Inverse Graded Relation between Alcohol Consumption and Active Infection with Hellcobacter pylori

Transcription

1 Amertean Journal of Epklemlotogy Copyright C 1999 by The Johns HopWns University School of Hygiene and Public Health Ail ngnis reserved Vol. 149, No. 6 Printed In USA. Inverse Graded Relation between Alcohol Consumption and Active Infection with Hellcobacter pylori Hermann Brenner, 1 Dietrich Rothenbacher, 1 Gunter Bode, 2 and Guido Adler 2 Alcoholic beverages are known to have strong antibacterial activity. It is unclear, however, to what degree their consumption affects colonization of the human stomach with the bacterium Helicobacter pylori, a risk factor of various chronic diseases. The authors assessed the relation between alcohol consumption and active infection with H. pylori in a cross-sectional study among employees of a health insurance company and their household members (n = 425) in southern Germany. Quantitative information on alcohol consumption by beverage type and other factors that were known or suspected to be related to infection status was collected by a standardized questionnaire, and active infection was measured by the 13 C-urea breath test. After control for confounding factors, there was a monotonic inverse graded relation between alcohol consumption and H. pylori infection (p for trend = 0.017). The odds of infection among subjects who consumed more than 75 g of alcohol per week compared with subjects who did not drink alcohol was 0.31 (95 percent confidence interval ). The inverse relation with H. pylori infection was stronger for alcohol consumed in the form of wine than for alcohol from beer. Notwithstanding its cross-sectional design, this study seems to support the hypothesis that alcohol consumption, particularly wine consumption, may reduce the odds of active infection with H. pylori. Am J Epidemiol 1999;149: alcoholic beverages; Helicobacter pylori The bacterium Helicobacter pylori is one of the world's most common pathogens. Colonization of the human stomach with H. pylori is the main cause of chronic gastritis and is strongly associated with other chronic diseases (1). The infection is acquired in childhood in most cases (2, 3), but acquisition and elimination of the infection occur throughout adulthood, although the rates appear to be low in developed countries (4). Little is known about factors that might affect acquisition or elimination of the infection among adults. One factor that is of particular interest in this context is alcohol consumption: Alcoholic beverages have a strong antimicrobial activity, as has been shown both in vitro and in vivo (5-7). Furthermore, they have diverse dose- and beverage type-specific effects on the gastric mucosa, on the emptying of the stomach (8), and on gastric acid secretion (9) that may affect the living conditions of H. pylori in the stomach. Received for publication December 18, 1997, and in final form June 9, Abbreviation: CHD, coronary heart disease. 1 Department of Epidemiology, University of Ulm, Ulm, Germany. 2 Department of Internal Medicine I, University of Ulm, Ulm, Germany. Reprint requests to Hermann Brenner, University of Ulm, Department of Epidemiology, Helmholtzstrasse 22, D Ulm, Germany. Several studies have addressed the relation of alcohol consumption to H. pylori infection, and they have yielded inconsistent results (10-20). No association was seen in population-based studies in which infection status was measured by serum immunoglobulin G antibody levels (10-15). Serologic tests are suboptimal, however, for assessing factors that might affect elimination of the infection because they may remain positive over prolonged time periods after elimination of the infection (1). Clinical studies in which active infection was determined by histopathology or culture among patients who underwent gastroscopy likewise did not find an association (16-19). Most of these studies had limited power due to small sample size (16-18), however, or were limited by crude assessment of alcohol consumption (such as a simple dichotomization of drinkers and nondrinkers) (17, 18) or failure to adjust for sociodemographic and lifestyle factors that might confound the results (16, 17). Furthermore, patients included in these studies typically suffered from major gastrointestinal disorders. Some of these patients may have changed their alcohol consumption subsequent to the development of those disorders. To avoid this problem, we excluded patients with major gastrointestinal disorders from the analysis in a study that we recently carried out among 447 consec- 571

2 572 Brenner et al. utive outpatients of a general practitioner who were aged years (20). In this study, in which current infection with H. pylori was measured by the l3 C-urea breath test, a clear inverse graded relation emerged between alcohol consumption and H. pylori infection after control for confounding variables by multivariable analysis. This relation was seen for both alcohol from beer and alcohol from wine. So far, to our knowledge, no study has assessed the relation of alcohol consumption with current H. pylori infection in population samples. In this study, we assessed this relation in a sample of healthy adults recruited in an occupational setting. MATERIALS AND METHODS Study design and study population All employees of a health insurance company in the city of Ulm, Germany, and their household members aged 5-69 years were invited to participate in a crosssectional study. Participation was voluntary, and informed consent was obtained in each case. The study was approved by the Ethics Board of the University of Ulm. The current analysis is restricted