J Am
Body
Size
and
GARY FRANK
Risk
of Kidney
C. CURHAN,*tII E. SPEIZER,*t
*Depart,nent Epidemiology,
and
Department
Channing
Laboratory,
Environmental
of Medicine,
Health,
Massachusetts
1998
Stones
WALTER C. WILLETT,*tt and MEIR J. STAMPFER*t
of Medicine,
Soc Nephrol 9: 1645-1652,
Brigham
Harvard
ERIC
Hospital;
Women
and
School
General
B. RIMM,*t ‘s Hospital;
of Public
Health;
and
Harvard
disease
Departments and
the
Medical
of Nutrition,
1Renal
School,
Unit,
Boston,
Massachusetts.
Abstract. cium
A variety
oxalate
excretion factors
be
a larger
stone
body
risk
factor
body
the
of kidney
risk
cohorts:
the the
exposures
during
during
horts,
the
Kidney calcium ily
diet
of
disease
vast
(4,5),
and
was
incident and
Several
).
oxalate
daily
stones,
urine
oxalate,
is derived
endogenous
acid,
both
from
production.
oxalate
generation
have
not been
it was
suggested
that
larger
is associated
with
higher
urinary
sumably due to increased endogenous Men, who on average have a larger have a lifetime risk of women. Therefore, body for
calcium
oxalate
stone
the
oxalate
in the contri-
it is likely endogenous endogenous Recently, lean excretion,
pre-
oxalate production (7). body size than women,
although
risk
to
decreases
the risk
this
found
the
varies
determine
but
These
results
risk
of stone
by
gender.
of stone
risks
for
for women associated
not
associated
suggest
that
formation
and
Additional
body
that
the
studies
a reduction
formation,
I .76
confiin men.
relative
was
mass
was
(95%
comparison
disease
whether
published
relation
data
in
body
particularly
are
weight
in women.
and
sparse.
Most
in weight
case-control
between
cases
studies
have
and controls
(8,9),
study that found a significantly compared with male control sub-
(10).
examine
between
and
cohorts:
the
Health
this
body
[BMI])
the
between
completed Nurses’ means
and
dietary
women
initial cohort
association mass
index
in two women)
=
5 1,529
121,700
female
(ti
large and
the
men).
The
Study male
first
dentists,
inquire
as well
population
for the current either
the
about
as the 1992
a history of kidney questionnaire since
the by
lifestyle
incidence
of
analysis or
the
was 1994
stones, and 1980 when
collected.
Follow-Up (HPFS)
which
states
and constitute cohort is followed
of interest, answered
registered
in one of I I U.S.
questionnaire (1 1). The
questionnaires,
who
was
Professionals 5 1,529
89,376
=
1976,
which inquired about at least one dietary
information
als Follow-up
In
exposures disease.
to the
Health among
mailed
other
questionnaires, who completed
formation
of 30 and 55 yr living
returned the Study (NHS)
diagnosed
the body
Methods Study.
of biennial
practices
studied and
stone Health Study (n Follow-up Study
the ages
and Health
we
weight,
of kidney
Nurses’
and
further,
(height,
risk
Study Population Nurses’ Health nurses
issue
size
Professionals
limited Received December 8, 1997. Accepted March 4, 1998. Correspondence to Dr. Gary C. Curhan, Channing Laboratory, Brigham Women’s Hospital, 181 Longwood Avenue. Boston, MA 021 15.
are
no difference
with the exception of one higher weight in male cases
newly
1046-6673/0909l645$03.00/0 Journal of the American Society of Nephrology Copyright 0 1998 by the American Society of Nephrology
of
Materials
body
stone formation three times that of size may be an important risk factor formation,
with
1 .38
1 .89 (1 .5 1 to 2.36) Height was inversely
stone
age-
body
kg/m2
but
the multivariate
formation.
and
the
with
21 to 22.9
to 2.07),
were in men. of
necessary
To
relative
formation,
stone
women
for the same
.
is associated
jects
concentra-
defined. (possibly
fam-
(1).
controversial, results from that influence size
as the
(2,3),
ingested
with
weight
Specifically,
for
1 .50
1 16 to 1.65)
prevalence
with
of the associations
women.
ratio
compared
the similar comparisons and 1.19 (0.83 to 1.70)
on
condi-
usually influence
oxalate
completely
body
costly
among
interval,
stone
associated
the magnitude
mci-
the
urinary
interval,
For incident
magnitude
stones
and
citrate
Although
bution of these two sources remains that at least half of urinary oxalate production (6). However, the factors
mass)
and
dence
directly
odds
kg/m2
size
cases in In both co-
diet
confidence
by mailed
of 956
factors
32
(95%
the
a total
and
index
incident
of kidney
greater
prevalence
with
=
were
However,
consistently
with
stone
calcium,
and
n
index.
forma-
kidney
including
large
women)
(HPFS;
history
volume,
uric
89,376
=
and
in two
confirmed.
important
oxalate
of
index)
cases
contain
adjusted
risk
Study
painful,
of stones
was
average
association
obtained
disease
is a common,
1
(
were
a stone
majority
of calcium,
Urinary
1078
8 yr of follow-up
of calcium
history
tions
of
prevalence
oxalate
formation
size,
14 yr of follow-up
stone The
body
of interest
A total
HPFS
tion.
on
The
mass
n
(NHS;
of these on
studied
Follow-up
mass
lifetime
body
was
of stone
body
be an important
formation.
