Sie sind auf Seite 1von 10

American Journal of Epidemiology Vol. 163, No.

Copyright ª 2006 by the Johns Hopkins Bloomberg School of Public Health DOI: 10.1093/aje/kwj129
All rights reserved; printed in U.S.A. Advance Access publication March 29, 2006

Original Contribution

Peptic Ulcer Disease in a General Adult Population

The Kalixanda Study: A Random Population-based Study

Pertti Aro1, Tom Storskrubb1,2, Jukka Ronkainen1, Elisabeth Bolling-Sternevald1,3, Lars

Engstrand4, Michael Vieth5,6, Manfred Stolte6, Nicholas J. Talley7,8, and Lars Agréus1
Centre for Family Medicine, Karolinska Institutet, Stockholm, Sweden.
Kalix Hospital, Kalix, Sweden.
Astra Zeneca R&D, Mölndal, Sweden.
Institute of Infectious Disease Control, Stockholm, Sweden.
Institute of Pathology, Medical Faculty, University of Magdeburg, Magdeburg, Germany.
Institute of Pathology, Bayreuth, Germany.
Centre for Enteric Neurosciences and Translational Epidemiological Research, Division of Gastroenterology and
Hepatology, Mayo Clinic, Rochester, MN.
Department of Medicine, Faculty of Medicine, University of Sydney, Sydney, Australia.

Received for publication August 31, 2005; accepted for publication January 4, 2006.

The authors’ aim in this study was to explore the prevalence, symptomatology, and risk factors for peptic ulcer in
a general adult population. Between December 1998 and June 2001, the authors surveyed a random sample (n ¼
3,000) of the adult population (n ¼ 21,610) in two communities in northern Sweden using a validated question-
naire, the Abdominal Symptom Questionnaire (response rate ¼ 74%). A subsample (n ¼ 1,001) of the responders
was randomly invited to undergo esophagogastroduodenoscopy and symptom assessment (response rate ¼
73%). The prevalence of peptic ulcer was 4.1% (20 gastric ulcers and 21 duodenal ulcers). Nausea and gastro-
esophageal reflux were significant predictors of peptic ulcer disease, but epigastric pain/discomfort was not. Six
persons with gastric ulcer and two persons with duodenal ulcer were asymptomatic. Eight subjects with duodenal
ulcer (38%) lacked evidence of current Helicobacter pylori infection. Five (25%) of the gastric ulcers and four (19%)
of the duodenal ulcers were idiopathic (no use of aspirin or nonsteroidal antiinflammatory drugs, no H. pylori
infection). Smoking, aspirin use, and obesity were risk factors for gastric ulcer; smoking, low-dose (160 mg)
aspirin use, and H. pylori infection were risk factors for duodenal ulcer. Peptic ulcer disease often coexists with
atypical symptoms or no symptoms at all, and idiopathic duodenal ulcer may be more common than anticipated.

abdominal pain; adult; dyspepsia; endoscopy, digestive system; gastrointestinal diseases; Helicobacter pylori;
peptic ulcer

Abbreviations: CI, confidence interval; NSAID(s), nonsteroidal antiinflammatory drug(s); OR, odds ratio; PUD, peptic ulcer

In cross-sectional, population-based studies, up to every third symptoms is reported to be 10–20 percent (5), but the preva-
adult in the Western world reports dyspepsia (1–4). The preva- lence in the general population is unknown. The probability of
lence of peptic ulcer among patients with upper gastrointestinal finding peptic ulcer disease (PUD) or any organic cause as an

Reprint requests to Dr. Pertti Aro, Center for Family and Community Medicine, Karolinska Institutet, Alfred Nobels allé 12, S-141 52
Huddinge, Sweden (e-mail:

1025 Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
1026 Aro et al.

FIGURE 1. Design of a randomized, population-based endoscopic study of peptic ulcer disease (the Kalixanda Study), Kalix and Haparanda,
Sweden, December 1998–June 2001. ID, identification number; ASQ, Abdominal Symptom Questionnaire; EGD, esophagogastroduodenoscopy.

explanation for these symptoms is higher in secondary-care socioeconomic status of these communities was only
patients, who are usually referred for treatment, than in slightly below the Swedish average (18, 19).
primary-care patients (6). In two Scandinavian surveys of A representative sample of adults (n ¼ 3,000) was ran-
primary-care patients with dyspepsia (7, 8), 13 percent were domly selected from the computerized national population
found to have PUD, while 43 percent and 64 percent, respec- register and given an identification number (1–3,000) in
tively, had no obvious explanation and were considered to random order (figure 1) (17).
have functional dyspepsia. In an Italian study of secondary- Of the original study population (n ¼ 3,000), 140 persons
care patients, the prevalence of PUD was 31 percent, and only turned out to be ineligible for screening (21 were deceased,
21 percent of the patients were considered to have functional 17 had mental retardation or dementia, 87 had moved or had
dyspepsia (9); the corresponding figures in a US survey were an incorrect address, and 15 were ineligible for unclear
12 percent and 20 percent, respectively (10). The prevalence reasons) (17). Thus, after screening, the eligible study pop-
of PUD in the general population cannot be reliably esti- ulation consisted of 2,860 persons. These persons were
mated from studies of patients, because health-care-seeking sent a mailed questionnaire on abdominal symptoms, the
behavior is driven by factors other than symptoms (2, 11–15). Abdominal Symptom Questionnaire; 2,122 responded, pro-
The development of idiopathic ulcers (i.e., ulcers not viding us with a response rate of 74.2 percent (17).
caused by Helicobacter pylori infection, nonsteroidal anti- The aim was to perform esophagogastroduodenoscopy
inflammatory drugs (NSAIDs), or aspirin) is an increasing in one third (n ¼ 1,000) of the original study population.
problem (16). However, their prevalence in the general pop- These persons were first approached by mail and asked to
ulation is unknown. complete the Abdominal Symptom Questionnaire (11). The
Our aim in this study was to investigate the prevalence of responders were then invited by telephone to undergo
PUD, including idiopathic ulcers, and concomitant symptoms esophagogastroduodenoscopy in a random order. The study
and risk factors in a randomly selected adult population. population was divided into five parts for logistic reasons
(figure 1).
In order to complete 1,001 esophagogastroduodenoscopies,
MATERIALS AND METHODS 1,563 responders to the Abdominal Symptom Questionnaire
Setting, sampling, and study design had to be approached. The overall response rate for those
eligible for investigation was 73.3 percent (figure 1) (17).
The study setting consisted of two communities in north- Persons who agreed to participate filled in a more
ern Sweden with a total of 28,988 inhabitants (17). The comprehensive Abdominal Symptom Questionnaire at the

