The diets of middle-aged men in two rural areas of Greece


Keys, A.; Aravanis, C.; Sdrin, H.

Voeding 27(11): 575-586

1966


Seven-day dietary surveys were made on random samples of middle-aged men in villages on the island of Crete (3 surveys) and on the island of Corfu (2 surveys) in Greece. All foods consumed were weighed and replicates of the meals were collected for chemical analy sis. In both areas the diets were dominated by olive oil and bread, with low intakes of foods of animal origin and of sugar. Calculations using tables of food composition gave average results in fair agreement with those from direct chemical analysis but the chemical analyses indicated lower values than the tables for calories, proteins and especially for fats. In both areas the intake of saturated and of poly-unsaturated fatty acids was low. There were no significant seasonal differences in the diets in regard to calories or percentage of calories from proteins or from fats. The diets of the men in these 2 areas of Greece did not differ significantly in cal. or protein but the diet in Crete provided a higher proportion of fat cal. (40.3 by calculation, 36.1 by analysis) than in Corfu (32.7 and 27.2, respectively). Repeated surveys on the same men showed intra-individual variability to be a large part of the total variability in regard to total cal., alcohol cal., and percentages of cal. from proteins and from fats. The individual values from a single survey are of very limited use in predicting individual values in a second survey.

The
Laboratory
of
Physiological
Hygiene,
School
of
Public
Health,
University
of
Minnesota,
Minneapolis
THE
DIETS
OF
MIDDLE-AGED
MEN
IN
TWO
RURAL
AREAS
OF
GREECE
*
by
A.
KEYS'),
C.
ARAVANIS
ond
H.
SDRIN
3
)
Surveys
of
the
habitual
diet
are
an
integral
part
of
a
cooperative
international
study
on
the
epidemiology
of
heart
disease
in
men
oged
40-59
when
first
examined
(6).
In
Greece,
one
sample
embraces
the
men
in
villages
on
the
islond
of
Crete
centered
about
30
kilometers
eost
of
the
port
city
of
Iraklion.
Another
sample
in
Greece
in
made
up
of
men
living
in
villoges
northeost
of
the
town
of
Corfu
on
the
island
of
the
same
name.
These
areas
have
been
described
and
some
charac-
teristics
of
the
men
have
been
reported
elsewhere
(6).
The
present
paper
reports
the
results
from
3
dietary
surveys
in
Crete
ond
2
in
Corfu.
In
Crete
the
villages
involved
were
Agies
Paraskies,
Thrapsano,
and
Kastelli.
In
Corfu
the
villages
were
Ano
Korakiana,
Skriperon,
and
San
Morco.
Subjects
and
field
operations
The
medical
studies
cover
proctically
all
(over
95
per
cent)
of
the
men
in
the
defined
areos
aged
40-59
at
the
start
of
the
program
in
1960
in
Crete
and
in
1961
in
Corfu.
From
these
cohorts
random
sub-samples
of
men
were
selected
for
detailed
studies
of
the
diet.
A
few
men
selected
at
random
for
this
work
were
unwilling
or
unable
to
cooperate
because
of
special
circumstances
in
their
homes
and
these
men
were
replaced
by
further
rondom
selections
from
the
cohort.
The
selections
of
the
subjects
ond
the
general
organization
of
the
dietary
studies
were
similar
to
the
parallel
programs
in
other
countries
(7,
2,
3,
4,
5).
Each
survey
involved
recording
the
weights
of
all
foods
eaten
by
the
individuol
subjects
during
7
successive
days
and
a
quontitative
replicate
of
each
meal
wos
collected
for
subsequent
chemical
analysis
of
the
composite
7-day
diets.
The
amounts
of
alcoholic
beverages
were
also
recorded
but
these
moterials
were
not
included
in
the
composites
for
chemical
analysis.
A
team
of
dietitians
from
Athens,
with
local
help,
carried
out
or
supervised
all
measurements
and
food
collections
in
the
hauseholds.
In
general,
each
dietitian
worked
in
only
two
neighboring
households
in
the
same
period
so
close
contoct
was
maintained
at
all
times.
In
several
of
the
surveys
this
work
was
checked
in
the
field
by
dietitians
engaged
in
similar
work
in
this
program
in
other
countries.
Covered
enamel
jars
were
provided
to
each
household
for
each
Boy's
collection
of
the
replicate
of
food
eoten
by
the
mon
in
the
study.
These
composites
were
collected
in
the
evenings,
stored
overnight
in
a
refrigerator,
and
in
the
morning,
offer
the
composite
was
finely
ground
and
well
mixed,
a
10
per
cent
sample
was
transferred
to
a
glass
jor
and
stored
in
a
low-temperature
refrigerator.
At
the
The
work
reported
here
was
aided
by
research
grants
from
the
U.S.
Public
Health
Service
(No.
HE-04697),
the
American
Heart
Association,
New
York,
and
the
Olive
Advisory
Board,
San
Francisco,
California.
I)
Stadium
Gate
27,
University
of
Minnesota,
Minneapolis,
Minnesota
55455.
2)
Greek
Society
for
the
Study
of
Atherosclerosis,
Athens,
Greece.