Study
Information
questionnaires. in NHS
Health
men
may
and
dence
and
Several
higher
formation
Professionals
other
acid.
size
weight,
stone
Nurses’
men).
and
body stone
(height,
diet,
Because
a threefold
oxalate
size
Health
5 1 ,529 tion,
and
women,
for calcium
between
and
size
uric
size.
of calurinary
formation
gender,
and
to body
than
the
including
oxalate,
related
formation
influence
stones,
of calcium, may
have
of factors
kidney
Study.
is a longitudinal optometrists,
The Health study
of diet
osteopaths,
Professionand disease pharmacists,
Journal
1646
podiatrists. participants
of the
American
and veterinarians
who
returned
mation
on
followed
diet,
medical
by means lifestyle
about
history, and
of newly
of kidney
were
of Nephrology
in 1986
and
diagnosed
disease
infor-
The
questionnaires,
information
The
providing
medications.
mailed other
J Am Soc Nephrol
40 to 75 yr of age in 1986.
questionnaire
of biennial
practices
the incidence nosis
a mailed
Society
cohort
which
of interest,
including
is
inquire
as well
a physician
as
diag-
stones.
occurrence
and
symptoms.
random
sample
the diagnosis were
stones.
stones
each
biennial
weight
questionnaire.
(in kilograms)
validation
studies
self-reported
ments
of current
weights
(r
We
were
BMI
in NHS
highly
by
of the height
weight
0.96 and 0.97.
=
calculated
by the square
correlated
dividing
(12)
and
with
actual
In
HPFS
(13).
weight
answer
Assessment
of Diet
We asked complete
the NHS
participants
semiquantitative
in 1980,
1984,
food-frequency
1986,
and 1990 to
questionnaires
on
which
they reported the average use of specified foods and beverages during the past year. The HPFS participants completed the dietary questionnaires in 1986 and 1990. The 1980 NHS dietary questionnaire contamed
a list of 6 1 items,
cohorts
contained
and
the
approximately
subsequent
questionnaires
in both
130
We
nutrient
items.
computed
on the amount
carbonate)
ingested,
multimineral
of supplemental
both
as individual
preparations.
questionnaires
have
and HPFS
calcium
The
been
(such
supplements
reproducibility
documented
and
and
previously
as part
validity
in both
of (15)
adjusted regression
for total model
energy intake by taking with total caloric intake
the as
the independent variable and nutrient intake as the dependent variable (14,17). Energy-adjusted values thus reflect the nutrient composition of the diet independent of the total amount of food consumed. Follow-Up
and
Ascertainment
and
for
1986
were
analyzed
stone
disease
from the
were
the
Incident
goal
The
subjects’
reports,
of 90 of the confirmed HPFS. nosed were
women
who
stone
on
each
as those
that
baseline questionnaire. after the return of the
and 1986
was
reported
tary
form
The
response
medical
reported
records
from
a kidney
on a follow-up to confirm
the
biennial occurred
incident cases questionnaire. questionnaire. self-report
the
stone.
rate
The
or May 3 1 1994,
was
and updated
person-months return of the
status
by
of kidfor NHS
disease
and
prevalence
of
size
numbers
categories
of individuals
are the standard of BMI.
The
unadjusted using
study
tests
date-
Categorical
for trend
the Mantel
design
and
across
extension
for the incident size
in 1984,
1986,
and
of follow-up 1986 questionnaire
allocated
occurred
of interest
dietary
the
For
each
1990.
were calculated to the date occurred 1986 HPFS
person-months
at the start
first.
from
test
cases
was
intake
was
category
for
at the start
of a time
records
that
time
diag-
category
cases
parison
I 986
a supplemen-
follow-up
time
period.
period,
tension multiple
HPFS
then
on diet,
questionnaire participant, of the death,
first. Diet, weight, and other questionnaire were updated according
period
(e.g.,
to ex-
indicated
size information was assigned
If dietary
information
the subject
was excluded
by was to the was for
period. relative
risk-the
of exposure category-was
incidence divided used
test size
was
used
( 19). We risk factors
among
individuals
by the corresponding as the
adjusted relative risks were calculated 5-yr age categories for the respective body
of
that
1980
of follow-up
of each
missing
Information
from the date of a kidney stone,
to the
date
The
body
of the
of the
the
prevalent
cases.
