Content uploaded by Pramil Singh
Author content
All content in this area was uploaded by Pramil Singh
Content may be subject to copyright.
Fax +41 61 306 12 34
E-Mail karger@karger.ch
www.karger.com
Original Paper
Ann Nutr Metab 2008;52:96–104
DOI: 10.1159/000121365
Meats, Processed Meats, Obesity, Weight
Gain and Occurrence of Diabetes among Adults:
Findings from Adventist Health Studies
Arnold Vang
a
Pramil N. Singh
b
Jerry W. Lee
a
Ella H. Haddad
c
Charles H. Brinegar
d
Departments of
a
Health Promotion and Education,
b
Epidemiology and Biostatistics, and
c
Nutrition,
School of Public Health, Loma Linda University, and
d
Department of Internal Medicine, School of Medicine,
Loma Linda University, Loma Linda, Calif. , USA
meat intake remained an important risk factor (OR = 1.38;
95% CI 1.06–1.08) for diabetes. Conclusions: Our findings
raise the possibility that meat intake, particularly processed
meats, is a dietary risk factor for diabetes.
Copyright © 2008 S. Karger AG, Basel
Introduction
The prevalence of diagnosed and undiagnosed diabe-
tes in the United States has been estimated at more than
16 million individuals
[1, 2] . This epidemic has led to
research into lifestyle modification (physical activity,
weight control, and diet)
[3–6] that can be effective in
primary prevention. Among the lifestyle practices, diet,
particularly a high-fat diet pattern
[7] , has long been
thought to play an important role in the development of
type 2 diabetes. This association between diet and type 2
diabetes has been attributed to the well-known mecha-
nisms of hyperglycemia, hyperinsulinemia, and insulin
resistance
[8, 9] . More recent evidence also raises the pos-
sibility that some of the diet-diabetes relation is mediated
through pathways related to chronic inflammation
[10]
or chronically elevated levels of glucagon and cortisol
[11,
12]
. Furthermore, higher levels of these hormones have
been found to be associated with animal product intake
[13–15] . Interestingly, meat protein has been implicated
Key Words
Diabetes mellitus ⴢ Vegetarian ⴢ Nutritional epidemiology ⴢ
Obesity ⴢ Adventist Health Study
Abstract
Aim: To examine the relation between meat intake and dia-
betes occurrence in adults. Methods: In a prospective co-
hort study we examined the relation between diet and inci-
dent diabetes recorded among 8,401 cohort members (ages
45–88 years) of the Adventist Mortality Study and Adventist
Health Study (California, USA) who were non-diabetic at
baseline. During the 17-year follow-up, we identified 543 in-
cident diabetes cases. Results: (1) Subjects who were week-
ly consumers of all meats were 29% (OR = 1.29; 95% CI 1.08,
1.55) more likely (relative to zero meat intake) to develop di-
abetes. (2) Subjects who consumed any processed meats
(salted fish and frankfurters) were 38% (OR = 1.38; 95% CI
1.05–1.82) more likely to develop diabetes. (3) Long-term ad-
herence (over a 17-year interval) to a diet that included at
least weekly meat intake was associated with a 74% increase
(OR = 1.74; 95% CI 1.36–2.22) in odds of diabetes relative to
long-term adherence to a vegetarian diet (zero meat intake).
Further analyses indicated that some of this risk may be at-
tributable to obesity and/or weight gain – both of which
were strong risk factors in this cohort. It is noteworthy that
even after control for weight and weight change, weekly
Received: April 19, 2007
Accepted: October 18, 2007
Published online: March 18, 2008
Dr. Pramil N. Singh
Department of Epidemiology, School of Public Health
Loma Linda University
Loma Linda, CA 92350 (USA)
Tel. +1 909 558 4590, Fax +1 909 558 0326, E-Mail psingh@llu.edu
© 2008 S. Karger AG, Basel
0250–6807/08/0522–0096$24.50/0
Accessible online at:
www.karger.com/anm
Meat Intake and Diabetes
Ann Nutr Metab 2008;52:96–104
97
in diabetic renal disease due to its effect on the glomeru-
lar filtration rate
[16] . Also, when people with diabetic
nerve damage switch to a vegan diet (no meat, dairy or
eggs), evidence has supported improvements in renal
function
[17, 18] and glucose tolerance [19] .
Ecologic data indicates that in countries with higher
per capita consumption of beef, there is a higher preva-
lence of diabetes
[20–22] . Three prospective cohort stud-
ies have examined the relation between animal product
consumption and incident diabetes
[23–25] . These stud-
ies, the Iowa Women’s Health Study
[23] , the Physician’s
Health Study
[24] , and the Nurses’ Health Study [25] in-
dicated modest 20–46% increases in risk of diabetes for
higher meat consumption (about 5 times per week or
more), particularly for consumption of processed meats
[25] . In the Adventist Health Study, higher meat intake
was attributed to a 37% increase in risk of diabetes-re-
lated mortality
[26] . One prospective study [27] of nor-
moglycemic elderly subjects did, however, indicate that
fish intake may decrease risk of type 2 diabetes during 3
years of follow-up, possibly due to the –3 fatty acid con-
tents of this specific meat. A few studies have raised the
possibility that dairy products may decrease the risk of
type 2 diabetes, and that early exposure to dairy products
may increase the risk of juvenile onset (type 1) diabetes
[16, 27–29] .
The purpose of this study was to examine the associa-
tion between all animal product consumption, specific
animal product consumption (red meat, poultry, fish,
processed meats, eggs, milk, and cheese) and diabetes in-
cidence among California Seventh-day Adventist adults
who were cohort members of the Adventist Mortality
Study (AMS)
[30] and the AHS [31] . Since about one-
third of these cohort members consumed no meat, we
had greater statistical power and hence a unique oppor-
tunity to evaluate whether meatless diets are associated
with a lower risk of diabetes. Moreover, the two reports
of diet (1960 and 1976) available for these cohort mem-
bers also enabled us to relate long-term meat intake
dietary patterns (over a 17-year interval) to the develop-
ment of diabetes.
