{"id":13002,"date":"2026-08-25T12:40:22","date_gmt":"2026-08-25T16:40:22","guid":{"rendered":"https:\/\/www.curingheartdisease.com\/?p=13002"},"modified":"2026-08-25T12:55:07","modified_gmt":"2026-08-25T16:55:07","slug":"the-history-of-cardiovascular-disease","status":"publish","type":"post","link":"https:\/\/www.curingheartdisease.com\/zh\/the-history-of-cardiovascular-disease\/","title":{"rendered":"\u5fc3\u8840\u7ba1\u75be\u75c5\u53f2"},"content":{"rendered":"

<\/a>The Hundred-Year Retreat of Coronary Heart Disease<\/h2>\n

<\/a>What Actually Happened, How Much Modern Cardiology Contributed, and What Should Happen Next<\/h4>\n

A narrative and systematic evidence review, c.\u00a01925\u20132026<\/em><\/p>\n

Peter Megdal, PhD August 24, 2026<\/p>\n

Table of Contents<\/p>\n

The Hundred-Year Retreat of Coronary Heart Disease<\/em>. 3<\/a><\/p>\n

What Actually Happened, How Much Modern Cardiology Contributed, and What Should Happen Next<\/em>\u00a0 3<\/a><\/p>\n

Plain-language executive summary<\/em>. 3<\/a><\/p>\n

The answer in one paragraph<\/em>.. 4<\/a><\/p>\n

Central quantitative conclusions<\/em>. 5<\/a><\/p>\n

Part I. The story: how a disease rose and fell<\/em> 7<\/a><\/p>\n

An epidemic that had to be discovered before it could be counted (c.\u00a01900\u20131949)<\/em>. 7<\/a><\/p>\n

The peak, and the turn (1950\u20131969)<\/em>. 7<\/a><\/p>\n

The steep descent (1970\u20131989)<\/em>. 8<\/a><\/p>\n

The pharmacological consolidation (1990\u20132009)<\/em>. 9<\/a><\/p>\n

The stall (2010\u20132026)<\/em>. 9<\/a><\/p>\n

Table 1 \u2014 The hundred-year timeline<\/em>. 10<\/a><\/p>\n

Figure 1 \u2014 U.S. age-adjusted coronary mortality trajectory<\/em>. 11<\/a><\/p>\n

Figure 2 \u2014 When the tools arrived<\/em>. 12<\/a><\/p>\n

Part II. Measuring the decline properly<\/em>. 13<\/a><\/p>\n

What the surveillance actually shows<\/em>. 14<\/a><\/p>\n

Part III. Why it happened: the attribution literature<\/em>. 15<\/a><\/p>\n

How the models work, and what they can and cannot tell you<\/em>.. 15<\/a><\/p>\n

The double-counting problem, and how it is handled<\/em>. 16<\/a><\/p>\n

Biological lag times: why attribution windows matter<\/em>. 16<\/a><\/p>\n

Table 2 \u2014 Major attribution studies<\/em>. 17<\/a><\/p>\n

A defensible synthesis<\/em>. 19<\/a><\/p>\n

Figure 3 \u2014 A concrete decomposition (U.S., 1980\u20132000)<\/em>. 20<\/a><\/p>\n

Which risk factors mattered most?<\/em>. 20<\/a><\/p>\n

Table 3 \u2014 Contribution of individual risk factors<\/em>. 20<\/a><\/p>\n

Part IV. Prevention: what the randomized evidence actually proves<\/em>. 23<\/a><\/p>\n

Blood-pressure lowering<\/em>. 23<\/a><\/p>\n

LDL and apolipoprotein B lowering<\/em>. 23<\/a><\/p>\n

Antiplatelet therapy and the secondary-prevention bundle<\/em>. 24<\/a><\/p>\n

Table 4 \u2014 Major medical advances<\/em>. 25<\/a><\/p>\n

Part V. Rescue: how a heart attack stopped being a death sentence<\/em>. 29<\/a><\/p>\n

The equation that separates prevention from rescue<\/em>. 29<\/a><\/p>\n

Table 5 \u2014 Acute MI survival across eras<\/em>. 30<\/a><\/p>\n

The interventions, in causal perspective<\/em>. 31<\/a><\/p>\n

Japan: the decisive counterexample<\/em>. 32<\/a><\/p>\n

Part VI. Revascularization: three different operations wearing one name<\/em>. 32<\/a><\/p>\n

Table 6 \u2014 Prevention versus rescue<\/em>. 33<\/a><\/p>\n

Part VII. Counterfactuals and lives saved<\/em>. 35<\/a><\/p>\n

What if a modern population had old medical care \u2014 or old risk factors?<\/em>. 35<\/a><\/p>\n

How many lives were saved?<\/em>. 36<\/a><\/p>\n

Part VIII. International natural experiments<\/em>. 36<\/a><\/p>\n

Finland and North Karelia<\/em>. 37<\/a><\/p>\n

Iceland<\/em>. 37<\/a><\/p>\n

Sweden<\/em>.. 37<\/a><\/p>\n

England and Wales<\/em>. 37<\/a><\/p>\n

United States<\/em>. 38<\/a><\/p>\n

Poland: an accidental national experiment in the food supply<\/em>. 38<\/a><\/p>\n

Russia and the post-Soviet surge: the experiment run in reverse<\/em>. 38<\/a><\/p>\n

The Netherlands<\/em>. 38<\/a><\/p>\n

Japan<\/em>.. 39<\/a><\/p>\n

Global divergence<\/em>. 39<\/a><\/p>\n

Table 7 \u2014 Natural experiments in cardiovascular epidemiology<\/em>. 39<\/a><\/p>\n

Part IX. The stall, and who has been left behind<\/em>. 40<\/a><\/p>\n

Part X. What comes next: doing better than we are doing today<\/em>. 41<\/a><\/p>\n

    \n
  1. Treat cumulative lifetime exposure, not middle-aged thresholds<\/em>. 41<\/a><\/li>\n
  2. Finally treat lipoprotein(a)<\/em>. 42<\/a><\/li>\n
  3. Extend the lipid-lowering toolkit and solve adherence<\/em>. 42<\/a><\/li>\n
  4. Treat the cardiometabolic driver directly<\/em>. 42<\/a><\/li>\n
  5. Address residual inflammatory risk<\/em>. 43<\/a><\/li>\n
  6. Find the people at risk before the event<\/em>. 43<\/a><\/li>\n
  7. Population policy retains exceptionally large leverage<\/em>. 43<\/a><\/li>\n<\/ol>\n

