IJE Advance Access originally published online on February 15, 2006
International Journal of Epidemiology 2006 35(3):720-730; doi:10.1093/ije/dyl014
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Coronary Heart Disease |
Profiling risk: the emergence of coronary heart disease epidemiology in the United States (1947–70)
1 Department of Health and Nutrition Sciences, Brooklyn College, City University of New York, 2900 Bedford Avenue, Brooklyn, NY 11210, USA
2 Center for the History and Ethics of Public Health, Department of Sociomedical Sciences, Columbia University, 722 West 168th Street, New York, NY 10032, USA
* Corresponding author. Center for the History and Ethics of Public Health, Mailman School of Public Health, 722 West 168th Street, New York, NY 10032, USA. E-mail: go10{at}columbia.edu
 |
Abstract |
|---|
 Top Abstract Designing heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
This historical study examines the development of coronary heart disease (CHD) research and its role in the evolution of post-1945 chronic disease epidemiology in the United States. To give the examination greater salience, it compares the pathway represented by CHD epidemiology with that of lung cancer. Historians have paid less attention to the differences between the two, which later merged into what we now call ‘risk factor epidemiology’. This study assesses why CHD epidemiology in the post-war period almost uniformly began with cohort studies and primarily stressed clinical variables as putative aetiological factors. It describes how CHD epidemiologists sought to justify the creation of a non-infectious chronic disease epidemiology, a position reinforced by the relative swiftness with which they obtained important results. It also follows the emergence of ‘risk factor thinking’ within CHD epidemiology. CHD epidemiology critically differed from its lung cancer counterpart in that it identified multiple factors of risk, each producing relatively small effects, rather than a single factor producing a strong and evident outcome. Consequently, it was difficult for CHD epidemiologists to demonstrate causality and to confirm scientifically that reducing risk factors would lower CHD rates. This had significant consequences for primary prevention and public health policy.
Accepted 12 January 2006
Historians studying the shift in epidemiology from infectious to chronic disease in the decades following the Second World War have stressed the forceful role of two major research programmes, one focusing on the aetiology of lung cancer the other coronary heart disease (CHD).1,2 Lung cancer and its relation to cigarette smoking have received somewhat more attention of the two, particularly from social and medical historians. Its tale includes the spectacle of a renascent epidemiology confronting a powerful, wealthy and, we now know, duplicitous industry.3,4 Important epidemiological research catalysed a shift in social mores, undermining a pleasant, licit, and very popular habit.2
The research itself produced strong objections, based as it was on observational studies and statistical correlations. It was also deeply puzzling in that it linked cigarette smoking, taken as a single agent, to a sweeping list of disease outcomes.5–7 The intense response of industry and scientific critics to the unfolding epidemiological work pressed those who supported a causal nexus between cigarettes and lung cancer to develop an increasingly sophisticated conceptual and quantitative framework for chronic disease research.2,7,8 During the 1950s and 1960s, government and university-based statisticians and epidemiologists carefully crafted intellectually rigorous positions on issues of research design, analyses of association, the validity of population-based data, the relations of epidemiological to laboratory and clinical studies, and the criteria of causal thinking.8 Historians of science have subsequently been drawn to the lung cancer-tobacco controversies because they generated so much of what is central to modern epidemiology.1,2,9–13
There has been, however, more than one pathway to contemporary epidemiology. A number of researchers have already stressed the importance of investigations into CHD.1,14–18 They have pointed to its influence, for example, on the epidemiological construction of the cohort study design.1,15 To understand CHD, moreover, epidemiologists were forced to think of disease as the outcome of multiple factors, a core idea of modern epidemiology. Then, to grasp the relationship of these many factors to CHD, researchers developed and applied ‘risk factor thinking’, the notion that certain behaviours or characteristics moderated the probability of disease. Epidemiologists may have borrowed that concept, in part, from the insurance industry, as William Rothstein14 has argued. By 1958, for example, 3 years before ‘risk factor’ entered the epidemiological vernacular, Jeremiah Stamler, a pioneering CHD epidemiologist, wrote of using ‘actuarial predictions of risk’ to identify those ‘with an inordinately high risk of developing clinical heart disease’. [p. 107 in Ref. (19); p. 418 in Ref. (20)].
With this paper, I aim to examine in greater depth the development of CHD epidemiology and its role in the evolution of post-war chronic disease epidemiology in the United States. To give the examination greater salience, this paper will compare the pathway represented by CHD epidemiology, particularly the Framingham study, with that of lung cancer. Historians have paid less attention to the differences between the two, which later merged into what we now call ‘risk factor epidemiology’. This paper will assess why CHD epidemiology in the post-war period almost uniformly began with cohort studies and focused primarily on clinical variables as putative aetiological factors; describe how epidemiologists sought to justify the creation of a non-infectious chronic disease epidemiology; and follow the emergence of ‘risk factor thinking’ within CHD epidemiology. Finally, given the discovery that CHD, unlike lung cancer, was the consequence of multiple variables, none of which were necessary or sufficient, each producing relatively small effects, this paper will discuss the relationship of ‘risk factor thinking’ to issues of causation and primary prevention in clinical medicine and public health. The paper will begin in the late 1940s, with the first CHD cohort studies, and end in 1970, when those maturing follow-up investigations had firmly established a pool of heart disease risk factors, and when clinical trials were beginning to demonstrate that modifying those factors reduced CHD morbidity and mortality.
|
Designing heart disease epidemiology |
|---|
|
TopAbstractDesigning heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
Post-war epidemiological research on lung cancer and cigarette smoking began with a series of case–control studies, most famously those of Ernst Wynder and Evarts Graham in the United States and of Austin Bradford Hill and Richard Doll in England.21,22 Doll and Hill's investigation, published in 1951, tested whether lung cancer patients, compared with controls matched on age, sex, and hospital of treatment, differed in terms of their smoking habits or in their exposure to atmospheric pollution. Only after completion of the case–control study, with critics sceptical of their claim of a causal relationship between smoking and lung cancer, did British and American scientists undertake research using a cohort design.10,23,24
Although a few case–control studies of smoking and lung cancer had been published previously, their weaknesses—small sample size and low response rate—made the choice of new case–control investigations appropriate.25–28 As Cuyler Hammond, one of the initiators of the US cohort study and a critic of the case–control approach wrote at the time, ‘Because of the cost and other difficulties, it is seldom possible to embark on a prospective study of a particular habit...unless there is good reason to suspect that it is important. For that reason, it is usually best to first investigate the matter by the retrospective method if warranted.’ [p. 182 in Ref. (6)] This view was reaffirmed by MacMahon and Pugh in their now classic textbook on modern epidemiology,29 and by several recent texts.30,31
It is, therefore, interesting and provocative to discover that early heart disease epidemiological investigations—initiated in the 1940s and 1950s in Framingham, Minneapolis, Albany, and Los Angeles—began, with rare exception,32 as prospective designs that developed into cohort studies. Why this apparent anomaly? Their ‘follow-up’ approach appears, in part, to have developed logically from contemporary understanding of CHD.
