Risk with

Meir

of symptomatic severe obesity3

J Stampftr, K Malcolm Although
gallstones, the

gallstones
Maclure, Graham

in women
A Colditz, risk JoAnn E Manson, and Walter C Willett

ABSTRACT tor for

obesity
excess risks

is a well-recognized
associated

with

of obesity and recent weight change are poorly evaluated these issues in the Nurses Health 90 302 women aged 34-59 y at baseline followed
1988, occurred 2122 cases during of 607 newly 104 diagnosed person-years symptomatic of follow-up.

of weight change in a 2-y interval on risk of clinically fac- impact gallstones in the subsequent 2-y interval. Previous levels symptomatic an association between smoking and gallstones quantified. Wework suggested this issue in our large prospective cohort of Study. Among (4). We also address women. from 1980 to higher gallstones From 1980

to 1986, 488 cases of newly diagnosed unremoved were documented. We observed a striking monotonic
in gallstone disease risk > 45 kg/m2 with had obesity; a sevenfold

gallstones increase

Subjects The Nurses

and

methods Health Study cohort

Downloaded from www.ajcn.org by guest on August 25, 2011

The Nurses Health Study began in 1976 when 12 700 female 1 index (BMI) registered nurses living in 1 1 large US states completed a mailed with those whose BMI was 24 kg/m2. < Women with a BMI questionnaire that included items about their medical history > 30 kg/m2 had a yearly gallstone incidence > of1% and those for coronary disease and cancer10) including ( with a BMI 45 kg/m2 had a rate of 2%/y. Recent weight and risk factors height, weight, current and past smoking habits, use of postand loss was associated with a modestly increased risk after adjusthormones. Height and weight were used to compute ment for BMI before weight loss. Current smoking was an in- menopausal BMI, defined as weight in kilograms divided by height in meters dependent risk factor; women smoking 35 cigarettes/d had a to the second power, as an index ofadiposity. Every 2 y, followrelative risk of 1.5 (95% CI 1.2-1 .9). Am J Clin Nuir l992;55: up questionnaires were mailed to update the information on 652-8. potential risk factors and to identify newly diagnosed cases of various illnesses. In 1980 a semiquantitative food frequency KEY WORDS Cholecystectomy, galstones obesity, women, questionnaire was added (1 1). cigarette smoking, weight loss Identification Introduction
Obesity

women with a body mass excess risk compared

ofsymptomatic with

gallstones

the 1982 follow-up questionnaire, we asked the nurses to report whether they had had a cholecystectomy or had been given a diagnosis of gallstones without reThe magnitude ofthe increased risk and the rates occurrence of moval. We also inquired whether symptoms were present and, of symptomatic gallstones, however, have not been well quancholecystectomy, whether the diagnosis tified, particularly among the most obese, who are at highest for the cases without by x ray or ultrasound. In the 1988 questionnaire risk. Women who embark on rapid weight-loss programs using was confirmed about untreated gallstones was omitted but the item very-low-calorie diets are apparently at increased risk for lith- the question concerning cholecystectomy was retained. ogenesis (8, 9). However, because such women are generally We assessed the validity ofthe self-report of cholecystectomy obese, the excess risk due to the extreme diet is difficult to quangallstones in two random samples of 50 nurses tify because rates of symptomatic gallstones among the very and unremoved
is a well-established risk factor for gallstones (1-7). participating obese have not been published. In an earlier Nurses Health Study based on 4 y of follow-up, compared with women with in kg/m2) < 20, those women risk of symptomatic gallstones

Starting

a body mass index with a BMI of 32 had of 6.0 [95% confidence

report from theeach we found that (BMI measured

a relative
I

who

reported

these

conditions.

Of

43

who

responded

for

From

the Channing

Laboratory,

Department

ofMedicine,

Brigham

(CI) 4.0-9.0] and were attributable

follow-up to

that 70% to obesity

examine

8 y and

and Womens Hospital and Harvard Medical School, and the Departinterval ments ofEpidemiology and Nutrition, Harvard SchoolofPublic Health, of the gallstones in those women Boston. (1). In this report we extend the 2 Supported by research grants CA 40356 and DK36798 from the both the rates of occurrence and National Institutes 02115. of Health

the relative risks by level of obesity, paying particular attention 3 Address to the highest levels of overweight. Boston, MA Data on the effect of weight loss on gallstone disease in a Received general population are also sparse. Hence, we also assess the Accepted 652
Am J C/in Nuir l992;55:652-8. Printed

reprint requests May 20, 1991.

for publication in USA.

and by Sandoz to MJ Stampfer,

Pharmaceuticals. 180 Longwood Avenue,

August 1992

15, 1991. Society for Ginical Nutrition

American

GALLSTONES
cholecystectomy confirmed
=

IN
report,

WOMEN
and the this records (n confirmed

WITH

SEVERE
were updated

OBESITY
allocated according For recently with returned women were the for assigned completed diagnosis a follow-up was assigned the analysis. with associated according who to the I 980

