Determinants of longevity and all-cause mortality among middle-aged men. Role of 48 personal characteristics in a 40-year follow-up of Italian Rural Areas in the Seven Countries Study


Menotti, A.; Lanti, M.; Maiani, G.; Kromhout, D.

Aging Clinical and Experimental Research 18(5): 394-406

2006


Forty-year all-cause mortality and its association with entry risk factor levels are reported for men enrolled in the Italian Rural Areas of the Seven Countries Study of Cardiovascular Diseases. Forty-eight potential risk factors were measured in 1712 men aged 40-59 at entry examination in 1960. Mortality data were collected during 40 years of follow-up. The relationship of entry risk factor levels with all-cause mortality was studied by univariate and multivariate approaches. Overall death rate was 83.7%. The main causes of death were cardiovascular diseases, followed by cancer and others. The 48 risk factors were tested with univariate and multivariate approaches. In the final model, 15 risk factors were strongly and significantly related to all-cause mortality and survival. They were age, father and mother history of premature mortality, cigarette smoking, job-related physical activity (protective), body mass index (BMI) (in an inverse J-shaped fashion), mid-arm circumference (protective), mean blood pressure, forced respiratory volume in 3/4 seconds (protective), serum cholesterol, corneal arcus, xanthelasma, presence of cardiovascular diseases, cancer and diabetes at entry examination. During a 40-year period 15 mainly cardiovascular risk factors were highly predictive of all-cause mortality and survival in middle-aged men.

Aging
Clinical
and
Experimental
Research
Determinants
of
longevity
and
all-cause
mortality
among
middle-aged
men.
Role
of
48
personal
characteristics
in
a
40-year
follow-up
of
Italian
Rural
Areas
in
the
Seven
Countries
Study
Alessandro
Menotti
1
,
Mariapaola
Lanti
1
,
Giuseppe
Maiani
2
,
and
Daan
Kromhout
3
lAssociazione
per
le
Ricerca
Cardiologica
-
Association
for
Cardiac
Research,
Roma,
2
Istituto
Nazionale
di
Ricerca
per
gli
Alimenti
e
la
Nutrizione,
Roma,
Italy,
3
Division
of
Human
Nutrition,
Wageningen
University,
Wageningen,
The
Netherlands
ABSTRACT.
Background
and
aims:
Forty-year
all-
cause
mortality
and
its
association
with
entry
risk
factor
levels
are
reported
for
men
enrolled
in
the
Ital-
ian
Rural
Areas
of
the
Seven
Countries
Study
of
Car-
diovascular
Diseases.
Methods:
Forty-eight
potential
risk
factors
were
measured
in
1712
men
aged
40-59
at
entry
examination
in
1960.
Mortality
data
were
col-
lected
during
40
years
of
follow-up.
The
relationship
of
entry
risk
factor
levels
with
all-cause
mortality
was
studied
by
univariate
and
multivariate
approaches.
Results:
Overall
death
rate
was
83.7%.
The
main
causes
of
death
were
cardiovascular
diseases,
followed
by
cancer
and
others.
The
48
risk
factors
were
tested
with
univariate
and
multivariate
approaches.
In
the
fi-
nal
model,
15
risk
factors
were
strongly
and
signifi-
cantly
related
to
all-cause
mortality
and
survival.
They
were
age,
father
and
mother
history
of
premature
mortality,
cigarette
smoking,
job-related
physical
ac-
tivity
(protective),
body
mass
index
(BMI)
(in
an
in-
verse
J-shaped
fashion),
mid-arm
circumference
(pro-
tective),
mean
blood
pressure,
forced
respiratory
vol-
ume
in
3
/
4
seconds
(protective),
serum
cholesterol,
corneal
arcus,
xanthelasma,
presence
of
cardiovascu-
lar
diseases,
cancer
and
diabetes
at
entry
examination.
Conclusions:
During
a
40-year
period
15
mainly
car-
diovascular
risk
factors
were
highly
predictive
of
all-
cause
mortality
and
survival
in
middle-aged
men.
(Aging
Clin
Exp
Res
2006;
18:
394-406)
©2006,
Editrice
Kurtis
INTRODUCTION
Population
studies
with
very
long
follow-up
periods,
whatever
the
initial
interest,
focus
attention
on
long-
term
survival,
longevity
and
all-cause
mortality,
indepen-
dently
of
the
single
causes
of
death.
In
other
words,
the
natural
question
emerges
regarding
which
are
the
causes
for
longer
or
shorter
survival
times,
beyond
the
single
dis-
eases
which
may
lead
to
death.
The
two
Italian
Rural
Areas
of
the
Seven
Countries
Study
were
enrolled
in
1960,
mainly
to
study
coronary
heart
disease
and
other
cardiovascular
diseases.
A
detailed
anal-
ysis
of
40-year
mortality
from
various
types
of
cardiovascular
diseases
has
already
been
produced
(1),
confirming
the
predictive
role
of
classical
cardiovascular
risk
factors,
even
during
such
a
long
follow-up
period.
Attention
to
survival
and
all-cause
mortality
represents
the
immediate
next
step.
This
analysis
therefore
deals
with
survival
and
all-
cause
mortality
in
two
cohorts
of
middle-aged
men
fol-
lowed
for
40
years
-
that
is,
for
almost
all
their
remaining
life-span.
The
hypothesis
is
that
a
number
of
personal
characteristics
measured
in
middle
age
are
associated
with
long-term
survival
and
mortality.
METHODS
The
epidemiological
material
used
for
this
analysis
derives
from
the
two
Italian
rural
cohorts
of
the
Seven
Countries
Study
of
Cardiovascular
Diseases,
made
up
of
men
aged
40
to
59
years,
enrolled
and
first
examined
in
1960
(2).
They
represented
98.8%
(n=1712)
of
defined
samples
belonging
to
the
rural
communities
of
Crevalcore
in
Northern
Italy
and
Montegiorgio
in
Central
Italy.
Field
surveys
with
measure-
ment
of
risk
factors
and
clinical
evaluation
were
conducted
at
entry
and
then
again
on
survivors
every
5
years
for
40
years.
For
the
purposes
of
this
analysis,
only
baseline
measurements
of
risk
factors
are
considered,
together
with
information
on
mortality
over
40
years.
Key
words:
Longevity,
mortality,
prediction,
risk
factors.
Correspondence:
A.
Menotti,
MD,
PhD,
VP
Association
for
Cardiac
Research,
Via
Latina
49,
00179
Roma,
Italy.
E-mail:
menottia@tin.it
Received
May
16,
2005;
accepted
in
revised
form
September
30,
2005.
394
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
Aging
Clin
Exp
Res
18:
394-406,
2006
Determinants
of
all-cause
mortality
in
40
years
©2006,
Editrice
Kurtis
Risk
factors
used
in
this
analysis
were
the
following:
age;
father
history,
mother
history,
family
history
of
car-
diovascular
diseases;
marital
status,
children;
job-related
physical
activity,
cigarette
smoking,
diet
of
any
type
followed
by
subjects;
height,
weight,
body
mass
index
(BMI),
trunk/height
ratio,
biacromial
diameter,
bicrystal
diameter,
shoulder/pelvis
shape,
laterality/linearity
in-
dex,
tricipital
skinfold
thickness,
subscapular
skinfold
thickness,
arm
circumference;
systolic
blood
pressure,
di-
astolic
blood
pressure,
mean
blood
pressure,
heart
rate;
vital
capacity,
forced
expiratory
volume;
serum
choles-
terol,
urine
protein,
urine
glucose;
baldness,
corneal
arcus,
xanthelasma;
diagnoses
of
cardiovascular
dis-
eases,
cancer,
diabetes,
chronic
bronchitis;
history
of
lung
tuberculosis,
bronchial
asthma,
peptic
ulcer,
in-
testinal
diseases,
liver
diseases,
gall
bladder
diseases,
kid-
ney
stones,
other
genito-urinary
diseases,
thyroid
disease;
minor
ECG
abnormalities
at
rest,
exercise
ECG
abnor-
malities.
Unit
of
measurements,
technical
details,
mean
levels,
and
use
of
values
for
univariate
and
multivariate
analysis
are
reported
in
the
Appendix,
together
with
spe-
cific
references
(2-4).
Collection
of
data
on
vital
status
and
causes
of
death
was
complete
for
40
years,
and
only
three
subjects
were
lost
to
follow-up
at
a
defined
censoring
date
(35
years).
Causes
of
death
were
allocated
by
reviewing
and
com-
bining
information
from
death
certificates,
hospital
and
medical
records,
interviews
with
physicians
and
relatives
of
the
deceased,
and
any
other
witnesses
of
fatal
events.
Causes
of
death
were
determined
by
a
single
reviewer
(AM)
following
defined
criteria,
employing
the
8th
Revision
of
the
WHO-ICD
(5).
In
the
presence
of
multiple
causes,
a
hierarchical
preference
was
adopted
with
violence,
cancer
in
advanced
stages,
coronary
heart
disease,
stroke,
and
others,
in
that
order.
The
baseline
survey
was
conducted
well
before
the
era
of
the
Helsinki
Declaration.
