Using Register Receipts in a Restaurant-Associated Escherichia coli O157:H7 Outbreak Investigation — San Francisco County, California, August 2013

BACKGROUND: From August 23-26, 2013, the San Francisco Department of Public Health (SFDPH) received eight reports of laboratory-confirmed Escherichia coli (E. coli) O157 infection; a marked increase over the background incidence in San Francisco of less than one O157 case per month.  We investigated to identify a source and prevent further illness. 

METHODS: Patients matching the outbreak strain by pulse-field gel electrophoresis (PFGE) were interviewed with a hypothesis generating questionnaire, which strongly suggested the illnesses originated from Restaurant X. Probable cases included those with diarrheal illness who ate food purchased from Restaurant X.  We re-interviewed all case-patients in a restaurant case-control study using non-ill meal companions as controls. We gathered copies of register receipts and itemized sales records from August 17^th to enumerate the exposed cohort of dining groups. We calculated the number of ill and non-ill dining groups that ordered suspect menu items, assuming media attention from the restaurant closure and press release contributed to full case ascertainment. We calculated menu-item specific attack rates, relative risks and chi-square p-values at the dining-group level. Sanitarians inspected the restaurant, collected product invoices and conducted traceback of implicated products. All employees submitted stool for testing. 

RESULTS: We identified 22 cases; seven (32%) were hospitalized and four (18%) developed hemolytic-uremic syndrome.  Twenty (91%) ate at Restaurant X during August 16-17. While no menu item was significantly associated with illness in the case-control study, 74% of cases reported eating garlic noodles; all reported “family-style” dining. In the receipt-level analysis, 8% of dining groups ordered garlic noodles on August 16-17. The attack rate among dining groups eating garlic noodles on August 17 was 24% (5/21) compared to 2% (3/196) among those not ordering garlic noodles (RR=15.6, p<0.001). Environmental assessment and traceback of suspect ingredients (e.g., scallions) did not reveal a source of contamination.  All employees denied illness and were culture negative for E. coli. 

CONCLUSIONS: While we could not definitively identify a single ingredient as the source of the outbreak, risk of illness among dining groups ordering garlic noodles was 15 times the risk among groups not ordering garlic noodles. Family-style dining likely contributed to insignificant findings from the case-control study. A receipt-level analysis allowed us to analyze the exposures of patrons who dined on a specific date, without the need to interview all controls. As restaurant register receipts become increasingly electronic, this method could prove a less resource-intensive alternative to a case-control study.