Journal of Human (2000) 14, 83–90  2000 Macmillan Publishers Ltd All rights reserved 0950-9240/00 $15.00 www.nature.com/jhh REVIEW ARTICLE Fifty years of Framingham Study contributions to understanding hypertension

WB Kannel Department of Preventive and Epidemiology, Evans Department of Clinical Research, School of Medicine, Boston, MA, USA

The Framingham Study established hypertension as a genic risk factors, with which it tends to cluster. This major cardiovascular risk factor and quantified its clustering with other metabolically linked risk factors atherogenic potential. An his- has been shown to reflect insulin resistance promoted torical perspective is presented on the epidemiological by weight gain and abdominal . Obesity was insights about hypertension derived from 50 years of shown to be one of the major determinants of hyperten- Framingham Study research into the prevalence, inci- sion in the general population. Left ventricular hyper- dence, determinants and hazards of hypertension. trophy was shown to be an ominous harbinger of car- Existing misconceptions about the presence of critical diovascular disease rather than an incidental com- levels of , the impact of the systolic and pensatory phenomenon. Multivariate risk profiles for diastolic components of blood pressure, the hazard coronary disease, , peripheral disease and ‘mild’ hypertension, the impact in advanced age and the failure have been devised to facilitate incorpor- hazard of left ventricular hypertrophy. The importance ation of elevated blood pressure in a global, multivariate of isolated systolic hypertension and the cardiovascular risk assessment. were demonstrated. It has been demonstrated that Journal of Human Hypertension (2000) 14, 83–90 hypertension seldom occurs in isolation of other athero-

