Vegetarian diets in the Adventist Health Study 2: A review of initial published findings

Abstract

The Adventist Health Study 2 is a large cohort that is well suited to the study of the relation of vegetarian dietary patterns to health and disease risk. Here we review initial published findings with regard to vegetarian diets and several health outcomes. Vegetarian dietary patterns were associated with lower body mass index, lower prevalence and incidence of diabetes mellitus, lower prevalence of the metabolic syndrome and its component factors, lower prevalence of hypertension, lower all-cause mortality, and in some instances, lower risk of cancer. Findings with regard to factors related to vegetarian diets and bone health are also reviewed. These initial results show important links between vegetarian dietary patterns and improved health.

Full text

Vegetarian diets in the Adventist Health Study 2: a review of initial
published findings
1–4
Michael J Orlich and Gary E Fraser
ABSTRACT
The Adventist Health Study 2 is a large cohort that is well suited to
the study of the relation of vegetarian dietary patterns to health and
disease risk. Here we review initial published findings with regard to
vegetarian diets and several health outcomes. Vegetarian dietary pat-
terns were associated with lower body mass index, lower prevalence
and incidence of diabetes mellitus, lower prevalence of the metabolic
syndrome and its component factors, lower prevalence of hyperten-
sion, lower all-cause mortality, and in some instances, lower risk of
cancer. Findings with regard to factors related to vegetarian diets and
bone health are also reviewed. These initial results show important
links between vegetarian dietary patterns and improved health.
Am J Clin Nutr 2014;100(suppl):353S–8S.
INTRODUCTION
Much of the current understanding of the health effects of
vegetarian diets has come from a few cohort studies, especially in
California Seventh-day Adventists and British vegetarians. The
Adventist Health Study 2 (AHS-2) is a relatively new largecohort
with a high proportion of vegetarians, which promises to add to
that understanding. Here, we review the characteristics of AHS-2
and the initial published findings related to vegetarian diets.
COHORT CHARACTERISTICS
The AHS-2 is a large North American cohort. Approximately
96,000 cohort members were enrolled throughout the United
States and Canada between 2002 and 2007. Recruitment for the
study was done in Seventh-day Adventist churches, and the vast
majority of cohort members identify themselves as Adventists.
There was a special effort to recruit black subjects (including
African Americans and Caribbean Americans) as an important
group that has been underrepresented in scientific studies of diet
and health. Approximately 27% of the cohort members are black
in AHS-2, with the vast majority of others identifying as white.
Sixty-five percent of subjects are women. The mean age at en-
rollment was 57 y. A calibration sample of more than 1100
participants was selected by using a 2-stage weighted random
process, with approximately equal numbers ofblacks and whites, in
which food and physical activity recalls, biometric measurements,
and biological samples for laboratory analysis were obtained for
the purpose of validation and calibration of the cohort question-
naire data. Butler et al (1) provide a more detailed description of
the cohort’s characteristics and recruitment.
DIETARY PATTERNS
In the AHS-2, dietary patterns were defined along a vegetarian
continuum, which can be thought of as an index of animal food
avoidance. Cohort members were not asked to self-identify as
vegetarians. Rather, they were categorized on the basis of their
reported intakes of key food items of animal origin (see Table 1 for
dietary pattern definitions). Defined in this manner, 7.7% of cohort
members are vegan, 29.2% are lactoovovegetarian, 9.9% are
pescovegetarian, 5.4% are semivegetarian, and 47.7% are non-
vegetarian. For some analyses, these 5 dietary patterns were col-
lapsed to yield fewer categories; for example, in some cases, the 4
vegetarian categories (vegan, lactoovovegetarian, pescovege-
tarian, and semivegetarian) were combined together as “vegetar-
ian” (see Table 2 for select demographic, lifestyle, and nutritional
characteristics for each dietary pattern category at baseline).
HEALTH OUTCOMES
The main aims of the AHS-2 are to examine the possible effects
of dietary factors on the risk of specific cancers. These analyses for
specific cancers will begin this year after the accrual of further
incident cases to provide sufficient power. Meanwhile, several
earlypublicationsfromAHS-2haveexamined the relation of diet to
certain other health outcomes. Herewe review findings relating diet
to prevalent obesity, prevalent metabolic syndrome, prevalent
hypertension, prevalent diabetes mellitus, incident diabetes mel-
litus, bone density and fracture risk, mortality, and incident cancer
(considered as all cancers combined and by organ system). A
summary of selected results is provided in Table 3.
OBESITY
As in earlier studies (10–12), vegetarians in the AHS-2 have
lower BMI values. Among 60,903 participants, crude mean baseline
1
From the Division of Cardiology, Department of Internal Medicine
(GEF) and Department of Preventive Medicine (MJO), School of Medicine,
and the Department of Epidemiology, Biostatistics, and Population Medicine
(GEF), School of Public Health, Loma Linda University, Loma Linda, CA.