Research Design and Methods
Study Population
The AMS and the AHS are previously described
[26, 30–34]
prospective cohort studies in which data were collected on diet
and lifestyle factors at baseline (1960 in the AMS, 1976 in the
AHS) and subjects were enrolled in surveillance for incident and
fatal diseases. The AMS and AHS cohort populations were identi-
fied in 1958 and 1974, respectively
[30–34] . Subject recruitment
was done through the Seventh-day Adventist churches in Califor-
nia where members were encouraged to complete the self-admin-
istered questionnaires. The AMS consisted of 27,530 non-Hispan-
ic white adults who completed a questionnaire in 1960 that was
identical to the American Cancer Society perspective study of 1
million persons
[30] . The population in the AHS consisted of
34,198 non-Hispanic white adults who completed a detailed life-
style questionnaire in 1976
[31] . For the present analysis, we iden-
tified 8,401 participants who completed both the 1960 and 1976
questionnaires and did not have diabetes in 1960. These 8,401
subjects became the analytic population for the present study. In
this analytic population, we identified 543 incident diabetes (dur-
ing 1960–1976) cases based on their disease history data from the
1976 questionnaire.
Study Questionnaires
The questions in the AMS were identical to the American
Cancer Society prospective cohort study of 1 million persons and
contained items on dietary intake, anthropometrics, disease his-
tory, and demographic characteristics
[26, 30] . Food frequency
items on this questionnaire measured frequency of consumption
in times per week. The food category items examined consisted
of the following foods that pertained to specific animal product
consumption: red meat/poultry, fish, eggs, cheese, milk con-
sumption, and processed meats (salted fish and frankfurters).
From these variables we computed: (1) an index of consumption
of all animal products (meats, eggs, cheese, milk) measured in
times per week; (2) an index of consumption of all meats (red
meat, poultry, fish) measured in times per week; (3) an index of
consumption of all processed meats (salted fish and frankfurters)
measured as a dichotomous variable (ever or never consumed in
usual diet); (4) an index of consumption of eggs measured in times
per week, and (5) an index of consumption of milk measured in
times per day. The index of meat consumption was used to further
classify subjects into three groups: vegetarians (those who did not
consume meat), individuals with occasional meat intake (those
who consumed meat less than once per week) and non-vegetari-
ans (those who consumed meat once or more per week). We also
classified subjects by vegetarian status (vegetarian, occasional
meat intake, non-vegetarian) based on their responses to the 1976
questionnaire. By cross-classifying the 1960 data on vegetarian
status with the 1976 data on vegetarian status, we were then able
to relate long-term adherence to a vegetarian diet pattern to dia-
betes occurrence. In a validity sub-study among 147 cohort mem-
bers
[33, 34] , the correlation between a total meat index deter-
mined from questionnaire data and the corresponding measures
from five 24-hour recalls was 0.83.
Demographic questions pertaining to age, gender, body mass
index (BMI, kg/m
2
), smoking status, alcohol consumption (beer,
wine, liquor combined), education, and prevalent disease were
obtained from the 1960 AMS study. The validity of self-reported
anthropometric data was tested in a random sample of 118 AHS
cohort participants and it was found that the correlation between
self-reported and measured weight and height were 0.93 and 0.96,
respectively
[33, 34] . The variable for smoking status was obtained
from responses to items on current and former use of cigarettes,
cigars, and pipe smoking. Prevalent disease history included heart
disease, cancer, or stroke.
Vang /Singh /Lee /Haddad /Brinegar
Ann Nutr Metab 2008;52:96–104
98
Since data on physical activity in 1960 was not available for
this analysis, we controlled for physical activity as a confounder
using a previously validated
[33] index of physical activity based
on variables from the 1976 questionnaire. This physical activity
index was calculated from the response to vigorous leisure-time
or occupational activities items and classifies subjects into a ‘high’
category for frequent (15 min or more per session, 3 times or more
per week) participation in vigorous activity, a ‘moderate’ category
for less frequent (less than 15 min per session, less than 3 times
per week) participation in vigorous activity, and a ‘none/low’ cat-
egory for a ‘rarely or never’ response to vigorous activity items.
Statistical Analyses
Descriptive analysis of pertinent dietary and other lifestyle
variables were completed by computing means or proportions of
these variables for categories of meat intake – an exposure of pri-
mary interest. To relate consumption of animal products in 1960
to diabetes occurrence after 17 years of follow-up (in 1976), we
used a multivariate logistic regression model with a dichotomous
diabetes variable as the outcome. Two statistical tests were per-
formed for each dietary variable. First, to assess the overall sig-
nificance of the individual food variables, we performed a log-
likelihood ratio test of the indicator food variables. Next, a mul-
tivariate test for linear trend across food intake levels was
performed by replacing the indicator food variables for k catego-
ries in each multivariate model, with a single k level variable rep-
resenting the median frequency of consumption in each of the k
categories. The p value for this trend test was then obtained by the
significance test of the regression coefficient for the single food
variable.
In additional multivariate models, we tested for confounding
by adding the following variables: age, gender, education, physical
activity, cigarette smoking, and alcohol consumption. Since BMI
is a possible intervening variable in the relation between diet and
diabetes, we did not test for confounding by weight or weight
change. We did, however, add variables for weight and weight
change to the model to examine the extent to which the diet-dia-
betes association was mediated through this pathway. We also
conducted further analyses where subjects with cardiovascular
disease, stroke, and cancer were excluded since these prevalent
diseases may have impacted dietary choices.
R e s u l t s
Baseline characteristics of the population (n = 8,401)
are presented in table 1 by level of meat intake (3,994 veg-
etarians, 224 occasional meat intake, 3,798 weekly meat
intake, 385 no response). These data indicate that among
the non-vegetarians there was a higher prevalence of cig-
arette smoking, alcohol use, and less years of education.
