ORIGINAL CONTRIBUTION

A Prospective Study of Egg Consumption

and Risk of Cardiovascular Disease
in Men and Women
Frank B. Hu, MD Context Reduction in egg consumption has been widely recommended to lower blood
Meir J. Stampfer, MD cholesterol levels and prevent coronary heart disease (CHD). Epidemiologic studies on
egg consumption and risk of CHD are sparse.
Eric B. Rimm, ScD
Objective To examine the association between egg consumption and risk of CHD
JoAnn E. Manson, MD and stroke in men and women.
Alberto Ascherio, MD Design and Setting Two prospective cohort studies, the Health Professionals Fol-
Graham A. Colditz, MD low-up Study (1986-1994) and the Nurses’ Health Study (1980-1994).
Bernard A. Rosner, PhD Participants A total of 37 851 men aged 40 to 75 years at study outset and 80 082
women aged 34 to 59 years at study outset, free of cardiovascular disease, diabetes,
Donna Spiegelman, ScD hypercholesterolemia, or cancer.
Frank E. Speizer, MD Main Outcome Measures Incident nonfatal myocardial infarction, fatal CHD, and stroke
Frank M. Sacks, MD corresponding to daily egg consumption as determined by a food-frequency questionnaire.
Charles H. Hennekens, MD Results We documented 866 incident cases of CHD and 258 incident cases of stroke
in men during 8 years of follow-up and 939 incident cases of CHD and 563 incident cases
Walter C. Willett, MD of stroke in women during 14 years of follow-up. After adjustment for age, smoking,
and other potential CHD risk factors, we found no evidence of an overall significant as-

E
LEVATED LOW-DENSITY LIPOPRO- sociation between egg consumption and risk of CHD or stroke in either men or women.
tein (LDL) cholesterol is a ma- The relative risks (RRs) of CHD across categories of intake were less than 1 per week (1.0),
jor risk factor for coronary heart 1 per week (1.06), 2 to 4 per week (1.12), 5 to 6 per week (0.90), and $1 per day (1.08)
disease (CHD).1 Dietary choles- (P for trend = .75) for men; and less than 1 per week (1.0), 1 per week (0.82), 2 to 4 per
terol raises LDL cholesterol levels and week (0.99), 5 to 6 per week (0.95), and $1 per day (0.82) (P for trend = .95) for women.
causes atherosclerosis in numerous In subgroup analyses, higher egg consumption appeared to be associated with increased
risk of CHD only among diabetic subjects (RR of CHD comparing more than 1 egg per
animal models.2 In controlled meta-
day with less than 1 egg per week among diabetic men, 2.02 [95% confidence interval,
bolic studies conducted in humans, di- 1.05-3.87; P for trend = .04], and among diabetic women, 1.49 [0.88-2.52; P for
etary cholesterol raises levels of total and trend = .008]).
LDL cholesterol in blood,3,4 but the ef-
Conclusions These findings suggest that consumption of up to 1 egg per day is un-
fects are relatively small compared with likely to have substantial overall impact on the risk of CHD or stroke among healthy
saturated and transfatty acids.5,6 Studies men and women. The apparent increased risk of CHD associated with higher egg con-
have found that individuals vary widely sumption among diabetic participants warrants further research.
in their responses to dietary cholesterol JAMA. 1999;281:1387-1394 [Link]
based on monitoring their plasma lev-
els.7,8 Prospective cohort studies on the 300 mg/d of cholesterol and limit con- tics (Drs Spiegelman and Rosner), Harvard School of
Public Health; Channing Laboratory (Drs Stampfer,
relationship of dietary cholesterol with sumption of eggs, which contain about Rimm, Manson, Colditz, Rosner, Speizer, and Wil-
risk of CHD have been inconsistent, with 213 mg of cholesterol per egg.1,18 How- let), and the Division of Preventive Medicine (Drs Man-
a significant association found in some,9,10 ever, eggs contain many other nutrients son and Hennekens), Department of Medicine,
Brigham and Women’s Hospital and Harvard Medi-
but not in most studies.11-17 cal School, Boston, Mass.
To avoid elevations in blood choles- Author Affiliations: Departments of Nutrition (Drs Hu, Corresponding Author and Reprints: Frank B. Hu, MD,
Stampfer, Rimm, Ascherio, Sacks, and Willet), Epide- Department of Nutrition, Harvard School of Public
terol and reduce CHD risk, the public has miology (Drs Stampfer, Rimm, Manson, Ascherio, Health, 665 Huntington Ave, Boston, MA 02115 (e-
been advised to consume no more than Spiegelman, Hennekens, and Willet), and Biostatis- mail: [Link]@[Link]).

©1999 American Medical Association. All rights reserved. JAMA, April 21, 1999—Vol 281, No. 15 1387
EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