3)
Characopeios
School
of
Home
Economics,
Athens,
Greece.
57
end
of
the
7
days
af
food
collection
the
combined
samples
were
well
mixed
with
a
heavy-duty
electric
mixer
and
240
grams,
os
a
7-day
sample,
was
flown
to
Athens
where
it
was
again
mixed
and
sub-samples
were
lyophilized
at
the
Department
af
Chemistry
af
the
University
af
Athens.
Aliquots
of
these
lyophilized
samples
were
sealed
in
small
glass
cantainers
and
shipped
to
the
University
of
Minnesota
for
chemical
analysis.
Chemical
analyses—proximate
analysis
Weighed
aliquots
of
the
lyophilized
homogenates
of
the
7-day
diets
were
analysed
for
water
by
drying
to
constant
weight
in
a
vacuum
oven,
for
nitrogen,
using
the
Kjeldahl
method,
for
lipids
(„fat")
by
Soxhlet
ether
extraction,
and
for
osh
by
incineration
in
a
muffle
furnace.
Protein
was
estimated
as
6.25
times
nitrogen
weight.
Carbohydrote
was
estimated
by
subtrocting
from
the
weight
of
lyophilized
hamagenate
the
weights
of
water,
protein,
fats,
and
ash.
Calories
were
computed
from
the
weights,
in
grams,
os
4
times
protein,
9
times
fat,
and
4
times
carbohydrate.
Total
calories
were
estimated
as
the
sum
of
protein,
fat
and
carbohydrate
calories
plus
7
times
grams
of
alcohol
the
weight
of
alcohol
being
computed
from
the
volumes
of
alcoholic
beverages
consumed
and
tables
of
alcohol
concentrotion
in
the
various
beverages.
The
diet
samples
were
analysed
in
the
United
States
but
in
some
cases
duplicates
were
also
onalysed
in
Athens
(Prof.
D.
S.
GALANOS)
and
in
Naples,
Italy
(Prof.
F.
FIDANZA).
Toble
1
summarizes
the
results
from
31
sets
of
duplicates
Table
1.
Comparison
of
the
results
of
chemical
analyses
of
31
sets
of
aliquots
of
diet
composites
in
Athens
and
in
the
U.S.A.
Item
%
of
total
calories
from:
Proteins
Fats
I.
Mean,
Athens
10.06
34.63
2.
Mean
difference,
Athens
minus
U.S.A.
0.75
0.43
3.
Standard
error
of
2)
-
1
-
0.256
±
0.739
4.
100
x
2)
/
I)
7.46
1.24
5.
t
=
2)
/
3)
2.93
0.58
6.
p
<
0.01
n.s.
onolysed
in
Europe
as
well
as
in
the
United
States.
The
average
calories
provided
by
proteins
is
indicated
to
be
7.46
per
cent
lower
from
the
European
than
from
the
U.S.
analyses
and
the
difference
is
statistically
significant
(p
=
less
thon
0.01).
For
fats,
the
European
results
are
also
lower
than
those
obtained
in
the
U.S.
but
the
difference,
1.24
per
cent,
is
not
statisticolly
significant.
In
the
subsequent
presentation
of
the
analytical
data
in
this
paper
only
the
U.S.
volues
are
used.
Chemical
analyses—fatty
acids
The
lipid
(ether)
extrocts
of
the
lyophilized
homogenates
were
methylated
by
treatment
with
methanol
and
sulfuric
acid
as
described
by
ANDERSON
(1).
The
resulting
methyl
esters
were
anolysed
by
gas-liquid
chromatography
(GLC)
in
three
different
systems,
the
carrier
gos
being
helium
in
all
cases.
58
Method
1
used
a
Beckman
GC-2
GLC
machine
with
a
column
150
cm.
long
packed
with
30
per
con!
butanediol
succinate
(Craig)
polyester
on
acid-washed
Celite,
60-130
mesh,
operated
at
220-230
°
C.
Method
2
used
an
F
and
M
model
810
GLC
machine
with
flame
ionization
detector,
the
column
being
a
6-foot
length
of
1/4
inch
(internal)
diameter
stainless
steel
tube
packed
with
10
per
cent
ethylene
glycol
succinate
on
Deotoport
5,
80-100
mesh.
The
column
temperature
was
200
°
C.,
detector
300
°
C.,
and
injection
port
280
°
C.
Method
3,
at
Athens,
used
an
F
and
M
model
720
GLC
machine,
the
column
being
6
feet
in
length,
1
/
2
inch
outside
diameter,
packed
with
15
per
cent
ethylene
glycol
succinate
on
Chromisorb
S.
80-100
mesh.
The
column
temperature
was
188
°
C.,
detector
315
°
C.,
injection
port
300
°
C.
Table
2
summarizes
results
from
duplicate
GLC
analyses.
In
general,
the
standard
error
of
measurement,
S.E.M.,
was
less
thon
one
per
cent
of
the
total
fatty
acid
except
for
oleic
acid
(18:1),
and
linoleic
acid
(18:2).
When
Method
1
was
compared
with
Method
2
the
mean
values
for
the
several
fatty
acids
were
not
significantly
different
except
for
oleic
and
linoleic
acids.
