per cent did. The share for whom medication was recommended barely moved — 34.3 to 36.2 per cent. Almost the entire change was in who now gets told they have something.
Nobody’s arteries changed on the day of publication. A line did. That is the thing about a risk factor: it is a measurement plus a threshold, and the threshold is a decision made by people in a room.
SPRINT’s 120 is not your clinic’s 120
The best modern test of a target is SPRINT, published in the New England Journal of Medicine in 2015: 9,361 adults with elevated cardiovascular risk and no diabetes, randomised to a systolic target below 120 or below 140. It was stopped early, at a median of 3.26 years. The primary composite outcome ran at 1.65 per cent a year in the intensive arm against 2.19 per cent in the standard arm — a hazard ratio of 0.75 (95% CI 0.64–0.89). Death from any cause: 0.73 (0.60–0.90).
It also found more hypotension, more fainting, more electrolyte disturbance and more acute kidney injury in the intensive arm. Both halves of that sentence are the result.
And then there is the measurement itself. SPRINT’s readings were taken by a protocol almost nobody experiences: five minutes seated and quiet, then three automated readings, averaged. Whether the patient was left alone in the room turns out to matter less than early commentary assumed, and researchers are still arguing about it. What is not in dispute is that the readings were taken with unusual care. A 120 produced that way and a 120 taken in a corridor, mid-sentence, after a flight of stairs, are not the same quantity wearing the same label — a problem that recurs with every number a device hands you.
What this story is actually evidence for
Three things, and not more.
Risk factors are population instruments. They forecast groups well and individuals loosely. A ten-year risk of eight per cent is not a prediction about you; it is a statement about a thousand people who look like you on paper. The same limit shows up whenever a score is sold as a verdict, which is why biological-age tests deserve the same scepticism.
A single reading is mostly noise. The entire apparatus rests on repeated, standardised measurement in the same people across decades. What made Framingham work was not the cuff. It was coming back, and coming back again.
And none of this licenses a stronger claim about anything you can buy. An association is not an intervention trial, and a history lesson is not evidence. What eighty years actually established is narrower and more useful: that carefully measured numbers in well people can carry real information about the future, and that only a trial can tell you whether acting on a given number helps. Everything else — including which biomarkers are worth tracking at all — has to be argued case by case.
The bottom line
Pick one number you have only ever seen measured badly, and this week have it measured properly once: seated, quiet, five minutes, arm supported, more than one reading, averaged. Not to optimise it. To find out whether the figure you have been carrying around in your head is even the right figure. Anything clinical belongs in a conversation with your doctor — the point here is only that the number you argue about should be one that was taken with care.
Roosevelt’s physicians had the instrument and not the idea. We have the instrument, the idea, and a phone full of forecasts — and most of us still take the reading in a hurry, and then believe it.