besides cholesterol, including unsatur- The Nurses’ Health Study on the frequency distribution of egg
ated fats, essential amino acids, folate, The Nurses’ Health Study (NHS) co- consumption. In a validation study in
and other B vitamins. In addition, con- hort was established in 1976 when a subsample of the participants, the cor-
sumption of eggs instead of carbohy- 121 700 women who were registered relation coefficent between intakes of
drate-rich foods may raise high-density nurses residing in 11 large states, aged eggs assessed by the diet question-
lipoprotein (HDL) cholesterol lev- 30 to 55 years, provided detailed infor- naire and by multiple week diet re-
els19,20 and decrease blood glycemic and mation about their medical history and cords was 0.8 for both men 28 and
insulinemic responses.21 For these rea- lifestyle characteristics.25 Every 2 years, women.29 We also computed intake of
sons, it is useful to study directly the re- follow-up questionnaires have been sent eggs included in other foods such as
lationship of egg consumption with risk to update information on potential risk cakes, cookies, pancakes, muffins, sweet
of CHD, but few epidemiologic studies factors and to identify newly diag- rolls, and donuts. The amount of eggs
have addressed this association.22-24 In nosed cases of CHD, stroke, and other estimated from other foods was rela-
this article, we examine prospectively the diseases. In 1980, a 61-item food fre- tively small in both cohorts (men, 0.4
association between egg consumption quency questionnaire was included to egg per week; women, 0.3 egg per
and risk of CHD and stroke in 2 large co- assess intake of specific fats and other week), so we used the reported egg con-
horts of men and women. nutrients. In 1984, the food frequency sumption as our primary exposure vari-
questionnaire was expanded to in- able. In a secondary analysis, we ex-
METHODS clude 116 items. Similar question- amined the effect of computed total egg
The Health Professionals naires were used to update diet in 1986 consumption.
Follow-up Study and 1990. The reproducibility and va-
The Health Professionals Follow-up lidity of the food frequency question- End Points
Study (HPFS) began in 1986 when naires have been described in detail else- The end points were incident CHD (in-
51 529 men who were US health pro- where.26,27 cluding nonfatal myocardial infarction
fessionals (dentists, optometrists, phar- After up to 4 mailings, 98 462 women [MI] and fatal CHD) and stroke occur-
macists, podiatrists, and veterinar- returned the 1980 diet questionnaire. ring between return of the baseline ques-
ians), aged 40 to 75 years, answered a We excluded those who left 10 or more tionnaires and January 31, 1994 (men),
detailed questionnaire that included a items blank, those with implausibly or June 1, 1994 (women). We inquired
comprehensive diet survey and items high or low scores for total food in- about occurrence of cardiovascular end
on lifestyle practice and medical his- take or energy intake (ie, ,2100 kJ/d points on each biennial questionnaire.
tory. Follow-up questionnaires were or .14 700 kJ/d) (n = 5994), and those Participants reporting an incident MI or
sent in 1988, 1990, 1992, and 1994 to with previously diagnosed cancer stroke were asked for permission to re-
update information on potential risk (n = 3526), cardiovascular disease view medical records. Nonfatal MI was
factors and to identify newly diag- (n = 1812), high blood cholesterol confirmed by symptoms plus either typi-
nosed cases of cardiovascular and other (n = 1821), or diabetes (n = 4122) at cal electrocardiographic changes or in-
diseases. We excluded from the analy- baseline. Incidence of CHD and stroke creased activities in cardiac enzymes
sis men who did not satisfy the a priori during the subsequent 14 years was (World Health Organization criteria).30
criteria of reported daily energy in- monitored for 80 082 women during Infarctions that required hospital admis-
take between 3360 and 17 640 kJ or follow-up. The follow-up rate for non- sion and for which confirmatory infor-
who left blank more than 70 items of fatal events was 98% of the total po- mation was obtained by interview or let-
the 131 total food items in the diet ques- tential person-years of follow-up. ter, but for which no medical records
tionnaire (n = 1152). We also excluded were available, were designated as prob-
men with prior diagnoses of cardiovas- Assessment of Egg Consumption able. We included all confirmed and
cular disease (n = 5182) or cancer Validated dietary questionnaires were probable cases in the analyses because
(n = 1644) at baseline. In the primary sent to the HPFS participants in 1986 results were the same after excluding
analyses, we also excluded men who re- and 1990 and the NHS participants in probable cases. Strokes were confirmed
ported diabetes mellitus (n = 1187) or 1980, 1984, 1986, and 1990. In all the if characterized by a typical neurologi-
hypercholesterolemia (n = 4458) at questionnaires, we asked the partici- cal defect of sudden or rapid onset, last-
baseline because these diagnoses could pants how often, on average, during the ing at least 24 hours, and attributable to
have led to changes in diet. Incidence previous year they had consumed eggs a cerebrovascular event. Strokes caused
of CHD and stroke during the sub- (unit of consumption was 1 egg). Nine by infection or neoplasia were ex-
sequent 8 years was monitored for responses were possible, ranging from cluded. Strokes were subclassified ac-
37 851 men during follow-up. The fol- never to 6 or more times per day. We cording to the criteria of the National Sur-
low-up rate for nonfatal events was 97% divided the participants into 5 catego- vey of Stroke as due to ischemia
of the total potential person-years of ries (,1 per week, 1 per week, 2-4 per (embolism or thrombosis), subarach-
follow-up. week, 5-6 per week, $1 per day) based noid hemorrhage, intracerebral hemor-
1388 JAMA, April 21, 1999—Vol 281, No. 15 ©1999 American Medical Association. All rights reserved.
 EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