Compared
with
Method
1,
Method
2
gave
lower
values
for
oleic
and
higher
values
for
linoleic
acid.
On
the
other
hand,
Method
3
results
(Athens),
are
higher
in
oleic
and
lower
in
linoleic
Table
2.
Duplicate
analyses
for
fatty
acids
os
percentage
of
total
fatty
acids
in
the
lipid
extrocts
of
the
lyophilized
diet
homogenates
Averoge
differences
Method
1
minus
Method
2,
and
Method
1
minus
Athens,
and
the
standard
error
of
meosurement
far
pairs
of
duplicates,
(5.E.M.)
2
(I,A
2
)/
2N,
where
.4
=
difference
between
duplicates
and
N
=
number
of
pairs
of
duplicates.
Fatty
Acid
Item
12:0
14:0
16:0
18:0
16:1
18:1
18.2
16:3
Method
1
-
method
2
0
0
0.8
0.4
0.4
3.5
±0.9
±0.6
±2.6
±0.7
Method
1
-
Athens
0.1
—0.4
—0.1
—0.4
—0.4
—3.9
1.8
0.7
S.E.M.
±0.2
±0.4
±0.3
±1.4
±0.9
acid
than
those
from
Method
1.
Pure
fatty
acid
standards
were
used
in
all
3
methods.
It
was
not
possible
to
make
sure
which
system
provided
the
most
accurate
results
but
there
was
indirect
evidence
that
Method
3
underestimated
the
linoleic
acid
content
of
the
diet.
Since
Method
1
and
Method
3
results
were
available
only
on
some
of
the
samples,
in
general
the
subsequent
presentation
cf
fatty
acid
data
relies
on
Method
2
data.
General
character
of
the
diets
As
indicated
in
table
3,
both
in
Crete
and
Corfu
the
diets
of
the
rural
men
are
dominated
by
olive
oil
and
bread,
these
two
items
alone
accounting
for
50
to
over
60
per
cent
of
the
total
calories.
Olives
os
such,
chiefly
the
black
kind,
are
eaten
only
in
small
amounts,
the
daily
average
being
of
the
arder
of
5
grams
(edible
portion).
Besides
the
use
in
bread,
cereal
grains
contribute
substantially
to
the
diet
in
the
form
of
macaroni,
rice,
and
barley
soup
or
porridge.
Alcohol
makes
an
appreciable
part
of
the
total
calories
for
most
of
the
men
but
around
one-fifth
of
the
men
are
abstainers.
Almost
all
of
the
alcohol
is
taken
in
the
form
of
light
wine,
averaging
about
12
per
cent
alcohol
by
volume.
Very
small
amounts
of
a
kind
of
brandy
(„raki")
and
liquors
(„ouzo"
etc.)
are
can-
59
Table
3.
Sources
of
calories
in
the
diets
of
the
men
in
Crete
and
Corfu
Calories
estimated
from
tables
of
food
composition
applied
to
the
weighed
quantities
of
all
foods
and
alcoholic
beverages
consumed
by
each
individual
in
7
consecutive
days.
Means
and,
in
parentheses,
standard
deviations.
Item
Period
1960
Sept.
Crete
1962
1965
May-June
Feb.
Cork
1961
1963
Sept.
Mar:
April
Number
of
men
32
34
30
40
34
%
Cal.,
bread
28.9
28.1
22.6
38.1
40.8
(7.1)
(6.6)
(7.9)
(7.0)
(6.8)
%
Cal.,
olive
oil
32.6
33.0
28.5
25.7
22.1
(6.8)
(6.5)
(7.2)
(6.0)
(5.5)
%
Cal.,
animal
protein
2.6
2.8
4.0
3.3
3.2
(1.6)
(2.8) (1.6)
(2.1)
(1.7)
%
Cal.,
animal
fat
5.5
5.8
6.7
3.9
3.9
(3.8)
(2.8)
(3.3) (3.2) (2.3)
V.
Cal.,
alcohol
2.1
5.2
5.7
8.4
7.2
(2.7)
(6.8)
(5.5) (5.7) (5.2)
All
other
sources
28.3
25.1
32.5
20.6
22.8
sumed.
The
use
of
beer
is
increasing
in
these
areas
but
the
overage
intake
is
still
trivial.
Animal
products
generally
contribute
less
than
10
per
cent
of
the
total
calories.
On
the
average,
only
3
to
4
per
cent
of
the
total
dietary
calories
are
provided
by
animal
proteins.
In
Crete
the
meat
is
mostly
goat,
beef
and
mutton,
with
an
occasionol
chicken
or
rabbit.
In
Corfu,
the
meat
is
mostly
beef
and
veal.
In
both
areos
fish
is
eaten,
chiefly
sardines
and
salt
codfish,
the
latter
especially
in
the
winter.
In
the
several
surveys
in
Crete
the
average
wos
from
15
to
23
grams
of
fish
a
doy.
In
Corfu
the
average
consumption
of
fish
was
more
than
twice
os
greot,
53
to
65
grams
a
day.
Milk
as
such
is
little
used
by
the
men
in
these
areas
and
is
generolly
considered
os
o
food
only
for
children
and
sick
people.
Cheese,
mostly
of
the
soft,
fresh
voriety,
is
eaten
regularly,
the
overoges
for
the
several
surveys
running
from
8
to
20
grams
per
day.