653 exposure with status but or Thus, up-to-date with these given folIf no if weight we only

all in was all cases reviewed). reiterated 16 medical for analysis was

reiterated for For by records all which

the unremoved 35 we

initial we could

obtain and (1).

person-months was variables and gallstones, in reported low-up questionnaire folon was

to subsequent were according

follow-up-quesdiagnosed to the gallstones. cycle

36

records the

gallstones obtain

original tionnaire

information. person-months the most was and weight. was in the terminated

diagnosis
1 5 of

respondents could

questionnaire of

Population The low-up and who Because cluded before (1, disease. agnosed the

population questionnaires provided we were women return main with

limited in 1982 in

to women

who

returned follow-up current illness,

the

or in subsequent on height and newly diagnosed or a diagnosis

information interested ofthe focus 1980 was on

cholecystectomy questionnaire.

cycles category weight. considered we exof body of gallstones reports

was

not reported,

that
not

follow-up
included the change

time
in

to the

missing
reports weight analyses interval

analysis the

follow-up in risk

In our

earlier

We

also assessed
during report

12) the

dietarydeterminants

To

avoid

the
might

possibility
influence all that

that
the women they

of gallbladder symptomatic but undiofdiet, we excluded on the symptoms

change required

the preceding 2-y interval. Because ofweight both at the beginning ofa

gallstones

report had

as well as on the previous

of follow-up. Missing

questionnaire,
values resulted

there

from those plementary

1988) and 3.6% from the analysis ofdiagnosed but unremoved ofgallstones before return ofthe 1980 questionnaire even if the gallstones (1980-1986). diagnosis was made after the return ofthat questionnaire. In the We used the relative risk as the measure ofassociation, defined present analysis the focus is on incidence rates and relative risks as the incidence rate of gallstones or cholecystectomy among by categories of overweight; hence, we only excluded women women in various categories of BMI (estimated as the number

earlier analyses questionnaire

who reported experienced

years sup-

were fewer personin exclusion of 4.8% (1980-

of

person-years

from

the

analysis

of cholecystectomy

Downloaded from www.ajcn.org by guest on August 25, 2011

whose
study

diagnosis
avoid

of gallstones
earlier because

preceded

the

return

of the

1980

questionnaire. to by the subject alter her diet.

The bias

exclusions symptoms correctly or in the present

were

justified

in a dietary to gallstones

ofevents

divided

by

person-years

of follow-up This

in

that

BMI

catin the as

attributed

(whether However,

not) may have her led to analysis such an exclusion

egory) divided by reference category the comparison that the relative

the corresponding of 24 kg/m2. <

rate among women category was chosen analysis not

would
pecially notoriously unnecessary gallbladder

result
because

in

gross
the

underestimation
report the of symptoms such an

of incidence
due exclusion to

rates,
gallstones would symptoms body weight. consideration followthe occurred be

es-

group because in our previous risk for gallstones was essentially of24 kg/m2 to a weight 5-y

we found associated of average Age-specific individually

inaccurate because disease

( 1 3). Also,
presence be would

with BMI < 24 kg/m2. A BMI is height (165 cm) corresponds

in a woman of65.4 kg. were

ofprediagnostic to influence

unlikely

Even
in the

if it did,
analysis

such
because

a change
weight

would

be taken
every who which

into

(or BMI-specific) rates using of calculated and used to compute relative risks with 95% CIs (15). simultaneously, Pvalues are diby the direct

categories

age-adjusted To adjust for

(or BMI-adjusted) multiple risk factors used. All standardized entire cohort.

is updated

2 y on the reported usually

up questionnaires. agnosis ofgallstones

We excluded women without symptoms,

proportional-hazards models two-tailed. The incidence rates method to the age distribution

(16) were were age ofthe

as an incidental another indication.

follow-up and

We validated area participants

an examination or surgery for were excluded from subsequent Results were not counted as cases this analysis. in the self-report of weight in a sample of BostonDuring by comparing their self-reported weights with
values taken 6The correlation 1 2 mo after between the questionthe self-report we newly tended 1986,

finding Such

during women

607
diagnosed from for

104 person-years
2122 the cases return of 474 symptomatic of the 768

offollow-up
but 1980 unremoved questionnaire

from
The

1980

to

1988
for exJune during 1,

technician-measured naire was completed. and measured weight was

identified

of cholecystectomy.

follow-up gallstones until

values was = 0.97 ( 14). The mean r self-reported 1 .5 kg lower than the measured weight, compatible

a total

person-years

of

follow-up

with the difference between a casual weight with and a morning postvoid nude weight. In 1980, 90 302 women entered the analysis
1982 interval. During the 8 y of follow-up from the 1980 person-years symptomatic of 474 the 768 1980 questionnaire until June 1, 1988, of follow-up. The follow-up but person-years unremoved gallstones until of follow-up. June questionnaire

light for
the

clothing the 1980return

which we ascertained 488 cases. We observed a strong linear association

incidence tectomy of rate and of symptomatic gallstones unremoved stones) and age-standardized 95% CI for gallstones. from

between
(combining

BMI

and

the

cholecys-

607 104 relative diagnosed moved extended from the return tectomy 1, 1986, for a total of

we accrued for newly

number

ofcases

(Fig Table I). 1 shows the rates and the age-adjusted cholecystectomy The incidence 202/100 rate and for unreof cholecys-

risks with symptomatic ranged

eightfold

reference
in

category
the 45+

ofBMI
kg/m2

of24
category.