Keywords: hypertension; epidemiological contributions and historical perspective

Introduction a ‘normal’ pressure was designated to be 100 mm Hg plus the age of the patient. It was also widely held Hypertension is now recognised worldwide as a that women tolerated hypertension well. major clinical and public health problem. However, At the time the Study was initiated, and to some when the Framingham Study was initiated in 1949 degree still tenaciously held by many physicians there was great uncertainty about the role of blood today, it was the prevalent concept that the diastolic pressure in the development of atherosclerotic car- component of the blood pressure is chiefly respon- diovascular disease. Before the data emerged from sible for the cardiovascular sequelae of hyperten- epidemiological investigation of hypertension at the sion.1 Isolated systolic hypertension was considered Framingham Study, opinions differed about the an innocuous accompaniment of arterial stiffening. value of and indications for the treatment of essen- There was also the notion that only a fixed ‘basal’ tial hypertension, the level at which treatment blood pressure elevation was important and the low- should be considered, or the age at which therapy est pressure recorded on the patient (often under should be instituted. There was no sound basis for sedation at bed rest) was used to determine the need selecting those hypertensive persons most likely to for treatment. Left ventricular hypertrophy, a com- develop cardiovascular disease. The common var- mon feature of long-standing hypertension, was iety of hypertension was labelled ‘benign essential regarded as an incidental compensatory phenom- hypertension’ implying that it was not particularly enon. hazardous and might even be a normal or beneficial For many years there was a debate as to whether condition in older persons. The rise in blood press- ‘hypertension’ was a discrete disease to be dis- ure with advancing age was considered necessary in tinguished from elevated blood pressure.2 It was felt order to ensure adequate tissue perfusion as the by many that there was some critical value of blood arterial circulation progressively narrowed. Because pressure where normal leaves off and hypertension the average pressure tended to increase with age begins. normal pressures for the elderly were designated at higher values than those for younger adults. Hence, Medical trivia Although there was agreement about the necessity Correspondence: WB Kannel for the early detection and treatment of overt athero- Fifty years of Framingham Study contributions WB Kannel 84 sclerotic cardiovascular disease (which was not con- also challenged the concept that normal systolic sidered preventable) there was skepticism about the pressure was 100 mm Hg plus the subject’s age. An need to detect and treat asymptomatic abnormalities investigation of London busmen soon after also con- such as elevated blood pressure and blood lipids.3 cluded that systolic pressure was a better predictor There were some who doubted the validity of the of hypertensive events following coronary disease.10 epidemiological approach and the risk factor con- In 1969 the Framingham Study disputed the cept.3 The perception prevailed that it was meddle- accepted notion that the adverse consequences of some to label, alarm and treat persons with these hypertension derive chiefly from the diastolic press- asymptomatic conditions, particularly when they ure for coronary disease and a year later for were within the usual range of values found in the stroke.11,12 Several years later these findings of an apparently healthy population. However, Fram- equal or greater influence of systolic pressure were ingham Study investigation of the way cardiovascu- extended to peripheral artery disease and heart fail- lar disease develops in the general population indi- ure.13 The Framingham Study showed that in per- cated that much of the premature cardiovascular sons whose diastolic pressures did not exceed 95 mortality occurs with little warning in populations mm Hg, cardiovascular disease risk increased pro- generally prone to and in relation to gressively with the systolic pressure at all ages in identifiable risk factors present well in advance of both men and women (Figure 1).14 The common symptoms.4 It was determined that a preventive clinical practice of relying on the diastolic pressure approach is essential because more than half of the to judge the need for treatment in persons with an coronary deaths occurred suddenly and unexpec- elevated systolic blood pressure was shown to be tedly outside the hospital and only 20% of myocar- invalid. In this circumstance, knowledge