Campylobacter illnesses were mostly attributed to dairy and Listeria illnesses were attributed to fruit but these areas had data limitations and estimation uncertainties.
Salmonella illnesses were broadly attributed across food commodities.
The US Food and Drug Administration (FDA), Centers for Disease Control and Prevention (CDC) and USDA’s Food Safety and Inspection Service (USDA-FSIS) developed a method for analyzing outbreak data to determine which foods are responsible for illness related to four foodborne bacteria.
The report was produced by the Interagency Food Safety Analytics Collaboration (IFSAC) - a partnership of the three agencies.
Illnesses from outbreaks caused by Salmonella, E. coli O157, Listeria monocytogenes (Lm), and Campylobacter were attributed to 17 food categories.
Pathogen and category findings
Salmonella illnesses were found across multiple food categories and Campylobacter infections also spanned a broad array but unpasteurized milk or cheese made from unpasteurized milk made up 60%.
E. coli O157 and Lm illnesses were attributed to fewer food categories, and two accounted for the largest point estimates: beef and vegetable row crops for E. coli O157 and dairy and fruits for Lm.
For Lm, the limited number of outbreaks dictates caution in interpreting the percentages for the two food categories – taking into account the 2011 cantaloupe outbreak.
IFSAC analyzed data from nearly 1,000 outbreaks from 1998 to 2012 to assess which categories of foods were most responsible for making people sick.
The dataset, after exclusions, included 597 outbreaks caused by Salmonella, 170 by E. coli O157, 161 by Campylobacter and 24 by Lm.
They determined four variables: pathogen, food category, food preparation setting (home, restaurant, schools), and whether exposures to the implicated food was in multiple states or not.
Seventy four percent of Campylobacter illnesses were attributed to dairy (66%) and chicken (8%), 82% of E. coli O157 illnesses were because of beef (46%) and vegetable row crops (36%), and 81% of Lm illnesses were due to fruits (50%) and dairy (31%).
Salmonella was more broadly attributed, as 77% of model‐estimated illnesses were from seeded vegetables (18%), eggs (12%), fruits (12%), chicken (10%), sprouts (8%), beef (9%), and pork (8%).
Method for data analysis
IFSAC focuses on foodborne illness source attribution, which is estimating the percentage of illnesses associated with specific foods.
It hopes determining the food that causes illnesses will improve safety and help identify opportunities to influence policy.
The team used regression modeling techniques to normalize distribution of outbreak size and minimize bias in estimated attribution percentages associated with large outbreaks.
Researchers applied an annual discounting scheme to decrease the weight of older data so information older than five years would contribute no more than 50% of total data weight, and data older than 10 years would contribute less than 8%.
They also calculated 90% credibility intervals using a method to adjust for small outbreak numbers.
“We summed the estimated number of illnesses by pathogen and food category and then divided by pathogen totals to obtain attribution percentages for each replicate; we used the 5% and 95% distribution values to define the estimated lower‐and upper‐credibility limits of the interval.”