2
Presented at the symposium “Sixth International Congress on Vegetarian
Nutrition” held in Loma Linda, CA, 24–26 February 2013.
3
Project support was obtained from the National Cancer Institute
(1U01CA152939; GEF).
4
Address correspondence and requests for reprints to MJ Orlich, Adventist
Health Study 2, 24951 North Circle Drive, Nichol Hall 2031, Loma Linda,
CA 92350. E-mail: morlich@llu.edu.
First published online June 4, 2014; doi: 10.3945/ajcn.113.071233.
Am J Clin Nutr 2014;100(suppl):353S–8S. Printed in USA. Ó2014 American Society for Nutrition 353S

BMIs (in kg/m
2
) were 23.6 for vegans, 25.7 for lactoovovegetarians,
26.3 for pescovegetarians, 27.3 for semivegetarians, and 28.8 for
nonvegetarians (4). After adjustment for age, sex, and race, mean
BMIs were 24.1 for vegans, 26.1 for lactoovovegetarians, 26.0 for
pescovegetarians, 27.3 for semivegetarians, and 28.3 for non-
vegetarians among 73,308 participants (2).
METABOLIC SYNDROME
Rizzo et al (7) examined the relation of dietary patterns to
metabolic syndrome and its component risk factors in the cali-
bration sample of the AHS-2 (n= 773). Diets were considered
in 3 categories: vegetarian (vegan plus lactoovovegetarian),
semivegetarian (pescovegetarian plus semivegetarian), and
TABLE 1
Definitions and prevalence of dietary patterns in the Adventist Health Study 2
Dietary pattern
Vegan Lactoovovegetarian Pescovegetarian Semivegetarian Nonvegetarian
Prevalence (%) 7.7 29.2 9.9 5.4 47.7
All meats, including fish (servings) ,1/mo ,1/mo $1/mo $1/mo but #1/wk .1/wk
Nonfish meat (servings) ,1/mo ,1/mo ,1/mo $1/mo but #1/wk $1/mo
Fish (servings) ,1/mo ,1/mo $1/mo #1/wk Any amount
Eggs and dairy products (servings) ,1/mo $1/mo Any amount Any amount Any amount
TABLE 2
Select baseline characteristics by dietary pattern category
Dietary pattern
Vegan Lactoovovegetarian Pescovegetarian Semivegetarian Nonvegetarian
Age,
1,2
(y) 57.9 613.6 57.5 613.9 58.8 613.7 57.8 614.1 55.9 613.1
Female sex
1
(%) 63.8 64.9 68.0 69.7 65.3
Race, black
1
(%) 21.0 13.6 39.1 17.8 34.0
Marital status, married
1
(%) 75.6 76.3 73.1 71.5 70.3
Educational level
1
(%)
High school or less 16.7 13.9 18.4 21.3 24.4
Trade, associate, some college 39.4 35.7 38.1 39.2 42.2
Bachelor’s degree 24.4 25.3 23.0 21.3 19.2
Graduate degree 19.5 25.1 20.5 18.3 14.1
Alcohol consumption
1
(%)
None 98.8 96.8 92.5 92.4 83.4
Rare 0.6 1.8 4.0 4.2 7.5
Monthly 0.2 0.5 1.1 1.1 3.1
Weekly 0.3 0.7 1.9 2.0 4.7
Daily 0.1 0.2 0.5 0.3 1.3
Smoking
1
(%)
Never 85.0 88.2 84.1 81.4 75.7
Former 14.9 11.7 15.5 18.3 22.3
Current 0.1 0.1 0.4 0.3 2.0
Exercise
1,3
(%)
None 15.1 17.3 18.0 20.6 23.4
1–20 min/wk 16.2 18.6 16.8 20.5 20.0
21–60 min/wk 16.1 16.5 16.2 16.1 15.8
61–150 min/wk 27.8 26.8 27.5 24.5 23.6
$151 min/wk 24.8 20.8 21.6 18.3 17.2
Energy intake
1,2
(kcal/d) 1897 6729 1912 6735 1939 6772 1720 6713 1884 6773
Macronutrients
4,5
(% of energy)
Carbohydrate 58.1 60.1 54.3 60.1 54.5 60.1 53.9 60.1 51.4 6,0.1
Fat 28.2 60.1 31.9 60.1 31.3 60.1 32.2 60.1 33.8 6,0.1
Protein 13.6 6,0.1 13.7 6,0.1 14.2 6,0.1 13.7 6,0.1 14.7 6,0.1
Select nutrients
4,5
(g/d)
Total fiber 46.7 60.1 37.5 60.1 37.7 60.1 34.9 60.1 30.4 6,0.1
SFAs 11.6 60.1 16.0 60.1 15.8 60.1 17.4 60.1 19.9 6,0.1
Animal protein 3.1 60.2 12.2 60.1 16.0 60.2 17.6 60.2 31.8 60.1
1
Results from reference 2 (n= 73,308). Adjusted for age, sex, and race (as appropriate) by direct standardization.