On
the
occasion
of
subsequent
examinations,
verbal
consent
was
obtained
in
view
of
col-
lecting
follow-up
data.
The
end-point
of
this
analysis
was
all-cause
mortality
in
40
years,
or
corresponding
survival
in
some
analyses.
A
survival
curve
during
40
years
of
follow-up
was
constructed
using
the
life
table
technique.
The
contribution
of
different
causes
of
death
to
over-
all
40-year
mortality
was
computed
based
on
selected
broad
categories
and
sub-categories
of
death.
Univariate
analysis
This
was
run
by
computing
cumulative
survival
using
the
life
table
technique
and
the
log-rank
test
for
all-cause
mortality
as
a
function
of
each
risk
factor.
Dichotomic
risk
factors
were
used
as
such.
For
physical
activity,
sedentary
and
moderately
active
workers
were
pooled
together
and
compared
with
heavy
workers.
For
cigarette
smoking,
two
classes
were
made:
current
smokers
and
all
the
oth-
ers.
For
each
of
the
other
continuous
risk
factors,
two
ar-
bitrary
classes
were
created,
using
as
cut-off
the
median
value,
approximately
(details
in
the
Appendix).
Selection
of
risk
factors
for
multivariate
analysis
Among
the
48
risk
factors
tested
in
univariate
anal-
ysis,
27
had
a
log-rank
test
with
p<0.05
and
were
therefore
chosen
for
multivariate
analysis.
However,
some
of
them
were
highly
correlated
with
each
other.
As
a
consequence,
a
preliminary
selection
of
risk
factors
was
made
before
running
the
final
multivariate
models.
In
particular:
-
the
three
blood
pressure
values
(systolic,
diastolic,
mean)
were
highly
correlated
with
each
other;
proportional
haz-
ard
models
were
run
with
all-cause
mortality
in
40
years
as
end-point
and,
as
covariates,
the
selected
risk
factors
but
using
the
three
blood
pressure
values
alter-
natively
in
three
different
models;
lastly,
mean
blood
pressure
was
chosen,
since
the
model
including
mean
blood
pressure
had
a
log-likelihood
statistic
with
a
chi-
square
significantly
larger
than
in
models
with
the
other
two
blood
pressure
values;
-
the
two
respiratory
measurements
were
also
highly
correlated
and
therefore
independent
models
were
run
with
each
of
the
two
variables,
plus
the
remaining
risk
factors;
forced
expiratory
volume
was
chosen
since
the
model
including
this
variable
had
a
statistically
larger
chi-
square
of
the
log-likelihood
statistic
compared
with
that
of
vital
capacity;
-
urine
glucose
and
diagnosis
of
diabetes
were
highly
correlated,
since
urine
glucose
was
part
of
the
diagnostic
criteria
for
diabetes;
independent
models
were
run
with
each
of
the
two
variables,
plus
the
remaining
risk
factors;
the
diagnosis
of
diabetes
was
chosen
since
the
model
in-
cluding
diabetes
had
a
larger
(although
not
significant)
chi-square
of
the
log-likelihood
statistic
compared
with
the
model
including
urine
glucose.
In
this
way,
four
out
of
27
risk
factors
were
prelimi-
narily
excluded.
Multivariate
analysis
This
was
run
using
the
proportional
hazard
model.
Risk
factors
were
entered
in
the
model
with
their
own
units
of
measurements,
as
defined
in
the
Appendix.
The
end-point
was
all-cause
morality
in
40
years.
Several
models
were
computed
in
succession.
Model
1
was
run
using
the
23
risk
factors
with
a
p-val-
ue
<0.05
in
the
univariate
log-rank
test
(after
the
exclusion
of
other
four,
as
explained
above).
Model
2
was
run
with
16
risk
factors,
after
exclusion
of
seven
risk
factors
with
a
p-value
>0.10
in
the
first
step.
Model
3
included
only
14
risk
factors
with
a
p-value
<0.05
in
the
previous
model,
which
meant
the
exclusion
of
two
more
risk
factors.
In
this
model,
all
risk
factors
had
a
coefficient
with
a
p-value
<0.05.
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
395
111111111111111111111111111111111111111
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
1617
1819
20
21
22
23
24
25 26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Years
of
follow-up
co
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
A.
Menotti,
M.
Lanti,
G.
Maiani,
et
al.
Table
1
-
Causes
of
death
in
40
years
from
selected
groups
and
subgroups
of
diseases.
Aging
Clin
Exp
Res
18:
394-406,
2006
©2006,
Editrice
Kurtis
Causes
of
death
Absolute
Rate
per
1000
Proportion
of
Proportion
within
number
in
40
years
all
causes
%
specific
groups
%
Cardiovascular
diseases
655
383
45.7
Within
cardiovascular
diseases
Coronary
heart
disease
268
157
18.7
40.9
Heart
disease
of
uncertain
etiology
126
74
8.8
19.2
Heart
diseases
of
defined
etiology
46
27
3.2
7.0
Cerebrovascular
diseases
195
114
13.6
29.7
Peripheral
artery
diseases
20
12
1.4
3.1
Cancer
418
244
29.1
Within
cancer
Lung
cancer
70
41
4.9
16.7
Other
cancers
348
203
24.3
83.3
Chronic
bronchitis
49
29
3.4
Infectious
diseases
17
10
1.2
Violent
causes
65
38
4.5
All
other
causes
230
134
16.0
All
causes
1434
838
100.0
Alive
278
All
1712
Model
4
included
all
the
risk
factors
used
in
Model
3,
with
the
addition
of
the
linear
and
quadratic
components
of
BMI,
since
a
curvilinear
(parabolic)
relationship
was
suspected
(overall,
15
factors,
with
BMI
with
two
components).
An
independent
model
was
then
run,
using
the
same
risk
factors
employed
in
Model
4,
but
all
of
them
in
di-
chotomized
form.
In
this
case,
BMI
was
entered
as
a
sin-
gle
risk
factor
in
a
dichotomized
fashion.
The
dichotomies
of
the
various
risk
factors
are
listed
in
the
Appendix.
Plots
of
Schoenfeld
residuals
over
time
were
produced
to
test
the
proportionality
of
hazard,
for
risk
factors
re-
maining
in
the
model
at
the
end
of
the
step-wise
procedure.
Survival
and
95%
confidence
limits
in
(40-year)
follow
up
Fig.
1
-
Survival
curve
during
40-year
follow-up.
Dotted
lines
represent
95%
confidence
intervals.
396
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
Aging
Clin
Exp
Res
18:
394-406,
2006
Determinants
of
all-cause
mortality
in
40
years
©2006,
Editrice
Kurtis
The
coefficients
and
constant
of
the
multivariate
Model
4
were
applied
back
to
the
original
risk
factor
levels
of
all
men,
to
obtain
an
estimated
risk
of
death.
Observed
cases
of
deaths
were
computed
in
decile
classes
of
estimated
risk,
to
obtain
a
relative
risk
be-
tween
the
extreme
deciles.
The
same
procedure
was
carried
out
for
subjects
who
survived
40
years.
A
regression
equation
between
estimated
risk
(inde-
pendent
variable)
and
length
of
survival
(dependent
vari-
able)
was
then
used
to
estimate
the
lost
years
of
life
as
a
Table
2
-
Univariate
analysis.
Kaplan-Meier
cumulative
survival
and
log-rank
test
for
all-cause
mortality
in
40
years
as
a
function
of
sin-
gle
risk
factors
arbitrarily
dichotomized
in
2
classes.
Classes
0
and
I
as
defined
in
the
Appendix.
