Share this article on:

Smoking and Alcohol Intake as Risk Factors for Bleeding and Perforated Peptic Ulcers: A Population-Based Cohort Study

Bak Andersen, Inger; Jørgensen, Torben; Bonnevie, Olaf; Grønbæk, Morten; Sørensen, Thorkild I. A

Original Articles

Both the incidence of and mortality from bleeding and perforated peptic ulcers are increasing. We assessed the association between smoking, intake of alcohol (including type of alcoholic beverage), and risk of a complicated peptic ulcer in a population-based study of 26,518 Danish subjects followed up for an average of 13.4 years. There were 214 cases of incident bleeding and 107 cases with perforated ulcers. We estimated relative risks (RRs) for incident bleeding and perforated peptic ulcers using Poisson regression analysis. Smoking more than 15 cigarettes per day compared with never smoking increased the risk of a perforated ulcer more than threefold [RR = 3.5; 95% confidence interval (CI) = 1.7–7.1)]. Drinking more than 42 drinks per week increased the risk of a bleeding ulcer fourfold (RR = 4.4; 95% CI = 2.3–8.3) compared with drinking less than one drink per week. Using the same comparison group, subjects who drank more than 21 drinks per week but no wine were at a higher risk of a bleeding ulcer (RR = 8.8; 95% CI = 2.2–35) than drinkers of the same amount of alcohol, but with more than 25% of their intake as wine (RR = 2.4; 95% CI = 1.0–6.0).

From the 1Copenhagen Centre for Prospective Population Studies, Danish Epidemiology Science Centre at the Institute of Preventive Medicine, Copenhagen University Hospital, H:S Kommunehospitalet; 2Centre of Preventive Medicine, Medical Department C/F, Glostrup University Hospital; and 3Department of Internal Medicine I, Bispebjerg University Hospital, Copenhagen, Denmark.

This study was supported in part by a grant from The Danish National Research Foundation (to T. I. A. S. and I. B. A.) and by a grant from Sygekassernes Helsefond (to I. B. A.).

Submitted May 24, 1999; final version accepted November 24, 1999.

Address correspondence to: Inger Bak Andersen, Institute of Preventive Medicine, H:S Kommunehospitalet, DK-1399 Copenhagen K, Denmark.

The Copenhagen Centre for Prospective Population Studies consists of the Centre of Preventive Medicine (formerly Glostrup Population Studies), the Copenhagen City Heart Study, and the Copenhagen Male Study.

In Northern Europe, both the incidence and mortality of complicated, that is, bleeding or perforated, peptic ulcers are rising, while the incidence of uncomplicated peptic ulcers is declining. 1,2 It is not known what promotes uncomplicated peptic ulcers to become potentially lethal with bleeding or perforation, although smoking and high alcohol intake have been proposed as possible risk factors. Several cohort studies have identified cigarette smoking as a risk factor for uncomplicated peptic ulcers, 3–6 although none evaluated specifically the relation of smoking and complicated peptic ulcers. A positive association between peptic ulcers and alcohol intake 3,5,7–9 has not been demonstrated. On the contrary, three studies have reported a protective effect of moderate alcohol intake. 7,9,10 Few studies have addressed the possible different effects of wine, beer, and spirits.

The aim of this prospective, population-based study was to assess the effect of smoking, alcohol intake, and type of alcoholic beverage on the risk of subsequent hospitalization for bleeding or perforated peptic ulcer.

Back to Top | Article Outline

Subjects and Methods


This study is based on eight Danish prospective cohort studies from the Copenhagen area, including six cohort studies from the Centre of Preventive Medicine (formerly the Glostrup Population Studies), the Copenhagen City Heart Study, and the Copenhagen Male Study. In 1993, the studies were combined into the Copenhagen Centre for Prospective Population Studies. These studies have been described in detail previously. 11–14 Owing to their participation in more than one cohort study, 272 participants in subsequent cohorts were excluded and were only included in the primary cohort. Responders who reported having had a peptic ulcer before study entry (N = 905) were also excluded, leaving 26,518 for this investigation.

