Gastroenterology 2016;150:156167

CLINICALBILIARY
A Prediction Rule for Risk Stratication of Incidentally
Discovered Gallstones: Results From a Large Cohort Study
Daniel Mnsted Shabanzadeh,1,2 Lars Tue Srensen,1,3 and Torben Jrgensen2,4,5
1

Digestive Disease Center, Bispebjerg University Hospital, Copenhagen, Denmark; 2Research Centre for Prevention and
Health, Centre for Health, Capital Region of Copenhagen, Copenhagen, Denmark; 3Institute for Clinical Medicine, 4Department
of Public Health, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark; 5Faculty of
Medicine, Aalborg University, Aalborg, Denmark

CLINICAL BILIARY

BACKGROUND & AIMS: No one knows exactly what proportion of gallstones cause clinical events among subjects
unaware of their gallstone status. We investigated the longterm occurrence of clinical events of gallstones and associations between ultrasound observations and clinical events.
METHODS: We analyzed data from 3 randomly selected
groups in the general population of urban Copenhagen (age,
3070 y) participating in an international study of cardiovascular risk factors (the Multinational mONItoring of trends
and determinants in CArdiovascular disease study). In this
study, participants (n 6037) were examined from 1982
through 1994, and underwent abdominal ultrasound examinations to detect gallstones. Our study population comprised
664 subjects with gallstones; subjects were not informed of
their gallstone status. Participants were followed up for clinical events through central registers until December 31, 2011.
Independent variables included ultrasound characteristics,
age, sex, comorbidity, and female-associated factors, which
were analyzed using Cox regression. RESULTS: Study participants were followed up for a median of 17.4 years (range,
0.129.1 y); 99.7% of participants completed the study. A total
of 19.6% participants developed events (8.0% complicated
and 11.6% uncomplicated). Ten percent had awareness of
their gallstones; awareness was associated with uncomplicated and complicated events. Stones larger than 10 mm were
associated with all events (hazard ratio [HR], 2.31; 95%
condence interval [CI], 1.453.69), acute cholecystitis (HR,
9.49; 95% CI, 2.0543.92), and uncomplicated events
(HR, 2.55; 95% CI, 1.384.71), including cholecystectomy
(HR, 2.69; 95% CI, 1.295.60). Multiple stones were associated with all events (HR, 1.68; 95% CI, 1.002.81), complicated events (HR, 2.52; 95% CI, 1.056.04), and common bile
duct stones (HR, 11.83; 95% CI, 1.5491). There was an association between gallstones more than 5 years old and acute
cholecystitis. Female sex was associated with all and uncomplicated events. We found a negative association between
participant age and all events, uncomplicated events, and
acute cholecystitis. Comorbidities and female-associated factors (intake of birth control pills or estrogens and number of
births) were not associated with events. Compared with men
with a single stone of 10 mm or smaller (reference), women
with multiple stones greater than 10 mm had the highest risk
for events (HR, 11.05; 95% CI, 3.7632.44; unadjusted
absolute risk, 0.0235 events/person-years). CONCLUSIONS:
Fewer than 20% of subjects with gallstones develop clinical
events. Larger, multiple, and older gallstones are associated
with events. Further studies are needed to conrm the prediction rules.

Keywords: Cholelithiasis; Gallbladder Disease; Longitudinal;

Ultrasonography.

he natural history of gallstones has been debated

since the beginning of the 20th century, and the
idea of an innocent gallstone has ranged from acceptance
to myth.14 Population-based cohort studies have reported different results on how gallstones develop to
symptoms,511 but all studies have been biased by awareness of gallstone status in the study populations. Accordingly, the true natural history of gallstones causing
symptoms is unknown, therefore we performed a longitudinal study of a random population that was unaware of
their gallstone status and we performed a thorough followup evaluation.
Generally, the clinical course for the majority of gallstones is asymptomatic. Uncomplicated gallstone disease
includes occasional abdominal pain attacks called biliary
colic. When complicated, the stones may cause acute
cholecystitis and common bile duct stones. The latter may
cause life-threatening cholangitis and acute pancreatitis.12
Previous studies have reported a wide variety in the rates
of symptomatic disease. Uncomplicated disease with biliary
colic has been reported in 13%44%, and cholecystectomy
has been reported in 15%28%. Complicated disease was
reported in 0%9% (Table 1).
A few prospective cohort studies have identied some
factors, such as female sex,5,6 younger age,5,9 high body
mass index,6 and larger stones,10 as being associated with
symptomatic disease. Overall, little knowledge exists about
the development into symptomatic disease,1315 although
this is of importance for doctors and patients in surgical
decision making.
Based on a population that is unaware of gallstone
status, the aims of this study were to determine the natural
history of gallstones over a total period of 29 years. The
cumulative incidence proportions of all uncomplicated and

Abbreviations used in this paper: CI, condence interval; HR, hazard ratio;
MONICA, Multinational mONItoring of trends and determinants in
CArdiovascular disease.
Most current article
2016 by the AGA Institute
0016-5085/$36.00
http://dx.doi.org/10.1053/j.gastro.2015.09.002

Study
Gracie and
Ransohoff,4 1982,
United States
McSherry et al,5
1985,
United States

Friedman et al,6
1989,
United States

Attili et al,7
1995, Italy

Angelico et al,8
1997, Italy

Male faculty
members,
University
of Michigan
Health Insurance
Plan of
Greater
New York
subscribers
The Keiser
Permanente
Medical Care
Program,
Northern
California
Civil servants
in Rome.
Group for
Epidemiology
and Prevention
of Cholelithiasis
(GREPCO)
Random sample
of females
from electoral
rolls in one town
of rural central
Italy, GREPCO
Random sample
aged 3585
from one
municipality

Subjects with
gallstone
disease
followed
up, N

Gallstone
diagnosis

123

Oral cholecystography

1124

13%
biliary colic

2.4%

691

Oral
cholecystography,
surgery

Median,
5.2

22.3% change
in intensity
of pain

8.7%

467

Oral
cholecystography,
surgery

NA (maximum, 37)