    And above all: close the implementation gap<\/em>. 44<\/a><\/p>\n

    Table 8 \u2014 The forward agenda<\/em>. 44<\/a><\/p>\n

    Evidence hierarchy and the evidence ledger<\/em>. 45<\/a><\/p>\n

    Evidence ledger<\/em>. 46<\/a><\/p>\n

    Structured summaries where abstract-level evidence was used<\/em>. 50<\/a><\/p>\n

    Evidence gaps and remaining uncertainty<\/em>. 51<\/a><\/p>\n

    Confidence grades for the main conclusions<\/em>. 52<\/a><\/p>\n

    Central synthesis<\/em>. 53<\/a><\/p>\n

    The bottom line<\/em>. 54<\/a><\/p>\n

    References<\/em>. 56<\/a><\/p>\n

    <\/a>Plain-language executive summary<\/h4>\n

    For most of medical history, nobody was diagnosed as dying of a heart attack \u2014 not because myocardial infarction did not occur, but because it had not yet been recognized as a clinical syndrome. Coronary thrombosis was described as a clinical syndrome in the living patient only in 1912.[1] Within four decades it had become the leading cause of death in the industrialized world. Within another four decades it was in headlong retreat. This is one of the most extraordinary reversals in the history of medicine, and it is routinely explained badly.<\/p>\n

    The popular version \u2014 that cardiology invented statins and stents and thereby conquered heart disease \u2014 is wrong on chronology alone. Coronary mortality in the United States peaked around 1968 and had already fallen by roughly a third before the first statin was approved, before ACE inhibitors, before primary angioplasty, and before anyone had run a randomized trial of aspirin in acute myocardial infarction. The opposite version \u2014 that the decline was purely a matter of people smoking less and eating better, with medicine along for the ride \u2014 is equally wrong. Modern therapeutics account for a very large minority, and in some countries and eras close to half, of the fall.<\/p>\n

    \"\"<\/p>\n

    The accurate version is that two different things happened at once, and they should never be confused with one another:<\/p>\n

    First, fewer people had heart attacks.<\/strong> Populations reduced their lifetime exposure to three of the dominant modifiable drivers of atherosclerosis: tobacco smoke, elevated blood pressure, and atherogenic cholesterol. These are not the only causal contributors \u2014 diabetes, adiposity, and metabolic dysfunction matter too, and their trends have moved in the wrong direction. Some of that came from individual behavior, much of it came from policy and the food supply, and an increasing share over time came from drugs \u2014 antihypertensives from the 1970s, statins from the mid-1990s.<\/p>\n

    Second, the people who did have heart attacks stopped dying from them.<\/strong> In 1960, being admitted to hospital with an acute myocardial infarction carried roughly a one-in-three to one-in-five chance of not leaving. Contemporary in-hospital mortality after ST-elevation infarction in high-income systems is often around 4\u20137%, though the figure varies substantially with case mix and is not a universal all-AMI value. Coronary care units, defibrillation, cardiopulmonary resuscitation, emergency medical systems, aspirin, thrombolysis, and primary percutaneous coronary intervention did this. These acute-rescue interventions primarily improve survival after an event rather than prevent first myocardial infarctions.<\/p>\n

    Both achievements lower the coronary death rate, and the death rate alone cannot tell you which one is responsible. The best international data suggest that during the classic decline decades roughly two-thirds of the fall in coronary mortality came from having fewer coronary events and roughly one-third from surviving them.<\/p>\n

    Then there is the part of the story that is not yet written into the textbooks. The decline has stalled.<\/strong> In the United States and the United Kingdom, progress slowed sharply after about 2011, and among adults under 55 some acute outcomes have begun to move the wrong way. Obesity and type 2 diabetes have been subtracting from the ledger for thirty years; national blood-pressure control has deteriorated; and the easy wins from tobacco control have largely been banked. The next hundred years of progress will not come from repeating the last hundred. It is likely to depend increasingly on treating cumulative lifetime atherogenic exposure rather than middle-aged risk thresholds, on lipoprotein(a) if outcome trials confirm benefit, on the cardiometabolic revolution now underway with incretin therapies, on residual inflammatory risk, and above all on closing the vast implementation gap between what we know works and what patients actually receive.<\/p>\n

    <\/a>The answer in one paragraph<\/h4>\n

    Coronary heart disease rose through the first half of the twentieth century, peaked in most high-income countries in the 1960s, and then underwent one of the largest sustained mortality declines documented for a major cause of death. In the United States, age-adjusted ischemic heart disease mortality fell from 693 to 135 deaths per 100,000 between 1970 and 2022 \u2014 an 81% reduction<\/strong> \u2014 while mortality coded specifically to acute myocardial infarction fell from 354 to 40 per 100,000, an 89% reduction<\/strong> [2]. The best available attribution models do not support crediting that achievement to \u201cmedicine\u201d alone or to \u201clifestyle\u201d alone. Across the United States and several European countries during the principal decline eras, roughly 44\u201373% of the fall was attributed to favorable population risk-factor change<\/strong>, while approximately 23\u201347% was attributed to medical and surgical treatment<\/strong>; acute coronary treatment itself usually accounted for only about 4\u201312% of the overall mortality decline<\/strong> in IMPACT-type models [3,4,5,6,7,8]. In parallel, WHO MONICA data indicate that roughly two-thirds of the decline in coronary mortality in the 1980s\u20131990s came from fewer coronary events and one-third from better survival after an event<\/strong>, with substantial variation between populations [9].<\/p>\n

    The practical conclusion: much of the historical reduction in coronary-event incidence appears to have resulted from changes in the population\u2019s underlying exposure to atherosclerotic risk \u2014 especially smoking, blood pressure, and cholesterol \u2014 while modern cardiology made a very large additional contribution by medically lowering those same risks, preventing recurrent events, and transforming a heart attack from an event with roughly 20\u201330% short-term mortality into one that is, in contemporary high-income healthcare systems, far more survivable.<\/strong><\/p>\n