rhage, or unknown cause.31 If no rec- 10-14, 15-29, and $30 g/d); (4) his- diets for those who developed these
ords could be obtained, strokes were tory of hypertension; (5) parental his- conditions.35
considered probable if they required hos- tory of MI; (6) body mass index (calcu-
pitalization and were corroborated by ad- lated as weight in kilograms divided by RESULTS
ditional information provided by letter the square of height in meters) (quin- In men, the average egg consumption
or interview. tiles); and (7) current multivitamin use, declined from 2.3 eggs per week in 1986
Deaths were reported by next of kin, and (8) vitamin E supplement use. In to 1.6 eggs per week in 1990. In women,
coworkers, postal authorities, or the Na- HPFS, we also adjusted for physical ac- the average egg consumption de-
tional Death Index. Using all sources tivity in metabolic equivalents per week clined from 2.8 eggs per week in 1980
combined, we estimate that follow-up (quintiles).34 In NHS, we adjusted for to 1.4 eggs per week in 1990. At base-
for the deaths was more than 98% com- regular vigorous exercise (once or more line, egg consumption was positively as-
plete.32 Fatal CHD was defined as fatal per week) and menopausal status and sociated with smoking in men but in-
MI if this was confirmed by hospital rec- postmenopausal hormone use. Tests of versely associated with smoking in
ords or autopsy, or if CHD was listed linear trend across increasing catego- women (TABLE 1). Those with higher
as the cause of death on the death cer- ries of egg consumption were con- egg intake consumed more dietary cho-
tificate and this was the underlying and ducted by treating the median values of lesterol and protein but less carbohy-
most plausible cause, and evidence of consumption in categories (servings per drates. Egg consumption was posi-
previous CHD was available. The state- day) as a continuous variable. tively associated with bacon intake
ment of the cause of death on the death To reduce intra-individual varia- among both men (r = 0.35) and women
certificate was never relied on by itself tion and best represent long-term diet, (r = 0.21). Men with higher egg con-
as providing sufficient confirmation of we used repeated measures of diet in sumption were more likely to con-
death due to CHD. Sudden death within our primary analyses.35 In particular, sume whole milk, red meat, and bread,
an hour of the onset of symptoms in CHD or stroke incidence was related to and less likely to consume skim milk,
subjects with no other plausible cause the cumulative average of egg consump- chicken, vegetables, and fruits. These
of death (other than coronary disease) tion from all available questionnaires relationships were less clear in women.
was also included. Fatal stroke was also prior to the beginning of each 2-year
confirmed by medical records or au- follow-up interval. For example, in Association With CHD
topsy reports, or considered probable HPFS, CHD or stroke incidence dur- We documented 866 incident cases of
if these were not obtainable but stroke ing the 1986-1990 period was related CHD in men during 8 years of fol-
was listed as the underlying cause on to egg consumption assessed in 1986, low-up (1986-1994) and 939 cases of
the death certificate. while the incidence during 1990- CHD in women during 14 years of fol-
1994 period was related to the average low-up (1980-1994). The age-adjusted
Statistical Analysis intake assessed in 1986 and 1990. For RR of CHD comparing more than 1 egg
Participants contributed follow-up time those who failed to complete the 1990 per day with less than 1 egg per week
from the return of the 1986 (men) or the diet questionnaire (24%), we used the was 1.15 (95% confidence interval [CI],
1980 (women) questionnaire up to the 1986 values to replace the missing val- 0.85-1.55; P for trend = .37) for men and
occurrence of a confirmed end point, ues and a missing data indicator was in- 0.85 (95% CI, 0.62-1.16; P for
death, or the end of follow-up (January cluded in the model to correct for po- trend = .50) for women (TABLE 2). Af-
31, 1994, for men and June 1, 1994, for tential bias. Average egg consumption ter adjustment for smoking and other co-
women). The relative risks (RRs) were at baseline was the same between the variates, the corresponding RRs were
calculated by dividing the incidence of respondents and nonrespondents. In al- 1.08 (95% CI, 0.79-1.48; P for
CHD or stroke among men and women ternative analyses, we analyzed the in- trend = .75) for men and 0.82 (95% CI,
in various categories of egg consump- cidence of CHD in relationship to egg 0.60-1.13; P for trend = .95) for women.
tion by the incidence among those in the consumption at baseline only. We also Additional adjustment for dietary fiber
lowest category of intake (,1 egg per related the incidence to the most re- intake had little impact on the RRs for
week), adjusting for age (5-year catego- cent diet by using the baseline diet to women but it further attenuated the as-
ries). To adjust for other risk factors, we predict incidence during 1986-1990, sociation for men (RR for .1 egg per
used pooled logistic regression,33 which and the 1990 diet to predict incidence day, 1.01; 95% CI, 0.74-1.38). Because
is asymptotically equivalent to Cox re- during 1990-1994. Because the diag- of the relatively strong correlation be-
gression for time-dependent covari- nosis of hypercholesterolemia, diabe- tween consumption of eggs and bacon,
ates. Multivariate models included as co- tes, or hypertension may lead to we further adjusted for bacon intake.
variates were (1) total energy intake changes in egg consumption and there- The adjusted RRs across categories of egg
(quintiles); (2) smoking (never, past, fore confound the associations be- consumption are less than 1 per week
current 1-14, 15-24, and $25 cigarettes/ tween egg consumption and risk of car- (1.0), 1 per week (1.00), 2 to 4 per week
d); (3) alcohol consumption (0-4, 5-9, diovascular disease, we did not update (1.04), 5 to 6 per week (0.78), and 1 or
©1999 American Medical Association. All rights reserved. JAMA, April 21, 1999—Vol 281, No. 15 1389
EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

more per day (0.93) (P for trend = .36) In further analyses egg consump- pants who reported changes in egg
for men; and less than 1 per week (1.0), tion had no significant relationship with consumption in the previous decade
1 per week (0.81), 2 to 4 per week either fatal CHD or nonfatal MI (data from baseline, we observed a signifi-
(0.96), 5 to 6 per week (0.91), and 1 or not shown). After excluding cases with cant inverse association for men but a
more per day (0.78) (P for trend = .73) events occurring during the first 2 years nonsignificant positive association for
for women. Additional adjustment for of follow-up to reduce the effect of women. The results from analyses in-
other foods including whole milk, fish, change in diet due to preclinical con- cluding eggs estimated from other foods
beef as main dish, chicken, or cereal had ditions, the results remained un- were similar to those from the main
little impact on the results. changed. In analyses excluding partici- analyses. We also observed no signifi-

Table 1. Baseline Characteristics According to Egg Consumption Levels in Men and Women*
Men in Health Professionals’ Women in Nurses’ Health
Follow-up Study (1986 Baseline) Study (1980 Baseline)
Egg Consumption Egg Consumption