of the dia- dial infarctions were preceded by and most stolic pressure was found to be misleading in evalu- were unheralded by transient ischaemic ating the cardiovascular risk.14 The incidence of all attacks.4,5 It was evident that atherosclerotic cardio- the cardiovascular sequelae of hypertension were vascular events must be dealt with in evolution if shown to be greater for isolated systolic than iso- substantial inroads are to be made against the epi- lated diastolic blood pressure elevations. Each stan- demic. One of the most common and powerful dard deviation increase in systolic blood pressure modifiable predisposing risk factors that needed increased the propensity to cardiovascular disease attention was hypertension. 40–50%, whereas for diastolic pressure, the increment was 30–35%. Following the Framingham Components of blood pressure Study many other population-based investigations, in a variety of population samples, confirmed that Clinical decisions and controlled trials for too long systolic pressure exerts a stronger influence than emphasized the diastolic pressure. Epidemiological diastolic pressure.1 investigation at the Framingham Study found Data from clinical trials also indicated that greater nothing to suggest a greater impact of the diastolic reliance should be placed on systolic pressure in than systolic pressure on the occurrence of cardio- evaluating hypertensive risk and control.15,16 How- . The preference for the diastolic ever, despite all the evidence, trials of hypertension component of the blood pressure as the more treatment continued to use diastolic pressure as important promoter of the adverse sequelae of entry and outcome criteria until the recent SHEP hypertension may derive from the proclamation and Syst-Eur studies.15,16 These studies demon- about hypertension in the 1927 edition of the classic strated the benefits of lowering systolic blood press- Cecil’s medical textbook that: ‘The blood pressure is ure elevations in the elderly to reduce the risk of found to be distinctly raised; systolic pressures of stroke and . It was only after 250 mm Hg are quite common. More important than these trials that practitioners have begun to relin- the systolic pressure is the increase in the diastolic quish their overemphasis on the diastolic pressure pressure, for it has been shown that as the diastolic in their evaluation and treatment of hypertension. pressure increases, the lose their normal elasticity and efficiency and greater work is conse- Arterial compliance quently thrown upon the heart. It is therefore important to bear in mind, in determining the prog- The foregoing epidemiological facts and clinical nosis, that those individuals whose diastolic blood trials have fostered new concepts questioning the pressures remain relatively low, in spite of high view that elevated peripheral resistance is the systolic pressures, are decidedly less apt to suffer underlying cause of hypertension and that the best cardiac decompensation and cerebral apoplexy.’6 indicator of hypertension is the diastolic blood This concept of the greater importance of the dia- pressure. Because mean arterial pressure is closer to stolic pressure was reinforced by the first JNC report diastolic than systolic pressure the former was used in 1977 that continued to recommend the diastolic as a surrogate for the mean arterial pressure and pressure as the basis for diagnosis and treatment of elevated diastolic pressure became the hallmark of hypertension.7 It was further reinforced by the use hypertension.17 The notion that the increase in sys- of diastolic pressure as the basis for therapeutic tolic pressure with age was innocuous persisted interventions in hypertension clinical trials. The until the SHEP and Syst-Eur studies reported their earliest indication that the systolic pressure was results, despite the Framingham Study data chal- more strongly related to mortality than diastolic lenging this for many years. pressure came from the insurance industry.8,9 They Aging is the most important cause of arterial stiff-

Journal of Human Hypertension Fifty years of Framingham Study contributions WB Kannel 85

Figure 1 Risk of cardiovascular disease by systolic blood pressure at diastolic blood pressure up to 95 mm Hg in the Framingham Heart Study.

Table 1 Cox proportional hazards regression models relating incidence of coronary heart disease to systolic blood pressure (SBP) together with either diastolic blood presssure (DBP) DBP (Model 1) or PP (Model 2) within grouped SBP categories, adjusted for index examination characterstics

Coefficient, ␤ a s.e.a Wald ␹2b Hazard ratio/10 mm Hg Hazard ratio/s.d. (CI)a increment (CI)c

