Family Practice Vol. 16, No. 2, 129-132
© Oxford University Press 1999
The prognostic value of gastrointestinal morbidity for gastric cancer
Department of Gastroenterology, University Hospital Nijmegen,
a Department of General Practice and
b Department of Epidemiology, University of Nijmegen, Nijmegen, The Netherlands.
R Laheij MSc, MIES (152), P.O. Box 9101, NL 6500 HB Nijmegen, The Netherlands.
Laheij RJF, Jansen JBMJ, van de Lisdonk EH, Severens JL and Verbeek ALM. The prognostic value of gastrointestinal morbidity for gastric cancer. Family Practice 1999; 16: 129–132.
Received 27 March 1998; Revised 25 August 1998; Accepted 19 November 1998.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Objective. A history of gastrointestinal disease is a risk factor for organic gastrointestinal disorders overall. We conducted a retrospective case-control study to explore whether a history of gastrointestinal disease is of prognostic value for gastric cancer.
Methods. Forty-six patients with gastric cancer were identified from a dynamic population of approximately 12 000 patients followed since 1971 in four general practices. The control subjects without gastric cancer were matched one-on-one according to age, general practice, sex and observation period. Data on gastrointestinal morbidity in the period before gastric cancer was diagnosed were obtained from the Continuous Morbidity Registration.
Results. The mean observation period between the date of enrolment in the registration and the first diagnosis of gastric carcinoma was 12 years. Although every patient with gastric cancer ultimately will develop gastrointestinal complaints, 28 of these patients had no previous gastrointestinal morbidity, in comparison with 26 control subjects. Furthermore, patients who developed gastric cancer did not have more frequent gastrointestinal morbidity in their past than the control subjects (odds ratio 0.80, 95% CI 0.32–2.03).
Conclusions. Our results suggest that a history of gastrointestinal morbidity is not of prognostic value for gastric cancer. Focusing attention on patients with a past history of gastrointestinal symptoms to detect gastric cancer may be of little value.
Keywords. Case-control study, gastric cancer, gastrointestinal disorders, prognosis..
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Although the incidence of gastric cancer has decreased over the past few decades, in The Netherlands and in other Western countries it is still a common disease.1 The incidence of gastric cancer in The Netherlands was 15.4 for males and 6.1 for females per 100 000 person-years in the period 1989–1990.2 The 5-year survival rate is low (±20%) but has remained relatively constant during the past few decades. However, relative survival is clearly better for patients discovered with an early stage of gastric cancer. If diagnosed in an early stage, gastric cancer is curable and has an excellent prognosis.3 Increased recognition of the early manifestations of gastric cancer should eventually lead to improvement in overall survival rates.
Previous studies have used a history of gastrointestinal diseases in order to differentiate patients with and without serious gastrointestinal disorders.4–6 A history of gastrointestinal disease was found to be a prognostic factor for organic gastrointestinal disorders besides other warning symptoms such as vomiting, nausea, loss of weight and smoking. In order to determine the prognostic value of previous gastrointestinal disease for gastric cancer, we conducted a retrospective case-control study to compare gastrointestinal morbidity between patients with gastric cancer and control subjects without gastric cancer.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Patients with gastric cancer were identified using the Continuous Morbidity Registration. The Continuous Morbidity Registration Nijmegen began in 1971 in four general practices with a combined patient population of approximately 12 000 people (with an annual turnover of about 5%) in and around Nijmegen, The Netherlands.7,8 This databank is used to study the epidemiological aspects of diseases in general practice, including the incidence, prevalence, and course of a disease over time. Every episode of illness seen by or reported to the GP is registered as soon as it is established. The recorded data have passed stringent quality controls and are consistent over the years of registration.9,10
In The Netherlands, specialist care is only available after referral by a GP. Subsequently, the specialist reports back to the GP about diagnosis and treatment of referred patients. The adapted E-list is used to register diagnosis, with use of the fourth digit extension to make it compatible with the International Classification of Health Problems in Primary Care.11 Causes of death, including those of patients dying in a hospital, are also recorded in the system.