Risk
factor
Cumulative
survival
p
of
log-rank
test
Class
0
Class
1
Age
0.2769
0.0485
<0.0001
Father
history
0.1735
0.1468
0.0002
Mother
history
0.1694
0.1388
0.0007
Family
history
of
cardiovascular
diseases
0.1750
0.1672
0.0897
Marital
status
0.1568
0.1959
0.0542
Children
0.1826
0.1624
0.4482
Physical
activity
0.1528
0.1750
<0.0001
Cigarette
smoking
0.2083
0.1297
<0.0001
Diet
0.1560
0.1915
0.3878
Height
0.1524
0.1953
0.1678
Sitting
height
0.1328
0.1807
0.0011
Weight
0.1569
0.1649
0.3643
Body
mass
index
(BMI)
0.1671 0.1536
0.8023
Trunk/height
ratio
0.1611
0.1606
0.2163
Biacromial
diameter
0.1684
0.1487
0.4285
Bicrystal
diameter
0.1749
0.1458
0.0513
Shoulder/pelvis
shape
(ratio)
0.1461
0.1731
0.0178
Laterality-linearity
index
0.2084
0.1501
0.1623
Tricipital
skinfold
0.1596
0.1621
0.9813
Subscapular
skinfold
0.1632
0.1885
0.7392
Mid-arm
circumference
0.1367
0.1861
<0.0001
Systolic
blood
pressure
0.2274
0.1004
<0.0001
Diastolic
blood
pressure
0.2043
0.1130
<0.0001
Mean
blood
pressure
0.2179
0.0957
<0.0001
Heart
rate
0.1663
0.1552 0.0262
Vital
capacity
0.1105
0.2138
<0.0001
Forced
expiratory
volume
in
3
/
4
sec
0.1073
0.2201
<0.0001
Serum
cholesterol
0.1924
0.1565
<0.0001
Urine
protein
0.1680
0.1061
<0.0001
Urine
glucose
0.1679
0.0519
<0.0001
Baldness
0.1641
0.1528
0.0856
Corneal
arcus
0.1793
0.0462
<0.0001
Xanthelasma
0.1657
0.0000
<0.0001
Diagnosis
of
cardiovascular
disease
0.1987
0.1658
<0.0001
Diagnosis
of
cancer
0.1612
0.0000
<0.0001
Diagnosis
of
chronic
bronchitis
0.1646
0.1215
<0.0001
Diagnosis
of
diabetes
0.1616
0.1446
0.0333
History
of
tuberculosis
0.1630
0.1111
0.1753
History
of
bronchial
asthma
0.1627
0.1026
0.0287
History
of
peptic
ulcer
0.1611
0.1632
0.3687
History
of
intestinal
disease
0.1532
0.2000
0.0362
History
of
liver
disease
0.1556
0.2879
0.0872
History
of
gall
bladder
disease
0.1597
0.4000
0.9015
History
of
kidney
stones
0.1605
0.1959
0.7288
History
of
genito-urinary
disease
0.1631
0.1045
0.5270
History
of
thyroid
disease
0.1610
0.3333
0.0414
Minor
ECG
abnormalities
0.1628
0.1471
0.0140
ECG
abnormalities
after
exercise
0.1629
0.1164
0.3636
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
397
A.
Menotti,
M.
Lanti,
G.
Maiani,
et
al.
Aging
Clin
Exp
Res
18:
394-406,
2006
©2006,
Editrice
Kurtis
function
of
arbitrary
levels
of
each
significant
risk
factor,
vs
arbitrary
reference
levels
of
the
same
factors.
RESULTS
Appendix
1
provides
a
list
of
all
risk
factors
mea-
sured
at
baseline
and
employed
in
this
analysis.
It
also
in-
cludes
details
of
their
definitions,
units
of
measurement,
mean
levels
and
standard
deviations
(or
proportions
with
their
standard
errors,
when
appropriate),
values
used
in
the
univariate
analysis
(all
risk
factors
dichotomized)
and
values
used
in
multivariate
analysis.
Mean
levels
reflect
the
values
of
two
Italian
rural
population
samples
in
1960,
when
these
data
were
the
only
epidemiological
infor-
mation
available
in
the
country.
In
the
40-year
follow-up,
there
were
1434
deaths
(83.8%).
Survivors
numbered
278
(16.2),
although
for
three
of
them
follow-up
was
stopped
at
35
instead
of
40
years
(Table
1).
The
major
causes
of
death
were
cardiovascular
diseases
(45.7%),
followed
by
cancer
and
other
causes.
Within
car-
diovascular
diseases
(100%),
more
than
40%
were
at-
tributed
to
coronary
heart
disease,
almost
30%
to
stroke,
and
almost
20%
to
heart
diseases
of
undefined
etiology.
Among
cancer
deaths,
almost
17%
were
lung
cancer.
Figure
1
shows
survival
during
the
40-year
follow-up,
revealing
a
slight
acceleration
after
7-8
years
of
follow-up
and
then
a
steady
decline.
The
outcome
of
univariate
analysis,
based
on
the
life
table
technique,
is
given
in
Table
2,
in
which
it
appears
that
27
out
of
the
48
risk
factors
were
significantly
associated
with
sur-
vival
(and
inversely
with
death)
(p<0.05).
In
general,
signif-
icant
factors
included
traditional
cardiovascular
risk
factors,
some
chronic
diseases
diagnosed
on
physical
examination
or
reported
by
subjects,
some
indicators
of
family
health
history,
some
physical
signs
(corneal
arcus,
xanthelasma),
plus
four
protective
characteristics
perhaps
associated
with
physical
fit-
ness
(physical
activity
at
work,
mid-arm
circumference,
vital
capacity,
forced
expiratory
volume).
Multivariate
Model
1
included
all-cause
mortality
with
40
years
as
the
end-point,
and
23
risk
factors
as
covariates,
af-
ter
exclusion
of
other
four,
as
described
in
Material
and
Methods.
In
this
multivariate
Model
1,
seven
risk
factors
produced
coefficients
with
p-values
>0.10
and
were
there-
fore
excluded
from
the
next
step
(heart
rate,
urine
protein,
sitting
height,
shoulder/pelvis
shape,
history
of
asthma,
his-
tory
of
thyroid
disease,
minor
ECG
abnormalities).
Multivariate
Model
2
was
run
on
16
risk
factors.
In
this
case,
two
risk
factors
showed
coefficients
with
p-value
>0.05
and
were
excluded
from
the
next
step
(diagnosis
of
chronic
bronchitis
and
history
of
intestinal
disease).
Multivariate
Model
3
included
the
remaining
14
risk
factors,
all
of
which
had
coefficients
with
p-values
<0.05.
Multivariate
Model
4
included
the
remaining
14
risk
factors
and
the
linear
and
quadratic
components
of
the
BMI.
All
had
coefficients
with
p-values
<0.05,
except
the
quadratic
component
of
BMI
which
had
a
p-value
of
0.057.
The
chi-square
likelihood
statistic
for
model
3
was
635.69
(p<0.001)
and
for
model
4
was
636.81
(p<0.001).
This
means
that
the
coefficients
of
the
BMI
add
very
little
to
prediction,
although
they
are
significant.
Table
3
-
Proportional
hazard
model
with
all-cause
mortality
as
end-point
and
15
risk
factors
as
predictors.
Risk
factor
Coefficient
and
(SE)
Difference
for
evaluation
of
hazard
ratio
Hazard
ratio
95%
confidence
intervals
Standardized
coefficients
and
(rank)
Age
0.1021
(0.0064)
5
years
1.67
1.56-1.77
0.5152
(1)
Father
history
0.1692
(0.0679)
No-yes
1.18
1.04-1.35
0.0689
(10)
Mother
history
0.2166
(0.0695)
No-yes
1.24
1.08-1.42
0.0875
(7)
Physical
activity
-0.0986
(0.0446)
1
grade
out
of
3
0.91
0.83-0.99
-0.0636
(13)
Cigarette
smoking
0.0194
(0.0029)
10
cigarettes
per
day
1.21
1.15-1.29
0.1841
(3)
Body
mass
index*
-0.1425
(0.0721)
4
units
0.57
0.33-1.00
-
Body
mass
index
2
*
0.0025
(0.0013)
200
units
1.65
0.99-2.74
-
Mid-arm
circumference
-0.0036
(0.0017)
25
mm
0.91
0.84-0.99
-0.0840
(8)
Mean
blood
pressure
0.0205
(0.0024)
15
mmHg
1.36
1.27-1.46
0.2764
(2)
Forced
expiratory
volume
in
3
/
4
sec
-0.5770
(0.1256)
0.25
L/m
2
0.87
0.81-0.92
-0.1439
(4)
Serum
cholesterol
0.0851
(0.0271)
1
mmol/L
1.09
1.03-1.15
0.0908
(6)
Corneal
arcus
0.1838
(0.0795)
No-yes
1.20
1.03-1.40
0.0643
(12)
Xanthelasma
0.5550
(0.2142)
No-yes
1.74
1.14-2.65
0.0685
(11)
Diagnosis
of
cardiovascular
disease
0.3714
(0.1315)
No-yes
1.45
1.12-1.88
0.0761
(9)
Diagnosis
of
cancer
2.1717
(0.4554)
No-yes
8.77
3.59-21.42
0.1232
(5)
Diagnosis
of
diabetes
0.2193
(0.1103)
No-yes
1.25
1.00-1.55
0.0549
(14)
*Body
mass
index
(BMI)
appears
with
both
linear
and
(
2
)
quadratic
components.
398
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
linear
_
quadratic
0.89
-
a)
0
0.87
-
c
is
0.85
uJ
cc
0.83
-
0.81
-
0.79
I I I I I I I I I I I I I I I I I I
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
BMI,
kg/m
squared
0.93
-
0.91
-
Aging
Clin
Exp
Res
18:
394-406,
2006
Determinants
of
all-cause
mortality
in
40
years
©2006,
Editrice
Kurtis
Table
4
-
Years
of
life
lost,
within
40-year
period,
as
a
function
of
ar-
bitrary
levels
of
risk
factors
vs
arbitrary
reference
levels.
