- Hermann Brenner, professor of epidemiologya,
- Dietrich Rothenbacher, assistant professor in epidemiologya,
- Günter Bode, assistant professor in internal medicinea,
- Guido Adler, professor of internal medicinea
- Correspondence to: Dr Brenner
- Accepted 29 July 1997
Objective: To assess the relation of smoking and alcohol and coffee consumption to active Helicobacter pylori infection.
Design: Cross sectional study of patients attending a general practitioner. Active H pylori infection was measured by the 13C-urea breath test and detailed quantitative information on smoking and on alcohol and coffee consumption was obtained by a standardised self administered questionnaire.
Setting: One general practice in Germany.
Subjects: 447 patients aged 15–79 who had not had peptic ulcer disease or treatment for H pylori infection.
Main outcome measures: Prevalence of H pylori infection according to smoking and alcohol and coffee consumption.
Results: Overall prevalence of infection was 21% (94/447). There was no significant relation between smoking and active H pylori infection. Alcohol consumption showed a negative dose-response relation and coffee consumption a positive dose-response relation with active infection. After adjustment for potential confounders, the odds ratios for patients who drank ≥3 cups of coffee per day compared with those who did not drink coffee were 1.49 (0.71 to 3.12) and 2.49 (1.23 to 5.03), respectively (P value for trend 0.007).
Conclusion: These results suggest a protective effect of alcohol consumption against active infection with H pylori and an opposite effect of coffee consumption.
Although H pylori infection is commonly acquired during childhood, active infection may be acquired and eliminated throughout adulthood
This study of patients in a general practice found no significant relation between smoking and active H pylori infection
Drinking alcohol seemed to protect against active H pylori infection
This protective effect was dose dependent and similar for beer and wine
Drinking coffee was associated with an increased prevalence of active H pylori infection
The protective effect of alcohol on active H pylori infection may be related to its antimicrobial effects
Helicobacter pylori is a major cause of type B gastritis and is strongly associated with duodenal and gastric ulcer disease.1 Seroprevalence studies have consistently identified socioeconomic conditions in childhood as major risk factors for infection.2 3 By contrast, results on the role of socioeconomic and lifestyle factors during adulthood have been inconsistent.
Most studies addressing the role of lifestyle factors during adulthood have measured IgG titres, which indicate chronic infection, 4 5 6 7 8 9 but infection may be acquired and eliminated throughout adulthood.10 11 Thus serological tests might not be best for assessing the role of factors that might primarily affect current (active) infection. Furthermore, most studies have been of patients with major gastrointestinal disorders, who might have changed their lifestyle since becoming ill.12 13 14 15 16 Other limitations of previous studies include crude assessment of lifestyle habits, such as a simple dichotomisation of alcohol consumption4 15 17; limited power due to a small sample size5 14 15 17; and failure to adjust for risk factors that might confound the association of lifestyle factors with H pylori infection.5 12 14 15 We investigated the relation of H pylori infection to smoking and alcohol and coffee consumption, taking care to overcome or at least minimise these problems.
Subjects and methods
Study design and population
We carried out a cross sectional study among patients who attended a general practitioner in Blaustein, a community of about 15 000 inhabitants in the south of Germany, at fixed weekdays between June and September 1996. Patients aged 15–79 years were recruited during usual visits to the practice. Participation was voluntary, and informed consent was obtained in each case. A total of 501 patients (94% of eligible patients) agreed to participate. Patients with a history of gastric or duodenal ulcer (32 patients) were excluded because they might have changed their lifestyle because of their gastrointestinal problems. To avoid false negative test results, we additionally excluded 16 patients who had been treated for H pylori infection and 6 patients receiving antibiotic treatment, leading to a final sample size of 447 patients.
Active infection with H pylori was determined by breath testing with carbon-13. An initial breath sample was collected in a plastic bag. The patients then received 200 ml of apple juice (pH 2.2-2.4) containing 75 mg non-radioactive 13C-urea (Mass Trace, Woburn, Massachusetts). Thirty minutes later a second breath sample was collected. The breath samples were analysed with an isotope selective non-dispersive infrared spectrometer (Wagner Analytical Systems, Worpswede, Germany). An increase in the ratio of 13C-carbon dioxide to 12C-carbon dioxide between the baseline sample and the 30 minute sample of more than 5 per 1000 was considered to indicate active infection.