Briefly, the Centre of Preventive Medicine was started in 1964 as the Glostrup Population Studies. 11,12 Several randomly selected sex- and age-specific cohorts from Copenhagen County were examined, and for the present study the “1897” (started in 1977) , “1914” (1974) , “1936” (1976) , moni toring trends and determinants in ca rdiovascular diseases (MONICA) I (1982) , MONICA II (1986) , and MONICA III (1991) cohorts were included (Table 1). All cohorts have similar distributions of men and women, with a total of 8,788 individuals included in our study. Table 1 shows the number of participants included from each study. The initial response rates were between 69% and 88%.

Table 1

Table 1

The Copenhagen City Heart Study 13 was started in 1976 and consists of a randomly selected sex- and age-stratified population of men and women from 20 to 91 years of age and resident in the city of Copenhagen (Table 1). In total, 14,218 were included in this investigation. The initial response rate was 74%.

The Copenhagen Male Study 14 began in 1970 with 5,249 men employed in 14 public and private companies in Copenhagen. This investigation included 3,514 men who were followed since 1976 (Table 1). The initial response rate was 85.7%.

Back to Top | Article Outline

Study Variables

At the initial general health interviews, researchers collected information on the following variables: smoking; intake of beer, wine, and spirits; physical activity during leisure time; level of education; history of peptic ulcer; and use of analgesics.

Smoking habits included information about ever smoking, which included smoking in the past or smoking at the time of interview. Current smokers were asked about the daily amount of tobacco smoked (1–14 gm or ≥15 gm of tobacco per day, with one cigarette containing on average 1 gm of tobacco). Smokers reported whether they inhaled while smoking as well as the number of years of smoking. For current smokers, we calculated the cumulative tobacco exposure as pack-years. One pack-year is equivalent to smoking 20 cigarettes per day for 1 year. Data on the tobacco variables were collected from all of the cohorts.

We categorized the total weekly intake of alcohol as <1 drink, 1–6 drinks, 7–13 drinks, 14–27 drinks, 28–41 drinks, and ≥42 drinks per week. One beer contains 12 gm of alcohol, which can be considered the amount of an average drink in Denmark. We categorized the intake of beer, wine, and spirits as <1 drink, 1–2 drinks, 3–6 drinks, and ≥7 drinks per week. Using the weekly intake of the different types of alcoholic beverages, we estimated the percentage of the total alcohol intake that was wine. Information on absolute alcohol intake per week and type of alcoholic beverage was not available in the Copenhagen Male Study and the 1914 cohort. For both cohorts, however, alcohol intake was categorized beforehand in the questionnaire as <6, 6–20, 21–30, 31–40, 41–50, and ≥51 drinks per week. We used the latter categorization to adjust for confounding by alcohol intake when evaluating the effect of smoking.

In addition to the subject’s age, sex, and cohort of origin, we included level of physical activity and education in the analyses to adjust for potential confounders. Information on leisure-time physical activity was available for all cohorts except the 1897 cohort. Subjects were categorized as physically active if they were engaged in physical activity for more than 2 hours per week; otherwise, they were categorized as sedentary. Level of education (<8 years, 8–11 years, or ≥12 years) was recorded for all cohorts.

Information on analgesic use within the last year (daily, weekly, monthly, or never) was collected from the 1936 cohort and in the three MONICA follow-up studies.

Back to Top | Article Outline

Hospital Discharges

The study population was linked to the Danish National Register of Hospital Discharges (“Landspatientregisteret”), in which all hospital discharges from somatic hospital wards in Denmark have been registered since 1976. The Register contains information on hospital, department, dates of admission and discharge, and diagnosis according to the International Classification of Diseases, 8th revision (ICD-8). 15 The ICD-8 diagnosis code numbers for bleeding (531.90/2/5) and perforated gastric ulcers (531.00/1/8/9), for bleeding (532.90) and perforated duodenal ulcers (532.09), and for bleeding (533.90) and perforated unspecified gastroduodenal ulcers (533.09) were selected for this study. Linkage was done by means of a civil registration number that is assigned to all inhabitants in Denmark, making linkage across registers and time highly accurate. Follow-up was until January 1, 1994.