2
Values are means 6SDs.
3
Exercise defined as “vigorous activities, such as brisk walking, jogging, bicycling, etc, long enough or with enough intensity to work up a sweat, get
your heart thumping, or get out of breath.”
4
Results from reference 3 (n= 71,751). Mean nutrient intake values standardized to 2000 kcal/d, adjusted for age, sex, and race.
5
Values are means 6SEs.
354S ORLICH AND FRASER

nonvegetarian. In ANCOVA analysis, with adjustment for age,
sex, ethnicity, smoking, alcohol intake, physical activity, and
dietary energy intake, significant differences between the dietary
groups were found for all of the metabolic syndrome compo-
nents except for HDL (triglycerides, diastolic blood pressure,
systolic blood pressure, waist circumference, BMI, and glucose),
with vegetarians having more favorable levels in each case.
Considering metabolic syndrome as a whole, the prevalence was
25.2%, 37.6%, and 39.7% for vegetarians, semivegetarians, and
nonvegetarians, respectively; and in logistic regression analysis
with adjustment for the same potential confounders, vegetarians
had 0.44 (95% CI: 0.30, 0.64) times the odds of having meta-
bolic syndrome as did nonvegetarians (7).
HYPERTENSION
Pettersen et al (5) examined the relation of dietary patterns to
prevalent hypertension among whites in the calibration sample
(n= 500). Diets were considered in 4 categories: vegans, lactoo-
vovegetarians, partial vegetarians (pescovegetarians plus semi-
vegetarians), and nonvegetarians. In a logistic regression analysis
that controlled for age, sex, and exercise, the adjusted ORs of
having hypertension were 0.37 (95% CI: 0.19, 0.74) and 0.57
(95% CI: 0.36, 0.92) for vegans and lactoovovegetarians, re-
spectively, compared with nonvegetarians (5). Additional adjust-
ment for BMI (a possible causal intermediate) attenuated the
results to 0.53 (95% CI: 0.25, 1.11) and 0.86 (95% CI: 0.51, 1.45),
respectively. A subsequent analysis (6) showed similar findings in
black subjects (n= 592). In a logistic regression analysis that
adjusted for age, sex, education, and physical activity, the OR for
prevalent hypertension among vegetarians (vegans and lactoo-
vovegetarians combined) was 0.56 (95% CI: 0.36, 0.87) compared
with nonvegetarians.
DIABETES MELLITUS
The relation of vegetarian diets to both prevalent and incident
diabetes mellitus has been examined in AHS-2. Prevalence of
type 2 diabetes was 2.9% among vegans, 3.2% among lac-
toovovegetarians, 4.8% among pescovegetarians, 6.1% among
semivegetarians, and 7.6% among nonvegetarians (4). In logistic
regression analysis, compared with nonvegetarians, the multi-
variate adjusted (for age, sex, ethnicity, education, income,
physical activity, television watching, sleep habits, alcohol use,
TABLE 3
Summary of the association of vegetarian dietary patterns with selected health outcomes in the Adventist Health Study 2
Health Outcome
1
Dietary pattern
Vegan Lactoovovegetarian Pescovegetarian Semivegetarian Nonvegetarian
Cross-sectional findings
BMI
2
(4) (kg/m
2
) 23.6 64.4 25.7 65.1 26.3 65.2 27.3 65.7 28.8 66.3
Diabetes
3
(4) [OR (95% CI)] 0.51 (0.40, 0.66) 0.54 (0.49, 0.60) 0.70 (0.61, 0.80) 0.76 (0.61, 0.80) Referent
Prevalence (%) 2.9 3.2 4.8 6.1 7.6
Hypertension [OR (95% CI)]
Nonblacks
4
(5) 0.37 (0.19, 0.74) 0.57 (0.36, 0.92) 0.92 (0.70, 1.50) Referent
Blacks
5
(6) 0.56 (0.36, 0.87) 0.94 (0.54, 1.63) Not reported Referent
Metabolic syndrome
6,7
(7) [OR (95% CI)] 0.44 (0.30, 0.64) Not reported Referent
Prevalence
6
(%) 25.2 37.6 39.7
Prospective findings
Diabetes
8
(8) [OR (95% CI)] 0.38 (0.24, 0.62) 0.62 (0.50, 0.76) 0.79 (0.58, 1.09) 0.49 (0.31, 0.76) Referent
n3545 14,099 3644 2404 17,695
Incident cases (%) 0.54 1.08 1.29 0.92 2.12
All cancers
9
(9) [HR (95% CI)] 0.84 (0.72, 0.99) 0.93 (0.85, 1.02) 0.88 (0.77, 1.01) 0.98 (0.82, 1.17) Referent
n4922 19,735 6846 3881 33,736
No. of events 190 878 276 182 1413
All-cause mortality
10
(2) [HR (95% CI)] 0.85 (0.73, 1.01) 0.91 (0.82, 1.00) 0.81 (0.69, 0.94) 0.92 (0.75, 1.13) Referent
n5548 21,777 7194 4031 35,359
No. of events 197 815 251 160 1147
1
Numbers in parentheses are reference numbers.