All newly diagnosed patients with gastric cancer from 1971 to 1995 were included in the study population. The diagnosis of all cases was verified using the data from the patients' files and reports from medical specialists to the GP. Patients were excluded whenever no X-ray or upper gastrointestinal endoscopy was performed to confirm the diagnosis of gastric cancer. For each patient one control subject was selected and matched for sex, social class, practice, observation period and age. Controls were selected from the practice list at the time of diagnosis of the gastric cancer. We retrieved information concerning gender, date of birth, date of entry or departure in the registration, social class, morbidity and both histological type and location of the tumour when present. Social class, defined according to profession as well as education, was classified in low, middle and high categories. Cases and controls were observed for the same time period. In the Continuous Morbidity Registration, gastrointestinal morbidity was registered as: esophagitis, peptic ulcer, gastroenteritis, gastritis, dyspepsia, other stomach disorders and constipation. The odds ratio of patients with gastric cancer versus control subjects was estimated by a matched data analysis, according to the formula of Mantel and Haenzel.12
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Fifty patients out of a total population of 12 000 patients were registered between 1971 and 1995 as newly diagnosed gastric cancer patients. The incidence of gastric cancer in the area of Nijmegen is approximately one patient every 2 years per general practice. Four patients were excluded from the study population because gastric cancer had been diagnosed without an X-ray or an upper gastrointestinal endoscopy. Thus, 46 patients with diagnostically verified stomach cancer were included in the study population (Table 1
).
|
Four patients had gastric cancer of the diffuse type and 12 patients had gastric cancer of the intestinal type; for 30 patients the histological type was not mentioned in the patient file or in the letters of medical specialists. Survival from stomach cancer was very poor, with only six patients (13%) surviving more than 5 years after the diagnosis. The average period between diagnosis and death was less than 2 years. Twenty-seven patients (59%) with gastric cancer were of the lowest social class, while only one patient (2%) was of the highest social class. Patients with gastric cancer registered an average of 30 disease episodes in the observation period, and the control subjects 39 episodes (P = 0.54).
We found no differences in previous gastrointestinal morbidity between patients and control subjects except for constipation, which was more frequent in the control population (Table 2). There were also no differences in morbidity when stratified by observation period, histological type, gender and tumour location (Table 3).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The most important finding of this case-control study is that patients with gastric cancer do not have a history of gastrointestinal diseases more often than do control subjects without gastric cancer. Other studies have reported that a hiatus hernia and a previous ulcer were prognostic factors for predicting major pathological conditions.4–6 Two reasons may account for this contradiction with our results. First, these studies tried to predict major pathological conditions such as peptic ulcers, complicated esophagitis and malignancies with a scoring system. Both peptic ulcers and complicated esophagitis are disorders with many relapses, which are probably responsible for selecting prior gastrointestinal morbidity as a warning factor.
Secondly, the data about morbidity in the studies of Mann et al.4 and Bytzer et al.6 were obtained by a questionnaire. Obtaining information with a questionnaire depends upon a patient's recall capability. In particular, when asked for a long observation period, recall bias will be introduced. As a result, many of the patients (33%) did not know whether they had had a previous ulcer or hiatus hernia.6 In the Continuous Morbidity Registration Nijmegen, no recall bias will occur because every episode of illness seen by or reported to the GP independent of future occurrence of gastric cancer is registered as soon as it is established. Therefore obtaining information with a questionnaire seems to be less reliable than using the Continuous Morbidity Registration Nijmegen.
We measured gastrointestinal morbidity as a proxy for gastrointestinal complaints. Not every gastrointestinal symptom is serious enough to warrant consultation with a GP. The decision to visit a GP for a complaint varies among individuals.13,14 Therefore, physician diagnosis may not reflect patient experience, but at least should be equal for cases and controls.