Risk
factor
Years
of
life
lost
within
40
years
vs
reference
Age
Reference:
40
years
50
years
10.3
-
4
60
years
14.61
Father
history
Reference:
no
yes
1.38
Mother
history
Reference:
no
yes
1.73
Physical
activity
Reference:
active
moderate
0.84
sedentary
1.60
Cigarette
smoking
Reference:
0
cigarettes
10
cigarettes
1.74
20
cigarettes
3.21
30
cigarettes
4.44
Body
mass
index
Reference:
30
kg/m
2
(BMI)
20
kg/m
2
25
kg/m
2
35
kg/m
2
2.6
-
5
0.95
0.05
smoking
habits,
mean
blood
pressure,
serum
choles-
terol,
the
protective
characteristics
of
mid-arm
circum-
ference
and
forced
expiratory
volume,
presence
of
chronic
diseases
such
as
cardiovascular
diseases,
dia-
betes
and
cancer,
presence
of
two
physical
signs,
i.e.,
corneal
arcus
and
xanthelasma,
and
the
curvilinear
re-
lationship
with
BMI.
Table
3
gives
also
hazard
ratios,
based
on
arbitrary
differences
in
the
levels
of
each
risk
factor,
and
standardized
coefficients
(the
coefficient
mul-
tiplied
by
the
standard
deviation
of
the
risk
factor),
which
offer
some
information
on
the
relative
impor-
tance
and
rank
of
the
various
risk
factors.
The
special
case
of
BMI
is
shown
in
Figure
2,
in
which
the
estimated
risk,
everything
else
being
equal,
is
drawn
using
the
two
linear
and
quadratic
components,
to-
gether
with
the
curve
derived
from
a
model
in
which
only
the
linear
component
is
shown,
everything
else
be-
ing
equal.
In
the
quadratic
curve,
it
appears
that
the
lev-
el
of
BMI
corresponding
to
the
lowest
mortality
risk
is
lo-
cated
over
the
value
of
30,
with
a
definitely
higher
al-
though
declining
risk
between
20
and
30,
and
a
rise
Mid-arm
circumference
Reference:
293
mm
269
mm
24-5
mm
0.7
-
3
1.41
Mean
blood
Reference:
90
mmHg
pressure
100
mmHg
110
mmHg
1.9
-
9
3.71
Forced
expiratory
volume
in
3
/
4
sec
Reference:
1.35
L/m
2
1.10
L/m
2
0.85
L/m
2
1.3
-
0
2.46
Serum
cholesterol
Reference:
4.1
mmol/L
5.1
mmol/L
6.1
mmol/L
7.1
mmol/L
0.7
-
5
1.46
2.11
Comeal
arcus
Reference:
no
yes
1.47
Xanthelasma
Reference:
no
yes
3.50
Diagnosis
of
Reference:
no
cardiovascular
disease
yes
2.66
Diagnosis
of
Reference:
no
cancer
yes
4.71
Diagnosis
of
diabetes
Reference:
no
yes
1.70
The
first
three
models
mentioned
above
are
not
re-
ported
in
detail,
since
they
are
bulky
and
not
necessari-
ly
informative.
The
final
multivariate
Model
4
is
report-
ed
in
Table
3.
It
includes
age,
family
history
of
both
par-
ents,
living
habits
of
physical
activity
(protective)
and
Fig.
2
-
Relationship
of
BMI,
using
only
linear
component
or
both
linear
and
quadratic
components,
with
all-cause
mortality
risk
in
40
years.
Estimate
adjusted
for
14
other
risk
factors.
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
399
B
90—
.c
80—
70—
o
60—
E
c'
3
50—
oc
o
40-
30—
E
20—
z
10-
78
59
A.
Menotti,
M.
Lanti,
G.
Maiani,
et
al.
Aging
Clin
Exp
Res
18:
394-406,
2006
©2006,
Editrice
Kurtis
Num
ber
o
f
dea
t
hs
in
40
y
ears
>
which
is
only
visible
around
the
level
of
32.
Together,
the
two
curves
are
quite
different,
with
higher
values
of
risk
for
the
quadratic
curve
in
the
lower
BMI
levels
and
viceversa
for
the
higher
BMI
levels.
Another
model
was
run
with
the
same
15
risk
factors
as
in
model
4,
all
of
them
being
defined
in
a
dichotomic
way.
In
general,
the
t-value
of
the
coefficients
were
larg-
er,
compared
with
the
basic
model,
but
the
overall
likeli-
hood
statistic
showed
a
definitely
smaller
chi-square
val-
ue
(507.28
vs
636.81).
The
coefficients
of
multivariate
Model
4
were
applied
back
to
the
original
risk
factor
levels
of
all
men,
to
obtain
an
estimated
risk.
Observed
cases
were
computed
in
decile
classes
of
estimated
risk,
as
shown
in
Figure
3a.
There
is
an
increasing
risk
with
increasing
deciles,
but
the
discrimination
is
not
very
good,
since
in
the
upper
deciles
almost
all
subjects
had
died.
The
relative
risk
be-
tween
decile
10
and
decile
1
is
of
the
order
of
2.
Figure
3b,
which
shows
the
number
of
those
still
alive
after
40
years
in
the
10
deciles
of
estimated
risk,
is
more
infor-
mative,
showing
that
the
probability
of
being
alive
after
150
152
153
154
77
0.4769
0.5279310.667410.74
5
54610.833410.896210.945710.976610.995610.9999
1
3
6
7
8
9
10
Deciles
of
estimated
risk
and
highest
value
of
risk
in
each
decile
51
0
0.227810.478
10.5
3
79710.668
10.755210.8335
6
10.896710.945810.976910.9956
10
1
1
2
4
5
7
8
9
I
23
23
18
I
6
3
2
Deciles
of
estimated
risk
and
lowest
value
of
risk
in
each
decile
Fig.
3
-
Number
of
subjects
who
died
(Section
A)
or
were
still
liv-
ing
(Section
B)
in
40
years,
distributed
in
decile
classes
of
estimated
risk
of
death.
Section
A
also
reports
the
highest
risk
value
in
each
decile;
Section
B
reports
the
lowest
risk
value
in
each
decile.
40
years
is
78
times,
comparing
decile
1
with
decile
10.
For
each
decile
class,
the
highest
and
lowest
risk
levels
are
also
shown.
The
years
of
life
lost
for
defined
levels
of
each
risk
fac-
tor
(vs
an
arbitrarily
defined
reference
level
for
each
of
them)
were
computed
for
the
15
risk
factors
included
in
Multivariate
Model
4
(Table
4).
It
appears
that
age
alone
plays
a
major
role
in
the
amount
of
lost
years,
as
a
function
of
increasing
levels
of
a
single
risk
factor.
However,
the
combination
of
4
or
5
undesirable
levels
of
the
other
risk
factors
corresponds
to
about
10-year
difference
in
age.
For
example,
combining
four
non-modifiable
risk
factors
(father
and
mother
history,
corneal
arcus,
xanthelasma)
the
lost
years
of
life
total
8.08;
combining
the
three
major
prevalent
diseases, although
an
exceptional
situation
(car-
diovascular
disease,
cancer
and
diabetes),
the
lost
years
of
life
are
9.07;
combining
the
worst
situation
of
each
of
the
other
modifiable
factors
(physical
activity,
cigarette
smok-
ing,
BMI,
mid-arm
circumference,
mean
blood
pressure,
forced
expiratory
volume,
serum
cholesterol),
the
lost
years
of
life
are
18.38.
DISCUSSION
In
the
40-year
follow-up,
83.8%
of
these
middle-aged
men
died.
Within
the
Seven
Countries
Study,
data
have
been
published
so
far
only
for
two
other
cohorts.
The
overall
death
rate
in
these
Italian
rural
areas
is
similar
to
that
recorded
in
the
US
Railroad
cohort
(83.2%)
(6)
and
that
of
the
cohort
on
the
island
of
Corfu
(Greece)
(86.8%)
(7).
Other
international
comparisons
are
not
available
for
the
same
initial
age
range
and,
mainly,
for
the
same
length
of
follow-up.
As
everywhere,
in
industrialized
nations
the
dominant
causes
of
death
are
cardiovascular
diseases,
followed
by
cancer,
and
then
other
causes
(8).
The
role
of
a
large
number
of
possible
predictors
of
death
or
survival
is
partly
obscured
by
the
facts
that
some
of
them
were
measured
in
an
approximate
way
(e.g.,
some
of
the
reported
diagnoses
at
entry)
and
that
several
of
them
are
somehow
correlated
with
each
other.
Despite
this,
15
factors
or
personal
characteristics
showed
a
definite
predictive
(or
protective)
role
for
all-cause
mor-
tality
in
such
a
long
follow-up,
although
they
were
mea-
sured
only
once
at
entry.
In
a
correlation
matrix
among
the
15
risk
factors
used
in
the
final
model,
only
in
six
out
of
105
cells
was
the
cor-
relation
coefficient
equal
to
or
greater
than
0.20
(i.e.,
ex-
plaining
at
least
4%
of
variance
of
one
variable
over
an-
other).
Three
of
these
cells
involved
age
and
three
involved
BMI.
As
a
consequence,
the
overall
system
seems
to
be
relatively
free
from
over-con-elation
and
the
risk
factors
are
relatively
independent.
The
relative
risks
found
for
various
risk
factor
com-
binations
may
have
been
overestimated,
since
the
co-
180
160-
140
120-
100-
80
60
40
20
0
I I I
132
132
137
96
104
1 I I
400
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
Aging
Clin
Exp
Res
18:
394-406,
2006
Determinants
of
all-cause
mortality
in
40
years
©2006,
Editrice
Kurtis
efficients
applied
for
the
estimates
were
those
derived
from
the
same
population.