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
Peptic Ulcer Disease in the General Population 1027

esophagogastroduodenoscopy visit, and blood samples for The above definitions allowed concomitant reporting of
H. pylori serologic analysis and measurement of gastrin-17 symptoms of gastroesophageal reflux, dyspepsia, and irrita-
and pepsinogen-1 levels were taken (17). ble bowel syndrome.
The study protocol was approved by the Umeå University Gastric and duodenal ulcers. Ulcer was defined as a mu-
ethics committee, and the study was conducted in accor- cosal break at least 3 mm in diameter, with or without a ne-
dance with the Declaration of Helsinki. crotic base in the middle of the lesion, in either the stomach
(gastric) or the duodenum (duodenal). In the case of several
Endoscopy ulcers/erosions, at least one had to fulfill this definition.
Endoscopy was performed by three experienced endoscop- Histology and H. pylori infection
ists who participated in regular quality assessment programs.
Internal validity was assessed by means of consensus ses- Biopsy samples were stained with hematoxylin and eosin.
sions (17, 20). The three endoscopists were unaware of the H. pylori infection was histologically detected by means of
subjects’ symptoms before and during endoscopy. The Warthin-Starry silver staining (27). The histologic param-
endoscopy findings were recorded, and biopsies were taken eters of the gastric mucosa were assessed using the updated
from the cardia, the corpus, the angulus (except for the first Sydney System score definitions (28). Gastritis, including
200 subjects), and the antrum for histologic analysis. In addi- features of former H. pylori (minimal chronic inactive or
tion, biopsies were taken from the antrum and the corpus for ex-H. pylori) gastritis, was diagnosed according to the
H. pylori culture. Any visible lesions were also biopsied (17). method of Oberhuber et al. (29). Chemical-reactive gastritis
proposed to be caused by aspirin, NSAIDs, or bile reflux was
Questionnaire defined according to the updated Sydney System definitions
(28, 30, 31).
The Abdominal Symptom Questionnaire has been vali- Two experienced pathologists (M. V. and M. S.) evaluated
dated (11, 21, 22). The participants indicated (yes/no) the biopsies and gave a common report, and then a third
whether they had been troubled by any of the listed gastro- experienced pathologist (Dr. M. Walker, Imperial College
intestinal symptoms (n ¼ 27) or by any of 11 listed descrip- London, London, United Kingdom) reevaluated the biopsies
tors of abdominal pain or discomfort. Symptom frequency from 100 randomly chosen subjects. The kappa value for
(daily, weekly, or the past 3 months) was also recorded, agreement between observers in the evaluation of H. pylori
as was the participants’ medication use in the previous infection was 0.76 (95 percent confidence interval (CI): 0.56,
3 months (17). 0.96) for the corpus and 0.78 (95 percent CI: 0.59, 0.98)
for the antrum. The corresponding figures for granulocyte
Definitions infiltration were 0.57 (95 percent CI: 0.37, 0.76) and 0.73
Symptom groups. ‘‘Gastroesophageal reflux symptoms’’ (95 percent CI: 0.53, 0.93), respectively.
were defined as troublesome heartburn and/or acid regurgi- Samples taken from the antrum and corpus were cultured
tation over the past 3 months (23). and analyzed as described previously (27, 32).
‘‘Dyspepsia’’ was defined as troublesome pain or discom- Current H. pylori infection was defined as a positive cul-
fort, expressed as one or more of the 11 listed types of pain ture or histologic finding. There was overall agreement of
or discomfort, in the epigastric part of the abdomen, or 99.3 percent, with a kappa value of 0.96 (95 percent CI:
reporting of one or more of the symptoms ‘‘uncomfortable 0.94, 0.98) for agreement between the tests (27).
feeling of fullness,’’ ‘‘early satiety,’’ or ‘‘nausea,’’ in accor-
dance with the Rome II definition of dyspepsia (24). ‘‘Upper Serology
abdominal bloating’’ was not recorded. A simple definition The presence of H. pylori immunoglobulin G antibodies
of dyspepsia labeled ‘‘epigastric pain or discomfort,’’ based was determined by enzyme immunoassay (Pyloriset EIA-G;
on the Rome I definition of dyspepsia, was also used (25). Orion Diagnostica, Espoo, Finland) (33). A positive test in
‘‘Abdominal pain’’ was defined as troublesome pain or the absence of H. pylori detection by culture or histology
discomfort in the abdomen. It was expressed as one or more was considered indicative of past infection.
of the 11 listed types of pain or discomfort anywhere in the Levels of gastrin-17 (cutoff, 10 pmol/liter) and pepsinogen-1
abdomen. (cutoff, <25 lg/liter) were analyzed using specific enzyme
‘‘Irritable bowel syndrome’’ was defined as one or more immunoassays (Biohit Plc, Helsinki, Finland).
of the 11 listed types of abdominal pain or discomfort at any
site, combined with reported disturbances in bowel habits. Covariates
This definition has been shown to have good diagnostic
agreement with both the Manning criteria and the Rome I Use of aspirin and NSAIDs. All participants were thor-
criteria (26). oughly interviewed face to face regarding their medication
‘‘Atypical PUD symptoms’’ were defined as other gastro- use. Reported use of aspirin or NSAIDs for all subjects with
intestinal symptoms, except dyspepsia or ‘‘epigastric pain or idiopathic ulcers was rechecked by means of a telephone
discomfort’’ concomitant with PUD. interview and a review of the subjects’ medical records.
‘‘No symptoms or minor symptoms’’ were defined as in- Body mass index. Body mass index (weight (kg)/height
dividual symptoms not fulfilling any of the above symptom (m)2) was calculated and categorized according to World
classifications, or an absence of symptoms. Health Organization recommendations (34).