To validate the information from the Register, we examined discharge reports of a sample of 366 hospitalizations for uncomplicated and 252 for complicated peptic ulcer. We drew the sample at random from the registered hospitalizations for peptic ulcer in the eight cohorts. For complicated peptic ulcers, a total of 67 (25%) hospitalizations for complicated ulcer were lost to follow-up owing to closing of a hospital or department or lost medical records. We reviewed all of the accessible discharge reports. To be registered as an ulcer, the diagnosis had to have been verified by endoscopy, operation, or necropsy. In total, 103 hospitalizations for bleeding and 82 hospitalizations for perforated peptic ulcer were reviewed. Of the 103 patients with bleeding peptic ulcer, 75 (73%) had a confirmed bleeding peptic ulcer and 13 (13%) had no ulcer diagnosed at the pertaining hospitalization; for perforated peptic ulcer, 57 (70%) had a confirmed perforated ulcer and 9 (11%) had no ulcer. Diagnosis of ulcer had not been confirmed by endoscopy, operation, or necropsy in the remainder.

Back to Top | Article Outline

Statistical Analysis

We computed the time at risk for each participant as time from initial examination to the first admission with a complicated peptic ulcer, emigration, death, or January 1, 1994, whichever occurred first. We conducted Poisson regression analyses using the statistical program STATA 5.0. 16,17 Age was chosen as time scale and was divided into 10-year age bands after the age of 40. A person observed in more than one age group contributed observation time in each of these age groups. We excluded 339 individuals with missing data.

We estimated relative risks (RRs) for bleeding or perforated peptic ulcers adjusted for cohort of origin, age, sex, alcohol intake, smoking habit, physical activity, and level of education as potential confounders. We stratified subjects by total weekly alcohol intake to estimate RRs for complicated peptic ulcers according to percentage in the total alcohol intake that was wine (0%, 1–25%, or ≥25% wine).

Back to Top | Article Outline


During the follow-up period representing 354,292 person-years at risk, there were a total of 321 patients hospitalized at least once after the initial examination with a complicated peptic ulcer, 214 bleeding and 107 perforated (Table 1). Table 2 shows baseline characteristics of subjects according to smoking and alcohol drinking habits.

Table 2

Table 2

There was a dose-response relation between amount of tobacco smoked per day and risk of hospitalization with a perforated peptic ulcer (Table 3). In smokers, inhalation was associated with perforated peptic ulcers. The risk of hospitalization with a perforated ulcer increased with years of smoking and, hence, with the number of pack-years, even after adjusting for amount smoked per day. Adjusting the smoking variables for the weekly intake of alcohol as well as wine did not result in important changes in the RRs. The participants in the Copenhagen Male Study and the 1914 cohort were excluded from the analyses of the alcohol risk estimates because of different categorization of the alcohol intake. As a sensitivity analysis, we calculated RRs for tobacco consumption without these two cohorts and found the estimates to be similar. Hospitalization with a bleeding peptic ulcer was only modestly associated with smoking status.

Table 3

Table 3

Drinking more than 14 drinks per week increased the risk of a bleeding peptic ulcer (Table 3). For bleeding peptic ulcers there was a dose-response relation with the weekly alcohol intake. Drinking beer increased the risk of hospitalization with bleeding ulcers. There was no increased risk associated with intake of wine; for spirits, there was a modest increased risk for drinking ≥7 drinks. When evaluating the effect of each type of alcoholic beverage, we adjusted for intake of the others. For perforated peptic ulcers, there was no association with alcohol intake.

The risk of a bleeding ulcer was further analyzed according to percentage of wine in total alcohol intake (Figure 1). Compared with the risk at similar levels of alcohol intake in non-wine drinkers, the risk of a bleeding ulcer decreased with increasing proportion of wine of total alcohol intake. In subjects drinking more than 21 drinks per week, the risk of a bleeding ulcer was 2.4 [95% confidence interval (CI) = 1.0–6.0] among wine drinkers (wine constituted >25% of total alcohol intake) compared with 8.8 (95% CI = 2.2–35) among non-wine drinkers.



The joint effect of alcohol drinking and smoking on the risk of complicated peptic ulcer was slightly more than additive (RR = 1.9 compared with RR = 1.3 for smokers who drank less than one weekly drink and RR = 1.4 for those who drank 14 or more weekly drinks and never smoked).

Back to Top | Article Outline


The results of this population-based cohort study indicate that perforated and bleeding peptic ulcers have different risk factors, with smoking being associated with perforation and moderate to high alcohol intake with bleeding.

There was a strong association (RR = 3.5; 95% CI = 1.7–7.1) between smoking more than 15 gm of tobacco per day and being hospitalized with a perforated peptic ulcer. Svanes et al, 18 in a recent Norwegian case-control study, also found current smokers to be at increased risk [odds ratio (OR) = 9.7; 95% CI = 5.9–15.8] for a perforated ulcer, along with a noticeable dose-response relation. Our estimate of the RR among ex-smokers (RR = 1.9; 95% CI = 0.8–4.3) differed from that of the Norwegian study (OR = 0.8; 95% CI = 0.2–2.2).

Bleeding peptic ulcers were not strongly associated with smoking in our study. The literature on the association between smoking and bleeding peptic ulcers is sparse. A small Italian case-control study 19 found current smoking to protect against duodenal ulcer bleeding (OR = 0.4; 95% CI = 0.2–0.9).

The available epidemiologic data are consistent with a causal relation between cigarette smoking and perforated ulcer, but specific components of the causal pathways have not been identified. Studies on the effect of cigarette smoking on gastric acid secretory activity, gastroduodenal mucosal prostaglandin concentrations, gastric motility, and pancreatic secretion are conflicting. 20 Some studies have found that smoking causes vasoconstriction in the mucosa and that ischemia reduces mucosal resistance. 21 The mucosal ischemia could be mediated further by carbon monoxide, as carboxyhemoglobin levels are increased in smokers. 22,23 These latter mechanisms might explain the strong association between smoking and perforated peptic ulcers.

In our study, we found an increased risk of hospitalization for bleeding peptic ulcer in persons drinking more than 14 drinks per week. Two case-control studies have likewise found positive associations between daily alcohol intake and major upper nonvariceal gastrointestinal bleeding. 24,25 In persons with a daily alcohol intake of more than five drinks per day, the RR was 6.3 (95% CI = 3.5–11) in the study by Kelly et al 24 and 2.8 (95% CI = 1.9–4.0) in the study by Henry et al. 25 As in our study, Kelly et al 24 found a dose-response relation between alcohol consumption and risk of bleeding peptic ulcer. These findings are in contrast with those of a large German case-control study that reported no association between alcohol intake and risk of bleeding peptic ulcer. 26

Acute exposure to high concentrations of ethanol is followed by severe gastric epithelial damage and necrosis of deeper layers of the mucosa, and microvascular damage leading to engorgement, increased permeability, and intramucosal hemorrhage. 27 Besides the local irritative effects of ethanol, it has been shown that moderate to high concentrations of ethanol delay gastric emptying. 28

The risk of hospitalization with bleeding peptic ulcer depended on the type of alcoholic beverage. Beer increased the risk, whereas wine drinking was associated with a much lower risk. In our study population, beer drinkers probably were more likely to binge than wine drinkers. Fourteen per cent of the study population drank more than seven beers per week compared with the 4% of the population who drank more than seven glasses of wine per week. At the same level of alcohol intake, this difference in the pattern of drinking does not explain the lower risk of bleeding peptic ulcer in wine drinkers compared with beer drinkers. No other study has shown an association between type of alcoholic beverage and risk of bleeding peptic ulcer. The mechanisms behind the apparent protective effect of wine and the damaging effect of beer are unknown. Wine has been used through the centuries as a digestive aid and has recently been shown to have pronounced antibacterial properties. 29 One could speculate that wine might have a bactericidal or bacteriostatic effect on Helicobacter pylori. A recent cross-sectional study found that alcohol consumption, primarily wine consumption, reduced the odds of active infection with H. pylori. 30 Another explanation could be that the phenol compounds in wine with antioxidant properties may have a protective effect on the gastric mucosa.

