Incidence of duodenal ulcers and gastric ulcers in a Western population : Back to where it started

1Department of Gastroenterology and Hepatology, Erasmus Medical Centre, Rotterdam; 2Department of Gastroenterology and Hepatology, Ikazia Hospital, Montessoriweg; 3Department of Epidemiology and Biostatistics, Erasmus Medical Centre Rotterdam, Rotterdam, The Netherlands Correspondence: Dr Marcel JM Groenen, Department of Gastroenterology and Hepatology, Rijnstate Hospital, Wagnerlaan 55, 6815 AD Arnhem, The Netherlands. Telephone 31-88-005-8952, fax 31-88-005-6737, e-mail mgroenen@alysis.nl Received for publication August 19, 2007. Accepted January 31, 2009 Over the past decades, important changes have occurred in the epidemiology of ulcer disease. The peak incidence was seen among individuals born at the end of 19th century. The peak for duodenal ulcers followed the gastric ulcer peak, with a 10-year to 20-year lag (1). As recently as 40 years ago, a decline in the incidence of peptic ulcers was suggested, based on a cohort analysis (2) showing that the mortality due to peptic ulcers decreased in successive birth cohorts. This observation was supported by decreasing hospitalization rates for peptic ulcers in the Unites States as well as in many European countries over the past decades (3,4). The discovery of Helicobacter pylori in the early 1980s as a major cause of peptic ulcer disease had a significant impact on the treatment of ulcer disease. The significance of the discovery led to the award of the 2005 Nobel Prize in Medicine to Robin Warren and Barry Marshall. H pylori eradication therapy was proven to cure patients with previous chronic, recurrent ulcer disease. Further research into the epidemiology of H pylori infection showed that the prevalence of this bacterium was decreasing over time in recent decades, presumably as a result of improvements in living conditions (5,6). A recent Asian study (7) showed that the prevalence of peptic ulcer disease – mainly duodenal ulcers – was reduced in association with a decreasing trend in the prevalence of H pylori, confirming that H pylori is especially involved in the development of duodenal ulcers. orIgInal artIcle

With the decreased incidence of ulcer disease, the incidence of ulcer complications may have been affected as well.A recent study in the Netherlands (8) showed that the incidence of the most important complication, ulcer bleeding, remained stable in the period between 1993 and 2003, notwithstanding the decreasing incidence of peptic ulcers.Mortality and rebleeding rates for peptic ulcer bleeding did not change either.Investigators from England (9) reported that the incidence of gastric and duodenal ulcer hemorrhage increased, even among older patients.This rise was shown to be correlated with oral anticoagulants, nonsteroidal anti-inflammatory drugs (NSAIDs) and oral corticosteroid intake (10,11).
The introduction of an endoscopy database allowed for closer investigation of the incidence and epidemiology of gastric and duodenal ulcers, complication rates and classifications.We aimed to define changes in the incidence of gastric and duodenal ulcer disease, and its complications over a 10-year period in a large, uninvestigated, dyspeptic population in the Netherlands.Moreover, use of the database enabled us to describe a more detailed incidence of ulcer complications over time in a large group of peptic ulcer patients for the first time, with the use of the 'Forrest' classification (12).

METHODs
All gastrointestinal endoscopies performed between June 1996 and January 2006, at the Ikazia Hospital (Rotterdam, The Netherlands), were evaluated.The Ikazia Hospital is a district hospital with an open-access endoscopy unit, where approximately 40% of the patients are referred by general practitioners.The large percentage of referrals by general practitioners from both the urban agglomeration and rural sites makes the investigated population a cross-section of the Dutch population.In July 1995, a computerized endoscopy report system, Endobase III, developed by Olympus Software (Olympus, Japan), was introduced.From June 1996 to present, each gastrointestinal endoscopy performed in the centre was recorded with this system.Four experienced endoscopists performed all endoscopy procedures recorded in the database, which enabled standardized methods for report writing (13).All data were labelled with a specific gastrointestinal endoscopic terminology code, which is an extension of the International Classification of Diseases -10th Revision (ICD-10) (14).
All codes in the database starting with 'K25' and 'K26', coding for gastric and duodenal ulcer, respectively, were included in the present study.Furthermore, a search for the word "ulcer" in all reports of upper gastrointestinal endoscopies in the database was performed.These reports were reviewed for the diagnosis of gastric and duodenal ulcers.Many patients with a gastric ulcer underwent a follow-up gastroscopy a few weeks later.These follow-up endoscopies were excluded from the analysis; consequently, all included cases were newly diagnosed ulcers in a previously uninvestigated dyspeptic population.
Patient characteristics such as age, sex and indication for endoscopy were retrieved from the database.Furthermore, the Forrest classification and H pylori status were investigated.All of the ulcers detected were classified according to the Forrest classification (12).From 2001 to present, the Endobase III program enabled entering the Forrest classification into the database, and was completed for all peptic ulcers.Ulcer cases entered before 2001 were reclassified using the Forrest classification based on the reports and available endoscopic images.