,1 Per 1 Per 2-4 Per 5-6 Per .1 Per ,1 Per 1 Per 2-4 Per 5-6 Per .1 Per
Characteristics Week Week Week Week Day Week Week Week Week Day
No. of subjects 11 247 8103 13 845 2280 2376 11 568 16 155 38 109 7160 7090
Current smoker, % 7.2 9.0 10.8 13.9 14.6 34.7 30.6 27.0 26.3 25.4
Current postmenopausal hormone use, % NA NA NA NA NA 15.2 15.8 16.9 17.4 16.0
Parental history of myocardial infarction, % 13.1 11.1 11.2 10.5 9.2 15.9 16.1 15.3 14.9 14.5
Multivitamin supplement use, % 41.5 39.9 41.1 41.6 43.3 29.8 30.2 34.1 38.4 40.5
Vitamin E supplement use, % 19.6 17.3 17.6 18.9 20.8 11.4 10.9 12.2 15.1 18.0
History of hypertension, % 17.8 18.4 17.7 15.9 14.3 14.2 13.8 14.1 13.8 14.0
Vigorous exercise $1/wk, % ... ... ... ... ... 42.8 43.8 45.8 46.2 46.7
Age, y 52.7 52.6 53.8 54.2 54.2 46.4 45.7 45.8 45.5 45.9
Body mass index, kg/m2 25.2 25.4 25.6 25.8 25.5 23.9 23.9 24.4 24.6 24.4
Alcohol use, g/d 10.1 11.0 12.1 13.1 12.3 6.4 6.5 6.3 6.8 6.7
Physical activity, metabolic equivalent/wk† 22.0 19.6 19.0 18.8 17.5 ... ... ... ... ...
Energy intake, kJ 7627 8089 8770 9559 9925 5851 6086 6703 7274 7505
Nutrient intakes (energy-adjusted)
Cholesterol, mg/d 237 266 330 404 536 228 258 342 436 557
Saturated fat, g/d 23.1 24.7 25.5 26.5 27.9 28.0 28.1 27.8 28.3 29.0
Monounsaturated fat, g/d 25.9 27.4 28.1 29.1 30.3 28.4 28.6 28.5 29.1 29.8
Polyunsaturated fat, g/d 13.0 13.2 13.2 13.3 13.4 9.0 9.3 9.4 9.6 9.6
Transfatty acid, g/d 2.7 2.9 2.9 2.9 2.9 4.0 4.1 4.0 4.0 3.8
Dietary folate, µg/d 368 354 348 336 336 240 246 261 264 272
Fiber, g/d 22.6 20.6 19.9 19.0 18.5 13.6 13.4 13.7 13.5 13.3
Protein, g/d 90.7 90.8 92.2 93.4 96.0 72.8 73.0 76.3 77.6 79.9
Carbohydrates, g/d 247 236 229 221 211 160 158 156 150 144
Food intakes, serving per 4200 kJ
Whole milk 0.05 0.07 0.08 0.10 0.12 0.16 0.17 0.16 0.17 0.18
Skim milk 0.38 0.39 0.36 0.32 0.29 0.36 0.38 0.44 0.39 0.36
Red meat‡ 0.27 0.31 0.33 0.34 0.37 0.51 0.51 0.50 0.48 0.48
Bacon 0.03 0.05 0.07 0.11 0.15 0.05 0.06 0.08 0.10 0.12
Chicken 0.19 0.18 0.17 0.16 0.15 0.19 0.17 0.17 0.17 0.17
Breakfast cereal 0.24 0.22 0.19 0.14 0.10 0.19 0.20 0.19 0.15 0.11
Dark bread 0.37 0.34 0.36 0.37 0.44 0.45 0.44 0.50 0.52 0.52
White bread 0.21 0.23 0.26 0.31 0.33 0.49 0.51 0.52 0.51 0.52
Vegetables§ 1.75 1.55 1.50 1.42 1.35 1.41 1.35 1.34 1.30 1.30
Fruits\ 1.09 0.95 0.88 0.82 0.77 1.37 1.34 1.37 1.30 1.31
Legumes¶ 0.26 0.23 0.22 0.22 0.23 0.33 0.31 0.31 0.30 0.29
*Values are means unless otherwise indicated. Data, except age, were directly standardized to the age distribution of each study population. Ellipses indicate data not provided;
NA, not applicable.
†The metabolic equivalent value is the energy needed per kilogram of body weight per hour of activity divided by the energy need per kilogram of body weight per hour at rest.
‡Composite score of beef, pork, or lamb as the main dish, beef as a sandwich or mixed dish, and hamburger.
§Composite score of 22 vegetable foods from the men’s study and 11 from the women’s study.
\Composite score of 12 fruits in the men’s study and 6 from the women’s study.
¶Composite score of peas or lima beans and beans or lentils.

1390 JAMA, April 21, 1999—Vol 281, No. 15 ©1999 American Medical Association. All rights reserved.
 EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

cant increase in risk when either only mass index, age, vitamin supplement 4200 kJ for men and 175.9 mg/4200 kJ
baseline or only recent egg consump- use, parental history of MI, and in- for women).
tion was analyzed. Only 4.8% women takes of saturated fat, polyunsaturated
reported almost never consuming eggs fat, and carbohydrates. We found no The Association With Stroke
and 1.6% women reported consuming evidence of a positive association with We documented 258 incident cases of
2 eggs or more per day. When these 2 higher consumption of eggs in any sub- stroke in men during 8 years of fol-
extreme groups were compared (us- group except a suggestion that the risk low-up and 563 cases of stroke in
ing never consumers as the reference might be elevated among individuals women during 14 years of follow-up.
group), the multivariate RR was 0.76 with diabetes. Among diabetic men, the In multivariate analyses (TABLE 3), egg
(95% CI, 0.43-1.35). The correspond- multivariate RRs of CHD across cat- consumption was not significantly as-
ing RR was 1.10 (95% CI, 0.67-1.79) egories of intake were less than 1 per sociated with risk of total stroke; the
for men. week (1.0), 1 per week (1.0), 2 to 4 per RRs for 1 egg or more per day were 1.07
In all above analyses, participants week (1.16), 5 to 6 per week (1.16), and (95% CI, 0.66-1.75; P for trend = .50)
with diabetes or hypercholesterol- 1 or more eggs per day (2.02); (95% CI, for men and 0.89 (95% CI, 0.60-1.31;
emia at baseline were excluded. When 1.05-3.87; P = .04 for trend and P = .18 P for trend = .77) for women. Adjust-
these subjects were included in the for interaction between egg consump- ment for intake of bacon, vegetables,
analyses, the results did not apprecia- tion and diabetes status). The corre- and fruit further attenuated the asso-
bly change. The multivariate RRs as in sponding RRs for diabetic women were ciation for men (RR for $1 egg per day,
Table 2 across categories of egg intake 1.0, 0.91, 1.05, 1.87, and 1.49 (95% CI, 1.00; 95% CI, 0.57-1.78; P for trend =
were less than 1 per week (1.0), 1 per 0.88-2.52; P = .008 for trend and P = .07 0.95). In both cohorts, no significant
week (1.05), 2 to 4 per week (1.04), 5 for interaction). To investigate the pos- association was observed between egg
to 6 per week (0.96), and 1 or more per sibility that an effect of egg consump- consumption and risk of ischemic or
day (1.10) (95% CI, 0.83-1.45; P for tion may be more apparent among those hemorrhagic stroke.
trend = .78) for men; and less than 1 per with a low-background cholesterol
week (1.0), 1 per week (0.87), 2 to 4 diet,36 we conducted analyses strati- COMMENT
per week (0.98), 5 to 6 per week (1.03), fied by dietary cholesterol from foods In these 2 large prospective cohort stud-
and 1 or more per day (0.95) (95% CI, other than eggs. The RRs for more than ies of men and women, we found no
0.74-1.22; P for trend = .54) for women. 1 eggs per day were 1.05 (95% CI, 0.61- overall significant association be-
1.79) for the men and 0.97 (95% CI, tween egg consumption (up to 1 egg per
Associations According 0.13-7.10) for the women whose non- day) and risk of CHD or stroke. We spe-
to Risk Factor Status egg cholesterol intake was low (mean cifically found no evidence for a sig-
To examine the possibility that a posi- intakes were 88.4 mg/4200 kJ for men nificant increase in risk with either re-
tive association with egg intake is lim- and 118.8 mg/4200 kJ for women), cent or relatively long-term (over the
ited to certain subgroups, we con- compared with 0.97 (95% CI, 0.64- past decade) egg consumption. De-
ducted additional multivariate analyses 1.46) for the men and 0.79 (95% CI, spite somewhat different patterns of egg
stratified by risk factor status includ- 0.57-1.11) for the women whose non- consumption in men and women, the
ing hypercholesterolemia, diabetes, hy- egg cholesterol intake was relatively results from the 2 cohorts were remark-
pertension, smoking, alcohol use, body high (mean intakes were 135.9 mg/ ably consistent.