2
Values are means 6SDs.
3
Logistic regression model, adjusted for age, sex, race, BMI, physical activity, education, income, sleep, television watching, and alcohol consumption.
4
Pescovegetarians and semivegetarians were considered together as partial vegetarians because of the small numbers in both categories (logistic
regression model, adjusted for age, sex, and exercise).
5
Vegans and lactoovovegetarians were considered together as vegetarians because of the small number of vegans (logistic regression model, adjusted for
age, sex, education, and physical activity).
6
Vegans and lactoovovegetarians were considered together as vegetarians because of the small number of vegans; pescovegetarians and semivegetarians
were considered together as semivegetarians because of the small numbers in both categories.
7
Logistic regression model, adjusted for age, sex, ethnicity, physical activity, smoking, alcohol consumption, and dietary energy.
8
Logistic regression model, adjusted for age, sex, race, BMI, physical activity, education, income, sleep, television watching, smoking, and alcohol
consumption (2-y follow-up).
9
Cox proportional hazards regression model, adjusted for age, race, family history of cancer, education, smoking, alcohol consumption, age at menarche,
pregnancies, breastfeeding, oral contraceptive use, hormone replacement therapy, and menopausal status (4.14-y average follow-up).
10
Cox proportional hazards regression model, adjusted for age, sex, race, smoking, exercise, personal income, educational level, marital status, alcohol,
geographic region, menopause (in women), and hormone therapy (in postmenopausal women) (5.79-y average follow-up).
REVIEW OF VEGETARIAN DIETS IN AHS-2 355S

and BMI) ORs for prevalent type 2 diabetes were 0.51 (95%
CI: 0.40, 0.66) for vegans, 0.54 (95% CI: 0.49, 0.60) for lac-
toovovegetarians, 0.70 (95% CI: 0.61, 0.80) for pescovegetarians,
and 0.76 (95% CI: 0.65, 0.90) for semivegetarians (4).
Among 41,387 participants who did not report having diabetes
mellitus at baseline, diabetes incidence was calculated from
a response to a follow-up questionnaire at 2 y. The percentage
who reported developing diabetes was 0.54% in vegans, 1.08% in
lactoovovegetarians, 1.29% in pescovegetarians, 0.92% in
semivegetarians, and 2.12% in nonvegetarians (8). In multivariate
adjusted (for age, sex, education, income, television watching,
physical activity, sleep, alcohol use, smoking, and BMI) logistic
regression analysis, ORs for developing diabetes compared with
nonvegetarians were 0.38 (95% CI: 0.24, 0.62) for vegans, 0.62
(95% CI: 0.50, 0.76) for lactoovovegetarians, 0.79 (95% CI: 0.58,
1.09) for pescovegetarians, and 0.49 (95% CI: 0.31, 0.76) for
semivegetarians (8). Similar analyses stratified by race found
reductions in odds among blacks for the vegan (0.30; 95% CI:
0.11, 0.84) and lactoovovegetarian (0.47; 95% CI: 0.27, 0.83)
dietary patterns and among nonblacks for the vegan (0.43; 95%
CI: 0.25, 0.74), lactoovovegetarian (0.68; 95% CI: 0.54, 0.86),
and semivegetarian (0.50; 95% CI: 0.30, 0.83) dietary patterns
(8).
OSTEOPOROSIS
The relation of diet to osteoporosis risk is complex, and the
scientific understanding of it is incomplete. In particular, there is
conflicting evidence with regard to the relation of protein intake
(particularly animal protein) with bone density and fracture risk
(13–18). Thorpe et al (19) examined the relation of protein-rich
foods of both animal and plant origin to the incidence of wrist
fracture over 25 y among 1865 women who were participants in
both the AHS-1 and AHS-2. Higher consumption of protein-rich
foods of both animal and plant origin was found to be protective.
In Cox proportional hazards regression analysis, among those
with the lowest consumption of animal protein (vegetarians),
those who consumed protein-rich plant foods more than once
per day had an HR of 0.32 (95% CI: 0.13, 0.79) for wrist
fracture compared with those consuming plant protein foods
,3 times/wk (19). Similarly, among those with the lowest
consumption of plant protein foods, those who consumed meat
.4 times/wk had an HR for wrist fracture of 0.20 (95% CI: 0.06,
0.66) compared with those not consuming meat (19).