In
oddition,
yogurt
is
popular
but
the
total
consumption
is
small.
Butter
is
rarely
used
and
then
only
in
trifling
omounts.
Margarine
is
also
seldom
used.
Instead
of
using
o
spread,
these
Greeks
prefer
to
dip
bread
in
olive
oil,
though
most
of
the
bread
in
eoten
plain.
Eggs
are
eaten
is
small
quantities
in
both
Crete
and
Corfu.
In
the
Crete
survey
of
1962
the
average
consumption
was
3
eggs
o
week.
In
Corfu
in
1961
a
totol
of
26
eggs
were
eoten
by
40
men
in
7
days.
Potatoes
are
eoten
in
forge
amounts
both
in
Crete
and
Corfu,
especially
in
the
foil
and
winter.
In
Corfu,
sweet
potatoes
as
well
as
white
potatoes
ore
populor.
In
the
severol
surveys
the
average
per
capita
daily
consumption
of
potatoes
ronged
from
100
to
197
grams.
Pulses
of
various
kinds,
also
form
an
important
port
of
the
diet,
especiolly
in
the
winter,
the
average
being
about
30
grams
daily.
Broad
beons
(Vicia
fava)
ore
most
common
but
dried
peas,
lentils,
ordinary
beans
(Phaseolus
vulgaris),
and
chick
peos
(Cicer
arietinum)
are
also
widely
used.
60
Nuts,
moinly
almonds
and
filberts,
are
popular
throughout
the
year
but
the
total
consumption
is
small,
6
or
8
nuts
being
eaten
at
a
time,
often
as
a
snack,
especially
when
there
are
visitors
in
the
household.
Raisins
ond
dried
figs
are
eaten
in
the
same
way,
commonly
with
nuts.
Sugar,
as
such,
is
used
in
very
small
amounts,
the
average
daily
per
capita
consumption
in
the
several
surveys
being
only
8
to
17
grams.
Honey
is
popular
and
is
frequently
used
on
special
occasions.
In
the
summer
an
occasional
small
dish
of
ice
cream
ond
a
sweetened
soft
drink
add
a
few
grams
of
sugar.
Both
in
Crete
and
Corfu
gorlic
ond
onions
are
generously
used
throughout
the
yeor.
Leaks
and
scallions
also
contribute
flavor
in
the
season.
The
average
intake
of
salt
(NaCI)
is
of
the
order
of
5
to
10
grams
daily.
The
similorities
between
the
diets
in
Crete
ond
those
in
Corfu
are
striking
but
there
are
a
few
differences.
The
Cretans
use
more
olive
oil
and
goat
milk;
the
men
of
Corfu
more
fish
and
macaroni.
The
bread
in
both
areos
is
mostly
rather
dark;
that
in
Crete
often
is
made
of
whole
wheat
and
barley
;
in
Corfu
85
to
95
per
cent
extraction
wheat
flour
is
more
commit.
Calories,
proteins
and
fats-calculations
versus
analyses
Table
4
gives
means,
and
standard
errors
of
those
means,
for
totol
calories
per
day
and
percentages
of
calories
from
proteins
and
from
fats.
The
averages
for
calories
from
calculotions
and
from
chemical
analysis
are
not
significantly
Table
4.
Men
in
Greece,
7-day
survey
data,
means
and
standard
errors
Area
Date
No.
Men
Cal./Day
Tables
Chem.
%
Protein
Col.
Tables
Chem.
%
Fat
Cal.
Tables
Chem.
Crete
Sept.
1960
30
2769
2654
10.2
10.6
41.8
35.7
±78
±91
±0.2
±0.4
+1.2
±1.2
Crete
May-June
1962
33
2848
2781
9.8
9.2
41.8
37.9
+97
±97
+0.3
±0.3
±1.0
±1.1
Crete
Feb.
1965
30
2626
2566
11.6
9.6
37.4
34.6
±98
±82
+0.4
±0.3
±0.9 ±0.9
Corfu
Sept.
1961
40
2796
2632
11.2
10.3
34.2
26.2
±89
±90
+0.4 +0.4
±0.9 ±0.9
Corfu
Mar.-April
1963
34
2877
2712
11.5
10.0
31.2
28.2
±62
±116
±0.2 ±0.2
±0.9
±0.9
different
in
ony
one
of
the
5
surveys
but
in
every
survey
the
average
calculated
value
is
somewhat
higher
than
the
chemical
value,
the
average
difference
for
5
surveys
being
114
Cal.
per
day
or
4.1
per
cent
of
the
average
calculated
value
and
for
all
surveys
combined
the
difference
is
statistically
significant.
The
agreement
in
regard
to
protein
is
less
satisfactory
and
in
three
out
of
five
surveys
the
calculated
values
for
percentages
of
calories
from
proteins
are
significantly
higher
than
the
values
from
chemical
analysis.
On
the
average,
the
calculations
indicate
5.1
per
cent
more
protein
calories
than
recorded
by
chemical
analysis.
For
fats,
the
calculated
values
are
consistently
higher
than
the
analyticol
values
and
this
difference
accounts
for
the
discrepancy
in
calories
noted
above.