    <\/a>Central quantitative conclusions<\/h4>\n\n\n\n\n\n\n\n\n\n\n\n\n
    Question<\/td>\nBest defensible synthesis<\/td>\nConfidence<\/td>\n<\/tr>\n<\/thead>\n
    How much has U.S. CHD\/IHD mortality fallen?<\/strong><\/td>\nAge-adjusted IHD mortality fell 81% from 1970\u20132022<\/strong>; acute-MI\u2013coded mortality fell 89%<\/strong>. Over the best-documented single interval, age-adjusted CHD mortality fell roughly 50% between 1980 and 2000<\/strong>. Earlier historical series show heart-disease mortality rising into a mid-century peak before the sustained decline.<\/td>\nHigh<\/strong><\/td>\n<\/tr>\n
    How much has MI incidence fallen?<\/strong><\/td>\nNo valid universal century-long figure exists. Framingham ECG-defined first MI fell about 50% from 1960\u20131999<\/strong>; Icelandic MI incidence fell 66% from 1981\u20132006<\/strong>; ARIC showed annual declines of roughly 3\u20134%<\/strong> in most sex\/race groups during 1987\u20132008. Japan demonstrates that incidence need not fall at all: Miyagi rose from 7.4 to 27.0 per 100,000<\/strong> between 1979 and 2008.<\/td>\nModerate<\/strong><\/td>\n<\/tr>\n
    How much has MI case fatality fallen?<\/strong><\/td>\nFramingham age- and sex-adjusted 30-day, 1-year, and 5-year case fatality each fell about 60% from 1960\u20131999<\/strong>; Miyagi in-hospital mortality fell from 20.0% to 7.8%<\/strong> across 1979\u20132008. Pre-coronary-care-unit hospital mortality of roughly 30%<\/strong> compares with contemporary in-hospital STEMI mortality of roughly 4\u20137%<\/strong>.<\/td>\nHigh<\/strong> for direction; Moderate<\/strong> for cross-era magnitude<\/td>\n<\/tr>\n
    Population risk-factor contribution<\/strong><\/td>\nGenerally ~45\u201370%<\/strong>, with individual IMPACT estimates ranging from 44% (U.S.)<\/strong> to 73% (Iceland)<\/strong>.<\/td>\nModerate<\/strong><\/td>\n<\/tr>\n
    All medical\/surgical treatment contribution<\/strong><\/td>\nGenerally ~25\u201345%<\/strong>, ranging approximately 23\u201347%<\/strong> in major IMPACT analyses.<\/td>\nModerate<\/strong><\/td>\n<\/tr>\n
    Acute\/emergency treatment contribution<\/strong><\/td>\nUsually ~5\u201310% of the total CHD mortality decline<\/strong> in attribution models (approximately 4\u201312% across countries). But improvements in case fatality \u2014 to which improvements in acute systems of care contributed substantially \u2014 account for roughly one-third of the fall in coronary mortality<\/strong> in MONICA-type decomposition.<\/td>\nModerate<\/strong><\/td>\n<\/tr>\n
    Other\/diagnostic\/health-system contribution<\/strong><\/td>\nNot independently identifiable. IMPACT residuals are commonly ~5\u201310%<\/strong>, but this residual must not<\/strong> be equated with EMS, diagnosis, or health-system improvement, because many such effects are already embedded within the treatment estimates.<\/td>\nLow<\/strong><\/td>\n<\/tr>\n
    Has progress continued?<\/strong><\/td>\nNo.<\/strong> The decline slowed markedly after ~2011 in the U.S. and U.K. and has stagnated in younger adults, with recent evidence of worsening in-hospital mortality after a first STEMI in adults aged 18\u201354. Obesity, diabetes, and hypertension control remain important population-level counterweights.<\/td>\nModerate\u2013High<\/strong><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    A critical terminology point governs everything that follows: the \u201crisk-factor\u201d category is not a synonym for personal lifestyle choice.<\/strong> Falling smoking prevalence reflects taxation, smoke-free legislation, advertising restriction, physician counseling, and individual decision. Falling population cholesterol reflects food-supply reformulation, dietary change, and statins. Falling blood pressure reflects both secular population shifts and antihypertensive prescribing. The strongest analyses explicitly separate these pathways rather than double-count them. In the Swedish IMPACT work, for example, a 0.64 mmol\/L population cholesterol decline was decomposed into approximately 5,210 deaths prevented or postponed through dietary change and 810 through statins<\/strong>, while a 2.6 mmHg fall in systolic pressure was divided into roughly 900 deaths from secular change and 575 from antihypertensive treatment<\/strong> [10].<\/p>\n

    <\/a>Part I. The story: how a disease rose and fell<\/h3>\n

    <\/a>An epidemic that had to be discovered before it could be counted (c.\u00a01900\u20131949)<\/h4>\n

    The first difficulty in writing a hundred-year history of myocardial infarction is that for the first quarter of it, the diagnosis barely existed. James Herrick\u2019s 1912 description of coronary thrombosis in the living patient \u2014 largely ignored for a decade \u2014 established that a blocked coronary artery was survivable and therefore diagnosable rather than simply a post-mortem finding.[1] The electrocardiograph made it identifiable at the bedside. Only then could anyone begin counting.<\/p>\n

    What we can say with confidence is that heart disease became the leading cause of death in the United States in 1921<\/strong>, and that coronary mortality continued to rise through the 1930s, 1940s, and 1950s. What we cannot say is how much of that apparent rise was a genuine increase in atherosclerotic events and how much was improved recognition, better death certification, and the simple fact that populations were surviving long enough \u2014 past infectious disease, past tuberculosis, past childbirth \u2014 to develop coronary disease at all. Almost certainly it was both, with mass cigarette adoption, adverse blood-pressure trends, and a changing food supply likely contributing substantially \u2014 though the causal partition for this early period is intrinsically uncertain.<\/p>\n

    Effective evidence-based coronary therapy was extremely limited. The standard treatment for acute infarction was six weeks of strict bed rest, on the theory that the damaged myocardium needed to be spared work. There was no monitoring, no way to treat the arrhythmia that killed a substantial share of patients in the first hours, no way to reopen the artery, and no validated modern risk-factor framework for coronary prevention \u2014 diet, exercise, tobacco, and blood pressure had all been discussed, but prospective risk-factor science did not yet exist. A patient who reached hospital alive had perhaps a two-in-three to four-in-five chance of leaving it.<\/p>\n

    <\/a>The peak, and the turn (1950\u20131969)<\/h4>\n

    Two things had to happen before the epidemic could be reversed, and both happened in this era.<\/p>\n

    The risk factors were discovered.<\/strong> The Framingham Heart Study, begun in 1948, gave the world the very concept of a \u201crisk factor\u201d and identified blood pressure, cholesterol, and smoking as the principal modifiable ones. Ancel Keys\u2019s Seven Countries Study linked population dietary patterns to serum cholesterol to coronary incidence.[11] The 1964 U.S. Surgeon General\u2019s report made tobacco\u2019s causal role official policy rather than academic opinion. None of this treated a single patient. All of it made everything downstream possible.<\/p>\n