21.2%
nonsevere
events

8.4%

19.7%
symptoms
and
prophylactic

151

Ultrasound

z10

18.5%
biliary colic

3.3%

7.3%
symptoms
9.9%
prophylactic

47

Ultrasound

z10

21.3%
biliary colic

123

Ultrasound

Median,
7.25 (range:
0.2512.2)

10.6%
developed
events

Follow-up
period, y

Symptom
development

Complicated
disease

0%

4.1%

8.9%
symptoms
24.4%
prophylactic
z21%
28%

14.9%
symptoms

NA

Natural History of Gallstones

Halldestam
et al,9 2004,
Sweden

Population

Cholecystectomy
owing to
symptomatic
uncomplicated
disease or
prophylactic
in asymptomatic

January 2016

Table 1.Previous Population-Based Studies of the Natural History of Gallstones

157

CLINICAL BILIARY

CLINICAL BILIARY

158

Table 1. Continued

Population

Festi
et al,10 2010, Italy

Systematic
sampling
from
electoral
rolls in 10
Italian
regions,
Multicenter
Italian
Study on
Cholelithiasis
(MICOL)
Random
sample from
National
Peoples
Registry in Bergen

Schmidt,12
2011. Norway

Gallstone
diagnosis

Follow-up
period, y

Symptom
development

793

Ultrasound

Mean,
8.7

18.9%
developed
new
symptoms
of pain

225

Ultrasound

24

44%
interim or
present pain

Complicated
disease
4.5%

NA

Shabanzadeh et al

Study

Subjects with
gallstone
disease
followed
up, N

Cholecystectomy
owing to
symptomatic
uncomplicated
disease or
prophylactic
in asymptomatic
15.3%
symptoms
7.9%
prophylactic

NA

z, exact data were not available; NA, not available.

Gastroenterology Vol. 150, No. 1

January 2016

Natural History of Gallstones

Materials and Methods

Sampling and Ultrasound Examination
The data in this study comprised 3 random samples of the
general population made up of persons aged 3070 years,
living in 11 municipalities in the Western part of the urban area
of Copenhagen, and were drawn by computer from the Civil
Registration System (Figure 1). The cohorts were part of an
international collaboration of the Multinational mONItoring of
trends and determinants in CArdiovascular disease (MONICA).
The aim of the MONICA studies was to examine cardiovascular
risk factors in the general population. The MONICA I (cohort 1)
consisted of persons aged 3060 years who were examined in
19821984.16 The 1914 cohort (cohort 2) consisted of persons
aged 70 years who were examined in 1984.17 The MONICA III
(cohort 3) consisted of persons aged 3070 years who were
examined in 19911992.18 Cohort 1 was re-examined after 5
years (19871988),19 and again after 10 years (19921994).18
The reason for the drawing of the 2 rst cohorts was to
represent a wide range of ages. A third cohort was drawn later
to reect potential time-dependent gallstone prevalence. All
participants were invited by letter and nonresponders were reinvited and contacted by telephone. Free transportation and
examinations that took place outside of ofce working hours
were offered if necessary. Participants appeared after 12 hours
of fasting and underwent a general health examination,
including a questionnaire about lifestyle and medical history,
and had an abdominal ultrasound examination to assess

gallstone status. Gallstone prevalence and incidence studies

from the 2 rst cohorts have been published.16,17,19
Gallstones were dened as acoustic shadows that moved
with gravity in a gallbladder lumen. Exceptions from the
mobility criteria included the following: if a stone was wedged
in the infundibulum of the gallbladder or otherwise impeded by
size, septae, or folds. These were classied as nonmobile stones.
Nonvisualization of the gallbladder lumen was present in some
gallbladders containing nonmobile stones.16 No intrahepatic or
common bile duct gallstones were identied. As accepted by
the local ethics committee, the participants were not informed
about the nding of gallstones or other benign conditions of
their gallbladder to avoid unnecessary surgical treatment in
asymptomatic participants.19 All cohorts were sampled, invited,
and examined according to the same methods.

Study Population
The 3 cohorts comprised 7847 persons. Of these, 6037
(76.9%) participated in the study and were examined by
ultrasound. Cholecystectomy was performed in 189 persons,
excluding them from the study population. A total of 664
persons had ultrasound-proven gallstones (Figure 1). Of these,
10% reported awareness of having gallstones.

Study Design
Data from the 664 individuals with gallstones were linked
to the Danish National Patient Registry by use of a unique
personal identication number for each participant. The National Patient Registry contains data on dates of contact with
hospitals and codes for diagnosis, treatments, and surgical
procedures, and has been validated several times.20 Entry in
this study was dened as the date gallstones were detected for
the rst time regardless if this was a baseline or follow-up
examination. All gallstone-positive individuals were followed
up until the date of a gallstone event, death, or emigration. The
follow-up evaluation was terminated on December 31, 2011
(Figure 1).

Figure 1. Study population ow and study design.

CLINICAL BILIARY

complicated clinical gallstone events were estimated and

the association between gallstone characteristics were
determined by ultrasound, and clinical events were determined. The hypotheses were that the development of
symptomatic disease might be less severe than was reported previously, that recently formed gallstones may
cause fewer events, and that awareness of gallstone status
would cause more clinical events.