In the mid-1940s, epidemiological knowledge of CHD was spotty. Morbidity incidence and prevalence rates from unbiased samples were almost non-existent. Mortality statistics collected by the US government and the insurance industry revealed the weight of heart disease, variously defined.14,33–35 (Between 1900 and 1948, categories of heart disease classification in the US increased from 4 to 40, adding new distinctions like degenerative and hypertensive heart disease.14) By using the most inclusive category, cardiovascular-renal diseases, analysts of federal data showed mortality rates rising since the early 1920s, almost exclusively in middle-aged white males.19,36 Jerry Morris,37 studying ischaemic heart disease in the UK, found similar trends in England and Wales. By mid-century, federal statisticians estimated that arteriosclerotic heart disease accounted for >22% of deaths, making it the country's leading cause of mortality.38 Such statistics produced alarm and concern for action.
The determinants of CHD, unfortunately, were poorly understood. Increasingly, however, heart disease experts stressed the role of atherosclerosis in the development of CHD and its relationship to nutrition.39,40 That link was inferred, in part, from clinical-pathological surveys of the prevalence of atherosclerotic disease in economically underdeveloped and developed countries, which found, ‘life-span dietary patterns’, respectively, low and high in total calories, saturated fat, and cholesterol [p. 20 in Ref. (19)]. Where diet was low in cholesterol and fat intake, whether comparing regions or social classes, atherosclerosis and heart disease tended to be low as well.19 Evidence also mounted that blood cholesterol levels could be altered through dietary change.19 From such data, researchers began to argue that the observed rates of CHD were the consequence of mutable environmental factors and to attack the older, still-influential theory that the disease was the inevitable result of senescence.1,35,40
Based on their experiences, clinicians and insurance actuaries offered additional possible determinants of CHD.15,35,41 At a symposium, Paul Dudley White, the dean of academic cardiologists, provided a partial list of possible variables.
We practitioners know from our own experience that basic factors behind CHD are of very great importance. These include the possible influence of race, the sure influence of heredity, of sex and age.... We have acquired some information...of environmental factors.... [They] include stress and strain, about which we know relatively little, exercise, which some of us believe important, but just how and why we do not know, local customs and personal habits, for example the use of tobacco and alcohol, and proper programs of rest and relaxation, and diet.... We have talked for years about the relationship of the ways of life to the development of coronary atherosclerosis. [p. 2 in Ref. (42)]
Along with the results of ecological, cross-sectional, and small group experimental studies, clinical and actuarial variables could be incorporated into hypotheses—focusing especially on adult lifestyles43—once individual-level epidemiological studies of CHD were inaugurated after 1946. In the US, these investigations were supported by the federal government, which had so successfully mobilized scientific resources and personnel during the War.44–46 Post-1945, the government committed itself to basic research on chronic disease by substantially increasing funding to the National Institute of Health and by adding new branches such as the National Heart Institute in 1948. In that same year in the UK, Jerry Morris became director of the Medical Research Council's Social Medicine Unit, where he pioneered research on physical activity and CHD, studies that at first drew more American than British interest.47
Unlike lung cancer, where a single factor soon dominated the aetiological discussion, epidemiologists, given these preliminary studies, usually assumed that CHD was the consequence of multiple causes.48,49 But how did one study this disease, whose unpredictable onset was preceded by years of exposure to a complex of hypothetical factors? If, as a growing number of experts believed, CHD resulted from the accrual of atheromatous lesions,39 how could one study the relation between their progression and those factors? (This was especially difficult in that researchers could not directly observe or measure atherosclerosis in the living, absent a clinical event.) Because heart disease was hardly uniformly fatal, CHD epidemiologists, more frequently than their cancer counterparts, sought to incorporate morbidity as well as mortality as endpoints in their research. But how did one design a study that included all ‘true’ cases of atherosclerotic heart disease, avoiding the ‘clinician's illusion’ of drawing inferences about the course of the disease from the sickest patients?50,51
The solution adopted by investigators in Framingham, Minneapolis, and elsewhere was to follow longitudinally a selected population initially free of overt heart disease in order to observe the emergence of CHD in that group. The originators of the Framingham Study held that it was ‘designed to study the expression of CHD in a normal...population’,48 by which they meant theirs was to be a study of the natural history of CHD that included time before the onset of a clinical endpoint.51
|
Apologia for a new epidemiology |
|---|
|
TopAbstractDesigning heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
In its earliest years, Framingham sought precedent for using the demanding, relatively untried longitudinal methodology in the work of Sir James MacKenzie (1853–1925).49,51 MacKenzie, a pioneering British cardiologist, championed long-term follow-up of patients as a means of studying all stages of disease, especially its predisposing and early phases.52,53 Allusions to MacKenzie probably reflected the influence of Paul Dudley White, once his student,54 now advisor to Framingham, who strongly advocated a similar approach.55
In the earliest years of CHD study, ‘long-term follow-up’ meant incorporating clinical and laboratory examinations to observe and measure factors in each individual at baseline and subsequent intervals, then retrospectively comparing those who did and did not develop CHD to ascertain those factors ‘predisposing to’ CHD.48,49 (The notion of a ‘retrospective comparison’, which one today associates with case–control studies, may reflect the clinical background of the CHD researchers; namely, thinking in terms of patient series or case-finding, then secondarily seeking common predisposing factors.) As factors of interest, Gilcin Meadors, first head of the Framingham Study, proposed, like White, to measure diet, occupation, psychological status, body build, and use of stimulants.16 Ancel Keys, principal investigator of the Twin Cities Business and Professional Men's study, also begun in 1947, measured blood pressure, body weight, serum cholesterol, and other physiological characteristics,56 as did Albany, Los Angeles, and Framingham in its later design iterations.16,57–59 In further laboratory and population studies, Keys came to privilege dietary cholesterol and fatty acids as factors of risk.60,61 Conceivably, all these studies could end with a series of correlations and inferences. However, Framingham realized a prospective design would permit it to test a number of hypotheses ‘with respect to factors associated with the development of arteriosclerotic or hypertensive heart disease’.49 It would also allow researchers to measure the rate of disease progression for those who became abnormal.49
Hypotheses, many of them derived from clinical experience and focusing ‘on the body’, helped drive the selection of study factors, among them gender, age, weight, blood lipids, and pressure levels.18,41,62,63 Most of the first heart disease researchers were physicians, another reason why clinical variables predominated. Choice of study factors was governed, too, by what researchers believed they could validly and reliably measure. Ostensibly for that reason, the new National Heart Institute (NHI), to which Framingham was transferred from the Public Health Service in 1949, dropped psychological status, diet, and occupation from Framingham's initial roster of variables.64
It is interesting to note which factors were consciously excluded or considered of no importance. Either intentionally or blindly, perhaps, the first US researchers gave short shrift to the social dimensions of their populations. For example, despite extensive preliminary surveys of the community's characteristics in 1948,49,65 Framingham limited social items in its baseline exam to subjects' names, addresses, educational attainment, and nationalities.66 In Framingham, at least, omission of certain variables may have derived from the study design itself: the researchers depended upon the goodwill of their study subjects. Thomas Dawber, who became director of the Framingham Study in 1950, made it a policy to exclude items he thought would disturb or alienate participants. These included questions about sexual dysfunction, psychiatric problems, and—perhaps given the conservative, cold war times—queries about income and social class.59,66,67 (In this respect, Dawber's reluctance was more extreme than contemporaries like Caroline Bedell Thomas and Jeremiah Stamler, who included, respectively, psychological testing and occupational level in their CHD prospective studies.)20,68 Finally, Dawber, who shaped Framingham over its first decades, had little interest in the work of social science or public health.69,70 For him, epidemiology was ‘clinical investigation on a community level’.71