For
all
61
5
surveys
combined,
the
calculations
indicate
an
average
of
37.3
per
cent
of
calories
from
fats
while
the
corresponding
analytical
average
is
only
32.5
per
cent.
Two
possible
explanations
for
this
discrepancy
con
be
entertained.
Con-
ceivably
the
food
composition
tables
used
may
not
allow
for
the
fact
that
the
meats
used
in
Crete
and
Corfu
are
very
lean.
However,
as
will
be
seen,
at
most
this
would
account
for
only
a
small
part
of
the
discrepancy
because
the
amount
of
meat
is
so
small
in
these
diets
and
olive
oil
accounts
for
much
the
largest
amount
of
fat
in
these
diets.
It
seems
more
likely
that
the
fat
values
from
chemical
analysis
are
erroneously
low.
Besides
the
passibility
that
the
ether
extraction
of
the
lyophilized
material
was
incomplete,
nate
should
be
made
of
the
problem
of
obtaining
true
aliquots
from
the
composites
before
lyophilization.
Loss
of
fat
an
the
walls
of
the
containers
is
a
distinct
possibility.
Area
and
season
comparisons
The
data
in
table
4
do
not
indicate
any
significant
differences
between
Crete
and
Corfu
in
regard
to
total
daily
calories
or
percentage
of
calories
from
proteins.
But
the
proportion
of
fat
in
the
diet
is
distinctly
higher
in
Crete
than
in
Corfu,
the
general
averages,
in
terms
of
percentage
of
tatal
calories,
being
40.3
for
Crete
and
32.7
far
Corfu.
These
values
are
from
calculation
;
the
chemical
analyses
indicate
a
similar
difference.
The
surveys
covered
three
seasons
in
Crete,
early
summer,
late
summer,
and
winter,
while
in
Corfu
the
surveys
covered
spring
and
late
summer.
In
regard
to
total
calories
and
percentages
of
calories
from
proteins
and
from
fats
the
data
indicate
no
significant
seasonal
differences.
It
should
be
observed,
however,
that
dietary
surveys
were
not
carried
out
during
the
periods
of
heaviest
farm
work
because
the
men,
and
their
wives,
were
too
busy
in
the
fields
at
those
times.
Fatty
acids
in
the
diet
As
would
be
expected
from
the
large
use
of
alive
oil,
oleic
acid
dominated
the
fatty
acids
in
the
diets
of
bath
Crete
and
Corfu.
In
Crete,
oleic
acid
represented
an
average
of
from
70.7
to
72.7
per
cent,
of
the
total
fatty
acids.
In
Corfu
the
overage
wos
67.7
per
cent.
Table
5
summarizes
the
principal
folly
acid
data,
expressed
as
per
cent
of
total
fatty
acids,
in
3
surveys
in
Crete
(N
=
30,
33,
and
30,
for
number
of
men)
and
in
one
in
Corfu
(N
=
34).
Data
from
Corfu
in
1961
are
not
included
because
it
was
evident
that
same
of
the
poly-unsaturated
fatty
acids
had
been
destroyed
in
processing.
In
general,
however,
the
Corfu
1961
data
seem
to
hove
been
similar
to
those
obtained
in
1963.
Oleic
acid
accounted
for
dose
to
70
per
cent
of
the
total
fatty
acids
in
all
cases
and
palmitic
was
the
next
mast
abundant
fatty
acid
in
the
diets.
The
third
most
important
fatty
acid,
linoleic,
may
hove
been
under-estimated
in
some
of
the
1962
Crete
samples.
On
the
average,
3
to
5
per
cent
of
the
total
fatty
acids
was
made
up
of
stearic
acid.
Myristic
acid,
not
tabulated
in
table
5,
and
palmitoleic
acid,
were
the
only
other
fatty
acids
that
amounted
to
as
much
as
1.0
per
cent
62
Table
5.
Fatty
acid
composition
of
the
ether
extracts
of
the
diets,
expressed
as
percentage
of
the
total
fatty
acids
Means
and,
in
parentheses,
standard
deviations.
Fatty
acid
Crete
1960
Crete
1962
Crete
1965
Corfu
1963
Short-chain
saturates
-
0.4
0.1
0.1
(10
or
fewer
carbons)
-
(0.4)
(0.1)
(0,1)
Palmitic
(16
:
0)
14.6
17.0
13.9
14.7
(3.3)
(3.8)
(2.0)
(2.6)
Stearic
(18
:
0)
5.2
5.1
4.1
3.2
(2.2)
(1.2)
(1.4)
(1.2)
Other
saturates
1.4
1.5
1.4
1.5
-
- -
-
Palmitoleic
(16
:
1)
1.0
1.2
1.3
1.6
(0.4) (0.4)
(0.4)
(0.5)
Oleic
(18
:
1)
69.6
69.7
71.3
65.9
(5.0) (5.2)
(4.2) (3.6)
Unpick
(18
;
2)
7.5
4.2
7.1
12.0
(1.5) (2.0)
(1.3) (2.4)
Linolenic
(18
r
3)
0.4
0.5
0.3
0.7
(0.3)
(0.9)
(0.3) (0.3)
of
the
total
fatty
acids.