159

160

Shabanzadeh et al

Study Outcome and Independent Variables

CLINICAL BILIARY

The study outcomes were all registered clinical gallstone

events dened by hospital admission with coded gallstone diagnoses or treatments, divided into complicated events and
uncomplicated events. Complicated events were categorized
into acute cholecystitis and common bile duct stones also
including pancreatitis. The uncomplicated events were divided
into those that had a cholecystectomy and those that
just received a clinical diagnosis of gallstone disease
(cholecystolithiasis/cholelithiasis) without treatment. Gallstone
events were dened using diagnosis codes or surgical treatment codes, either as single entities or in combination. Some
individuals had multiple codes and, in these individuals,
complicated diagnoses were prioritized before uncomplicated
and treatment codes were prioritized over diagnosis codes.
Persons with solitary treatment codes indicating complicated
events such as endoscopic retrograde cholangiopancreaticography or endoscopic stents were regarded as patients with
common bile duct stones and cholecystostomy as acute cholecystitis. Solitary cholecystectomy with no diagnose codes indicating a complicated event was considered as an uncomplicated
event. Chronic cholecystitis diagnosed after cholecystectomy
was considered an uncomplicated event because it rather denes the surgical ndings during nonacute surgery or a histopathologic diagnosis than it does the clinical course of gallstone
disease.21 Treatment codes followed the Nordic Classication of
Surgical Procedures, which was revised in 1995.22 Diagnosis
codes followed the International Classication of Diseases, 8th
or 10th revisions (International Classication of Diseases, 9th
revision, was never used in Denmark). Codes for surgical or
endoscopic treatments of the liver or biliary ducts in participants with concomitant diagnoses of hepatic metastasis or
cancer of the pancreas, liver, or bile ducts during the same
admission were not regarded as gallstone events. The detailed
coding of gallstone events is outlined in Supplementary Table 1.
Gallstone age could be explored in cohort 1 owing to the 2
re-examinations. A gallstone age of 5 years or less was dened
in participants with ultrasound-proven gallstones at one reexamination, but not at the former examination. Gallstones
detected at baseline could not be given an age, but they were
regarded to be older than 5 years. Gallstone characteristics
such as age, number, size, and mobility were the primary independent variables of interest. Awareness of gallstone status
was examined as an independent variable as well. The
comorbidities of diabetes and hypertension, as diagnosed by a
physician, and the sex-specic female factors such as number of
births and use of birth control pills or estrogens, were explored
for all gallstone events. A priori covariates were age, sex, and
body mass index as measured by the body mass divided by the
square of the body height. These variables were regarded as
confounders and were adjusted for when possible. Participant
age and sex also were studied as independent variables. All
variables were assessed at the time of the participants
entrance into the study (Table 2).

Data Management
The primary author harmonized all variables in the 3 cohorts based on original questionnaires and the ultrasound information sheets. There was a discrepancy in the gallstone size
variable between the cohorts. Cohort 3 included it as a
continuous variable and the other 2 cohorts included it as a

Gastroenterology Vol. 150, No. 1

Table 2.Study Population Characteristics
Variable
Participant age
Sex
Body mass index
Stone number
Largest stone size
Smallest stone size
(only cohort 3)

Stone mobility
Gallstone age
(only cohort 1)

Awareness of
gallstone status
Diabetes
Hypertension
Birth control pills
Estrogen

Levels

Value (n 664) Missing

Median [IQR]
Female
Male
Median [IQR]
2
1
10
>10
5

60.0 [50.065.0]
386 (58.1)a
278 (41.9)a
25.6 [23.129.2]
305 (51.9)a
283 (48.1)a
421 (67.0)a
207 (33.0)a
148 (80.4)a

76

>5
Not measured
Mobile
Nonmobile
>5 years

36 (19.6)a
480
566 (85.6)a
95 (14.4)a
216 (45.2)a

5 years
No re-examination
No

262 (54.8)a
186
595 (90.0)a

Yes
No
Yes
No
Yes
Ever used
Never used
Ever used
Never used

66
635
29
498
166
140
201
73
311

(10.0)a
(95.6)a
(4.4)a
(75.0)a
(25.0)a
(41.1)a
(58.9)a
(19.0)a
(81.0)a

36

45
2

NOTE. The random samples of the general population

comprised 7847, with 6037 participants. Of these, 664 had
gallstones, which comprised the study population.
a
The numbers in parentheses are % of total population.
categoric variable. It was agreed to change the continuous
variable from cohort 3 into a categoric variable for harmonization among cohorts. One of the co-authors (T.J.) was
responsible for the design and conduction of all 3 original
cohort studies. A data manager handled all of the data in the
databases and registries. This included data extraction from the
study databases, the linking of participants to the National
Patient registry, and retrieving participant status data from the
Civil Registration System. All the data were extracted to Excel
(Microsoft, Redmond, WA) sheets and the primary author then
performed all subsequent data management. There was no new
contact with the participants and, therefore, the study required
no new research ethics approval.

Statistical Analysis
Descriptive analyses were reported as medians with interquartile ranges for continuous data and as counts and percentages for categoric data. Time-to-gallstone-event analyses
were performed as competing-risk analyses with cumulative
incidence proportions and death as competing outcomes for all
outcomes. In these, participants were right-censored when they
had a gallstone event, died, or emigrated, or they completed
follow-up evaluation until December 31, 2011. Separate analyses for prestudy awareness of gallstones were performed and

January 2016

Natural History of Gallstones

161

Table 3.Counts of Gallstones Events During Complete Follow-Up Evaluation and Cumulated Incidence Proportions Over 20
Years
Event type
All

Complicated

Uncomplicated

Total
No awareness
Awareness
Total
Common bile duct stones
Pancreatitis
Endoscopic retrograde
cholangiopancreaticography
Endoscopic stent
Acute cholecystitis
Cholecystostomy
No awareness
Awareness
Total
Cholecystectomy
Clinical diagnosis
No awareness
Awareness

No events
Total

n (% of total)
130
99
30
53
26
3
3

(19.6)
(15.0)
(4.5)
(8.0)
(3.9)
(0.5)
(0.5)

1 (0.2)
19 (2.9)
1 (0.2)
43 (9.0)
9 (1.9)
77 (11.6)
44 (6.6)
33 (5.0)
56 (8.4)
21(3.2)
534 (80.4)
664 (100)

CIP20

a
years

0.18
0.15
0.42
0.08

P valueb
<.0001

0.07
0.16
0.113

.02

0.09c
0.32c

<.000

CIP20 years, cumulative incidence proportion over 20 years.

