BACKGROUND: An estimated 1.2 million salmonellosis cases occur in the United States annually. Salmonella serotype Enteritidis (SE) accounts for nearly 20% of salmonellosis cases reported to FoodNet. Identifying exposures associated with illness is essential for developing effective regulations and interventions. Exposure data can be obtained by case-control studies, but they are resource-intensive and not conducted regularly. To better understand exposures, FoodNet began routine collection of standardized exposure data for salmonellosis. We compared exposures reported by Salmonella serotype Enteritidis patients to those of patients infected with other Salmonellaserotypes (non-SE) and to healthy controls from the 2006–2007 FoodNet Population Survey (PS) to assess the utility of different comparison groups.
METHODS: Cases and non-SE cases were defined as sporadic, domestically-acquired, culture-confirmed Salmonellainfections with documented serotype in patients over 1 year of age reported to FoodNet during 2014–2015. We excluded PS controls <1 years of age and those who reported diarrhea in the 7 days before interview. We compared 35 food, water, and animal exposures common to PS and FoodNet and calculated odds ratios (aOR), adjusted for site and season. Statistical significance was defined as α<0.05.
RESULTS: Our analysis included 1,481 SE cases, 5,511 non-SE cases, and 16,377 PS controls. Increased risk of SE infection was significantly associated only with consumption of chicken outside the home (45% vs. 42%; aOR=1.2), lettuce (58% vs. 50%; aOR=1.2), and spinach (22% vs. 18%; aOR=1.2) compared with non-SE cases. Compared with PS controls, increased risk of SE infection was significantly associated only with consumption of chicken (80% vs. 67%; aOR=2.0), chicken outside the home (45% vs. 26%; aOR=2.3), fish (30% vs. 25%; aOR=1.4), and contact with live poultry (5% vs. 3%; aOR=1.4). Protective exposures were also identified, and differed among the two analyses.
CONCLUSIONS: Risky exposures associated with SE infection were similar to those identified in case-control studies and outbreaks. More risky exposures were identified using PS controls, indicating that the exposure profile of non-SE cases differs from that of population controls from 10 years ago. These findings support the utility of routinely collected standardized exposure data. Monitoring these exposures over time could provide an on-going, readily available alternative to case-control studies for a limited set of exposures.