Model 1: SBP and DBPd mm Hg SBP 0.22 0.06 13.2 1.25 (1.11–1.41)** 1.54 (1.19–1.99)** DBP coefficients for individual SBP groups DBP within SBP Ͻ120 mm Hg −0.18 0.08 4.8 0.84 (0.71–0.98)* 0.85 (0.74–0.97)* DBP within SBP 120–139 mm Hg −0.14 0.07 3.6 0.87 (0.76–1.00) 0.85 (0.75–0.98)* DBP within SBP 140–159 mm Hg −0.16 0.07 5.3 0.85 (0.75–0.98)* 0.85 (0.75–0.98)* DBP within SBP у160 mm Hg −0.17 0.07 6.1 0.84 (0.74–0.97)* 0.85 (0.75–0.97)*

Model 2: SBP and PPd mm Hg SBP 0.05 0.08 0.40 1.05 (0.90–1.23) 1.12 (0.80–1.55) PP coefficients for individual SBP groups PP within SBP Ͻ120 mm Hg 0.11 0.09 1.5 1.12 (0.94–1.33) 1.28 (1.04–1.57)* PP within SBP 120–139 mm Hg 0.19 0.07 6.8 1.20 (1.05–1.38)* 1.29 (1.04–1.58)* PP within SBP 140–159 mm Hg 0.16 0.07 5.4 1.18 (1.03–1.35)* 1.28 (1.04–1.58)* PP within SBP у160 mm Hg 0.15 0.07 4.1 1.16 (1.00–1.34)* 1.28 (1.04–1.58)* aAssociated with a 10 mm Hg increment in the corresponding blood pressure component. bWald ␹2 = (␤/SE)2. cAssociated with a 1-s.d. increment in the corresponding blood pressure component. dAdjusted for age, sex, body mass index, cigarettes smoked per day, glucose intolerance, and total /HDL. Model 1 includes DBP and model 2 includes pulse pressure (PP), both specific to SBP categories (SBP Ͻ120, 120–139, 140–159, and у160 mm Hg). *P Ͻ 0.05, **P Ͻ 0.001. (Reprinted with permission from Franklin et al. Circulation 1999; 100: 354–360). ness, but hypertension accelerates this aging process tinuing rise in systolic pressure so that the pulse by increasing the stretching of the artery rendering pressure increases dramatically in later life. Systolic it stiffer.17 Increased aortic stiffness increases aortic and pulse pressure have been shown to be of greater pulse pressure for any given stroke volume and predictive value than diastolic pressure in the Fram- raises the peak pressure in the central arteries and ingham Study.20 left ventricle predisposing to left ventricular The Framingham Study established by 1980 that hypertrophy and failure and reduces the diastolic there is a more robust relationship between cardio- pressure predisposing to myocardial ischaemia by vascular disease and systolic than diastolic pressure. reducing coronary perfusion which occurs during Indeed, recent Framingham Study reports show that diastole.17,18 The Framingham Study confirmed con- for any given systolic pressure cardiovascular events clusively that the mean arterial blood pressure are inversely related to diastolic pressure.20 Risk of increases only slightly throughout adult life19 and cardiovascular events in the Framingham Study has showed that a marked reduction in diastolic press- been shown to be related to pulse pressure at any ure occurs beyond age 50, accompanied by a con- level of blood pressure (Table 1). Isolated systolic

Journal of Human Hypertension Fifty years of Framingham Study contributions WB Kannel 86 Table 2 Risk of cardiovascular events in subjects with hypertension. (Persons aged 35–64 years—Framingham Study 36-year follow-up)

Cardiovascular events Age-adjusted biennial rate per 1000 Age-adjusted risk ratio Excess risk per 1000

Men Women Men Women Men Women

Coronary heart disease 45 21 2.0* 2.2* 23 12 Stroke 12 6 3.8* 2.6* 9 4 Peripheral arterial disease 10 7 2.0* 3.7* 5 5 Cardiac failure 14 6 4.0* 3.0* 10 4 Cardiovascular events 65 35 2.2* 2.5* 36 21

*P Ͻ 0.0001.

Table 3 Myocardial infarctions unrecognised by hypertensive status

Hypertensive status Percent of myocardial infarctions unrecognised

Excluding diabeticsa Excluding anti-HBP RXa Excluding LVHa

Men Women Men Women Men Women

Normal 18.5 30.7 17.8 26.6 19.6 29.0 Mild 28.3 36.1 30.2 35.5 30.1 35.3 Definite 33.2 48.1 34.8 48.5 32.7 50.5

aAlso excludes persons with coronary heart disease at exam immediately preceding myocardial infraction. LVH, left ventricular hypertrophy; HBP RX, high blood pressure treatment. (From: Kannel WB et al. Unrecognised myocardial infarction and hypertension. The Framingham Study. Am Heart J 1985; 109: 581).