The communities that Dawber and his contemporaries studied were overwhelmingly white, underscoring the continued invisibility of other ethnic groups in most CHD and cancer epidemiological research. Los Angeles was the exception, with almost 16% of its participants being ‘Negro’ or ‘Oriental’.58 Both the Minnesota and Albany investigations were limited to men; CHD, at the time, was considered primarily a male disease.51 Women, over 50% of Framingham's sample, were of interest because they appeared to have a protective advantage which, if identified, might prove helpful to men.66,69
The linkage of multiple independent variables to a disease outcome was associated with an epidemiological theory current in the 1940s. In the Study's first publication, Thomas Dawber, wrote ‘epidemiology deals with "the fundamental questions as to where a given disease is found, when it thrives, where and when it is not found.... [I]n other words it is the ecology of disease."’ [p. 279 in Ref. (49)] Dawber here quotes John Paul, an epidemiologist at Yale, but he might also have referred to contemporaries like George Armstrong, Surgeon General of the Army, and John Gordon of Harvard, both of whom presented disease as a process generated by mutual interactions between individuals or communities (the host) and a complex of factors they subsumed under the rubrics of ‘agent’ and ‘environment’.72–74 The multiple aspects of the host included the dimensions of age, sex, race, genetically inherited susceptibility, and differences in social attitudes and psychological attributes.75 For these theorists, environment had its own dynamic interaction, including not only geographical and biological variables but also socioeconomic forces like housing, income, and occupation.75 The conceit of an ecology of disease—a vast, potentially totalizing, play of inter-relationships between and within agent, host, and environment—was initially applied to communicable disorders, providing a sophisticated analysis of their origins and development in populations.73 Perhaps most importantly for the period of the 1940s and 1950s, Gordon and others argued the ecological approach universally fitted all classes of disease. It provided a justification, as John Paul would maintain, for shifting epidemiology from an almost exclusive focus on infectious to non-infectious phenomena: ‘All forms of disease have their ecology or epidemiology.... There is an epidemiology of cancer, and an epidemiology of benzol poisoning, of alcoholism, and drug addiction, or at least there should be.’ [p. 53 in Ref. (76)]
For Paul and his colleagues,76 the metaphor of ecology implied a balancing of forces. Because the equilibrium attained was dynamic and fragile, disease occurred when the response of a host to other elements of the triad becomes maladaptive. Heart disease could be perceived as just such an imbalance, a discordance between a modern, industrialized way of life and a human body that evolved under very different conditions.77–79 Ancel Keys, writing after the Second World War, hypothesized that, compared with developed countries, significantly lower rates of cardiovascular disease and cancer occurred in Asia and Africa because of cultural differences, particularly diets rich in cereals and lower in calories.80 Ten years later, George Mann, an early investigator in (and later critic of) the Framingham Study, asked whether the ‘progress which man has made in altering his environment inadvertently introduced or exaggerated the agents which cause CHD?’81 Looking at a secular rise in serum cholesterol in a rural Georgian population, the leaders of the Evans County study ascribed those changes to ‘modernization’ of ‘predominantly agrarian communities’. [p. 544 in Ref. (82)] In brief, despite their ultimate dependence upon individual-level solutions, many epidemiologists conceived of CHD as a mass phenomenon; the origins of the ‘epidemic’ of heart disease lay in a broad, historic social transformation,83 the agent of which remained vague.84
The appeal to an ecological theory that transcended infectious disease or to an exemplary forebear like MacKenzie was part of the attempt by epidemiologists to justify their entry into chronic heart disease research. At the National Institutes of Health and the American Heart Association, for instance, there were powerful elements who questioned the scientific value of epidemiological studies, stressing instead basic laboratory research.17,66,85 During the 1950s, those who took the epidemiological path often assumed a defensive posture.
For example, the authors of Framingham's first publication in 1951 defined epidemiology for their readers, gave it a pedigree that included John Snow and Wade Hampton Frost, and cited its importance for studies of beriberi, pellagra, scurvy, and other non-infectious disorders. They faulted traditional methods of chronic disease research, i.e. clinical studies, for excluding those without diagnosed CHD from their purview, and actuarial analyses of mortality statistics, which shed only partial light on long-term disorders.49 Not only could epidemiology follow the emergence of CHD in an ‘unbiased’ population of normal composition, it could ‘explore certain relationships in health and disease which, with present technological methods, cannot be observed directly.’[p. 284 in Ref. (49)] With this allusion to the hypothesized underlying development of atherosclerosis, Dawber suggested a ‘black box’ approach that turned a weakness into a virtue.
At the same time that Dawber gave epidemiology special properties, he appealed to his fellow clinicians to view it as congruent with their approach to medical decision-making. Quoting Frost's observation that epidemiology ‘includes the orderly arrangement of facts into chains of inference which extend more or less beyond the bounds of direct observation’, [p. 289 in Ref. (49)] he claimed this ‘essence’ of epidemiology's method ‘has been used by the physician since the time of Hippocrates to arrive at clinical diagnosis.’ [p. 284 in Ref. (49)] As he would later in his career, Dawber sought here to reduce the suspicion physicians had of quantitative, population-based studies.71
Also in defence of epidemiology, Jeremiah Stamler, a physician and bench scientist turned epidemiologist, wrote an extensive piece on ‘Epidemiology as an Investigative Method for the Study of Human Atherosclerosis’.86 Using pellagra as a methodological model, Stamler stressed epidemiology's focus on patterns of disease in populations, especially those of person, place, and time, to find strong promising correlations between possible aetiological factors and CHD. He firmly advocated for epidemiology as an alternative to the laboratory for the study of disease. Only when he came to epidemiology's ultimate aim, determining causation, did he find the field insufficient in itself. Reflecting his own background and Goldberger's approach to pellagra, Stamler argued that for scientists to discover the aetiology of CHD, they needed to link ‘descriptive-analytic’ epidemiology with clinical and animal-experimental research. ‘The indispensability of ... shuttling back and forth among three basic methodologies is one of the crucial lessons to be learned from the history of medical advance. It is a basic research principle.’ [pp. 165–6 in Ref. (86)]
|
Proof is in the testing |
|---|
|
TopAbstractDesigning heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
The value of epidemiology as a method of researching heart disease was enhanced by the relative swiftness with which the cohort studies reported results. By 1957, a special issue of the American Journal of Public Health highlighted findings of a number of these investigations. The Los Angeles study, following a random sample of municipal employees since 1949, found a statistically significant association between blood pressure and CHD in men <54 years of age and suggested a relationship between heart disease and study subjects in the highest weight categories. Based on 3 years of data, an investigation centred in Albany, NY, again composed of civil servants, provided early evidence of trends: CHD rose with serum cholesterol and gross overweight. Framingham, with its supportive cadre of NIH statisticians, presented the most sophisticated results.16,59 Dawber and colleagues reported that arteriosclerotic heart disease showed a statistically significant association with age and male sex; in men 45–62, for whom they had sufficient data, heart disease was associated with hypertension, hypercholesterolaemia, and obesity. They also found that when two or three of these variables were elevated, risk of CHD increased greatly.