Frequency of consumption of animal products in the co-
hort we studied is given in table 2 .
Table 3 shows the age- and sex-adjusted estimates
from multivariate logistic regression models relating in-
dividual food variables to diabetes status at follow-up.
Among the animal product variables, the strongest asso-
ciations were for intake of all meats (red meat, poultry,
fish) in which those who consume meat once or more per
week were 29% more likely to develop diabetes during the
follow-up, and a significant (p ! 0.005) positive trend for
meat intake was also identified. We also found a similar
increase in risk for weekly intake of red meat and poultry
(OR = 1.27; 95% CI 1.06–1.53), but not for weekly intake
of fish (OR = 1.12; 95% CI 0.88–1.44). In further analyses
(data not shown) we found no association between an in-
dex of intake of all animal products (all meats, dairy,
eggs) and diabetes incidence.
Table 1. Selected demographic and lifestyle characteristics of the
study population at baseline in 1960 by level of meat intake
Variable Vegetarian
1
(n = 3,994)
Occasional
meat intake
2
(n = 224)
Non-
vegetarian
3
(n = 3,798)
Mean age, years, at time
of questionnaire return 66 66 63
Gender, %
Male 60 66 62
Female 40 34 37
BMI (mean, kg/m
2
)
4
24 25 25
Alcohol consumption, %
4
00 1
Smoking history, %
4, 5
Never 77 76 65
Current <1 3 2
Past 23 21 33
Heart disease, % 4 5 4
Cancer, %
Male 3 4 3
Female 2 2 2
Stroke, % <1 1 1
Physical activity, %
4, 6
None 25 32 30
Slight 9 10 10
Moderate 22 18 22
Heavy 44 40 39
Education
(some college), %
4, 7
61 54 48
1
No current meat intake.
2
Current meat intake less than once per week.
3
Current meat intake greater than or equal to once per week.
4
p < 0.005 by
2
, multivariate logistic regression for the distri-
bution of variables across levels of meat intake.
5
Never indicates a ‘rarely or never’ response to items describ-
ing smoking (cigarette and/or cigar) status: never smoked, pres-
ent, or past smoker.
6
Data available for males only.
7
Identifies individuals with college level education.
Meat Intake and Diabetes
Ann Nutr Metab 2008;52:96–104
99
For processed meats, the strongest association was for
salted fish in which those who consumed salted fish were
55% more likely (OR = 1.55; 95% CI 1.0–2.39) to develop
diabetes than those who did not consume salted fish.
Those who consumed frankfurters weekly were 29%
more likely (OR = 1.29; 95% CI 0.94–1.76) to develop dia-
betes during the follow-up than those who did not con-
sume frankfurters. Individuals who consumed both pro-
cessed meats (salted fish and frankfurters) were 38%
more likely (OR = 1.38; 95% CI 1.05–1.82) to develop dia-
betes relative to those who did not consume any pro-
cessed meats. When we further adjusted for all other
meat consumption, we continued to find a reduced, al-
beit significant association where individuals who con-
sumed processed meats were 27% more likely (OR = 1.27;
95% CI 1.02–1.16) to develop diabetes.
Using data on meat intake at baseline in 1960 and at the
end of the follow-up in 1976, we also examined the relation
between diabetes and long-term adherence (concordant
reports of meat intake in 1960 and 1976) to the following
diet patterns: non-vegetarian, occasional meat intake, and
vegetarian. These data in table 4 indicate that subjects who
were weekly consumers of meat over the 17-year interval
(1960–1976) were 74% (OR = 1.74; 95% CI 1.36–2.22) more
likely to have developed diabetes during the follow-up
than those who had been vegetarian over the 17-year in-
terval. In analyses that stratified by gender (not shown) we
found that female long-term non-vegetarians were 80%
(OR = 1.8; 95% CI 1.20–2.84) more likely to develop dia-
betes while male non-vegetarians were 60% (OR = 1.6; 95%
CI 1.20–2.24) more likely to develop diabetes.
In additional multivariate models that included con-
founders for education, physical activity, smoking, and
alcohol consumption the findings depicted in tables 3
and 4 were not substantially altered. Moreover, the posi-
tive associations with a non-vegetarian diet were not sub-
stantially altered by exclusion of smokers (OR = 1.28; 95%
CI 1.05–1.56), alcohol users (OR = 1.27; 95% CI 1.03–
1.57), and those with preexisting disease (heart disease,
stroke, cancer) (OR = 1.27; 95% CI 1.05–1.54).
In tables 4 and 5 , we also present additional analyses
where variables for BMI and BMI by weight change were
added to the model. In these analyses the significant 74%
increase in risk due to weekly attenuated to a significant
34–38% increase in risk after adjustment for obesity and/
or weight gain (during the 17-year interval). Table 5 fur-
ther indicates the strong risk factor association with obe-
sity and weight gain that was evident in this population.
Weight loss among the obese was also identified as a par-
ticularly potent risk factor.