Table 2. Coronary Heart Disease According to Categories of Egg Consumption*

Men Women

Amount Age-Adjusted Multivariate Age-Adjusted Multivariate

of Eggs No. of Person- Relative Risk Relative Risk No. of Person- Relative Risk Relative Risk
Consumed Cases Years (95% CI) (95% CI) Cases Years (95% CI) (95% CI)
Per week
,1 252 90 118 1.0 1.0 189 175 144 1.0 1.0
1 210 69 091 1.09 (0.90-1.31) 1.06 (0.88-1.27) 346 366 625 0.75 (0.61-0.91) 0.82 (0.67-1.00)
2-4 315 88 077 1.16 (0.98-1.37) 1.12 (0.95-1.33) 331 409 896 0.95 (0.80-1.14) 0.99 (0.82-1.18)
5-6 38 12 216 0.97 (0.68-1.36) 0.90 (0.63-1.27) 43 52 584 0.97 (0.71-1.31) 0.95 (0.70-1.29)
Per day
$1 51 14 059 1.15 (0.85-1.55) 1.08 (0.79-1.48) 30 52 971 0.85 (0.62-1.16) 0.82 (0.60-1.13)
P for trend .37 .75 .50 .95
*Adjusted for age (5-year categories), body mass index (quintiles), 2-year time periods, cigarette smoking (never, past, and current [1-14, 15-24, and $25 cigarettes/d]), parental
history of myocardial infarction, multivitamin supplement use, vitamin E supplement use, alcohol consumption (0-4, 5-9, 10-14, 15-29, $30 g/d), menopausal status and post-
menopausal hormone use (women), history of hypertension, physical activity (quintiles of metabolic equivalents in men and vigorous activities at least once a week in women), and
total energy intake (quintiles). CI indicates confidence interval.

©1999 American Medical Association. All rights reserved. JAMA, April 21, 1999—Vol 281, No. 15 1391
EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

In subgroup analyses, egg consump- Several metabolic studies have sug- intake, which was positively associ-
tion appeared to be associated with in- gested a hypocholesterolemic effect of ated with risk of CHD in our cohorts,
creased risk of CHD among individu- decholesterolized eggs (ie, Eggbeaters) the RRs became weakly inverse. This
als with diabetes. This result should be on blood cholesterol levels compared speaks to the importance of consider-
interpreted cautiously due to numer- with whole eggs.39,40 We were not able ing overall eating patterns when exam-
ous subgroup analyses, but the consis- to examine the effect of such products ining the effects of egg consumption.
tency of the association in the 2 co- on the risk of CHD because they were It is possible that participants with
horts argues against a chance finding. not included in the dietary question- high serum cholesterol levels were more
The increased risk may be related to ab- naires on the printed form. However, our likely to reduce their egg intake than oth-
normal cholesterol transport due to de- questionnaires permit respondents to re- ers, which might obscure a positive as-
creased levels of apolipoprotein E37 and port other foods that are frequently con- sociation between egg consumption and
increased levels of apolipoprotein C- sumed. In NHS, only 48 women re- risk of CHD. However, in our primary
III38 among patients with diabetes. corded consuming Eggbeaters in the analyses, we excluded subjects with di-
We considered the possibility that in- 1984 dietary questionnaire and 105 re- agnosed hypercholesterolemia at base-
accurate self-reports of egg consump- corded this in the 1990 questionnaire. line. Also, in our analyses by using up-
tion or confounding by intake of other Egg consumption was positively as- dated dietary information, we stopped
foods could explain the observed null sociated with smoking, lower physical updating egg consumption at the begin-
results. Egg consumption was re- activity, and a generally unhealthy eat- ning of the time interval during which
ported on food frequency question- ing pattern (ie, more whole milk, red individuals reported hypercholesterol-
naires with relatively high accuracy meat, and bacon and less skim milk, emia to avoid confounding due to change
(correlations were 0.8 between the self- vegetables, and fruits) in men. Con- in diet during follow-up.41 Finally, we did
report and multiple week dietary re- founding due to these factors would ar- not measure blood cholesterol levels in
cords in our validation studies28,29). tifactually produce an elevated risk for our cohorts. However, blood choles-
Also, egg consumption was assessed egg consumption. As expected, an ap- terol should not be controlled in the
several times in both cohorts so that our parent positive association with higher analyses as it is an intermediate variable
analyses using updated dietary infor- egg consumption in the age-adjusted when assessing the relationship be-
mation could dampen measurement er- analysis in men was attenuated after ad- tween dietary cholesterol and CHD.
ror and take into account changes in justment for smoking and other covar- In controlled metabolic studies, in-
eating behavior. iates. After further adjusting for bacon gestion of cholesterol by eating egg

Table 3. Stroke According to Categories of Egg Consumption*

Type of Stroke

Total Ischemic Hemorrhagic

Amount Age-Adjusted Multivariate Age-Adjusted Multivariate Age-Adjusted Multivariate

of Eggs No. of Relative Risk Relative Risk No. of Relative Risk Relative Risk No. of Relative Risk Relative Risk
Consumed Cases (95% CI) (95%CI) Cases (95% CI) (95% CI) Cases (95% CI) (95% CI)
Men
Per week
,1 76 1.0 1.0 51 1.0 1.0 16 1.0 1.0
1 65 1.09 (0.78-1.52) 1.06 (0.76-1.49) 42 1.04 (0.69-1.57) 0.99 (0.66-1.50) 13 1.05 (0.51-2.19) 1.02 (0.49-2.14)
2-4 82 0.99 (0.73-1.36) 0.95 (0.69-1.31) 52 0.94 (0.63-1.38) 0.86 (0.58-1.28) 19 1.12 (0.57-2.18) 1.01 (0.51-2.00)
5-6 18 1.52 (0.91-2.56) 1.43 (0.85-2.43) 11 1.39 (0.72-2.69) 1.25 (0.64-1.87) 4 0.77 (0.26-2.30) 0.67 (0.22-2.07)
Per day
$1 17 1.25 (0.74-2.12) 1.07 (0.66-1.75) 10 1.09 (0.55-2.15) 0.93 (0.46-1.87) . . .† . . .† . . .†
P for trend .28 .50 .70 .88 .78 .55
Women
Per week
,1 110 1.0 1.0 63 1.0 1.0 33 1.0 1.0
1 161 0.83 (0.65-1.05) 0.89 (0.70-1.13) 82 0.70 (0.51-0.98) 0.78 (0.56-1.09) 51 0.89 (0.67-1.38) 0.93 (0.60-1.44)
2-4 222 0.81 (0.64-1.01) 0.83 (0.66-1.05) 114 0.75 (0.55-1.03) 0.79 (0.57-1.08) 69 0.80 (0.53-1.21) 0.86 (0.56-1.31)
5-6 34 0.90 (0.61-1.32) 0.89 (0.60-1.32) 21 1.08 (0.66-1.78) 1.11 (0.67-1.84) 9 0.72 (0.34-1.51) 0.75 (0.35-1.58)
Per day
$1 36 0.92 (0.63-1.35) 0.89 (0.60-1.31) 16 0.80 (0.46-1.39) 0.81 (0.46-1.42) 14 1.09 (0.58-2.03) 1.07 (0.56-2.03)
P for trend .96 .77 .90 .81 .77 .81
*See footnote to Table 2 for multivariate list. CI indicates confidence interval.
†Ellipses indicate that data in this category are combined with the previous category due to the small number of cases.

