Original Contributions A Large Community Outbreak of Salmonellosis Caused by Intentional Contamination of Restaurant Salad Bars Thomas J. To¨ro¨k, MD; Robert V. Tauxe, MD, MPH; Robert P. Wise, MD, MPH; John R. Livengood, MD; Robert Sokolow; Steven Mauvais; Kristin A. Birkness; Michael R. Skeels, PhD, MPH; John M. Horan, MD, MPH; Laurence R. Foster, MD, MPH† Context.—This large outbreak of foodborne disease highlights the challenge of OUTBREAKS of foodborne infection investigating outbreaks caused by intentional contamination and demonstrates the are caused by foods that are intrinsically vulnerability of self-service foods to intentional contamination. contaminated or that become contami­ Objective.—To investigate a large community outbreak of Salmonella Typhi- nated during harvest, processing, or murium infections. preparation.Itisgenerallyassumedthat such contamination events occur inad­ Design.—Epidemiologic investigation of patients with Salmonella gastroenteri- vertently; intentional contamination tis and possible exposures in The Dalles, Oregon. Cohort and case-control inves- with a biologic agent is rarely suspected tigations were conducted among groups of restaurant patrons and employees to or reported.1,2 identify exposures associated with illness. On September 17, 1984, the Wasco- Setting.—A community in Oregon. Outbreak period was September and Octo- Sherman Public Health Department in ber 1984. Oregon began to receive reports of per­ Patients.—A total of 751 persons with Salmonella gastroenteritis associated with sonsillwithgastroenteritiswhohadeaten eating or working at area restaurants. Most patients were identified through pas- at either of 2 restaurants in The Dalles, sive surveillance; active surveillance was conducted for selected groups. A case Ore, several days before symptom onset. was defined either by clinical criteria or by a stool culture yielding S Typhimurium. Local and state public health officials con­ firmed an outbreak of Salmonella Typhi­ Results.—The outbreak occurred in 2 waves, September 9 through 18 and murium associated with the 2 restaurants September 19 through October 10. Most cases were associated with 10 restau- and then noted an abrupt increase in re­ rants, and epidemiologic studies of customers at 4 restaurants and of employees ports of gastroenteritis the following at all 10 restaurants implicated eating from salad bars as the major risk factor for week among persons who had eaten or infection. Eight (80%) of 10 affected restaurants compared with only 3 (11%) of the worked at other restaurants in The 28 other restaurants in The Dalles operated salad bars (relative risk, 7.5; 95% con- Dalles. Because many patients reported fidence interval, 2.4-22.7; P<.001). The implicated food items on the salad bars eating food from salad bars, the local differed from one restaurant to another. The investigation did not identify any water health department closed all salad bars in supply, food item, supplier, or distributor common to all affected restaurants, nor the town on September 25, 1984, and the were employees exposed to any single common source. In some instances, Oregon Health Division requested assis­ tance from the Centers for Disease Con­ infected employees may have contributed to the spread of illness by inadvertently trol (CDC) for further evaluation and con­ contaminating foods. However, no evidence was found linking ill employees to ini- trol of the outbreak. tiation of the outbreak. Errors in food rotation and inadequate refrigeration on ice- The epidemiologic investigation identi­ chilled salad bars may have facilitated growth of the S Typhimurium but could not fied the vehicles of transmission as foods have caused the outbreak. A subsequent criminal investigation revealed that on multiple self-service salad bars and members of a religious commune had deliberately contaminated the salad bars. An probable times when contamination oc­ S Typhimurium strain found in a laboratory at the commune was indistinguishable curred. Common mechanisms by which from the outbreak strain. salad bars could have become contami­ Conclusions.