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
1028 Aro et al.

Statistical analysis 12
The significance of age and gender in the prevalence of 10 Men
both individual symptoms and combined symptoms (gastro-

No. with PUD

esophageal reflux symptoms, ‘‘epigastric pain or discom- 8
fort,’’ dyspepsia, abdominal pain, and irritable bowel
syndrome) was tested by applying a logistic regression 6
model. The significance of individual symptoms and com-
bined symptoms, H. pylori, use of acid-reducing drugs (ant-
acids, histamine-2 receptor antagonists, and proton pump 2
inhibitors), obesity, use of NSAIDs, use of aspirin, and
smoking in the risk of PUD was analyzed by applying a mul- 0
tivariate logistic regression model adjusting for age and GU GU GU GU DU DU DU DU
20–34 35–49 50–64 ≥65 20–34 35–49 50–64 ≥65
gender. The results are presented as odds ratios with 95
percent confidence intervals. The goodness of fit of the mod- Age group (years)
els was judged from the Pearson v2 test. The fit of the model
was considered acceptable if the p value was 0.05. Fisher’s FIGURE 2. Age and gender distribution of persons with ulcers in
a randomized, population-based endoscopic study of peptic ulcer
exact test was applied in appropriate cases. A two-sided disease (the Kalixanda Study), Kalix and Haparanda, Sweden,
p value less than 0.05 was regarded as statistically significant, December 1998–June 2001. PUD, peptic ulcer disease; GU, gastric
and 95 percent confidence intervals were computed using ulcer; DU, duodenal ulcer.
a logistic regression model. The Intercooled Stata 8 program
was used for the analyses (35).

with PUD. The age and gender distributions of subjects with

gastric and duodenal ulcers are shown in figure 2.
Original study population
Gastric cancer
The mean age in the original study population was 50.4
years, and 1,560 participants (52.0 percent) were men (as One 78-year-old woman who did not report alarm symp-
compared with 50.0 years and 49.7 percent in the corre- toms (i.e., difficulties in swallowing, weight loss, early satiety,
sponding Swedish population). or blood in the stool) was found to have an adenocarcinoma
upon histologic analysis in a benign-appearing gastric ulcer.
Symptoms at endoscopy and their relation to PUD
Altogether, 1,001 subjects had an esophagogastroduode-
noscopy performed. These subjects were slightly older than The 3-month prevalences of the 27 individual symptoms
the original study population (age 54.1 years vs. 50.4 years; are shown in table 1, and the 3-month prevalences of
p < 0.001); 488 (48.8 percent) were men. grouped symptoms (gastroesophageal reflux symptoms, epi-
gastric pain, dyspepsia, overall abdominal pain, and irritable
Peptic ulcer disease bowel syndrome) are shown in table 2. Thirty-three persons
with PUD (80.5 percent) reported symptoms. Nausea was
Twenty subjects (2.0 percent, 95 percent CI: 1.1, 2.9) had significantly associated with duodenal ulcer and PUD, as
gastric ulcer. A single ulcer was found in 12 subjects (60.0 were gastroesophageal reflux symptoms and dyspepsia.
percent); two ulcers were found in two subjects (10.0 per- Dyspepsia was the only weekly symptom associated
cent); and one subject (5.0 percent) had three ulcers, one with PUD (odds ratio (OR) ¼ 2.16, 95 percent CI: 1.11,
subject (5.0 percent) had four ulcers, and one subject (5.0 4.19). Daily abdominal pain was associated with duodenal
percent) had five ulcers. Three subjects (15.0 percent) had ulcer (OR ¼ 3.96, 95 percent CI: 1.39, 11.29) and with PUD
more than five ulcers. Fifteen of the subjects (75.0 percent) (OR ¼ 3.26, 95 percent CI: 1.49, 7.13).
had their ulcers located in the prepyloric/antral area, while Eleven subjects (1.1 percent, 95 percent CI: 0.5, 1.7)
four (20.0 percent) had ulcers in the middle of the stomach with PUD—four with gastric ulcer and seven with duodenal
at either the angulus (n ¼ 1) or the curvatura major (n ¼ 3). ulcer—reported ‘‘atypical PUD symptoms’’ but not dyspepsia
One subject (5.0 percent) had ulcers in both the fundus and or epigastric pain. Eight (72.7 percent) of these persons were
the antrum. The mean age of the subjects with gastric ulcer aged 50 years or more, and nine (81.8 percent) were women.
was 58.1 years. The prevalence of asymptomatic PUD was 0.8 percent
Duodenal ulcer. Twenty-one subjects (2.1 percent, 95 (95 percent CI: 0.2, 1.4) (six gastric ulcers and two duodenal
percent CI: 1.2, 3.0) were found to have duodenal ulcer. ulcers).
Fourteen (66.7 percent) had a single ulcer, five (23.8 percent)
had two ulcers, and two (9.5 percent) had three ulcers. The Risk and protective factors for PUD
mean age of the subjects with duodenal ulcer was 53.3 years.
No one had both gastric ulcer and duodenal ulcer. Thus, Of the 1,001 subjects in the esophagogastroduodeno-
there were 41 subjects (4.1 percent, 95 percent CI: 2.9, 5.3) scopy study, 62 had taken NSAIDs during the past 3 months,

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
TABLE 1. Three-month period prevalence (%) of individual gastrointestinal symptoms and their associations with age, gender, and peptic ulcer disease in a randomized,
Am J Epidemiol 2006;163:1025–1034
population-based endoscopic study of peptic ulcer disease (the Kalixanda Study), Kalix and Haparanda, Sweden, December 1998–June 2001

Men (48.8%) Women (51.2%) All subjects (n ¼ 1,001)

