Introduction. On June 28, 2016, the Chicago Department of Public Health (CDPH) received five reports of Shiga Toxin-producing Escherichia Coli (STEC)1 through routine surveillance. By June 29, routine interviews conducted by the CDPH Communicable Disease (CD) Program revealed that three of the five cases reported consuming food items from Carbon (Restaurant A) within 2-3 days before illness onset. That evening, three separate hospitals reported an increase in the number of patients that presented to the ED with complaints of diarrhea and had preliminary positive STEC diagnostic laboratory tests. By July 1, seven cases reported eating at Restaurant A prior to their illness onset. Carbon Final Outbreak Summary – City of Chicago Dept. of Public Health

Restaurant. Restaurant A has two Chicago locations, one on the south side and another on the west side of the city. The restaurant is open 7 days a week and serves Mexican-style foods. Both locations serve the same menu and use the same food suppliers. The majority of food preparation is performed out of the south side location; most food for the west side location is transported after preparation at the south side kitchen. Catering is also available. Overall, approximately 40% of food orders are placed by phone or through online ordering websites (i.e. GrubHub, Eat24, etc.) for delivery or pickup. Catering and other delivery orders are prepared in the same kitchen and by the same staff as dine-in orders at both locations. Staff members at each location reported regularly consuming restaurant food.

Epidemiological investigation. Case finding was conducted through public messaging and disease surveillance. On June 30, 2016, CDPH issued a health alert to all Chicago hospitals to notify them of the outbreak, to request prompt reporting of STEC cases, and to discourage use of antibiotics and encourage aggressive hydration if suspecting a diagnosis of STEC. Concurrently, the Illinois Department of Public Health (IDPH) issued an alert via the Food borne Outbreak Network to state health departments to notify them of any STEC cases with travel to Chicago and mention of Restaurant A.

A standard questionnaire was created to collect information about signs and symptoms of illness, food consumption and other potential exposures occurring in the seven days prior to the case’s onset of illness, and meal companions. A case-control study was conducted to determine risk factors for infection with STEC. Case definitions were in accordance with the Centers for Disease Control and Prevention (CDC) and Council of State and Territorial Epidemiologists standards2. A confirmed case was defined as isolation of E. coli O157:H7 (STEC) from a clinical specimen in a person with illness onset between June 3-July 23, 2016, with either reported exposure to Restaurant A or a pulsed-field gel electrophoresis (PFGE) pattern indistinguishable from one of 14 patterns associated with the outbreak. Confirmed cases with reported Restaurant A exposure and onset dates that preceded others within their household were considered confirmed primary cases. A probable case was defined as a person with clinically compatible illness (bloody diarrhea or ~3 days of diarrhea with ~3 stools in a 24-hour period) in the absence of laboratory confirmation, and exposure to Restaurant A or shared household with a primary case. Secondary cases were defined as household contacts of primary confirmed or probable cases, with onset of diarrhea one to eight days after the primary case’s symptom onset date. Case-control analysis was limited to primary confirmed cases and well controls. To identify controls, CD Program staff asked confirmed cases about their meal companions and obtained a list of individuals who placed orders through the online delivery service GrubHub. Controls were frequency matched 4:1 to cases by meal date (June 17th-June 30th) and restaurant location.

Contingency tables were arranged to evaluate the bivariate relationships between case status and individual food items, and odds ratios (OR) with 95% confidence intervals {95% Cl) were estimated for each. Chi-Square tests were performed to identify statistically significant associations, except when expected cell counts were less than or equal to 5, in which case Fisher’s Exact test was used. P-values <0.05 were considered statistically significant. The independent effects of variables found to be significantly associated with disease in the bivariate analyses were further evaluated using multivariable logistic regression, adjusted for age and gender. All statistical analyses were carried out with SAS version 9.3 (SAS Institute, Cary, NC).