Over the next decade, Framingham and other contemporary studies would reproduce and extend their results; most importantly, they often corroborated each other's findings providing greater confidence in the outcomes achieved.71,87,88 At times, initially ambiguous or contradictory results were resolved as the studies matured and analyses became more precise. During much of the 1950s, for example, Framingham could demonstrate no statistically significant relationship between cigarette smoking and CHD, a finding at odds with that of the larger lung cancer cohort studies.24,59,89 By 1959, it could report a non-statistically significant association between cigarette smoking, myocardial infarction (MI), and sudden death. Shortly thereafter, the Albany and Framingham studies combined their data, providing sufficient statistical power to demonstrate a significant association between cigarettes and MIs or CHD-related deaths, both rising with increased cigarette consumption.90 Researchers also reported that ex-smokers developed the same low risk of CHD incidence as non-smokers. For yet other variables, results remained equivocal, but suggestive, during the 1950s and 1960s. Evidence that physical activity was protective against heart disease, for example, proved inconsistent.71,88,91,92 A few putative variables, suggested in the 1950s, later showed very weak, if any, relationship with CHD mortality, among them water softness and type A behaviour.93–97
By 1957, Framingham data also served to strengthen its case for an epidemiological approach to CHD.59 When all study deaths through 1956 were combined, Framingham ascertained that nearly half of all heart attacks were fatal, a striking finding.59 Of the 43 men who died of MIs over the study's first 4 years, almost a third were ‘sudden’ fatalities (defined as death within minutes). Because they usually perished before receiving acute medical care, most of these individuals would not have entered hospital statistics or clinical case series. In effect, case fatality rates for heart attacks were under-reported.
Dawber and his colleagues also uncovered cases of ‘silent’ infarcts, individuals lacking sufficient history of symptoms to have been diagnosed, but who were found, on the basis of the study's electrocardiographic (EKG) exams, to have suffered heart attacks. Almost 20% of new myocardial infarcts before 1957 were ‘silent’. Dawber used the existence of this group, and of those who died suddenly without previous symptoms of heart disease, to underscore the challenge of diagnosing CHD. Although this difficulty might have been used to question the validity of his investigation, Dawber employed it to justify carefully designed cohort studies whose case definitions were more clinically sensitive than was usual in medical pratice.59
|
Risk factors and the coronary prone individual: reading a new subject |
|---|
|
TopAbstractDesigning heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
Given CHD's high case fatality rate and the absence of curative treatment, epidemiologists strongly advocated for primary preventive methods.98 The audience they chiefly addressed were physicians, the dominant group in American health care, and people like themselves. Most of the investigators initiating CHD epidemiology, as already pointed out, were clinicians or laboratory scientists, not professional epidemiologists.85 William Kannel, Associate Medical Director of the Framingham Study under Dawber, later his successor, referred to his colleagues and himself as ‘little e’ epidemiologists, people without initial training or academic prominence in the field, who learned statistics and created epidemiology in the pursuit of research goals.99 In that respect, they differed from most of the researchers who designed the cigarette smoking and lung cancer studies, men like Austin Bradford Hill in Britain, head of the Medical Research Council's Statistical Research Unit, or, in the US, Cuyler Hammond, professor of biometry at Yale and Director of Statistical Research at the American Cancer Society.6,10,100
The message that Keys, Dawber, and their colleagues sent to physicians focused on the notion of susceptibility, easily understood by those trained in germ theory.56,98,101 Beginning around 1961, they advanced the metaphor of the ‘coronary-prone’ individual, the patient burdened by a greater probability of developing CHD. The assumption was that such individuals suffered more rapid advances in atherosclerosis, the existence of which would only be heralded by sudden death, MI, or angina pectoris. Because such events often marked advanced or fatal disease, physicians were urged to intervene early, based on their reading of patients' risky characteristics. Clues as to what to read was what the new epidemiology had to offer:
From the study of the characteristics of persons who later develop CHD and comparison with the characteristics of those who remain free of this disease, it is possible many years before any overt symptoms or signs become manifest, and at a time in life when a preventive program might be instituted with some reasonable hope of success, to put together a profile of those persons in whom there is a high risk of developing CHD. [p. 672 in Ref. (98)]
In this fashion, CHD epidemiology sought to turn itself into a handmaiden of medical practice, allaying the indifference with which many contemporary physicians may have held it.71,102 Epidemiologists appealed to doctors by creating a new medical ‘subject’ requiring their clinical intervention.15 Ideally, the subject, the ‘coronary-prone patient’,56,98 could be easily identified using normal office procedures.103 As an ideal type, the patient was an obese middle-aged male, a heavy smoker with high blood pressure and serum cholesterol, low vital capacity, and EKG abnormalities.77 In reality, the clinical picture was less clear. Most individuals presented with only a few such characteristics, which had then to be factored in with family history and other issues. In addition, various combinations of characteristics appeared to increase the probability of CHD, possibly doubling, tripling, or quadrupling it.101 Finally, epidemiological studies observed that, from ‘low’ to ‘high’, progressive levels of blood cholesterol or pressure were linked to increasing risk of disease. Contrary to physicians' training, there was no cut-off, below which either was ‘normal’ or ‘risk-free’.101 Consequently, almost every adult patient was a susceptible, particularly in modern cultures like the United States.104 The art and science of medicine was then to determine for patients, given their composite of characteristics and experiences, their degree of ‘risk’.15
The language of risk, specifically, the probability of developing CHD, given the presence and magnitude of certain personal attributes, permeated early Framingham publications. The idea of ‘risk’, if not the term, was implicit in the study's initial 28 hypotheses developed in 1949, e.g. hypothesis 8, ‘degenerative cardiovascular disease appears earlier and progresses more rapidly in persons who habitually use tobacco’.105 In their 1957 report, Dawber and coauthors used the term ‘risk’ at least 18 times, generally to mean ‘probability’, as in ‘there is an increased risk of [CHD] in persons with elevated cholesterol’.59 Over the following 2 years, Dawber began systematically referring to clinical and social characteristics possibly associated with the development of CHD—hypercholesterolaemia, hypertension, place of residence—as ‘factors’.106,107 In a July, 1961 publication, Kannel and Dawber employed ‘factors of risk’, meaning ‘characteristics of increased risk’ for a given disease. In the same article, they created a contracted phrase-‘risk factor’. This was done in a rather casual manner. Kannel, the lead author, claims to have soon forgotten that they had coined the term, or at least first applied it to epidemiology,14 until reminded some years later.99 In fact, in another article the same month, Dawber used the term ‘risk characteristic’ to express the same idea.98
In 1961, ‘risk factor’ seemed to capture a moment. It was a phrase that signalled a new approach to disease. It was more felicitous than terms like ‘actuarial predictions of risk’, a short-hand formula that could be applied aptly to each dimension of CHD epidemiological thought, a term that joined them together. In the absence of a known agent for CHD or the clinical ability to observe the development of underlying disease processes, ‘risk factors’ represented discrete characteristics linked to the probability of overt heart disease. Within a few years, ‘risk factors’ were also perceived as tools for inferring, in general, the degree of arterial atherosclerosis in populations and individuals, an external sign of an inner process.104
By seeking out ‘risk factors’, epidemiologists hoped to analyse inter-connections leading to an understanding of CHD's pathogenesis and natural history.103 Second, in countries experiencing an epidemic of heart disease, ‘risk factors’ could be used to describe the distribution of populations at greater risk of CHD. Finally, ‘risk factors’ had clinical application, defining the coronary-prone person for whom medical intervention was most advisable. Nevertheless, perhaps because epidemiologists took heart disease to be a mass phenomenon, the result of a shift in ‘ways of life’, individual responsibility or blame was almost entirely absent from their discussion of risk factors during the 1950s and 1960s.