hypertension has been shown to carry a substantial ingham Study examined the hypothesis of a J-curve risk for development of cardiovascular morbidity relation between diastolic blood pressure and coron- and mortality.21,22 These epidemiological findings ary mortality and found it confined to persons with support the contention that large artery stiffness is a myocardial infarction.25 Thus it appears that there an important feature of hypertensive cardiovascular is a continuous graded influence of blood pressure disease. Thus, in studies of hypertension, recent on coronary morbidity and mortality in healthy per- attention is being focused on as a sons with no critical value where ‘normal’ leaves off causal factor and target of therapy, diverting atten- and ‘hypertension’ ensues. It appears that it is blood tion away from peripheral resistance as the hallmark pressure that kills and not ‘hypertension’. of the disease. This arterial stiffness is usually mani- fested as an increased pulse pressure and isolated Basal or casual pressure systolic hypertension. At the initiation of the Framingham Study it was Blood pressure or hypertension considered appropriate to ignore labile and casual blood pressure elevations.26 Clinicians tended to Before epidemiological research, ‘hypertension’ was disregard casual office blood pressures and rely on considered more as a disease than as a risk factor. ‘basal’ pressures obtained after a period of bed rest There was the perception that there was a critical under sedation to establish the presence of ‘hyper- blood pressure value that distinguished hyper- tension’ requiring treatment.26,27 Epidemiological tensive persons from those who did not have the investigation at Framingham revealed that an aver- disease. This would require that there be a bimodal age of a series of pressures provided a valid estimate distribution of blood pressure in the general popu- of the risk regardless of how labile it was.28 Attempt lation, which has never been convincingly demon- to improve the estimate of risk by using the lowest strated. Also, the distributions of blood pressure in of three recorded blood pressures proved disap- those with and without cardiovascular disease over- pointing. This refuted the common clinical practice lap to such a degree that no blood pressure can dis- of using the lowest blood pressure recorded as the tinguish cases from non-cases. Investigation of the index of the hypertensive hazard. The average of a blood pressure relationship to development of car- series of pressures appeared to be the best indicator diovascular disease revealed a continuous graded of the cardiovascular risk. influence extending down into the level of pressures considered ‘normal’. Cardiovascular hazards It has been claimed that there is an excess risk of mortality at both high and low levels of diastolic Epidemiological research demonstrated that because blood pressure. Some have claimed that this applies of its high prevalence in the general population and only to persons with coronary disease whereas sizeable risk ratio, hypertension is a powerful risk others claim it applies in general.23,24 The Fram- factor for cardiovascular disease accounting for

Journal of Human Hypertension Fifty years of Framingham Study contributions WB Kannel 87 about 35% of atherosclerotic cardiovascular ECG-LVH, and antihypertensive treatment.33 In male events.29 Hypertension was shown to predispose hypertensives about 35% of infarctions went unre- powerfully to all of the major atherosclerotic cardio- cognised—double the rate of normotensive men and vascular events including coronary disease, stroke, in hypertensive women close to half the infarctions peripheral artery disease and .29,30 Risk were unrecognised. These observations led to the ratios were shown to be larger for stroke and heart recommendation that hypertensive persons should failure as clinical studies suggested, but the Fram- be periodically ECG monitored for the occurrence of ingham Study data indicated that coronary disease a silent myocardial infarction. was in fact the most common and most lethal It was widely believed that mild hypertension sequela, equalling in incidence all the other cardio- promotes brain infarctions, whereas severe hyper- vascular disease outcomes combined (Table 2). The tension leads to intracerebral haemorrhage. Investi- less than expected efficacy of antihypertensive ther- gation of the relation of hypertension to the occur- apy for avoiding coronary disease led some to doubt rence of strokes in the Framingham Study indicated whether there was a direct causal relationship that the preponderance of hypertension-related between hypertension and coronary disease. How- strokes in the Framingham Study were atherthrom- ever, the epidemiological data indicated that the botic brain infarctions, whether the hypertension incidence of every clinical manifestation of coronary was severe (70%) or mild (56%). The proportion of disease, including angina, myocardial infarction, strokes due to intracerebral haemorrhage in mild and sudden death increased in proportion to the hypertension (5%) was virtually identical to that for severity of the antecedent hypertension.31 In con- severe hypertension (4%). Furthermore, the pro- trast to blood pressures measured before myocardial portion of strokes due to brain infarction increased infarctions, the blood pressure measured after the event was found to be unrelated to survival over the at the expense of subarachnoid haemorrhage and next 5 years.32 However, the blood pressure mea- cerebral embolus, whereas the proportion due to sured prior to the infarction was related to survival intracerebral haemorrhage did not change. after infarction with hypertensive persons having Hypertension is recognised as a major precursor almost three times the mortality of previously of heart failure but some have assigned a more domi- normotensive infarct patients. These puzzling find- nant role to coronary disease. Investigation of the ings were explained by the two-fold greater risk of role of hypertension in heart failure in the Fram- death in hypertensive patients who experienced a ingham Study revealed that adjusting for age and spontaneous substantial decrease in their pressure other relevant risk factors, hypertension imposed a after their myocardial infarction compared to that of two-fold increased risk in men and increased the those who remained hypertensive. A decrease in hazard three-fold in women. Hypertension was pressure with interim myocardial infarction was found to account for 39% of the cases in men and found to occur frequently and must be taken into 59% in women. Survival following onset of hyper- account in evaluating hypertensive risk after a myo- tensive heart failure was bleak with only 24% of cardial infarction. men and 31% of women surviving 5 years.34 Thus The Framingham Study revealed that approxi- hypertension was the most common predisposing mately 30% of the myocardial infarctions that occur factor for heart failure, contributing a large pro- in the general population go unrecognised. For portion of the cases in the general population reasons that are unclear, it was found that like dia- (Table 4). It is likely that earlier and more aggressive betics, hypertensive persons were unusually suscep- blood pressure control offer the greatest promise for tible to these unrecognised infarctions (Table 3). reducing the incidence of heart failure and its asso- This hypertensive propensity to unrecognised ciated mortality. infarctions was found to persist even excluding per- sons with possible confounders such as diabetes,

