Role of Life-style and Dietary Habits in Risk of Cancer among

Seventh-Day Adventists
Roland L. Phillips
Cancer Res 1975;35:3513-3522.

Updated version

E-mail alerts
Reprints and
Subscriptions
Permissions

Access the most recent version of this article at:

http://cancerres.aacrjournals.org/content/35/11_Part_2/3513

Sign up to receive free email-alerts related to this article or journal.

To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at pubs@aacr.org.
To request permission to re-use all or part of this article, contact the AACR Publications
Department at permissions@aacr.org.

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

(CANCER RESEARCH 35,3513-3522,Novemberl975J

Role of Life-style and Dietary Habits in Risk of Cancer among

Seventh-Day Adventists'
Roland L PhiHip.
Department of Biostatisticsand Epidemiology, Loma Linda University School of Health, Loma Linda, California 92354

unusual

Summary

life-style

characteristics

(Table

1) which suggest

that they may be a particularly fruitful study population in

The Seventh-Day Adventist population abstains from
smoking
and drinking;
about
50% follow a lac
to-ovo-vegetanian
diet; and most avoid the use of coffee,

elucidating the role of dietary factors and possibly other

life-style variables in either promoting or preventing the
carcinogenic process.

tea, hot condiments, and spices. Existing data on cancer

mortality in Seventh-Day Adventists clearly document

Seventh-Day Adventists are an evangelical religious

denomination with about 2.5 million members worldwide

mortality

and about

rates that are 50 to 70% of general population

rates for most cancer sites that are unrelated to smoking

and drinking. Several approachesto determining whether
this

reduced risk

is due to the unique Seventh-Day

500,000

members

in North

America.

About

100,000 of these live in California. Aside from the lack of

smoking and drinking, dietary habits appear to be the key
distinguishing characteristic of the Seventh-Day Adventist

Adventist life-style or selective factors related to who life-style. Approximately one-half of the Seventh-Day Ad
chooses to become and remain a Seventh-Day Adventist ventist population follow a lacto-ovo-vegetanian diet, and
virtually all of them abstain from using pork products. They
are described.
A comparison ofthe mortality experienceof Seventh-Day use vegetables, fruits, whole grains, and nuts abundantly
Adventist and non-Seventh-DayAdventist physiciansshows and avoid the use of coffee, tea, hot condiments, and,
equal cancer mortality, which is consistentwith the hypoth
possibly, highly refined foods. This dietary pattern has been
esis that the apparent reduced risk of cancer death in all
Adventists may be due to selective factors. However, the

results of a small case-control study of colon and breast

cancer among Adventists show statistically significant rela
tive risks for colon cancer of 2.8 for past use of meat. For
current food use, the significant relative risks are 2.3 for

beef, 2.7 for lamb, and 2. 1 for a combined group of highly

saturated fat foods. This strongly suggeststhat the lacto
ovo-vegetarian diet may protect against colon cancer.
However, the evidence linking diet to breast cancer is less
clear. Becauseof the marked variability in dietary habits
within

the Seventh-Day

Adventist

population,

they will

strongly recommended
among Adventists for over 100
years. With the exception of abstinence from alcohol and
pork products, the degree of adherence to these dietary

recommendationsis somewhat variable.

Comparative data for the frequency of use of 3 types of
foods hypothesized to be etiologically related to certain
cancer sites (6, 7, 16, 28), have been obtained for a small
sample of Adventists and non-Adventists enrolled in a study
of stool bacteria, currently being carried out in collaboration
with Dr. Sydney Finegold. Table 2 shows that the intake of

meat and coffee among Adventists is markedly less than

that among the general population.

It is important

to note

be a productive group for further study of diet and cancer.

that the Adventist population demonstrates significant

Introduction

variability of dietary habits. It is also worthwhile to recall

that observational investigations have been unsuccessful in
demonstrating
a relationship between dietary habits and

Numerous components of life-style have been suggested

or demonstrated to have a relationship to the major sites of

CHD2 despite the great likelihood that a significant bio

logical association

exists. Lack of significant

variation

in

cancer in this country, but aside from the well-established intake of the suspected causative nutrients within a single
relationship of smoking to risk of certain cancer sites, few of

the hypothesized relationships with life-style are backed by

sufficient evidence to allow firm conclusions to be made
(46). Interest in studying cancer occurrence among Seventh

Day Adventists was initially generated becausethey are a

nonsmoking and nondrinking population by Church pro
scription (33, 52). However, Adventists have many other

study population

is the most likely explanation

for failure

to observe a significant association. No doubt this same

difficulty will plague observational studies of dietary habits
and cancer. For this reason, the significant variability of
dietary habits among Adventists may provide a unique
opportunity for investigating dietary determinants of can
cer.

This report presents a reanalysis of data collected in a

1 Presented

at

the

Conference

on

Nutrition

in

the

Causation

of

Cancer,

May 19to 22, 1975,Key Biscayne,Fla. This work waspartially supported

by grant number I ROI CA 14703-02from National Cancer Institute and
funds from The American Health Foundation.

2 The

abbreviations

used

are:

CHD,

coronary

heart

disease;

LLU,

Loma Linda University; USC, University ofSouthern California.

NOVEMBER 1975

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

3513

R. L. Phillips
Table I
Unique characteristics of typical Seventh-Day Adventist life-style

as recorded on the death certificates. Their data show that

mortality from most cancer sites in Adventists is con

siderably below cancer mortality in the general popula

Abstinence from smoking and alcoholic beverages (one-third of males

tion. Their studies also documented reduced mortality
previous smokers)
from numerous other causes, with resultant greater life
Lacto-ovo-vegetarian diet ( 40-50%)
Abstinence from pork products and other biblically defined uncleanexpectancy among Adventists. Table 3 shows the age-sex
meats
composition of their study population, compared with the
Avoid caffeine-containing beverages
general California population.
Chart I summarizes the
Avoid hot condiments and spices
8-year
cancer
mortality
experience
of this Seventh-Day
? Avoid highlyrefinedfoods
Abundant use of fruits, whole grains, vegetables, and nuts
Adventist population. The age-sex standardized cancer
Great stress on quality education (including health education) and family
mortality in Adventists is one-half to two-thirds of the
life
generalpopulation
cancermortality rates.
? Conservative
sexualmores
The
obvious
initial
question is to what extent their absti
?? Regularexercise
and rest
nence from smoking and drinking accounts for the lower
7'?Decreasedstress,
anxiety,
etc.