Grays test.
c
Uncomplicated events included a clinical diagnosis of gallbladder stones without further treatment or a cholecystectomy
without codes for complicated events. Complicated events were censored from the cumulative incidence proportions of the
uncomplicated events.
b

Grays test was used for testing differences between awareness

and no awareness with a signicance level of P < .05.23 The
cumulative incidence curves were reported. For hypothesis
testing, Cox proportional hazards regressions were performed.
The risk time was set at a maximum of 10 years of observation
to obtain a clinically relevant time period.
The independent variables and the confounders associations with outcome were rst estimated in unadjusted models
separately. In multiple regression analyses, a threshold of 10
events per included parameter was set.24 For the all gallstone
event outcome, all ultrasound gallstone variables and all a
priori confounders were included in one large model. For the
smaller models of complicated and uncomplicated outcomes
only, the confounders that had the best statistical association in
the univariate analysis were adjusted for. In all analyses of
uncomplicated events, the complicated events were censored.
Scaled Schoenfeld residuals were used to test the proportional
hazards assumption in Cox regression.25 The assumption of
proportional hazards was not violated in any of the analyses.
Analyses of the variables that were measured in only one
cohort were restricted to that cohort. Hazard ratios (HRs) with
95% condence intervals (95% CIs) were reported. A 95% CI
not including 1 was set for the purpose of rejecting the null
hypothesis.
To make this studys ndings more clinically applicable, a
clinical prediction score was created. The aim was to translate
the obtained gallstone event estimates to relevant person
proles seen in clinical practice. First, a multivariable regression model was built from the clinically measurable and
signicant predictors identied in this study. The subjects who
had only a clinical diagnosis of gallstones without treatment

(cholecystolithiasis/cholelithiasis) during follow-up evaluation

were of no importance in terms of the clinical prediction score
and therefore were censored for the purpose of this analysis.
The lowest-risk person was identied by inspecting the HRs
from the multivariable model and was chosen as the reference
for the calculation of the HRs for the remaining person proles
in the prediction score table. The rates of events and the unadjusted absolute risks (events/person-years) of every person
prole category also was reported.
The statistical software R Studio (RStudio Inc, Boston, MA)
was used for all analyses, including the package cmprsk for the
competing-risk analysis, survival for the Cox regression analysis
and testing proportional hazards, and lava for the prediction
scores.

Results
Among the 664 subjects with gallstones, a total of 130
(19.6%) had gallstone events (Table 3). The median followup period was 17.4 years, with a range of 0.129.1 years,
comprising a total of 10,348 person-years at risk. At followup evaluation, 308 (46.4%) participants were alive, 354
(53.3%) had died, 1 had emigrated, and 1 had changed his
personal identication number (0.3%). Follow-up evaluation thus was 99.7% complete. No participants developed
gallbladder cancer.

All Events
The cumulative incidence proportion at 20 years was
18% (Table 3). The steepest curve with the most gallstone

CLINICAL BILIARY

162

Shabanzadeh et al

Gastroenterology Vol. 150, No. 1

CLINICAL BILIARY

Figure 2. Cumulative incidence proportions (CIP) for

(A) all, (B) complicated,
and (C) uncomplicated
gallstone events. Awareness of gallstone status
shown separately.

events was seen during the rst 15 years of follow-up

evaluation (Figure 2A). Multiple stones and stone size
greater than 10 mm caused 23 times higher signicant
associations for gallstone events compared with single
stones and smaller sizes of the largest stone. In addition,
participant age was associated inversely with gallstone
events (Table 4). When participant age was split into decades, the youngest (3040 and 4050 years of age) participants were associated signicantly with gallstone events
when compared with the oldest (6070 years of age) in an
unadjusted model (data not shown). Gallstone age of 5 years
or less or the comorbidities of diabetes and hypertension
had no associations with all events. Finally, female sex was a
predictor of all gallstone events. In the female population,
use of female hormones (birth control pills or estrogens) or
the total number of births had no signicant associations
with all gallstone events (Table 4). No signicant associations were found when the 3 sex-specic female factors (any

use of birth control pills or estrogens and number of births)

were combined in an age-adjusted multivariable model
(data not shown). Therefore, the comorbidities and sexspecic female factors were not explored in further analyses of complicated or uncomplicated events.

Complicated Events
During follow-up evaluation, 8% of the participants
developed a complicated gallstone event. Of these, 5% had
common bile duct stones or pancreatitis and 3% had acute
cholecystitis. The cumulative incidence proportion at 20
years for complications was 8% (Table 3). Multiplicity of
stones had a signicant association with complicated events
in both unadjusted and age-adjusted models. Female sex or
participant age had no signicant associations with all
complicated events. Common bile duct stones were associated with multiple stones. Acute cholecystitis was

January 2016

Natural History of Gallstones

163

Table 4.Gallstone Events Cox Regression Over 10 Years

All

All (only female population)

Complicated

Uncomplicated

Variable
Participant age
Sex, female
Stone number  2
Largest stone size > 10 mm
Smallest stone size  5 mm
Stone mobility, mobile
Gallstone age  5 y
Gallstone awareness
Hypertension
Diabetes
Use of birth control pills
Use of estrogens
Number of births
All complicated events
Participant age
Sex, female
Stone number  2
Largest stone size > 10 mm
Smallest stone size  5 mm
Stone mobility, mobile
Gallstone age, 5 y
Gallstone awareness
Acute cholecystitise
Participant age
Sex, female
Stone number  2
Largest stone size > 10 mm
Stone mobility, mobile
Gallstone age, 5 y
Gallstone awareness
Common bile duct stonese
Participant age
Sex, female
Stone number  2
Largest stone size > 10 mm
Stone mobility, mobile
Gallstone age, 5 y
Gallstone awareness
All uncomplicated events
Participant age
Sex, female
Stone number  2
Largest stone size > 10 mm
Smallest stone size  5 mm
Stone mobility, mobile
Gallstone age, 5 y
Gallstone awareness
Cholecystectomy
Participant age
Sex, female
Stone number  2
Largest stone size > 10 mm
Stone mobility, mobile
Gallstone age, 5 y
Gallstone awareness

Unadjusted HR (95% CI)

0.97
2.27
1.68
2.31
1.12
0.77
0.67
4.25
0.88
0.62
1.16
0.79
1.03

(0.960.99)
(1.343.85)
(1.002.81)
(1.453.69)
(0.323.94)
(0.421.40)
(0.401.14)
(2.567.05)
(0.501.52)
(0.152.52)
(0.672.02)
(0.371.66)
(0.871.23)