Discussion
Among 8,401 cohort members of the AMS and the
AHS, we examined the relation between animal product
intake and diabetes. Our major findings from this cohort
are as follows: (1) Subjects who consume all meats once
or more per week were 29% more likely to develop diabe-
tes, and this association was primarily due to high intake
of red meat and/or poultry. (2) Subjects who consumed
processed meats (salted fish and frankfurters) were 38%
Table 2. Frequency of consumption of animal products by sub-
jects in the study population
Food variable %
Total meats
Never 50
>0 to <1/week 3
≥1/week 47
Red meat/poultry
Never 50
>0 to <1/week 5
≥1/week 45
Fish
Never 77
>0 to <1/week 8
≥1/week 15
Eggs
Never 10
>0 to <1/week 3
≥1/week 87
Cheese
Never 14
>0 to <1/week 4
≥1/week 82
Milk
Never 18
1/day 29
2/day 26
3/day 16
>4/day 11
Total processed meats
Never 90
Consumed weekly 10
Frankfurters
Never 92
Consumed weekly 8
Salted fish
Never 97
Consumed weekly 3
Vang /Singh /Lee /Haddad /Brinegar
Ann Nutr Metab 2008;52:96–104
100
Table 3. Age- and sex-adjusted associations between animal products and diabetes among 8,401 subjects in the
study population
Food variable Cases Odds
ratio
1
95% CI
1
p for trend
across 3
2
p for log-
likelihood ratio
3
Total meats
4
Never 237 1.00 referent 0.005 0.012
>0 to <1/week 11 0.80 0.43–1.49
≥1/week 267 1.29 1.08–1.55
Red meat/poultry
Never 248 1.00 referent 0.008 0.028
>0 to <1/week 24 0.99 0.64–1.52
≥1/week 260 1.27 1.06–1.53
Fish
Never 399 1.00 referent 0.35 0.318
>0 to <1/week 48 1.24 0.91–1.69
≥1/week 84 1.12 0.88–1.44
Eggs
Never 54 1.00 referent 0.54 0.534
>0 to <1/week 20 1.32 0.77–2.25
≥1/week 461 1.15 0.85–1.54
Cheese
Never 89 1.00 referent 0.41 0.710
>0 to <1/week 23 0.96 0.60–1.55
≥1/week 423 0.90 0.71–1.15
Milk
Never 107 1.00 referent 0.83 0.645
1/day 158 0.93 0.72–1.19
2/day 135 0.92 0.70–1.19
3/day 87 1.01 0.75–1.36
>4/day 52 0.92 0.65–1.29
1
The odds ratio and 95% CI for each intake category of a specific food or food group were computed relative
to the low intake category from a logistic regression model with adjustment for age and gender.
2
Determined from a logistic regression model in which the indicator variables for food frequency categories
were replaced with a single variable representing the median frequency at a given intake level.
3
A log-likelihood ratio test of the indicator variables for food frequency categories.
4
Current intake red meat, poultry, fish.
Table 4. Relation between long-term (over a 17-year interval) diet patterns and diabetes among subjects in the
study population who maintained stable diet patterns over a 17-year interval
Variable Age-sex adjusted
1
Age, sex, BMI adjusted
2
OR 95% CI OR 95% CI
Long-term vegetarian
3
1.00 referent 1.00 referent
Long-term occasional meat intake
4
1.00 0.40–2.53 0.91 0.36–2.34
Long-term non-vegetarian
5
1.74 1.36–2.22 1.34 1.03–1.75
1
Multivariate logistic regression model adjusting for age and sex.
2
Multivariate logistic regression model adjusting for age, sex and BMI.
3
Participants who identified themselves as eating no meat in both 1960 and 1976.
4
Participants who identified themselves as eating meat less than once per week in 1960 and meat intake
in 1976.
5
Participants who identified themselves as eating meat once or more per week in both 1960 and 1976.
Meat Intake and Diabetes
Ann Nutr Metab 2008;52:96–104
101
more likely to develop diabetes than those who did not
consume any processed meats. (3) Long-term adherence
(over a 17-year interval) to a non-vegetarian diet (weekly
meat intake) was associated with a 74% increase in odds
of diabetes relative to a vegetarian diet (zero meat intake).
No statistically significant associations were found be-
tween dairy consumption, egg consumption, or an index
of all animal products (all meats, dairy, eggs) and diabe-
tes.
Our findings relating meat intake to diabetes are con-
sistent with the findings from three prospective cohort
studies that have examined the relation between meat in-
take and diabetes incidence
[23–25] . In a prospective
study among 35,988 Iowa women, total meat intake (red
meat and poultry) was positively associated with a sig-
nificant 35% increase in risk of diabetes, and further
analysis revealed that some of this association may be at-
tributable to a relation between meat fat and diabetes
[23] .
In the Health Professionals Study, higher intake of un-
processed red meat and poultry was associated with a
27% increase risk for diabetes incidence
[24] . In the Nurs-
es’ Health Study, a 26% increase in risk of diabetes was
found per serving of red meat
[25] .
Clinical and experimental evidence provides some
support for the biologic plausibility of red meat and poul-
try intake as risk factors for diabetes. Specifically, meats
tend to be higher in saturated fat, and higher saturated fat
intake has long been associated with increased risk of
type 2 diabetes
[36–40] . Several studies have also shown
a positive association between saturated fat intake and
hyperinsulinemia
[41–43] and, in these studies, the high
levels of saturated fats were primarily from red meat and
dairy fat. Evidence also links the low polyunsaturated to
saturated fat ratio in meats, particularly red meats, to di-
abetes incidence
[44] . More recent evidence also raises
the possibility of a pathway where insulin resistance is
produced from chronically elevated levels of glucagon
and cortisol
[11, 12] – hormones that have been found to
be associated with higher meat intake
[13–15] . Our null
findings for fish intake may have resulted from a protec-
tive effect caused from –3 fatty acids and are also con-
sistent with studies indicating that those consuming fish
are less likely to develop chemically induced diabetes
[45,
46]
.