1392 JAMA, April 21, 1999—Vol 281, No. 15 ©1999 American Medical Association. All rights reserved.
 EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

yolks or whole eggs raises serum total erides by 0.09 mmol/L (7.97 mg/dL) by One potential alternative explana-
and LDL cholesterol levels.5-7,42 In most adding 2 eggs to the usual diet. Pack- tion for the null finding is that back-
egg feeding studies, intakes of other nu- ard et al20 observed a reduction of tri- ground dietary cholesterol may be so
trients such as fatty acids, carbohy- glycerides by 0.19 mmol/L (16.8 mg/ high in the usual Western diet that add-
drates, and protein were balanced be- dL) by adding 6 eggs to the diet. In ing somewhat more has little further ef-
tween egg and no egg groups so that addition, egg intake decreased blood fect on blood cholesterol. In a random-
only dietary cholesterol varied. In our glycemic and insulinemic responses,47 ized trial, Sacks et al36 found that adding
cohorts, participants who consumed especially when egg yolk was ingested 1 egg per day to the usual diet of 17 lac-
more eggs had lower intakes of carbo- (compared with whole egg or egg tovegetarians whose habitual choles-
hydrates, suggesting that, in reality, white).21 This might result from fur- terol intake was very low (97 mg/d) sig-
people often substitute eggs for carbo- ther delayed gastric emptying after yolk nificantly increased LDL cholesterol
hydrate-rich foods such as breakfast ce- ingestion. Moreover, adding eggs to level by 12%. In our analyses, differ-
reals. The effects of egg cholesterol on pasta produced lower insulin and C- ences in non-egg cholesterol intake did
blood cholesterol can be predicted from peptide responses.48 Holt et al49 tested not appear to be an explanation for the
well-established equations derived from 38 common foods and found that eggs null association between egg consump-
metabolic studies. A 50-g egg contains were among the foods that have the tion and risk of CHD. However, we can-
about 213 mg of cholesterol, 6 g of pro- lowest glycemic and insulin indexes. not exclude the possibility that egg con-
tein, and 5 g of fat.43 Of the fat, nearly Based on the 116 food items in the sumption may increase the risk among
half is monounsaturated fat and 16% is 1984 food frequency questionnaire in participants with very low back-
polyunsaturated fat. The equation de- NHS, we estimated that egg consump- ground cholesterol intake. Also, we
rived by Keys and Parlin4 predicts that tion contributed to 32% of total dietary have limited power to examine the ef-
adding 1 egg to an average diet (as- cholesterol. Eggs also made contribu- fect of high egg consumption (eg, $2
sume 200 mg background cholesterol tions to dietary intakes of many other eggs per day). Nevertheless, the aver-
and 7560 kJ/d) will result in about a 4% nutrients, including vitamin D (5%), age egg consumption was 0.64 eggs per
increase in total serum cholesterol for retinol (4%), folate (4%), a-tocopherol day in 1995 in the United States, 50
a normocholesterolemic person (as- (3%), monounsaturated fat (3%), vita- which is well within the range of con-
sume total blood cholesterol of 5.17 min B12 (3%), vitamin B2 (3%), protein sumption in our analyses.
mmol/L [200 mg/dL]). If we assume (3%), saturated fat (2.5%), linoleic acid We found no significant association
that raising cholesterol levels is the only (2%), calcium (1.3%), vitamin B1 (1.2%), between egg consumption and risk of to-
effect of egg consumption, this would and vitamin B2 (1.2%). It is conceiv- tal stroke or its subtype. Epidemiologic
translate into about an 8% increase in able that the small adverse effect of cho- studies51 have revealed no clear pat-
CHD risk,44 an effect generally too small lesterol in an egg on plasma LDL levels terns between blood cholesterol and
to be detectable in this and most epi- is counterbalanced by potential benefi- stroke although secondary prevention
demiologic studies or clinical trials. In cial effects on HDL and triglycerides, and trials52,53 showed significant reductions
NHS, dietary cholesterol (but not eggs) of other nutrients including antioxi- in stroke incidence among patients with
was nonsignificantly associated with dants, folate, other B vitamins, and un- MIs and who were treated with choles-
CHD,17 raising the possibility that eggs saturated fats. Since there is no single terol-lowering drugs (statins). A recent
contain other nutrients that may be ben- biochemical measurement that can rep- report from the Framingham Heart
eficial in preventing CHD. resent the effects of various nutrients, it Study54 indicated that intakes of total fat,
The equation developed by Men- is important to examine the direct rela- saturated fat, and monounsaturated fat
sink and Katan45 predicts that substi- tionship between egg consumption and were associated with reduced risk of
tuting fatty acids from 1 egg for carbo- risk of CHD. Only 2 previous prospec- ischemic stroke, but the study did not
hydrates would raise HDL cholesterol tive cohort studies reported on the as- examine the effects of dietary choles-
by about 2% (assume HDL level of 1.03 sociation between egg consumption and terol or egg consumption.
mmol/L [40 mg/dL]) and decrease tri- risk of CHD. In the Framingham Study, In conclusion, our data suggest that
glycerides also by about 2% (assume tri- egg consumption was not significantly consumption of up to 1 egg per day is un-
glyceride level of 2.82 mmol/L [250 mg/ associated with either serum choles- likely to have substantial overall impact
dL]). The effects of egg consumption terol or risk of CHD (RR was not pro- on the risk of cardiovascular disease
on raising HDL levels have been ob- vided).22 In the California Adventists among healthy men and women. The ap-
served in some metabolic studies,19,20 Study, the RR for higher intake ($3 per parent increased risk of CHD associated
but not in others.36,40 Also, in egg feed- week) vs lowest intake (,1 per week) with higher egg consumption among
ing studies, triglyceride levels were sig- was 1.01.23 In a case-control study con- diabetics warrants further research.
nificantly reduced in some,19,20 but not ducted in Italy,24 the RR comparing
Funding/Support: This research was supported by Na-
other studies.36,46 In particular, Schnohr women in the upper third of egg intake tional Institutes of Health grants HL35464, HL24074,
et al19 observed a reduction of triglyc- with those in the lower third was 0.8. HL34594, DK 46200, CA40356, and CA55075.

