Professional Documents
Culture Documents
S INCE the Fall of 1957, a long-term study iien developing clinical coronary disease in
of coronary heart disease has been in the course of the study were to be compared
progress at the Hawthornie Works of the to those men not so afflicted. The Hawthorne
WTesterril Electric Company in Chicago under Works of the Western Electric Conmpany was
the auspices of the IUniversity of Illinois Col- selected both because of its reasonable prox-
lege of Medicine and Presbyterian-St. Luke 's imity to the Medical Center, and because of
Hospital. The study was unidertaken in the the expressed interest of its Medical Depart-
belief that coronary heart disease was a dis- rient in such a project. Over 20,000 persons
ease resulting from the interplay of multiple were einployed in the Hawthorne Works in
factors and that there was ineed to delineate 1937, imainly in clerical and light assembly
these factors further. The data presented work, although a significant group was emn-
herein represent the initial compilation of ployed at positionls requiring vigorous labor.
data centering on this problemrl; a discrimi- The group was comuposed chiefly of second-
niate funietioln analysis is to be undertaken anid third-generationi Americans of Polish and
iiext. The report which follows has been made Bohemian ancestry.
possible through the efforts of a large group All mnen aged 40 to 55 who had been em-
of physicians aiid other scientists who have ployed at the Works for 2 or muore years (N
volunteered their time in the project to per- 3,397) were assigoned an idenitificationi nunmber
Downloaded from http://ahajournals.org by on March 17, 2024
init the accumulation of a large body of data, by a ranidomnizing process. Melubers of the
a portion of whieh is presenited below. randomly selected group were then invited to
Acknowledgment must also be made to the participate inl the study participation was
W\estern Electric Company, which has beenl entirely voluntary-by beinig called from the
nost cooperative and helpful throughout, to random list. In this process the medical rec-
its employees who have participated in the ords of the Company were inspected and those
study, and to the agencies and doniors par- men for whom there was evidence of angina
tially recognized below* who have finaneed pectoris by description or of myocardial in-
the survey since its inception. faretion confirmed by history anid electrocar-
Plan of Study diographic chaniges were excluded, as were
It was planned to study the eharacteristies
mnen in certain additioiial categories (deaths
of a nlale population believed to be free from:i subsequelnt to preparation of the list, trans-
clinieal coronary disease through annual inter- fers, permanient disabilities, leaves of ab-
views an-d examinations carried oni over a peri- senee); this group numbered 127 men.
od of at least 3 years. The attributes of those Invitations to participate were sent by letter
after preliminary discussion and explanation
Fronm the Department of Medicine, University of at the top management and supervisor level,
Illinois College of Medicilne, Chicago, Illinois.
*Gratitude is expressed to the Aniierican Heart As- and publicity through the plant bulletin and
sociation, the Chicago Heart Associationl, the Illilnois company magazine. It was planned to invite
Heart Association, The United States Public Health a sufficient number to provide a population
Service, The Otho S. Sprague Foundation, The Re- of approximately 2,000 men. Of the 3,102
search and Education Coiniiittee of The Presbyteriani-
St. Luke's Hospital, The Illini Founidation, and muany
men asked to participate, 2,080 or 67 per cent
private donors. Particular thainks are dlue to the gein- of those iiivited agreed to participate and
erous help of Mrs. Tiffany ]Blake. wvere exanuilned the first year. Of this group,
20 Circulation, Volume XXVIII, July 1963
CORONARY HEART DISEASE 21
44 were excluded from subsequent anialysis and physical examination were modeled after
because the first examiniation indicated the that employed in the Framinghanm Project.2
presence of coronary heart disease. Forty- The dietary data were obtained by a 1-hour
seven others were not included in the analysis interview with a trained dietitian.3 Supple-
because of special problems in follow-up un- mentary information was obtained from
related to the eardiovascular system. The base forms completed by the wives, and by inter-
population therefore available at the end of views with a sample of the wives. Plastic food
the first year for analysis amounted to 1,989 models were employed to assist in portion
men. At the end of 4 years, 38 mnen had died estimates, as were data regarding food pur-
and 105 men had withdrawn from the project chases and preparationi obtainied from the
or left the Company. Over 94 per cent of the Company cafeteria.
