[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Citations 0
News From the Centers for Disease Control and Prevention
October 13, 2010

Multiple-Serotype Salmonella Gastroenteritis Outbreak After a Reception—Connecticut, 2009

JAMA. 2010;304(14):1547-1549. doi:

MMWR. 2010;59:1093-1097

1 table omitted

In September 2009, the Connecticut Department of Public Health (DPH) identified an outbreak of Salmonella gastroenteritis among attendees at a reception. A case-control study and environmental and laboratory investigations were conducted. Nine case-patients and 14 control subjects were identified. Potato salad consumption was strongly associated with illness (odds ratio [OR] = 84.0). During the investigation, food service workers were observed to have bare-handed contact with ready-to-eat food. Five case-patients and one asymptomatic food service worker had stool samples positive for Salmonella species. Two Salmonella serotypes were identified, Salmonella enterica serovar Schwarzengrund and Salmonella enterica serovar Typhimurium variant O:5–, including coinfection in one case-patient and one food service worker. The isolates of each respective serotype (S. Schwarzengrund and S. Typhimurium variant O:5–) had indistinguishable pulsed-field gel electrophoresis (PFGE) patterns. Potato salad was the likely source of the outbreak but the contamination mechanism is unclear. Control measures included exclusion of the food service worker with Salmonella -positive stool from the restaurant until two consecutive stool samples yielded no bacterial growth. Standard public health laboratory practices in Connecticut and testing techniques used specifically during this investigation led to the rapid identification of the two serotypes. Multiple-serotype Salmonella outbreaks might occur more frequently than recognized; knowledge of all Salmonella serotypes involved in an outbreak might help implicate the outbreak source, define the scope of the outbreak, and determine the selection of appropriate control measures.

On September 18, 2009, a physician notified the DPH Epidemiology and Emerging Infections Program of a laboratory-confirmed Salmonella infection in a person who had attended a reception at a banquet hall on September 6. Preliminary information indicated that other attendees became symptomatic with gastrointestinal illness after the reception. Food served at the reception was prepared at an off-site licensed restaurant, delivered to the banquet hall by restaurant staff, and set up as a self-serve buffet. DPH and the local health department conducted an investigation to determine the source and extent of the illnesses and to recommend control measures.

A case-control study was conducted among attendees. A case was defined as diarrhea (three or more loose stools during a 24-hour period) in a reception attendee within 5 days after the reception. A control subject was defined as an attendee who did not experience gastrointestinal illness. Because no guest list existed, contact information for ill attendees was provided by the reception host; control subjects and additional case-patients were recruited by asking known attendees to identify and provide contact information for other attendees. Contact information was obtained for 25 (17%) of the approximately 150 attendees. DPH conducted telephone interviews during September 21-25 regarding illness history and food consumed at the reception; an itemized list of foods served at the reception was used to obtain food consumption histories. Of the 25 interviewed attendees, nine (36%) met the case definition, 14 qualified as control subjects, and two were excluded because they reported gastrointestinal illness that did not meet the case definition. Of the nine case-patients, eight (89%) had abdominal cramping, seven (78%) had subjective fever, six (67%) had muscle aches, and four (44%) had bloody stools. Median age was 31 years (range: 25-51 years); five (56%) were male. The median incubation period* was 13.5 hours (range: 9.5-95.5 hours); median illness duration was 8.5 days (range: 0.5-14 days). A case-control analysis revealed that case-patients were significantly more likely than control subjects to have consumed potato salad (88% versus 8%, respectively; OR = 84.0; 95% confidence interval = 3.3-4,077; p<0.001).

During September 21–October 1, the local health department and the DPH Food Protection Program conducted an environmental investigation of the restaurant in which the food served at the reception had been prepared. Of the four persons who worked at the restaurant, two were directly involved in food preparation for the reception. All four were interviewed, and none reported experiencing gastrointestinal illness. During the investigation, food service workers were observed to have bare-handed contact with ready-to-eat food and did not practice adequate hand washing. Preparation procedures of items served at the reception, including the potato salad, were reviewed, and environmental samples of food contact surfaces and spices used in preparation of the reception food were collected for testing. The environmental and spice samples were obtained >3 weeks after the outbreak occurred and after the facility had been cleaned; Salmonella was not detected in these samples . No leftover potato salad was available for testing.