An advantage of the present study is its size and prospective design, where assessment of smoking and alcohol drinking habits took place before the diagnosis of the complicated peptic ulcer. Recent symptoms due to peptic ulcer may affect smoking and alcohol habits, so case-control and cross-sectional studies of this disease have inherent limitations. Lack of follow-up information on smoking and alcohol consumption presumably result in underestimated RRs, because part of the effect of alcohol and smoking may be acute. At the time baseline measures on alcohol were collected, wine drinking was probably associated with higher socioeconomic status. We have used level of education as an indicator of socioeconomic status. This crude adjustment for socioeconomic status renders the possibility of residual confounding as an alternative explanation for the protective effect of wine. The present study only included ulcers identified from hospital discharges. Complicated peptic ulcers are usually life-threatening diseases resulting in hospitalization in almost all patients. Our validation study showed that about 70% of the persons registered with a complicated peptic ulcer had a confirmed, complicated ulcer, whereas in 20% the diagnosis had not been confirmed.

We have no information on H. pylori status among participants at study entry. H. pylori is a known risk factor for uncomplicated peptic ulcers, but the significance of H. pylori in the etiology of complicated peptic ulcers is uncertain. 31 Eradication of the bacteria may reduce the risk of subsequent episodes with peptic ulcer bleeding, however. 31

Studies have reported a population attributable risk of 17–34% for nonsteroidal anti-inflammatory drug (NSAID) use in complicated peptic ulcer disease. 25 The combined exposure to NSAIDs and alcohol increases the risk of a complicated peptic ulcer. 25 For the 1936 cohort and the MONICA I, II, and III cohorts, we have information on frequency of analgesic use that can be considered a proxy of NSAID use. In this subpopulation, 14% of the persons with a weekly alcohol intake of ≥21 drinks used analgesics daily or weekly compared with 21% of persons who had an intake of <1 drink per week. Among smokers, 18% used analgesics frequently compared with 15% of never-smokers. These small differences indicate that use of NSAIDs does not account for the fairly strong association between alcohol intake and risk of a bleeding peptic ulcer in our study.