survivors of the original group had thus re- Physical activity on the job was assessed
mained in the study. The group labeled "non- by reference to a Company eode, the validity
coronary " in subsequent references will be of which was confirmed for work in the shops
the base population less men not followed, by oxygen studies on the job in a small sample
the new coronary cases, and in certain in- of 36 men. The code included five grades of
stances a small number of men for whom physical activity in the shops and four of en-
adequate data were lacking. After the first vironmental factors in the offices. Physical
year the men were, in general, examined on activity off the job was estimated by use of a
an anniversary-month basis, although in many questionnaire administered by a trained in-
inistances, this was not possible due to conm- terviewer fromn the personnel department of
mitments at work, travel, or illniess. the Company.
No therapeutic suggestions of any kinid A routine 7-foot chest x-ray was taken, as
were given to the men participating in the were an electrocardiogram with a direct-writ-
study, and the study itself was labeled a ing instrument with standard 12 leads plus
Downloaded from http://ahajournals.org by on March 17, 2024
Health Survey to draw attention away fron lead V4R, urinalysis, blood hemoglobin level
the cardiovascular system. (with a colorimeter), and blood cholesterol
The diagnosis of angina pectoris was made (Abell-Kendall method). Body fatness was
by two or more internists in each ease and measured with a skin-fold caliper. Somato-
was based upon published criteria.1 The diag- typing was performed by standard measure-
nosis of myocardial infaretioni was made with ments plus photographs in three views of each
use of the clinical history plus characteristic man clothed in a brief supporter, with typing
electrocardiographic findings.1 Where differ- by the Sheldon method.4 ljipoprotein lipase
ence of opinion occurred, decisioln as to diag- measuremenlts were made the third year by
nosis rested with the senior investigator, who the technic of Pilgeram.5
undertook an additional interview and phys- Blood pressures were taken at the begin-
ical examination. The assumption of death ning of each physical examination with the
from coronary disease was made on the basis participan-t seated; left and right arm meas-
of the most reasonable conclusion to be drawn urements were found to be similar on statis-
from information obtained from the family, tical analysis and the left arm readings are
physicians' and hospital records, death cer- referred to below.
tificates, anid coroners' reports. Unfortu- Psychologic studies were also made and will
nately, postnmortem examinations were rarely be reported separately.
made. For the purpose of this report, the diagno-
A family history was provided by each par- sis of angina pectoris is given wheni a man
ticipant, who completed a form with the help exhibited only this manifestation of coronary
of his family. Annual additions to the history disease in the absence of aortic stenosis or
were made by interview. The medical history other obvious underlying factors. The diag-
Circulation, Volume XXVIII, July 1963
22 PAUL ET AlI.
nosis of myocardial infaretion is given when respectively. The total mortality for the 2,168
myocardial infarction was adequately docu- nmen in the group of 5,397 whose numnber
rnented whether or not angina had been pres- never came up for consideration was 3.3 per
ent on previous examinations (this occurred cent, which compares with a total mortality
in onie of 28 infarets). The diagnosis of death of 3.1 per cent for the 3,102 invitees plus the
from coronary disease is listed regardless of 127 men referred to previously who were con-
previous documentation of angina pectoris or sidered but not called.
myocardial infaretion, providing adequate Historical Data
evidenee for such listingf was available. A Family History
history of sudden death was not considered
There was no significant difference betweeni
adequate evidenee by itself.
the coronary and noneoronary groups when
In the data presented below, differences
analyzed for the historical evidence of both
between the groups are investigated with re-
parents dying over 70 years of age, over 65
spect to data obtained from the first exam- years of age, or dying over 60 years of age.
inationl unless otherwise stated. The ages of living parents were adjusted to
Results expected age at death aceordiing to the Ameri-
New Coronary Cases canl Experienee Mortality Table. There was
After 4 years anid 3 months of the study, ino significant differenee between the average
88 meii had beeii elassified as having devel- age of death of the mother of these two
oped clinieal coronary heart disease as fol- groups, but there was with the age of the
lows: angina pectoris, 47 men; myocardial father (the age of the fathers in the noneor-
infaretion, 28 men; deaths from coronary onary group exceeded that of the coronary
disease, 13 men. Approximately one new case group by 3.4 years, on the average). Infor-
per 100 men per year was thus diagnosed. mation on the grandparents' age at death did
Downloaded from http://ahajournals.org by on March 17, 2024
These coronary men had a mean age 6 months not appear to be a factor associated with the
older than the men in the noneoronary group, incidence of coronary disease.