Environmental investigation. On July 1, 2016, the Food Protection Division (FPD) conducted an environmental inspection of Restaurant A and collected the following: food samples, initial information about restaurant employees and food preparation, and copies of invoices for food items. Food items collected included steak, chicken, cilantro, elote (corn), elote mix, cheese, sour cream, grilled corn & pineapple salsa, salsa fresca, tequila lime sauce, red and green salsas. CD Program staff performed in depth interviews of the owners of the restaurant and employees. Because employees at both locations often functioned in multiple roles, all on-site restaurant employees were considered food handlers for the purposes of this outbreak investigation. Food handlers were asked to submit stool specimens to screen for STEC.

Laboratory investigation. Clinical culture or polymerase chain reaction tests were performed by hospital and commercial laboratories and results were reported to the CD program. Specimens from cases, food handlers, and food were sent to the Illinois Department of Public Health Division of Laboratories for culture, and for serotyping and PFGE analysis of STEC isolates. PFGE patterns were uploaded to the national Pulse Net database and compared by the Centers of Disease Control (CDC). Sixteen isolates selected to represent all outbreak-associated PFGE patterns and a variety of source patients (primary and secondary cases as well as food handlers), restaurant locations, and meal dates were sent to CDC for characterization by multiple locus variable number of tandem repeats analysis (MLVA).

Epidemiologic findings. Sixty-nine confirmed and 37 probable cases were identified as part of this outbreak. Among the confirmed, 55 met the primary case definition, four were secondary cases, and ten of the confirmed cases matched the outbreak PFGE pattern but their association with the restaurant was not identified (five denied eating at the restaurant and five were unable to locate). One additional case, identified after the restaurant closure and reopening, was unable to be classified due to multiple

Restaurant A meal dates and a PFGE pattern that was similar but not identical to other outbreak patterns. Illness onset dates of the 55 confirmed primary cases ranged from June 19-July 3.

Median age was 29 years (range, 3 to 69 years); 29 (53%) of the cases were female. Median incubation period was 3 days (range 12 hours-5 days). Twenty-one primary and one secondary case were hospitalized. No cases developed hemolytic uremic syndrome, and none died. Among the 55 confirmed primary cases, 50 (91%) ate at the south side location (meal date range 6/17 to 6/30) and 5 (9%) ate at the west side location (meal date range 6/19 to 6/26).

Multiple food items were associated with illness on bivariate analysis (Table 1) including consumption of cilantro (odds ratio [OR] 3.5, 95% Cl: 1.5-8.1), salsa fresca (OR 3.1, 95% Cl: 1.6-6.1), chicken taco (OR 3.1, 95% Cl: 1.6-6.0), and lettuce (OR 2.01, 95% Cl: 1.1-3.8). Multivariable analysis using logistic regression (Table 2) revealed that consumption of cilantro (adjusted OR [aOR] 4.64, 95% Cl: 1.87-12.011.6), salsa fresca (aOR 2.85, 95% Cl: 1.31-6.05.4), and lettuce (aOR 2.57, 95% Cl: 1.23-5.26) remained independently associated with illness after adjusting for age and gender. The observed epidemiologic association with chicken tacos may reflect collinearity between chicken tacos and cilantro, meaning that an association was identified because the chicken tacos are prepared and served with raw cilantro. All cases who reported eating a chicken taco also reported eating cilantro. Other chicken-containing items (e.g., chicken burritos, chicken salad bowls) were not associated with illness. Because salsa fresca was known to contain raw cilantro, an additional multivariable logistic regression analysis was performed including a combined variable indicating consumption of either cilantro or salsa fresca. In this model, consumption of cilantro or salsa fresca was associated with an adjusted odds ratio of 6.9 [Cl: 2.0-24.0]

Lettuce was associated with illness in both multivariable models but was consumed by only 44% of cases. In comparison, cilantro was consumed by 87% of cases, and either cilantro or salsa fresca were consumed by 95% of cases.