    Rescue was invented.<\/strong> The coronary care unit \u2014 Desmond Julian\u2019s proposal, realized in units in Kansas City, Toronto, Philadelphia, and Sydney in the early 1960s \u2014 was one of the most consequential innovations in the acute treatment of myocardial infarction, and it involved no drug at all. Its insight was simple: a large fraction of early MI deaths were caused by ventricular fibrillation, ventricular fibrillation is instantly reversible if you are standing there with a defibrillator, and therefore patients should be watched continuously by people trained to shock them. Closed-chest cardiac massage was described in 1960; external defibrillation had been demonstrated in the preceding decade. Historical series suggest hospital mortality for acute MI fell from roughly 30% toward 15%<\/strong> with the arrival of coronary care.[12]<\/p>\n

    The evidence base for this is weaker than its importance would suggest \u2014 most of it is before-and-after observational work, and a famous strand of British research argued that home care was no worse for selected patients. A prospective randomized comparison of coronary care unit versus general ward admission found overall mortality of 15.3% (17\/111) versus 29.3% (27\/92)<\/strong>, with the risk of dying on the general ward relative to the unit of 2.3 (95% CI 1.1\u20134.8)<\/strong>, and sudden death more frequent on the general ward.[13] That trial is small, single-centre, and drawn from a care context unlike today\u2019s, and the wider evidence base is largely observational;[14] but the temporal signal across many institutions is consistent and large.<\/p>\n

    Meanwhile, effective antihypertensive therapy arrived. The Veterans Administration Cooperative Studies (1967, 1970) demonstrated that treating severe and then moderate hypertension prevented strokes, heart failure, and death \u2014 landmark randomized evidence that pharmacologically modifying a cardiovascular risk factor changed clinical outcomes.[15]<\/p>\n

    U.S. coronary mortality peaked around 1968<\/strong> and began to fall.<\/p>\n

    <\/a>The steep descent (1970\u20131989)<\/h4>\n

    This is the era that most decisively refutes the \u201cmodern cardiology did it\u201d narrative, because for most of it modern cardiology did not yet exist. There were no statins until 1987 and no statin outcome trial until 1994. There was no ACE inhibitor trial in heart failure until 1987 or post-infarction until 1992. Balloon angioplasty was performed for the first time in 1977 and remained a niche procedure for a decade. Aspirin was not proven in acute MI until 1988.<\/p>\n

    The decline was noticed before it was explained. The 1978 Bethesda Conference, convened by the National Heart, Lung, and Blood Institute, was called specifically to document and account for an unexpected downturn in U.S. coronary and stroke mortality that nobody had predicted and nobody could yet attribute. That conference is the origin point of the entire attribution literature reviewed below.[16]<\/p>\n

    And yet U.S. age-adjusted ischemic heart disease mortality, 693 per 100,000 in 1970<\/strong>, fell steeply and continuously throughout the period. What changed instead was exposure. Male smoking prevalence fell substantially in the wake of the Surgeon General\u2019s report and the subsequent advertising restrictions. Serum cholesterol fell as the food supply shifted. The National High Blood Pressure Education Program (1972) drove detection and treatment of hypertension into routine primary care. And in North Karelia, Finland \u2014 then recording among the highest coronary mortality rates documented anywhere, particularly among working-age men \u2014 an explicitly community-based prevention program began in 1972 and demonstrated that a population\u2019s risk-factor profile could be deliberately changed.<\/p>\n

    Toward the end of the era the therapeutic revolution finally arrived, and it arrived in the emergency department. GISSI (1986)[17] and then ISIS-2 (1988)[18] established that streptokinase and aspirin each independently reduced death in acute myocardial infarction and that together they reduced it dramatically. The randomized trials of coronary artery bypass grafting conducted between 1972 and 1984 established that survival benefit was concentrated in left main and extensive multivessel disease, with little demonstrable survival benefit in less extensive disease \u2014 though benefit also varied with left ventricular function and anatomy.[19]<\/p>\n

    <\/a>The pharmacological consolidation (1990\u20132009)<\/h4>\n

    If the 1970s and 1980s belonged to population prevention and acute rescue, this era belonged to chronic pharmacotherapy and to systems of care.<\/p>\n

    The 4S trial (1994) showed that simvastatin reduced mortality in patients with coronary disease, and within a decade statins had become among the most prescribed drugs on earth.[20] ACE inhibitors moved from heart failure into post-infarction care. Beta-blockers, aspirin, and lipid lowering consolidated into a standard secondary-prevention bundle. Coronary stents reduced acute recoil and abrupt closure and substantially reduced restenosis compared with balloon angioplasty \u2014 though restenosis was reduced, not eliminated, and drug-eluting stents later reduced it further. Dual antiplatelet therapy substantially reduced thrombotic complications and helped make routine stenting practical. Primary PCI progressively displaced thrombolysis for ST-elevation infarction[21], and organized STEMI networks \u2014 prehospital ECG, catheterization laboratory activation from the ambulance, measured door-to-balloon times \u2014 made timely reperfusion increasingly a systems-and-logistics challenge that health systems could optimize.<\/p>\n

    Both incidence and case fatality fell measurably during this period, and this is the era in which quantitative attribution is strongest, because treatment uptake was documented, trial efficacy was known, and risk-factor surveillance was systematic. It is also the era in which the counterweights became visible: obesity and type 2 diabetes began subtracting from the ledger, and many major attribution models from this period recorded them as negative contributions.<\/p>\n

    <\/a>The stall (2010\u20132026)<\/h4>\n

    Mortality remains far below its historical peak \u2014 U.S. ischemic heart disease mortality reached 135 per 100,000 in 2022<\/strong> \u2014 but the trajectory has changed character. In the United States and the United Kingdom, the rate of decline slowed markedly after roughly 2011. In adults under 55, and particularly women, the long-term decline has slowed or stagnated, with some studies reporting adverse trends in specific coronary outcomes.[22] The absolute burden remains enormous: the United States recorded 915,973 cardiovascular deaths in 2023<\/strong>, an age-adjusted CVD death rate of 218.3 per 100,000<\/strong>[23], and globally GBD 2023 estimated 19.2 million cardiovascular deaths<\/strong>[24], because population growth and aging can drive absolute deaths up even while individual age-adjusted risk falls.<\/p>\n