to employees and household members age 15 years and older who were of German nationality. We excluded children and adolescents below age 15 who rarely drink alcohol regularly and among whom a different questionnaire (filled out by the parents) was used. Individuals of a foreign nationality were excluded to avoid language problems. We further excluded subjects with a history of gastric or duodenal ulcer, since they might have changed their alcohol consumption due to their gastrointestinal problems. To avoid false-negative test results, we additionally excluded patients with a history of H. pylori treatment and those under current antibiotic treatment. Data collection Employees were enrolled in the study at the headquarters of the health insurance company in November 1996 during working hours. Active infection with H. pylori was determined by the l3 C-urea breath test. First, an initial breath sample was collected in a plastic bag. The employees then received 200 ml of apple juice (ph ), which contained 75 mg of 13 C-urea (Mass Trace, Wobum, Massachusetts). Thirty minutes later, a second breath sample was collected. The breath samples were analyzed with an isotope-selective, nondispersive, infrared spectrometer (Wagner- Analytical Systems, Bremen, Germany). A change in the Xi COJ l2 CO 1 over baseline of more than five per thousand was considered to indicate active infection. Sensitivities and specificities of the 13 C-urea breath test close to 100 percent have consistently been reported, suggesting that the test is the gold standard in subjects for whom endoscopy is not indicated (1, 21-23). Employees filled out a standardized questionnaire between collection of the first and the second breath samples. The questionnaire included questions on current alcohol consumption (average amount of different alcoholic beverages per week, expressed in terms of the number of typical consumption units in southern Germany) and other factors that were suspected to be potentially related to H. pylori infection, including sociodemographic factors and other lifestyle habits. Employees were given a urea breath test kit and a standardized questionnaire for each household member to be used at home by following simple instructions in the same manner as described above for the employees. These materials were collected at the workplace during the following workday. Statistical analysis The study participants were first described with respect to basic sociodemographic factors and thenalcohol consumption habits. The average total and beverage-specific weekly alcohol consumption was calculated assuming that half a liter of beer or onequarter liter of wine contain 25 g of ethanol (typical consumption units and alcohol contents of beer and wine in southern Germany) and that the alcohol content of one unit of liquor (0.02 liters) is approximately 8 g of ethanol. We then assessed the prevalence of infection with H. pylori among employees by levels of alcohol consumption (none, ^75 g/week, and >75 g/week; the cutpoint for categorization of drinkers reflects the median alcohol consumption in this population). The association between alcohol consumption and active infection with H. pylori was quantified by crude and adjusted odds s and their 95 percent confidence intervals. Adjustment was made by logistic regression with the employee's infection status as the dependent variable and alcohol consumption and the following potential confounders as the independent variables: age (in years), sex, school education (<9, 10-12, and >12 years, reflecting standard categories of the German school system), smoking (never, former, and current), and coffee consumption (none, <2, and >2 cups per day). The latter two covariates were controlled for because of their known correlation with alcohol consumption habits. Additional analyses were carried out in which alcohol consumption from beer and from wine (the two main sources of alcohol in this population) were included as separate variables in the same

3 Alcohol and Helicobacter pylori Infection 573 model to assess beverage-specific effects. All analyses were carried out with the SAS statistical software package (24). A basic assumption in multiple logistic regression is the independence of observations. This assumption is likely to be violated to some extent in our study due to the correlation of observations within households. We therefore carried out additional analyses in which this correlation was taken into account using the generalized estimation equation approach of Liang and Zeger (25, 26). However, the results did not differ materially from those obtained with the more-familiar logistic regression, and therefore, only the latter are reported. RESULTS Overall, 211 of 276 eligible employees (76.5 percent) and 253 members of their households aged years participated in the study. After exclusion of individuals with foreign or unknown nationality (n = 19), history of gastric or duodenal ulcer (n = 15), previous treatment for H. pylori infection (n = 1), or current antibiotic treatment (n = 4), a final sample of 425 persons remained for analysis. TABLE 1. Study population by sex, age, nationality, and school education, Ulm, Germany, 1996 Sex Female Male Age (years) School education (years)* >12t No % Information on school education was missing for two subjects, t Or ongoing school education. Basic sociodemographic characteristics of the study population are given in table 1. A slight majority (55.3 percent) of study participants were women. The mean age was 35.3 years, and most subjects (89.4 percent) were less than age 55 years. About half of the subjects had a medium level of school education. Table 