However,
no
other
similar
model
from
other
studies
could
be
found
with
the
same
end-point,
the
same
risk
factors,
and
the
same
length
of
follow-up.
It
is
within
expectations
that
some
recognized
and
strong
cardiovascular
risk
factors
(e.g.,
age,
blood
pressure,
cigarette
smoking,
serum
cholesterol)
are
also
predic-
tive
of
all-cause
mortality,
since
a
large
proportion
of
all
causes
of
death
is
made
up
of
cardiovascular
diseases.
Cigarette
smoking
is
also
related
to
a
number
of
cancer
deaths
and
to
chronic
bronchitis.
For
parallel
although
different
reasons,
it
may
also
be
reasonable
to
assume
that
some
severe
conditions,
such
as
cardiovascular
diseases,
cancer
and
diabetes,
are
related
to
early
death,
although
these
conditions
were
rel-
atively
rare
at
entry.
Instead,
there
were
two
risk
factors
(mid-arm
circum-
ference
and
forced
expiratory
volume),
partly
correlated
to
physical
activity,
and
physical
activity
itself,
which
showed
an
important
protective
role,
corresponding,
for
high
levels,
to
a
substantial
increase
in
the
expectancy
of
life.
The
interesting
but
unexplained
fact
is
that,
in
sepa-
rate
multivariate
models
not
reported
in
detail,
mid-arm
circumference
was
not
significantly
protective
for
car-
diovascular
disease
and
cancer
deaths,
whereas
forced
ex-
piratory
volume
was
also
protective
for
non-cardiovascular
deaths
combined
with
non-cancer
deaths.
The
early
death
of
parents,
due
to
non-violent
or
non-infectious
causes,
were
also
strongly
associated
with
the
risk
of
death.
However,
in
separate
analyses
(not
shown),
the
association
did
not
hold
for
cardiovascular
and
cancer
deaths,
probably
due
to
lack
of
statistical
power
as-
sociated
with
smaller
numbers.
Lastly,
we
found
an
important
predictive
role
of
xan-
thelasma
and
corneal
arcus,
which
should
indicate
disor-
ders
in
the
area
of
lipid
metabolism.
However,
strangely
enough,
in
a
separate
model
both
of
these
were
also
pre-
dictors
of
death
from
cancer.
The
coefficient
for
BMI
was
negative
and
non-sig-
nificant,
when
considered
in
a
linear
fashion,
while
it
as-
sumed
a
parabolic
(inverse
J-shaped)
relationship
when
both
linear
and
quadratic
terms
were
considered.
The
re-
lationship
of
body
fatness
(at
least
as
expressed
by
BMI)
with
mortality
is
a
complex
issue,
with
contrasting
evidence
in
the
literature,
and
it
also
depends
on
the
ar-
tificial
cut-off
point
chosen
to
define
obesity.
In
univariate
analysis,
survival
was
slightly
longer
for
those
with
a
BMI
<25
units,
but
this
advantage
was
marginal
and
without
statistical
significance.
The
association
between
BMI
and
mortality
may
be
more
closely
related
to
specific
dis-
tribution
than
to
the
absolute
value
of
BMI.
This
popu-
lation
also
had
a
high
level
of
working
physical
activity,
which
may
have
protected
even
subjects
with
relatively
high
levels
of
BMI.
In
fact,
the
inverse
association
(or
parabolic
but
substantially
inverse
until
around
30
units)
was
of
course
adjusted
for
other
risk
factors,
but
included
an
overall
high
mean
level
of
physical
activity.
The
large
contribution
of
unfitness
and
obesity
to
mortality-
longevity
is
clearly
shown
in
another
recent
contribution
(9).
Another
possibility
is
that
the
association
between
obesity
and
mortality
is
linked
to
secular
trends,
as
shown
in
a
population-based
analysis
in
the
USA
(10).
The
transformation
of
risk
estimates
into
lost
years
of
life,
as
a
function
of
various
levels
of
the
risk
factors,
sim-
ply
offers
a
different,
probably
more
friendly,
way
of
evaluating
the
role
of
the
selected
risk
factors
on
the
expectancy
of
life.
Similar
reports
linking
cardiovascular
risk
factors
to
all-
cause
mortality
are
relatively
rare
in
the
literature
or
they
are
frequently
attached
to,
or
mixed
with,
analyses
dealing
with
cardiovascular
disease
end-points.
A
very
long
follow-up,
comparable
to
the
one
described
here,
is
also
rarely
reported.
MRFIT
analysis
of
primary
screenees
provided
extensive
data
on
the
association
of
cardiovascular
risk
factors
and
all-cause
mortality
(11).
Diastolic
blood
pressure,
smoking
habits,
and
serum
total
cholesterol,
were
all
strongly
as-
sociated
with
all-cause
mortality
risk,
although
the
strength
of
this
association
declined
with
age.
A
comparison,
re-
ported
in
the
same
paper,
with
findings
of
other
non-recent
US-based
population
studies
in
men
(Framingham,
Albany,
Chicago-Gas,
Chicago-Western
Electric,
Tecumseh)
sug-
gests
that
smoking
habits
and
diastolic
blood
pressure
are
significant
predictors
of
all-cause
mortality
in
all
those
studies,
while
serum
cholesterol
was
so
only
in
Tecumseh.
This
raises
uncertainties
about
the
role
of
serum
cholesterol
as
a
predictor
of
all-cause
mortality.
In
the
Chicago
population
studies,
there
were
some
conflicting
findings
(12).
In
the
Western
Electric
sample,
for
example,
a
positive
relationship
was
found
between
blood
pressure,
cigarette
smoking,
serum
cholesterol
and
all-cause
mortality,
but
not
for
BMI
in
either
linear
or
parabolic
transformations.
Instead,
in
the
People's
Gas
co-
hort,
serum
cholesterol
was
not
significantly
predictive,
but
a
significant
U-shaped
relationship
was
found
for
BMI.
A
more
recent
analysis
on
the
same
material
showed
the
im-
pact
of
risk
factor
combinations
on
all-cause
mortality,
call-
ing
for
urgent
public
health
action
in
this
group
of
indi-
viduals
(13).
In
a
Dutch
longitudinal
study
with
a
12-year
follow-up,
smoking habits
and
high
blood
pressure
were
directly
related
to
all-cause
mortality
(14),
whereas,
as
in
our
findings,
BMI
was
inversely
related
to
risk.
The
same
findings
were
observed
in
the
Belgian
Bank
study
with
men
aged
40-59,
followed
for
25
years.
There,
blood
pressure
and
cigarette
smoking
were
pre-
dictive
of
all-cause
mortality,
but
serum
cholesterol,
was
so
only
for
coronary
deaths
(15).
Also
in
a
study
on
a
special
group,
the
Seventh-Day
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
401
A.
Menotti,
M.
Lanti,
G.
Maiani,
et
al.
Aging
Clin
Exp
Res
18:
394-406,
2006
©2006,
Editrice
Kurtis
Adventist
men,
cigarette
smoking
and
high
blood
pressure
were
directly
associated
with all-cause
mortality
but
not
BMI
or
serum
cholesterol,
whose
associations
with
the
risk
of
death
were
rather
complex
(16).
An
analysis
conducted
in
the
Framingham
Study
iden-
tified
factors
associated
with
reaching
75
years
of
age
in
men
originally
aged
50.
It
was
found
that
lower
blood
pressure
and
fewer
cigarettes
smoked,
but
not
BMI
or
cholesterol
levels,
were
associated
with
longevity,
not
completely
matching
our
findings
(17).
In
the
25-year
follow-up
analysis
of
all
the
cohorts
of
the
Seven
Countries
Study,
age,
cigarette
smoking,
systolic
blood
pressure
and
serum
cholesterol
were
risk
factors
for
all-cause
mortality,
whereas
BMI
and
job-related
physical
activity
were
protective
factors
(18).
A
study
of
middle-aged
male
twins
in
the
United
States
showed
that
a
more
favorable
cardiovascular
disease
risk
factor
profile,
including
longevity
of
parents,
was
associ-
ated
with
a
lower
risk
of
death
in
a
25-year
follow-up
(19).
A
study
somewhat
similar
to
our
own
was
conducted
on
a
sample
of
elderly
Japanese
men
and
women
(age
65
years
+)
in
which
30
risk
factors
were
available.
The
authors
stressed
the
role
of
health
behavior
and
social
role,
but
found
that,
among
men,
age,
low
serum
albu-
min,
high
diastolic
blood
pressure
and
ECG
abnormal-
ities
were
the
only
predictive
factors
for
12-year
follow-
up
all-cause
mortality
(20).
There
is
at
least
one
other
contribution
dealing
with
the
estimated
expectancy
of
life
as
a
function
of
risk
factor
modifications
(21).
Other
studies
or
analyses
have
focused
on
single
risk
factors
in
greater
detail.
In
cohorts
studied
in
France
and
Northern
Ireland,
parental
longevity
was
inversely
related
to
the
incidence
of
major
cardiovascular
events
(22).