Dairy products are generally thought to be good sources of
dietary protein and calcium, raising the concern that reduced
dairy product consumption among vegetarians, particularly
vegans, may increase the risk of osteoporosis. Many vegetarians
(and many nonvegetarians) use soy milk or other types of milk
substitutes to replace dairy consumption. Matthews et al (20)
examined whether soy milk consumption might confer similar
benefits on bone health as dairy product consumption. Among
337 postmenopausal white women from AHS-2 evaluated for
osteoporosis by broadband ultrasound attenuation of the calca-
neus, the multivariate adjusted OR for osteoporosis for those
consuming $1 servings dairy products/d compared with those
consuming dairy less than twice per week was 0.38 (95% CI:
0.17, 0.86) (20). These analyses come from a logistic regression
model in which both soy milk consumption and dairy product
consumption were included. The OR for those consuming $1
servings soy milk/d compared with those not consuming soy
milk was 0.44 (95% CI: 0.20, 0.98) (20). Thus, soy milk ap-
peared to be associated with improved bone health to a similar
degree as dairy products, suggesting that it may provide a useful
alternative to dairy in certain vegetarian diets. This finding may
be related to the protein content of soy milk and, in the case of
many fortified soy milks, the calcium content. The protein
content of unfortified soy milk is 3.27 g/100 g, compared with
3.15 g/100 g for whole milk; the calcium contents of unfortified
and fortified soy milks are 25 mg/100 g and 123 mg/100 g,
respectively, compared with a calcium content of 113 mg/100 g
for whole milk (21).
CANCER
Tantamango-Bartley et al (9) recently published an initial
analysis of the association of dietary patterns with cancer in-
cidence in AHS-2. Because this was early follow-up, there was
not yet sufficient power to analyze the effect on specific cancers.
However, interesting results were shown in analyses of all in-
cident cancers and of cancers categorized by organ system.
Among 69,120 participants included in the analysis, there were
2939 incident cancers. In multivariate adjusted (for age, race,
family history of cancer, education, smoking, alcohol, age at
menarche, pregnancies, breastfeeding, oral contraceptives, hor-
mone replacement therapy, and menopause status) Cox pro-
portional hazards regression analyses comparing all vegetarians
combined (vegans, lactoovovegetarians, pescovegetarians, and
semivegetarians) with nonvegetarians, significant reductions in
risk were found for all cancers (HR: 0.92; 95% CI: 0.85, 0.99) and
gastrointestinal system cancers (HR: 0.76; 95% CI: 0.63,0.90)
(9). When the 4 vegetarian groups were compared separately with
the nonvegetarian referent group, reduced risk was found in
vegans for all cancer (HR: 0.84; 95% CI: 0.72, 0.99) and for
female-specific cancers (HR: 0.66; 95% CI: 0.47, 0.92) and in
lactoovovegetarians for gastrointestinal system cancers (HR:
0.75; 95% CI: 0.60, 0.92) (9).
MORTALITY
A longevity advantage for those who consume vegetarian diets
was previously shown in the AHS-1 cohort (12, 22). On the other
hand, a reduction in all-cause mortality has not been associated
with vegetarian dietary patterns in the European Prospective
Investigation into Cancer and Nutrition–Oxford cohort (23).
Orlich et al (2) examined the possible association of vegetarian
dietary patterns with all-cause mortality and broad categories of
cause-specific mortality in AHS-2. After a mean follow-up of
5.79 y (n= 73,308), Cox proportional hazards regression anal-
ysis (adjusting for age, race, sex, smoking, exercise, education,
marital status, alcohol, geographic region, menopause, and
hormone therapy) showed reduced all-cause mortality for all
vegetarians compared with nonvegetarians (HR: 0.88; 95% CI:
0.80, 0.97). For specific dietary patterns, the HRs were 0.85
(95% CI: 0.73, 1.01) for vegans, 0.91 (95% CI: 0.82, 1.00) for
lactoovovegetarians, 0.81 (95% CI: 0.69, 0.94) for pescovege-
tarians, and 0.92 (95% CI: 0.75, 1.13) for semivegetarians. Ef-
fects were stronger in men and less often significant in women.
Apparent beneficial associations were seen in some cases for
mortality from cardiovascular, renal, and endocrine diseases (2).
356S ORLICH AND FRASER

DISCUSSION
Because of its relatively large number of vegetarians, the AHS-
2 is a valuable cohort for the study of the possible effects of
vegetarian dietary patterns on various health outcomes. The
initial published results, reviewed previously, show a number of
apparent health benefits of vegetarian diets. Vegetarian diets in
AHS-2 are associated with lower BMI values, lower prevalence
of hypertension, lower prevalence of the metabolic syndrome,
lower prevalence and incidence of diabetes mellitus, and lower
all-cause mortality. Initial analyses also showed possible mod-
erate reductions in the rates of certain cancer outcomes for some
vegetarians. The bone health research presented here links in-
adequate protein amounts to an increased risk of osteoporosis and
fractures; however, it appears to show that plant sources of
protein, like animal sources, decrease this risk.
As with all observational research, caution must be exercised
in inferring causation from the results reviewed here. Although
appropriate attempts at adjustment for possible confounders were
made in each case, it remains possible that some uncontrolled
confounding may explain all or part of these findings. Mea-
surement error is another challenge and potential source of bias in
nutritional studies (24), but this would seem less likely to affect
analyses by broad dietary pattern than analyses according to the
intake of specific foods or nutrients.