0.98
1.58
2.52
2.01

(0.951.01)
(0.753.34)
(1.056.04)
(0.974.17)
1.13 (0.403.23)
0.81 (0.351.86)
3.07 (1.327.13)

0.98
2.50
2.08
2.62
0.69
0.78
4.12
0.86
0.68
0.91
0.85
1.05

(0.951.00)a
(1.344.67)a
(1.213.59)a
(1.564.40)a
(0.331.43)a
(0.451.34)b
(2.466.90)b
(0.491.51)b
(0.162.82)b
(0.481.73)c
(0.401.81)c
(0.871.27)c

1.55 (0.733.27)d
2.74 (1.136.62)d
2.00 (0.974.15)d
1.12 (0.393.18)d
0.88 (0.382.06)d
3.22 (1.387.51)d

0.95
1.93
0.96
9.49
0.94
0.20
4.38

(0.900.99)
(0.517.28)
(0.283.32)
(2.0543.92)
(0.273.20)
(0.040.96)
(1.1316.97)

0.99
1.24
11.83
0.98
1.37
2.14
2.82

(0.961.03)
(0.493.14)
(1.5491)
(0.342.83)
(0.325.94)
(0.578.07)
(0.948.51)

0.97
3.12
1.31
2.55
0.91
0.61
0.60
5.25

(0.940.99)
(1.456.73)
(0.682.51)
(1.384.71)
(0.194.36)
(0.291.27)
(0.301.19)
(2.779.94)

0.97
3.00
1.53
2.44

0.95
2.86
0.91
2.69
1.05
0.59
5.37

(0.920.98)
(1.176.99)
(0.422.01)
(1.295.60)
(0.373.00)
(0.271.29)
(2.5111.48)

0.95 (0.930.98)g
0.94 (0.432.06)g
2.66 (1.285.53)g
1.13 (0.393.23)g
0.61 (0.281.32)g
4.82 (2.2510.35)g

Adjusted for age, sex, body mass index, stone number, largest stone size, and stone mobility.
Adjusted for age, sex, and body mass index.
c
Adjusted for age and body mass index.
d
Adjusted for age.
e
No adjustment owing to the low number of events.
f
Adjusted for age and sex.
g
Adjusted for sex.
b

Adjusted HR (95% CI)

(0.95 to <1.00)f
(1.396.48)f
(0.792.95)f
(1.324.52)f
0.65 (0.311.36)f
0.70 (0.341.41)f
5.28 (2.7510.11)f

CLINICAL BILIARY

Gallstone events

164

Shabanzadeh et al

associated with a stone size greater than 10 mm, but was

associated inversely with gallstone age of 5 years or less, in
addition to participant age. Adjusted analyses could not be
performed for common bile duct stones and acute cholecystitis because of the low number of events.

Uncomplicated Events
Uncomplicated gallstone events occurred in 11.6% of the
participants during follow-up evaluation. Cholecystectomy
for uncomplicated events was performed in 6.6% of the
participants. The cumulative incidence proportion at 20
years was 11% (Table 3). Stone size greater than 10 mm
was associated with all uncomplicated events and with
cholecystectomy in both the unadjusted and adjusted
models. Female sex and younger age were associated with
uncomplicated events and cholecystectomy (Table 4).

Prestudy Awareness of Gallstones

CLINICAL BILIARY

In all of the analyses, prestudy awareness compared

with no awareness was associated with signicantly higher
cumulative incidence proportions of gallstone events
(Table 3). The difference was most obvious for the uncomplicated subgroup (Figure 2B and C). Awareness was
associated independently with all, complicated, and uncomplicated gallstone events in both unadjusted and
adjusted analyses, with the exception of common bile duct
stones (Table 4).

Sensitivity Analysis
Five participants were categorized as having complicated events solely based on 1 treatment code for endoscopic retrograde cholangiopancreaticography, endoscopic
stent, or cholecystostomy without any diagnosis code to
conrm the type of clinical event (Table 3). For validation
purposes, a sensitivity analysis thus was performed after
these 5 cases were recategorized as uncomplicated events.
However, none of the HRs and 95% CIs were altered
signicantly in this analysis.

Clinical Prediction Score

Age, female sex, stone size greater than 10 mm, and
multiple stones were the 4 clinically measurable and signicant variables identied in this study. When only these 4
predictors were combined in a multiple regression model,
all remained signicantly associated with gallstone events
(complicated events and cholecystectomy for uncomplicated
events). A clinical prediction score table was created by
stratifying the 3 dichotomous predictors, and each row
therefore represented a person prole category seen in the
clinic based on sex (female/male), stone size greater than 10
mm (yes/no), and multiple stones (yes/no). The lowest-risk
person was identied as the male, with a single gallstone of
no more than 10 mm. By using the estimates from the
multiple model including the mentioned 4 predictors, this
lowest-risk person served as the reference for the calculation of the other person prole categories prediction estimates. The person with the highest risk was female, with

Gastroenterology Vol. 150, No. 1

multiple gallstones, and with the largest stone size greater

than 10 mm (HR, 11.05; 95% CI, 3.7632.44; unadjusted
absolute risk, 0.0235 events/person-years). Age was
included as a continuous variable (HR, 0.97; 95% CI,
0.950.99). To get a prediction score HR for an individual
patient, the patients prole category prediction HR must be
multiplied by a factor of 0.97 for each additional year of age.
The 10-year follow-up rate of events, absolute risks
(events/person years), prediction HRs with 95% CIs for all
clinical prediction score person proles, and a calculation
example for a prediction score HR can be found in Table 5.
Because age had an inverse association with gallstone
events the prediction risk will decrease with age.