©1999 American Medical Association. All rights reserved. JAMA, April 21, 1999—Vol 281, No. 15 1393
EGG CONSUMPTION AND CARDIOVASCULAR DISEASE

Acknowledgment: We are indebted to the partici- 18. Krauss RM, Deckelbaum RJ, Ernst N, et al. Di- 37. Fielding CJ, Castro GR, Donner C, Fielding PE,
pants in the NHS and HPFS for their outstanding level etary guidelines for healthy American adults: a state- Reaven GM. Distribution of apolipoprotein E in the
of cooperation; and to Al Wing, MBA, Gary Chase, ment for health professionals from the nutrition com- plasma of insulin-dependent and noninsulin-
Karen Corsano, MSL, Lisa Dunn, Barbara Egan, Lori mittee, American Heart Association. Circulation. 1996; dependent diabetics and its relation to cholesterol net
Ward, Jill Arnold, Mira Kaufman, MS, Betsy Frost- 94:1795-1800. transport. J Lipid Res. 1986;27:1052-1061.
Hawes, Kerry Pillsworth-Demers, and Mildred Wolff, 19. Schnohr P, Thomsen O, Riis Hansen P, Boberg- 38. Venkatesan S, Imrie H, Read S, Halliday D. Apo
MA, for their unfailing help. Ans G, Lawaetz H, Weeke T. Egg consumption and C subclasses from non-insulin-dependent diabetic pa-
high-density lipoprotein cholesterol. J Intern Med. tients: a quantitative comparison with control sub-
1994;235:249-251. jects. Biochem Soc Trans. 1995;23:278S.
REFERENCES
20. Packard CJ, McKinney L, Carr K, Shepherd J. Cho- 39. Roberts SL, McMurry MP, Connor WE. Does egg
1. National Cholesterol Education Program. Report of lesterol feeding increases low density lipoprotein syn- feeding (ie, dietary cholesterol) affect plasma choles-
the Expert Panel on Population Strategies for Blood thesis. J Clin Invest. 1983;72:45-51. terol levels in humans? the results of a double-blind
Cholesterol Reduction: executive summary. Arch In- 21. Pelletier X, Thouvenot P, Belbraouet S, et al. Ef- study. Am J Clin Nutr. 1981;34:2092-2099.
tern Med. 1991;151:1071-1084. fect of egg consumption in healthy volunteers: influ- 40. Ginsberg HN, Karmally W, Siddiqui M, et al. A
2. Stamler J, Shekelle R. Dietary cholesterol and hu- ence of yolk, white or whole-egg on gastric empty- dose-response study of the effects of dietary choles-
man coronary heart disease. Arch Pathol Lab Med. ing and on glycemic and hormonal responses. Ann Nutr terol on fasting and postprandial lipid and lipoprotein
1988;112:1032-1040. Metab. 1996;40:109-115. metabolism in healthy young men. Arterioscler Thromb
3. Hegsted DM, McGandy RB, Myers ML, Stare FJ. 22. Dawber TR, Nickerson RJ, Brand FN, Pool J. Eggs, Vasc Biol. 1994;14:576-586.
Quantitative effects of dietary fat on serum choles- serum cholesterol, and coronary heart disease. Am J 41. Shekelle RB, Stamler J, Paul O, Shryock AM, Liu
terol in man. Am J Clin Nutr. 1965;17:281-295. Clin Nutr. 1982;36:617-625. S, Lepper M. Dietary lipids and serum cholesterol level:
4. Keys A, Parlin RW. Serum-cholesterol response to 23. Fraser GE. Diet and coronary heart disease: be- change in diet confounds the cross-sectional associa-
changes in dietary lipids. Am J Clin Nutr. 1966;19: yond dietary fats and low-density-lipoprotein choles- tion. Am J Epidemiol. 1982;115:506-514.
175-181. terol. Am J Clin Nutr. 1994;59(suppl):1117-1123. 42. Hopkins PN. Effects of dietary cholesterol on se-
5. Clarke R, Frost C, Collins R, Appleby P, Peto R. Di- 24. Gramenzi A, Gentile A, Fasoli M, Negri E, Parazzini rum cholesterol: a meta-analysis and review. Am J Clin
etary lipids and blood cholesterol: quantitative meta- F, La Vecchia C. Association between certain foods Nutr. 1992;55:1060-1070.
analysis of metabolic ward studies. BMJ. 1997;314: and risk of acute myocardial infarction in women. BMJ. 43. US Department of Agriculture. Composition of
112-117. 1990;300:771-773. Foods: Raw, Processed, and Prepared, 1963-1988.
6. Howell WH, McNamara DJ, Tosca MA, Smith BT, 25. Colditz GA, Manson JE, Hankinson SE. The Nurses’ Washington, DC: Dept of Agriculture, Government
Gaines JA. Plasma lipid and lipoprotein responses to Health Study: 20-year contribution to the understand- Printing Office; 1989. Agricultural Handbook No. 8.
dietary fat and cholesterol: a meta-analysis. Am J Clin ing of health among women. J Womens Health. 1997; 44. Rich-Edwards JW, Manson JE, Hennekens CH, Bur-
Nutr. 1997;65:1747-1764. 6:49-62. ing JE. The primary prevention of coronary heart dis-
7. McGill HC. The relationship of dietary cholesterol 26. Willett WC, Sampson L, Stampfer MJ, et al. Re- ease in women. N Engl J Med. 1995;332:1758-
to serum cholesterol concentration and to atheroscle- producibility and validity of a semiquantitative food 1766.
rosis in man. Am J Clin Nutr. 1979;32:2664-2702. frequency questionnaire. Am J Epidemiol. 1985;122: 45. Mensink RP, Katan MB. Effect of dietary fatty ac-
8. Hegsted DM. Serum-cholesterol response to di- 51-65. ids on serum lipids and lipoproteins: a meta-analysis
etary cholesterol: a re-evaluation. Am J Clin Nutr. 1986; 27. Willett WC, Sampson L, Browne ML, et al. The of 27 trials. Arterioscler Thromb Vasc Biol. 1992;12:
44:299-305. use of a self-administered questionnaire to assess diet 911-919.
9. McGee DL, Reed DM, Yano K, Kagan A, Tillotson four years in the past. Am J Epidemiol. 1988;127: 46. Vorster HH, Benade AJS, Barnard HC, et al. Egg
J. Ten-year incidence of coronary heart disease in the 188-199. intake does not change plasma lipoprotein and co-
Honolulu Heart Program: relationship to nutrient in- 28. Feskanich D, Rimm EB, Giovannucci EL, et al. Re- agulation profiles. Am J Clin Nutr. 1992;55:400-
take. Am J Epidemiol. 1984;119:667-676. producibility and validity of food intake measure- 410.
10. Shekelle RB, Shryock AM, Paul O, et al. Diet, se- ments from a semiquantitative food frequency ques- 47. Villaume C, Beck B, Rohr R, Pointel J-P, Debry G.
rum cholesterol, and death from coronary heart dis- tionnaire. J Am Diet Assoc. 1993;93:790-796. Effect of exchange of ham for boiled egg on plasma
ease: the Western Electric Study. N Engl J Med. 1981; 29. Salvini S, Hunter DJ, Sampson L, et al. Food- glucose and insulin responses to breakfast in normal
304:65-70. based validation of a dietary questionnaire: the ef- subjects. Diabetes Care. 1986;9:46-49.
11. Kromhout D, de Lezenne Coulander C. Diet, fects of week-to-week variation in food consump- 48. Granfeldt Y, Bjorck I, Hagander B. On the impor-
prevalence and 10-year mortality from coronary heart tion. Int J Epidemiol. 1989;18:858-867. tance of processing conditions, product thickness and
disease in 871 middle-aged men: the Zutphen Study. 30. Rose GA, Blackburn H. Cardiovascular survey egg addition for the glycaemic and hormonal re-
Am J Epidemiol. 1984;119:733-741. methods. WHO Monograph Series No. 58. Geneva, sponses to pasta: a comparison with bread made from
12. Kushi LH, Lew RA, Stare FJ, et al. Diet and 20- Switzerland: World Health Organization; 1982. “pasta ingredients.” Eur J Clin Nutr. 1991;45:489-
year mortality from coronary heart disease: the Ireland- 31. Walker AE, Robins M, Weinfeld FD. The Na- 499.
Boston Diet-Heart Study. N Engl J Med. 1985;312: tional Survey of Stroke: clinical findings. Stroke. 1981; 49. Holt SHA, Brand-Miller JC, Petocz P. An insulin
811-818. 12(suppl 1):I13-I44. index of foods: the insulin demand generated by
13. Posner BM, Cobb JL, Belanger AJ, Cupples A, 32. Stampfer MJ, Willett WC, Speizer FE, et al. Test 1000-KJ portions of common foods. Am J Clin Nutr.
D’Agostino RB, Stokes J III. Dietary lipid predictors of of the national death index. Am J Epidemiol. 1984; 1997;66:1264-1276.
coronary heart disease in men. Arch Intern Med. 1991; 119:837-839. 50. Federation of American Societies for Experimen-
151:1181-1187. 33. D’Agostino RB, Lee M-L, Belanger AJ, Cupples tal Biology. Third Report on Nutrition Monitoring in
14. Ascherio A, Rimm EB, Giovannucci EL, Spiegel- LA, Anderson K, Kannel WB. Relation of pooled lo- the United States. Vol 1. Washington, DC: US Gov-
man D, Stampfer MJ, Willett WC. Dietary fat and risk gistic regression to time dependent cox regression ernment Printing Office; 1995.
of coronary heart disease in men: cohort follow up analysis: the Framingham Heart Study. Stat Med. 1990; 51. Bronner LL, Kanter DS, Manson JE. Primary preven-
study in the United States. BMJ. 1996;313:84-90. 9:1501-1515. tion of stroke. N Engl J Med. 1995;332:1392-1400.
15. Esrey KL, Joseph L, Grover SA. Relationship be- 34. Chasan-Taber S, Rimm EB, Stampfer MJ, et al. Re- 52. Pedersen T, Kjekshus J, Pyorala K, et al. Effect of
tween dietary intake and coronary heart disease mor- producibility and validity of a self-administered physi- simvastatin on ischemic signs and symptoms in the
tality: Lipid Research Clinics Prevalence Follow-up cal activity questionnaire for male health profession- Scandinavian Simvastatin Survival Study. Am J Car-
Study. J Clin Epidemiol. 1996;49:211-216. als. Epidemiology. 1996;7:81-86. diol. 1998;81:333-335.
16. Pietinen P, Ascherio A, Korhonen P, et al. Intake 35. Hu FB, Stampfer MJ, Rimm E, et al. Dietary fat 53. Plehn JF, Davis BR, Sacks FM, et al. Reduction of
of fatty acids and risk of coronary heart disease in a and coronary heart disease: a comparison of ap- stroke incidence after myocardial infarction with pravas-
cohort of Finnish men: the Alpha-Tocopherol, Beta- proaches for adjusting for total energy intake and mod- tatin: the Cholesterol and Recurrent Events (CARE)
Carotene Cancer Prevention Study. Am J Epidemiol. eling repeated dietary measurements. Am J Epide- Study. Circulation. 1999;99:216-223.
1997;145:876-887. miol. 1999;149:531-540. 54. Gillman MW, Cupples LA, Millen BE, Ellison RC,
17. Hu FB, Stampfer MJ, Manson JE, et al. Dietary fat 36. Sacks F, Salazar J, Miller L, et al. Ingestion of egg Wolf PA. Inverse association of dietary fat with de-
intake and the risk of coronary heart disease in women. raises plasma low density lipoproteins in free-living sub- velopment of ischemic stroke in men. JAMA. 1997;
N Engl J Med. 1997;337:1491-1499. jects. Lancet. 1984;8378:647-649. 278:2145-2150.

1394 JAMA, April 21, 1999—Vol 281, No. 15 ©1999 American Medical Association. All rights reserved.