Table 4 Risk factors for cardiac failure in the Framingham Offspring and Cohort Studya

Risk factor Age and RFA hazard ratio Prevalence, % Population attributable risk

Hypertension (у140/90 mm Hg) Men 2.1 60 39 Women 3.4 62 59 Myocardial infarction Men 6.3 10 34 Women 6.0 3 13 Angina Men 1.4 11 5 Women 1.7 9 5 Diabetes Men 1.8 8 6 Women 3.7 5 12 ECG-LVH Men 2.2 4 4 Women 2.9 3 5 Valvular heart disease Men 2.5 5 7 Women 2.1 8 8 aSubjects were 40 to 89 years old; follow-up period was 18 years. ECG-LVH, electrocardiographically detected left ventricular hypertrophy; RFA, risk factor adjusted. (From: Levy D et al. JAMA 1996; 275: 1553–1562).

Journal of Human Hypertension Fifty years of Framingham Study contributions WB Kannel 88 Left ventricular hypertrophy heart rate is followed by an increase in blood press- ure and development of hypertension.41 The Fram- Left ventricular hypertrophy is a response of the ingham Study in 1987 confirmed this and showed heart to the chronic pressure overload imposed by that the predictive power of the heart rate rivalled hypertension. The electrocardiographic pattern of that of obesity.42 increased R-wave voltage, and ST and T-wave The Framingham Study in 1987 showed the heart repolarisation abnormalities was well recognised as rate to be related to the rate of development of car- an important clinical diagnostic entity. However, diovascular and all-cause mortality in men, even before epidemiological investigation of this entity in taking into account other risk factors.43 The relation- the Framingham Study in 1969 there were no pro- ship to sudden coronary fatalities was particularly spective precise estimates of its prevalence, inci- strong with a linear increase with the heart rate. In 35 dence, and prognosis in the general population. the Framingham Study the association of the heart Before this it was believed that the hypertrophy was rate with cardiovascular mortality was also observed a compensatory phenomenon in response to the in the hypertensive segment of the population.44 increased workload imposed by the hypertension. Heart rate has been shown to be a marker of There was even fear that reversal of this hypertrophy increased sympathetic tone by Julius and Palatini.45 would be detrimental to cardiac function. High heart rates are also found in persons with fea- Epidemiological investigation at the Framingham tures of the insulin resistance syndrome. A large Study revealed that left ventricular hypertrophy, proportion of hypertensive persons has increased rather than an asset in hypertension, was an omin- sympathetic activity, tachycardia, and the metabolic ous harbinger of disabling and lethal cardiovascular syndrome. The implications for treatment of hyper- 35,36 events. When it occurred, it was found to be tension is to select a therapy that not only reduces associated with a marked increase in the risk of cor- the blood pressure but also decreases the heart rate onary disease, stroke, and heart failure. The risk of and improves the metabolic abnormalities associa- cardiovascular sequelae of hypertension was shown ted with insulin resistance. The efficacy of so doing to increase with the severity of the hypertrophy needs to be tested by randomised trials. whether evaluated by ECG, echocardiogram, or chest film.37,38 No critical amounts of hypertrophy have been found that delineate compensatory from patho- The elderly and women logical hypertrophy. Risk of cardiovascular sequelae Before epidemiological insights were provided by was shown to increase progressively with the the Framingham Study hypertension was con- increase in echocardiographically measured left sidered to be less dangerous in the elderly and ventricular mass, and severity of ECG manifestations women. Because blood pressures tended to increase of the condition.37,38 Each 50 g/m increment in left with age higher pressures were accepted as more ventricular mass was associated with a 50% greater normal in the elderly than in the middle-aged.30 Epi- risk of developing cardiovascular disease and nearly demiological data from the Framingham Study has twice the risk of a cardiovascular disease death. It corrected this misconception by demonstrating that has been shown that antihypertensive therapy can hypertension risk ratios do not diminish greatly regress left ventricular hypertrophy in hypertensive with advancing age and that the hazard of a given persons, but it is not established whether such elevation of blood pressure in the elderly is actually regression reduces the associated high risk of cardio- higher than the same pressure in middle-age.30 The vascular events. Investigation of the prognostic Framingham Study also showed that although implications of reversal of ECG indications of left women have a lower incidence of cardiovascular ventricular hypertrophy in the Framingham Study events than men do at the same blood pressure, the cohort indicated that improvement in the ECG fea- risk ratio is just as large in women (Table 5). tures of left ventricular hypertrophy was associated with a substantial reduction in the risk of cardio- vascular events.38 These results strongly suggest that Concomitant risk factors regression of left ventricular hypertrophy by anti- A tendency for other risk factors to cluster with hypertensive therapy may well reduce the risk of hypertension was long noted in the Framingham cardiovascular events. The Framingham Study showed that left ventricu- Table 5 Cardiovascular disease risk imposed by hypertension: by lar hypertrophy is an important feature of the evol- age in each sex. (The Framingham Study—36 year follow-up) ution of hypertension on to heart failure. Cardio- vascular risk ratios for hypertensive persons who Age (years) Age-adjusted Risk ratio Excess risk rate per 1000 developed left ventricular hypertrophy were found to be greatest for cardiac failure but coronary disease Men Women Men Women Men Women was shown to be the most common lethal hazard.39 35–64 65 35 2.2* 2.5* 36 21 Heart rate 65–94 125 81 1.8* 1.8* 56 35 In 1975 four Chicago epidemiological surveys indi- Hypertension = Ͼ160/90 mm Hg. CVD’s = CHD, Stroke, CHF, IC. Excess risk = Def. HBF-Normal. cated a strong correlation between the heart rate and *P Ͻ 0.0001 40 blood pressure. As early as 1945 it was shown in (From: Kannel WB. Potency treatment of vascular risk factors. a group of US army officers that elevation of the Euro H J 1992; 13: 34–42).