The stool sample from the index case-patient was collected on September 14 and processed at a private laboratory; the clinical isolate was then sent to the DPH laboratory for confirmation. Stool specimens from five additional case-patients and all four food service workers were collected during September 21–October 7 and tested at the DPH laboratory. The specimens were first plated to selective media to test for the presence of Salmonella, Shigella, Campylobacter, and Escherichia coli O157. After incubation, presumptive Salmonella colonies were serotyped† and subtyped genetically by PFGE. Serotyping and PFGE testing were not sequential.‡

The isolate from the index case-patient was serotyped as S. Typhimurium variant O:5–. Initial serotyping steps performed on Salmonella isolates obtained from stool specimens revealed a preliminary antigen result consistent with the S. Typhimurium variant O:5– already identified for the index case-patient. Consequently, investigators assumed that S. Typhimurium variant O:5– was the only outbreak serotype. Next, while final serotyping was pending, Salmonella isolates were submitted for PFGE. Testing of the first five isolates yielded two distinct PFGE patterns (PFGE XbaI patterns JPXX01.0456 and JM6X01.0036§). One PFGE pattern appeared to be consistent with S. Typhimurium; the other appeared to be consistent with S. Schwarzengrund. The results of serotyping verified the presence of both S. Typhimurium variant O:5– and S. Schwarzengrund.

The identification of both S. Typhimurium variant O:5– and S. Schwarzengrund in reception attendees raised the possibility that two different Salmonella serotypes might be involved in the outbreak. Therefore, laboratory staff systematically collected multiple single-colony picks from original media to screen for the presence of an additional Salmonella serotype. After all testing was complete, including isolation, serotyping, and PFGE, two of the six case-patients with specimens at the DPH laboratory were determined to be infected with S. Typhimurium variant O:5–, another two with S. Schwarzengrund, and one with both; no pathogens were isolated from the stool specimen of the sixth case-patient. A seventh case-patient's stool specimen was tested at a private laboratory; no Salmonella was detected. Of the four food service worker specimens tested, one yielded both S. Schwarzengrund and S. Typhimurium variant O:5– and the other three were negative. All respective S. Schwarzengrund isolates and S. Typhimurium variant O:5– isolates had indistinguishable PFGE patterns.

On September 25, the food service worker with positive stool findings was reinterviewed and reaffirmed the absence of recent gastrointestinal illness, including around the time of the reception. This food service worker had been responsible for transporting food to the banquet hall and ensuring that the food was maintained at the correct temperature before serving, but reported not having prepared, consumed, nor served any of the food.

Control measures implemented by the local health department included exclusion of the Salmonella -positive food service worker from the restaurant for approximately 2 weeks until two consecutive stool cultures obtained ≥24 hours apart had no bacterial growth. Health department staff members provided information about employee health policies and employee hygiene to the restaurant owners and reviewed the information with them.

Reported by:

L Mank, MS, M Mandour, Connecticut Dept of Public Health Laboratory; T Rabatsky-Ehr, MPH, Q Phan, MPH, J Krasnitski, MPH, J Brockmeyer, MPH, L Bushnell, C Applewhite, M Cartter, MD, Connecticut Dept of Public Health. J Kattan, MD, EIS Officer, CDC.

CDC Editorial Note:

Epidemiologic and laboratory data demonstrate that an outbreak of Salmonella infection with two different serotypes occurred among guests who attended a reception; potato salad was the likely source of this outbreak, but the contamination mechanism is unclear. Likewise, whether the food service worker might have unknowingly contaminated a food item or whether the food service worker also was infected through the same source as the case-patients remains unclear.

Salmonella is the most common bacterial cause of foodborne disease outbreaks in the United States.1 However, outbreaks of Salmonella infection with multiple serotypes are reported less commonly in the literature.2,3,4 Standard public health laboratory practices in Connecticut,∥ as well as testing techniques used specifically in the context of this outbreak investigation, led to the rapid identification of two distinct serotypes. Connecticut requires all identified Salmonella isolates to be submitted to the DPH laboratory, where serotyping and PFGE are routinely performed; private laboratories in Connecticut do not have the capacity to perform full serotyping and PFGE testing. Initiating PFGE testing before finalization of serotyping led to more rapid recognition of the two different serotypes; complete serotyping can take days, whereas PFGE testing can take as little as 1 day after pure isolate is available for analysis.

Systematically screening and testing multiple single-colony picks on each original culture plate, a time-intensive practice that is usually not part of routine laboratory protocol, facilitated identifying both outbreak serotypes. This approach particularly aided discovery of coinfection with two Salmonella serotypes in one case-patient and the food service worker. The recognition of coinfection helped investigators conclude that a multiple-serotype outbreak had occurred. Furthermore, testing multiple colonies is dependent upon availability of stool specimens; had all of the case-patients' stool been first tested at a private laboratory, such that only single clinical isolates were available for testing at the DPH laboratory, coinfection in the case-patient would not have been discovered.

Not all states require that all Salmonella isolates be submitted to the public health laboratory for serotyping and PFGE. Additionally, in an outbreak setting, some states with resource limitations might only perform comprehensive testing on a very limited number of case-patient specimens. If the outbreak described in this report had taken place in a state without a requirement for submission of Salmonella isolates to the public health laboratory or in a state in which the number of specimens tested was strictly limited, the discovery of both Salmonella serotypes might not have occurred. In those public health laboratories that perform both serotype and PFGE testing, but do not do so simultaneously, multiple-serotype infections would not be identified as quickly as they were in this outbreak.