Our findings relating certain processed meats to a
higher risk of diabetes are similar to data from the Health
Professionals Study
[24] . In the Health Professionals co-
hort, consumption of processed meats (bacon, frankfurt-
ers, sausage, salami, bologna) at least 5 times per week
was associated with a 45% increase (RR = 1.45; 95% CI
1.14–1.86) in risk of type 2 diabetes
[24] . In the Nurses’
Health Study, each serving of processed meats contrib-
uted to a 38% increase in risk of diabetes, and indepen-
dent associations with specific processed meats (i.e. ba-
con, hot dogs) were also identified
[25] . Is the processing
method used in salting and curing meats an independent
risk factor for diabetes? Lijinsky
[47] noted that the ni-
trates, nitrites, and nitrosamines that are used for meat
preservation can all increase the body’s nitrosamine ex-
posure (i.e. nitrite and amine interaction). Some nitrosa-
mines are known to be  -cell toxins and several studies
Table 5. Multivariate model relating long-term (over a 17-year in-
terval) diet patterns, BMI, and weight change (over a 17-year in-
terval) to diabetes among subjects in the study population
Variable OR
1
95% CI
Long-term vegetarian
2
1.00 [referent]
Long-term occasional
meat intake
3
0.94 0.37–2.42
Long-term non-vegetarian
4
1.38 1.06–1.80
Baseline BMI by weight change
5
Weight gain >10 kg
<25 kg/m
2
0.62 0.33–1.16
25–30 kg/m
2
0.82 0.41–1.63
>30 kg/m
2
2.96 1.50–5.84
Weight gain 5–9 kg
<25 kg/m
2
0.26 0.14–0.50
25–30 kg/m
2
0.63 0.32–1.23
>30 kg/m
2
1.90 0.83–4.38
Weight stable (<5 kg loss or gain)
<25 kg/m
2
0.47 0.34–0.65
25–30 kg/m
2
1.00 [referent]
>30 kg/m
2
1.79 1.10–2.92
Weight loss 5–9 kg
<25 kg/m
2
0.83 0.51–1.34
25–30 kg/m
2
1.52 1.02–2.27
>30 kg/m
2
3.10 1.65–5.85
Weight loss >10 kg
<25 kg/m
2
1.17 0.61–2.25
25–30 kg/m
2
1.25 0.75–2.09
>30 kg/m
2
5.32 3.35–8.44
1
Odds ratios from a multivariate logistic regression model
that includes the variables in the table, age, and sex.
2
Participants who identified themselves as eating no meat in
both 1960 and 1976.
3
Participants who identified themselves as eating meat less
than once per week in 1960 and meat intake in 1976.
4
Participants who identified themselves as eating meat once
or more per week in both 1960 and 1976.
5
Weight change between 1960 and 1976.
Vang /Singh /Lee /Haddad /Brinegar
Ann Nutr Metab 2008;52:96–104
102
have identified a positive association between the con-
sumption of foods with high content of nitrites and nitro-
samines and risk of type 1 diabetes
[48–50] .
When interpreting the findings of our study it is im-
portant to note that the risk factor association with meats
may be an indicator of risk due to a diet pattern, and not
simply the independent effect of meats. For example, in
the Health Professionals Study
[51] the authors evaluated
the diabetes risk associated with two diet patterns: (1) a
‘prudent’ diet pattern (characterized by high consump-
tion of vegetables, legumes, fruit, whole grains, fish and
poultry), and (2) a ‘Western’ diet pattern (characterized
by high consumption of red meat, processed meat, re-
fined grains, french fries, high-fat dairy products and
desserts). The prudent dietary pattern was associated
with modestly reduced risk for type 2 diabetes (RR = 0.84;
95% CI 0.70–1.00), while the Western diet pattern was as-
sociated with a 59% increased risk for type 2 diabetes
(RR = 1.59; CI 1.32–1.93).
These findings suggest that some of the risk due to
meat intake observed in our study may be due to the ab-
sence of certain plant foods that contribute to a preven-
tive profile for this disease. Plant foods that have been
identified as possible protective against diabetes include:
legumes
[52] , soy [53] , whole grains [54] , nuts [55] and
some fruits
[56, 57] . Vegetable fat (found in non-animal
products including fruits, vegetables, grains, and nuts)
has also been associated with a lower risk of diabetes
[58] .
Also noteworthy are dietary intervention trial data indi-
cating a decrease in inflammatory markers (CRP, IL-6,
IL-7, IL-18) among metabolic syndrome patients who in-
creased their plant food consumption during a 2-year pe-
riod
[59] .
When interpreting the risk factor association with
meat intake reported by our study it is also important to
note that further analyses revealed the possibility of a
pathway whereby meat intake contributes to obesity – a
strong risk factor for diabetes. Specifically, when we in-
cluded BMI in the models depicted in tables 4 and 5 , we
found that the significant 74% increase in the risk of dia-
betes for long-term meat eaters (vs. long-term vegetari-
ans) reduced to a significant 34–38% increase in risk. Van
Dam et al.
[51] obtained similar findings indicating that
the positive association between dietary fat (total, satu-
rated) and diabetes was not evident after multivariate ad-
justment for BMI adjustment.
Furthermore, there have been studies that have identi-
fied a correlation between vegetarian diet (zero or low
meat intake) and lower levels of adiposity. Specifically,
Key et al.
[54] have shown that vegetarians have a signif-
icantly lower BMI than non-vegetarians. This putative
antiobesity effect of the vegetarian diet may be attribut-
able to any of the following possible mechanisms: (1) The
lower total energy intake (i.e. due to lower total fat, zero
animal fat, zero animal protein) of many vegetarian diet
patterns. (2) The lower saturated fat intake of many veg-
etarian diet patterns (i.e. where a high-fat diet promotes
obesity to a greater extent than an isocaloric low-fat in-
take). (3) The contribution of higher fiber intake (as much
as twice the levels in non-vegetarians
[51] ) among vege-
tarians to increased satiety (leading to appetite depres-
sion) and decreased between-meal snacking
[52] . (4) The
contribution of higher complex carbohydrate intake in
vegetarians to an increased resting metabolic
[50] rate or
insulin-mediated thermogenesis that can retard weight
gain
[53] . Sabate and Blix [60] have reported similar find-
ings from the AHSs indicating a particularly high preva-
lence of obesity among non-vegetarian cohort members.