—This outbreak of salmonellosis was caused by intentional con- natedwereexcluded.Asubsequentcrimi­ tamination of restaurant salad bars by members of a religious commune. nal investigation found that members of a JAMA. 1997;278:389-395 nearby religious commune had intention­ ally contaminated the salad bars on mul­ tiple occasions. From the National Center for Infectious Diseases ruary 28, 1985; 99th Congress, 1st Session: H901­ and Epidemiology Program Office, Centers for H905, and at the Epidemic Intelligence Service 34th BACKGROUND Disease Control and Prevention, Atlanta, Ga annual conference, Atlanta, Ga, April 23, 1985. The Dalles, population 10 500 (1980 (Drs To¨ro¨k, Tauxe, Wise, Livengood, and Horan and Trade names are used for identification only and Ms Birkness); and the Oregon Health Division, does not imply endorsement by the US Department census), is the county seat of Wasco Portland (Messrs Sokolow and Mauvais and of Health and Human Services or the US Public County, population 21 000,aregion of or- Drs Skeels and Foster). Dr Wise is now with the US Health Service. chards and wheat ranches. Located near Food and Drug Administration. Reprints: Thomas J. To¨ro¨k, MD, Centers for Disease †Deceased. Control and Prevention, Mailstop G-17, 1600 Clifton Rd the Columbia River on Interstate 84, Presented in part in the Congressional Record, Feb- NE, Atlanta, GA 30333. The Dalles is a frequent stop for travel- JAMA, August 6, 1997—Vol 278, No. 5 Intentional Restaurant Salad Bar Contamination—To¨ro¨k et al 389 ©1997 American Medical Association. All rights reserved. ers. Two independent water systems Group 3 restaurants were not affected Oregon Public Health Laboratory cul­ serve The Dalles: a smaller system sup­ and had no case customers with an SRE. tured suspected foods. plied by a well and a larger system that The Dalles outbreak strain was com­ serves most restaurants and uses sur­ Outbreak Investigation pared with human isolates included in 2 face water augmented by well water national surveys of salmonellae in 1979 during the summer. From 1980 through Caseswere identifiedthrough passive and 1980 and in 1984 and 1985.8,9 To iden­ 1983,only16isolatesof salmonellaewere surveillance. Press releases encouraged tify a possible animal reservoir, CDC reportedby thelocal healthdepartment; reporting by case patients and health characterized all available veterinary 8 isolates were S Typhimurium. No case care professionals. We interviewed pos­ isolates of S Typhimurium identified be­ of salmonellosis was reported in the first sible case patients about symptoms and tween October 1, 1984, and September 8 months of 1984. riskfactorsandobtainedcomprehensive 30, 1985, by the US Department of Ag­ In 1981, followers of Bhagwan Shree foodhistoriesforrestaurantmeals eaten riculture National Veterinary Services Rajneesh purchased a large ranch in during the 3-day period before onset of Laboratory in Ames, Iowa. Wasco County to build a new interna­ symptoms. Case customerswithanSRE tional headquarters for the Indian were asked to identify all other persons Environmental Studies guru.3-5 Construction of the commune with whom they had eaten at the restau­ Local health department sanitarians was controversialfrom its inception; cul­ rant. Histories were obtained from per­ and US Food and Drug Administration turalvalues andland-useissues werethe sons so identified, and those who were representativesinvestigatedthedistribu­ major areas of conflict. Part of the ranch not ill and reported no other restaurant tors and original suppliers of foods used in was incorporated as the city of Raj­ exposure served as controls for food- group 1 restaurants. All group 1 restau­ neeshpuram, but the charter was chal­ specific case-control analyses. Poten­ rants were inspected by sanitarians. Rec­ lenged in the courts, effectively limiting tially exposed cohorts, such as banquet ords of the city water system were re­ new construction. Commune members participants and take-out food patrons, viewed for the month of September 1984. believed that the outcome of the Novem­ wereidentifiedfromrestaurant records, Tap water samples were collected during ber 6, 1984, elections for Wasco County and attempts were made to interview the outbreak from restaurants for analy­ commissioners would have an important these persons. sis. Temperatures maintained by ice- impact on further land-use decisions.3-5 Employees of group 1 restaurants were chilled salad bars were evaluated. interviewed twice. During the out­ METHODS break, investigators interviewed em­ Statistical Analysis ployees when restaurant involvement Food exposure data were analyzed Case Definition was first suspected. In October 1984, im­ separately by restaurant and by date of A case was defined as an illness with mediately following the outbreak, all em­ onset of illness at the 2 restaurants that diarrhea and at least 3 of the following ployees were asked to complete a self- had recurrent outbreaks. Univariate symptoms: fever, chills, headache, nau­ administered questionnaire. Work analyses were performed and odds ra­ sea, vomiting, abdominal pain, or bloody schedules were obtained from review of tios (ORs) with 95% confidence intervals stools, or by a stool culture yielding S time cards, interviews with restaurant (CIs) were calculated using the Epi Info Typhimurium. A patient was considered managers, and review of insurance claims computerprogramVersion6.03.10