with a distribution curve skewed somewhat History of Chest Discomfort
to the right. It was of considerable interest to follow the
The group under study was probably not men who at the time of the first examinationl
more or less coronary prone than the men who gave a history of soime type of chest discom-
were invited but did not accept or respond. fort that was not considered adequate for a
This is shown by the fact that at the end of diagnosis of angina pectoris. Table 1 shows
4 years and 5 months, 21 of the 1,022 who a significantly higher development of clinical
declined or did not respond had died from coronary disease in those men who gave such
all causes and 45 of the 2,080 who had ac- a history, as contrasted with those who did
cepted had died from all causes. The mor- not give such a complaint (p < 0.001).
tality percentages are 2.1 and 2.2 per cent History of Cough and Shortness of Breath
A chronic cough was stated to be present
Table 1 by 25 per cent of the men in the coronary
History of "Noncardiac" Chest Discomfort group (88 cases) as contrasted with 12 per
Noncoronary cases Coronary cases cent of the noneoronary men (1,778 cases).
(1764) (87)
Likewise, 18 per cent of the men in the coro-
Yes 16 31a/c
nary group indicated shortness of breath on
Co
No 84%l 69%c
effort as compared with 11 per cent of the
Data relating to history of "'noneardiac " chest
discomfort as obtained at the time of the initial
men in the noneoronary category. Both of
exasninationi anid its relation to subsequent recogni- these differences are significant (p < 0.001
tiosi of clinical coronary heart disease. for the former and < 0.025 for the latter).
Circulation, Volume XXVIII, July 196.q
CORONARY HEART DISEASE 2M3
Table 2
Somatotype Distribution in Noncoronary and Coronary Groups
Noncoronary cases Coronary cases
Category (1790) (87)
Obs./Exp. Obs./Exp.
Eiido. dominanit 253/259 19/13
Meso. dominant (endo. > ecto.) 742/743 37/36
Meso. dominant (enido. = ecto.) 156/153 4/7
Meso. dominant (endo. < ecto.) 124/120 2/6
Eiido. = Meso. 217/221 15/11
Meso. = Ecto. 101/99 3/5
Ecto. dominiant 197/195 7/10
Somatotype distribution in the noneoronary and coronary groups, listing observed
and expected numbers. In the mesomorphic dominant category are listed the secondary
dominant or equal components.
-~~~~~~~~4h R
disease.
Laboratory Observations
250-
Hemoglobin Levels
No significant difference was shown among --,0958
Cholesterol Levels
220-
As anticipated, a definite relationship was
demonstrated between higher blood choles- WINTER SUMMER
terol levels as found the first year, and sub- SPRING AUTUMN
below 240 mg. per cent. A third observationi in the cholesterol de-
A second and perhaps more important ob- terminations supports the work of Thomas et
servation was the demnonstration of a signifi- al.8 These authors founld a seasonal variation
can-t difference in behavior of the coronary
in cholesterol levels in a small group of pris-
oilers who were studied throughout the year.
subgroups. A comparison of cholesterol levels
in the noneoronary cases with the levels of As will be seen in figure 2, our data show a
47 angina cases, the 28 infaret cases and 13
consistent seasona[ variationi, the highest levels
menl who died of coronary disease, is indicated
beinig encountered in the winter with the
in table 6. The stepwise progression of the lowest in the springfo or sum-mer. In this
mean cholesterol levels is clearly seen, fron
study, cholesterol determinations were not
made on each man for each season; rather,
the lowest levels in the men as vet not iden-
tified as having coronary disease to the high- the means for each seasoni were determined
with the data procured from annual examina-
tions. The men were often but not invariabl-
Table 5 examined in the same seasoni each year. That
Cholesterol Levels in Noncoronary and Coronary
the variatioll is not identical with the seasoni
Groups
variation in blood pressure recordings can be
Noncoronary cases Coronary cases seen by comparisoni with figure 1. Our dieti-
Mg. % (1794) (88)