kg/rn2

Our procedures are Brigham and Womens

and the Data The months the Harvard School Declaration. Helsinki analysis primary of analysis follow-up

approved Hospital
ofPublic

by and

the ethics committees HarvardMedical School,

and are in accord

Health,

cm woman corresponds to a weight rates of unremoved symptomatic with in a similar manner with increasing

of

years

000 person-years in the 1622/100 to 000 person(BMI of 45 kg/m2 for a 165 kg.) are stable The incidence lower but rise ofthe than only of gallstones smaller those minor

of 1 22.5 gallstones BMI. points less

Because we observed and risk

number
is based as the on incidence For rates each using person- differences flecting

of cases

the

rates For

are both

statistically end

for cholecystectomy.

between
the weak

the
relation

crude
between

and

age-standardized
age

rates,

rein

denominator.

participant,

654
Inc#{232}dince Rat#{149}. f o Clinically Symptomatic (p.r 100.000 person-years) 260C Gallstones

STAMPFER

ET weight y interval

AL loss; slender (1 7). women Moreover, are in unlikely a previous to

>lose

10 kg in a 2of predictors

analysis

--

of weight change weight gain was

women

2000

women
weight founding
dardized

we found that the best predictor of 2-y future weight loss in the previous 2-y interval. Obese are also more likely to have substantial weight than gain of average relative weight. Hence, the extremes of both
and gain are enriched with obese women. in and with the was This conwhen differences with the agethe association For example group to the age-stanBMI-adjusted weight change rate for chole000 734.7/100

loss

1500

1000
E

+ +
*
c
(

relative apart cystectomy person-years This

I

is quite apparent rates are compared risks, which estimate from the in the obesity 10+ rate adjusting effect.

kg weight-gain

500

compared
after

with

298. 1 for those

is reduced for BMI. After

with
that

stable

weight.
4% inwe

2.5-fold-higher in risk

a nonsignificant adjustment

crease
-45 45+observed

moderate but significantly increased risks ofboth catof gallstones with loss of 4 kg in the previous 2 y. with a weightloss of 10 kg had a BMI-adjusted relative FIG 1 Incidence rates of cholecystectomy or newly diagnosed tomatic unremoved gallstones combined by categories ofbody mass index risk of cholecystectomy of 1 .99 (95% CI 1 .55-2.56) compared (BMI) in the Nurses Health Study. The vertical lines represent the95% with women with a weight change < of 4 kg and 1.97 (95% CI
24

-25

-26

27

-29

Body

Mass

Index

-30 -35 (kg/rn)

-40

egories symp- Women

confidence intervals around tribution. The upper bound 100,000 person-years.

the estimated rates, assuming a Poisson dis- 1.57-2.47) for the highest BMI category is 3016 per unremoved

for

the

combined

endpoint gallstones.

of cholecystectomy

plus

Downloaded from www.ajcn.org by guest on August 25, 2011

symptomatic

The analyses rates of clinically

change effect the plus 30 ofobesity was far more important

of weight change in Tables 2 and symptomatic gallstones according

for current attained

3 compare to weight
weight. that issue,

this

cohort

(1).

Thus,

the

However,
may

in the previous 2 y adjusting part ofthe increase in risk to their excess weight

among
2 y before.

those
To

who
assess

lost weight 2 y

than that Incidence symptomatic 1 . Women and women dence rate

of age. rates for gallstones

be due

combined endpoint cholecystectomy are kg/m2 category 2%/y. according had an annual of 45+

we compared of unremoved shown in Table before who

rate of> had an 1%, subsequent interval mcithe weight after

incidence rates for women had maintained their weights

2 y. The relative a 10-kg weight loss,

of similar weight or lost10 kg in the

BMI did

with

a BMI

in the highest that exceeded

kg/m2

risk of cholecystectomy loss, adjusting for 1 .67) 85-kg (Table 4). women

in the 2-y before not lose

Table
tomatic

2 shows
unremoved

the

incidence
gallstones

of cholecystectomy
to weight

and
change

symp- lack of increase in the weight (compared

catfinding attributable rates It is ofinterest andof symptomatic

was 1 .17 (0.82in risk in the

The
who

apparent likely a chance rates

be at-

with

the 65-kg the

category) extent
the

is most the

previous 2 y. The highest rates were found in the extreme egories of weight change. However, the age-standardized are confounded by the strong association between obesity

to small

numbers.

