National study refines proportion of illness due to microbes in food, water
Jan. 5, 2021: UF researchers led a comprehensive study estimating the proportion of food- and waterborne diseases in the US attributable to five major transmission pathways.
UF researchers led the nation’s first comprehensive study of the proportion of domestic illness caused by both water- and foodborne transmission pathways.
Approximately every 10 years, the US Centers for Disease Control and Prevention updates their estimates for the burden of food- and waterborne diseases in the US. The agency estimates the number of cases, hospitalizations and deaths based on epidemiologic data for more than 30 pathogenic bacteria, viruses and parasites.
But these data don’t always capture the many ways that these pathogens can be transmitted to humans by multiple exposure pathways; a limitation that calls for including other data to estimate the proportion of illnesses transmitted by food or water. These include data on the occurrence of pathogens in food and water, the amount of food consumed, exposure to different water sources, the ecology of the pathogens and more. There are currently no models that integrate such diverse data into one framework.
A new paper published this week in Emerging Infectious Diseases used novel methods designed for assessing risk in the face of missing data to produce current estimates for five transmission pathways of 33 major food- and water-related pathogens in the US.
University of Florida food safety expert Arie Havelaar, a microbiologist in UF's Institute of Food and Agricultural Sciences who specializes in microbial risk assessment and food safety, is the study’s senior author. Havelaar is also a faculty member in UF’s Emerging Pathogens Institute and the Food Systems Institute. His then-doctoral researcher, Elizabeth Beshearse, is the study’s first author; she is currently an Epidemic Intelligence Service officer with the CDC.
“Ideally, our data will help inform a public health response,” Beshearse says. “They were created to inform food and waterborne disease burden estimates which can be used to understand the actual amount of illness that is occurring in the US, and to create targeted public health control and prevention strategies.”
The research team identified five major transmission pathways: foodborne, waterborne, person-to-person, animal contact, and environmental. Most of the pathogens included were bacteria, but estimates were also produced for relevant viruses and protozoa.
The CDC, along with other federal partners, previously estimated which food products are most frequently involved in foodborne transmission. However, researchers lack data to estimate the role of infected food handlers, presumed person-to-person associated transmission, presumed animal contact associated environmental transmission, and several modes of transmission via water including recreational water exposure, drinking water, and nondrinking or nonrecreational water exposures. The researchers included these subpathways in the study, alongside the major transmission pathways.
“We did not include any pathogens that were known based on their ecology and microbiology, to be transmitted through just a single pathway,” Beshearse says. “And we worked with the CDC on which pathogens might be priorities.”
Structured expert judgment
Beshearse and her research colleagues used a method, Cooke’s classical model, that relies on the ability of experts in the field to integrate diverse lines of evidence into systematic estimates while acknowledging uncertainty due to incomplete knowledge.
“It is the most mathematically rigorous way to create estimates when data is missing or can’t be directly observed,” Beshearse says.
The same method was recently used by the World Health Organization to estimate the global burden of foodborne disease. Havelaar led the WHO effort and built on this experience for the current study. Roger Cooke, the researcher who designed and validated the risk assessment method used by the team also contributed to the study. Additional coauthors are experienced in attribution studies and in Cooke’s classical model. The study also involved CDC subject matter experts who facilitated providing information to the experts to inform their estimates and contributed to the study design.
Structured expert judgment relies on mining the knowledge of experts while providing extensive training on how to create estimates under uncertainty and document their level of uncertainty around each estimate. Each pathogen was assessed by a panel of experts, and their estimates for the transmission pathways of targeted pathogens were then combined into a single distribution estimate per infectious agent.
“The human mind is vastly superior to computer algorithms in combining disparate types of data into consistent estimates,” Havelaar says.
The study estimated the proportion of Salmonella from foodborne sources to be lower than previously estimated, just 66%, compared to prior estimates ranging as high as 95%. Similarly, the new study also lowered estimates for Campylobacter acquired from foodborne sources from 80% to 57%.
But the study also significantly raised the estimates for the proportion of some pathogens acquired via food, including for astrovirus (15%), sapovirus (13%), and rotavirus (5%) which were all previously estimated at less than 1%.
The researchers also increased the estimate for the proportion of hepatitis A virus, acquired from foodborne pathways, from 7% to 42%. The authors note that rising person-to-person hepatitis A transmission may further alter this estimate, however.
When estimating the proportion of specific pathogens from waterborne transmission pathways, the researchers note that the vast majority is bacterial in nature with viruses comprising only about 6-8%. The new study estimated a lower proportion of diarrheagenic E. coli in the US at 9%, compared to 15% in Canada; and for the enterotoxin-producing type of E. coli, the new study pegged the US estimate at 9% (compared with 42% in a study combining the US, Canada, and Cuba).
The new study also raised the estimate of Shiga-toxin producing E. coli from animal contact in the US forSTEC O157 to 12%, and for STEC non-O157 to 21%, from 6% and 8%, respectively.
Their new, lower estimate of Salmonella spp., for example, was more in line with estimates provided by other countries, which lends validation to the method.
The results from this study were used to inform a second paper published in the same issue of Emerging Infectious Diseases, which estimated the burden and direct healthcare costs of infectious waterborne diseases in the US. Both Havelaar and Beshearse are coauthors of this paper as well.
The researchers hope the results of their study will be used as a roadmap for agencies to target research and surveillance with the aim of preventing infections transmitted by food and water pathways.
Written by DeLene Beeland