tians have also analyzed seasonal dietary
100-204 20C 7%
205-239 2 9 thf
changes and find that in this population, they
240-256 15%G 15cc are apparently insignificant although more
257-274 11%o 17% beer is consumed in the months with higher
275-309 14%l 22% temperatures and somewhat more fruit is in-
> 310 12Cc 18%Co cluded in the diet. There is no change in the
Circulation, Volume XXVIII, July 1963
26 PAULI ET AL.
Table 6
Cholesterol Levels in Noncoron-ary asnd Coronary Subygroups
Noncoronary cases Angina Infarct Deceased
(1794) (47) (28) (13)
Mean 247.3 263.6 7V
'- _284.5
K 256 mIg. %C 63Xc- 48%'c 43( c 277%
3
> 256 mng. Cc 37'%o 5 2 Vc 57cc 7 7 -'I
men are more physically active in the sumnler observations has shown as yet iio significant
than the winter. The magnitude of the varia- association between the presence of -ventrieu-
tionis present is considerable and inmportant in lar prenmature beats or the finding of a coini-
any study in which attempts to manipulate plete left or right bundle-branch block with
the blood cholesterol levels are made. the development of eliiiical coronary disease.
Lipoprotein Lipase Measurements The numbers in these categories are as yet
During the third year anld during the first small. However, there was an association
half of the fourth year, lipoprotein lipase de- found between all types of ST-segment or T-
wave abnormalities (in the absence of a coni-
terminations were made on the whole group.
ILipoprotein lipase cofactor levels were also plete bundle-branch block or QRS abnormali-
m-iade. These studies were carried out through ties consistent with myocardial infaretion).
the courtesy of Dr. John Olwin of the Coagu- Thus 10 (11 per cent) of the 88 electrocar-
lation laboratory of Presbyterian-St. Luke's diograms of the nlen subsequently identified
as coronary cases showed these changes on the
Hospital. There was no evidenee of a rela-
Downloaded from http://ahajournals.org by on March 17, 2024
tionship between the presence of clinical cor- first electrocardiographic tracinig, whereas
oniary disease and levels of either lipoprotein
only 49 (2.7 per cent) of the 1,797 electrocar-
lipase or lipoprotein lipase cofactor. diograms in the as yet noneoronary group
revealed this type of abnormality. This is
Blood Sugar Determinations highly significant (p < 0.005).
During the third year of the project, each
Environmental Factors
man was asked to drink a solution containing
100 Gm. of glucose, and a blood sample for Place in, the Company
glucose determination was drawn approxi- There was no indication that current type
mately 2 hours later. The mean blood-glucose of work in the Company was a factor predis-
levels showed no significant differences be- posiiig to the development of coronary disease
tween the noneoronary and coronary groups (table 7). A study of the prior work history
(mean noneoronary blood-glucose level 106.7 likewise failed to demonstrate any significant
mg. per cent, mnean coronary level 106.8 lug. differences.
per cent). Of interest was a significant het- Physical Activity off the Job
erogeneous variation within each of the four Physical activity off the job was estimated
coronary groups. The angina group was the from the answers to series of questions posed
most variable, followed by the noncoronarv to each mnan by a trained interviewer. These
anid infaret groups, respectively. The deceased questions included reference to blocks walked
group appeared to be the least variable, but to and from work, flights of stairs climbed,
the estimated standard deviation is based on hours devoted to household chores and yard
only five observations. The differenee in vari- work, and past and present participation in
ability between the angina and infarct group sports. No striking differences were encoun-
was particularly striking. tered between the coronary and noneoronarv
Circulation, Volume XXVIII, July 1963
CORONARY HEART DISEASE 27
Tobacco
No difference was found between the groups was a relation between the number of cig,ar-
relative to the habit of cigar smoking, or to ettes smoked and the manifestation of coro-
the habit of pipe smoking. A highly signifi- nary disease identified (table 9), which rela-
cant relation to the use of cigarettes was ap- tion was significant at the 0.005 level. The
parent. The data secured in answer to the stepwise trend analogous to that observed
question asked of the participants to indicate with cholesterol levels is to be noted.
the number of cigarettes smoked most of their Diet
adult life are seen in table 8. The association The assessment of diet3 was done by inter-
of cigarette smoking most of adult life with view as indicated above. It is recognized that
the subsequent development of clinical coro- such a method yields at best only an approxi-
nary disease is significant, with a p value of mation of food intake. In a population of this
0.005. It is also significant in regard to size, metabolic balance studies were clearly
current smoking habits, also with a p value
Downloaded from http://ahajournals.org by on March 17, 2024
elsewhere), we believe the dietary data to be etable fat, so that this is in essenee a coni-
a reasonable approximatioii of the truth. parison of high and low aniimal fat groups.