to assess
gallstones

to which
obese

elevated
may

among

women

TABLE

1
incidence rates unremoved gallstones per 100,000 person-years, in the Nurses Health and Study relative risks and 95% confidence intervals for cholecystectomy t and

Age-standardized symptomatic

Cholecystectomy Cholecystectomy Symp tomatic gallsto nes, not removed symptomatic

plus gallstones

BMI
<24 24-<25 25-<26 26-<27 27-cz29 29-cz3O 30-<35

Cases
642 196 181 148 266 121 384

Incidence rate
202 318 370 461 1 .0

Relative (95%

risk CI)

Cases
18 1 32 41 29 54 22 69

Incidence rate
75 71 112 122 145 1 .0

Relative (95%

risk CI)

Incidence rate
277 389 482 582 685 1 .0

Relative (95%

risk CI)

(referent)

(referent)

(referent) (1.24-1.66) 1 (1.80-2.47) (2.23-2.86)

1.57(1.34-1.84) 1.82(1.55-2.14) 2.25 2.56 2.87 3.98 (1.89-2.67) (2.23-2.94) (2.38-3.46) (3.54-4.47)

0.94(0.64-1.36) 1.48(1.05-2.07) 1.62 1.92 (1.10-2.38) (1.43-2.59)

1.43 2.1 2.53

1.74(1.51-2.02)

540
590 814

140
198

1.95
2.64

(1.26-3.01)
(2.03-3.45)

729
1012

2.67
3.69

(2.25-3.17)
(3.32-4.11)

35-<40 40-<45
45+
*

127 36
21

928 912
1622

4.58 4.46

(3.86-5.45) (3.29-6.06)

39 16
5

360 591
469

5.24 7.56

(3.85-7.13) (4.90-1 1.67)

1287 1504
2091

4.72
5.1 7.36

(4.06-5.49)
1 (3.97-6.56) (5.28-10.26)

7.56(5.23-10.92)

6.64(3.08-14.31)

Age-standardized

to the distribution

of the cohort,

1980-1988.

t Findings from 1980-1988. j: Follow-up from 1980-1986.

GALLSTONES

IN

WOMEN

WITH

SEVERE

OBESITY

655

TABLE
Weight

2 change

in previous

2 y5 and

risk

of cholecystectomy

and

symptomatic

unremoved

gallstonesf Cholecystectomy or unremoved gallstones Rate per 100 000 person-years

1099.1

Cholecystectomy Rate per 100 000 person-years

839.4

Sym ptomatic BMI-adjusted relative risk (95% CI)II

1.99 (1.55-2.56)

gallston es, not

removed

Weight
10+ kg

change
loss

Cases
63

Cases
14

Rate per 100 000 person-years

259.7

BMI-adjusted relative risk (95% CI)II

1.87 (1.10-3.19)

BMI-adjusted relative risk (95% CI)II

1.97 (1.57-2.47)

4.0-9.9 3.9 kg 3.9-9.9 10+ kg

a

kg loss gain or loss kg gain gain with from from

194 13 1 3 365 76 data

505.7 298. 1 466.7 734.7 are excluded.

1.38 (1.19-1.61) 1.0 (referent) 0.94 (0.83-1 .06) 1.04 (0.80-1.35)

58 3 11 74 22

197.8 92.7 126. 1 269.6

1.75 (1.33-2.32) 1.0 (referent) 0.87 (0.66-1 . I 3) 1.12 (0.69-1.81)

703.5 392.6 577.2 1004.3

1.45 (1.27-1.66) 1 .0 (referent) 0.92 (0.83-1.03) 1.06 (0.84-1.33)

Women

missing

t Follow-up

1980-1988. Follow-up 1980-1986. Age-standardized to the distribution II Adjusted for BMI at the beginning

of the cohort, 1980-1988. ofeach follow-up interval

for the

categories

listed

in Table

I using

stratified

analyses.

Downloaded from www.ajcn.org by guest on August 25, 2011

tributable to other gallstone with adiposity. Thus, for heavier (18).

gallstones

women Because
(1), part

tend alcohol

risk factors that might be associated cystectomy in the different categories of BMI and recent weight example, it is well established that change after simultaneous adjustment for known or suspected to drink less alcohol than leaner women gallstone risk factors: age, parity, postmenopausal hormone use, intake is associated with decreased risk of alcohol intake, polyunsaturated fatty acid intake (12), total en-

of the

effect

of obesity
heavier frequently the rates the the

may
women than in the association relative

reflect
tend leaner

decreased
to use women

alcohol intake. Conversely, menopausal hormones less

ergy, and post-adjusted (19).observed

smoking relative
in Table

(1), using risks for

1, suggesting

proportional BMI are

that

hazards

generally
only

models. These similar to those

portion of the

a small

Estrogen
confounding Also, loss these and

use increases
factor factors gallstones. may

the
may Table

risk ofgallstones
bias

(20),

so this

potential
direction. for

opposite risks

dramatic rise to confounding

with

in risk with increasing factors. Similarly,

weight loss were

adiposity the elevated

unchanged

can be attributed relative risks asin the multi-

confound 3 shows

between weight

sociated chole- variate

largely

analyses

that

adjusted

for current

attained

weight.