the category of BMI and other variables). If body missing at the start of a time period, the individual
were those that occurred Whenever a kidney stone to ascertain
variables
missing
Prevalent return
other
sample
newly
(1980
of our analyses
whichever
.
weight,
The any
we mailed and
medical
a random
questionnaire. before
subject a to obtain
the validity
the diagnosis in all but one of the cases. The participants were asked to report
kidney defined
we obtained
if known.
the
collected before the onset of the kidney stone. For each NHS participant, person-months of follow-up were calculated from the date of the return of the 1980 questionnaire to the date of a kidney stone,
on the date
was 92%. To confirm
a history
questionnaire
on body
information
of stone,
subjects
to have
BMI
information
We
the type
Similarly,
examined
MA).
in 1990.
questionnaire
Only
using the test, and confidence interthe Robins-Breslow-Greenland variance
posure
and
prevalent
categories
was
supplementary
did not
with
sufficient
information supplementary conditions,
provide
participants
of having
3 1 , 1994, whichever of interest from the
relevant
to
they
period.
the incident
were calculated are two-tailed.
thus.
a participant
relevant
The
Cambridge,
Analyses.
for
imputed.
at baseline. The categories of height and the start of the analysis according to their
were performed calculated using
size categories All P values
was
who reported
dietary
was
BMI) before
not reported “missing”
or because
analyses.
considered
categories.
( 18) (StatXact,
death,
symptoms.
by the
stone.
from
or January variables
of occurrence,
Thus,
we use for the majority
comparisons vals were
the
the incident
baseline
with
prospective;
of Cases
the and
for
incident
separately
were labeled
the analyses of
Individuals
HPFS) at
for
failure
on the baseline
weight, and were selected
that
used
analyzed.
NHS. On the 1992 questionnaire. we inquired whether each woman had ever been diagnosed with a kidney stone and the date of the first occurrence. The 1994 biennial questionnaire asked about a new diagnosis of a kidney stone since 1992. Prevalent cases were defined as those that occurred before the return of the 1980 questionnaire. If a kidney stone reportedly occurred in 1980 (when dietary first collected) or later, we mailed questionnaire to confirm the diagnosis
confirmed
measurements
from
Analyses. disease
body (19).
3%
of self-reported
cases.
on a questionnaire,
was
after
ney stone
gories
(16).
Nutrient values were residuals from a linear
Prevalent
values
result
excluded
in the extreme
of the
NHS
were
distributions,
as calcium
confirmed
the remaining rate
to a category
questionnaire
event
censored
(height, weight
intake from the reported frequency of consumption of each specified unit of food or beverage and from published data on the nutrient content of the specified portions (14). Beginning in 1984, we collected information
disease
could
size
assigned
of the
biennial
stone
were
supplementary To confirm the records from a
records
all reports
to be true
body
or height
the
first
we assumed
were
None
values
their
respectively).
period
period.
Missing
measure-
The
of the cases;
the confirmation
so high,
for whom
a time
that
to the
1998
Analyses
Participants during
rate
cases.
in 97%
questionnaire
Statistical
the
(in meters).
stone
Because
was
supplementary
from the 1976 NHS and the on weight was obtained from
response
period was more than 92%. we obtained the medical
of 60 of the confirmed of a kidney
bladder
kidney
Assessment of Body Size We obtained information on height 1986 HPFS questionnaires. Information
The
questionnaire for each 2-yr validity of the self-report,
9: 1645-1652,
to evaluate
measure
in a particular rate
of association
in the com(20).
after stratification according cohorts (20). The Mantel linear
trends
across
used a proportional hazards model simultaneously (21). The selection
categories
Ageto exof
to adjust for of variables
J Am
Soc Nephrol
9: 1645-1652,
considered
in these
to calcium
stone
models
considered categories
in the models for NHS and
ries),
(seven
BMI
(five (0,
calcium,
animal
phosphorus.
P values
C (five
(eight
to be related
literature.
500+
potassium,
fiber.
D, and
we calculated
and
sodium,
vitamin
95%
vitamin
fluid
sucrose,
magnesium,
(quintile
groups).
For
(95%
Cl).
confidence
intervals
for
of
lence odds of kidney
to categories 1 . Of
the
question
about
lifetime
a history
of kidney
disease
decreased
89,376
with
in NHS weight,
women
history
stones with
increased
stones
of height,
in Table
and
who
before
stones,
1980.
The
height
increasing
BMI
responded
of kidney
increasing
at baseline
women
inches
shorter
were
1.96
respectively. to
The
189
pounds
than (1.13
weight
59
inches
to
3.38)
age-adjusted and
1. Frequency according
in men 21
(P. trend
and and
odds
190
to 234
of history
ratios pounds
of kidney
to categories
ratios
of stone
than