Ages Ages Ages Ages All peptic ulcer

Symptom Total Total Prevalence p valuey Gastric ulcer Duodenal ulcer
20–49 years 50–81 years 20–49 years 50–81 years 95% CI* disease
(n ¼ 488) (n ¼ 513) (%)
(n ¼ 178) (n ¼ 310) (n ¼ 188) (n ¼ 325) Age Sex OR* 95% CI OR 95% CI OR 95% CI
Loss of weight 0 1.3 0.8 4.8 2.8 3.5 2.2 1.3, 3.1 NS* 0.008z 5.10 1.18, 24.01 NS 3.99 1.11, 14.38
Poor appetite 5.7 1.3 2.9 7.5 2.5 4.3 3.6 2.4, 4.8 <0.001 NS NS NS NS
feeling of
fullness 21.5 8.1 13.0 24.3 20.7 22.0 17.6 15.2, 20.0 <0.001 <0.001z NS NS NS
Difficulties in
swallowing 2.8 7.8 6.0 7.5 7.7 7.6 6.8 5.2, 8.4 NS NS 3.35 1.09, 10.37 NS NS
Retching 26.6 19.1 21.8 20.9 28.5 25.7 23.8 21.2, 26.4 NS NS 2.58 1.05, 6.32 NS NS
Acid regurgitation 32.2 18.8 23.6 27.7 29.4 28.8 26.3 23.6, 29.0 0.02 NS NS NS 2.33 1.23, 4.41
Early satiety 11.4 9.4 10.1 17.6 14.2 15.4 12.9 10.8, 15.0 NS 0.01z NS NS NS
Nausea 18.1 6.8 10.9 20.9 13.6 16.2 13.6 11.5, 15.7 <0.001 0.02z 4.49 1.76, 11.45 NS 2.61 1.25, 5.45
Vomiting 4.5 1.9 2.9 4.3 3.4 3.7 3.3 2.2, 4.4 0.03 NS NS NS NS
Heartburn 36.4 27.0 30.4 39.0 34.3 36.0 33.3 30.4, 36.2 NS NS NS NS 2.34 1.24, 4.44
Pain behind the
breastbone 25.1 16.1 19.4 17.0 27.6 23.7 21.6 19.1, 24.1 NS NS NS NS NS
Burning feeling
rising§ 20.1 14.9 16.8 21.7 19.6 20.4 18.6 16.2, 21.0 0.02 NS NS NS NS
Constipation 11.4 17.7 15.4 34.8 30.3 32.0 23.9 21.3, 26.5 NS <0.001z NS NS NS
Diarrhea 34.4 22.8 27.9 25.7 19.3 21.8 24.8 22.1, 27.5 <0.001 0.03 NS NS NS
and diarrhea 10.3 9.5 9.8 21.4 13.0 16.1 13.0 10.9, 15.1 NS 0.004z NS NS NS
Feeling of incomplete

Peptic Ulcer Disease in the General Population

evacuation 24.3 21.1 22.3 37.0 35.4 36.0 29.3 26.5, 32.1 NS <0.001z NS NS NS
upon defecation 11.9 5.5 7.9 18.7 9.4 12.8 10.4 8.5, 12.3 <0.001 0.01z NS NS NS
relieved by
defecation 25.0 12.0 16.7 32.5 19.5 24.3 20.6 18.1, 23.1 <0.001 0.004z NS NS NS
Straining 14.1 18.9 17.2 34.0 30.1 31.6 24.6 21.9, 27.3 NS <0.001z NS NS NS
Urgency 23.7 18.8 20.6 23.7 21.9 22.5 21.6 19.1, 24.1 NS NS NS NS NS
Flatus 31.6 21.1 25.0 26.1 28.4 27.5 26.3 23.6, 29.0 NS NS NS NS NS
(gurgling sounds) 35.2 21.8 26.7 38.8 28.1 32.1 29.4 26.6, 32.2 <0.001 NS NS NS NS
distension 32.2 19.7 24.2 51.1 40.2 44.2 34.5 31.6, 37.4 <0.001 <0.001z NS NS NS
Urge to defecate
during the night 9.6 5.5 7.0 5.4 6.5 6.1 6.5 5.0, 8.0 NS NS NS 3.53 1.15, 10.88 2.67 1.07, 6.61
Black stools 1.7 1.0 1.2 4.8 1.0 2.4 1.8 1.0, 2.6 0.003 NS NS NS NS
Blood stains in stool 10.9 5.8 7.6 8.0 6.5 7.1 7.3 5.7, 8.9 0.02 NS NS NS NS
Mucus 5.7 5.2 5.4 9.6 12.4 11.4 8.5 6.8, 10.2 NS 0.001z NS NS NS

* CI, confidence interval; OR, odds ratio; NS, not significant.

y p value from logistic regression analysis (significance level: p < 0.05).

z More common in women.
§ A burning feeling rising from the stomach or lower chest towards the neck.

Downloaded from

by guest
on 19 August 2018
1030 Aro et al.

two had used a cyclooxygenase-2 inhibitor, 107 had taken

1.3, 4.68

1.10, 3.95
All peptic ulcer

95% CI
aspirin, and 108 had taken acetaminophen. Of the persons
TABLE 2. Three-month period prevalence (%) of grouped gastrointestinal symptoms and their associations with age, gender, and peptic ulcer disease in a randomized,

consuming aspirin, 59 had used low-dose aspirin (160 mg/
day) and 48 had used standard-dose aspirin (>160 mg/day)


either daily (n ¼ 11) or on demand (n ¼ 37). Two subjects


had used bisphosphonates but did not have PUD.

1.06, 6.30

1.01, 6.08
Antacids had been taken by 115 subjects during the pre-
Duodenal ulcer

95% CI

vious 3 months, histamine-2 receptor antagonists had been

taken by 31 subjects, proton pump inhibitors had been taken
by 49 subjects, and any of the above had been taken by 190



subjects. The corresponding numbers of subjects who had
used these drugs during the week before esophagogastro-
population-based endoscopic study of peptic ulcer disease (the Kalixanda Study), Kalix and Haparanda, Sweden, December 1998–June 2001

All subjects (n ¼ 1,001)

95% CI
Gastric ulcer

duodenoscopy were 55, 14, 36, and 102, respectively.

A total of 187 subjects smoked cigarettes, and 118 used
moist snuff; 22 persons used both. In total, 339 esophago-




gastroduodenoscopy subjects (33.9 percent) were H. pylori-

positive upon culture and/or histologic analysis. Of the 20


persons with gastric ulcer, 10 (50.0 percent) were H. pylori-

p valuey

positive, as were 13 (61.9 percent) of the 21 persons with

duodenal ulcer.