The
fundamental
role
of
smoking
in
influencing
ex-
pectancy
of
life
was
clearly
shown
also
by
a
re-analysis
of
the
British
Doctors
Study
(23).
In
the
35-year
follow-up
of
the
Finnish
cohorts
of
the
Seven
Countries
Study,
smok-
ing
habits
were
closely
associated
with
all-cause
mortali-
ty
(24).
In
a
special
study
dealing
with
centenarians
living
in
Rome,
Italy,
83%
had
never
been
smokers
and
14%
were
former
smokers,
suggesting
that
the
smoking
habit
is
incompatible
with
longevity
(25).
High
levels
of
leisure-time
physical
activity
were
in-
versely
related
to
all-cause
mortality
in
several
cohort
studies
dealing
with
middle-aged
men
(26-28).
The
beneficial
role
of
physical
activity
has
also
been
found
in
special
groups,
such
as
diabetic
patients
(29)
and
Sev-
enth-Day
Adventist
men
(16).
In
a
recent
analysis
deal-
ing
with
the
long-term
follow-up
of
the
LRC
Study
in
the
United
States,
fatness
and
unfitness
were
both
re-
lated
to
all-cause
mortality
in
both
sexes
(9).
An
ex-
tensive
review
of
the
relationship
between
physical
activity
and
health
has
confirmed
its
favorable
role
for
survival
and
the
limitation
or
delay
of
specific
morbid
conditions
(30).
All
this
agrees
with
our
findings,
al-
though
our
subjects
were
classified
for
their
working
physical
activity.
However,
it
should
be
noted
that
high
levels
of
leisure-time
physical
activity
were
rare
and
not
fashionable
in
the
early
1960s.
The
uncertain
role
of
BMI
and
other
indices
of
obesi-
ty
in
predicting
all-cause
mortality
is
revealed
in
other
stud-
ies
(31-33).
Some
of
the
problems
have
been
discussed
above
or
are
linked
to
the
presence
of
parabolic
rela-
tionships,
and
some
to
the
possible
confounding
effect
of
prevalent
diseases,
smoking
habits,
and
other
risk
factors.
The
direct
association
of
blood
pressure
level
with
all-cause
mortality,
at
least
until
the
age
of
60
(the
time
of
baseline
measurement)
was
clearly
shown
in
a
large
Nor-
wegian
study
with
20-year
mortality
data
(34).
The
direct
association
of
blood
pressure
level
with
all
kinds
of
vascular
disease
mortality
until
the
age
of
80
(the
time
of
baseline
measurement)
was
recently
confirmed
in
a
large
meta-
analysis
involving
almost
1
million
persons
(35).
Among
special
analyses
on
the
role
of
serum
choles-
terol,
one
conducted
in
Framingham
showed
that
mea-
surements
taken
at
age
40
are
directly
associated
with
all-
cause
mortality;
those
taken
at
age
80
are
inversely
related
to
all-cause
mortality;
and
those
taken
at
years
50,
60
or
70
show
no
association
(36).
In
the
MRFIT
primary
screenees,
with
a
12-year
follow-up,
the
predictive
role
of
serum
cholesterol
with
all-cause
mortality
was
weak
or
un-
certain,
due
to
the
variation
in
direction,
strength,
degree
and
persistence,
with
respect
to
various
causes
of
death
(37).
Similar
findings
were
reported
from
the
White-Hall
study
in
the
UK
(38).
The
literature
on
the
subject
raises
issues
of
the
socio-
economic
causes
of
total
mortality
and
specific
mortality.
Low
income
and
poor
education
are
factors
frequently
identified
as
predictive
of
higher
death
rates
(39,
40),
al-
though
in
some
circumstances
adjustment
for
the
levels
of
traditional
cardiovascular
risk
factors
(mainly
smoking
habits)
largely
reduces
the
relative
risk.
In
our
case,
the
so-
cio-economic
classes
roughly
corresponded
to
the
three
levels
of
physical
activity
and,
from
this
point
of
view,
we
may
conclude
that,
in
the
specific
location
and
time,
the
higher
the
class,
the
higher
the
mortality.
Overall,
the
identification
of
a
relatively
small
number
of
risk
factors
is
relevant
to
long-term
survival
and
the
mor-
tality
perspective.
They
are
easy
to
measure
and,
proba-
bly
by
proper
intervention
on
modifiable
factors,
the
ex-
pectancy
of
life
can
be
raised.
CONCLUSIONS
A
few
personal
characteristics,
partly
already
identified
as
cardiovascular
risk
factors,
measured
once
in
middle
age,
are
strongly
associated
with
long-term
mortality
and
survival
during
a
time
period
covering
almost
all
the
re-
maining
life-span.
402
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
Aging
Clin
Exp
Res
18:
394-406,
2006
Determinants
of
all-cause
mortality
in
40
years
©2006,
Editrice
Kurtis
This
does
not
state
for
how
long
each
risk
factor
is
pre-
dictive,
but
simply
that,
within
40
years,
a
group
of
risk
factors
is
associated
with
death
and
survival.
Only
complex
and
bulky
partitioned
analyses,
which
may
deserve
a
separate
analysis,
can
establish
for
how
long
and
at
which
strength
each
risk
factor
is
predic-
tive.
Analyses
of
this
type
have
already
been
published
on
the
same
material
for
shorter
follow-up
periods,
specific
end-points
and
few
risk
factors
(41).
Similarly,
the
association
of
risk
factors
with
mortality
may
also
have
been
influenced
by
long-term
changes
in
risk
factor
levels,
the
general
trends
for
some
of
which
have
partly
been
documented
(42).
Again,
complex
and
dedicated
analyses
are
needed
for
the
purpose.
APPENDIX
Risk
factors
measured
at
entry.
Definitions,
units
of
measurement,
mean
levels
and
use
in
analyses.
Risk
factor
Definition
or
details
Unit
of
Mean
and
(SD)
measurement
or
Proportion
%
and
(SE)
Values
in
univariate
analysis
(low-high)
(0-1)
Values
in
multivariate
analysis
Age
Approximated
to
the
nearest
birthday
Years
49.8
(5.1)
0=<50
years
1=50+
years
Continuous
Father
history
Father
died
<65
years
for
0=no
21.1%
(0.99)
0=no 0=no
non-infectious
or
non-violent
causes
1=yes
1
=yes
1=yes
Mother
history
Mother
died
<65
years
for
0=no
20.1%
(0.98)
0=no 0=no
non-infectious
or
non-violent
causes
1=yes
1=yes
1=yes
Family
history
of
cardiovascular
diseases
History
of
myocardial
infarction,
stroke
or
other
defined
cardiovascular
diseases
in
0=no
1=yes
37.9%
(1.17)
0=no
1=yes
0=no
1=yes
1st-degree
siblings
Marital
status
Presently
married
0=no
90.4%
(0.71)
0=no 0=no
(first
marriage)
1=yes
1=yes
1=yes
Children
Has
children
0=no
87.6%
(0.80)
0=no 0=no
1=yes
1=yes
1=yes
Physical
activity
Job-related,
derived
from
questions
matched
with
reported
occupation
1=sedentary
2=moderate
3=heavy
2.59
(1.60)
0=non
heavy
1=heavy
Discrete,
3
levels
Cigarette
smoking
Current
cigarette
smoking
derived
from
ad-hoc
questionnaire
N
per
day
on
average
8.71
(9.51)
0=non
smoker
1=smoker
Continuous
Diet Diet
prescribed
by
a
doctor
0=no
13.3%
(0.82)
0=no 0=no
for
any
possible
condition
and
followed
by
subject
1=yes
1=yes
1=yes
Height
Without
shoes.
Cm
166.3
(6.4)
0=<167
cm
Continuous
Method
as
from
(3)
1=167+
cm
Sitting
height
Method
as
from
(3)
Cm
87.5
(9.1)
0=<88
cm
Continuous
1=88+
cm
Weight
Without
shoes,
light
undergarment.
kg
69.8
(12.1)
0=<69
kg
1=69+
kg
Continuous
Method
as
from
(3)
Body
mass
index
Weight/height
squared
kg/m
2
25.2
(3.7)
0=<25
units
Continuous
1=25+
units
Trunk/height
ratio
(sitting
height/height)`
100
ratio
53.3
(1.5)
0=<53
units
Continuous
1=53+
units
Biacromial
diameter
Obstetric
calliper.
cm
39.1
(2.1)
0=<40
cm
Continuous
Between
the
2
acromions
1=40+
cm
Bicrystal
diameter
Obstetric
calliper.
cm
28.8
(1.8)
0=<29
cm
Continuous
Between
2
pelvic
crystal
spines
1=29+
cm
Shoulder/pelvis
shape
(ratio)
Biacromial
diameter/
bicrystal
diameter
ratio
1.35
(0.12)
0=<1.36
units
1=1.36+
units
Continuous
Laterality-linearity
index
(Sum
of
2
diameters/height)`100
formula
40.9
(1.8)
0=<41
units
Continuous
1=41+
units
Continued
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
403
A.
Menotti,
M.
Lanti,
G.
Maiani,
et
al.