Environmental findings and food handler interviews. Meats, salsas, and marinades were fully or partially prepared at the south side location and transported daily to the west side location. Most fresh produce items, including cilantro and lettuce, were received by each location in separate deliveries and chopped and prepared on-site. Several critical violations were identified during the sanitarians’ inspection of Restaurant A on July 1, including improper temperatures for several food items (i.e. red & green salsas, tequila lime sauce, raw fish, guacamole, and cheese), and improper hand hygiene practices among food handlers. Because of concern for a potential ongoing public health threat associated with food served by Restaurant A, CDPH recommended that the restaurant voluntarily cease operations and withdraw from a large outdoor food festival until more information about the source of the contamination was known. The owner agreed, and Restaurant A voluntarily closed both locations. CD staff subsequently interviewed and tested forty food handlers from both locations. According to the restaurant owner, there was no cross-over of food handlers at the two locations. Among the forty food handlers interviewed none reported any history of gastrointestinal illness in the two weeks preceding or during the outbreak period, though absenteeism was reported for one. Nearly all food handlers had stool tests performed within one week after the restaurant closure.

Laboratory findings. Specimens from 69 cases and 16/40 (40%) food handlers yielded STEC isolates. From primary case isolates, 10 PFGE patterns were identified. An additional four similar patterns were identified among food handler isolates. The 16 isolates analyzed by MLVA displayed four unique MLVA patterns. One predominant MLVA pattern was shared by 10 isolates. Food handler and case isolates displayed a variety of MLVA patterns, with some food handlers sharing MVLA patterns indistinguishable from restaurant patrons despite differing PFGE patterns. There were no distinct pattern groupings according to restaurant location. None of the 12 food items cultured were positive for STEC.

Food product traceback. In collaboration with CDPH, FPO and IDPH department of Food, Drugs and Dairy, invoices collected from the restaurant for the outbreak period meal dates were reviewed. Cilantro was purchased from a distributor serving multiple other restaurants throughout Illinois. The distributor repackaged cilantro from multiple sources, including suppliers in Mexico and Illinois. Of the five laboratory-confirmed cases who denied Restaurant A exposure, none reported cilantro consumption, although cooperation with re-interview was limited. In the absence of confirmed cases reporting consumption of implicated food items from another restaurant, it was not possible to perform further traceback to assess for a common source of contamination. No other restaurants serviced by the distributor were linked to the outbreak.

Re-inspection and reopening. FPO performed re-inspections at both locations, during which instruction and guidance were provided on hand hygiene and it was ensured that food preparation and storage areas were adequately sanitized. Only food workers who had two consecutive negative tests for STEC were permitted to work at the reopened locations, which delayed reopening of the south side location. After passing FPO re-inspections, the south and west side restaurants re-opened on July 9 and 29, respectively.

An additional case of STEC was subsequently identified with a meal date of July 11, 2016 at the reopened west side location, prompting imposed restriction of all food handlers who had been involved with preparation of the case’s meal. All of these food handlers and the case’s meal companions were tested for STEC, with negative results. The case had also eaten food from the restaurant on June 21 with a household member, prior to the restaurant’s closing; neither reported symptoms of illness at that time. PFGE analysis revealed an additional pattern not previously identified in this outbreak, but which appeared related to other outbreak patterns. Because of the multiple meal dates, negative test results of all meal companions and involved food handlers, and absence of additional reported restaurant-associated cases after the re-opening, we could not definitively determine if the case was primary, secondary or unrelated. Food handlers were permitted to return to work after repeated negative test results. No additional complaints were received in association with the restaurant in the two months following this case’s illness onset

Conclusion. This was a large restaurant-associated outbreak of Shiga toxin-producing E. coli O157:H7 infections. Closure of the restaurant during the early stage of the investigation prevented additional cases of illness from occurring. Cilantro was the most likely food-vehicle causing this outbreak, based on the strong statistical association of raw cilantro consumption with illness, and the high percentage of cases explained by cilantro consumption. The large number of PFGE patterns associated with the outbreak was suggestive of a heavily contaminated food item rather than introduction from a point source such as an ill food worker at the restaurant. However, STEC was not isolated from cilantro or cilantro-containing food items collected from the restaurant or the restaurant’s distributor. Inability to isolate STEC from food samples may have been hindered by imperfect sensitivity of testing, imperfect representativeness of food samples, or turnover of produce items through the distribution chain leading to items no longer being contaminated at the time of collection. Additionally, cross-contamination during food preparation and transmission by food handlers who were found to have STEC infection likely contributed to the outbreak.