Subjects who had used acid-reducing drugs during the


past 3 months had an increased risk of PUD (OR ¼ 2.37,

95 percent CI: 1.16, 4.86). Smoking, obesity, and overall
37.0, 43.0

18.4, 23.4
34.6, 40.6

26.8, 32.4
48.5, 54.5
95% CI*

aspirin intake were independent risk factors for gastric ulcer

(OR ¼ 3.12 (95 percent CI: 1.13, 8.64), OR ¼ 4.15 (95
percent CI: 1.31, 13.13), and OR ¼ 7.44 (95 percent CI:

2.78, 19.93), respectively). Smoking, overall aspirin intake,




and H. pylori infection were independent risk factors for

duodenal ulcer (OR ¼ 2.84 (95 percent CI: 1.11, 7.27),
OR ¼ 4.28 (95 percent CI: 1.52, 12.10), and OR ¼ 3.56
(n ¼ 513)

(95 percent CI: 1.40, 9.09), respectively).





The presence of esophagitis was an independent risk fac-

tor for duodenal ulcer (OR ¼ 3.39, 95 percent CI: 1.17,
9.86) and PUD (OR ¼ 3.47, 95 percent CI: 1.57, 7.69).
Women (51.2%)

50–81 years
(n ¼ 325)

Low-dose aspirin use was an independent risk factor for





both gastric ulcer (OR ¼ 8.88, 95 percent CI: 2.64, 29.88)

and duodenal ulcer (OR ¼ 9.38, 95 percent CI: 2.71, 32.46),
while standard-dose aspirin use was a risk factor for gastric
20–49 years
(n ¼ 188)

ulcer only (OR ¼ 4.85, 95 percent CI: 1.25, 18.83). Use of

y p value from logistic regression analysis (significance level: p < 0.05).




NSAIDs or acetaminophen did not change the outcome.

One person with gastric ulcer (5.0 percent), a 57-year-old
woman, had taken NSAIDs, and eight persons with gastric
(n ¼ 488)

* CI, confidence interval; OR, odds ratio; NS, not significant.

ulcer (40.0 percent) had taken aspirin. None of the subjects





with duodenal ulcer had used NSAIDs; six (28.6 percent)

had used aspirin.
Fifty-nine persons who underwent esophagogastroduode-
50–81 years
Men (48.8%)

(n ¼ 310)

noscopy (5.9 percent) reported former, previously treated





PUD (28 gastric ulcers, 21 duodenal ulcers, and 10 with

no given localization) before the study started, and 15 of
them had received H. pylori eradication therapy. Seven of
20–49 years
(n ¼ 78)

these 59 subjects had PUD (four gastric ulcers, three duo-





denal ulcers) in this study, and none had received eradica-

z More common in women.

tion therapy before.

reflux symptoms
Symptom group


Idiopathic ulcers
Abdominal pain
Irritable bowel

Altogether, five (25.0 percent) of the persons with gastric


ulcer and four (19.0 percent) of the persons with duodenal ulcer
were found to have no known risk factors (NSAID/aspirin use
or H. pylori infection) for PUD, and hence their cases were

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
Peptic Ulcer Disease in the General Population 1031

considered idiopathic. The prevalence of idiopathic PUD was alence of most symptoms was higher than has been reported
0.9 percent (95 percent CI: 0.3, 1.5), and six of the nine before in Sweden (22), and it was also higher than in some
subjects (0.6 percent, 95 percent CI: 0.1, 1.1) did not have investigations conducted elsewhere (38–40), although prev-
histologic signs of former H. pylori infection or serologic alence rates of a similar magnitude have been reported by
evidence of former H. pylori infection. Five of them had other investigators (41). Atypical symptoms in PUD pa-
chemical-reactive gastritis in the antrum, and one had normal tients, especially among the elderly, have also been reported
histology. None of the nine subjects had any antral granulo- before (42).
cyte activity, but one of them had the lowest degree of activity There is debate as to whether the concept of dividing
in the corpus. Only four of the nine subjects with idiopathic dyspepsia symptoms into ‘‘ulcer-like’’ and ‘‘dysmotility-
ulcer smoked, and one had an elevated gastrin-17 level (76 like’’ symptoms is valid (24). The proportion of patients
pmol/liter) but a low pepsinogen-1 level (7.2 lg/liter), sug- with PUD has been found to be approximately the same in
gesting a low gastric acid output. both symptom groups (10, 43), suggesting that those symp-
There was no significant association between idiopathic tom profiles are not useful predictors of PUD. Our study
PUD and gastroesophageal reflux symptoms, epigastric supports the concept that ‘‘epigastric pain or discomfort’’
pain, dyspepsia, irritable bowel syndrome, obesity, or smok- does not predict PUD, while the dysmotility-like symptom
ing. The only individual symptoms significantly associated nausea was a weak predictor. We also found, as have other
with idiopathic PUD were stated weight loss (p ¼ 0.015) investigators (10), that PUD was common in patients with
and loss of appetite (p ¼ 0.041) (Fisher’s exact test). gastroesophageal reflux symptoms. This suggests that treat-