Taken together, the findings from the AHSs ( tables 4 ,
5 )
[60] suggest that much of the mechanism by which
long-term meat intake produces an increased risk may be
due to the obesity that results from this diet pattern. It
remains noteworthy that we continued to find an asso-
ciation with weekly meat intake even after very detailed
control for obesity and weight gain ( table 5 ) – a finding
that supports causal pathways for meat intake that are
unrelated to adiposity. Further study of these effects is
needed.
There are a number of limitations of our study that
need to be considered. First, our endpoint was an indica-
tor of diabetes that was based on self-report, and the end-
point does not identify the type of diabetes. In a validity
study of self-reported diabetes, Goldman et al.
[61] found
that 96% of those who reported diabetes were confirmed
to have diabetes. Also, sensitivity
[62] and specificity of
self-reported diabetes have also been shown to be high
( 1 80%) in validity studies
[63] . Since the average age of
our study population at diagnosis was approximately 56
years, we posit that virtually all of our cases were type
2 diabetics. The measurement error in the assessment
of diet is also noteworthy; however, our validity studies
from Adventists indicate excellent validity for the mea-
surement of all meats and specific meats
[30, 31, 64] . Wil-
lett
[65] has noted that measurement error of diet-disease
relations would tend to attenuate associations, indicating
that the risk estimates reported for meat intake here could
be stronger. Finally, we also note that the small size of the
occasional meat intake category (11 events among 224
subjects) limited our power to detect significant diet-dia-
betes relations in this group. Further studies of the ‘semi-
Meat Intake and Diabetes
Ann Nutr Metab 2008;52:96–104
103
vegetarians’ are needed given the growing popularity of
this diet pattern in Western nations.
In conclusion, our findings raise the possibility that
red meat, poultry, and factors in processed meats are risk
factors for diabetes. Some of the risk due to meat intake
seen in our study may be attributable to a pathway where-
by adults following this diet pattern are not maintaining
healthy weight.
A c k n o w l e d g m e n t s
This study has been possible with the support from the Adven-
tist Mortality Study (grant: CA 14703, HL 26210) and the Adven-
tist Health Study (grant: CA 18186, AG 01582). Thanks go out to
the staff and faculty of Loma Linda University, School of Public
Health for their support.
References
1 Mokdad AH, Ford ES, Bowman BA, Nelson
DE, Engelgau MM, Vinicor F, Marks JS: Dia-
betes trends in the US: 1990–1998. Diabetes
Care 2000;
23: 1278–1283.
2 King H, Aubert RE, Herman WH: Global
burden of diabetes, 1995–2025: prevalence,
numerical estimates, and projections. Dia-
betes Care 1998;
21: 1414–1431.
3 Astrup A: Healthy lifestyles in Europe: pre-
vention of obesity and type II diabetes by diet
and physical activity. Public Health Nutr
2001;
4: 499–515.
4 Ford ES, Mokdad AH: Fruit and vegetable
consumption and diabetes mellitus inci-
dence among US adults. Prev Med 2001;
32:
33–39.
5 Tuomilehto J, Lindstrom J, Eriksson JG, Val-
le TT, Hamalainen H, Ilanne-Parikka P,
Keinanen-Kiukaanniemi S, Laakso M,
Louheranta A, Rastas M, Salminen V,
Uusitupa M, Finnish Diabetes Prevention
Study Group: Prevention of type 2 diabetes
mellitus by changes in lifestyle among sub-
jects with impaired glucose tolerance. N
Engl J Med 2001;
344: 1343–1350.
6 Knowler WC, Barrett-Connor E, Fowler SE,
Hamman RF, Lachin JM, Walker EA: Na-
than DM, Diabetes Prevention Program Re-
search Group: Reduction in the incidence of
type 2 diabetes with lifestyle intervention or
metformin. N Engl J Med 2002;
346: 393–
403.
7 Rewers M, Hamman RF: Risk factors for
non-insulin-dependent diabetes; in Diabe-
tes in America, ed 2. National Diabetes Data
Group, National Institutes of Health Nation-
al Institute of Diabetes and Digestive and
Kidney Diseases, NIH Publ No 95-1468,
1995.
8 Reaven GM: Insulin resistance, the insulin
resistance syndrome, and cardiovascular
disease. Panminerva Med 2005;
47: 201–210.
9 Ferrannini E, Galvan AQ, Gastaldelli A,
Camastra S, Sironi AM, Toschi E, Baldi S,
Frascerra S, Monzani F, Antonelli A, Nan-
nipieri M, Mari A, Seghieri G, Natali A: In-
sulin: new roles for an ancient hormone. Eur
J Clin Invest 1999;
29: 842–852.
10 Schulze MB, Hoffmann K, Manson JE, Wil-
lett WC, Meigs JB, Weikert C, Heidemann C,
Colditz GA, Hu FB: Dietary pattern, inflam-
mation, and incidence of type 2 diabetes in
women. Am J Clin Nutr 2005;
82: 675–684.
11 Provonsha S, Wade C, Sherma A: Syndrome-
AC: non-insulin-dependent diabetes melli-
tus and the anabolic/catabolic paradox. Med
Hypotheses 1998;
51: 429–438.
12 Oppert JM, Nadeau A, Tremblay A, Despres
JP, Theriault G, Deriaz O, Bouchard C: Plas-
ma glucose, insulin, and glucagon before and
after long-term overfeeding in identical
twins. Metabolism 1995;
44: 96–105.
13 Charlton MR, Adey DB, Nair KS: Evidence
for a catabolic role of glucagon during an
amino acid load. J Clin Invest 1996;
98: 90–
99.
14 Gray R, Butler P, Beers T, Kryshak E, Rizza
R: Comparison of the ability of bread versus
bread plus meat to treat and prevent subse-
quent hypoglycemia in patients with insu-
lin-dependent diabetes mellitus. J Clin En-
docrinol Metab 1996;
81: 1508–1511.
15 Hubbard R, Kosch C, Sanchez A, Sabate J,
Berk L, Shavlik G: Effects of dietary protein
on serum insulin and glucagon levels in hy-
per- and normocholesterolemic men. Ath-
erosclerosis 1989;
76: 55–61.