Discussion
This was a cohort study of the natural history of gallstones with a long-term follow-up evaluation of a population that was unaware of having gallstones. More than 4 of 5
participants with gallstones remained uneventful during the
20-year follow-up period. Determinants of clinical events
over a 10-year period were female sex, young age, awareness of gallstones, and large, multiple, and older stones. A
score for estimation of individual clinical prediction of
gallstone events has been suggested based on patient sex,
age, and gallstone size and number obtained through ultrasound. In this prediction score, the person who had the
highest risk for an event was the female with multiple and
large stones and she had an 11 times greater relative risk of
cholecystectomy or complications over 10 years when
compared with a male with a single gallstone of no more
than 10 mm in size.
This study had some methodologic advantages when
compared with other population-based cohort studies of
gallstone disease411 (Table 1). Many of the previous studies
selected populations with no random sampling.47 In addition, the reported follow-up data were not sufcient in some
studies to calculate the annual symptom or complication
rate and, therefore, only the total rates are reported in
Table 1. Contrary to all previous studies, the participants of
this study were not informed about gallstone status.
Consequently, comparisons between this study and others
should be made with caution. In previous studies, the
ndings of higher cholecystectomy rates for uncomplicated
disease suggested a more aggressive attitude toward cholecystectomy in uncomplicated symptomatic disease and
even in prophylactic surgery in asymptomatic patients when
the gallstone status is known. The most obvious explanation
is a pathologization of the gallstone condition and introduction of a protopathic bias, as Attili et al7 suggested.
Accordingly, 2 studies even referred participants to their
family doctor after gallstones were detected using ultrasound.7,8 In the present study, only 10% were aware of
having gallstones before entry into the study. This fraction
of the population had signicantly higher rates of all outcomes, especially uncomplicated events (Table 3). This
nding may reect a protopathic bias. It also may reect the
hypothesis that the participants who were aware of their
gallstones had suffered biliary colic attacks before entry into

January 2016

Natural History of Gallstones

165

Table 5.Clinical Prediction Score of Gallstone Events (Cholecystectomy or Complicated Events) During 10 Years of Follow-Up
Evaluation

Male (1[1;1])

Female
(2.16 [1.10;4.25])

Largest stone
> 10 mm
(2.79 [1.57;4.96])

Multiple stones
(1.83 [1.02;3.30])

Rate of events, N
events/N persons in
prole categorya

Absolute risk, N
events/person years
in prole category

Prediction
HR (95% CI)

2/67
4/97
2/47
3/29
4/102
11/120
12/67
10/53

0,0032
0,0046
0,0048
0,0126
0,0041
0,0105
0,0211
0,0235

Reference
1.83 (1.023.30)
2.79 (1.574.96)
5.12 (2.1512.19)
2.16 (1.104.25)
3.96 (1.639.64)
6.02 (2.5314.33)
11.05 (3.7632.44)

NOTE. A multiple model was built including the 4 identied clinical relevant determinants of clinical events and each person
prole category prediction HR was calculated through the 3 dichotomous variables estimates from this model. The lowest-risk
person was chosen based on the estimate size and was identied as the male, with a single stone, and of a size no more than
10 mm. He served as the reference for the remaining person prole categories. The total rate of events, unadjusted absolute
risks, and prediction estimates are reported for each person prole category. The Largest stone size greater than 10 mm,
multiple stones, and sex were included as dichotomous variables and age was included as a continuous variable with a
linearity assumption. The age estimate (HR, 0.97 [0.95;0.99]) indicates that for each year older the risk is reduced by a factor of
0.97. An example of a prediction score HR calculation for a 60-year-old woman with the largest stone greater than 10 mm and
multiple stones was as follows: prediction score HR 11.05 * (0.97 60) 1.78. The values in parenthesis are hazard ratios
from the multiple model. The values in brackets are 95% condence intervals.
a
The total number of events was 48, the total sample size was 582.

with no diagnosis codes were excluded, helping to reduce

the likelihood of altered results. Misclassication is always a
potential bias in databases and cannot be corrected for
entirely by post hoc analyses. However, the database used
was validated several times,20 and the limitations of this
database were outweighed by its unique possibility to
enable a 99.7% complete long-term follow-up evaluation of
this large population sample with gallstones. Finally, some
of the 95% CIs of the estimates for complicated outcomes
were wide owing to low event counts. However, all 95% CIs
were signicant and the conclusions drawn from the estimates are thereby not caused by type II error.
Participants 3040 years of age were more prone to
have events when compared with older individuals.
Accordingly, cohort studies have reported that age was
associated inversely with symptomatic stones and treatment.5,9 Such ndings may suggest that younger participants may have different pain perceptions than the elderly.
It also may show that the risk factors for gallstone formation
are not necessarily the same as for the development of
symptoms or simply the fact that younger individuals have a
higher risk of having gallstones detected because of their
longer longevity. The female sex was found to be associated
with all gallstone events and uncomplicated events, similar
to previous studies.5,6 This association was independent of
common factors related to the female sex. Some practitioners may be more prone to send women than men to
ultrasound examinations in case of abdominal complaints
and this nding therefore may reect a detection bias.16
The larger stones were associated with uncomplicated
events, cholecystectomy, and acute cholecystitis. These
ndings are supported by another cohort study that found
that stone size greater than 1.5 cm was associated with a

CLINICAL BILIARY

the study and, therefore, had a higher risk of clinical events,

as observed in a previous study.7 Subjects with complicated
gallstone events before study start would, ultimately, have
undergone a cholecystectomy and, thereby, would be
excluded from this study. As such, aware participants in this
study most likely represent subjects with uncomplicated
gallstones with a history of biliary colic.
In all, 8% of the study population developed complicated
events. This proportion is slightly higher than was reported
in other studies and probably is owing to this studys long
and comprehensive follow-up period with as good as no
attrition.4,710 The overall rate of events is comparable with
previous studies (Table 1).
In this study design, registered clinical events were
interpreted as the development from asymptomatic to
symptomatic disease, whereas previous studies used
symptom questionnaires and hospital records (Table 1).
This might cause under-reporting of symptoms in this
studys population. However, asking study participants or
patients about symptoms can introduce recall bias when
gallstone status is known. Therefore, studies based on
patient-reported outcomes might be of limited use in this
context. This studys limitation was that we assessed all
variables only once at the time of entry into the study.
Gallstone characteristics may change over the years.19,26 We
partly addressed this problem by only including the rst 10
years of follow-up evaluation in all Cox regression models.
The smallest stone size was measured only in cohort 3 and,
therefore, no inferences could be made about this gallstone
characteristic for the entire cohort. Gallstone age could only
be explored only for cohort 1. We used data from a database, which might have introduced misclassication. This
was addressed by the sensitivity analysis in which patients