2 provides information on alcohol consumption in the entire sample. About half of the subjects reported drinking beer (50.6 percent) and wine (46.8 percent), respectively, but beer, which was consumed in much larger amounts, was by far the most important source of alcohol in this study population. Spirits provided only a very minor contribution to overall alcohol consumption. As shown in table 3, alcohol consumption varied greatly by age and sex. Alcohol consumption was least common in the youngest age group, and men drank much more alcohol than did women. Overall, 71 among 417 individuals with complete information on the amount of total alcohol consumption (17 percent) were infected with H. pylori, with little variation between categories of alcohol consumption (table 4). The relation between alcohol consumption and infection was strongly confounded by age and gender, however. After control for these factors, consumption of 75 g or less and more than 75 g alcohol per week was associated with a 37 and 66 percent reduction of the odds of H. pylori infection, respectively, compared with abstinence from alcohol (p for trend = 0.024). Control for additional covariates did not materially alter the results. If anything, the observed relation was strengthened further. The multivariable analysis also revealed a strong association of the infection with age (odds per 10-year increase in age = 2.06, 95 percent confidence interval ) and gender (odds for men compared with women = 2.00, 95 percent confidence interval ). Furthermore, there was an approximately twofold increased odds of H. pylori infection among subjects who reported drinking coffee compared with those who did not (odds s for consumption of 2 or less and more than 2 cups of coffee per day = 2.13 and 1.76, respectively). Smoking and H. pylori infection were unrelated in this sample (odds s for current TABLE 2. Alcohol consumption in the study population by beverage type, Ulm, Germany, 1996 Beverage Beer Wine Spirits* Any alcoholic beverage No Drinkers' % Meant Amount of alcohol (g/week) Mediant * Participants who drank more than one type of beverage are listed in all applicable categories. t Among drinkers of each respective beverage. X Information on consumption of spirits was missing for three subjects Range

4 574 Brenner et al. TABLE 3. Alcohol consumption In the study population by age and sex, Ulm, Germany, 1996 Dtnoiayo Age (years) (n= 98) (n= 168) 40-54(n= 108) (n = 43) Sex Female (n = 230) Male (n = 187) None * Information on the amount of incomplete for eight subjects. Alcohol consumption (%) 75gAveek >75 gfweek total alcohol consumption was smokers and former smokers compared with nonsmokers = 0.81 and 1.12, respectively). The inverse graded relation between alcohol consumption and H. pylori infection was particularly strong for alcohol consumed in the form of wine (table 5). Although a monotonic inverse graded relation was also seen between alcohol consumption from beer and H. pylori infection, this relation was weaker and was not statistically significant. DISCUSSION This study among healthy employees and their household members reveals an inverse graded relation between alcohol consumption, in particular, alcohol consumption in form of wine, and active infection with H. pylori. These findings confirm and extend the results of another cross-sectional study we recently carried out among patients of a general practitioner in southern Germany to assess the relation of H. pylori infection to a variety of medical conditions (27, 28). The latter study had also found an inverse graded relation between alcohol consumption and active H. pylori TABLE 5. Relation of alcohol consumption wtth active Hellcobacter pylori Infection by source of alcohol,* Ulm, Germany, 1996 Alcohol from beer (g/week) Nonet 5T75 >75 Alcohol from wine (g/week) Nonet 75 >75 No.f % confidence interval P value * Alcohol consumptions of beer and wine were included as separate independent variables in the same model. s are adjusted for alcohol consumption from another source and for sex, age, school education, smoking, and coffee consumption. t Information on the amount of alcohol consumption from beer and wine was incomplete for two and six subjects, respectively. t Reference category. p value for trend in multivariable analysis. TABLE 4. Relation of total alcohol consumption to active Hellcobacter pylori Infection (HP+), Ulm, Germany, 1996 Alcohol consumption (g/week) infection that was somewhat stronger for alcohol from wine than for alcohol from beer (adjusted odds s for consumption of 75 g or less and more than 75 g alcohol per week compared with abstinence = 0.90 and 0.33, respectively) (20). The consistency of results in both studies is reassuring. Furthermore, the possibility of bias due to disease- related changes of alcohol consumption is even less likely in the current study, which was conducted among healthy adults in an occupational setting, than in our previous study. In contrast to these studies, most previous studies did not find an association between alcohol consumption and H. pylori infection. There are several reasons that might explain this apparent discrepancy. To our knowledge, ours are the only two studies reported to date that address the impact of alcohol con- No. of participants* Nonet > No HP Crude 95% confidence interval Adjusted for age and sex 95% confidence interval p = Adjusted for multiple covariatesf J> % ct^, fiden f 8 Interval ) = * Information on the amount of total alcohol consumption was incomplete for eight subjects, t Adjusted for sex, age, school education, smoking, and coffee consumption. X Reference category. p value for trend.