risk of Adventists to death from cancer. A large share of

Table 2

Table 3

Percentage ofpersons using difftrent amounts of meat and coffee among

California Adventists and non-Adventists

FrequencyofcurrentuseAdventists
34)Meat,

Age-sexdistribution of the Adventist population at the beginningof the

study, compared with the 1960 California population

(n = 151)Non-Adventists
(n =

times/wk10.88.85-6times/wk3.926.57+
times/wk3.458.8Total100.0100.0BeefNever

Calif.
popula

Calif.
popula

fishNever
poultry, or
(life-time)20.50Never
users)25.80<
(past
time/month12.80l-2times/mo.10.30l-2times/wk12.35.93-4
I

tionin

Ageat

tionin

19603544No.
1960Adventists
%%
No.
entry(yr)MaleFemaleAdventists
28.3
24.2
27.5
18.7
19.3
17.1
12.7
9.6
5.0
2.234.6 0.95,619

45-54
2,959
2,287
5564
65-74
2,099
75-84
1,173
1.4All
271
85+3,461

ages
never43.30<
or almost
time/wk19.88.8l-2times/wk12.064.73-4times/wk17.523.55-6times/wk4.82.97+
I

5,347
4,838
4,441
2,368
597

100.0100.023,210

%%
24.2
23.0
25.7
19.1
20.8
14.3
19.1
10.2
6.5
2.633.0
100.0100.0

>3512,250

a Includes

those

who entered

study

in 1958 (29,745) plus those who

enteredin 1960(5,715).
times/wk2.40Total100.0100.0Coffee
@-MALE

110TH

@oo
- -@@@

__ @
-.--- f

- ---

(cups/day)Never76.611.8<I6.65.8I9.620.52-33.629.44-51.820.55+1.811.8Total100.0100.0

@
@

liv.

LU@J@IJ
__

aim tsi,u. sLaoouI si

NM@1X

n,esmsuivci? u@trn* LYMPIIOMA

ALL
OTN(I ALL

@V(1Nf

I (AN@I

(ANLIl
I

L@@110161@
RELATED@@'-Ul@El.ATEDtoClGAlETTES

t. CIGARETTES
5/orALCOHOL

ALCOHOL

Chart I. Standarized mortality ratios (SMR)for various cancers

previously published study of Adventists as well as prelimi

nary summary results of 2 other studies we have recently

among California Seventh-Day Adventists by sex, aged 35+, from 1958 to

completed.

CNSb, includes benign and unspecified central nervous system neoplasms;

obs., observed.

Cancer Mortality in Seventh-Day Adventists

1965. RESP., respiratory; ESOPH., esophageal;GI., gastrointestinal;

3 In

all

charts,

standardized

mortality

ratios

(O/E)

100

where

number of expected deaths derived by accumulating the yearly expected

Lemon and Walden (32), and Lemon et a!. (33) carefully

followed a defined cohort of 35,460 Adventists living in
California during 1958 to 1965 by maintaining annual con
tacts with each subject. They tabulated the causes of death

3514

deaths obtained by applying the yearly age-sex-specificCalifornia death

rates to the corresponding age (10-year) and sex groups in the Adventist
population at risk at the beginning of each year, and 0 = number of
observed deaths in Adventists from 1958 to 1965; 95% confidence limits are
indicated.

CANCER RESEARCH

VOL. 35

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

Cancer in Seventh-Day Adventists

this reduction

in mortality

is clearly due to the expected

very low mortality rates of cancersknown to be related to

smoking and/or alcohol. However, other major sites, such
as gastrointestinal
and reproductive cancers, which are not
strongly related to smoking and alcohol, are significantly
below the general population.
This suggests that other
components of the Seventh-Day Adventist life-style may be
protective for these sites. It is noteworthy that the sites that
differ significantly are those most suspected of being related

to dietary habits. Kidney and central nervous system can

cers, as well as lymphoma, male leukemia, and all other
cancers (males only) do not differ significantly from those
ofthe general population.
Chart 2 details the Adventists' experience by sex for
selected gastrointestinal cancers. The rates are all in the
vicinity of 60 to 70% of general population rates, except for
gallbladder, which is not significantly below that of the
general population. With the exception of pancreas cancer,
no significant sex differentials are seen.
Table 4 shows the results of a recent hospital record
review of a sample of Seventh-Day Adventist and non-Sev
enth-Day Adventist colon cancer cases discharged from 2
California hospitals during 1965 to 1974. The anatomic
location within the colon was recorded using surgery,
sigmoidoscopy,
and pathology reports. Previous studies
have suggested that colon cancers tend to occur more

frequently in the right colon in low-risk populations and in

the left colon and rectum in high-risk populations (9, 17,
18). The data in Table 4 provide suggestive evidence that
Seventh-Day Adventists are indeed a low-risk population,
since their lesions are shifted somewhat to the right,
compared with those of non-Adventists, but the difference
does not quite reach significance at the 5% level.
Chart 3 indicates the mortality ratios for specific repro
ductive cancers for 2 age groups where numbers are
adequate. Postmenopausal
cancer of the breast, ovary, and
other uterus (endometrium and uterus cancer of unspecified
type) are significantly below the general population, while
prostate cancer mortality is not significantly different from
that of the general population. The low rates for cancer of
the cervix are probably a result of conservative sexual
practices in Seventh-Day Adventists.
Selection versus Life-style

Seventh-Day Adventists are by no means a representative

sample of the general population; indeed, they are a very
select group of the general population. For example, the
proportion of college-educated persons among Seventh-Day
Adventists is twice that of the general population. Thus,
before we attribute the low Adventist mortality risk for
cancer sites unrelated to smoking and alcohol to some other
aspect of the Adventist life-style, a crucial issue to be re
solved is whether the reduction in risk can be explained by
selective factors related to who chooses to become and re
main a Seventh-Day Adventist. Selective factors such as
socioeconomic status could conceivably be quite unrelated
to dietary habits or other life-style characteristics and, yet,
significantly influence cancer mortality rates (47). Clearly,
the alternative explanation is that one or more components
of the Adventist life-style are protective against cancer.
COLON
STAMEN
PANCIFAS GALL
Education Adjustment. A simplistic approach to this
SLADGAI
question is to adjust the mortality ratios for educational
E1F.m@.@MahUkAbs.ua
status. Chart 4 shows that adjustment for education does
Chart 2. Standardized mortality ratios (SMR) for gastrointestinal not significantly change the mortality ratios, thus eliminat
cancers among California Adventists by sex, age 35+, 1958 to 1965; obs., ing one very general measure of socioeconomic
status as a
observed.
possible explanation of the low Seventh-Day Adventist
mortality rates.
Appropriate Comparison Group. Another approach would
b,d.,,h.