For Dawber, Kannel, Stamler, and their colleagues, it was important to initiate a primary preventive programme.87,98 Tavia Gordon, a major NHI statistician who worked closely with Dawber and Kannel, recalled, in a rather critical 1966 memo to his NIH colleague, the mathematician, Jerome Cornfield, ‘The initial impulse for looking at variables like serum cholesterol as ’risk factors' came from Stamler, who was thinking of the Framingham results as a guide to interceptive treatment....I suspect, however, that Framingham usage has tended to carry everyone along too quickly into terminology tinted with the impulse of action.‘108 It is consistent with Gordon's recollection that the first articles on ’susceptibility' and treating ‘the coronary prone patient’ appeared almost simultaneously with the first published use of the term ‘risk factor’.98,101
To practicing physicians, Dawber and his colleagues argued (although avoiding any definitive statement) that heart disease epidemiology was successfully accumulating evidence showing that ‘the way of life of individuals affects their susceptibility to CHD’. [p. 88 in Ref. (87)] Environmental factors appeared to be most responsible; and environmental factors were open to intervention. In effect, maladaptation between host and environment could be adjusted by altering risk factors like serum lipids, blood pressure, cigarette smoking, and obesity.77,87
|
Causal uncertainty and epidemiological activism |
|---|
|
TopAbstractDesigning heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
In making their recommendations, these epidemiologists realized they were moving beyond their evidence. They based their conclusions on observational research, what had occurred in a population they were following, not a controlled experiment. Their work had shown a relationship between levels of behaviours or characteristics and outcome. Unlike the instance of cigarette smoking, there was no scientific confirmation that reducing factors in those with high blood pressure, cholesterol, or weight would lower heart disease rates, producing outcomes like those in persons with habitually lower numbers.77 Perhaps essential hypertension or hypercholesterolaemia had other underlying physical determinants that would remain untouched. There was also no way of knowing what effect changing one risk factor would have on other risk factors or outcomes. And there was no way of predicting whether the relationships between risk factors and CHD, based on retrospective examination of follow-up data, would hold true in the future. In effect, epidemiologists still lacked an understanding of the underlying causal chain of events that led to heart disease; the links they held, if links they were, were correlations only.
This did not prevent ‘risk factors’ from being understood as causal agents.99 Once ‘risk’ migrated from ‘risk of disease’ to a noun modifier of ‘factor’, it seemed to indicate, that blood pressure, for example, was a hazardous vector that led to heart disease. Such an understanding of ‘risk factor’ might be all the more probable in a clinical context, one in which physicians acted to prevent disease by profiling their patients and prescribing medications or behaviour modification. In a period when the germ theory of disease was dominant, it would be even more likely that individuals would understand a ‘risk factor’ to be analogous to a microbe. The very ambiguity of ‘risk factor’, resonating ‘cause’ while denoting ‘correlation’, gave it added imaginative power.
Publications by CHD epidemiologists during the 1950s and 1960s contained little formal discussion of causality, certainly compared with the literature on cigarette smoking and lung cancer. And contemporaries like Tavia Gordon can recall no ‘philosophical discourse, serious or otherwise’, on ‘cause’ by Stamler, the Framingham team or other epidemiologists.109
On occasion, some articles published during the 1960s presented arguments that parallel the criteria of judgement for causality produced in 1964 by the Surgeon General's Committee on Smoking and Health.110 In 1961, for example, Kannel et al.101 could claim their cohort design demonstrated that hypertension and hypercholesterolaemia, clinically suspected to be associated with CHD, preceded the appearance of the disease in humans (temporal relationship). They noted, too, that the gradient of risk of CHD increased with rising levels of cholesterol and blood pressure (dose–response). Finally, they observed their findings showed consistency, both with other longitudinal studies, and with current biological knowledge. Such arguments, however, were as interesting as they were rare.
Instead of pursuing proof of causality, CHD epidemiologists fell back on good clinical practice and the need for action. By combining medical knowledge and experience with newly garnered information about risk factors, doctors, like epidemiologists, could proceed to ‘the orderly arrangement of facts into chains of inference which extend more or less beyond the bounds of direct observation’.64,111 Designating those most prone to CHD, she/he should manage susceptible patients' risk factors, even absent proof of the efficacy of such actions. In so doing, doctors were treating heart disease like other disorders. Wrote Kannel, ‘If physicians were to wait for unquestioned proof of the efficacy of every therapeutic measure they are called upon to employ, they would never treat cancer, arthritis, or most of the degenerative and metabolic disorders. In short, they would be reduced to therapeutic nihilists.’ [p. 689 in Ref. (77)] Moreover, the risks here were small compared with the potential benefits.
Epidemiologists hoped that clinical interventions would be scientifically studied to test for efficacy.87 For the moment, CHD epidemiologists argued, their therapeutic recommendations, reducing blood pressure, weight, fats in diets, were consistent with standard medical practice.87
By 1964, the Surgeon General's Advisory Committee on Smoking and Health added its considerable symbolic and political weight to the position staked out by Dawber, Stamler, and their allies. In the same Report on Smoking and Health that authoritatively outlined criteria for judging causality in epidemiology, the Committee called for action against CHD risk factors despite the lack of demonstrated causal association. Discussing cigarette smoking, high blood pressure, high serum cholesterol, and excessive obesity, the Advisory Committee observed: ‘The causative role of these factors in coronary disease, though not proven, is suspected strongly enough to be a major reason for taking countermeasures against them’. [p. 38 in Ref. (110)] Both the Advisory Committee and CHD epidemiologists in general appeared to be reacting pragmatically, given the consistency of evidence across epidemiological studies, to the enormous toll of CHD morbidity and mortality.