    The candidate explanations for the stall are unflattering and largely non-technological: obesity and diabetes at levels no previous generation faced; the exhaustion of easy tobacco gains in populations where smoking is already uncommon; and, in the United States, a documented deterioration in hypertension control from 53.8% (2013\u20132014) to 43.7% (2017\u20132018)<\/strong>[25]. The COVID-19 pandemic then produced an abrupt reversal \u2014 the U.S. age-adjusted CVD death rate was 330.8 in 2019, 346.0 in 2020, and 359.6 in 2021 per 100,000 among adults aged 25 and over \u2014 erasing several years of progress in two. This series covers adults aged 25 and over and is not directly comparable with the all-adult, differently standardized rate of 218.3 per 100,000 cited above; the two must not be read as a single trend.<\/strong><\/p>\n

    <\/a>Table 1 \u2014 The hundred-year timeline<\/h4>\n\n\n\n\n\n\n\n\n\n
    Era<\/td>\nCHD\/MI epidemiology<\/td>\nPopulation and risk-factor developments<\/td>\nMedical and emergency developments<\/td>\nEvidential interpretation<\/td>\n<\/tr>\n<\/thead>\n
    1925\u20131949<\/strong><\/td>\nCHD rising; comparable population MI-incidence surveillance largely unavailable. Heart disease became the leading U.S. cause of death by 1921.<\/td>\nCigarette consumption and other adverse coronary exposures rose steeply; risk-factor theory did not yet exist.<\/td>\nECG increasingly used clinically; treatment of MI was bed rest and supportive care.<\/td>\nTier 3\/5.<\/strong> Mortality data are broadly reliable; causal attribution is far weaker than after 1950.<\/td>\n<\/tr>\n
    1950\u20131969<\/strong><\/td>\nCHD mortality remained high and rose to its U.S. peak around 1968.<\/td>\nFramingham and Seven Countries identified smoking, blood pressure, and cholesterol as modifiable risks; 1964 Surgeon General\u2019s report.<\/td>\nExternal defibrillation, CPR, coronary care units, effective hypertension therapy, first CABG. CCUs associated with hospital MI mortality falling from ~30% toward ~15%.<\/td>\nTier 2\u20135.<\/strong> The early survival change is compelling but largely observational.<\/td>\n<\/tr>\n
    1970\u20131989<\/strong><\/td>\nSustained steep decline begins; U.S. age-adjusted IHD mortality was 693\/100,000 in 1970<\/strong>.<\/td>\nSmoking falls substantially; population BP and cholesterol improve; North Karelia launches intensive community prevention.<\/td>\nWidespread antihypertensive treatment, CABG, PTCA (1977), beta-blockers, aspirin, cardiac rehabilitation, thrombolysis; first statins late in the era. ISIS-2 establishes aspirin and streptokinase.<\/td>\nTier 1\u20134.<\/strong> Population prevention and evidence-based cardiology begin contributing simultaneously.<\/td>\n<\/tr>\n
    1990\u20132009<\/strong><\/td>\nLarge falls in both MI incidence and case fatality become measurable in ARIC and Framingham.<\/td>\nFurther smoking, cholesterol, and BP improvement, increasingly offset by obesity and diabetes.<\/td>\nStatins, ACE inhibitors, stents, dual antiplatelet therapy, modern EMS, primary PCI mainstream. Primary PCI outperforms thrombolysis in randomized synthesis.<\/td>\nTier 1\u20134.<\/strong> The strongest era for quantitative attribution.<\/td>\n<\/tr>\n
    2010\u20132026<\/strong><\/td>\nMortality far below peak but decline stalls; stagnation in younger adults, with recent worsening in hospitalized first-STEMI mortality at ages 18\u201354<\/strong>. U.S. IHD mortality 135\/100,000 in 2022<\/strong>.<\/td>\nTobacco continues downward in high-income settings; obesity, diabetes, and metabolic risk counteract gains; U.S. hypertension control deteriorates.<\/td>\nIntensive statins, ezetimibe, PCSK9 inhibition, optimized ACS systems, radial PCI, high-sensitivity troponin, GLP-1 receptor agonists, colchicine.<\/td>\nTier 1\u20133.<\/strong> Strong evidence for individual treatment efficacy; much weaker evidence partitioning recent<\/em> national trends.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    <\/a>Figure 1 \u2014 U.S. age-adjusted coronary mortality trajectory<\/h4>\n

    Observed national mortality data. Historical and modern series are deliberately not numerically spliced, because age standards and disease coding changed.<\/strong><\/p>\n

    Panel A \u2014 Key observed data points<\/strong><\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
    Series<\/td>\nYear<\/td>\nAge-adjusted rate per 100,000<\/td>\nChange<\/td>\n<\/tr>\n<\/thead>\n
    Diseases of the heart (1940 standard)<\/td>\n1950<\/td>\n307.4<\/td>\n\u2014<\/td>\n<\/tr>\n
    Diseases of the heart (1940 standard)<\/td>\n1996<\/td>\n134.6<\/td>\n\u221256%<\/td>\n<\/tr>\n
    Ischemic heart disease (modern series)<\/td>\n1970<\/td>\n693<\/td>\n\u2014<\/td>\n<\/tr>\n
    Ischemic heart disease (modern series)<\/td>\n2022<\/td>\n135<\/td>\n\u221281%<\/strong><\/td>\n<\/tr>\n
    Acute myocardial infarction<\/td>\n1970<\/td>\n354<\/td>\n\u2014<\/td>\n<\/tr>\n
    Acute myocardial infarction<\/td>\n2022<\/td>\n40<\/td>\n\u221289%<\/strong><\/td>\n<\/tr>\n
    All CVD, adults 25+<\/td>\n2019<\/td>\n330.8<\/td>\n\u2014<\/td>\n<\/tr>\n
    All CVD, adults 25+<\/td>\n2020<\/td>\n346.0<\/td>\n+4.6%<\/td>\n<\/tr>\n
    All CVD, adults 25+<\/td>\n2021<\/td>\n359.6<\/td>\n+8.7% vs 2019<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    Panel B \u2014 Shape of the trajectory<\/strong><\/p>\n

    Heart disease becomes leading US cause of death ………. 1921
    \nMortality rises through ……………………………. 1930s-1960s
    \nPEAK (coronary heart disease) ………………………. ~1968
    \nSteep sustained decline ……………………………. 1968-2010
    \nDecline slows\/stagnates in younger adults;
    \nselected acute outcomes worsen …………………….. ~2011 onward
    \nPandemic interruption ……………………………… 2020-2021<\/p>\n