Patients were asked to fill out a standardised questionnaire on medical history, sociodemographic factors, and other factors that were suspected to be potentially related to H pylori infection, including smoking, alcohol consumption (average amount of different alcoholic beverages a week), and coffee consumption (average number of cups a day).
We obtained summary statistics relating to smoking and to alcohol and coffee consumption and assessed the bivariate associations between the three variables. The average total and beverage specific weekly alcohol consumption were calculated assuming that 1 litre of beer or 0.5 litres of wine contain 50 g ethanol (which are typical ethanol contents of the beer and wine consumed in Southern Germany) and that the alcohol content of one unit of liquor (0.02 litres) is about 8 g ethanol. We then estimated the relation of smoking and alcohol and coffee consumption to active H pylori infection by crude and adjusted odds ratios. Adjusted odds ratios and their 95% confidence intervals were estimated by multiple logistic regression. The following covariates which were suspected to be related to infection on the basis of previous knowledge were included in the multivariable modelling: age (in years), sex, nationality (German, other), schooling (≥12 years, or ongoing schooling, reflecting standard categories of the German school system), and parental history of gastric or duodenal ulcer (yes, no). Initially, potential two factor interactions between smoking and alcohol and coffee consumption were assessed. None of these interactions was significant (P>0.18 in all cases), however, and therefore no interaction terms were included in the final model. All analyses were carried out with the sas statistical software package.18
Table 1 shows the sociodemographic factors of the study population. About two thirds of the patients were women. The mean age was 42.8 years, and slightly more than half of the patients (57%) were under 45. Most patients (97%) were of German nationality.
About a quarter of the patients were current (26%) or former (25%) smokers. The average number of cigarettes smoked a day by current smokers was 13.5. Other forms of smoking were negligibly rare in this study population.
Table 2 shows alcohol consumption. About half of the patients reported that they drank beer (48%) or wine (49%), but beer, being the main source of alcohol in Germany, was on average consumed in larger amounts. Only 17% reported that they drank spirits. Overall, two thirds of the patients reported drinking some form of alcohol.
Most patients (73%) reported drinking coffee regularly. Mean consumption was 3.3 cups daily (median 3 cups daily).
Smoking and alcohol and coffee consumption were strongly related to each other. For example, current smoking was about twice as common among participants who consumed more than 75 g of alcohol a week (52/141, 37%) than among non-drinkers (30/152, 20%), or among participants who drank 3 or more cups of coffee a day (64/187, 34%) than among participants who did not drink coffee (22/120, 18%). Coffee and alcohol consumption were likewise positively related.
Table 3 shows the relation of smoking and alcohol and coffee consumption to active H pylori infection. The overall prevalence of infection was 21% (94/447). Infection was slightly more common among former smokers (25%) and current smokers (22%) than among those who had never smoked (18%), and it was considerably higher among participants who drank 3 or more cups of coffee a day (28%) than among other participants. These patterns persisted after control for potential confounding in multivariable analysis, which showed a positive dose-response relation between coffee drinking and infection (P value for trend 0.007), with a moderate but not significant increase in the odds of infection among former and current smokers (odds ratios 1.48 and 1.57, respectively). By contrast, we observed a negative association between alcohol consumption and H pylori infection that was considerably strengthened after controlling for smoking, coffee consumption, and the other covariates in multiple logistic regression (P value for trend 0.005). In particular, there was a clear, highly significant reduction in the prevalence of infection among participants who reported drinking more than 75 g ethanol a week compared with non-drinkers (adjusted odds ratio 0.33 (95% confidence interval 0.16 to 0.68)). There was also a significant positive relation between age and foreign nationality and the risk of infection.