Aging
Clin
Exp
Res
18:
394-406,
2006
©2006,
Editrice
Kurtis
APPENDIX
-
Continued.
Risk
factor
Definition
Unit
of
Mean
and
(SD)
Values
in
Values
in
or
details
measurement
or
Proportion
univariate
analysis
multivariate
%
and
(SE)
(low-high)
(0-1)
analysis
Harpenden
calliper.
Midway
mm
9.4
(5.4)
between
acromion
and
olecranon,
right
side.
Method
as
from
(3)
Harpenden
calliper.
Below
tip
mm
11.8
(5.8)
of
right
scapula.
Method
as
from
(3)
Right
arm.
Method
as
from
(3).
mm
268.6
(23.7)
Mathematically
cleaned
from
skin
and
subcutaneous
tissue
using
the
value
of
tricipital
skinf
old
thickness
Supine.
Average
of
2
measurements.
Method
as
from
(3)
Supine.
5th
phase
of
Korotkoff.
Average
of
2
measurements.
Method
as
from
(3)
Diastolic
blood
pressure+1/3
mmHg
104.9
(13.5)
of
pulse
pressure
From
ECG,
average
rate
in
lead
beats/minute
71.3
(12.9)
I
and
V6
Best
of
2
tests.
L/m
2
1.65
(0.24)
Adjusted
(divided)
for
height
2
Method
as
from
(3)
Best
of
2
tests.
L/m
2
1.10
(0.25)
Adjusted
(divided)
for
height
2
Method
as
from
(3)
Method
of
Abel-Kendal
modified
mmol/L
5.21
(1.07)
by
Anderson
and
Keys
(4).
Casual
Spot
urines.
Semiquantitative
0=no
7.8%
(0.65)
stix
method.
1=yes
Definitely
present
0=<9
mm
1=9+
mm
0=<11
mm
1=11+
mm
0=<269
mm
1=269+
mm
Continuous
Continuous
Continuous
0=<105
mmHg
1=105+
mmHg
0=<70
beats/minute
1=70+
beats/minute
0=<1.65
L/m
2
1=1.65+
L/m
2
0=<1.08
L/m
2
1=1.08+
L/m
2
0=<5.12
mmol/L
1=5.12+
mmol/L
0=no
1
=yes
mmHg
143.6
(21.0)
0=<140
mmHg
1=140+
mmHg
mmHg
85.5
(11.2)
0=<85
mmHg
1=85+
mmHg
Tricipital
skinf
old
Subscapular
skinf
old
Mid-arm
circumference
Systolic
blood
pressure
Diastolic
blood
pressure
Mean
blood
pressure
Heart
rate
Vital
capacity
Forced
expiratory
volume
in
3
/
4
sec
Serum
cholesterol
Urine
protein
Urine
glucose
Baldness
Comeal
arcus
Xanthelasma
Diagnosis
of
any
cardiovascular
disease
Diagnosis
of
cancer
Diagnosis
of
chronic
bronchitis
Diagnosis
of
diabetes
0=no
4.6%
(0.51)
1
=yes
0=no
29.4%
(1.10)
1
=yes
0=no
13.9%
(0.84)
1
=yes
0=no
1.6%
(0.31)
1
=yes
0=no
4.4%
(0.50)
1
=yes
0=no
0.3%
(0.13)
1=yes
0=no
6.3%
(0.59)
1=yes
0=no
4.8%
(0.52)
1=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
Spot
urines.
Semiquantitative
stix
method.
Definitely
present.
Bald
dome
or
receding
forehead+bald
spots
Clinical
judgment
Clinical
judgment
Any
clinically
overt
cardiovascular
disease
by
combining
history,
physical
examination,
ECG
findings.
Seven
Countries
criteria
(2)
Clinical
judgment
based
on
diagnosis
and
specific
treatment
Clinical
judgment,
based
on
ad
hoc
questionnaire
and
physical
examination
History
of
diabetes
or
use
of
anti-diabetic
diet
or
specific
drugs
or
glucose
definitely
present
in
urine
0=no
1
=yes
0=no
1
=yes
0=no
1=yes
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
Continued
404
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
Aging
Clin
Exp
Res
18:
394-406,
2006
Determinants
of
all-cause
mortality
in
40
years
©2006,
Editrice
Kurtis
APPENDIX
-
Continued.
Risk
factor
Definition
Unit
of
Mean
and
(SD)
Values
in
Values
in
or
details
measurement
or
Proportion
univariate
analysis
multivariate
%
and
(SE)
(low-high)
(0-1)
analysis
History
of
tuberculosis
History
of
bronchial
asthma
History
of
peptic
ulcer
History
of
intestinal
disease
History
of
liver
disease
History
of
gall
bladder
disease
History
of
kidney
stones
History
of
genito-urinary
disease
History
of
thyroid
disease
Minor
ECG
abnormalities
ECG
abnormalities
after
exercise
Clinical
judgment
Clinical
judgment
Clinical
judgment
Clinical
judgment
Clinical
judgment
Clinical
judgment
Clinical
judgment
Clinical
judgment
Clinical
judgment
From
Minnesota
code,
in
absence
of
clinical
diagnosis
of
heart
disease.
Code
11
in
(2)
From
Minnesota
Code.
Codes
11.1-2
or
12.1-2
or
13.1-2
or
14.1-2
or
15.1-2
(3)
2.1%
(0.35)
0=no 0=no
1
=yes
1
=yes
2.3%
(0.36)
0=no 0=no
1
=yes
1
=yes
11.0%
(0.76)
0=no 0=no
1
=yes
1
=yes
16.1%
(0.89)
0=no 0=no
1
=yes
1
=yes
3.9%
(0.47)
0=no 0=no
1
=yes
1
=yes
11.7%
(0.26)
0=no 0=no
1
=yes
1
=yes
5.7%
(0.56)
0=no 0=no
1
=yes
1
=yes
3.9%
(0.47)
0=no 0=no
1
=yes
1
=yes
0.2%
(0.10)
0=no
0=no
1
=yes
1
=yes
6.0%
(0.57)
0=no 0=no
1
=yes
1
=yes
4.3%
(0.49)
0=no 0=no
1
=yes
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
0=no
1
=yes
ACKNOWLEDGEMENTS
The
authors
thank
the
Istituto
Superiore
di
Sanit,
Roma,
Italy
(Na-
tional
Institute
of
Public
Health)
for
its
partial
contribution
in
collecting
mortality
data.
The
company
Medrisk
of
Roma,
Italy,
contributed
fi-
nancially
to
this
analysis.
REFERENCES
1.
Menotti
A,
Lanti
M,
Maiani
G,
Kromhout
D.
Forty-year
mortal-
ity
from
cardiovascular
diseases
and
their
risk
factors
in
men
of
the
Italian
Rural
areas
of
the
Seven
Countries
Study.
Acta
Cardiol
2005;
60:
521-31.
2.
Keys
A,
Blackburn
H,
Menotti
A,
et
al.
Coronary
heart
disease
in
seven
countries.
Circulation
1970;
41
(Suppl
1):
1-211.
3.
Rose
G,
Blackburn
H.
Cardiovascular
survey
methods.
Geneva,
World
Health
Organization,
1968.
4.
Anderson
JT,
Keys
A.
Cholesterol
in
serum
and
lipoprotein
fractions:
its
measurement
and
stability.
Clin
Chem
1956;
2:
145-
59.
5.
World
Health
Organization.
International
classification
of
dis-
eases
and
causes
of
death.
8th
Revision.
Geneva,
World
Health
Organization,
1965.
6.
Menotti
A,
Kromhout
D,
Blackburn
H,
Jacobs
D,
Lanti
M.
Forty-year
mortality
from
cardiovascular
diseases
and
all
causes
of
death
in
the
US
Railroad
cohort
of
the
Seven
Countries
Sudy.
Eur
J
Epidemiol
2004;
19:
417-24.
7.
Panagiotakos
DB,
Chrysoholou
C,
Pitsavos
C,
et
al.
Forty-years
(1961-2001)
of
all-cause
and
coronary
heart
disease
mortality
and
its
determinants:
the
Corfu
cohort
from
the
Seven
Countries
Study.
Int
J
Cardiol
2003;
90:
73-9.
8.
Murray
CJ,
Lopez
AD.
Alternative
projection
of
mortality
and
dis-
ability
by
cause
1990-2020:
global
burden
of
disease
study.
Lancet
1997;
349:
1498-504.
9.
Stevens
J,
Cai
J,
Evenson
KR,
Thomas
R.
Fitness
and
fatness
as
predictors
of
mortality
from
all
causes
and
from
cardiovascular
dis-
eases
in
men
and
women
in
the
lipid
Research
Clinics
Study.
Am
J
Epidemiol
2002;
156:
832-41.
10.
Su
D.
Body
mass
index
and
old-age
survival:
a
comparative
study
between
Union
Army
Records
and
the
NHANES-I
epi-
demiological
follow-up
sample.
Am
J
Hum
Biol
2005;
17:
341-
54.
11.
Kannel
WB,
Neaton
JD,
Wentworth
D,
et
al,
for
the
MRFIT
Re-
search
Group.