    Ischemic heart disease, age-adjusted, per 100,000:
    \n1970\u00a0 693\u00a0 ####################################
    \n2022\u00a0 135\u00a0 #######<\/p>\n

    Acute myocardial infarction, age-adjusted, per 100,000:
    \n1970\u00a0 354\u00a0 ##################
    \n2022\u00a0\u00a0 40\u00a0 ##<\/p>\n

    Sources: historical U.S. national mortality analyses (CDC\/NCHS)[26]; King et al.\u00a02025[2] for the 1970\u20132022 IHD and AMI series.<\/em><\/p>\n

    <\/a>Figure 2 \u2014 When the tools arrived<\/h4>\n

    Approximate clinical adoption timeline. Dates indicate emergence or the arrival of major evidence, not a single universal start date.<\/strong><\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
    Period<\/td>\nPrevention and chronic therapy<\/td>\nAcute rescue and systems<\/td>\n<\/tr>\n<\/thead>\n
    1950s<\/strong><\/td>\nEffective antihypertensive drugs emerging<\/td>\nExternal defibrillation<\/td>\n<\/tr>\n
    Early 1960s<\/strong><\/td>\nRisk-factor concept established (Framingham)<\/td>\nCPR; coronary care units and continuous rhythm monitoring<\/td>\n<\/tr>\n
    Late 1960s<\/strong><\/td>\n1964 Surgeon General\u2019s report on smoking<\/td>\nCABG enters clinical practice<\/td>\n<\/tr>\n
    1970s<\/strong><\/td>\nBeta-blockers; National High Blood Pressure Education Program; North Karelia<\/td>\nBalloon coronary angioplasty (1977); organized EMS expands<\/td>\n<\/tr>\n
    1980s<\/strong><\/td>\nStatins introduced (1987); cardiac rehabilitation formalized<\/td>\nAspirin and thrombolysis established in acute MI (GISSI 1986, ISIS-2 1988)<\/td>\n<\/tr>\n
    1990s<\/strong><\/td>\nStatin outcome trials (4S 1994); ACE-inhibitor-based secondary prevention<\/td>\nCoronary stents; primary PCI increasingly replaces thrombolysis<\/td>\n<\/tr>\n
    2000s<\/strong><\/td>\nIntensive statin therapy; polypill concept<\/td>\nPublic-access AED programs; modern STEMI networks; prehospital ECG<\/td>\n<\/tr>\n
    2010s<\/strong><\/td>\nEzetimibe outcomes (2015); PCSK9 inhibitors (2017\u201318); SGLT2 inhibitors<\/td>\nHigh-sensitivity troponin widespread; radial access standard<\/td>\n<\/tr>\n
    2020s<\/strong><\/td>\nVery-low-LDL secondary prevention; colchicine; GLP-1 receptor agonists (SELECT 2023); inclisiran; bempedoic acid<\/td>\nMature integrated ACS\/EMS systems; Lp(a)-lowering agents in outcome trials<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    <\/a>Part II. Measuring the decline properly<\/h3>\n

    Almost every popular error about this subject comes from conflating four different quantities. They are related, but they move independently and for different reasons.<\/p>\n

    Coronary event incidence<\/strong> is how often people have heart attacks. Case fatality<\/strong> is the proportion who die once they have one. CHD mortality<\/strong> is approximately the product of the two, plus recurrent events and sudden death outside hospital. Total cardiovascular mortality<\/strong> additionally includes stroke, heart failure, and other vascular death and is therefore a broader and less specific measure. A therapy that halves case fatality and a policy that halves incidence can produce identical falls in CHD mortality while representing entirely different achievements.<\/p>\n

    Four methodological cautions apply throughout.<\/p>\n

    Age standardization is mandatory.<\/strong> Crude death rates in an aging population can rise while individual risk falls steeply. Every trend statement in this review is age-adjusted or age-specific unless explicitly stated otherwise. It is also why absolute global cardiovascular deaths continue to increase while age-standardized rates fall.<\/p>\n

    Diagnostic criteria changed repeatedly, and this is not a minor artifact.<\/strong> Myocardial infarction has passed through successive biomarker eras: AST, LDH and total CK in the 1970s; CK-MB mass and activity assays in the 1980s and 1990s; conventional cardiac troponin I and T in the early 2000s; and high-sensitivity troponin assays from the 2010s \u2014 layered on top of the successive Universal Definitions of MI (2000, 2007, 2012, 2018).[27] As the analytical limit of detection fell from micrograms to sub-nanograms per litre, troponin elevations became detectable in patients who previously met no biomarker criterion. These categories must be kept distinct. Myocardial injury<\/strong> \u2014 any troponin elevation above the 99th percentile \u2014 is not itself myocardial infarction; it becomes infarction only when accompanied by clinical evidence of acute myocardial ischemia. Type 1 MI<\/strong> reflects atherothrombosis; type 2 MI<\/strong> requires an acute rise and\/or fall in troponin plus<\/em> clinical evidence of ischemia arising from supply\u2013demand mismatch; and non-ischemic myocardial injury<\/strong> (sepsis, myocarditis, renal disease, heart failure) remains injury and is not MI at all. What the assay transition did was enlarge the population in whom troponin is measurable and abnormal, increasing recorded NSTEMI diagnoses while also increasing recognition of injury that is not infarction. This is sometimes called troponin-induced diagnostic reclassification<\/strong>, and it has two opposing effects: it artifactually raises<\/em> recorded NSTEMI incidence in administrative databases, and it lowers<\/em> apparent aggregate case fatality by enlarging the denominator with lower-risk patients. When standardized historical algorithms are applied longitudinally \u2014 strict Q-wave ECG criteria and constant biomarker thresholds, as in MONICA \u2014 the decline in standardized ECG-defined infarction appears steeper<\/strong> than raw administrative coding suggests. Framingham demonstrated the magnitude directly: between 1960 and 1999, ECG-defined MI incidence fell roughly 50% while infarctions detected predominantly through increasingly sensitive biomarkers approximately doubled<\/strong>.[28] Surveillance that standardizes on ECG criteria or explicitly adjusts for biomarker era \u2014 Framingham, ARIC, MONICA \u2014 therefore deserves far more weight than unadjusted hospital discharge coding.<\/p>\n