Most
of
the
diets
showed
traces
of
caprylic,
capric
and
lauric
acids,
as
well
as
of
negligible
amounts
of
fatty
acids
with
odd
numbers
of
carbon
atoms
in
the
chain.
Table
6
summarizes
the
major
fatty
acid
data,
as
averages
for
all
surveys,
expressed
as
percentage
of
total
calaries
in
the
diets.
Both
in
Crete
and
Corfu
the
diets,
though
moderately
high
to
high
in
total
fats,
are
definitely
low
in
both
Table
6.
Average
percentage
of
calories
provided
by
specified
groups
of
fatty
acids
in
the
diets
of
the
men
of
Corfu
and
Crete
in
the
period
1960-65
Calculated
from
chemical
analyses.
Fatty
acids
Crete
Corfu
Saturates
with
fewer
than
18
carbon
atoms
6.0
4.5
Stearic
acid
1.7
0.9
Mona-eines
(palmitoleic
plus
oleic)
25.8
18.3
Linoleic
2.3
3.3
Total
poly-unsaturates
2.5
3.5
saturated
and
poly-unsaturated
fatty
acids.
Compared
with
typical
U.S.
diets,
as
well
os
those
in
Finland,
saturated
fatty
acids
are
of
the
order
of
only
one-third
as
high
in
these
Greek
diets.
Though
the
averages
for
linoleic
and
total
poly-
unsaturates
in
table
6
may
be
somewhat
under-estimated
for
Crete,
it
is
clear
that
these
Greek
diets
ore
lower
in
these
constituents
than
most
U.S.
diets.
Diet
of
the
man
versus
that
of
the
family
In
most
dietary
surveys
in
which
the
foods
are
actually
weighed
in
the
household,
the
family,
sometimes
excluding
infants,
is
the
unit.
Though
it
is
realized
that
the
individuals
differ
in
total
colories,
it
is
generally
implicitly
assumed
that
the
63
composition
of
the
diet
is
the
some
for
all
members
of
the
family,
or
at
least
for
all
adults.
The
question
of
the
distribution
of
the
nutrients
between
the
man
of
the
family
and
the
rest
of
the
family
was
Investigated
in
a
separate
survey
in
Crete
in
1957.
Table
7.
Percentage
of
total
calories
provided
from
specified
sources
in
the
dieh
of
middle-aged
men
and
of
their
families
in
vil!ages
of
central
Crete
in
1957
Data
from
records
of
food
weights
consumed
in
7
successive
days
in
October-November
Item
Man
Rest
of
family
Calories
frcrn
proteins
10.2
10.4
total
fats
32.9
38.6
olive
oil
26.6
31.6
The
method
was
essentially
the
some
as
that
used
in
the
loter
surveys
summarized
above
except
thot
records
of
food
weights
were
kept
separately
for
the
family
as
a
whole
and
for
the
man
of
the
family.
Food
volues
were
estimoted
by
the
use
of
tobles
of
food
composition.
Salient
results
are
summarized
in
table
7.
Proteins
provided
substantially
the
same
proportion
of
the
total
calories
for
the
man
and
the
other
members
of
his
family
but
there
was
a
significant
difference
in
regard
ta
fats,
the
men
having
a
diet
relatively
lower
in
fats
than
eaten
by
the
rest
of
the
family.
The
data
show
that
olive
oil
was
responsible
far
this
difference.
Compared
with
the
rest
of
the
family,
the
men
consumed
relatively
less
olive
oil
and
more
bread
and
potatoes.
This
result
is
in
contrast
with
the
findings
in
similar
surveys
in
Finland
and
in
Yugoslavia.
Both
in
East
and
in
West
Finland
the
diets
of
the
men,
compared
with
those
of
the
other
family
members,
provided
a
larger
proportion
of
total
calories
from
proteins
and
from
fats
(7).
For
both
areas
of
Finland
combined,
the
averages
were
12.4
and
37.4
per
cent
of
colories
from
proteins
and
from
fats,
respectively,
for
the
men,
the
correspanding
figures
for
the
other
members
of
the
families
were
11.7
and
33.9.
In
Yugoslavia,
however,
five
similar
surveys
in
Dalmatia
and
Slavonia
failed
to
show
any
significant
difference
between
the
heads
of
families
and
the
other
members
of
the
families
in
regard
to
proportion
of
calories
from
either
proteins
or
from
fats
(2).
Intro-
versus
inter-individual
variations
The
principal
purpose
of
these
dietary
surveys
was
ta
characterize,
as
groups,
the
middle-aged
men
in
the
populations
concerned.
At
the
same
time
it
was
interesting
to
examine
the
stability
of
the
positions
of
the
individuals
within
the
group
so
some
of
the
same
men
were
included
in
repeated
surveys.
The
question
at
issue
is
as
to
how
well
the
individual
men
are
distinguished
from
their
fellows
in
terms
of
their
dietary
characteristics.
64
C
L;
30—
a.
a-
25—
18
MEN
OF
CORFU
'4
CALORIES
FROM
FATS
IN
REPEATED
SURVEYS
SO-
45-
40-o.
C.,
35
C
0
O
0
0
20
I
1
20
25
30
35
40
'
45
Percentage
in
1961
Figure
1
Figure
I
gives
an
example.