Journal of Human Hypertension Fifty years of Framingham Study contributions WB Kannel 89 Table 6 Sixteen-year coronary heart disease incidence by number of other risk factors. (Framingham Heart Study Offspring with Elev- ated Blood Pressure, ages 30–65 years at baseline)

Number of other risk Relative risk (95% CI) Prevalence CHD events number (%) Population attributable risk factors (multivariate)

Men 0 1.0 (referent) 22% 10 (14%) — 1 1.33 (0.57, 3.06) 29% 17 (24%) 0.09 у2 2.28 (1.09, 4.78) 49% 45 (63%) 0.39 Total 72 (100%)

Women 0 1.0 (referent) 18% 2 (5%) — 1 2.05 (0.41, 10.18) 28% 7 (18%) 0.23 у2 4.93 (1.14, 21.27) 54% 31 (78%) 0.68 Total 40 (100%)

Study.46 Many of the risk factors that tend to cluster it was considered appropriate to ignore labile and with elevated blood pressure were shown to predict systolic elevations of blood pressure. Clinicians its occurrence and greatly influence its impact on tended to disregard office blood pressure elevations the occurrence of atherosclerotic cardiovascular dis- in favour of ‘basal’ pressures. Isolated systolic ease.46,47 The Framingham Study determined the hypertension was rarely taken seriously before epi- prevalence of occurrence of one or more risk factors demiological evidence to the contrary was provided. in association with elevated blood pressure and Hypertension was considered less dangerous in the compared this with the rate of clustering expected elderly and women and because blood pressure by chance.48 We also examined the influence of tended to increase with age, higher pressures were weight and weight gain on the tendency for risk fac- accepted as normal in the elderly. Left ventricular tors to cluster in hypertensive persons. Clusters of hypertrophy was considered an incidental com- three or more of the metabolically linked risk factors pensatory phenomenon. Hypertension was not (elevated cholesterol, triglyceride, glucose, and body recognised as an ingredient of a cardiovascular risk mass index and reduced HDL-cholesterol) occurred that tended to cluster with other metabolically at 4–5 times the rate expected by chance in men and linked risk factors. It is clear that accepted teachings women respectively. Only 22% of men and 18% of that were based on case studies and clinical women had elevated blood pressure in isolation impression have not stood the test of prospective with other risk factors. The tendency for risk factors epidemiological investigation. Physicians have cor- to cluster in persons with hypertension was shown rected many misconceptions about hypertension to increase stepwise with weight gain. A 5-lb weight because of epidemiological insights provided by the gain imposed a 30% increase in the extent of clus- Framingham Study. tering. Risk of coronary disease in persons with elevated blood pressure was shown to vary widely in relation Acknowledgements to the amount of risk factor clustering. Among per- Framingham Study research is supported by sons with elevated blood pressure, it was estimated NIH/NHLBI Contract N01-HC-38038 and the Visit- that 63% of coronary events in men and 78% in ing Scientist Program which is supported by Servier women occur in conjunction with two or more risk Amerique. Reprints are regretfully not available. factors and 39% of the coronary events in hyperten- sive men are attributable to clustering of two or more other risk factors. In women, 68% of coronary events References are attributable to the presence of two or more accompanying risk factors (Table 6). The Fram- 1 Rutan GH, McDonald RH, Kuller LH. A historical per- spective of elevated systolic Vs diastolic blood press- ingham Study has long suggested that hypertension ure from an epidemiological and clinical trial view- is best regarded as a component of metabolic syn- point. J Clin Epidemiol 1989; 42: 663–673. 46–48 drome and cardiovascular risk profile. To assist 2 Pickering GW. Genetic factor in essential hyperten- in risk stratification of hypertension, multivariate sion. Ann Intern Med 1955; 42: 457–464. risk formulations have been contrived to make it 3 Werko L. Risk factors and coronary heart disease: facts convenient for physicians to quantify the cardio- or fancy? Am Heart J 1976; 91: 87–91. vascular hazards faced by their hypertensive 4 Gordon T, Kannel WB. Premature mortality from cor- patients.49 onary heart disease. The Framingham Study. JAMA 1971; 215: 1617–1625. 5 Kannel WB, Barry P, Dawber TR. Immediate mortality Summary in coronary heart disease: The Framingham Study. In: Proceedings of the IV World Congress of Cardiology. Before data emerged from the epidemiological inves- Mexican Int’l Soc. Cardiology 1963; IV-B: 176–188. tigation of hypertension by the Framingham Study, 6 Goodridge M. . In: Cecil RL, Kennedy most hypertension was considered a benign con- F (eds). A Textbook of Medicine. WB Saunders: Phila- dition. At the initiation of the Framingham Study delphia, 1927, p 1091.