TABLE change

3 in the past
5 relative risks for cholecystectomy, symptomatic unremoved gallstones, 2 y, adjusting simultaneously for various risk factors for gallstonest

and

both

endpoints

together

by BMI

and

weight

Cholecystectomy

or

Cholecystectomy BMI <24

24-<25 25-<26 26-<27 27-<29 29-<30

Gallstones,

unremoved

unremoved

gallstones

1.0
1.48 1.61 2.00 2.38 2.82

(referent)
(1.24-1.77) (1.33-1.93) (1.63-2.46) (2.03-2.79) (2.27-3.51)

1.0
1.01 1.54 1.66 1.96 1.96

(referent)
(0.70-1.46) (1.10-2.15) (1.12-2.45) (1.44-2.65) (1.25-3.05)

I .0
1.36 1.60 1.92 2.32 2.63

(referent)
(1.16-1.60) (1.36-1.88) (1.60-2.30) (2.02-2.66) (2.16-3.19)

30-<35 35-<40
40-<45 45+

3.77 4.35
4.60 7.08

(3.26-4.37) (3.48-5.44)
(3.15-6.72) (4.22-1 1.86)

2.62 4.98
6.93 5.91

(1.96-3.49) (3.46-7. 17)

(4.07-1 1.81) (2.37-14.75)

3.52 4.64
5.42 6.99

(3.1 1-3.98) (3.86-5.57)

(4.01-7.34) (4.48-10.90)

Weight 10+ 4.0-9.9 3.9 4.0-9.9 10+

S

change kg loss kg loss kg gain or loss kg gain kg gain adjusted for BMI,

1.91 (1.43-2.54) 1.32 (1.1 1-1.56) 1.0 (referent) 0.90 (0.78-1.03) 0.95 (0.72-1.26) weight change, age (5-y categories),

2.03 (1.18-3.47) 1.91 (1.44-2.53) 1.0 (referent) 0.93 (0.71-1.21) 1.24 (0.78-1.98) alcohol intake (five categories),

1.94 1.44 I .0 0.90 1.01

(1.50-2.50) (1.25-1.67) (referent) (0.80-1.02) (0.80-1.28) hormone use (current,

Simultaneously

postmenopausal

past, never), smoking(never, past, and current fatty acid intake (five categories).

at four amounts

ofcigarettes perday),

parity(six

categories),

energy intake(five

categories),

polyunsaturated

95%

CI

in

parentheses.

656 TABLE 4

STAMPFER

ET

AL

The risk of cholecystectomy in relation prevous 2 y, adjusting for BMI before

to a 10-kg weigh in the t loss the weight loss* 2 y ago

85kg 75 75 85

Weight
65kg Current weight (kg) 20 2319 55 65 65 75kg

newly symptomatic gallstone disease, an entity ofdirect clinical and public health importance. The findings are based entirely on self-report of weight and the occurrence ofgallstone disease on the biennial questionnaires. The self-report of weight in this cohort is quite accurate when
compared (r = 0.97) 4
10

with (14).

measured weights By comparing

the

in a sample questionnaire

ofthe

participants reports with

Weight Cases

loss (kg)

10+
22

4
190 137 10 1048

10+
40 5341

4 10+
6 389

medical records in a sample of participants we also found that these registered nurses had a very high level of accuracy in reporting cholecystectomy and diagnostically confirmed but unremoved gallstones (1). Similar accuracy of reporting was dem-

Person-years

1412

Risk next year (%) Relative risk for 10 kg weight losst (96% CI)
a

0.862 1.00 (0.63-1.59)

0.858

0.954 1.27 (0.64-2.55)

0.749 1.544 2.18 (0.79-5.99)

0.708

Projected the

for a woman for

140 cm in height. 10 kg weight method. loss 1.17 =

(0.82-1.67), by

t Overall
using

relative risk Mantel-Haenszel

In this cohort
smokers

we observed

a higher

risk ofgallstones

for current

onstrated for several other diagnoses (24). Although we do not know the composition of the stones, we may estimate that 8590% were cholesterol stones (25, 26). Our results are consistent with the limited data from previous studies. The Rome Group (7) and Barbara et al (6) both found that BMI is a significant predictor ofgallstone prevalence among women in a screened population but provided no prevalence rates by level ofobesity. In a small screening study of249 obese and 60 nonobese women the prevalence of gallstones was 31% among the obese vs 10% among the controls (23). Jorgensen

Downloaded from www.ajcn.org by guest on August 25, 2011

(21) screened 1680 Danish women and provided gallstone prey(Table 5), particularly for heavy smokers. For women alence rates specific for categories of BMI. Women with a BMI smoking 35 cigarettes per day, the relative risk for symptomatic of 30 kg/m2 had a prevalence of 30% (41 of 135) as compared gallstones (cholecystectomy and unremoved) compared with that with 6% (13 of 231) among women with a BMI20 kg/m2. < for women who had never smoked was 1.5 1 (95% CI 1.20-1.89)

after

adjustment

for BMI,

weight

change,

and

other

risk

factors.