Among
the
men
in
the
surveys
in
Corfu,
18
men
were
included
in
the
studies
both
in
1961
and
1963.
The
average
percentages
of
calories
from
fats
in
the
diets
of
these
18
men
were
31.9
in
1961,
30.6
in
1963,
so
the
averages
ogree
very
well.
But
figure
1
shows
that
the
individual
variation
from
survey
to
survey
was
large.
Among
the
9
men
wha
were
below
the
median
in
1961,
4
were
above
the
median
in
1963.
This
means
that
the
ranking
of
the
men
from
lowest
to
highest
fat
diets
in
the
first
survey
is
practically
useless
for
predicting
their
rank
in
this
respect
in
the
next.
This
is
the
result
using
nutrient
values
calculated
from
tables
of
food
composition;
the
data
from
chemical
analysis
of
the
diets
produces
almost
exactly
the
same
result.
Figure
2
concerns
the
individual
data
an
the
proportion
of
calories
from
proteins
in
repeated
surveys
on
the
same
15
men
in
Crete.
In
this
case
there
is
a
substantial
difference
between
the
averages
for
the
same
men
in
the
two
surveys
9.2
in
1962,
11.5
in
1965
but
it
is
again
clear
that
the
relative
positions
of
the
individuals
in
regard
to
protein
calories
in
1965
bear
little
relationship
to
those
observed
in
1962.
O
0
O
0
0
a
0
0
0
ro
0.41
65
15
MEN
OF
CRETE
'/.CALORIES
FROM
PROTEINS
IN
REPEATED
SURVEYS
ts-
O
14--
O
in
to
m
13—
12
C
cro
al
11—
O
C
ao
a.
O
O
10-
r=0.42
9—.
0
0
8
7
8
9
10
11
12
Percentage
in
1962
Figure
2
Table
8
summarizes
the
data
from
repeated
surveys
in
regard
to
total
calories
and
to
percentages
of
colaries
from
proteins,
fats
and
alcohol.
Essentially,
the
degree
to
which
an
individual
can
be
distinguished
among
his
fellows
in
regard
to
the
diet,
ar
any
other
characteristic
that
may
be
the
item
of
measurement,
depends
an
his
intra-individual
variability
as
compared
with
the
variability
between
the
individuals
of
the
group
of
which
he
is
a
part.
With
the
present
material,
the
standard
deviation
about
the
mean
value
in
a
survey
is
o
measure
of
the
inter-individual
variobility
of
the
group.
When
the
survey
is
repeated,
the
geometric
mean
of
the
2
values
for
the
S.D.
may
be
taken
to
be
a
better
estimate
of
the
average
inter-individual
variability;
in
fable
8
this
value
is
designoted
as
8.D.
b
.
The
average
infra-individual
variobility
can
be
similarly
computed
from
all
the
pairs
of
first
and
second
survey
values
far
the
individuals.
For
most
items
in
table
8
this
statistic,
S.D.
w
,
is
of
much
the
same
order
of
magnitude
as
S.D.
b
.
66
Table
8.
Comparison
of
intro-
and
inter-individual
variations
in
the
dietary
data
on
the
same
men
surveyed
on
each
of
2
occasions
N
en
number
of
men.
S.D.
w
and
S.D.
b
are
the
average
standard
deviations
within
and
between
individuals,
respectively.
The
coefficient
of
correlation
between
values
from
the
same
men
on
the
2
surveys
is
given
in
the
lines
designated
by
r.
Area
Years
N
Calories
Food
table
data
%
Calories
Protein
Fot
Alcohol
Calories
Chemical
analysis
%
Calories
Calories
Protein
Fat
Crete,
1960-'65
9
S.D.
w
200
1.9
4.7
63
236
1.3
5.3
S.D.
b
643
1.2
4.7
84
463
1.2
5.4
r
0.83"
0.01
0.02
0.58
0.71
6
-0.21
-0.04
Crete,
196265
15
S.D.
w
598
2.0
7.3
40
494
1.6
4.9
S.D
525
1.6
6.1
186
451
1.6
7.7
r
-0.08
0.42
0.17
0.91"
0.13
0.05
0.47
Corfu,
1961-'63
18
S.D.
w
572
1.3
4.5
139
414
2.1
5.0
S.D
'b
688
1.5
5.9
203
731
2.6
6.5
r
0.34
0.31
0.41
0.51*
0.66"
0.33
0.41
Statistically
significant
at
p
C
0.05
Cl
Statistically
significant
at
p
C
0.01
These
findings
mean,
of
course,
that
only
a
relatively
small
part
of
the
differences
observed
between
individuals
in
any
one
survey
is
truly
dependent
upon
consistent
differences
between
individuals.
Similar
results
were
obtained
in
repeated
surveys
on
the
same
men
elsewhere
(2).
The
values
for
the
correlation
coefficient,
r,
in
table
8
further
show
the
importance
of
intro-individual
variability.
Using
food
table
data,
the
average
value
of
r
is
0.30
for
calories,
0.29
for
protein
per
cent
calories,
0.24
for
fat
per
cent
calories,
and
0.67
far
alcohol.