H pylori status was based on gastric biopsy specimens assessed with a rapid urease test (Campylobacter-like organism [CLO] test).
Data were analyzed with SPSS version 11.5 (SPSS Inc, USA).Regression analysis of ulcer incidence over time (years) was plotted against the observed incidence of ulcers in percentage per year.Data are described by mean ± SD or 95% CIs.A P<0.05 (based on a two-sided test) was considered to be statistically significant.Student's t test and linear regression analysis were used to compare groups and linear relations.

REsULTs
In total, the data from 20,006 upper gastrointestinal endoscopies were accessible for analysis, of which 10,784 (53.9%) were performed in women.The mean (± SD) age of the patients was 55.6±17.9years.The total number of endoscopies performed yearly increased gradually over the 10 years, from 1800 to 2200 per year.
Duodenal ulcers were found in 696 of the investigated patients (3.5%) and gastric ulcers in 487 patients (2.4%).The majority of ulcer patients were men, who comprised 57.6% of duodenal ulcer and 51.5% of gastric ulcer patients.
Duodenal ulcer patients were significantly younger than gastric ulcer patients.The mean age of duodenal ulcer patients was 57.4 years (95% CI 55.9 years to 58.9 years), gastric ulcer patients had a mean age of 66.3 years (95% CI 64.9 years to 67.7 years; P<0.001).The distribution with respect to age group is shown in Figure 1.From 1996 to 2006, the mean age of the ulcer patients increased: for duodenal ulcer patients by 0.992 years per year (95% CI 0.44 to 1.4; P<0.001) and for gastric ulcers patients by 0.274 years per year, a difference that was not statistically significant (95% CI -0.34 to 0.89) (Figure 2).
Classification according to the Forrest system (Table 1) demonstrated that gastric ulcers showed signs of bleeding (Forrest 1 and 2) less frequently than duodenal ulcers (P<0.001).Ulcers classified as Forrest 3 (ie, no signs of recent bleeding) were seen in 77.3% of the duodenal ulcer cases and 87.1% of the gastric ulcer cases.Active bleeding (Forrest 1) was seen in 6.5% of all duodenal ulcers and 3.9% of all gastric ulcers, and Complicated ulcers, defined as Forrest 1 and 2, were more frequently seen in elderly patients, with a mean age of bleeding duodenal ulcer patients of 69.7 years (95% CI 66.8 years to 72.7 years) versus 54.3 years (95% CI 52.8 years to 55.9 years) for nonbleeding duodenal ulcer patients (P<0.001).For bleeding gastric ulcer patients, the mean age was 71.6 years (95% CI 67.6 years to 75.5 years), compared with 65.6 years (95% CI 64.1 years to 67.1 years) for uncomplicated ulcer disease (P=0.077).Of the patients with complicated ulcers, 60.4% with duodenal and 63.4% with gastric ulcers were older than 70 years of age.The sex distribution was not different for various Forrestclassified ulcers, nor was the sex distribution in patients with complicated ulcers different from that in patients with uncomplicated ulcers.For both groups, the majority of patients were men (57.7% and 54.5%, respectively; P=0.395).
In patients with complicated ulcer disease, H pylori infection was assessed in 22.6%, with a positive result in 37.1% of the tested patients.
The incidence of gastric ulcers remained stable over the 10-year observation period.In contrast, duodenal ulcers showed a decreasing incidence, with a linear regression coefficient of -0.169 (P=0.040)(Figure 3).This decline was due to a decline in the incidence of uncomplicated ulcers only.The incidence of complicated duodenal and gastric ulcers increased over this 10-year period, as shown in Figure 4.This increase was more dramatic for duodenal ulcers.
The CLO test performed on biopsy specimens taken during endoscopy for the presence of H pylori was completed in 20.5% of gastric ulcer patients and 36.4% of duodenal ulcer patients.
Thirty-four per cent of the tested gastric ulcer patients and 56.5% of the tested duodenal ulcer patients were positive for H pylori.The overall incidence of CLO-positive tests declined significantly from 57.7% in the period between 1996 and 2000, to 45.1% in the period between 2001 and 2006 (P=0.02).