ing all patients with symptomatic gastroesophageal reflux
DISCUSSION empirically by acid suppression may not represent optimal
management. In addition, we found that 34 percent of sub-
In this randomly selected population of adults aged 20 jects with unknown PUD were taking acid-reducing drugs.
years or more, we found a point prevalence of 4 percent for The high risk of gastric ulcer among obese people has not
PUD, but the symptomatology did not conform to a classical been described before, to our knowledge. H. pylori infec-
pattern; that is, ‘‘epigastric pain or discomfort’’ alone did tion, use of NSAIDs or aspirin, serum gastrin-17 level, and
not predict PUD, while nausea and gastroesophageal reflux smoking habits did not appear to explain this observation. It
symptoms did, as did loss of weight. An unexpectedly high remains unknown whether higher acid secretion rates, in-
proportion of persons with duodenal ulcer were H. pylori- creased stress, or mechanical factors could explain excess
negative; 25.0 percent of the gastric ulcers and 19.0 percent gastric ulcer disease in obese persons.
of the duodenal ulcers were idiopathic. Continuous use of The reason why some ulcers are asymptomatic is un-
low-dose aspirin was a risk factor for PUD. In addition, known. In controlled trials, both asymptomatic duodenal
obesity was a risk factor for gastric ulcer. ulcers and gastric ulcers have been found in a substantial
The subjects who underwent endoscopy had a mean age proportion of cases (44, 45). We found that eight subjects
that was approximately 4 years higher than the mean ages of (19.5 percent of all ulcer patients and 0.8 percent of the
the original adult study population and the Swedish general study population) had asymptomatic PUD. Similarly, in
population. Most of the difference was due to a lower re- the Sørreisa Gastrointestinal Disorder Study, 1 percent of
cruitment rate among the youngest quarter of participants persons who underwent upper endoscopy had asymptomatic
(age <35 years) and the fact that symptomatic subjects un- PUD (39). The clinical implications of asymptomatic peptic
der 50 years of age were somewhat more willing to respond ulcers in our study are uncertain; evaluation of this issue
than asymptomatic people (17). This selection bias might would require a longitudinal natural history study.
have caused a slight overestimation of PUD prevalence. To our knowledge, there have been no large-scale studies
The upper age limit of 80 years in this study was decided of PUD in a randomly selected adult population that have
by the ethical committee because of the risk of complica- had a satisfactory participation rate (46). The Sørreisa Gastro-
tions during esophagogastroduodenoscopy and concerns intestinal Disorder Study, from the 1980s, was population-
about obtaining informed consent. based (47), but a case-control design was applied for the
The two study communities, located in northern Sweden, endoscopies. Those investigators found prevalences of
have a slightly lower socioeconomic status than the Swedish PUD of 4 percent among controls and 8 percent among
average (17–19). Most other relevant studies have shown that persons with dyspepsia (47). The only other comparable
socioeconomic differences of similar magnitude do not affect study was performed in Sweden (48) and had a participation
gastrointestinal morbidity (17). H. pylori prevalence, another rate of 25 percent; 3 percent of those subjects had current
indirect indicator of socioeconomic status, decreases with PUD, and a further 3 percent had evidence of past ulcers.
greater prosperity. The prevalence of H. pylori seropositivity Few other data are available. Ihamäki et al. (49) found a less
in this study was 43.0 percent, and this is comparable with than 2 percent prevalence of PUD and a 4 percent preva-
other Northern European countries (36). We do not know of lence of duodenal scars among healthy controls matched to
any population-based data from Northern Europe on the cancer patients. Khuroo et al. (50) found a point prevalence
prevalence of current infection, which was 34 percent in this of 4.7 percent for PUD and a lifetime prevalence of 11.2
study, although available data from Southern Europe show percent in a population-based case-control study; most of
prevalence that is markedly higher (37). those subjects had duodenal ulcer. Among monks, Katelaris
The minimum requirement for symptom reporting in this et al. (51) found a 6 percent prevalence of duodenal ulcer,
study was that symptoms should be troublesome. The prev- a 2 percent prevalence of gastric ulcer, and a 7 percent