The
higher
reliability
of
the
alcohol
values
for
characterizing
the
individual
would
be
expected
because
same
of
these
men
never
drink
and
a
few
are
always
heavy
drinkers.
The
men
are
mare
variable
in
other
aspects
of
the
diet.
From
time
to
time
there
are
reports
of
dietary
surveys
made
in
the
interview
method
that
suggest
less
intro-individual
variability
than
found
in
Greece
or
in
the
comparable
surveys
in
Yugoslavia.
This
would
be
expected
because
when
people
simply
are
queried
about
their
diets
their
answers
from
time
to
time
necessarily
reflect
their
own
ideas
of
their
stereotypes;
they
tend
to
repeat
the
same
answers
whether
or
not
they
truly
correspond
to
current
reality.
Summary
Seven-day
dietary
surveys
were
made
on
random
samples
of
middle-aged
men
in
villages
on
the
island
of
Crete
(3
surveys)
and
on
the
island
of
Corfu
(2
surveys)
in
Greece.
All
foods
consumed
were
weighed
and
replicates
of
the
meals
were
collected
for
chemical
analysis.
67
In
both
areas
the
diets
were
dominated
by
olive
oil
and
bread,
with
law
intakes
of
foods
of
animal
origin
and
of
sugar.
Calculations
using
tables
of
food
composition
gave
average
results
in
fair
agreement
with
those
from
direct
chemical
analysis
but
the
chemical
analyses
indicated
lower
values
than
the
tables
for
calories,
proteins
and
especially
for
fats.
In
both
areas
the
intake
of
saturated
and
of
poly-unsatu-
rated
fatty
odds
was
low.
There
were
no
significant
seasonal
differences
in
the
diets
in
regard
to
calories
or
percentage
of
calories
from
proteins
or
from
fats.
The
diets
of
the
men
in
these
two
areas
of
Greece
did
not
differ
significantly
in
calories
or
pratein
but
the
diet
in
Crete
provided
a
higher
proportion
of
fat
calories
(40.3
by
calculation,
36.1
by
analysis)
than
in
Corfu
(32.7
and
27.2,
respectively).
Repeated
surveys
on
the
same
men
showed
infra-individual
variability
to
be
a
large
part
of
the
total
variability
in
regard
to
total
calories,
alcohol
calories,
and
percentages
of
calories
from
proteins
and
from
fats.
The
individual
values
from
a
single
survey
are
of
very
limited
use
in
predicting
individual
values
in
a
second
survey.
Acknowledgments
We
are
grateful
to
Dr.
J.
T.
Anderson,
Dr.
D.
Galanos
and
Prof.
F.
Fidanza
for
collaboration
in
chemical
analyses
and
to
Prof.
George
Michoelides
and
Dr.
E.
Stophylakis
for
encouragement
and
facilitation
of
the
program.
The
dietary
work
in
the
field
was
done
by
Misses
V.
Geranta,
A.
Choniataki,
T.
Victoratou,
M.
Sponou,
A.
Marinau,
S.
Mellisari,
T.
Moschopoulou,
P.
Monolia,
M.
Karimboka
and
P.
Pe-
risteridau.
Mrs.
A.
Brodorec
and
Mrs.
Nedra
Foster
aided
in
the
supervision
of
the
field
work.
References
I.
Anderson,
J.
T.,
F.
Grande,
Y.
Matsumoto
and
A.
Keys:
Glucose,
sucrose
and
lactose
in
the
diet
and
blood
lipids
in
man.
J.
Nutr.
79
(1963),
349-359.
2.
Burin°,
R.,
E.
Ferber,
A.
Keys,
A.
Brodarec,
B.
Agneletto
and
A.
Horvat:
Diets
of
rural
families
and
heads
of
families
in
twa
regions
of
Yugoslavia.
Voeding
25
(1964),
629-639.
3.
Buzina,
R.,
A.
Keys,
A.
Brodorec,
J.
T.
Anderson
and
F.
Fidanza:
Dietary
surveys
in
rural
Yugoslavia.
II.
Chemical
analyses
of
diets
in
Dalmatia
and
Slavonia.
Voeding
27
(1966),
31-36.
4.
Fidanza,
F.,
A. A.
Fidanza,
G.
Ferro-Luzzi
and
M.
Proja:
Dietary
studies
in
connection
with
the
epidemiology
of
heart
disease:
Results
in
Italy.
Voeding
25
11964),
502.509.
5.
Hartog,
C.
Den,
Th.
F.
S.
M.
van
Schalk,
L.
M.
Dalderup,
E.
F.
Drion
and
T.
Mulder:
The
diet
of
volunteers
participating
in
a
long
term
epidemiological
field
survey
on
coronary
heart
disease
at
Zutphen,
the
Netherlands.
Vaeding
26
(1965),
184-208.
6.
Keys,
A.
e.a.:
Epidemiological
studies
related
to
coronary
heart
disease:
Characteristics
of
men
aged
40-59
in
seven
countries.
Ada
med.
Scand.
In
press.
7.
Roine,
P.,
M.
Pekkorinen
and
M.
J.
Karvonent
Dietary
studies
in
connection
with
epidemiology
of
heart
diseases:
Results
in
Finland.
Voeding
25
(1964).
383-393.
68