Robert Aronowitz has pointed out that ‘risk factors’, as a concept, were eventually absorbed into mainstream medicine because they appeared to be objective, measurable biomedical additions to existing ideas and usage.15 Strong advocacy on the part of CHD epidemiologists and their allies to make primary prevention central to clinical practice may have been at least as important in the eventual acceptance of risk factors by clinicians. However, in the 1960s, despite the recommendations of the Surgeon General's Advisory Committee, many physicians were still reluctant to go along. They differed among themselves over the importance of treating essential hypertension,87 usually intervening only if a patient's blood pressure was very high or if complications had set in.81 They debated as well whether cholesterol level affected CHD, a scepticism fostered by disputes over what determined the level and what the pathogenesis of atherosclerosis was. Such questions persisted for some time.87,99
Multicentre studies, testing whether lowering one or more risk factors would reduce rates of heart disease, took off in the late 1960s and 1970s. Decades of clinical trials helped establish a broad consensus that, in addition to smoking cessation, modifying levels of serum cholesterol and blood pressure could have beneficial effects on CHD rates.112,113 In particular, the Lipid Research Clinics Coronary Primary Prevention Trial, following on the heels of other cholesterol lowering studies, demonstrated that for every 1% reduction in cholesterol, there was a corresponding 2% decline in CHD incidence.114,115 The results of these series of trials in many countries, reported in the late 1970s and 1980s, played a substantial role in eventually shifting the opinion of many physicians and policy makers toward favouring primary prevention interventions against cardiovascular disease.112
The epidemiological dispute over the association between cigarette smoking and lung cancer was marked by an acute onset and a fevered course, distinguished by sharp scientific debate and frequent re-examination. But by the mid-1960s, retrospective and prospective studies, supplemented by laboratory research, had developed a case for a causal relationship, buttressed by institutional support and a growing consensus among major scientific and political bodies in the US and abroad.3 In contrast, CHD epidemiologists ended the 1960s with a much weaker causal case. The long period of causal uncertainty, stretched into the 1980s, could have forestalled action. Clearly, it did not. Summoning data from field investigations, perhaps supplemented by clinical and laboratory studies, many in CHD epidemiology found reason to assume causality well before the ‘definitive’ trial results were in.116–119 The position of these epidemiologists was possibly a blend of science, faith, and a commitment to medical or public health intervention. One of them, William Kannel, would colourfully convey his mixture of caution and credo in the 1960s and 1970s, telling his audiences of physicians and public health professionals:
Not everything that looks like a risk factor is a causal condition.... If something looks like a goat, it is not necessarily a goat. But, if it smells like a goat, feels like a goat, tastes like a goat, sounds like a goat, you better start thinking in terms of goat.... [Y]ou better take action on it....There are many diseases that have been prevented by removing associated causes before we ever learned what was the fundamental biological relationship.99
KEY MESSAGES
|
|
Acknowledgments |
|---|
The author would like to thank four anonymous referees and Ronald Bayer, Dana March, Harry Marks, Sharon Schwartz, Ezra Susser, Mervyn Susser and Colin Talley for their helpful comments. This work was supported in part by the National Institutes of Health/National Library of Medicine (award 1-G13LM07932-01), the American Legacy Foundation, and the Professional Staff Congress-City University of New York (grant 663160035).
|
References |
|---|
|
TopAbstractDesigning heart disease...Apologia for a new...Proof is in the...Risk factors and the...Causal uncertainty and...References |
|---|
1 Susser M. Epidemiology in the United States after World War II: The evolution of a technique. Epi Rev 1985;7:147–77.
2 White C. Research on smoking and lung cancer: a landmark in the history of chronic disease epidemiology. Yale J Biol Med 1990;63:29–46.[Web of Science][Medline]
3 Kluger R. Ashes to Ashes: America's Hundred-Year Cigarette War, the Public Health, and the Unabashed Triumph of Philip Morris. New York: Vintage Books, 1997.
4 Glantz SA, Slade J, Bero LA, Hanauer P, Barnes DE. The Cigarette Papers. Berkeley: University of California Press, 1996.
5 Doll R. Tobacco: A medical history. Bull N Y Acad Med 1999;76:289–313.
6 Hammond CE. Cause and effect. In: Wynder E (ed.). The Biological Effects of Tobacco. Boston: Little, Brown and Company, 1955, pp. 171–96.
7 Berkson J. Smoking and lung cancer: some observations on two recent reports. J Am Stat Assoc 1958;53:28–38.[CrossRef][Web of Science]
8 Cornfield J, Haenszel W, Hammond EC, Lilienfeld AM, Shimkin MB, Wynder EL. Smoking and lung cancer: recent evidence and a discussion of some questions. J Natl Cancer Inst 1959;22:173–203.[Web of Science][Medline]
9 Brandt AM. The cigarette, risk, and American culture. Daedalus 1990;119:155–75.
10 Berlivet L. ‘Association or causation?’ The debate on the scientific status of risk factor epidemiology, 1947–c.1965. Clio Med 2005;25:39–74.
11 Talley C, Kushner H, Sterk C. Lung cancer, chronic disease epidemiology, and medicine, 1948–1964. J Hist Med 2004;59:329–74.
12 Parascandola M. Two approaches to etiology: the debate over smoking and lung cancer in the 1950s. Endeavour 2004;28:81–86.[CrossRef][Medline]
13 Parascandola M. Epidemiology: Second rate science? Public Health Rep 1998;113:312–20.[Web of Science][Medline]
14 Rothstein WG. Public Health and the Risk Factor, A History of an Uneven Medical Revolution. Rochester, New York: University of Rochester Press, 2003.
15 Aronowitz RA. The social construction of coronary heart disease risk factors. In: Aronowitz RA (ed.). Making Sense of Illness: Science, Society and Disease. New York: Cambridge University Press, 1998, pp. 111–44.
16 Oppenheimer GM. Becoming the Framingham study: 1947–1950. Am J Pub Health 2005;95:602–10.
17 Levy D, Brink S. A Change of Heart: How the People of Framingham, Massachusetts Helped Unravel the Mysteries of Cardiovascular Disease. New York: Knopf, 2005.
18 Kuh D, Davey Smith G. The life course and adult chronic disease: an historical perspective with particular reference to coronary heart disease. In: Kuh D, Ben-Shlomo Y (eds). A Life Course Approach to Chronic Disease Epidemiology, 2nd Edn. Oxford: Oxford University Press, 2004, pp. 15–37.
19 Katz LN, Stamler J, Pick R. Nutrition and Atherosclerosis. Philadelphia: Lea & Febiger, 1958.
20 Stamler J, Lindberg HA, Berkson DM, Shaffer A, Miller W, Poindexter A. Prevalence and incidence of coronary heart disease in strata of the labor force of a Chicago industrial corporation. J Chronic Dis 1960;11:405–20.[CrossRef][Medline]
21 Wynder EL, Graham EA. Tobacco smoking as a possible etiologic factor in bronchiogenic carcinoma. JAMA 1950;143:329–36.
22 Doll R, Hill AB. Smoking and carcinoma of the lung. Preliminary report. BMJ 1950;2:739–48.
23 Doll R, Hill AB. The mortality of doctors in relation to their smoking habits, a preliminary report. BMJ 1954;1:1451–55.
24 Hammond EC, Horn D. The relationship between human smoking habits and death rates. A follow-up study of 187,766 men. JAMA 1954;55:1316–28.
25 Müller FH. Tabakmissbrauch und lungencarcinoma. Zeitschrift für Krebsforschung 1939;49:57–85. [Abstract in English: Abuse of Tobaccco and Carcinoma of Lungs. JAMA 1939;113:1372.]
26 Schairer E, Schöniger E. Lung cancer and tobacco consumption. Int J Epidemiol 2001;30:24–27. [trans. from Lungenkrebs und Tabakverbrauch. Zeitschrift für Krebsforschung 1943;54:261–69.]
27 Wassink WF. Onstaansvoorwarden voor Langkanker. Nederlands Tijdschrift voor Geneeskunde 1948;92:3732–47.[Medline]
28 Doll R. The first reports on smoking and lung cancer. In: Lock S, Reynolds LA, Tansey EM (eds). Ashes to Ashes: The History of Smoking and Health. Amsterdam-Atlanta, GA: Rodopi, 1998, pp. 130–40.