To assess whether the association between alcohol consumption and infection was confined to a specific type of alcoholic beverage, we carried out additional analyses in which alcohol from beer and wine were entered as separate variables in the multivariable modelling. A negative association between the amount of alcohol consumed and H pylori infection was seen for both beer and wine (P value for trend 0.027 and 0.023, respectively), although the association was slightly more pronounced for wine (Table 4).
We found a moderate but not significant increase in the prevalence of H pylori infection among smokers, a negative dose-response relation between alcohol consumption and active H pylori infection, and a positive dose-response relation between coffee consumption and active infection. The negative dose-response relation between alcohol consumption and active infection applied to both beer and wine.
A limitation of our study is the fact that information on smoking and on alcohol and coffee consumption was ascertained by a self administered questionnaire without verification by biological markers. Self reports of smoking are typically reasonably accurate,19 but we believe that alcohol consumption, as in other epidemiological studies, has probably been underreported to some extent. Information on alcohol consumption was given before H pylori infection was known, and potential underreporting would most likely have been non-differential with respect to infection. Such underreporting could not have produced the observed patterns if alcohol consumption was unrelated to H pylori infection.20
Our results on the relation between smoking and H pylori infection are similar to those from other studies. Despite differences in the populations studied and the methods used, most studies reported slightly increased odds of H pylori infection among smokers,4 6 12 13 15 but these odds were significant in only two studies.12 13 These patterns are consistent with a possible weak increase in active H pylori infection by smoking, which might be too subtle to be verified in most single epidemiological studies because of limited power.
The observed net association of smoking with active H pylori infection may result from various mechanisms with partly antagonistic effects on the risk of infection. Potentially relevant effects of smoking include an increase in acid and pepsin secretion and changes in gastric motility, prostaglandin synthesis, gastric mucosal blood flow, and mucus secretion.21
Several studies have reported the relation between alcohol consumption and H pylori infection.4 6 9 13 14 15 16 17 Most of them did not find a significant association. A recent study from Finland found a positive association between alcohol consumption and H pylori infection among younger people (aged 18-35) and a negative association among older age groups (≥46).9 Yet, interpretation and generalisation of these findings are difficult because the study was conducted among military staff who had endoscopy for gastric complaints and who drank heavily.
In contrast to most previous studies, we ascertained alcohol consumption in a strictly quantitative manner by type of beverage. Our results suggest a major protective effect of alcohol at moderate and high consumption but not at low consumption, which might explain the failure to detect any effect of alcohol in studies in which participants were simply classed as drinkers or non-drinkers. Furthermore, previous studies did not control14 15 or only partly17 controlled for factors that may confound the association between alcohol consumption and H pylori infection. Our analysis shows that the negative association between alcohol consumption and H pylori infection may easily remain undetected unless relevant confounders are adjusted for.
Another reason for the differences between our results and previous results on the role of alcohol consumption may be how infection was measured. Most previous studies used IgG titres to measure infection.4 6 9 High IgG titres may persist after elimination of active infection,10 so elimination of infection resulting from alcohol consumption might be missed in studies using IgG titres as markers of infection.
Several mechanisms might mediate the apparent protective effects of alcohol consumption. Alcoholic beverages have many direct and indirect effects on the gastric mucosa, on gastric emptying,22 and on gastric acid secretion,23 and these may affect the living conditions of H pylori in the stomach. In particular, moderate alcohol consumption might invigorate mucosal defence by its effects on prostaglandins.24 Both beer and wine are potent stimulants of acid secretion and gastrin release,23 and wine has strong antibacterial activity.25
The positive relation between coffee consumption and H pylori infection identified in our study is consistent with results from a cohort study among epidemiologists in which the risk of seroconversion (change from negative to positive results for antibodies to H pylori in serum was 4.6 times higher among those who drank more than 2 cups of caffeinated drinks a day than among the others.11 The mechanisms underlying this association require further research.
Our study in patients of a general practitioner suggests a protective effect of alcohol consumption against active H pylori infection and an opposite effect of coffee consumption. Further research should address the observed associations in representative population samples and aim to obtain a more complete understanding of their underlying mechanisms.
Funding: No external funding.
Conflict of interest: None.