Secondary Logo

Share this article on:

The Protective Effect of Alcoholic Beverages on the Occurrence of a Salmonella Food-Borne Outbreak

Bellido-Blasco, Juan B.1; Arnedo-Pena, Alberto1; Cordero-Cutillas, Enrique1; Canós-Cabedo, Manuel2; Herrero-Carot, Concha1; Safont-Adsuara, Lourdes1


Background.  There are few studies about the effect of alcoholic beverages in food-borne disease outbreaks.

Methods.  We studied this effect during an outbreak of Salmonella ohio infection that occurred in Castellón (Spain) in May 2000. S. ohio was isolated in stools of patients and also in implicated foods.

Results.  Analysis was restricted to the group that was exposed to contaminated food (51 persons). We compared attack rates by level of alcohol intake (none, 1–40 gm, and >40 gm). The adjusted relative risks in the two alcohol categories were 0.73 [95% confidence interval (CI) = 0.46–1.15] and 0.54 (95% CI = 0.31–0.96), respectively.

Conclusions.  We discuss the possible protective effect of alcoholic beverages in some food-borne infectious outbreaks.

From the 1Center of Public Health and

2Gran Vía Hospital, Castellón, Spain.

Address correspondence to: Juan B. Bellido-Blasco, Sección de Epidemiología, Centro de Salud Pública de Castellón, Avda del Mar 12 12003 Castellón, Spain;

Submitted May 7, 2001; final version accepted November 15, 2001.

Very few studies have examined the effects of alcoholic beverages when a food-borne outbreak occurs. 1,2 Recently, we had the opportunity to conduct such a study during a food-borne infection that took place in Castellón (Spain). The infectious agent, etiology, and mechanism of transmission were well established. The causal agent was Salmonella ohio, isolated in stool samples and also in epidemiologically incriminated foods.

Back to Top | Article Outline


The outbreak occurred in May 2000, after a celebration banquet attended by 120 people. One hundred two (85%) people who had attended the banquet gave their consent to be interviewed about food consumption (18 items) and symptoms. Fifty-nine reported symptoms and 43 did not. The case definition was vomiting or diarrhea (≥2 loose stools per day) plus abdominal cramps, fever (≥37.5°C), or headache. We excluded 12 sick people who did not meet this case definition.

The analysis was conducted with the remaining 90 people (53% female; mean age 34 years, range 5–70 years). First, for each of 18 food items rates of disease and their relative risks (RRs) were calculated. Then we used multiple logistic regression in a model that identified four food items as being associated with disease, after adjustment by age and sex. Stool samples from 20 cases and food handlers, as well as samples of six food items consumed, were analyzed in the Public Health Laboratory of Castellón, the Hospital Gran Vía, and the National Center of Microbiology in Madrid (Spanish reference laboratory for Salmonella).

To analyze the possible protective effect of consumption of alcoholic beverages during the banquet, we asked all 90 people about drinks containing alcohol (beer, wine, champagne, or liquors) and the ingested quantity of each (number of glasses), using a self-administered questionnaire. Subjects were not aware of our hypothesis. Seventy-eight people (87%) responded to these questions; of these, 58 had consumed at least one of the contaminated foods.

We evaluated the risk of disease in drinkers and nondrinkers among the group exposed to contaminated foods (45 sick and 13 not sick, 51 of whom provided sufficient information about the alcohol consumption). We restricted analysis to the exposed group because the protective effect against Salmonella, if any, must be present in this subsample. Alcohol was considered in two ways: qualitative (yes/no) and by level (1–40 gm, >40 gm), based on the estimated grams of alcohol per serving (beer, 10 gm per 250 ml; wine, 12 gm per 100 ml; champagne, 8 gm per 100 ml; and liquors, 36 gm per 100 ml). 3

We used Epi Info, version 6.0, and EGRET for logistic regression or LogXact as appropriate.

Back to Top | Article Outline


The global attack rate was at least 40% (48/120). Demographic and clinical characteristics of the 90 contacted people are presented in Table 1. Age and sex differences between sick and unaffected people were small. Eighteen food items were studied in a simple analysis. After adjustment, we found the main effect for potato salad with mayonnaise and for tuna sandwiches, with an RR of 22.3 [95% confidence interval (CI)= 3–154]. Salmonella ohio 6.7.:b:lw was isolated in 15 of 20 stool samples, and in both the potato salad and tuna sandwiches.

Table 1

Table 1

There were only small differences in age, sex, and attack rate between the 78 respondents and the 12 nonrespondents to the alcohol questions (mean age, 34 years vs 40; percentage female, 53%vs 58%; and percentage cases 54%vs 42%). Analysis restricted to those who answered the alcohol questions show the same implicated foods.