Although cidence
could

these numbers cannot rates for symptomatic

for example,

be compared directly gallstones, a prevalence

a net incidence

with inof 30%

of silent

result,

from

rate

formation of 2%/y followed by an average duration of gallstones of 17 y. Using cross-sectional data from a large In this large prospective cohort study we confirmed the im-cohort ofweight-conscious women, Bernstein et al (27) estimated incidence rates ofdiagnosed gallstones by using life-table methportance of obesity as a risk factor for gallstone disease and, ods. They found that women in the highest obesity category because of the size of the study, we could estimate rates for BMI 38.3) had an estimated incidence 2-6.25-fold clinically diagnosed gallstones and cholecystectomy at various (mean depending on age than the reference category (mean BMI BMIs. Rates of gallstone disease for obese people are known tohigher Discussion
be high defined but have population. and Substantial not previously The incidence been quantified rates increased in a large, monotonically well-23.0). These results are similar to our findings.

gallstone

with
35 kg/m2.

increasing
kg/m2

obesity,
exceeding weight

l%/y
2%/y loss

for women
for during women the

with
with preceding

Data from cohort studies are also sparse. a BMI of 30- Study (28) the risk of clinically diagnosed a BMI 45of higher than average for women with weight
2 y was also 20% higher than the cohort median (49

In the Framingham gallstones as 77% w (adjusted for height)

cases), but for those

a significant BMI, with

of loss
>

risk factor for gallstones, a relative risk of 1 .94 (95%

Comparing of the same the risk original among weight,

independent CI 1.50-2.50)
women the with impact

of current in the for a loss

that was weight much

10-19%

excess-weight

category

(15

cases)

the

risk

was

10 kg. to those

less

with

a nonsignificant

relative

risk

of

1. 17 (0.82-1.67).

TABLES WeMultivariate-adjusted5

also confirmed smoking and

21,

an independent gallstone disease, in this techniques

association as others study for

between cigarette symptomatic unremoved have observed (4, cigarette smoking

relative risks gallstones,

for cholecystectomy, and both endpoints

together

by

22). It was ultrasound

not

possible or other unremoved

to screen participants with the presence of gallstones.

Unremoved,

symptomatic

Cholecystectomy
1.0
1.06

gallstones

Both

endpoints

Hence,
firmed that there

we focused
but was

on cholecystectomy
symptomatic underascertainment of prevalent For example, of gallstones symptomatic that in many

and

diagnostically
We of gallstones be-

conrecognize
Never
Past Current

gallstones. stones Jorgensen were

(referent)

1.0
1.09

(referent)

1.0
1.06

(referent)

substantial proportion (6, 1 3, 23). or more to distinguish it is likely

(0.94-1.18)

(0.89-1.33)

(0.96-1.17)

cause a large asymptomatic that two-thirds that it is difficult stones. Moreover, during the asymptomatic the incidence

are completely (13) observed

1-14 cigarettes/d 15-24 cigarettes/d and 25-34

35+ cigarettes/d cigarettes/d

1.13 (0.94-1.36)
1.21 (1.04-1.42) 0.93

1.06 (0.74-1.50) 1.10 (0.93-1.29)

(0.68-1.29) 1.03 (0.90-1.19)

asymptomatic

1.36 (1.1 1-1.67)

1.59 (1.24-2.05)
adjusted

1.21 (0.81-1.31)
1.30 (0.78-2.16)

1.31 (1.09-1.58)
1.51 (1.20-1.89)
categories), alcohol smoking

from asymptomatic ofthe cases diagnosed were attempted rather the

follow-up interval, at baseline. We of gallstone formation

gallstones have not but

present but intake (five categories), to estimate (never, past, and current incidence
energy of intake (five

Simultaneously

for BMI,

weight hormone

change, age(5-y
use (current, per day),

poatmenopausal

past, never),
parity

at four amounts ofcigarettes

polyunsaturated

categories),

fatty

acid

intake

(sixcategories), (five categories).

GALLSTONES
36% tending women years lower cases) lower than average. In a large clinic crude in during rates steadily 30 kg/m2 cohort Oxford, nearly in that from (22 study

IN
of UK, 1 55

WOMEN
women atthe

WITH
loss.