16 Pereira MA, Jacobs DR Jr, Van Horn L, Slat-
tery ML, Kartashov AI, Ludwig DS: Dairy
consumption, obesity, and the insulin resis-
tance syndrome in young adults: the CAR-
DIA Study. JAMA 2002;
287: 2081–2089.
17 Cohen D, Dodds R, Viberti G: Effect of pro-
tein restriction in insulin dependent diabet-
ics at risk of nephropathy. Br Med J (Clin Res
Ed) 1987;
294: 795–798.
18 Evanoff G, Thompson C, Brown J, Weinman
E: Prolonged dietary protein restriction in
diabetic nephropathy. Arch Intern Med
1989;
149: 1129–1133.
19 Gin H, Aparicio M, Potaux L, Merville P,
Combe C, de Precigout V, Bouchet JL, Au-
bertin J: Low-protein, low-phosphorus diet
and tissue insulin sensitivity in insulin-de-
pendent diabetic patients with chronic renal
failure. Nephron 1991;
57: 411–415.
20 Bjelke E: Epidemiologic studies of cancer of
the stomach, colon, and rectum; with special
emphasis on the role of diet. Scand J Gastro-
enterol Suppl 1974;
31: 1–235.
21 West K: Epidemiology of Diabetes and Its
Vascular Lesions. New York, Elsevier North-
Holland, 1978.
22 Gear JS, Mann JI, Thorogood M, Carter R,
Jelfs R: Biochemical and haematological
variables in vegetarians. Br Med J 1980;
280:
1415.
23 Meyer K, Jacobs D, Kushi L, Folsom A: Di-
etary fat and incidence of type 2 diabetes in
older Iowa women. Diabetes Care 2001;
24:
1528–1535.
24 Van Dam R, Stampfer M, Willett W, Hu F,
Rimm E: Dietary fat and meat intake in rela-
tion to risk of type 2 diabetes in men. Diabe-
tes Care 2002;
25: 417–424.
25 Fung TT, Schulze M, Manson JE, Willett
WC, Hu FB: Dietary patterns, meat intake,
and the risk of type 2 diabetes in women.
Arch Intern Med 2004;
164: 2235–2240.
26 Snowdon DA: Animal product consumption
and mortality because of all causes com-
bined, coronary heart disease, stroke, diabe-
tes, and cancer in Seventh-day Adventists.
Am J Clin Nutr 1988;
48: 739–748.
27 Feskens E, Bowles C, Kromhout D: Inverse
association between fish intake and risk of
glucose intolerance in normoglycemic elder-
ly men and women. Diabetes Care 1991;
14:
935–941.
28 Esfarjani F, Azar MA, Gafarpour M: IDDM
and early exposure of infant to cow’s milk
and solid food. Indian J Pediatr 2001;
68: 107–
110.
29 Meloni T, Marinaro AM, Mannazzu MC,
Ogana A, La Vecchia C, Negri E, Colombo C:
IDDM and early infant feeding. Sardinian
case-control study. Diabetes Care 1997;
20:
340–342.
30 Kahn HA, Phillips RL, Snowdon DA, Choi
W: Association between reported diet and
all-cause mortality. Twenty-one-year follow-
up on 27,530 adult Seventh-day Adventists.
Am J Epidemiol 1984;
119: 775–787.
Vang /Singh /Lee /Haddad /Brinegar
Ann Nutr Metab 2008;52:96–104
104
31 Beeson WL, Mills PK, Phillips RL, Andress
M, Fraser GE: Chronic disease among Sev-
enth-day Adventists, a low-risk group. Ratio-
nale, methodology, and description of the
population. Cancer 1989;
64: 570–581.
32 Lemon F, Walden R, Woods R: Cancer of the
lung and mouth in Seventh-day Adventists.
Cancer 1964;
17: 485–497.
33 Singh PN, Haddad E, Knutsen SF, Fraser GE:
The effect of menopause on the relation be-
tween weight gain and mortality among
women. Menopause 2001;
8: 314–320.
34 Singh PN, Lindsted KD: Body mass and 26-
year risk of mortality from specific diseases
among women who never smoked. Epidemi-
ology 1998;
9: 246–254.
35 Kaye S, Folsom A, Sprafka J, Prineas R, Wal-
lace R: Increased incidence of diabetes mel-
litus in relation to abdominal adiposity in
older women. J Clin Epidemiol 1991;
44: 329–
334.
36 Feskens E, Virtanen S, Ransanen L, Tuo-
milehto J, Stengard J, Pekkanen J, Nissinen
A, Kromhout D: Dietary factors determin-
ing diabetes and impaired glucose tolerance.
A 20-year follow-up of the Finnish and
Dutch cohorts of the Seven Countries Study.
Diabetes Care 1995;
18: 1104–1112.
37 Feskens EJ, Kromhout D: Habitual dietary
intake and glucose tolerance in euglycemic
men: the Zutphen Study. Int J Epidemiol
1990;
19: 953–959.
38 Marshall JA, Hamman R, Baxter J: High-fat,
low-carbohydrate diet and the etiology of
non-insulin-dependent diabetes mellitus:
the San Luis Valley Diabetes Study. Am J Ep-
idemiol 1991;
134: 590–603.
39 Marshall JA, Hoag S, Shetterly S, Hamman
RF: Dietary fat predicts conversion from im-
paired glucose tolerance to NIDDM. The San
Luis Valley Diabetes Study. Diabetes Care
1994;
17: 50–56.
40 Fujimoto WY, Bergstrom RW, Boyko EJ,
Chen K, Kahn SE, Leonetti DL, McNeely MJ,
Newell LL, Shofer JB, Wahl PW: Type 2 dia-
betes and the metabolic syndrome in Japa-
nese Americans. Diabetes Res Clin Pract
2000;
50(suppl):S73–S76.