Among the 51 people in the subanalysis, we observed a mean difference of 33.2 gm of alcohol intake between the 40 ill people and the 11 others (95% CI = 12.1–54.3). Table 2 shows the risk of disease among people who ate the contaminated food, by level of alcohol consumption. There was a protective effect of alcohol that was stronger among the people who drank more than 40 gm of alcohol. Attack rates were 95%, 78%, and 54%, respectively, for those drinking 0, 1–40, and more than 40 gm (chi-squared linear trend P = 0.04). There was little difference in attack rates by type of alcoholic beverage (liquor compared with beer or wine; data not shown).

Table 2

Table 2

Age could confound these results if young people (who were nondrinkers) were also at higher risk of illness. Excluding all subjects who were less than 15 years old did not alter the analysis.

Back to Top | Article Outline


Our data suggest that the intake of alcoholic beverages during or immediately after consumption of contaminated food may be a protective factor for Salmonella acute gastrointestinal infection. A dose-response effect was observed.

Information bias is a potential issue. It is possible that ill people reduce their declared alcohol exposure more than non-ill. To reduce disease misclassification, we excluded 12 individuals who did not meet the case definition. Selection bias could also be present. Eighteen people of 120 (15%) were not contacted. Among those who answered questions about alcohol (N = 78), repeated analyses of food items show the same results for implicated foods as with the entire group of 90. Therefore, the relation between disease and food exposure was not different in the subsample that responded to the alcohol questionnaire and generated the comparison groups of drinkers and nondrinkers. Also, alcohol was not substantially different consumption between those exposed and those not exposed to the contaminated food (21.1 vs 24.5 gm).

Confounding by age and sex seems not to be present. Age could be a confounder if children (all nondrinkers) had different attack rates from those of nondrinking adults, but we did not find this. We did not ask for underlying medical conditions that could be related to the risk of acquiring salmonellosis. This was in general a group of healthy people, and no illness other than gastrointestinal illness was described during the interviews.

Results of this study are consistent with our previous work during an S. enteritidis outbreak 4 and also with other studies about hepatitis A virus 5 and Helicobacter pylori infection. 6 A dose response was not always present in these studies, 4,5 with some suggesting a protective effect of alcohol only at high doses.

Changes in stomach pH, bactericidal effects of ethanol, or other substances in the alcoholic beverages 7–10 could explain the acute protective effect against infection seen in outbreaks, or the more persistent protective effects seen in other settings (H. pylori). 6 The quantity and nature of contaminated food ingested with alcoholic beverages and the type of infectious agent also have to be considered.

These findings do not mean that alcoholic beverages should be recommended as a protective factor against infectious food-borne diseases. The universal recommendations for good practices among food handlers and food safety are the best way to prevent such outbreaks.

We thank Joan B. Soriano for the revision of this manuscript and his counseling, and Ferran Ballester and Josep R. Guzman for his help in the English version.

Back to Top | Article Outline


1. Klonz KC. Does imbibing alcohol protect against enteric pathogens? Epidemiology 1999; 10: 207–209.
2. San Molina L. ¿Tiene el alcohol algún efecto beneficioso? Med Clin (Barc) 1996; 107: 655–656.
3. Barros C. El consumo de alcohol a través de distintas bebidas alcohólicas. Alimentaria 1995; (suppl): 17–32.
4. Bellido-Blasco JB, González-Morán F, Arnedo-Pena A, et al. Brote de infección alimentaria por Salmonella enteritidis: posible efecto protector de las bebidas alcohólicas. Med Clin (Barc) 1996; 107: 641–644.
5. Desenclos JCA, Klonz KC, Wilder MH, Gunn RA. The protective effect of alcohol on the occurrence of epidemic oyster-borne hepatitis A. Epidemiology 1992; 3: 371–374.
6. Brenner H, Berg G, Lappus N, Kliebsch U, Bode G, Boeing H. Alcohol consumption and Helicobacter pylori infection: result from the German National Health and Nutrition Survey. Epidemiology 1999; 10: 207–209.
7. Ingram LO, Buttke TM. Effects of alcohols on micro-organisms. Adv Microb Physiol 1984; 25: 253–300.
8. Peterson WL, Mackowiak PA, Barnett CC, Marling-Cason M, Haley ML. The human gastric bactericidal barrier: mechanisms of action, relative antibacterial activity, and dietary influences. J Infect Dis 1989; 159: 979–983.
9. Notermans S, Soentoro PS, Delfgou-Van Asch EH. Survival of pathogenic microorganisms in an egg-nog-like product containing 7% ethanol. Int J Food Microbiol 1990; 10: 209–217.
10. Bujanda J. The effects of alcohol consumption upon gastrointestinal tract. Am J Gastroenterol 2000; 95: 3374–3382.

protective effect; alcoholic beverages; food-borne; Salmonella

© 2002 Lippincott Williams & Wilkins, Inc.