Although large, high-quality clinical trials examining the ef-
fects of vegetarian dietary patterns on major health outcomes
have not been conducted as they have for the Mediterranean
dietary pattern (25, 26), small interventional studies provide
indirect support for some findings presented here, particularly in
regard to reduced weight (27–32), improvements in serum lipid
concentrations (33–37), and improvements in control of diabetes
mellitus (27, 38, 39) with vegetarian diets.
The dietary patterns described here are defined according to
the avoidance of certain foods of animal origin. However, the
shown associations may not always be related to reduced animal
product consumption. They may also result from an increase in
nutritional components related to plant foods, such as the in-
creased fiber intake (Table 2). There may also be considerable
heterogeneity of food and nutrient consumption within each
vegetarian-spectrum dietary pattern, as we have previously
discussed (40), so additional analyses by food, nutrient, or dietary
indexes will be of value. As with all diets, vegetarian diets should
be carefully planned for nutritional adequacy. Nutrients of
possible concern for vegetarian diets include vitamin B-12
(particularly for vegans), iron, calcium, zinc, vitamin D, and
protein (41). Rizzo et al (3) analyzed the nutrient profiles of the
5 dietary patterns described here in detail and reported consid-
erable variation by diet pattern. In no cases were mean values of
potentially marginal nutrients less adequate among vegetarians
than among nonvegetarians, but some individuals in the tails of
the distributions may have had inadequate intakes.
POTENTIAL MECHANISMS
Although analysis by dietary pattern is advantageous in terms
of real-world relevance and avoids many of the problems of re-
ductionist models, a major disadvantage of this approach is its
remoteness from specific mechanistic hypotheses. Various mech-
anisms, known and unknown, may link vegetarian dietary pat-
terns to improved health outcomes, and a full discussion of these
is beyond the scope of this brief review; however, we offer a few
comments.
Adiposity is a core feature of the metabolic syndrome and an
important risk factor for diabetes mellitus, cardiovascular dis-
ease, and certain cancers. Thus, the stepwise increase in BMI
values from vegan (lowest) to nonvegetarian (highest) presented
here is noteworthy and may serve as an important intermediate in
pathways of causation leading from dietary pattern to disease.
The reason for this BMI gradient is not well understood. Caloric
intakes are similar among the 5 dietary pattern groups (3).
Significant differences in BMI persist after control for both di-
etary energy intake and physical activity (7). Vegetarian diets
may result in differences in energy absorption and utilization that
lead to differences in BMI. The results for diabetes mellitus
reviewed here are interesting in that significant reductions in risk
for vegetarians remained after BMI was controlled for. Some of
this remaining effect may still be mediated by differences in
adiposity not fully captured by BMI (central adiposity, visceral
adiposity); however, mechanisms entirely independent of adi-
posity may also be in effect.
Differences in the intake of specific nutrients may mediate
some of the effects of vegetarian dietary patterns. For example,
vegetarians have higher intakes of potassium (3), which is
considered an important micronutrient for the prevention of
hypertension. Tantamango-Bartley et al (9) provided a discussion
of many possible mechanisms linking vegetarian dietary patterns
to reduced cancer risk; in particular, they discussed the possibility
that increased soy consumption among vegetarians could be
relevant to their finding of a reduction in risk of female-specific
cancers among vegans (9).
ONGOING AHS-2 RESEARCH
The primary aim of the AHS-2 is to investigate potential
connections between dietary factors and the risk of specific
cancers. To this end, we are attempting record linkages with the
cancer registries of all 50 states and all Canadian provinces,
something that, to our knowledge, has not previously been done.
This process is well advanced, and we anticipate important
publications on the relation of diet to specific major cancers
starting in 2014. We are hopeful that these ongoing and future
analyses will add to our understanding of the relation of vege-
tarian dietary patterns to health and longevity.
The authors’ responsibilities were as follows—GEF and MJO: designed
the research; MJO: wrote the manuscript; and GEF: had primary responsi-
bility for the final content. Neither of the authors declared a possible conflict
of interest.
REFERENCES
1. Butler TL, Fraser GE, Beeson WL, Knutsen SF, Herring RP, Chan J,
Sabate
´J, Montgomery S, Haddad E, Preston-Martin S, et al. Cohort
profile: the Adventist Health Study-2 (AHS-2). Int J Epidemiol 2008;
37:260–5.
2. Orlich MJ, Singh PN, Sabate J, Jaceldo-Siegl K, Fan J, Knutsen SF,
Beeson WL, Fraser GE. Vegetarian dietary patterns and mortality in
Adventist Health Study 2. JAMA Intern Med 2013;173:1230–8.
3. Rizzo NS, Jaceldo-Siegl K, Sabate J, Fraser GE. nutrient profiles of
vegetarian and nonvegetarian dietary patterns. J Acad Nutr Diet 2013;
113:1610–9.