166

Shabanzadeh et al

CLINICAL BILIARY

change in symptoms, but were, however, not associated

with complications.10 The clinical interpretation of this association could be that larger stones cause more severe
pain, leading patients to seek medical help. The mechanism
may include an obstruction of the cystic duct by the large
stone, causing bile obstruction and gallbladder distention,
leading to acute cholecystitis or pain, as previously suggested.27,28 This study also found multiple stones to be
associated with complicated events and common bile duct
stones. No previous cohort studies have found this association.6,9 Previous cross-sectional screening studies have
found no association between stone size or multiplicity of
stones with abdominal symptoms.2931 Some clinical studies
have investigated the association between gallstone size and
number with clinical outcome.26,3239 A large stone size was
associated with acute cholecystitis in the pre-ultrasound
era.33,34 More recent clinical studies have not found stone
size to predict complications.36 Multiple stones have been
associated with abdominal pain,26,35,3739 however, solitary
stones also have been linked to complications.32 One study
did not nd any association with stone number.36 All of
these ndings are not directly comparable with the ndings
of this study owing to the difference in study design and the
fact that clinical studies include selected populations.
Gallstone age of less than or equal to 5 years was
associated negatively with acute cholecystitis in this study.
The association between gallstone age and outcome has not
been studied previously. We found no association between
stone mobility and gallstone events, which is in accordance
with previous cross-sectional screening studies.30,31

Conclusions
In a random sample of the population that is unaware of
gallstone disease, less than 1 of 5 with gallstones will
develop clinical events, and less than half of the clinical
events will be complicated. Predictors of all gallstone events
were found to be female sex, younger age, and having
multiple older and larger stones. This knowledge has clinical
implications when it comes to advising patients with gallstones. Patients with no symptoms require no further
treatment. A clinical prediction score, such as suggested
here, shows that some gallstone carriers have higher risks of
clinical events compared with others based on the identied
risk factors. This score can be used to make assessments of
relative risks for future clinical events in patients with
gallstones. However, such a score requires validation in a
similar large population of subjects with gallstones. Future
research should focus on long-term modiable risk factors
for clinical events, how to predict complications, and the
long-term consequences of surgery for gallstone disease.

Supplementary Material
Note: To access the supplementary material accompanying
this article, visit the online version of Gastroenterology at
www.gastrojournal.org, and at http://dx.doi.org/10.1053/
j.gastro.2015.09.002.

Gastroenterology Vol. 150, No. 1

References
1. Naunyn B, Garrod AE. A treatise on cholelithiasis. London: New Sydenham Society, 1896.
2. Osler W. The principles and practice of medicine,
designed for the use of practitioners and students of
medicine. New York: Appleton, 1909.
3. Mayo W. Innocent gall-stones a myth. JAMA 1911;56:
10211024.
4. Gracie WA, Ransohoff DF. The natural history of silent
gallstones: the innocent gallstone is not a myth. N Engl J
Med 1982;307:798800.
5. McSherry CK, Ferstenberg H, Calhoun WF, et al. The
natural history of diagnosed gallstone disease in symptomatic and asymptomatic patients. Ann Surg 1985;
202:5963.
6. Friedman GD, Raviola CA, Fireman B. Prognosis of
gallstones with mild or no symptoms: 25 years of followup in a health maintenance organization. J Clin Epidemiol
1989;42:127136.
7. Attili AF, De Santis A, Capri R, et al. The natural history of
gallstones: the GREPCO experience. The GREPCO
Group. Hepatology 1995;21:655660.
8. Angelico F, Del Ben M, Barbato A, et al. Ten-year incidence and natural history of gallstone disease in a rural
population of women in central Italy. The Rome Group for
the Epidemiology and Prevention of Cholelithiasis
(GREPCO). Ital J Gastroenterol Hepatol 1997;29:
249254.
9. Halldestam I, Enell EL, Kullman E, et al. Development of
symptoms and complications in individuals with
asymptomatic gallstones. Br J Surg 2004;91:734738.
10. Festi D, Reggiani ML, Attili AF, et al. Natural history of
gallstone disease: expectant management or active
treatment? Results from a population-based cohort
study. J Gastroenterol Hepatol 2010;25:719724.
11. Schmidt M, Hausken T, Glambek I, et al. A 24-year
controlled follow-up of patients with silent gallstones
showed no long-term risk of symptoms or adverse
events leading to cholecystectomy. Scand J Gastroenterol 2011;46:949954.
12. Harris H. Biliary system. In: Norton J, ed. Essential
practice of surgery: basic and clinical evidence. New
York: Springer-Verlag, 2003.
13. Friedman GD. Natural history of asymptomatic and
symptomatic gallstones. Am J Surg 1993;165:399404.
14. Sakorafas GH, Milingos D, Peros G. Asymptomatic
cholelithiasis: is cholecystectomy really needed? A critical reappraisal 15 years after the introduction of laparoscopic cholecystectomy. Dig Dis Sci 2007;52:
13131325.
15. Capron JP, Franco D. [Management of asymptomatic
lithiasis]. Rev Pract 1992;42:14741477.
16. Jorgensen T. Prevalence of gallstones in a Danish population. Am J Epidemiol 1987;126:912921.
17. Jorgensen T, Kay L, Schultz-Larsen K. The epidemiology
of gallstones in a 70-year-old Danish population. Scand J
Gastroenterol 1990;25:335340.
18. Jorgensen T. Treatment of gallstone patients. A health
technology assessment. Copenhagen: Danish Institute

19.
20.

21.

22.

23.

24.

25.

26.

27.

28.
29.

