In general, we define a cluster as two or more 2 cases in different households with isolates of the same pulsed-field gel electrophoresis (PFGE) subtype and with specimen collection dates within 2 weeks (however, in practice that 2-week period is a bit arbitrary in that sometimes we’ll compare exposures for cases that are more than 2 weeks apart).
All PFGE clusters are investigated. At a minimum this investigation entails comparing the routine interviews of the initial cases. A dynamic cluster investigation model is used (Figure 1). The model is described in Chapter 5 of the CIFOR Guidelines for Foodborne Outbreak Response. In brief, specific questions about suspicious exposures identified during initial case interviews are added to the standard interview for subsequent cases. Similarly, initial cluster cases may be reinterviewed to ensure uniform ascertainment of suspicious exposures identified in interviews of later cases. This iterative approach is used to identify exposures for further evaluation with formal hypothesis testing, product sampling, or product tracing.
The 2007 multistate outbreak of Salmonella 4,5,12:i:- infections associated with pot pies provides a great example of the utility of the dynamic cluster investigation model (Figure 2). The first two Minnesota cases both reported eating various types of microwaveable entrees, but nothing specific in common. The third Minnesota cluster case also reported frequently eating microwaveable entrees, including eating a Banquet pot pie every day for lunch. This prompted the addition of a specific question about Banquet pot pie consumption to the routine questionnaire used for the fourth case, and the previous two cases were re-interviewed about this specific exposure. This quickly led to the documentation that all four cases had eaten Banquet pot pies in their exposure period. When other states then asked their cases specifically about this exposure, such a high percentage of cases reported it that it quickly became apparent that the pot pies must be the outbreak vehicle.
Another, more general, example of this model relates to restaurant meals reported by cases. Once the first few cases in a cluster have been interviewed, a complete list of all restaurants reported by any of the cases is used to reinterview all of the initial cases. Cases often cannot remember all restaurant meals in their exposure off the top of their head, but when prompted with specific restaurant names can easily remember the meal. This practice has frequently led to the detection of restaurant outbreaks of salmonellosis.