SEVERE
Also, part

OBESITY
of the risk associated with weight

657
loss may

a family-planning had a cholcystectomy of follow-up than for

<

227 young be attributable to the tendency 000 person- given 2-y interval to gainweight were(17). (25000 However, The impact of obesity on

of women who in the subsequent

risk is far stronger

lost weight in a 2-y interval

than that of

(22). observed and kg/rn2

The

population in differences part 96/100 cases).

we 20

(perhaps increased for to 626

reflecting

in surgical the relative given the yielded did and not

practice) risks modest

weight loss in our cohort. Although women on a very-low-calorie diet appear to be at increased risk for gallstones, much of the excess risk is probably attributable to the underlying obesity. If
sustained level risk term of the and for the expert assistance ofKaren Corsano, Bechtel, and Debbie OSullivan. The Nurses general direction of its principal investigator, in of the impact weight attained ofweight reduction weight, loss eventually as could one decreases might expect, the the in the risk net to the long-

were similar to what we observed, particularly number of cases the Oxford in study, which rates. a relative obesity ofbile They until risk found, a BMI of6.5 of24 was as we did, kgjm2 observed is explained in obese people that was for the exceeded, BMI by increase

fairly materially

unstable study

be a reduction

incidence

of gallstones

despite

a transient

increase

in short-term

risk. #{163}3

in their 30 kg/m2. link

The increased 3 1). The

between saturation saturation

rapid
gallstone his ments

weight
population were

loss(29,

(29appears to be increased still further during 31), suggesting that weightloss could induce
(32). Jorgensen survey with an and increased (21) found addressed that prevalence this slimming issue

and gallstones with cholesterol

We are grateful

for the continuingconscientious

help ofthe participants

Mark Health Frank Shneyder, Stefanie Study is under the Speizer.

formation associated

screening

treat- References FE, in

Semin

of gallstones

in both sexes; moreover, was also associated with treatments analyses ments were adjusting was attenuated significant. association associated for BMI, and His

the number of slimming treatments 1. Maclure MK, Hayes KC, Colditz GA, Stampfer MJ, Speizer Willett WC. Weight, diet, and the risk ofsymptomatic gallstones gallstone prevalence. However, such middle-aged women. N EngI J Med 1989;32 1:563-9. with obesity and in multivariate the association with slimming treat- 2.
Kern F Jr. Epidemiology and natural history of gallstones.

Downloaded from www.ajcn.org by guest on August 25, 2011

in women findings we observed

statistically the moderate weight loss.

Liver Dis l983;3:87-96. (but not men), no longer 3. Scragg RK, McMichael AJ, Baghurst PA. Diet, are quite compatible with weight in gallstone disease: a case-control study. for substantial previous
1 1 13-9.

alcohol, Br Med

and relative J l984;288:

4.

The
weight

incidence
loss has

ofclinical
been 1 76-kJ/d free from

gallstone
estimated in

formation
several diets.

during
recent Among

very
small

rapid
studies 51

Diehl AK, Hakkner SM, Hazuda HP, Stern MP. Coronary and clinical gallbladder disease: an approach to the
gallstones? Am

risk factors
prevention of

J Public

S. Wysowski DK, EL. A study ofa possible association between breast cancer and gallbladder disease. Am J Epidemiol l986;l23:532-43. Liddle et al (8) found that after 8 wk of dieting (mean weight L, Sama C, Morselli-Labate AM, et al. A population study loss, 16.5 kg), 13 had developed gallstones and 3 of those had 6. Barbara prevalence ofgallstone disease: the Simione study. Hepatology symptoms severe enough to require cholecystectomy. In a 16- ofthe 1987;7:9 13-7. wk trial of ursodeoxycholic acid in rapid weight loss (22.6 kg 7. The Rome Group for Epidemiology and Prevention of Cholelithiasis mean loss), 7 of 33 patients on placebo or aspirin developed that used 2092-2 patients initially (500-520-kcal/d)

Health l987;77:84l-S. Goldberg EL, Comstock GW, Diamond

gallstones

according

to

ultrasound,

(GREPCO).

The

epidemiology

ofgallstone

disease

in

Rome,

Italy.

new

gallstones, that free

and

2 of them

were

symptomatic who by proportion for men,

(33). program, completed ultrasound for

Yang 15

et

al (9) found 1 54 subjects program

screening.

in a 16-wk from gallstones gallstones were reported the risk was and for thosewho ratio same

rapid-weight-loss at baseline diagnosed on higher the

developed
No details

symptomatic more Ifone stones

Broomfield

but loss, the

that those

of the 8. on were 9. with 10.

weight assumes in

were

more obese
observed the three

at baseline.
to asymptomatic by Liddle studies subjects, (8) and together 1 1. or 3.8%

ofsymptomatic

the Yang et al study as (33) et al, then putting incidence of

yields (95% should were

a cumulative

9 per estimate and

238

CI

2.0-7.0%)

over

a 2-4-mo

weight-loss

program.

This
12.