Table 10 lists the dietary information for Table 11 gives the important findings in these
the coronary and noncoronary groups. No two populations. Worthv of comnment is the
significant differences are apparent for these fact that of the 88 coronary eases, 14 have
nutrients. appeared in the high-fat intake group and
In an attempt to isolate the high- and low- 16 in the low-fat group.*
fat groups from the point of view of risk, the There was no demonstrable relation between
296 men eomprising the top 15 per cent of responses to questions regarding salt intake
the whole population in terms of calories from in the coronary and the noncoronary groups.
fat were contrasted with 296 men represent- The second year, the mnen were, in addition,
ing the bottom 15 per cent in terms of fat asked to taste three samples of soup contain-
intake. It will be seen from table 10 that only ing no salt, a moderate amount of salt, and a
20 per cent of the fat calories came front veg- heavy amount of salt to indicate their prefer-
Downloaded from http://ahajournals.org by on March 17, 2024
Animal fat (Gm.) 118 116 secutively during the last 9 months of the first
Vegetable fat (Gm.) 34 32 year are listed in table 12. For this group, a
Carbohydrate (Gm.) 304 299 significant correlation between the use of cof-
Calcium (Gin.) 1.08 1.10 fee and the later developmenit of coronary
Phosphorus (Gm.) 2.33 2.16
Iron (Mg.) 17.'2 16.6 disease is seen (p < 0.025).
Cholesterol (Mg.) 757 721
Total sat. fatty acids (Gnm.) 59 59
*We have attempted to assess the constanecy of diet
in these two groups. In the third year of the project,
Total unsat. fatty acids (Gm.) 83 80
a complete diet interview was performed iii a sample
Linoleic acid (Gm.) 12.28 11.57
.69 .66
conmposed of 60 men from each category. At this
Linolenic acid (Gm.) time, the samnple group initially characterized as
Arachidonic acid (Gin.) .98 .97
having a high-fat intake obtaiined 45 per cent of its
Vitanmin "Al" (I.U.) 12,470 11,750 calories fronm fat, ancd the sample group from the
Thiamine (Mg.) 1.66 1.55
original low-fat categor-y obtained 39 per cent of
Riboflavin (Mg.) 2.52 2.51 its calories fronm fat. The difference iil the fat
Niacin (Mg.) 22.8 22.2 intake of the tw o groups has thus lessened, a result
Vitamin "'C" (I.U.) 105 101 not unexpected in viewi- of the pheniomenon of regres-
Vitamin "'D"' (LU.) 198 196 sion ton-ard the mean upon repeated testing.
Circulation, Volume XXVIII, July 1963
CORONARY HEART DISEASE 29
Table 11
Comparison of Groups wvith High and Low Fat Intake
High fat Low fat
(296 cases) (296 cases)
Median C% calories from fat 49.3 36.1
Average cholesterol (Mg. c) 253.8 246.8
Average height (inches) 69.0 68.4
Average body weight (lb.) 168.9 175.0
Average triceps skin fold (cm.) 1.4 1.4
Average scapular skin fold (cm.) 1.8 1.9
Average blood pressure (left arm) 133/86 139/90
Shop work 171 Cases-median 176 Cases-nmedian
grade 2.3 grade 2.4
Office work 121 Cases-median 120 Cases-median
grade 1.2 grade 1.2
New coronlary cases 14 16
No positive association was foulnd between ticular importance is the link between a his-
the development of coronary disease and the tory of any type of chest discomfort and the
ilntake of alcoholic beverages. later recognition of coronary disease. This
Discussion finding confirms what many elinicians have
The data presented clearly cainnot have too observed. It should be commented that it is
broad an applieation. The population under probably more difficult to elicit a history of
survey was limited to men in an age group classical symptoms of disease in men who
known to be susceptible to coroniary disease. come for this type of screening examination
Further, the popula-tion is an industrial one as compared with those who present them-
selves at the office of a physician proffering a
Downloaded from http://ahajournals.org by on March 17, 2024