29 MacMahon B, Pugh TF. Epidemiology, Principles and Methods. Boston: Little, Brown and Co., 1970.
30 Gordis L. Epidemiology. Philadelphia: WB Saunders Company, 1996.
31 Rothman KJ. Modern Epidemiology. Boston: Little, Brown and Company, 1986.
32 Gertler M, White PD. Coronary Heart Disease in Young Adults. Cambridge: Harvard University Press, 1954.
33 Moriyama IM, Gover MI. Statistical studies of heart disease, I. Heart disease and allied causes of death in relation to age changes in the population. Public Health Rep 1948;63:537–45.[Web of Science][Medline]
34 Woolsey TD, Moriyama IM. Statistical studies of heart disease, II. Important factors in heart disease mortality trends. Public Health Rep 1948;63:1247–73.[Web of Science][Medline]
35 Dublin LI. Health Progress, 1936–1945. New York: Metropolitan Life Insurance Company, 1948.
36 Moriyama IM, Woolsey TD, Stamler J. Observations on possible factors responsible for the sex and race trends in cardiovascular-renal mortality in the United States. J Chronic Dis 1958;7:401–12.[CrossRef][Medline]
37 Morris JN, Dale RA. Epidemiology of coronary atherosclerosis. Proc R Soc Med 1955;48:667–72.[Medline]
38 National Heart Institute. Circular number 7. October 1951. Cited in Katz LN, Stamler J. Experimental Atherosclerosis. Springfield, IL: Charles C. Thomas, 1953.
39 Marks HM. The Progress of Experiment: Science and Therapeutic Reform in the United States, 1900–1990. Cambridge, England: Cambridge University Press, 1997.
40 Katz LN, Stamler J. Experimental Atherosclerosis. Springfield, IL: Charles C. Thomas, 1953.
41 White PD. Heart Disease. 3rd edn. New York: The Macmillan Co, 1944.
42 White PD. The cardiologist enlists the epidemiologist. Am J Pub Health 1957;47:1–3.
43 Kuh D, Davey Smith G. When is mortality risk determined? Historical insights into a current debate. Soc Hist Med 1993;6:101–23.
44 Swain DC. The rise of a research empire: NIH, 1930–1950. Science 1962;138:1233–35.
45 Strickland SP. Politics, Science, and Dread Disease. Cambridge, MA: Harvard University Press, 1972.
46 Fox DM. The politics of the NIH extramural program, 1937–1950. J Hist Med Allied Sci 1987;42:447–66.
47 Davey Smith G. A conversation with Jerry Morris. Epidemiology 2004;15:770–73.[Medline]
48 Meadors G. to Boone B. July 19, 1947. Papers of the National Heart, Lung, and Blood Institute. Folder: Epidemiology Correspondence 1947.
49 Dawber TR, Meadors GF, Moore FE. Epidemiological approaches to heart disease: The Framingham study. Am J Pub Health 1951;41:279–86.
50 Cohen P, Cohen J. The clinician's illusion. Arch Gen Psychiatry 1984;41:1178–82.
51 Meadors G. Proposed study of the epidemiology of cardiovascular disease. ND [1948]. Papers of the National Heart, Lung and Blood Institute. Folder: 1948–1949 Heart Disease Epidemiology Study, Dr. Meadors.
52 MacKenzie J. The Future of Medicine. London: Oxford University Press, 1919.
53 MacKenzie J. The Basis of Vital Activity; Being a Review of Five Years' Work at the St. Andrews Institute for Clinical Research. London: Faber and Gwyer, 1926.
54 White PD, My Life and Medicine. Boston: Gambit Incorporated, 1971.
55 White PD, Donoven H. Hearts: Their Long Follow-Up. Philadelphia: WB Saunders Co., 1967.
56 Keys A, Taylor HL, Blackburn H, Brozek J, Anderson JT, Simonson E. Coronary heart disease among Minnesota business and professional men followed 15 years. Circulation 1963;28:381–95.
57 Doyle JT, Heslin AS, Hilleboe HE, Formel PF, Korns RF. A prospective study of degenerative cardiovascular disease in Albany: report of three years experience. I. ischemic heart disease. Am J Public Health 1957;4:25–32.
58 Chapman JM, Goerke LS, Dixon W, Loveland DB, Phillips E. Measuring the risk of coronary heart disease in adult population groups. The clinical status of a population group in Los Angeles under observation for two to three years. Am J Public Health 1957;47:33–42.
59 Dawber TR, Moore FE, Mann GV. Coronary heart disease in the Framingham study. Am J Public Health 1957;47:4–24.
60 Keys A. Human atherosclerosis and the diet. Circulation 1952;5:115–18.[Web of Science][Medline]
61 Keys A. Atherosclerosis: a problem in newer public health. J Mt Sinai Hosp N Y 1953;20:118–39.[Medline]
62 Levine SA, Brown CL. Coronary thrombosis: its various clinical features. Medicine 1929;8:245–418.
63 Levine SA. Clinical Heart Disease. 3rd edn. Philadelphia: WB Saunders Co., 1945.
64 Memo, Moore F. to Van Slyke CJ. August 26, 1949. Papers of the National Heart, Lung and Blood Institute. Folder: Framingham-1949–50.
65 Gordon T, Kannel WB. The Framingham Study, An Epidemiological Investigation of Cardiovascular Disease. Bethesda: National Institute of Heart, Lung and Blood Institute. June 1968. (Unpublished report.)
66 Dawber TR. The Framingham Study. The Epidemiology of Atherosclerotic Disease. Cambridge, MA: Harvard University Press, 1980.
67 Schrecker E. No Ivory Tower: McCarthyism and the Universities. New York: Oxford University Press, 1986.
68 Thomas CB. The Precursors of Essential Hypertension and Coronary Heart Disease: Characteristics of the Johns Hopkins Medical Students, Collected Papers. Vol. II. Baltimore: The Johns Hopkins School of Medicine, 1962.
69 Dawber TR. Personal communication, May 23, 2002.
70 Dawber TR. To Watt J. April 4, 1956. Papers of the National Heart, Lung, and Blood Institute. Folder: Zukel Files, Framingham Evaluations: Correspondence.
71 Dawber T. Risk factors in young adults: The lessons from epidemiological studies of cardiovascular disease-Framingham, Tecumseh, and Evans County. J Am Coll Health Assoc 1973;22:84–95.[Web of Science][Medline]
72 Armstrong, GE. How the Army practices the epidemiology of health. In: Galdston I (ed.). The Epidemiology of Health. New York: Health Education Council, 1953, pp. 74–87.
73 Gordon JE, Augustine DL. Tropical environment and communicable disease. Am J Med Sci 1948;216:343–57.[Medline]
74 Gordon JE. The epidemiology of accidents. Am J Pub Health 1949;39:504–15.
75 Gordon JE. The World, the flesh and the devil as environment, host and agent of disease. In: Galdston I (ed.). The Epidemiology of Health. New York: Health Education Council, 1953, pp. 60–73.
76 Paul J. Epidemiology. In: Green DE, Knox WE (eds). Research in Medical Science. New York: The Macmillan Co., 1950, pp. 52–62.