5 Alcohol and Helicobacter pylori Infection 575 sumption on current infection with H. pylori as measured by the urea breath test. In previous studies, infection status was determined either by Immunoglobulin G serum antibody levels (10-15) or by histopathology or culture (15-19). The serologic tests may remain positive over prolonged time periods after elimination of the infection (1) and may, therefore, be suboptimal for detecting effects of lifestyle habits that might enhance elimination of infection in adulthood. The studies using biopsy-based diagnoses have to be interpreted with caution, as they were typically restricted to patients with gastrointestinal diseases that were severe enough to warrant gastroscopy. Some of these patients may have changed their alcohol consumption because of their gastrointestinal problems, which may bias cross-sectional associations of alcohol consumption with infection status. To avoid this type of bias, we excluded subjects with a history of peptic ulcer from our analyses. Several previous studies examined bivariate associations between alcohol consumption and infection status only, without control for potential confounders (13, 16, 19). Our analysis reveals, however, that control for confounding is essential. In particular, the inverse dose-response relation between alcohol consumption and H. pylori infection would have remained entirely undetected without control for age and sex at the least. There are several mechanisms by which alcohol may affect the living conditions of H. pylori in the stomach. Alcoholic beverages are known to have multiple direct and indirect effects on the gastric mucosa, on gastric emptying (8), and on gastric acid secretion (9). In particular, moderate alcohol consumption may invigorate the mucosal defense by its effects on prostaglandins (29). Last, but not least, alcoholic beverages are known to have strong direct antibacterial activity (5-7). The inverse association between alcohol consumption and active infection with H. pylori therefore appears to be biologically plausible. Recent studies suggest that chronic H. pylori infection and the associated inflammatory response may be risk factors for coronary heart disease (CHD) (30-32). Furthermore, there is now consensus that moderate alcohol consumption lowers the risk of CHD, but the mechanisms underlying this association are not fully understood (33). The results of our study would be consistent with the hypothesis that the protective effect of alcohol on CHD might at least partly be mediated by the impact of alcohol consumption on active H. pylori infection. This hypothesis requires further study. Although our study was larger than the majority of previous studies that addressed the relation of alcohol consumption with active H. pylori infection, it had limited power to assess beverage-specific effects (which may explain the lack of significance of the relation between alcohol consumption from beer and H. pylori infection) or to assess interactions between alcohol consumption and other risk factors. Furthermore, while current alcohol consumption was ascertained in a quantitative manner, there was no information on drinking patterns and past drinking habits, which should also be relevant. We further believe that, as in most other studies, alcohol consumption may have been somewhat underreported. Information on alcohol consumption was given before the infection status was known, however, and potential underreporting should have been nondifferential with regard to infection status, that is, it should have been similar among infected and uninfected subjects. This type of underreporting could not have erroneously produced the patterns observed in our study (34). The same applies to potential changes of alcohol consumption due to dyspeptic symptoms that were unrelated to H. pylori infection in this sample. Therefore, despite its limitations, our study seems to support the hypothesis that alcohol consumption may reduce the odds of active infection with H. pylori. Further studies involving larger numbers of subjects should address in more detail the impact of additional factors, such as drinking patterns, the type of alcoholic beverages consumed, lifetime history of alcohol consumption, and potential interactions between alcohol consumption and other factors that might affect active H. pylori infection among adults. REFERENCES 1. Goodwin CS, Mendall MM, Northfield TC. Helicobacter pylori infection. Lancet 1997;349: Mendall MA, Goggin PM, Molineaux N, et al. Childhood living conditions and Helicobacter pylori seropositivity in adult life. Lancet 1992;339: Webb PM, Knight T, Greaves S, et al. Relation between infection with Helicobacter pylori and living conditions in childhood: evidence for person to person