Back to Top | Article Outline


1. Bak Andersen I, Bonnevie O, Jørgensen T, Sørensen TIA. Time trends for peptic ulcer disease in Denmark 1981–93. Scand J Gastroenterol 1998; 38:260–266.
2. Sonnenberg A. Factors which influence the incidence and course of peptic ulcer. Scand J Gastroenterol 1988; 155suppl:119–140.
3. Paffenbarger RS, Wing AL, Hyde RT. Chronic disease in former college students. Am J Epidemiol 1974; 100:307–315.
4. Anda RF, Williamson DF, Escobedo LG, Remington PL. Smoking and risk of peptic ulcer disease among women in the United States. Arch Intern Med 1990; 150:1437–1441.
5. Kato I, Nomura AMY, Stemmermann GN, Chyon P-H. A prospective study of gastric and duodenal ulcer and its relation to smoking, alcohol, and diet. Am J Epidemiol 1992; 135:521–530.
6. Johnsen R, Førde OH, Straume B, Burhol PG. Etiology of peptic ulcer: a prospective population study in Norway. J Epidemiol Community Health 1994; 48:156–160.
7. Piper DW, McIntosh, JH, Grieg M, Shy M. Environmental factors and chronic gastric ulcer. Scand J Gastroenterol 1982; 17:721–729.
8. Chou SP. An examination of the alcohol consumption and peptic ulcer association: results of a national survey. Alcohol Clin Exp Res 1994; 18:149–153.
9. Schubert TT, Bologna SD, Nensey Y, Schubert AB, Mascha EJ, Ma CK. Ulcer risk factors: interactions between Helicobacter pylori, nonsteroidal use, and age. Am J Med 1993; 94:413–418.
10. Rosenstock SJ. Alcohol consumption and peptic ulcer disease. In: Rosenstock SJ, ed. The Epidemiology of Peptic Ulcer Disease in a Danish County (PhD thesis). Copenhagen: FADL Publishers, 1996; 135–148.
11. Hagerup L, Schroll M, Hollnagel H, Agner E, Larsen S. The Glostrup Population Studies: collection of epidemiological tables: reference values for use in cardiovascular population studies. Scand J Soc Med 1981:1–1250.
12. Schroll M. The World Health Organization MONICA project (monitoring trends and determinants in cardiovascular disease): a major international collaboration. J Clin Epidemiol 1988; 41:105–114.
13. Appleyard M, Hansen AT, Schnohr P, Jensen G, Nyboe J. The Copenhagen City Heart Study: Østerbroundersøgelsen: a book of tables with data from the first examination (1976–78) and a five year follow-up (1981–83). Scand J Soc Med 1989; 170:1–160.
14. Hein HO, Suadicani P, Gyntelberg F. Alcohol consumption, serum low density lipoprotein cholesterol concentration, and risk of ischaemic heart disease: six year follow up in the Copenhagen Male Study. BMJ 1996; 312:736–741.
15. Jürgensen HJ, Frølund C, Gustafsen J, Mosbech H, Guldhammer B, Mosbech J. Registration of diagnoses in the Danish National Registry of Patients. Methods Inf Med 1986; 25:158–164.
16. Clayton D, Hills M. Statistical models in epidemiology. Oxford: Oxford University Press, 1993.
17. StataCorp. Stata Statistical Software. Release 5.0. College Station, TX: Stata Corporation, 1997.
18. Svanes C, Søreide JA, Skarstein A, Fevang BT, Bakke P, Vollset SE, Svanes K, Søreide O. Smoking and ulcer perforation. Gut 1997; 41:177–180.
19. Di Mario F, Battaglia G, Leandro G, Dotto P, Dal Bo N, Salandin S, Ferrana M, Grassi SA, Vianello F. Risk factors of duodenal ulcer bleeding: the role of smoking and nicotine. Ital J Gastroenterol 1994; 26:385–391.
20. Kikendall JW, Evaul J, Johnson F. Effect of cigarette smoking on gastrointestinal physiology and non-neoplastic digestive disease. J Clin Gastroenterol 1984; 6:65–78.
21. Sørbye H, Svanes K. The role of blood flow in gastric mucosal defense, damage and healing. Dig Dis 1994; 12:305–317.
22. Nolan J, Jenkins RA, Kurihara K, Schultz RC. The acute exposure on experimental skin flaps. Plast Reconstr Surg 1985; 75:544–549.
23. Smith CJ, Guy TD, Stiles MF, Morton MJ, Collie BB, Ingebrethsen BJ, Robinson JH. A repeatable method for determination of carboxyhemoglobin levels in smokers. Hum Exp Toxicol 1998; 17:29–34.
24. Kelly JP, Kaufman DW, Koff RS, Laszlo A, Wiholm B-E, Shapiro S. Alcohol consumption and the risk of major upper gastrointestinal bleeding. Am J Gastroenterol 1995; 90:1058–1064.
25. Henry D, Dobson A, Turner C. Variability in the risk of major gastrointestinal complications from nonaspirin nonsteroidal anti-inflammatory drugs. Gastroenterology 1993; 105:1078–1088.
26. Imhof M, Ohmann C, Hartwig A, Thon K-P, Hengels K-J, Röher H-D, Düsuk Study Group. Which peptic ulcers bleed? Results of a case control study. Scand J Gastroenterol 1997; 32:131–138.
27. Tarnawski A, Hollander D, Stachura J, Klimczyk B, Mach T, Bogdal J. Alcohol injury to the normal human gastric mucosa: endoscopic, histological and functional assessment. Clin Invest Med 1987; 10:259–263.
28. Jian R, Cortot A, Ducrot F, Jobin G, Chayvialle JA, Modigliani R. Effect of ethanol ingestion on postprandial gastric emptying and secretion, biliopancreatic secretions, and duodenal absorption in man. Dig Dis Sci 1986; 31:604–614.
29. Weisse ME, Eberly B, Person DA. Wine as a digestive aid: comparative antimicrobial effects of bismuth salicylate and red and white wine. BMJ 1995; 311:1657–1660.
30. Brenner H, Rothenbacher D, Bode G, Adler G. Inverse graded relation between alcohol consumption and active infection with Helicobacter pylori. Am J Epidemiol 1999; 149:571–576.
31. Laine LA. Helicobacter pylori and complicated ulcer disease. Am J Med 1996; 100:52S–59S.

peptic ulcer hemorrhage; peptic ulcer perforation; cohort study; smoking; alcoholic beverages

Copyright © 2000 Wolters Kluwer Health, Inc. All rights reserved.