Journal of Human Hypertension Fifty years of Framingham Study contributions WB Kannel 90 7 Report of the Joint National Committee on Detection, 29 Kannel WB. Blood pressure as a cardiovascular risk Evaluation and Control of High Blood Pressure. JAMA factor: prevention and treatment. JAMA 1996; 275: 1977; 273: 255–261. 1571–1576. 8 Gubner RS. Systolic hypertension: a pathogenetic 30 Kannel WB. Rationale for treatment of hypertension in entity. Am J Cardiol 1962; 9: 773–776. the elderly. Am J Geriatr Cardiol 1994; 3: 33–45. 9 Lew EA. High blood pressure, other risk factors and 31 Wilson PWF, Kannel WB. Hypertension, other risk fac- longevity: the insurance viewpoint. Am J Med 1973; tors and the risk of cardiovascular disease. In: Laragh 55: 281–294. JH, Brenner BM (eds). Hypertension: Pathophysiology, 10 Morris JN, Kagan A, Pattison DC, Gardner MJ. Inci- Diagnosis and Management. 2nd edn. Raven Press: dence and prediction of ischemic heart disease in New York, NY, 1995, pp 99–114. London busmen. Lancet 1966; 2: 553–559. 32 Kannel WB, Sorlie P, Castelli WP, McGee D. Blood 11 Kannel WB, Schwartz MJ, McNamara PM. Blood press- pressure and survival after myocardial infarction: the ure and risk of coronary heart disease. The Fram- Framingham Study. Am J Cardiol 1980; 45: 326–330. ingham Study. Dis Chest 1969; 56: 43–52. 33 Kannel WB, Dannenberg AI, Abbott RD. Unrecognized 12 Kannel WB, Wolf PA, Verter J, McNamara PM. Epide- myocardial infarction and hypertension: the Fram- miologic assessment of the role of blood pressure in ingham study. Am Heart J 1985; 109: 581–585. stroke. JAMA 1970; 214: 301–310. 34 Levy D et al. The progression from hypertension to 13 Kannel WB. Role of blood pressure in the development congestive heart failure. JAMA 1996; 275: 1557–1562. of cardiovascular morbidity and mortality. Prog Car- 35 Kannel WB, Gordon T, Offut D. Left ventricular diovasc Dis 1974; 17: 5–24. hypertrophy by electrocardiogram: prevalence inci- 14 Kannel WB. Epidemiology of : dence, and mortality in the Framingham Study. Ann the Framingham Experience. Proc R Coll Phys Edinb Intern Med 1969; 71: 89–105. 1991; 21: 273–287. 36 Kannel WB, Cobb J. Left ventricular hypertrophy and 15 SHEP Cooperative Research Group. Prevention of mortality: results from the Framingham Study. Cardi- stroke by antihypertensive drug treatment in older per- ology 1992; 81: 291–298. sons with isolated systolic hypertension. Final results 37 Levy D et al. Prognostic implications of echocardio- of the Systolic Hypertension in the Elderly Program graphically determined left ventricular mass in the Fra- (SHEP). JAMA 1991; 265: 3255–3266. mingham Study. NEnglJMed1990; 322: 1561–1566. 16 Staesson JA et al. Randomized double-blind compari- 38 Levy D et al. Prognostic implications of baseline elec- son of placebo and active treatment for older patients trocardiographic features and their serial changes in with isolated systolic hypertension. The Systolic subjects with left ventricular hypertrophy. Circulation Hypertension in Europe (Syst-Eur) Trial Investigators. 1994; 90: 1786–1793. Lancet 1997: 350: 757–764. 39 Kannel WB, Dannenberg AL, Levy D. Population 17 O’Rourke MF. Isolated systolic hypertension, pulse implications of electrocardiographic left ventricular pressure, and arterial stiffness as risk factors for car- hypertrophy. Am J Cardiol 1987; 60: 851–931. diovascular disease. Curr Hypertens Rep 1999; 3: 40 Stamler J et al. Relationship of multiple variables to 204–211. blood pressure: findings from four Chicago epidemiol- 18 Mitchell GF et al. Sphygmomanometrically determ- ogic studies. In: O Paul (ed). Epidemiology and Control ined pulse pressure is a powerful independent predic- of Hypertension. Symposia Specialists: Miami, 1975, tor of recurrent events after myocardial infarction in pp 307–352. patients with impaired ventricular function. Circu- 41 Levy RI et al. Transient tachycardia: prognostic sig- lation 1997; 96: 4254–4260. nificance alone and in association with transient 19 Franklin SS et al. Hemodynamic patterns of age- hypertension. JAMA 1945; 129: 585–588. related change in blood pressure: the Framingham 42 Garrison RJ, Kannel WB, Stokes J III. Incidence and Study. Circulation 1997; 96: 308–315. precursors of hypertension in young adults. Prev Med 20 Franklin SS et al. Is pulse pressure useful in predicting 1987; 16: 235–251. risk for coronary heart disease? The Framingham Heart 43 Kannel WB et al. Heart rate and cardiovascular mor- Study. Circulation 1999; 100: 354–360. tality: the Framingham Study. Am Heart J 1987; 113: 21 Kannel WB, Dawber TR, McGee DL. Perspectives on 1489–1494. systolic hypertension. The Framingham Study. Circu- 44 Gillman MW et al. Influence of heart rate on mortality lation 1980; 61: 1179–1182. among persons with hypertension: the Framingham 22 Wilking SV et al. Determinants of isolated systolic Study. Am Heart J 1993; 125: 1148–1154. hypertension. JAMA 1988; 260: 3451–3455. 45 Palatini P, Julius S. Relevance of heart rate as a risk 23 Cruickshank JM. Coronary flow reserve and the J-curve factor in hypertension. Curr Hypertens Rep 1999; 3: relation between diastolic blood pressure and myocar- 219–224. dial infarction. BMJ 1988; 297: 1227–1230. 46 Kannel WB. Implications of Framingham Study data 24 Alderman MH, Ooi WL, Modhaven S, Cohen H. Treat- for treatment of hypertension: Impact of other risk ment-induced blood pressure reduction and the risk of factors. In: Laragh JH, Buhler FR, Seldin DW (eds). myocardial infarction. JAMA 1989; 262: 920–924. Frontiers in Hypertension Research. Springer–Verlag: 25 D’Agostino RB, Belanger AJ, Kannel WB, Cruickshank New York, 1981, pp 17–21. JM. Relation of blood pressure to coronary heart dis- 47 Kannel WB. Potency of vascular risk factors as the ease in the presence of myocardial infarction. BMJ basis for antihypertensive therapy. Eur Heart J 1992; 1991; 303: 385–389. 13: (Suppl G): 34–42. 26 Smirk FH, Veale A, Alstad KS. Basal and supplemental 48 Kannel WB, Wilson PWF, Silbershatz H, D’Agostino blood pressures in relation to life expectancy and hyper- RB. Epidemiology of risk factor clustering in elevated tension symptomatology. NZMedJ1959; 58: 711. blood pressure. In: Gotto AM et al. (eds). Multiple Risk 27 Vaisrub S. Labile and systolic hypertension: a re- Factors in Cardiovascular Disease. Klewer Academic appraisal. (Editorial). JAMA 1981; 245: 1250. Publishers: The Netherlands, 1998, pp 325–333. 28 Kannel WB, Sorlie P, Gordon T. Labile hypertension: 49 Wilson PWF et al. Prediction of coronary heart disease a faulty concept? The Framingham Study. Circulation using risk factor categories. Circulation 1998; 97: 1980; 61: 1183–1187. 1837–1847.

Journal of Human Hypertension