4. Tonstad S, Butler TL, Yan R, Fraser GE. Type of vegetarian diet, body
weight, and prevalence of type 2 diabetes. Diabetes Care 2009;32:
791–6.
REVIEW OF VEGETARIAN DIETS IN AHS-2 357S

5. Pettersen BJ, Anousheh R, Fan J, Jaceldo-Siegl K, Fraser GE. Vegetarian
diets and blood pressure among white subjects: results from the Adventist
Health Study-2 (AHS-2). Public Health Nutr 2012;15:1909–16.
6. Fraser GE, Katuli S, Anousheh R, Knutsen SF, Herring P, Fan J.
Vegetarian diets and cardiovascular risk factors in black members of
the Adventist Health Study-2. Public Health Nutr 2014;Mar 17 (Epub
ahead of print; DOI:10.1017/S1368980014000263).
7. Rizzo NS, Sabate J, Jaceldo-Siegl K, Fraser GE. Vegetarian dietary
patterns are associated with a lower risk of metabolic syndrome: the
Adventist Health Study 2. Diabetes Care 2011;34:1225–7.
8. Tonstad S, Stewart K, Oda K, Batech M, Herring RP, Fraser GE.
Vegetarian diets and incidence of diabetes in the Adventist Health
Study-2. Nutr Metab Cardiovasc Dis 2013;23:292–9.
9. Tantamango-Bartley Y, Jaceldo-Siegl K, Fan J, Fraser GE. Vegetarian
diets and the incidence of cancer in a low-risk population. Cancer
Epidemiol Biomarkers Prev 2013;22:286–94.
10. Davey GK, Spencer EA, Appleby PN, Allen NE, Knox KH, Key TJ.
EPIC–Oxford: lifestyle characteristics and nutrient intakes in a cohort
of 33 883 meat-eaters and 31 546 non meat-eaters in the UK. Public
Health Nutr 2007;6(3):259–68.
11. Spencer E, Appleby PN, Davey G, Key TJ. Diet and body mass index
in 38 000 EPIC-Oxford meat-eaters, fish-eaters, vegetarians and
vegans. Int J Obes Relat Metab Disord 2003;27:728–34.
12. Fraser GE. Associations between diet and cancer, ischemic heart dis-
ease, and all-cause mortality in non-Hispanic white California Seventh-
day Adventists. Am J Clin Nutr 1999;70(Suppl):532S–8S.
13. Feskanich D, Willett WC, Stampfer MJ, Colditz GA. Protein consumption
and bone fractures in women. Am J Epidemiol 1996;143:472–9.
14. Misra D, Berry SD, Broe KE, McLean RR, Cupples LA, Tucker KL,
Kiel DP, Hannan MT. Does dietary protein reduce hip fracture risk in
elders? The Framingham Osteoporosis Study. Osteoporos Int 2010;22
(1):345–9.
15. Munger RG, Cerhan JR, Chiu BC. Prospective study of dietary protein
intake and risk of hip fracture in postmenopausal women. Am J Clin
Nutr 1999;69:147–52.
16. Sellmeyer DE, Stone KL, Sebastian A, Cummings SR. A high ratio of
dietary animal to vegetable protein increases the rate of bone loss and
the risk of fracture in postmenopausal women. Study of Osteoporotic
Fractures Research Group. Am J Clin Nutr 2001;73:118–22.
17. Weikert C, Walter D, Hoffmann K, Kroke A, Bergmann MM, Boeing
H. The relation between dietary protein, calcium and bone health in
women: results from the EPIC-Potsdam cohort. Ann Nutr Metab 2005;
49:312–8.
18. Zhong Y, Okoro CA, Balluz LS. Association of total calcium and di-
etary protein intakes with fracture risk in postmenopausal women: The
1999–2002 National Health and Nutrition Examination Survey
(NHANES). Nutrition 2009;25:647–54.
19. Thorpe DL, Knutsen SF, Beeson WL, Rajaram S, Fraser GE. Effects of
meat consumption and vegetarian diet on risk of wrist fracture over 25
years in a cohort of peri- and postmenopausal women. Public Health
Nutr 2008;11:564–72.
20. Matthews VL, Knutsen SF, Beeson WL, Fraser GE. Soy milk and dairy
consumption is independently associated with ultrasound attenuation of
the heel bone among postmenopausal women: the Adventist Health
Study-2. Nutr Res 2011;31:766–75.
21. USDA National Nutrient Database for Standard Reference. Release 26.
Nutrient Data Laboratory Home Page. USDA, Agricultural Research
Service. Available from: http://www.ars.usda.gov/ba/bhnrc/ndl (cited
15 Dec 2013).
22. Fraser GE, Shavlik DJ. Ten years of life: is it a matter of choice? Arch
Intern Med 2001;161:1645–52.
23. Key TJ, Appleby PN, Spencer EA, Travis RC, Roddam AW, Allen NE.
Mortality in British vegetarians: results from the European Prospective
Investigation into Cancer and Nutrition (EPIC-Oxford). Am J Clin Nutr
2009;89(suppl):1613S–9S.