    Competing risks matter.<\/strong> As cancer, stroke, and other causes of death shift, the population at risk of a coronary death changes. Falling coronary mortality partly redistributes deaths to other causes and to later ages rather than eliminating them; \u201cdeaths prevented or postponed\u201d is the scientifically honest phrase.<\/p>\n

    A century-long MI incidence series does not exist and should not be manufactured.<\/strong> The honest statement is that reliable national incidence surveillance begins in the 1960s\u20131980s depending on the country, and that everything before that is mortality data plus inference.<\/p>\n

    <\/a>What the surveillance actually shows<\/h4>\n

    United States, mortality.<\/strong> Age-adjusted ischemic heart disease mortality fell from 693 to 135 per 100,000 between 1970 and 2022 (\u221281%), and AMI-coded mortality from 354 to 40 (\u221289%) [2]. Within that span, the best-documented single interval is 1980\u20132000, over which age-adjusted CHD death rates fell from 542.9 to 266.8 per 100,000 in men and from 263.3 to 134.4 in women<\/strong> [3]. Using the older 1940 age standard, \u201cdiseases of the heart\u201d mortality fell from 307.4 per 100,000 in 1950 to 134.6 in 1996<\/strong>; these values must not be numerically spliced onto modern 2000-standard series.[26]<\/p>\n

    United States, incidence and case fatality.<\/strong> Framingham, following 9,824 participants over 212,539 person-years, found ECG-defined first-MI incidence fell approximately 50% from 1960\u20131999<\/strong>, while crude 30-day case fatality moved from 20% to 14%<\/strong>, one-year from 24% to 21%, and five-year from 45% to 41%; after age and sex adjustment, case fatality at all three horizons fell by about 60%<\/strong> [28]. ARIC, surveilling four U.S. communities from 1987\u20132008 with explicit biomarker adjustment for the enzyme-to-troponin transition, found incident hospitalized MI declining annually by about 4.3% in White men, 3.8% in White women, 3.4% in Black women \u2014 but only 1.5% in Black men<\/strong>, with 28-day case fatality declining approximately 2.6\u20133.6% per year<\/strong> by group [29]. ARIC also documented that infarctions themselves became smaller and less severe over time.[30]<\/p>\n

    Contemporary burden.<\/strong> Despite the decline, the United States recorded 915,973 cardiovascular deaths in 2023<\/strong>, with an age-adjusted CVD death rate of 218.3 per 100,000<\/strong> versus 224.3 in 2022.[23] Globally, GBD estimated approximately 9.0 million ischemic heart disease deaths in 2021<\/strong> with an age-standardized IHD mortality rate of about 108.7 per 100,000<\/strong>, representing a roughly 32% decline in age-standardized rate since 1990<\/strong> \u2014 and 19.2 million total cardiovascular deaths in 2023<\/strong>.[24] Age-standardized risk is falling; absolute burden is not.<\/p>\n

    <\/a>Part III. Why it happened: the attribution literature<\/h3>\n

    <\/a>How the models work, and what they can and cannot tell you<\/h4>\n

    The strongest direct answer to \u201chow much was medicine and how much was prevention\u201d comes from IMPACT-type models<\/strong>, developed by Capewell and colleagues from 1996 and applied in more than twenty countries[3,4]. The method takes observed changes in population risk factors, observed treatment uptake, trial-derived efficacy estimates, patient numbers, and case fatality, and asks how many \u201cdeaths prevented or postponed\u201d relative to a baseline year can be assigned to each component. The models have been shown to account for most of the observed mortality change in England, Scotland, Ireland, New Zealand, Finland, Sweden, and the United States, in the sense that the sum of modeled components typically reproduces 85\u2013100% of the observed change. This is calibration and internal accounting, not independent validation of the causal attribution itself: a model can reproduce an aggregate total while misallocating it among components.<\/p>\n

    These are Tier 4 attribution models incorporating externally estimated causal effect sizes, not randomized experiments<\/strong>, and the percentages they produce are model-dependent. Their value is that the same broad finding recurs in populations with very different histories: risk-factor change usually explains somewhat more of the mortality decline than treatment does, but both are large.<\/strong><\/p>\n

    <\/a>The double-counting problem, and how it is handled<\/h4>\n

    This is the single most important methodological issue in the field, and it is where most casual analyses fail. A fall in population mean blood pressure is partly caused by antihypertensive drugs[10], partly by diet and sodium intake, partly by body weight, and partly by other secular influences. A fall in population cholesterol is partly statins, partly dietary change, partly trans-fat removal from the food supply. If you credit \u201cmedicine\u201d with the drug effect and separately credit \u201clifestyle\u201d with the entire observed risk-factor change, you have counted the same biological improvement twice.<\/p>\n

    Well-implemented IMPACT models handle this in three ways. First, in well-specified implementations such as the Swedish analysis, individuals receiving a relevant treatment are removed from the population risk-factor calculation, so a statin-treated patient\u2019s cholesterol fall is credited to statins, not to \u201cdiet.\u201d This handling is implementation-specific and is not applied identically across all IMPACT models. Second, benefits from concurrent therapies are combined multiplicatively rather than added, using the Mant\u2013Hicks approach, so that a patient on aspirin plus a beta-blocker plus a statin is not credited with the arithmetic sum of three relative risk reductions. Third, overlapping treatment populations are explicitly enumerated and adjusted.<\/p>\n

    The Swedish analysis is the most transparent worked example. Between 1986 and 2002, total cholesterol fell 0.64 mmol\/L in the Swedish population. The model attributed approximately 5,210 deaths prevented or postponed to dietary change and 810 to statins<\/strong>. Smoking reduction accounted for about 1,195 deaths. A 2.6 mmHg fall in systolic pressure was divided into roughly 900 deaths from secular change and 575 from antihypertensive treatment<\/strong>. About three-quarters of the mortality reduction attributed to the three major risk factors occurred among people without diagnosed coronary disease \u2014 that is, in primary prevention.[10]<\/p>\n

    Models also apply compliance discounts<\/strong> to reflect real-world persistence rather than trial adherence \u2014 conventionally assuming 100% compliance during acute hospitalization, approximately 70% in symptomatic outpatients, and approximately 50% in asymptomatic primary-prevention populations. This matters: applying trial efficacy to prescription data without discounting systematically overstates the treatment share.<\/p>\n

    Residual double-counting remains possible even in good implementations, and this is a stated limitation rather than a solved problem.<\/p>\n