40

78

28

14

27

18

be to compare Adventists with an easily identifiable sub

Table4
Anatomic location ofcolon-rectal cancer among a sample ofAdventist and
non-Adventist colon-rectal cancer cases in 2 California hospitals

group of the general population which might be more

similar to Adventists and, more particularly, to Adventists

% ofall
colon-rectal
cancersNon
@

Ioo@
AdventistAdventistsbSite(n
104)Cecum17.422.1Ascending
= 195)(n

and transverse1
.820.2Descending,

----

--- -@---.

sigmoid,53.844.2and
rectumLowrectum(<8cm)16.9
high
@

100.013.5
a A systematic
a All Adventist

sample of all non-Adventist

cases during 1965 to 1974.

0807

(11818

100.0
cases during

ShALT

@Ag@i3554
UA@s55+

1965 to 1974.

Chart 3. Standardized mortality

01818

PIOSTATI

mum
Ilflll..a@+

ratios (SMR)

for reproductive can

cers amongCalifornia Adventists, by age, 1958to 1965,obs., observed.

NOVEMBER 1975

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

3515

R. L. Phillips

ported at this conference by Hirayama (27) that stomach

cancer is related to both smoking and drinking.

The differential between Seventh-Day Adventists and the

general population for all causes ofdeath is reduced but not

@i
11111,1
CIIN

@c1@

ITOMACH

IDIALT

Stondordiz.d
@

Ag..Ssx

UTLIUS

OVAIY

PIOSTAfl

eliminated by using a comparable cohort of nonsmokers

from the general population as the comparison group. This

suggests that life-style factors other than nonsmoking may
contribute to the low general mortality among Adventists.

ML

c*iiai

br:

The persistence of the differentials between Seventh-Day

Chart 4. Standardizedmortality ratios (SMR) for variouscancers Adventists

and the general

population

for cancer

of the

colon, breast, ovary, and prostate, as well as leukemia and

according to whether education was accounted for in standardized
lymphoma, is consistent with the hypothesis that one or
mortality ratio calculation among California Adventists, age 35+, 1958 to
1965.Education adjustmentwas accomplishedby partitioning the annual more components of the Adventists life-style (other than
age-sex-cause-specific
California mortality rates for the period 1958to nonsmoking) may protect against these cancer sites.
1965 according to the 1960 education mortality

ratios for United States

whites given by Kitagawa and Hauser(30). obs., observed.

Furthermore, this observation tends to weaken the hypothe

sis that selective factors account for the differential. How

ever, the unique method of selectionof Hammond's popula

who choose to enroll in a prospective study. There is
convincing evidence that persons who enroll in large popula
tion studies by completing a questionnaire are at lower risk
of death than the rest of the population, many of whom are
institutionalized, etc. (48, 5 1). It occurred to us that such
data are readily available from the published data of

tion is dissimilar

to that of the Adventist

study. It is quite

conceivable that selective factors that may be related to

participation in Hammond's study are entirely different
from the selective factors that relate to choosing to become

and remain a Seventh-Day Adventist.

Time and Duration of Exposure to the Seventh-Day
Hammond's prospective study (20) and that it might be Adventists' Life-style. Another approach to assessing the

quite appropriate to compare Adventists with the nonsmok

ers in Hammond's population. The solid bars in Chart 5
depict standardized mortality ratios for Adventists recal
culated using Hammond's nonsmokers as the standard
(comparison) population. Due to the age groups for which
site-specific mortality rates are reported in Hammond's
publication, these data are limited to the indicated sites
among persons age 45 to 79. One would expect this to
virtually eliminate the differential between Adventists and
the general population for such sites as lung, mouth, and

relative role of life-style in explaining the low Adventist

rates is to compare thosewhojoined the Church at different

ages (Chart 6). If life-style differences are the key factor
explaining the gradient between Adventists and the general

population, then the earlier in life a person assumes the

Adventist life-style, the lower should behis risk. Conversely,
persons joining the Church late in life should have a risk

more nearly equal to the general population, especially if

@oo

pharynx. However, it is fascinating to note that an apparent

differential still exists, especially in females. This suggests
the possibility that factors other than smoking (possibly
dietary) could be related to these sites. However, the small
numbers and the wide confidence limits certainly restrict
@

any inferences.
In the lower part of Chart 5, the bars on the right, in
effect,
willing
cancer
cation

@:@1i
@i
Hi@J2
LUNG

IOTA 11810

LUNG

lUNG

MAIl

118*11

MOUTH

LADYIX

adjust for the selective factors that relate to who is

or able to enroll in a prospective study. Stomach
is limited to males only because Hammond's publi
did not give age-specific death rates for female

stomach cancer. Stomach and uterus are the only sites for
which the gradient between Seventh-Day Adventists and
the general population
was reduced by utilizing Ham
mond's nonsmokers as the comparison population. The

pattern for uterus cancer is possibly explained by either

the known inverse relationship

of endometrial

cancer

to

socioeconomic status (47) or by the possibility that female

nonsmokers in the general population are less sexually
promiscuous than smoking females (29) and thus are
more like the Seventh-Day Adventists. The elimination of
@
@

the gradient for male stomach cancer is consistent with the

possibility that dietary factors unique to the Adventist life

!1d@i@i

@I

15'1j@t

STOMACH COLON ShAlT

UTIIUS

OVADY

(MALI OMLY@IICTAL

HII@1
@1
PDOSTATI LIUKIMIA
ALL
ALL CALISIS
LYMPIIOMA CANCER

Ezp.ct.d
bas.don:

G.@e@alCoI,I
pop@Iono@

@p-iaa@,ord,
No@smoker@

style play only a minor role in stomach cancer or act as

Chart 5. Standardized mortality ratios (SMR) based on different

standard populations or various cancers among California Adventists,

promoters.

age45 to 79, 1958to 1965.obs., observed.

3516

It may

also emphasize

the observation

re

CANCER RESEARCH VOL. 35

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

Cancer in Seventh-Day Adventists

life-style very early in life were the critical factor. There are
no significant gradients except for stomach cancer and,
possibly, for all cancers. The latter is probably a reflection

of the near absence of smoking-related sites among persons

who joined early in life who are lesslikely to have smoked
in the past than those who joined at a later age. The picture

for stomach cancer is consistent with the importance of

exposure to etiological agents early in life, as suggested by
the migrant studies (19, 49). Persons who joined the
@@Li14iMT@
5@
I5D@
for4IThsrs
Church late in life are more likely to have been exposed
Chart 7. Standardized mortality ratios (SMR) for various cancers
early in life to foods that Adventists usually avoid (meat, amongCalifornia Adventists by length of church membership,both sexes,
coffee, etc.). Alternatively, if risk of certain sites of cancer age 35+, 1958to 1965.Lifetime
5DM, church membershipbeganat
is determined
more by habits in the recent past than
during childhood, one would not expect a gradient by age

18 yearsof age.obs.,observed.

that church membership began, as shown for colon and

primary

breast cancer in Chart 6.

among Adventists.
Comparison of Adventist versus non-Adventist Physicians.
One further approach to determining whether selective

Chart 7 compares life-time Adventists to recent converts,

irrespective of the age at which they became members. This
is consistent with the above observation for stomach cancer.

explanation

for the low cancer mortality

rates

The results for colon and breast, which show a lower risk for

factors or life-style explains the low cancer mortality in

Adventists is to compare the mortality experience of a

recent converts than life-time

group of Adventists and of non-Adventists who are very

members, are quite unex

pected. If it can be assumedthat recent converts are more

zealous about adhering to the Church recommendations
regarding diet and other aspects of life-style and that recent
habits are more important
than early life habits, this
observation would be consistent with the dietary hypotheses

for these sites. Although unsubstantiated

by any data,

long-time Adventists suspect a secular trend of an increas

ing proportion
of vegetarians among Adventists which
might also partially explain the pattern of lower risk in

recent converts. However, since these assumptions are

questionable, it may simply reflect the likely possibility that
persons who choose to become Adventists relatively late in
life are a highly select group who may already be at low risk
of colon and breast cancer before they adopt the Adventist
life-style, or they may indeed be living a life-style similar to
that of Adventists before joining and thus are more likely to
join a select group with a somewhat stringent life-style.
Because of the strong possibility that selective factors
may relate more strongly to joining the Church in later life

than in the teen years, Charts 6 and 7 do not provide much

evidence

for or against

@():

the hypothesis

that life-style is a

similar

in many respects, except for certain aspects of

life-style. It recently occurred to us that Adventist and

non-Adventist physicians would fulfill these criteria and
that existing records would allow easy follow-up of 2 such
groups of physicians. We are just completing a study of the
complete mortality experience of all male medical graduates
of LLU (3867) and USC (2416) medical schools who
graduated during the years 1914 to 1971 and 1901 to 1971,
respectively. We have so far obtained death certificates on
96.5% of the 863 deaths. These 2 groups of physicians are
clearly very similar in regard to socioeconomic status,
medical care, knowledge of health, etc., yet they differ in
other key life-style variables. We do not have information
on current or past religious affiliation of these 2 groups, but
we can safely assume that over 75% of the LLU graduates

are Adventists and lessthan 5% of the USC graduates are

Adventists. Thus, this is roughly a comparison of Adventist
versus non-Adventist physicians.

Chart 8 shows the results of this study presented as

standardized
mortality ratios derived from a life table
analysis. As expected, the death rates in physicians are
much lower than general population rates, but the startling
finding was that mortality rates in LLU and USC graduates

are essentially the same for all causesand all cancer. It is

interesting to note that the difference in cancer mortality
between Adventist physicians and the general population is
essentially the same as that observed between all Adventists
and the general population (Chart 1). Thus cancer mortality

in Adventist physicians is equal to the population from

@J
fluid
@

(WY.on

ceti@
iicou

if*n

AU aa,

*u CMSB

1934Y.ors

Chart 6. Standardized mortality ratios (SMR) for various cancers

among California Adventists by age at which church membershipbegan,
both sexes, age 35+, 1958 to 1965. The total observed (obs.) deaths for the
3 categories for any given cause is somewhat less than the total observed

deaths on previous tables becauseinformation regarding age at which

church membershipbeganwas available for only a subgroupof the total
Adventists study population.

which they come, whereas non-Adventist physicians have a

cancer mortality considerably below the general population.

The higher gastrointestinal and colon-rectal cancer deaths

in LLU graduates were unexpected. Differences between
LLU and USC for the indicated types of cancer are not
statistically significant because of small numbers. However,
the direction of these differences is certainly opposed to

the current dietary hypotheses for these sites, which sug

gests that intake of fat, meat, beef, or low-fiber foods are re

NOVEMBER 1975

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

3517

R. L. Phillips

which we will later relate to site-specific cancer incidence

rates among subgroupsthat do and do not adhereto certain

dietary patterns. However, we felt it would be worthwhile to
try to get some quicker, although somewhat less reliable,
information on the topic of diet and cancer from a
case-control study. Preliminary results are now available
from a recent interview study of 41 Adventist colon-rectal
cancer and 77 Adventist breast cancer patients discharged
from 2 Adventist-operated
hospitals during 1969 to 1973
AU
AU@
rnp.
bi
tillS
(N
cm@ caim CMkR (@ft RECTAL
and 3 types of age-, sex-, and race-matched Adventist con
.
CAKER
LLU
USC
trols. The food questions used in the structured inter
Chart 8. Standardized mortality ratios (SMR) for various causesof view were specifically designed to test the hypothesis that
death among all male graduates of LLU and USC medical schools. intake of high fat, low fiber, or both are associated with risk
Standardized mortality ratios = (O/E) x 100 where E = number of cx
of colon or breast cancer.

@i@iL

pected deaths derived by accumulating the yearly expected deaths oh

tamed by applying the age-decade (1920, 1930, 1940, etc.) cause-specific
death rates for white males in the United States to the corresponding
5-year age group in the physician population at risk and 0 = number of
observed deaths during 1914 to 1971; 95% confidence limits are shown for
each bar.

lated to colon cancer (4, 16, 28). We do not yet have any
data on how well Adventist physicians follow the typical
Adventist life-style or diet, but it seems highly unlikely that
these physicians eat more fat, meat, beef, or low-fiber foods
than USC physicians. This assumption is also upheld by the
gradient for CHD (Chart 8, far right), which is in the ex

pected direction and is statistically significant. One might

think this CHD gradient could be entirely explained by dif
ferences in smoking habits, but the extremely low mortality
ratios for respiratory cancer in both groups suggest that the
USC physicians are not smoking very much either. Thus it
seems likely that the CHD gradient is due to dietary differ
ences, which indeed makes the gradients for colon cancer
unexpected and contradictory to the dietary hypotheses for
this site. However, the small numbers of colon-rectal deaths
preclude drawing any firm conclusions.
The fact that the cancer mortality differential between
all Adventists and the general population is not seen be
tween Adventist and non-Adventist physicians is con
sistent with the hypothesis that selective factors are the
prime determinants of this differential. However, the dia
gram below indicates that 2 sets of selective factors (A
and B, which may be quite different) could relate to be
coming an Adventist physician, whereas the selective fac
tors relating to an individual in the general population be
coming a physician (C) may be less complex.
General
@

@@Adventists..j.@Adventist physicians

population

non-Adventist

physicians

If the selective factors for Steps B and C are dissimilar,

it would

be difficult

to infer

from

these data

that

then
the

selection hypothesis is a likely explanation of the reduced

cancer mortality among all Seventh-Day Adventists. It is
also quite possible that the life-style and dietary habits of
these 2 groups of physicians are much more similar than the
habits of all Adventists versus the general population.
Case-Control

Study of Seventh-Day

Adventists

We are presently 2 years into a 6-year prospective study

of the 100,000 Adventists currently living in California. At
the outset, we are collecting detailed information on diet
3518

Colon Cancer. The relative risks for reported usage of

food items by servings

per day, week, or month among

colon cancer cases and controls are shown in Table 5. For

the foods listed in Table 5, the results calculated separately
for the cases matched to the 3 different types of controls

showed an association with colon cancer of approximately

the same magnitude and direction. Thus, all controls were
combined for calculation of relative risks. Using relative
risk as a measure of association and x2 as a test of statistical
significance, it appears that any use of beef, lamb, fish, and
the heavy use of dairy products other than milk and other
high-fat foods are significantly related to risk of colon
cancer.
While current use of meat, poultry, or fish did not show a
relationship, such use 20 years ago gives one of the highest
relative risks. Although it is clearly impossible quantita
tively to determine frequency of food use 20 years ago, it is
possible, in Seventh-Day Adventists, to reliably assess
vegetarian versus nonvegetarian status in the distant past.
Adventists can usually distinctly remember when they
changed from nonvegetarian to vegetarian status, as many
did soon after joining the Church.
Based on the concept that the important variable may be
the total pattern of food used, rather than specific individual
foods, the results for an approximate index of intake for
several combined food groups are also presented in Table 5
(dairy products; fried foods; beef products = beef hambur
ger + beefsteak and other beef; highly saturated fat foods =
cheese + cheese + eggs + all meats and poultry). The
relative risks for these food groups are in the same general
direction but are not significantly higher than those for
individual components
of the groups. Milk,4 vegetarian
protein products, and green leafy vegetables show a non
significant negative association with colon cancer. These
are all foods that Adventist lacto-ovo-vegetarians are
likely to consume in large amounts, so the observed nega
tive associations
do not necessarily indicate that these
foods are protective. Vegetarian protein products are
highly processed, purified, fiber-free meat analogs de
rived from soy or gluten protein. Many of these products
contain little or no fat. Although the current pattern
of food use in colon cancer cases is likely to be different
from the pattern prior to diagnosis, it is quite possible
4 The

interview

form

did

not

determine

the

type

of

milk

used

and

recent diet survey shows that only 23% of the milk used by Adventists is

whole milk.

CANCER RESEARCH VOL. 35

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

Cancer in Seventh-Day Adventists

Table 5
Relative risk for use of individualfoods reported by Adventist colon cancer casesand Adventist controls

Forty-onecolon cancercaseswereeachmatchedby age,sex,and raceto 3 controls. Wherepossible,2 of the

controls were selected, respectively, from hospitalized cases of hernia and osteoarthritis, and the 3rd, from the
general Seventh-Day Adventist population.
data (case+,case,cont+,cont

)Dichotomy risk@Raw
vs.
(X

extremecPositively
Type of foodRelative

2.Y
2.5'

= ad/bc

3.1'
2.3

low

vs. <Xi'High

(22,19,36,70)
(21,20,31,74)

(11,19,13,70)
(I 1,20,18,74)

(12,29,14,91)

2.7'
I .6
2.3
1.9
1.7

3.4'
3.8
1.9

2.5'

( 18,23,34,71)
(34,6,75,30)
(34,6,79,26)
(30,1 1,65,41)

( 12,23, 11,71)
(9,6,12,30)
(18,6,41,26)
(23,11,34,41)

2.1

2.V

(31,10,63,43)

(20,10,32,43)

1.9

2.7'

(28,13,56,49)

(15,13,21,49)

1.9

2.3

(28,13,56,50)

(18,13.30,50)

2.0

1.7

(24,17,44,61)

(24,6,44,19)

1.8

2.0

(34,7,77,29)

(19,7,39,29)

(10,30,14,90)

2. I

associatedd
Milk ( 1/day vs. < 1/day)
Vegetarian protein products
(l/wk vs.<l/wk)
Green leafy vegetables ( l/wk
vs. <l/wk)0.5
risk (RR)

vs.

extremecDichotomy (X

<,@)bHigh

associate&'
Meat, any type, 20 yr ago (any
vs. none)
Beefproductsh (any vs. none)
Beefhamburger(any vs. none)
Lamb (any vs. none)
Fish (any vs. none)
Cheese( l/wk vs. < l/wk)
Icecream (any vs. none)
Dairy products except milk
(heavy + medium vs. light)
High saturated fat foods@
(heavy + medium vs. light)
Fried potatoes ( l/wk
vs.
< l/wk)
Fried foods(heavy + medium
vs. light)
Cake or pie ( l/wk vs. < 1/
wk)
Fresh fruit ( 1/day vs. < I/day)
White bread (most of the time)
(yesvs.no)
(7,26,4,31)Negatively
Pepper (frequent vs. seldom or
never)2.8'

a Relative

low

vs.

2.23.6'

2.1(26,9,51,50)'

(7,34,9,97)(9,9,14,50)

0.4

0.4

(4,36,24,82)

0.40.3

0.5(28,13,86,19)

(33,8,96,10)(28,7,86,6)(9,8,25,10)

where a(c)

cases reporting

higher

(lower)

(4,14,24,31)

usage, and b(d)

= all controls

combined reporting higher (lower) usage. A relative risk of I .0 indicates no association (i.e., subjects with high

and low usagehavethe samerisk).

0 The

reported

frequency

of

use

of

each

food

(times/week

or

day)

was

arbitrarily

divided

into

2 categories

at

the point X (indicated under each food) to approximate above and below the median of all controls except for

foods reported as never

used
by a sizable group where any versusnone was preferentially usedon the
assumption that recall was more accurate for this than for average number of times used.
C High

and

low

extremes

arbitrarily

selected

from

distribution

of

food

usage

in

all

controls

to

provide

adequatenumbersin both categories.It approximatesupper one-third vs. lower one-third.

d Positive

(negative)

2.O (O.5)
e Includes
I p is
g (a,

association

meat, poultry,

0.05, based on

c,

RR

1 .5 (

0.66)

for

both

hospital

and

general

population

controls

and

with either type of control.

b,

d)

see

Footnote

and fish.
@2corrected for continuity.
a

above.

The

inequality

of

c and

d between

individual

foods

is due

to

missing data.
A An

index of intake was constructed

by combining

the reported intake of individual foods in this category.

Reported usage of each food item was weighted by g/serving for the nutrient of interest (or the weight of an
average serving for beef). This resulting distribution of individual indices of usage was approximately divided
into tertiles, heavy, medium, and light (none).

(particularly
among Adventists) that the pattern after
diagnosis would shift toward less meat and fat con
sumption, which would simply decrease the likelihood of
finding associations with these foods.

It is quite clear that these results are supportive of the

hypothesis

that

beef, meat,

and saturated

fat or fat in

general are etiologically related to colon cancer. However, it

should also be noted that some highly refined but not
necessarily high-fat foods also show a positive but nonsig
nificant relationship to colon cancer (e.g., cake, pie, and
white bread). Green leafy vegetables that are quite high in
fiber are negatively associated. These associations
are

NOVEMBER 1975

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

3519

R. L. Phillips
consistent with Burkitt's hypothesis that low fiber intake is
a primary determinant of colon cancer (4). However, the
lack of statistical significance, the strong likilihood that
intake of these foods in Adventists is highly correlated with
low meat and low fat intake, and the strong negative corre
lation between fat and fiber content of most foods suggest
that the association with low fiber intake may be secondary

to a primary relationship with fat, meat, or beef intake.

Since the use of black pepper is strongly discouraged among
Adventists, its use may be somewhat of an indirect index of
how well a person follows the recommended diet, which may

well account for the observedassociationwith colon cancer.

Table 6 lists individual foods and combined food groups
Table 6
Foods not associatedwith colon cancer
Meat, poultry or fish (current use)
Beef steak

which were not associated with colon cancer. There is a lack

of association for several foods which might be predicted to
show an association, on the basis of the fat or fiber
hypothesis. However, the strong association with the high
fat combined food groups which include these individual
high-fat foods and the lack of association with the high- or
low-fiber combined food groups tends to support the
relative importance and credibility of the fat hypothesis.
Breast Cancer. Table 7 shows the results for the 77 breast

cancercasesand controls. Only 5 foods wereassociatedwith

breast cancer, but the association with fried potatoes was
highly significant (p < 0.01). It is important to note that the
next highest relative risk was for the use of hard fat
(margarine, butter, or shortening) for frying. This associa
tion together with the association with fried potatoes and
fried foods in general raises the possibility that carcinogens
may be produced by excessive heating of fat during frying
(1, 13). The distribution of type of fat used for frying was

Otherbeef

not significantly different in heavy and light usersof fried

Poultry
Eggs
Sweet rolls

foods, so it is likely that the observed associations with fried

foods and type of fat used for frying are independent.

Except for the foods listed in Table 7, all the others studied

Candy

in relation to colon cancer (Tables 5 and 6) were not

associated with breast cancer. Four of the 5 foods associated
with breast cancer are consistent with the hypothesis that fat
intake is related to breast cancer. However, in comparing
the data for colon and breast cancer, it seems that the
evidence implicating dietary factors in the etiology of breast
cancer is less convincing than that for colon cancer.

Jam or jelly
Sweets and desserts0
Green beans
Dried beans
Dried fruits
Tossed salad
High-fiber foods0
Low-fiber foods0
Routine use of butter or margarine on cooked vegetables
Usually add sugar to cereal
Type of fat used for frying

Discussion
a Do not meet

criteria

specified

for positive

(negative)

association

in

Table 5, Footnote d. Many showed positive association with one type of

control (hospital or general population) and negative or no association
with the other type of control.
0 See Table

5, Footnote

h.

Overall, the currently available evidence on cancer among

Seventh-Day Adventists is consistent with the hypothesis
that one or more components of the typical Adventist

Table 7
Relative risk for use of individual foods reported by Adventist breast cancer cases and Adventist controls
Seventy-seven breast cancer cases and controls were selected, as described in Table 5, legend.
)DichotomyHigh
Relative risk'sRaw

data (case+,case,cont+,cont

vs.(X
vs.Type
extremecFried
offood<X)0extremec(X

vs.lowDichotomyHigh

medium1.61.8(56,21,l03,63)e(33,21,54,63)vs.
foods(heavy +
light)Fried
7,20,28,79)Hard
potatoes (any vs. none)2.6'2.4t(56,20,85,79)(
vs.2.0(12,63,14,148)oil)Dairy
fat for frying (hard fat

vs <X)@low

milkd1.61.8(56,21,103,63)(33,21,54,63)(heavy
products except
light)White+ medium vs.
time)1.6(14,62,20,144)(yes
bread (most
ofthe
vs. no)
a See

Footnote

a.

0 See Table

Table

5, Footnote

5,

d.

C See

c.

Table

5, Footnote

d See Table

5, Footnote

h.

e See Table

5, Footnote

g.

p 0.01 by x2 corrected for continuity.

@

3520

@5
by

x2

corrected

for

continuity.

CANCER RESEARCH VOL. 35

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

Cancer in Seventh-Day Adventists

life-style

account

for a large portion

of their

apparent

reduced risk of the types of cancer which are unrelated to

cigarette smoking and alcohol consumption. Aside from
abstinence from smoking and drinking, the most distinctive
feature of the typical Adventist life-style is a unique diet
whose principal feature is lacto-ovo-vegetarianism. Har
dinge et al. (21, 22) and Sanchez et al. (44, 45) have shown
that the typical lacto-ovo-vegetarian
diet has about 25%
less fat and 50% more fiber than the average nonvege
tarian diet, and the proportion of saturated fat ratio is
just about doubled.
If the relative effect of selective factors is minimal, the
prospective mortality data for Adventists are clearly in
agreement with the hypotheses that either fat or fiber intake
is related to many gastrointestinal cancers, as well as breast,
endometrial, and ovarian cancer. The case-control data

would tend to favor the fat hypothesisfor colon cancer,and

the comparison of LLU to USC physiciansdoesnot clearly
refute or confirm either hypothesis. It seems clear that
additional studies on diet and cancer in the Adventist
population will be productive.
Several aspects of the typical Adventist diet might affect
their risk of cancer. The relatively low content of fat
(especially saturated fat) and cholesterol and the relatively
high content of fiber could influence other physiological or
metabolic phenomena which may relate to cancer risk. Such
phenomena could include rapid bowel transit time (4, 15)
and lower output of bile acids potentially convertible to
carcinogens (43, 25), as well as altered type and metabolic
activity of intestinal bacteria (2, 24, 26, 42).

The relatively low intake of protein (23) and the possible

lower frequency of obesity (8) in Seventh-Day Adventist
vegetarians, which suggests a lower caloric intake, may well
delay the onset of menarche (14, 31) and also influence
hormone status at other periods of life (1 1, 12, 36, 37). If
such effects were documented, they could explain all or part
of the reduced risk of Adventists for breast, ovary, and
endometrial cancer. The lack of coffee consumption by
most Seventh-Day Adventists could account for a good

share of their reducedbladder cancer risk (6, 7).

Primarily on the basis of animal experiments, one could
speculate that the relatively low intake of protein and fat by
members of this religious group may favorably alter the

body's responseto chemical carcinogens(5, 40). It is also

conceivable that exposure to potentially carcinogenic food
additives or contaminants among Adventists may be quite
different from that of the general public. Furthermore, their
response to potential carcinogens such as nitrosamines,

aflatoxin, polycyclic hydrocarbons, etc., might be less

detrimental because of a relatively high intake of vitamin C
and vitamin A (23), both of which are potentially protective
against certain chemical carcinogens (39, 41). Although
relatively little is known regarding the environmental
influences on the human microsomal hydroxylation enzyme
system (38, 40, 50), it is worth noting that several fruits and
vegetables that are abundantly used by Adventists contain
compounds (such as flavones) that are potent inducers of
this enzyme system which is intimately involved in the
detoxification of absorbed carcinogens (40). The lower use

of meats also reduces the use of backyard charcoal broilers

which are capable of producing considerable benzo(a)
pyrene content in broiled meat (34, 35).
Although the evidence is quite scanty (3), it seems
reasonable
to assume that the Adventists'
diet might
influence the functioning ofthe immunological system. Both
humoral and cellular immunity are involved in the body's
defense against cancer, and it is quite conceivable that a

very low intake of foreign animal protein could influence the

ability of the immunological surveillance system to recog
nize and destroy small, early clones of tumor cells. This
possibility is particularly attractive because of the rather
general decrease in cancer mortality from almost all cancer
sites in Adventists, which is more suggestive of a stronger
defense system against cancer than lack ofexposure to a few
of the known multiple environmental carcinogens.
Our current prospective study will not be able to elucidate
many of these potential mechanisms but, hopefully, our
findings will help to establish priorities for which hypotheses
would be most productive for further detailed epidemiologi
cal, laboratory, or clinical investigation.
References
I . Arffmann, E. Heated Fats and Allied Compounds as Carcinogens: A
Critical Review of Experimental Results. J. NatI. Cancer Inst., 25:
893-926, 1960.
2. Aries, V. C., Crowther, J. S., Drasar, B. S., Hill, M. J., and Ellis, F. R.
The Effect of a Strict Vegetarian Diet on the Faecal Flora and Faecal
Steroid Concentration. J. Pathol., 103: 5456,1971.
3. Bull, D. M. Nutrition and Tumor Immunity: Divergent Effects of
Antitumor Antibody. Cancer Res., 35: 3317-3319, 1975.
4. Burkitt, D. P. Epidemiology of Cancer of the Colon and Rectum.
Cancer, 28: 3-13, 1971.
5. Clayson, D. B. Nutrition and Epxerimental Carcinogenesis: A Review.
Cancer Res., 35: 3292-3300, 1975.
6. Cole, P. Coffee-Drinking and Cancer of the Lower Urinary Tract.

Lancet1:13351337,
1971.
7. Cole, P. A Population-Based Study of Bladder Cancer. In: R. Doll and
I. Vodopija (eds.), Host Environment Interactions in the Etiology of
Cancer in Man, pp. 83-87. Lyon, France: International Agency for
Research on Cancer, 1973.
8. Darnell, C. Obesity and Vegetarianism in Seventh-day Adventist
Women. Master's Thesis, Loma Linda University, Loma Linda,
Calif., 1975.
9. deiong, U. W., Day, N. E., Muir, C. S., Barclay, T. H. C., Bras, G.,
Foster, F. H., Jussawalla, D. J., Kurihara, M., Linden, G., Martinez,
I., Payne, P. M., Pedersen, E., Ringertz, N., and Shanmugaratnam, T.

TheDistributionof CancerwithintheLargeBowel.Intern.J. Cancer,

10: 463477,1972.
10. de Waard, F. Nurture and Nature in Cancer of the Breast and the
Endometrium. In: R. Doll and I. Vodopija (eds.), Host Environment
Interactions in the Etiology of Cancer in Man, pp. 121- 130. Lyon,
France: International Agency for Research on Cancer, 1973.
11. de Waard, F. Breast Cancer Incidence and Nutritional Status with
Particular Reference to Body Weight and Height. Cancer Res., 35:
33513356,1975.
12. Dickinson, L. E., MacMahon, B., Cole, P., and Brown, J. B. Estrogen
Profiles of Oriental and Caucasian Women in Hawaii. New EngI. J.

Med.,291:1211-1213,
1974.
13. Fieser, L., and Fieser, M. OrganicChemistry,

pp. 814-816. Boston: D.

C. HeathandCompany,1944.
14. Frisch, R. E. Critical Weight at Menarche, Initiation ofthe Adolescent

NOVEMBER 1975
Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.

3521

R. L. Phillips
Growth Spurt, and Control of Puberty. In: M. M. Grumbach, G. D.
Grave, and F. E. Mayer (eds.), Control of the Onset of Puberty, pp.
403-423. New York: John Wiley & Sons, 1974.
15. Glober, G. A., Klein, K. L., Moore, J. 0., and Abba, B. C. Bowel
Transit-Times
in Two Populations Experiencing Similar Colon
Cancer Risks. Lancet, 2: 80-81, 1974.
16. Haenszel, W., Berg, J. W., Segi, M., Kurihara, M., and Locke, F. B.
Large Bowel Cancer in Hawaiian Japanese. J. NatI. Cancer Inst., 51:
1765-1779,1973.
17. Haenszel, W., and Correra, P. Cancer of the Large Intestine:
Epidemiologic Findings. Diseases Colon Rectum, 16: 371 -377, 1973.
18. Haenszel, W., Correa, P., and Cuello, C. Social Class Differences in
Large Bowel Cancer in Cali, Colombia. J. NatI. Cancer Inst., 54:
1031-1035,1975.
19. Haenszel, W., and Kurihara, M. Studies of Japanese Migrants. I.
Mortality from Cancer and Other Diseases among Japanese in the
United States. J. NatI. Cancer Inst., 40: 43@68, 1968.
20. Hammond, E. C. Smoking in Relation to the Death Rates of One
Million Men and Women. NatI. Cancer Inst. Monograph, 19: 127204, 1966.
21. Hardinge, M. G., Chambers, A. C., Crooks, H., and Stare, F. J.
Nutritional Studies of Vegetarians. Ill. Dietary Levels of Fiber. Am.
J. Clin. Nutr., 6: 523-525. 1958.
22. Hardinge, M. G., Crooks, H., and Stare, F. J. Nutritional Studies of
Vegetarians. IV. Dietary Fatty Acids and Serum Cholesterol Levels,
Am. J. Clin. Nutr., 10: 516-524, 1962.
23. Hardinge, M. G., and Stare, F. J. Nutritional Studies of Vegetarians.
I. Nutritional, Physical, and Laboratory Studies. J. Clin. Nutr., 2:
73-82, 1954.
24. Hill, M. J. Colon Cancer: A Disease of Fibre Depletion or of Dietary
Excess? Digestion, 11: 289-306, 1974.
25. Hill, M. J., Drasar, B. S., Aries, V. C., Crowther, J. S., Hawksworth,