24. Fraser GE. A search for truth in dietary epidemiology. Am J Clin Nutr
2003;78(suppl):521S–5S.
25. Estruch R, Ros E, Salas-Salvado
´J, Covas M-I, Corella D, Aro
´sF,
Go
´mez-Gracia E, Ruiz-Gutie
´rrez V, Fiol M, Lapetra J, et al. Primary
prevention of cardiovascular disease with a Mediterranean diet. N Engl
J Med 2013;368:1279–90.
26. de Lorgeril M, Salen P, Martin J-L, Monjaud I, Delaye J, Mamelle N.
Mediterranean diet, traditional risk factors, and the rate of cardiovas-
cular complications after myocardial infarction. Circulation 1999;99:
779–85.
27. Mishra S, Xu J, Agarwal U, Gonzales J. A multicenter randomized
controlled trial of a plant-based nutrition program to reduce body
weight and cardiovascular risk in the corporate setting: the GEICO
study. Eur J Clin Nutr 2013;67:718–24.
28. Ferdowsian HR, Barnard ND, Hoover VJ, Katcher HI, Levin SM,
Green AA, Cohen JL. A multicomponent intervention reduces body
weight and cardiovascular risk at a GEICO Corporate Site. Am J
Health Promot 2010;24(6):384–7.
29. Burke LE, Warziski M, Styn MA, Music E, Hudson AG, Sereika SM.
A randomized clinical trial of a standard versus vegetarian diet for
weight loss: the impact of treatment preference. Int J Obes (Lond)
2008;32:166–76.
30. Turner-McGrievy GM, Barnard ND, Scialli AR. A two-year random-
ized weight loss trial comparing a vegan diet to a more moderate low-
fat diet. Obesity (Silver Spring) 2007;15:2276–81.
31. Jenkins DJA, Wong JMW, Kendall CWC, Esfahani A, Ng VWY, Leong
TCK, Faulkner DA, Vidgen E, Greaves KA, Paul G, et al. The effect of
a plant-based low-carbohydrate (“Eco-Atkins”) diet on body weight
and blood lipid concentrations in hyperlipidemic subjects. Arch Intern
Med 2009;169(11):1046–54.
32. Barnard ND, Cohen J, Jenkins DJA, Turner-Mcgrievy G, Gloede L,
Jaster B, Seidl K, Green AA, Talpers S. A low-fat vegan diet improves
glycemic control and cardiovascular risk factors in a randomized
clinical trial in individuals with type 2 diabetes. Diabetes Care 2006;
29:1777–83.
33. Ferdowsian HR, Barnard ND. Effects of plant-based diets on plasma
lipids. Am J Cardiol 2009;104:947–56.
34. Ornish D, Brown SE, Billings JH, Scherwitz LW, Armstrong WT, Ports
TA, McLanahan SM, Kirkeeide RL, Gould KL, Brand RJ. Can lifestyle
changes reverse coronary heart disease? Lancet 1990;336:129–33.
35. Ornish D, Scherwitz LW, Billings JH, Gould KL, Merritt TA, Sparler S,
Armstrong WT, Ports TA, Kirkeeide RL, Hogeboom C, et al. Intensive
lifestyle changes for reversal of coronary heart disease. JAMA 1998;
280(23):2001–7.
36. Jenkins DJ, Popovich DG, Kendall CW, Vidgen E, Tariq N, Ransom
TP, Wolever TMS, Vuksan V, Mehling CC, Boctor DL, et al. Effect of
a diet high in vegetables, fruit, and nuts on serum lipids. Metabolism
1997;46:530–7.
37. Barnard ND, Scialli AR, Bertron P, Hurlock D, Edmonds K, Talev L.
Effectiveness of a low-fat vegetarian diet in altering serum lipids in
healthy premenopausal women. Am J Cardiol 2000;85:969–72.
38. Barnard ND, Cohen J, Jenkins DJA, Turner-Mcgrievy G, Gloede L,
Green A, Ferdowsian H. A low-fat vegan diet and a conventional diabetes
diet in the treatment of type 2 diabetes: a randomized, controlled, 74-wk
clinical trial. Am J Clin Nutr 2009;89(suppl):1588S–96S.
39. Nicholson AS, Sklar M, Barnard ND, Gore S, Sullivan R, Browning S.
Toward improved management of NIDDM: a randomized, controlled,
pilot intervention using a lowfat, vegetarian diet. Prev Med 1999;29:
87–91.
40. Fraser GE. Vegetarian diets: what do we know of their effects on
common chronic diseases? Am J Clin Nutr 2009;89(suppl):1607S–
12S.
41. Craig WJ, Pinyan L. Nutrients of concern in vegetarian diets. In: Sa-
bate
´J, ed. Vegetarian nutrition. Boca Raton, FL: CRC Press, 2001:
299–332.
358S ORLICH AND FRASER