    <\/a>Biological lag times: why attribution windows matter<\/h4>\n

    A further complication rarely handled explicitly is that different interventions act on different timescales, so the same calendar window can flatter or penalize a given intervention depending on when it was adopted.<\/p>\n

    Weeks to months.<\/strong> Smoking cessation and antiplatelet therapy reduce coronary risk rapidly \u2014 before major anatomical plaque change would be expected.<\/p>\n

    One to two years.<\/strong> Pharmacological lipid lowering produces separating event curves in randomized trials within roughly 12\u201318 months[31], which is earlier than large changes in luminal geometry would be expected \u2014 the mechanism is generally attributed to plaque stabilization rather than anatomical regression, though the trial evidence establishes the outcome effect rather than the mechanism.<\/p>\n

    Decades.<\/strong> Cumulative ApoB exposure[32] and long-term blood-pressure burden operate over decades to influence atherosclerotic development and vascular injury. This is the timescale on which the population-level cholesterol decline of the 1970s and 1980s was operating \u2014 and it is why a model comparing 1980 with 2000 may capture the mortality consequence<\/em> of dietary change that began two decades earlier, while assigning it to the wrong window.<\/p>\n

    <\/a>Table 2 \u2014 Major attribution studies<\/h4>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
    Study \/ population<\/td>\nPeriod<\/td>\nObserved CHD mortality decline<\/td>\nRisk-factor contribution<\/td>\nTreatment contribution<\/td>\nOther \/ unexplained<\/td>\nAccess, tier, identifier<\/td>\n<\/tr>\n<\/thead>\n
    Ford et al., United States<\/strong><\/td>\n1980\u20132000<\/td>\nMen 542.9\u2192266.8\/100k; women 263.3\u2192134.4; 341,745 fewer deaths in 2000<\/strong><\/td>\n44%<\/strong><\/td>\n47%<\/strong><\/td>\n~9%<\/td>\nFull text reviewed; Tier 4<\/strong>. NEJM<\/em> 2007; DOI 10.1056\/NEJMsa053935; PMID 17554120[3]<\/td>\n<\/tr>\n
    Unal et al., England & Wales<\/strong><\/td>\n1981\u20132000<\/td>\n\u221262% men, \u221245% women<\/strong>; 68,230 fewer deaths<\/td>\n58%<\/strong><\/td>\n42%<\/strong><\/td>\nModel approximately complete after adverse trends<\/td>\nAbstract reviewed; Tier 4<\/strong>. Circulation<\/em> 2004; PMID 14993137[4]<\/td>\n<\/tr>\n
    Laatikainen et al., Finland<\/strong><\/td>\n1982\u20131997<\/td>\n\u221263%<\/strong><\/td>\n53\u201372%<\/strong><\/td>\n23%<\/strong><\/td>\nRemainder\/model uncertainty<\/td>\nAbstract reviewed; Tier 4<\/strong>. Am J Epidemiol<\/em> 2005; DOI 10.1093\/aje\/kwi274; PMID 16150890[5]<\/td>\n<\/tr>\n
    Bj\u00f6rck et al., Sweden<\/strong><\/td>\n1986\u20132002<\/td>\n\u221253.4% men, \u221252.0% women<\/strong>; 13,180 fewer deaths<\/td>\n55%<\/strong><\/td>\n36%<\/strong><\/td>\n~9%<\/td>\nAbstract + open follow-up full text; Tier 4<\/strong>. DOI 10.1093\/eurheartj\/ehn554; PMID 19141562[6]<\/td>\n<\/tr>\n
    Aspelund et al., Iceland<\/strong><\/td>\n1981\u20132006<\/td>\n\u221280%<\/strong>; MI incidence \u221266%<\/strong><\/td>\n73%<\/strong> (sensitivity 54\u201393%)<\/td>\n25%<\/strong> (8\u201340%)<\/td>\n~2%<\/td>\nAbstract\/results reviewed, open full text; Tier 4<\/strong>. DOI 10.1371\/journal.pone.0013957; PMID 21103050[7]<\/td>\n<\/tr>\n
    Ireland IMPACT<\/strong><\/td>\n1985\u20132000<\/td>\n\u221247%<\/strong><\/td>\n48.1%<\/strong><\/td>\n43.6%<\/strong><\/td>\n~8.3%<\/td>\nAbstract reviewed; Tier 4<\/strong>. PMID 16537349[8]<\/td>\n<\/tr>\n
    Bandosz et al., Poland<\/strong><\/td>\n1991\u20132005<\/td>\n\u221254%<\/strong>; 26,200 fewer deaths in 2005<\/td>\n54%<\/strong> (cholesterol\/dietary fat 39%, physical activity 10%, smoking in men 15%)<\/td>\n37%<\/strong><\/td>\nRemainder<\/td>\nAbstract reviewed; Tier 4<\/strong>. BMJ<\/em> 2012<\/td>\n<\/tr>\n
    Koopman et al., Netherlands<\/strong><\/td>\n1997\u20132007<\/td>\n\u221248%<\/strong> (269\u2192141\/100,000); ~11,200 fewer deaths<\/td>\n~36%<\/strong> (SBP 30%, cholesterol 10%, smoking 5%)<\/td>\n~37%<\/strong><\/td>\n~27% (model explained ~72%)<\/td>\nAbstract reviewed; Tier 4<\/strong>. PLoS One<\/em> 2016<\/td>\n<\/tr>\n
    North Karelia (within-community)<\/strong><\/td>\n1972\u20131986<\/td>\nSubstantial early IHD incidence decline<\/td>\nObserved changes in the three major risk factors predicted the great majority of the early incidence decline, far exceeding the reference area (exact percentage attribution requires primary-source verification)<\/td>\nNot separately assigned<\/td>\n\u2014<\/td>\nTier 2\/4<\/strong> natural experiment. Int J Epidemiol<\/em> 1989; DOI 10.1093\/ije\/18.3.595[35]<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    These percentages must not be averaged into a universal constant.<\/strong> They differ because baseline smoking, cholesterol, and hypertension prevalence differed; because treatment uptake differed; because countries were studied during different phases of technological diffusion; because age ranges differed; and because the models include somewhat different components. England and Wales looks smoking-dominated because the British smoking epidemic was unusually severe and its reversal unusually rapid. The United States shows the largest treatment share partly because it was studied during precisely the decades in which American treatment uptake expanded fastest.<\/p>\n

    <\/a>A defensible synthesis<\/h4>\n

    For high-income countries during the principal 1970s\u20132000s decline:<\/p>\n