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
1032 Aro et al.

prevalence of prepyloric or duodenal deformity. Lond et al. Stockholm, Sweden) for valuable help with the statistical
(52) found a prevalence of 9 percent for duodenal ulcers and analysis; Else-Maj Sundbaum-Lomakka, Åsa Storskrubb,
4 percent for gastric ulcers in a random population sample of Dr. Timo Tanner, and Dr. Bo Wikström for their partic-
persons reporting dyspepsia. In a recent preliminary report ipation and skillful help; Dr. Madeline Frame for as-
from Italy (53), the prevalence of PUD among adults was sistance with manuscript preparation; the employees of
4.5 percent, and one third (six of the 18 persons with duo- Biohit Plc (Helsinki, Finland) for analyzing gastrin-17 and
denal ulcer and four of the 12 persons with gastric ulcer) pepsinogen-1 levels; and Dr. Marjorie Walker (Department
were asymptomatic. of Histopathology, Faculty of Medicine, Imperial College
The proportion of H. pylori-negative duodenal ulcers was London, London, United Kingdom) for reevaluating the
surprising and worrying, since such results could alter cur- results of histologic analysis.
rent dyspepsia management algorithms. A rising proportion Dr. Elisabeth Bolling-Sternevald is employed by Astra-
of idiopathic ulcers among patients has been shown in recent Zeneca AB. Dr. Lars Agréus receives a research allowance
studies (16, 54–56). Lanas et al. (57) have shown that the from the Karolinska Institutet.
number of idiopathic ulcers may be overestimated if the
participants with PUD underreport aspirin or NSAID use.
In our study, aspirin or NSAID use for subjects with idio-
pathic ulcers was double-checked through an extra tele- REFERENCES
phone interview and a review of those subjects’ medical
records. In total, use of either aspirin or NSAIDs during 1. Jones R, Lydeard SE, Hobbs FD, et al. Dyspepsia in England
and Scotland. Gut 1990;31:401–5.
the past 3 months was confirmed for 169 subjects (16.9
2. Talley NJ, Zinsmeister AR, Schleck CD, et al. Dyspepsia
percent), which is consistent with estimated use for the re- and dyspepsia subgroups: a population-based study. Gastro-
gion (58). A small possibility of underreporting of NSAID enterology 1992;102:1259–68.
use in older women remains, but this is unlikely. 3. Talley N, Boyce P, Jones M. Identification of upper and lower
H. pylori infection and use of NSAIDs are well-recognized gastrointestinal symptom groupings in an urban population.
causes of PUD, and some studies have suggested a syner- Gut 1998;42:690–5.
gistic effect of these risk factors (59). We did not identify 4. Agréus L, Svärdsudd K, Talley NJ, et al. Natural history of
any synergism in the development of PUD. A higher risk gastroesophageal reflux disease and functional abdominal
of both gastric ulcer and duodenal ulcer with low-dose as- disorders: a population-based study. Am J Gastroenterol
pirin use as compared with use of standard-dose aspirin can 2001;96:2905–14.
probably be explained by the continuous use of low-dose 5. Talley NJ, Silverstein MD, Agréus L, et al. AGA technical
review: evaluation of dyspepsia. American Gastroenterologi-
cal Association. Gastroenterology 1998;114:582–95.
Another indicator of aspirin or NSAID use can be antral 6. Agréus L, Talley NJ. Dyspepsia: current understanding and
chemical-reactive gastritis (30), although this can also be management. Annu Rev Med 1998;49:475–93.
caused by bile reflux or excessive alcohol consumption 7. Heikkinen M, Pikkarainen P, Takala J, et al. Etiology of
(60). Earlier reports suggested that the sensitivity of this dyspepsia: four hundred unselected consecutive patients in
histologic finding for NSAID use is 73 percent (30), but in general practice. Scand J Gastroenterol 1995;30:519–23.
this study only 32 percent of our 169 subjects with reported 8. Kagevi I, Löfstedt S, Persson LG. Endoscopic findings and
use of aspirin or NSAIDs during the previous 3 months diagnoses in unselected dyspeptic patients at a primary health
had chemical-reactive gastritis. The three cases in which care center. Scand J Gastroenterol 1989;24:145–50.
H. pylori was detected by serologic analysis but not by 9. Stanghellini V, Tosetti C, Paternico A, et al. Risk indicators
histologic analysis or culture are another potential source of delayed gastric emptying of solids in patients with func-
of bias. None of those subjects had any granulocyte activity tional dyspepsia. Gastroenterology 1996;110:1036–42.
10. Talley NJ, Weaver AL, Tesmer DL, et al. Lack of discriminant
to indicate an ongoing but hidden infection. value of dyspepsia subgroups in patients referred for upper
We conclude that both gastric ulcer and duodenal ulcer endoscopy. Gastroenterology 1993;105:1378–86.
are common in the general population and that persons with 11. Agréus L, Svärdsudd K, Nyrén O, et al. Reproducibility and
ulcer frequently show atypical symptoms. In this general validity of a postal questionnaire. The Abdominal Symptom
adult population from Sweden, 22 percent of all cases of Study. Scand J Prim Health Care 1993;11:252–62.
PUD were idiopathic, and almost 40 percent of duodenal 12. Lydeard S, Jones R. Factors affecting the decision to consult
ulcers were H. pylori-negative. Obesity was a risk factor for with dyspepsia: comparison of consulters and non-consulters.
gastric ulcer, as was use of low-dose aspirin for PUD. J R Coll Gen Pract 1989;39:495–8.
13. Howell S, Talley NJ. Does fear of serious disease predict
consulting behaviour amongst patients with dyspepsia in
general practice? Eur J Gastroenterol Hepatol 1999;11:881–6.
ACKNOWLEDGMENTS 14. Talley NJ, Fung LH, Gilligan IJ, et al. Association of anxiety,
neuroticism, and depression with dyspepsia of unknown cause.
This study was supported in part by the Swedish Research A case-control study. Gastroenterology 1986;90:886–92.
Council, the Swedish Society of Medicine, Mag-Tarm Sjukas 15. Kane F Jr, Strohlein J, Harper RG. Nonulcer dyspepsia asso-
Förbund, the Norrbotten County Council (Norrbotten County, ciated with psychiatric disorder. South Med J 1993;86:641–6.
Sweden), and AstraZeneca R&D (Mölndal, Sweden). 16. Quan C, Talley NJ. Management of peptic ulcer disease not
The authors thank Dr. Sven-Erik Johansson (Center for related to Helicobacter pylori or NSAIDs. Am J Gastroenterol
Family and Community Medicine, Karolinska Institutet, 2002;97:2950–61.

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
Peptic Ulcer Disease in the General Population 1033