33. Lemon, F. R., Walden, R. T., and Woods, R. W. Cancer ofthe Lung
and Mouth in Seventh-day Adventists. Cancer, 17: 486-497, 1964.
34. Lijinsky, W., and Ross, A. E. Production ofCarcinogenic Polynuclear
Hydrocarbons in the Cooking of Food. Food Cosmet. Toxicol., 5:
343-347, 1967.
35. Lijinsky, W., and Shubik, P. Benzo(a)pyrene and Other Polynuclear
Hydrocarbons in Charcoal-Broiled Meat. Science, 145: 53-55, 1964.
36. Lipsett, M. B. Hormones, Nutrition, and Cancer. Cancer Res., 35:

33593361,
1975.
37. MacMahon, B., Cole, P., and Brown, J. Etiology of Human Breast
Cancer: A Review. J. Nail. Cancer Inst., 50: 21-42, 1973.
38. Magee, P. N. Metabolic Aspects of Chemical Carcinogenesis. In: R.
Doll and I. Vodopija (eds.), Host Environment Interactions in the

Etiology of Cancerin Man, pp. 213-221. Lyon, France:International

39.
40.

41.

42.
43.

44.
45.

G., andWilliams,R. E. 0. Bacteriaandthe Aetiologyof Cancerof

Large Bowel. Lancet, 1. 95-100, 1971.
26. Hill, M. J., Goddard, P. and Williams, R. E. 0. Gut Bacteria and
Aetiology of Cancer of the Breast. Lancet, 2: 472-473, 1971.
27. Hirayama, T. Epidemiology of Cancer of the Stomach with Special
Reference to Its Recent Decrease in Japan. Cancer Res., 35:
3460-3463,1975.
28. Howell, M. A. Diet as an Etiological Factor in the Development of
Cancers of the Colon and Rectum. J. Chronic Diseases 28: 67-80,
1975.
29. Kessler, I. I. Perspectives on the Epidemiology ofCervical Cancer with
Special Reference to the Herpesvirus Hypothesis. Cancer Res., 34:
1091-1110,1974.
30. Kitagawa, E., and Hauser, P. Differential Mortality in the United
States: A study in Socioeconomic Epidemiology, pp. 73-92. Cam
bridge, Mass.: Harvard University Press, 1973.
3 1. Kralj-Cercek, L. The Influence of Food, Body Build, and Social Origin
on the Age at Menarche. Human Biol., 28: 393-406, 1956.
32. Lemon, F. R., and Walden, R. T. Death from Respiratory System
Disease among Seventh-day Adventist Men. J. Am. Med. Assoc., 198:
117-126,1966.

3522

46.
47.

48.
49.

50.
51.

52.

Agency for Research on Cancer, 1973.

Maugh, T. H., 11. Vitamin A: Potential Protection from Carcinogens.
Science, 186: 1198, 1974.
McLean, A. E. M. Diet and the Chemical Environment as Modifiers
of Carcinogenesis. In: R. Doll and I. Vodopija (eds.), Host Environ
ment Interactions in the Etiology of Cancer in Man, pp. 223-230.
Lyon, France: International Agency for Research on Cancer, 1973.
Mirvish, S. S., Wallcave, L., Eagen, M., and Shubik, P. Ascorbate
Nitrite Reaction: Possible Means of Blocking the Formation of
Carcinogenic N-Nitroso Compounds. Science, 177: 65, 1972.
Reddy, B. S., Weisburger, J. H., and Wynder, E. L. Fecal Bacterial
@-Glucoronidase:Control by Diet. Science, 183: 416-417, 1974.
Reddy, B. S., and Wynder, E. L. Large Bowel Carcinogenesis: Fecal
Constituents of Populations with Diverse Incidence Rates of Colon
Cancer. J. NatI. Cancer Inst., 50: 1437-1442, 1973.
Sanchez, A., Porter, G. G., and Register, U. D. Effect of Entree on Fat
and Protein Quality of Diets. J. Am. Dietet. Assoc., 49: 492-496, 1966.
Sanchez, A., Scharffenberg, J. A., and Register, U. D. Nutritive Value
of Selected Proteins and Protein Combinations. I. The Biological
Value of Proteins Singly and in Meal Patterns with Varying Fat
Composition. Am. J. Clin. Nutr., 13: 243-249, 1963.
Schottenfeld, D. Cancer Epidemiology and Prevention. Current Con
cepts. Springfield, Ill.: Charles C Thomas, Publisher, 1975.
Seidman, H. Cancer Death Rates by Site and Sex for Religious and
Socioeconomic Groups in New York City. Environ. Res., 3: 234-250,
1970.
Seltzer, C. C., and Jablon, S. Effects ofSelection on Mortality. Am. J.
Epidemiol., 100: 367-372, 1974.
Staszewski, J., and Haenszel, W. Cancer Mortality
Among the
Polish-Born in the United States. J. NatI. Cancer Inst., 35: 291-297,
1965.
Wattenberg, L. W. Effects of Dietary Constituents on the Metabolism
ofChemical Carcinogens. Cancer Res., 35: 33263331,1975.
Worth, R. M., and Kagan, A. Ascertainment of Men of Japanese
Ancestry in Hawaii through World War II Selective Service Registra
tion. J. Chronic Diseases, 23: 389-397, 1970.
Wynder, E. L., Lemon, F. R., and Bross, I. J. Cancer and Coronary
Artery
Disease among Seventh-day Adventists.
Cancer, 12:

1016-1028,
1959.

CANCER RESEARCH VOL. 35

Downloaded from cancerres.aacrjournals.org on July 3, 2013. 1975 American Association for Cancer Research.