Overall
and
coronary
heart
disease
mortality
rates
in
relation
to
major
risk
factors
in
325,348
men
screened
for
the
MRFIT.
Am
Heart
J
1986;
112:
825-36.
12.
Stamler
J,
Dyer
AR,
Shekelle
RB,
Neaton
J,
Stamler
R.
Rela-
tionship
of
baseline
major
risk
factors
to
coronary
and
all-cause
mortality,
and
to
longevity:
findings
from
long-term
follow-up
of
Chicago
cohorts.
Cardiology
1993;
82:
191-222.
13.
Lowe
LP,
Greenland
Ph,
Ruth
KJ,
Dyer
AR,
Stamler
R,
Stamler
J.
Impact
of
major
cardiovascular
disease
risk
factors,
particularly
in
combination,
on
22-year
mortality
in
women
and
men.
Arch
Intern
Med
1998;
158:
2007-14.
14.
Hoes
AW,
Grobbee
DE,
Valkenburg
HA,
Lubsen
J,
Hofman
A.
Cardiovascular
risk
factors
and
all-cause
mortality:
a
12
year
fol-
Aging
Clin
Exp
Res,
Vol.
18,
No.
5
405
A.
Menotti,
M.
Lanti,
G.
Maiani,
et
al.
Aging
Clin
Exp
Res
18:
394-406,
2006
©2006,
Editrice
Kurtis
low-up
study
in
the
Netherlands.
Eur
J
Epidemiol
1993;
9:
285-92.
15.
Komitzer
M,
Dramaix
M,
Beriot
I,
Lannoy
M,
Gheyssens
H,
Kit-
tel
F.
Twenty-five-year
mortality
follow-up
in
the
Belgian
Bank
Study.
Cardiology
1993;
82:
153-71.
16.
Lindsted
KD,
Tonstad
S,
Kuzma
JW.
Self-report
of
physical
ac-
tivity
and
patterns
of
mortality
in
Seventh-Day
Adventist
men.
J
Clin
Epidemiol
1991;
44:
355-64.
17.
Goldberg
RJ,
Larson
M,
Levy
D.
Factors
associated
with
survival
to
75
years
of
age
in
middle-aged
men
and
women.
Arch
Intern
Med
1996;
156:
505-9.
18.
Menotti
A,
Blackburn
H,
Kromhout
D,
Nissinen
A,
Adachi
H,
Lanti
M.
Cardiovascular
risk
factors
as
determinants
of
25-year
all-
cause
mortality
in
the
seven
countries
study.
Eur
J
Epidemiol
2001;
17:
337-46.
19.
Reed
T,
Carmelli
D,
Robinson
TS,
Rinehart
SA,
Williams
CJ.
More
favorable
midlife
cardiovascular
risk
factor
levels
in
male
twins
and
mortality
after
25
years
of
follow-up
is
related
to
longevity
of
their
parents.
J
Gerontol
2003;
58:
367-71.
20.
Goto
A,
Yasumura
S,
Nishise
Y,
Sakihara
S.
Association
of
health
behavior
and
social
role
with
total
mortality
among
Japanese
elders
in
Okinawa,
Japan.
Aging
Clin
Exp
Res
2003;
15:
443-50.
21.
Tsevat
J,
Weinstein
MC,
Williams
LM,
Tosteson
AN,
Goldman
L.
Expected
gains
in
life
expectancy
from
various
coronary
risk
factor
modifications.
Circulation
1991;
84:
1194-201.
22.
Yarnell
J,
Yu
S,
Patterson
C,
et
al.
Family
history,
longevity,
and
risk
of
coronary
heart
disease:
the
PRIME
study.
Int
J
Epidemiol
2003;
32:
71-7.
23.
West RR.
Smoking:
its
influence
on
survival
and
causes
of
death.
J
R
Coll
Physicians
Lond
1992;
26:
357-66.
24.
Quiao
Q,
Tervahauta
M.
Nissinen
A,
Tuomilehto
J.
Mortality
from
all
causes
and
from
coronary
heart
disease
related
to
smoking
and
changes
in
smoking
during
a
35-year
follow-up
of
middle-aged
Finnish
men.
Eur
Heart
J
2000;
21:
1621-6.
25.
Tafaro
L,
Cicconetti
P,
Tedeschi
G,
et
al.
Smoking
and
longevi-
ty:
an
incompatible
binomial?
Arch
Gerontol
Geriatr
2004;
(Suppl):
425-30.
26.
Sandvik
L,
Erikssen
J,
Thaulow
E,
Erikssen
G,
Mundal
R,
Rodahl
K.
Physical
fitness
as
a
predictor
of
mortality
among
healthy
mid-
dle-aged
Norwegian
men.
N
Engl
J
Med
1993;
328:
533-7.
27.
Leon
AS,
Connett
J.
Physical
activity
and
10.5
year
mortality
in
the
Multiple
Risk
Factor
Intervention
Trial
(MRFM.
Int
J
Epidemiol
1991;
20:
690-7.
28.
Paffenbamer
RS
Jr,
Jung
DL,
Leung
RW,
Hyde
RT.
Physical
ac-
tivity
and
hypertension:
an
epidemiological
view.
Ann
Med
1991;
23:
319-27.
29.
Ford
ES,
DeStefano
F.
Risk
factors
for
mortality
from
all
causes
and
from
heart
disease
among
persons
with
diabetes.
Findings
from
the
National
Health
and
Nutrition
Examination
Survey
I
Epidemiologic
follow-up
Study.
Am
J
Epidemiol
1991;
133:
1220-30.
30.
US
Department
of
Health
and
Human
Services.
Physical
Activ-
ity
and
Health.
A
Report
for
the
Surgeon
General.
US
Dept
of
Health
and
Human
Services,
Centers
for
Disease
Control
and
Pre-
vention.
Atlanta
GA,
National
Center
for
Chronic
Disease
Pre-
vention
and
Health
Promotion,
1996.
31.
Keys
A.
Longevity
of
man:
relative
weight
and
fatness
in
middle
age.
Ann
Med
1989;
21:
163-8.
32.
Menotti
A,
Descovich
GC,
Lanti
M,
Spagnolo
A,
Dormi
A,
Sec-
carecda
F.
Indexes
of
obesity
and
all-causes
mortality
in
Italian
epi-
demiological
data.
Prey
Med
1993;
22:
293-303.
33.
Seccarecda
F,
Lanti
M,
Menotti,
Scanga
M.
Role
of
body
mass
in-
dex
in
the
prediction
of
all
cause
mortality
in
over
62,000
men
and
women.
The
Italian
RIFLE
Pooling
Project.
J
Epidemiol
Comm
Health
1998;
52:
20-6.
34.
Selmer
R.
Blood
pressure
and
twenty-year
mortality
in
the
city
of
Bergen,
Norway.
Am
J
Epidemiol
1992;
136:
428-40.
35.
Prospective
Studies
Collaboration.
Age-specific
relevance
of
usu-
al
blood
pressure
to
vascular
mortality:
a
meta-analysis
of
individual
data
for
one
million
adults
in
61
prospective
studies.
Lancet
2002;
360:
1903-13.
36.
Kronmal
RA,
Cain
KC,
Ye
Z,
Omenn
GS.
Total
serum
cholesterol
levels
and
mortality
risk
as
a
function
of
age.
A
report
based
on
the
Framingham
data.
Arch
Intern
Med
1993;
153:
1065-73.
37.
Neaton
JD,
Blackburn
H,
Jacobs
D,
et
al.
Serum
cholesterol
lev-
el
and
mortality
findings
for
men
screened
in
the
Multiple
Risk
Fac-
tors
Intervention
trial.
MRFIT
Research
Group.
Arch
Intern
Med
1992;
152:
1490-500.
38.
Smith
GD,
Shipley
MJ,
Marmot
MG,
Rose
G.
Plasma
cholesterol
concentration
and
mortality.
The
Whitehall
Study.
JAMA
1992;
267:
70-6.
39.
Clay
CM,
Dyer
AR,
Liu
K,
et
al.
Education,
smoking
and
non-car-
diovascular
mortality:
findings
in
three
Chicago
epidemiological
studies.
Int
J
Epidemiol
1988;
17:
341-7.
40.
Davey-Smith
G,
Neaton
JD,
Wentworth
D,
Starnler
R,
Starnler
J.
Socioeconomic
differential
in
mortality
risk
among
men
screened
for
the
Multiple
Risk
Factor
Intervention
trial:
I.
white
men.
Am
J
Public
Health
1996;
86:
486-96.
41.
Menotti
A,
Lanti
M.
Coronary
risk
factors
predicting
early
and
late
coronary
deaths.
Heart
2003;
89:
19-24.
42.
Lanti
M,
Menotti
A,
Nedeljkovic
S,
Nissinen
A,
Kafatos
A,
Kromhout
D.
Long
term
time
trends
in
major
cardiovascular
risk
factors
in
cohorts
of
aging
men
in
the
European
cohorts
of
the
Seven
Countries
Study.
Aging
Clin
Exp
Res
2005;
17:
306-15.
406
Aging
Clin
Exp
Res,
Vol.
18,
No.
5