Gastric cancer is not homogenous. Chronic gastritis and intestinal metaplasia are the precursors of the intestinal type of gastric cancer.15 Unfortunately, the histological type of the tumour was not available in patients diagnosed by X-ray or with an incomplete registry. It is unclear, however, whether differences in gastrointestinal morbidity after stratifying for histological type, tumour location and gender were not found due to the small patient population and missing patient data. Besides, other risk factors (Helicobacter pylori, diet, smoking, etc.) which are supposed to play a role in the pathogenesis of gastric cancer were not assessed in this study. Constipation was the only disorder we found that was different between patients with gastric cancer and the control subjects. An episode of constipation was registered in 8 of the 46 control subjects (17%) and in only one patient with gastric cancer (2%). Comparison of all registered disorders in the registration would probably give more differences in morbidity between patients and control subjects. According to our results, constipation is not associated with gastric cancer.
Our study indicates that most patients with gastric cancer did not present more frequently with a history of gastrointestinal diseases than did control subjects. Although the patient population was small, no indication was found that gastrointestinal morbidity was different in comparison to control subjects. Given the limitations of this study, our results indicate that prior gastrointestinal morbidity should not be considered a warning sign for gastric cancer. Early detection of gastric cancer is not yet possible, but focusing attention on patients with a history of gastrointestinal disease to detect gastric cancer may be of little value.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
1 Coleman MP, Esteve J, Damiecki P, Arslan A, Renard H. Trends in Cancer Incidence and Mortality. Lyon: International Agency for Research on Cancer, 1993.
2 Sanden GAC van der, Coebergh JJW, Visser O, Leeuwen FE van. Cancer incidence in The Netherlands in 1989 and 1990: first results of the nationwide Netherlands cancer registry. Eur J Cancer 1995; 11: 1822–1829.
3 Kodama Y, Inokuchi K, Soejima K et al. Growth patterns and prognosis in early gastric carcinoma: superficially spreading and penetrating growth types. Cancer 1983; 51: 320–326.[Web of Science][Medline]
4 Mann J, Holdstock G, Harman M, Machin D, Loehry CA. Scoring system to improve cost effectiveness of open access endoscopy. Br Med J 1983; 287: 937–940.
5 Numans ME. Gastroscopie op verzoek van de huisarts. Rijks-universiteit Utrecht, PhD Thesis, 1992.
6 Bytzer B, Schaffalitzky de Muckadell OB. Prediction of major pathologic conditions in dyspeptic patients referred for endoscopy. Scand J Gastroenterol 1992; 27: 987–992.[Web of Science][Medline]
7 Weel C van, Bosch WJHM van den, Hoogen HJM van den, Smits AJA. Development of respiratory illness in childhood—a longitudinal study in general practice. J R Coll Gen Pract 1987; 37: 404–408.[Medline]
8 Grauw WJC de, Lisdonk E van de, Hoogen HJM van den, Weel C van. Cardiovascular morbidity and mortality in type 2 diabetic patients: a 22-year historical cohort study in Dutch general practice. Diabet Med 1995; 12: 117–122.[Web of Science][Medline]
9
Lisdonk EH van de, Weel C van. New referrals, a decreasing phenomenon in 1971–94: analysis of registry data in the Netherlands. Br Med J 1996; 313: 602.
10 Weel C van. Validating long term morbidity recording. J Epidemiol Community Health 1995; 49 (Suppl 1): 29–32.
11 ICHPPC-2 Defined. WONCA. 3rd edn. Oxford: Oxford University Press, 1983.
12 Rothman KJ. Modern Epidemiology. Boston/Toronto: Little, Brown and Company, 1986.
13 Lydeard S, Jones RH. Factors affecting the decision to consult with dyspepsia: a comparison of consulters and non-consulters. J R Col Gen Pract 1989; 39: 495–498.
14 Kettell J, Jones Rh, Lydeard S. Reasons for consultation in irritable bowel syndrome: symptoms and patient characteristics. Br J Gen Pract 1992; 42: 459–461.[Web of Science][Medline]
15 Sipponen P, Kekki M, Siurala M. Age related trends of gastritis and metaplasia in gastric carcinoma patients and in controls representing the population at large. Br J Cancer 1984; 49: 521.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||