30.

for Health Technology Assessment and National Institute

of Public Health, 2000.
Jensen KH, Jorgensen T. Incidence of gallstones in a
Danish population. Gastroenterology 1991;100:790794.
Lynge E, Sandegaard JL, Rebolj M. The Danish
National Patient Register. Scand J Public Health 2011;
39:3033.
Fong Y, Dalal K. Diagnostic considerations in biliary and
liver disease. In: Fischer J, ed. Fishers mastery of surgery. Philadelphia, PA: Lippincott Williams & Wilkins,
2012:11631182.
National Sundheds-it SKS klassikations System.
SKS - download. Available at: www.ssi.dk/Sundheds
dataogit/National%20Sundheds-it/TerminologiOgKlassi
kationer/SKS/SKS_Download.aspx. Accessed December 17, 2014.
Gray RJ. A class of K-sample tests for comparing the
cumulative incidence of a competing risk. Ann Stat 1988;
16:11411154.
Peduzzi P, Concato J, Feinstein AR, et al. Importance of
events per independent variable in proportional hazards
regression analysis. II. Accuracy and precision of
regression estimates. J Clin Epidemiol 1995;48:
15031510.
Fox J. Cox proportional-hazards regression for survival
data. Available: http://cran.r-project.org/doc/contrib/FoxCompanion/appendix-cox-regression.pdf.
Accessed:
December 5, 2014.
Thistle JL, Cleary PA, Lachin JM, et al. The natural history
of cholelithiasis: the National Cooperative Gallstone
Study. Ann Intern Med 1984;101:171175.
Schein CJ, Hurwitt ES, Rosenblatt MA. The signicance
of calculus size in determining the indication for elective
cholecystectomy. Gastroenterology 1955;29:377380.
Portincasa P, Moschetta A, Palasciano G. Cholesterol
gallstone disease. Lancet 2006;368:230239.
Rome Group for the Epidemiology and Prevention of
Cholelithiasis. Radiologic appearance of gallstones and
its relationship with biliary symptoms and awareness of
having gallstones. Observations during epidemiological
studies. Rome Group for the Epidemiology and Prevention of Cholelithiasis (GREPCO). Dig Dis Sci 1987;
32:349353.
Brasca A, Berli D, Pezzotto SM, et al. Morphological and
demographic associations of biliary symptoms in subjects with gallstones: ndings from a population-based
survey in Rosario, Argentina. Dig Liver Dis 2002;34:
577581.

Natural History of Gallstones

167

31. Jorgensen T. Abdominal symptoms and gallstone

disease: an epidemiological investigation. Hepatology
(Baltimore, MD) 1989;9:856860.
32. Mechling RS, Watson JR. The solitary gallstone. Surg
Gynecol Obstet 1950;91:404408.
33. Schein CJ, Rosenblatt MA. Biliary calculus size as a
factor in prognosis of silent stones. Geriatrics 1957;
12:257260.
34. Carveth SW, Priestley JT, Gage RP. Size and number of
gallstones in acute and chronic cholecystitis. Proc Staff
Meet Mayo Clin 1959;34:371374.
35. Juvonen T, Niemela O, Makela J, et al. Characteristics of
symptomatic gallbladder disease in patients with either
solitary or multiple cholesterol gallstones. Hepatogastroenterology 1994;41:263266.
36. Persson GE. Expectant management of patients with
gallbladder stones diagnosed at planned investigation. A
prospective 5- to 7-year follow-up study of 153 patients.
Scand J Gastroenterol 1996;31:191199.
37. Csendes A, Becerra M, Rojas J, et al. Number and size
of stones in patients with asymptomatic and symptomatic gallstones and gallbladder carcinoma: a prospective study of 592 cases. J Gastrointest Surg 2000;
4:481485.
38. Venneman NG, Buskens E, Besselink MG, et al. Small
gallstones are associated with increased risk of
acute pancreatitis: potential benets of prophylactic
cholecystectomy? Am J Gastroenterol 2005;100:
25402550.
39. Andrews S. Gallstone size related to incidence of post
cholecystectomy retained common bile duct stones. Int J
Surg 2013;11:319321.

Received February 19, 2015. Accepted September 3, 2015.

Reprint requests
Address requests for reprints to: Daniel Mnsted Shabanzadeh, MD, Digestive
Disease Center, Bispebjerg University Hospital, Copenhagen, Denmark. e-mail:
dmshaban@gmail.com; fax: (45) 3531-3911.
Acknowledgments
The authors would like to acknowledge Anja Lykke Madsen for data management,
Mette Bach Mosebo and Karin Mnsted Shabanzadeh for linguistic proofreading, and, nally, Rikke Kart Jacobsen and Torben Martinussen for statistical
supervision.
Conicts of interest
The authors disclose no conicts.
Funding
Supported by the University of Copenhagen Faculty of Health and Medical
Sciences.

CLINICAL BILIARY

January 2016

167.e1
Shabanzadeh et al

Supplementary Table 1.Clustering of Diagnosis and Surgical Treatment Codes for the Denition of Complicated and Uncomplicated Gallstone Disease
Priority
1

Study outcome
Complicated gallstone disease

Included codes
Surgerya

Diagnosisb

Papillotomy
Stone extraction

47860, 47870, JKE02

47740, 48090, JKE12

Cholecystostomy
Endoscopic stent in bile duct
Endoscopic retrograde
cholangiopancreaticography
Acute cholecystitis

JKA10
JKE18
91050, UJK02

Combination surgery
and diagnosis

Pancreatitis
Cholecystectomy
Laparoscopic cholecystectomy
Chronic cholecystitis
cholecystectomy

57401, 57500, 57501, 57502,

57503, 57509, K800, K810, K819
57403, 57404, 57405, K803,
K804, K805B
57700, 57704, 57709, K859
47360, JKA20
47365, JKA21
57402, 57501, K801, K811
cholecystectomy

Diagnosisb

Cholecystolithiasis, cholelithiasis

57400, 57409, K802, K805, K808

Common bile duct stones

Uncomplicated gallstone
disease, cholecystectomy

Uncomplicated gallstone
disease, clinical diagnosis

Surgerya

Concomitant codes for:

1. Pancreatic, biliary,
or hepatic cancer
2. Hepatic metastasis

Concomitant codes for:

1. Complicated gallstone
disease
2. Pancreatic, biliary,
or hepatic cancer
3. Hepatic metastasis

Gastroenterology Vol. 150, No. 1

Nordic Classication of Surgical Procedures.

International Classication of Diseases, 8th or 10th revisions.

Exclusion criteria