DIsCUssION
The investigated population consisted of patients referred for endoscopy, for the most part based on either serious recurring complaints, alarming symptoms or older age, according to the 1996 Dutch General Practitioners' Guideline on Dyspepsia (15).Thus, the investigated population is not a cross-section of the total population and incidences from this investigation are not fully comparable with findings concerning total populations.Trends over time are, however, clear.We found a decreasing incidence of duodenal ulcers and a stable incidence of gastric ulcers.This suggests a further amplification of the trend observed from 1980 onward, implying a decreasing incidence of both gastric and duodenal ulcers, with only a slight decline for gastric ulcer incidence (8).Our data suggest that the former, albeit slighter, decline in gastric ulcer incidence has stopped by now, whereas duodenal ulcer incidence continues to decrease further.This will lead to a situation in which, similar to 100 years ago, gastric ulcers will become more common than duodenal ulcers, whereas for the past 100 years, duodenal ulcer disease has been more common than gastric ulcer disease in Western countries (1).
Other trends that we found were an increasing incidence of complicated ulcers for both duodenal and gastric ulcers, an increasing age at diagnosis for patients with duodenal ulcers and a declining incidence of H pylori-positive ulcers.
The overall decline of peptic ulcer disease is likely to be due to a combination of factors including the introduction of acid suppressive medication, a decreasing prevalence of H pylori in subsequent birth cohorts and the development of eradication treatment for H pylori-positive ulcer patients, which prevents chronic relapsing ulcer disease.The introduction of newer NSAIDs and a tendency for the prescription of lower doses of acetylsalicylic acid for patients with cardiovascular disease may also have contributed to the changing epidemiology of ulcer disease.
The Forrest classification can predict the risk of rebleeding from peptic ulcer disease and, therefore, is an important tool in the selection of patients for endoscopic treatment (16,17).In the present study, a minority (12.3%) of the gastric and 22.7% of the duodenal ulcers were complicated by bleeding.As in  other studies (10), bleeding ulcers were more prevalent in older patients and in men.While we observed a decreasing incidence of ulcers, there was a rise in the proportion of complicated ulcers -an observation consistent with data presented in other studies (8,9).The incidence of complicated ulcers rose in both gastric and, to a greater extent, duodenal ulcers in the time period investigated (Figures 3 and 4).The decline in the incidence of duodenal ulcers, in particular, can be completely attributed to a decrease of uncomplicated Forrest 3 ulcers.
The changes in the proportional distribution of complicated versus uncomplicated ulcers in the investigated period may have been influenced by the more accurate description of ulcers in endoscopic reports in recent years.The endoscopy report program Endobase III enabled more detailed standardized report writing as a result of improvements in the program from 2001 onward.Consequently, the need for free text was minimized in the updated version of the program.This may have influenced the description of the ulcer classification.We support the suggestion from other publications showing a stable incidence or slight increase of ulcer bleeding compared with overall peptic ulcer disease.It was suggested that the use of NSAIDs are the main cause of this change (8,10,11).The high proportion of negative CLO tests in bleeding ulcers supports the hypothesis that bleeding ulcers are more often related to NSAID use, although it may also partly be due to the low sensitivity of the CLO test in patients with upper gastrointestinal bleeding (18).Adherence to guidelines to prevent NSAID-related upper gastrointestinal events is low in the Netherlands and in the United States (9,19).This may be an important factor in the opposite trends in the incidence of complicated versus uncomplicated peptic ulcer patients.
Both ulcer types show an increase in incidence rate with increasing age.Duodenal ulcers occur in our population at a younger age than gastric ulcers.The incidence of duodenal ulcers starts to rise at approximately 20 years of age, in contrast with 50 years of age for gastric ulcers.This again indicates an etiological factor for duodenal ulcers early in life, such as infection with H pylori.These findings are not completely comparable with other studies (10) that report an older age for duodenal ulcer patients than in the present study.
In 1996, an update of the 1993 Dutch General Practitioners Guidelines on Dyspepsia was published (15).In this guideline, it was advised to diagnose H pylori status by endoscopic biopsy because only H pylori infection in patients with ulcers should be treated.The role of H pylori in other dyspeptic symptoms was stated as unclear.
In 2004, a new multidisciplinary guideline for dyspepsia was published in the Netherlands by The Dutch Institute for Healthcare Improvement CBO and the Dutch General Practitioners Society (20).In this guideline, an H pylori test is recommended if there are persistent nonreflux-like complaints without alarming symptoms.If the test for H pylori is negative, treatment with proton pump inhibitors for two to four weeks is recommended.If symptoms persist or diagnostic certainty is desired, endoscopy is recommended.Because these guidelines were published in 2004, and the total number of endoscopies performed was still rising up to that year, these guidelines were not suspected to have influenced the incidence and complications of peptic ulcer disease in the present study.However, it is possible that noninvasive testing for H pylori and the use of trials of therapy for dyspeptic patients have had a significant effect on the rate of endoscopies performed in duodenal ulcer patients, resulting in an apparent decline in incidence.
A CLO test was performed only in a small proportion of the patients.Other tests such as the carbon-13 urea breath test or serology may have been used in patients included in the present study, particularly in cases of bleeding.However, those test results were not included in the endoscopy database and, thus, could not be analyzed in the present study.(Forrest 3); DU with complications of bleeding (Forrest 1 and 2).Right panel: Gastric ulcers (GU) without complications of bleeding (Forrest 3); GU with complications of bleeding (Forrest 1 and 2)

Figure 1 )
Figure 1) Age-related proportion of patients with ulcer disease

Figure 2 )
Figure2) Linear regression of mean age of duodenal and gastric ulcer patients in a 10-year period.The mean age of the duodenal ulcer patients increased by 0.992 years per year (95% CI 0.44 to 1.4) (P<0.001) compared with a nonsignificant increase of 0.274 years per year for gastric ulcer patients (95% CI −0.34 to 0.89)

Figure 3 )Figure 4 )
Figure 3) Trends in the incidence of duodenal ulcer (DU) and gastric ulcer (GU) disease