17. Aro P, Ronkainen J, Storskrubb T, et al. Valid symptom re- asymptomatic subjects in 17 populations. Gut 1993;34:
porting at upper endoscopy in a random sample of the Swedish 1672–6.
adult general population. The Kalixanda Study. Scand J 37. Bazzoli F, Palli D, Zagari RM, et al. The Loiano-Monghidoro
Gastroenterol 2004;39:1280–8. population-based study of Helicobacter pylori infection:
18. Central Statistical Bureau, Statistics Sweden. Snabbfakta om prevalence by 13C-urea breath test and associated factors.
Kalix. Stockholm, Sweden: Statistics Sweden, 2002. (http:// Aliment Pharmacol Ther 2001;15:1001–7. 38. Jones R, Lydeard S. Prevalence of symptoms of dyspepsia in
19. Central Statistical Bureau, Statistics Sweden. Fakta om the community. BMJ 1989;298:30–2.
Haparanda. Stockholm, Sweden: Statistics Sweden, 2002. 39. Bernersen B, Johnsen R, Straume B, et al. Towards a true
( prevalence of peptic ulcer: The Sørreisa Gastrointestinal
20. Ronkainen J, Aro P, Storskrubb T, et al. High prevalence of Disorder Study. Gut 1990;31:989–92.
gastroesophageal reflux symptoms and esophagitis with or 40. Talley NJ, Holtmann G, Agréus L, et al. Gastrointestinal
without symptoms in the general adult Swedish population: a symptoms and subjects cluster into distinct upper and lower
Kalixanda Study report. Scand J Gastroenterol 2005;40:275–85. groupings in the community: a four nations study. Am J
21. Agréus L. The Abdominal Symptom Study: an epidemiolog- Gastroenterol 2000;95:1439–47.
ical survey of gastrointestinal and other abdominal symptoms 41. Heading RC. Prevalence of upper gastrointestinal symptoms
in the adult population of Östhammar, Sweden. (Doctoral in the general population: a systematic review. Scand J
thesis). Uppsala, Sweden: Uppsala University, 1993. Gastroenterol Suppl 1999;231:3–8.
22. Agréus L, Svärdsudd K, Nyrén O, et al. Irritable bowel syn- 42. Pound SE, Heading RC. Diagnosis and treatment of dyspepsia
drome and dyspepsia in the general population: overlap and in the elderly. Drugs Aging 1995;7:347–54.
lack of stability over time. Gastroenterology 1995;109:671–80. 43. Bytzer P, Hansen JM, Havelund T, et al. Predicting endo-
23. Klauser AG, Schindlbeck NE, Müller-Lissner SA. Symptoms scopic diagnosis in the dyspeptic patient: the value of clinical
in gastro-oesophageal reflux disease. Lancet 1990;335:205–8. judgement. Eur J Gastroenterol Hepatol 1996;8:359–63.
24. Talley NJ, Stanghellini V, Heading RC, et al. Functional gas- 44. Gibinski K, Rybicka J, Nowak A, et al. Seasonal occurrence
troduodenal disorders. Gut 1999;45(suppl 2):II37–42. of abdominal pain and endoscopic findings in patients with
25. Talley NJ, Colin-Jones D, Koch KJ, et al. Functional dys- gastric and duodenal ulcer disease. Scand J Gastroenterol
pepsia: a classification with guidelines for diagnoses and 1982;17:481–5.
management. Gastroenterol Int 1991;4:145–60. 45. Jorde R, Bostad L, Burhol PG. Asymptomatic gastric ulcer:
26. Agréus L. Rome? Manning? Who cares? (Editorial). Am J a follow-up study in patients with previous gastric ulcer dis-
Gastroenterol 2000;95:2679–81. ease. Lancet 1986;1:119–21.
27. Storskrubb T, Aro P, Ronkainen J, et al. A negative Helico- 46. Bytzer P. Gastro-oesophageal reflux disease: epidemiologic
bacter pylori serology test is more reliable for exclusion of challenges. Scand J Gastroenterol 2005;40:247–9.
premalignant gastric conditions than a negative test for current 47. Johnsen R, Bernersen B, Straume B, et al. Prevalences of
H. pylori infection: a report on histology and H. pylori de- endoscopic and histological findings in subjects with and
tection in the general adult population. Scand J Gastroenterol without dyspepsia. BMJ 1991;302:749–52.
2005;40:302–11. 48. Borch K, Jonsson KA, Petersson F, et al. Prevalence of gas-
28. Dixon MF, Genta RM, Yardley JH, et al. Classification and troduodenitis and Helicobacter pylori infection in a general
grading of gastritis. The updated Sydney System. International population sample: relations to symptomatology and life-style.
Workshop on the Histopathology of Gastritis, Houston 1994. Dig Dis Sci 2000;45:1322–9.
Am J Surg Pathol 1996;20:1161–81. 49. Ihamäki T, Varis K, Siurala M. Morphological, functional and
29. Oberhuber G, Puspok A, Dejaco C, et al. Minimal chronic immunological state of the gastric mucosa in gastric carci-
inactive gastritis: indicator of pre-existing Helicobacter pylori noma families. Comparison with a computer-matched family
gastritis? Pathol Res Pract 1996;192:1016–21. sample. Scand J Gastroenterol 1979;14:801–12.
30. Stolte M, Panayiotou S, Schmitz J. Can NSAID/ASA-induced 50. Khuroo MS, Mahajan R, Zargar SA, et al. Prevalence of peptic
erosions of the gastric mucosa be identified at histology? ulcer in India: an endoscopic and epidemiological study in
Pathol Res Pract 1999;195:137–42. urban Kashmir. Gut 1989;30:930–4.
31. Dixon MF, Mapstone NP, Neville PM, et al. Bile reflux gastritis 51. Katelaris PH, Tippett GH, Norbu P, et al. Dyspepsia, Helico-
and intestinal metaplasia at the cardia. Gut 2002;51:351–5. bacter pylori, and peptic ulcer in a randomly selected popu-
32. Enroth H, Nyren O, Engstrand L. One stomach–one strain: lation in India. Gut 1992;33:1462–6.
does Helicobacter pylori strain variation influence disease 52. Lond E, Litvinenko T, Hermlin E. Main digestive diseases
outcome? Dig Dis Sci 1999;44:102–7. diagnosed for the first time in a rural district. (Abstract A80).
33. Lerang F, Moum B, Mowinckel P, et al. Accuracy of seven Gut 1994;35(suppl 4):1194.
different tests for the diagnosis of Helicobacter pylori infec- 53. Zagari RM, Pozzato P, Nicolini G, et al. Prevalences of
tion and the impact of H2-receptor antagonists on test results. asymptomatic endoscopic lesions of the upper gastrointestinal
Scand J Gastroenterol 1998;33:364–9. tract. Preliminary results of the Loiano-Monghidoro popula-
34. World Health Organization. Obesity: preventing and managing tion study. (Abstract). Gastroenterology 2002;122(suppl 1):
the global epidemic. Report of a WHO consultation on obesity, A-208.
3–5 June 1997, Geneva, WHO/NUT/NCD/98.1. Geneva, 54. Ciociola AA, McSorley DJ, Turner K, et al. Helicobacter
Switzerland: World Health Organization, 1997. (http://www. pylori infection rates in duodenal ulcer patients in the United States may be lower than previously estimated. Am J
pdf). Gastroenterol 1999;94:1834–40.
35. Stata Corporation. The Intercooled Stata 8 program. College 55. Xia HH, Wong BC, Wong KW, et al. Clinical and endoscopic
Station, TX: Stata Corporation, 2003. characteristics of non-Helicobacter pylori, non-NSAID
36. The EUROGAST Study Group. Epidemiology of, and risk duodenal ulcers: a long-term prospective study. Aliment
factors for, Helicobacter pylori infection among 3194 Pharmacol Ther 2001;15:1875–82.

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018
1034 Aro et al.

56. Konturek SJ, Bielanski W, Plonka M, et al. Helicobacter 58. Apoteket AB. National Corporation of Swedish Pharmacies
pylori, non-steroidal anti-inflammatory drugs and smoking in drug sales statistics for year 2000. Stockholm, Sweden:
risk pattern of gastroduodenal ulcers. Scand J Gastroenterol Apoteket AB, 2004.
2003;38:923–30. 59. Huang JQ, Sridhar S, Hunt RH. Role of Helicobacter pylori
57. Lanas A, Sekar MC, Hirschowitz BI. Objective evidence infection and non-steroidal anti-inflammatory drugs in peptic-
of aspirin use in both ulcer and nonulcer upper and lower ulcer disease: a meta-analysis. Lancet 2002;359:14–22.
gastrointestinal bleeding. Gastroenterology 1992;103: 60. Borchard F. Chemical-reactive gastritis. (In German).
862–9. Pathologe 2001;22:44–55.

Am J Epidemiol 2006;163:1025–1034

Downloaded from

by guest
on 19 August 2018