Although not specifically illustrated by the findings in this report, not knowing about all Salmonella serotypes involved in an outbreak might hinder the epidemiologic investigation and the public health response. Certain Salmonella serotypes are known to be likely associated with particular food types or animal sources. Consequently, knowledge of multiple serotypes involved in an outbreak can help focus the investigation on potential outbreak sources. Databases, such as PulseNet,¶ can identify and link infected persons to a particular outbreak, thereby defining the scope. In a recent outbreak, PulseNet matched two different Salmonella serotypes to an outbreak linked to peppers used in making salami.4 If an outbreak were detected through PulseNet, not knowing all involved serotypes might result in cases not being associated with the outbreak. If only cases with a single serotype were included in such responses, sampled cases might not be representative of all cases. Furthermore, identifying a greater number of cases associated with multiple serotypes in an outbreak might increase the statistical power of the study to implicate a food vehicle or other outbreak source through epidemiologic analysis. Implementation of appropriate control measures relies on knowing the implicated source and the scope of the outbreak, particularly if multiple serotypes are involved.

The findings in this report are subject to at least three limitations. First, lack of a comprehensive guest list prohibited a cohort analysis. Second, recruitment of control subjects through known attendees might have introduced selection bias; attendees who knew each other might have had similar food preferences, potentially increasing the likelihood that case-patient and control subject food histories were similar. However, such a tendency would bias the results toward showing no association. Finally, the time lag between the reception and collection of environmental samples limited their usefulness.

Multiple-serotype Salmonella outbreaks might occur more frequently than recognized. Health departments should be aware of the possible occurrence of such outbreaks to better characterize their epidemiology. This outbreak demonstrates the importance of capacity to perform Salmonella serotyping and PFGE testing at public health laboratories. In outbreak settings, obtaining stool samples and performing comprehensive serotyping and PFGE at public health laboratories facilitate detection of multiple Salmonella serotypes. When more than one Salmonella serotype is suspected in an outbreak, screening and testing multiple single-colony picks could be considered, if resources permit, as an important technique for identifying multiple serotypes, including coinfection, among cases.


This report is based, in part, on contributions by local health department staff; K Desy, MPH, K Purviance, MPH, L Sosa, MD, Connecticut Dept of Public Health; A Kinney, D Barden, J Fontana, PhD, Connecticut Dept of Public Health Laboratory; J Hadler, MD, Connecticut Emerging Infections Program; and J Magri, MD, and I Williams, PhD, CDC.

What is already known on this topic?

Salmonella commonly causes foodborne illness; however, Salmonella outbreaks involving multiple serotypes are reported less commonly.

What does this report add?

Epidemiologic and laboratory data demonstrate that an outbreak of Salmonella infection with two different serotypes occurred among guests who attended a reception; rapid identification of the multiple serotypes was facilitated by confirmatory testing at the state laboratory, specifically the use of stool samples for subsequent serotyping and pulsed-field gel electrophoreses testing.

What are the implications for public health practice?

Multiple-serotype Salmonella outbreaks might occur more frequently than recognized; if resources permit, health departments can better characterize the epidemiology of Salmonella outbreaks by performing serotyping and PFGE, and by testing multiple single-colony picks when multiple Salmonella serotypes are suspected.

*Meal service began at approximately 6 p.m. The incubation period was calculated using 7:30 p.m. as the likely time by which all case-patients had eaten food.

Salmonella serotypes are based on the immunoreactivity of two surface structures, the O antigen and the H antigen. Serotyping was performed according to the Kauffmann-White Scheme. Additional information available at http://www.cdc.gov/ncidod/dbmd/phlisdata/salmtab/2006/salmonellaannualsummary2006.pdf.

Salmonella isolates were first screened for O antigens by using the slide agglutination method, a process that usually takes <1 minute to perform. Screening for H antigens was done by a tube agglutination test, a process that can take days to complete. While the H antigen test was pending, a fresh culture generated from a single-colony pick from the selective media underwent PFGE testing; single-colony picks from different persons' samples were run on the same PFGE gel. Because H antigen and PFGE testing ran concurrently, the PFGE results were typically available before H antigen testing was complete, and therefore, before the final serotype was known.

§PFGE pattern names were assigned by CDC's PulseNet database.

∥Connecticut is a participant in the Foodborne Diseases Active Surveillance Network (FoodNet), the principle foodborne disease component of CDC's Emerging Infections Program (EIP). FoodNet is a collaborative project between CDC, 10 EIP sites, the U.S. Department of Agriculture, and the Food and Drug Administration. As part of FoodNet, Connecticut conducts active, laboratory-based surveillance of foodborne bacterial and parasitic pathogens. Additional information is available at http://www.cdc.gov/foodnet.

¶PulseNet is a national network of laboratories in which participants submit PFGE results on certain types of bacterial isolates; the database is available on demand to participants, allowing for rapid comparison of PFGE patterns. Additional information is available at http://www.cdc.gov/pulsenet.


4 Available.