41 Maron D, Fair J, Haskell W: Saturated fat in-
take and insulin resistance in men with cor-
onary artery disease. The Stanford Coronary
Risk Intervention Project. Circulation 1991;
84: 2020–2027.
42 Parker D, Weiss S, Troisi R, Cassano P, Vo-
konas P, Landsberg L: Relationship of dietary
saturated fatty acids and body habitus to se-
rum insulin concentrations: the Normative
Aging Study. Am J Clin Nutr 1993;
58: 129–
136.
43 Virtanen SM, Feskens EJ, Rasanen L, Fidan-
za F, Tuomilehto J, Giampaoli S, Nissinen A,
Kromhout D: Comparison of diets of diabet-
ic and non-diabetic elderly men in Finland,
The Netherlands and Italy. Eur J Clin Nutr
2000;
54: 181–186.
44 Nickander KK, McPhee BR, Low PA,
Tritschler H: Alpha-lipoic acid: antioxidant
potency against lipid peroxidation of neural
tissues in vitro and implications for diabetic
neuropathy. Free Radic Biol Med 1996;
21:
631–639.
45 Axelrod L, Camuso J, Williams E, Kleinman
K, Briones E, Schoenfeld D: Effects of a small
quantity of omega–3 fatty acids on cardio-
vascular risk factors in NIDDM. A random-
ized, prospective, double-blind, controlled
study. Diabetes Care 1994;
17: 37–44.
46 Krishna Mohan I, Das UN: Prevention of
chemically induced diabetes mellitus in
experimental animals by polyunsaturated
fatty acids. Nutrition 2001;
17: 126–151.
47 Lijinsky W: N-Nitroso compounds in the
diet. Mutation Res 1999;
443: 129–138.
48 Vir tanen S, Ja ak kola L, Ra sa nen L , Ylonen K,
Aro A, Lounamaa R, Akerblom H, Tuo-
milehto J: Nitrate and nitrite intake and the
risk of type 1 diabetes in Finnish children:
Childhood Diabetes in Finland Study Group.
Diabetes Med 1994;
11: 656–662.
49 Helgason T, Jonasson MR. Evidence for a
food additive as a cause of ketosis-prone dia-
betes. Lancet 1981;
2: 716–720.
50 Dahlquist GG: Dietary factors and the risk of
developing insulin dependent diabetes in
childhood. BMJ 1990;
300: 1302–1306.
51 Van Dam RM, Rimm EB, Willett WC,
Stampfer MJ, Hu FB: Dietary patterns and
risk for type 2 diabetes mellitus in US men.
Ann Intern Med 2002;
136: 201–209.
52 Salmeron J, Ascherio A, Rimm EB, Colditz
GA, Spiegelman D, Jenkins DJ, Stampfer MJ,
Wing AL, Willett WC: Dietary fiber, glyce-
mic load, and risk of NIDDM in men. Diabe-
tes Care 1997;
20: 545–550.
53 Friedman M, Brandon DL: Nutritional and
health benefits of soy proteins. J Agric Food
Chem 2001;
49: 1069–1086.
54 Key TJ, Thorogood M, Appleby PN, Burr
ML: Dietary habits and mortality in 11,000
vegetarians and health conscious people: re-
sults of a 17-year follow-up. BMJ 1996;
313:
775–779.
55 Jiang R, Manson JE, Stampfer MJ, Liu S, Wil-
lett WC, Hu FB: Nut and peanut butter con-
sumption and risk of type 2 diabetes in wom-
en. JAMA 2002;
288: 2554–2560.
56 Salmeron J, Manson JE, Stampfer MJ, Col-
ditz GA, Wing AL, Willett WC: Dietary fi-
ber, glycemic load, and risk of non-insulin-
dependent diabetes mellitus in women.
JAMA 1997;
6: 472–477.
57 Sargeant LA, Khaw KT, Bingham S, Day NE,
Luben RN, Oakes S, Welch A, Wareham NJ:
Fruit and vegetable intake and population
glycosylated haemoglobin levels: the EPIC-
Norfolk Study. Eur J Clin Nutr 2001;
5: 342–
348.
58 Feskens, E: Can diabetes be prevented by
vegetable fat? Diabetes Care 2001;
24: 1517–
1518.
59 Esposito K, Marfella R, Ciotola M, Di Palo C,
Giugliano F, Giugliano G, D’Armiento M,
D’Andrea F, Giugliano D: Effect of a Medi-
terranean-style diet on endothelial dysfunc-
tion and markers of vascular inflammation
in the metabolic syndrome: a randomized
trial. JAMA 2004;
292: 1440–1446.
60 Sabate J, Blix G: Vegetarian diets and obesity
prevention; in Sabate J (ed): Vegetarian Nu-
trition. Boca Raton, CRC Press, 2001, pp 91–
108.
61 Goldman N, Lin I-F, Weinstein M, Lin YH:
Evaluating the quality of self-reports of hy-
pertension and diabetes. J Clin Epidemiol
2003;
56: 148–154.
62 Wu SC, Li CY, Ke DS: The agreement be-
tween self-reporting and clinical diagnosis
for selected medical conditions among the
elderly in Taiwan. Public Health 2000;
114:
137–142.
63 Bensen JT, Liese AD, Rushing JT, Province
M, Folsom AR, Rich SS, Higgins M: Accu-
racy of proband reported family history: the
NHLBI Family Heart Study (FHS). Genet
Epidemiol 1999;
17: 141–150.
64 Fraser GE, Lindsted KD, Knusten SF, Beeson
WL, Bennett H, Shavlik D: Validity of di-
etary recall over 20 years among California
Seventh-day Adventists. Am J Epidemiol
1998;
148: 810–818.
65 Willett WC: Epidemiologic studies of diet
and cancer. Prog Clin Biol Res 1990;
346: 59–
68.