77 Kannel WB, Kagan A, Dawber TR, Revotskie N. Epidemiology of coronary heart disease. Implications for the practicing physician. Geriatrics 1962;17:675–90.[Web of Science][Medline]
78 Dawber TR. Heart attack—what's the risk? Today's Health 1965;43:46–47;62–64.
79 Blackburn H. The evolution of mass atherosclerotic diseases: the rationale for primary prevention. Ann Clin Res 1988;20:4–9.[Medline]
80 Keys A. Nutrition in relation to the etiology and the course of degenerative diseases. J Am Diet Assoc 1948;24:281–85.[Medline]
81 Mann GV. The epidemiology of heart disease. Am J Med 1957;23:463–80.[CrossRef][Web of Science][Medline]
82 Tyroler H, Hames CG, Krishan I, Heyden S, Cooper G, Cassel JC. Black-white differences in serum lipids and lipoproteins in Evans County. Prev Med 1975;4:541–49.[CrossRef][Web of Science][Medline]
83 Stamler J. Cigarette smoking and atherosclerotic coronary heart disease. Bull N Y Acad Med 1968;44:1476–94.[Medline]
84 Krieger N. Epidemiology and the web of causation: has anyone seen the spider? Soc Sci Med 1994;39:887–903.[CrossRef][Web of Science][Medline]
85 Blackburn H, Epstein FH. A history of the American Heart Association (AHA) Council on Epidemiology and Prevention: the pursuit of epidemiology within the AHA. Prehistory and early organization. Epidemiology and prevention. Council Newsletter 1995; No. 52:1–35.
86 Stamler J. Epidemiology as an investigative method for the study of human atherosclerosis. J Natl Med Assoc 1958;50:161–200.[Medline]
87 Dawber TR, Kannel WB. Application of epidemiology of coronary heart disease to medical practice. Mod Med 1962;30:85–101.
88 Dawber TR, Thomas HE. Prophylaxis of coronary heart disease, stroke, and peripheral atherosclerosis. Ann N Y Acad Med 1968;149:1038–57.[Medline]
89 Doll R, Hill AB. Lung cancer and other causes of death in relation to smoking. BMJ 1956;2:1071–81.
90 Doyle JT, Dawber TR, Kannel WB, Heslin AS, Kahn H. Cigarette smoking and coronary heart disease: combined experience of the Framingham and Albany studies. N Engl J Med 1962;266:796–801.[Web of Science]
91 Morris JN, Heady JA, Raffle PA, Roberts CG, Parks JW. Coronary heart disease and physical activity of work. Lancet 1953;2:1053–57.
92 Kahn HA. The relationship of reported coronary heart disease mortality to physical activity of work. Am J Pub Health 1963;53:1058–67.[Medline]
93 Morris JN, Crawford MD, Heady JA. Hardness of local water supplies and mortality from cardiovascular disease in the County Boroughs of England and Wales. Lancet 1961;1:860–62.[CrossRef][Medline]
94 Pocock SJ, Shaper AG, Cook DG et al. British Regional Heart Study: variations in cardiovascular mortality, and the role of water quality. BMJ 1980;280:1243–49.
95 Shaper AG, Elford J. Regional variations in coronary heart disease in Great Britain: risk factors and changes in environment. In: Marmor M, Elliot P (eds). Coronary Heart Disease Epidemiology: From Aetiology to Public Health. Oxford: Oxford University Press, 1992, pp. 127–39.
96 Friedman M, Rosenman RH. Association of specific overt behavior pattern with blood and cardiovascular findings. JAMA 1959;169:1286–97.
97 Theorell T. The psycho-social environment, stress, and coronary heart disease. In: Marmor M, Elliot P (eds). Coronary Heart Disease Epidemiology: From Aetiology to Public Health. Oxford: Oxford University Press, 1992, pp. 256–73.
98 Dawber TR, Kannel WB. Susceptibility to coronary heart disease. Mod Concepts Cardiovasc Dis 1961;30:671–77.[Medline]
99 Kannel WB. Personal communication. July 11, 2002.
100 Armitage P. Fisher, Bradford Hill, and randomization. Int J Epidemiol 2003;32:925–28.
101 Kannel WB, Dawber TR, Kagan A, Revotskie N, Stokes J. Factors of risk in the development of coronary heart disease—six year follow-up experience. The Framingham Study. Ann Intern Med 1961;55:33–50.
102 Keys A. Editorial: The risk of coronary heart disease. Circulation 1961;23:805–12.
103 Kannel WB, Dawber TR, McNamara PM. Detection of the coronary-prone adult: The Framingham Study. J Iowa Med Soc 1966;56:26–34.[Medline]
104 Kannel WB, Castelli WP, McNamara PM. The coronary profile: 12-year follow-up in the Framingham study. J Occup Med 1967;9:611–19.[Web of Science][Medline]
105 National Heart Institute. Heart Disease Epidemiology Study, Manual of operations. 1949. Papers of the National Heart, Lung and Blood Institute. Folder: Historical for Soc 1,2.
106 Dawber TR, Kannel WB. An epidemiologic study of heart disease: The Framingham Study. Nutr Rev 1958;16:1–4.[Web of Science][Medline]
107 Dawber TR, Kannel WB, Revotskie N et al. Some factors associated with the development of coronary heart disease. Am J Public Health 1959;4:1349–56.
108 Gordon T to Truett J, Cornfield J. Nov. 1, 1966. Papers of the National Heart, Lung and Blood Institute. Folder: July–December 1966 Correspondence.
109 Gordon T. Personal communication. February 23. 2002.
110 US Department of Health, Education and Welfare. Surgeon General's Report, Smoking and Health: Report of the Advisory Committee to the Surgeon General of the Public Health Service. PHS Publication No. 1103. Washington, DC: Government Printing Office, 1964, pp. 19–20.
111 Frost WH. Introduction. In: Snow J. Snow on Cholera. New York: The Commonwealth Fund, 1936, p. ix.
112 Rifkind BM. From observation to policy: cholesterol. In: Marmor M, Elliot P (eds). Coronary Heart Disease Epidemiology: From Aetiology to Public Health. Oxford: Oxford University Press, 1992, pp. 435–50.
113 Fraser GE. Preventive Cardiology. New York: Oxford University Press, 1986.
114 Lipid Research Clinics Program. The Lipid Research Clinics Coronary Primary Prevention Trial Results. I. Reduction in incidence of coronary heart disease. JAMA 1984;251:351–64.
115 Lipid Research Clinics Program. The Lipid Research Clinics Coronary Primary Prevention Trial Results. II. The relationship in incidence of coronary heart disease to cholesterol lowering. JAMA 1984;251:365–74.
116 Stamler J. Lectures on Preventive Cardiology. New York; London: Grune & Stratton, 1967.
117 Tibblin G, Keys A, Werkö L. Preventive Cardiology. Stockholm: Almqvist & Wiksell Förlag, 1972.
118 Intersociety Commission for Heart Disease Resources. Primary prevention of the atherosclerotic diseases; report of Inter-Society Commission for Heart Disease Resources. Circulation 1970;42:A55–95.
119 Blackburn H. The development of cardiovascular disease epidemiology. In: Holland W (ed.). The Development of Modern Epidemiology. Oxford: Oxford University Press (In press).
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. EBRAHIM The future of modern epidemiology: genetics, methods, and history Int. J. Epidemiol., June 1, 2006; 35(3): 511 - 512. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||