Part II. Factors associated with the disease. Hepatology l988;8:90713. Liddle RA, Goldstein RB, Saxton J. Gallstone formation during weight reduction dieting. Arch Intern Med l989;l49:1750-3. Yang HY, Petersen GM, Marks JW, Roth MP, Schoenfield U. Risk factors for gallstone formation during rapid weight loss. Gastroenterology l990;98(suppl S):A266(abstr). Colditz GA. The Nurses Health Study: findings during 10 years of follow-up of a cohort of US women. Curr Probl Obstet Gynecol Fertil 1990; 13:129-74. Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol l985;l22:S 1-65.
Maclure KM, Hayes KC, Colditz GA, Stampfer MJ, Willett WC.

be considered a maximum all under intense surveillance

because these subjects any symptoms

would

be likely to be attributed to the stones. We have no data from the Nurses

cidence any ofthe that factor suggest is a risk of gallstones participants a history for clinically with were rapid enrolled symptomatic weight

Health
loss in such and loss gallstones

Study
do over programs.

on
not

the
know

Dietary predictors ofsymptom-associated women. Am J Gin Nutr l990;52:916-22. 13. T. Abdominal symptoms and in- Jorgensen
Hepatology

gallstonesin middle-aged gallstone disease: WC. and an epi-

ofsubstantial

weight

investigation. if demiologic 14. Rimm EB, Stampfer Our data of self-reported waist a 2-y period

1989;9:856-60.

MJ, Colditz GA, Litin and hip circumferences

1:466-73.

L, Willett in men

Validity women.

independent

Epidemiology

1990;

of the
impact weight weight

effect
ofweight (Table loss

of current

obesity.

The

analysis

that

assesses
attained effect

15. Rothman Ki, Boice JD Jr. Epidemiologic analysis with a programthe mable calculator. Washington, DC: Public Health Service, 1979. of 16.

loss adjusting 5) suggests that be attributable

for previous much of to the excess

rather than the apparent weight present

[NIH publication 79-1649). Cox DR. Regression models and life-tables.

187-220.

J R Stat Soc [B) 1972;34:

may

before

658
17. Colditz GA, Willett WC, Stampfer MJ, London

STAMPFER
SL, Segal 25. MR.

ET

AL
BW, Soloway RD. Pigment vs cholesterol cholelithiasis:

Trotman

Speizer FE. Patterns of weight change and their relation to diet in a cohort of healthy women. Am J Clin Nutr l990;Sl:l 100-S. 26. 18. Colditz GA, Giovannucci E, Rimm EB, et al. Alcohol intake in relation to diet and obesity among men and women. Am J Gin 27. Nutr 199 1;54:49-55. 19. Stampfer MJ, Colditz GA. Estrogen replacement therapy and coronary heart disease: a quantitative assessment of the epidemiologic evidence. Prey Med l990;20:47-63. 28. 20. A Report from the Boston Collaborative Drug Surveillance Program, Boston University Medical Center. Surgically confirmed gallbladder

clinical 40. Soloway

troenterology Bernstein

and RD,

epidemiological Trotman
1977;72: RA, Giefer

aspects. BW,
167-82. EE, Vieira JJ,

Am JD.
Wener

J Dig Pigment
LH,

Dis

1975;20:735gallstones.
Rimm AA.

Ostrow

GasGall-

bladder
clinically women. Friedman

disease.

II. Utilization

ofthe

like table method

in obtaining

21.

useful information; a study of 62,739 weight-conscious J Chronic Dis 1977;30:529-4l. GD, Kannel WB, Dawber TR. The epidemiology of gallbladder disease: observations in the Framingham Study. Chronic J Dis 1966;19:273-92. disease, venous thromboembolism, and breast tumors in relation to 29. Bennion U, Grundy SM. Effects of obesity and caloric intake on postmenopausal estrogen therapy. N Engl J Med l974;290: 15-9. biliary lipid metabolism in man. J Gin Invest l975;56:996-lOll. Jorgensen T. Gallstones in a Danish population. Relation to weight, 30. Freeman JB, Meyer PD, Printen KJ, Mason EE, DenBesten L. physical activity, smoking, coffee consumption, and diabetes mellitus.

Gut
22. Layde

l989;30:528-34.
PM, Vessey MP, Yeates D. Risk

Analysis
163-6.

ofgallbladder

bile in morbid

obesity.

Am J Surg l975;l29:

a cohort study of young J Epidemiol Community

23. 24. Tucker Colditz LE, GA, Tangedahl

women Health
TN,

factors gall-bladder for disease: 3 1. Mok HYI, Von Bergmann K, Crouse JR. Grundy SM. Biliary lipid attending family planning clinics. metabolism in obesity. Effects ofbile acid feeding before and during l982;36:274-8. weight reduction. Gastroenterology 1979;76:556-67. SR. Prevalence of gallstones

Newmark women.

32.

in obese Caucasian information cohort study

l982;6:247-Sl. Martin P, Stampfer MJ, et al. Validation questionnaire of on risk factors and disease outcomes in a prospective in women. Am J Epidemiol l986;l23:894-900.
American

mt J Obes

Amaral JF, Thompson WR. Gallbladder disease in the morbidly obese. Am J Surg l985;l49:55l-7. 33. Broomfield PH, Chopra R, Sheinbaum RC, et al. Effects of ursodeoxycholic acid and aspirin on the formation oflithogenic bile and gallstones during loss ofweight. N EngI Med l988;3 19:1567-72. J

Downloaded from www.ajcn.org by guest on August 25, 2011