transmission in early life. BMJ 1994;308: Xia HH-X, Talley NJ. Natural acquisition and spontaneous elimination of Helicobacter pylori infection: clinical implications. Am J Gastroenterol 1997;92: Ingram LO, Buttke TM. Effects of alcohols on microorganisms. Adv Microb Physiol 1984;25: Peterson WL, Mackowiak PA, Barnett CC, et al. The human gastric bactericidal barrier mechanisms of action, relative antibacterial activity, and dietary influences. J Infect Dis 1989;159: 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: Jian R, Cortot A, Ducrot F, et al. Effect of ethanol ingestion on postprandial gastric emptying and secretion, biliopancreatic secretions, and duodenal absorption in man. Dig Dis Sci 1986;31: Singer MV, Leffmann C, Eysselein VE, et al. Action of ethanol

6 576 Brenner et al. and some alcoholic beverages on gastric acid secretion and release of gastrin in humans. Gastroenterology 1987; 93: Graham DY, Malaty HM, Evans DG, et al. Epidemiology of Helicobacter pylori in an asymptomatic population in the United States. Effect of age, race, and socioeconomic status. Gastroenterology 1991;100: The EUROGAST Study Group. Epidemiology of, and risk factors for, Helicobacter pylori infection among 3194 asymptomatic subjects in 17 populations. Gut 1993;34: Tsugane S, Tei Y, Takahashi T, et al. Salty food intake and risk of Helicobacter pylori infection. Jpn J Cancer Res 1994; 85: Martin-de-Argila C, Boixeda D, Cant6n R, et al. Helicobacter pylori infection in a healthy population in Spain. Eur J Gastroenterol Hepatol 1996;8: Murray LJ, McCrum EE, Evans AE, et al. Epidemiology of Helicobacter pylori infection among 4742 randomly selected subjects from Northern Ireland. Int J Epidemiol 1997; 26: Paunio M, HOCk-Nikanne J, Kosunen TU, et al. Association of alcohol consumption and Helicobacter pylori infection in young adulthood and early middle age among patients with gastric complaints. Eur J Epidemiol 1994; 10: Rokkas T, Pursey C, Uzoechina E, et al. Campylobacterpylori and non-ulcer dyspepsia. Am J Gastroenterol 1987; 82: Chodos JE, Dworkin BM, Smith F, et al. Campylobacter pylori and gastroduodenal disease: a prospective endoscopic study and comparison of diagnostic tests. Am J Gastroenterol 1988;83: Dooley CP, Cohen H, Fitzgibbons PL, et al. Prevalence of Helicobacter pylori infection and histologic gastritis in asymptomatic persons. N Engl J Med 1989;321: Fontham ETH, Ruiz B, Perez A, et al. Determinants of Helicobacter pylori infection and chronic gastritis. Am J Gastroenterol 1995;90: Brenner H, Rothenbacher D, Bode G, et al. Relation of smoking, alcohol and coffee consumption to active infection with Helicobacter pylori. BMJ 1997;315: Cutler AF, Havstad S, Ma CK, et al. Accuracy of invasive and noninvasive tests to diagnose Helicobacter pylori infection. Gastroenterology 1995;109: Braden B, Schafer F, Caspary WF, et al. Nondispersive isotope-selective infrared spectroscopy: a new analytical method for "C-urea breath tests. Scand J Gastroenterol 1996; 31: Thijs JC, van Zwet AA, Thijs WJ, et al. Diagnostic tests for Helicobacter pylori: a prospective evaluation of their accuracy, without selecting a single test as the gold standard. Am J Gastroenterol 1996;91: SAS Institute, Inc. SAS language: reference. Version 6. Ed 1. Cary, NC: SAS Institute, Inc., Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73: GrOmping U. GEE: a SAS macro for longitudinal data analysis. Version Dortmund, Germany: University of Dortmund, Rothenbacher D, Bode G, Adler G, et al. Use of commonly prescribed antibiotics is not associated with prevalence of Helicobacter pylori infection in adults. Scand J Gastroenterol 1997;32: Brenner H, Rothenbacher D, Bode G, et al. The individual and joint contributions of Helicobacter pylori infection and family history to the risk of peptic ulcer. J Infect Dis 1998; 177: Robert A, Nezamis JE, Lancaster C, et al. Mild irritants prevent gastric necrosis through "adaptive cytoprotection" mediated by prostaglandins. Am J Physiol 1983;245:G Mendall MA, Goggin PM, Molineaux N, et al. Relation of Helicobacter pylori infection and coronary heart disease. Br Heart J 1994;71: Patel P, Mendall MA, Carrington D, et al. Association of Helicobacter pylori and Chlamydia pneumoniae infections with coronary heart disease and cardiovascular risk factors. BMJ 1995;311: Mendall MA, Patel P, Ballam L, et al. C reactive protein and its relation to cardiovascular risk factors: a population based cross sectional study. BMJ 1996;312: Rimm EB, Klatsky A, Grobbee D, et al. Review of moderate alcohol consumption and reduced risk of coronary heart disease: is the effect due to beer, wine, or spirits? BMJ 1996;312: Brenner H, Loomis D. Varied forms of bias due to nondifferential error in measuring exposure. Epidemiology 1994; 5: