Researchers have used an outbreak a decade ago to look at how decisions on excluding sick and healthy people from investigations could change the conclusions reached.

Criteria to include or exclude cases can help increase the efficiency of epidemiological analyses and traceback, but they can also affect the investigator’s ability to implicate a suspected food vehicle.

A 2010 outbreak of Salmonella Hvittingfoss associated with Subway in Illinois was the case study.

Inaccurate or ambiguous findings in outbreak investigations can have significant financial implications for industry, according to the study published in the journal Epidemiology and Infection.

Exclusion due to multiple visits changed conclusions
Researchers examined how the exclusion of cases and people who ate with them but did not get sick with multiple eating dates impacted the findings from epidemiological analyses.

In the original investigation, a case-control study of restaurant-associated cases and non-sick meal companions was done at the ingredient level to identify a suspected food vehicle; however, 21 percent of cases and 22 percent of well meal companions were excluded for eating at the chain of quick-service restaurants more than once during the outbreak.

In May 2010, the Illinois Department of Public Health (IDPH) investigated an outbreak of Salmonella Hvittingfoss associated with multiple Subway restaurants in many counties between April and June. There were 97 cases and 12 food handlers with confirmed infections, with illness onset ranging from April 25 to June 30.

In early June, Subway’s in the area wher the outbreak occurred were told to pull four suspected produce items: onions, lettuce, tomatoes and green peppers based on an early review of items most frequently consumed by cases.

Spotlight on green peppers
Early in the investigation, green peppers were suspected based on interviews with cases and well meal companions, as well as product traceback data. However, cases and well people who ate at Subway multiple times during the outbreak were excluded because it could not be determined which meal date led to exposure. In the final analysis, green peppers were not statistically associated with illness, while lettuce, olives, and tomatoes were. All excluded cases who consumed green peppers became ill.

“While the ingredient-specific analysis cannot clearly implicate a single food vehicle, including those with multiple meal dates, showed that green peppers were associated with illness as was found early in the initial outbreak investigation. This finding could have helped inform the outbreak investigation in real-time and, in conjunction with traceback and/or laboratory evidence, informed a less ambiguous conclusion,” said researchers.

Of 85 cases and 32 well meal companions, IDPH excluded 18 cases and seven well companions with multiple eating dates with interview information. The ultimate IDPH case-control analysis included 67 cases and 25 well meal companions. In some cases only positive food exposures were recorded.

Excluding people who had eaten at Subway more than once during the outbreak three foods were statistically associated with illness – lettuce, tomatoes, and olives. Including those with multiple meals, green peppers were also significantly associated with illness.

Removal of the four items (lettuce, tomatoes, green peppers, and onions) that had been eaten by at least 36 percent of cases appeared to stop illness. The original outbreak investigation did not implicate a single food vehicle but listed lettuce, tomatoes or olives as possibilities because they were statistically associated with illness.

Produce including tomatoes and lettuce were from a central Illinois distribution center that served multiple restaurant customers including Subway.

Green peppers were only delivered to Subway. Confirmed cases reported eating at 49 of these restaurants in 28 Illinois counties. Combining traceback data with epidemiology strongly suggested green peppers were the likely vehicle but they were not implicated when those with multiple eating dates were excluded.

“This study showed that excluding patrons with multiple meal dates from case-companion analyses changed the conclusions that could be drawn from the results of the original investigation. Using all available information to construct a coherent narrative of what happened and why is a critical component of an outbreak investigation,” said researchers.