No food processor wants pathogens contaminating the product in the plant for at least one obvious reason: the product on sale at retail might carry the risk of foodborne illness. Beyond that, it is unclear what are the chances that a consumer will become ill.

Helen Jensen and colleagues are seeking to connect the dots to determine how changes in the pork production process affect the predicted number of people who become ill with salmonellosis because of pork and how food safety interventions affect risk as well as industry costs. By learning that information, the meat industry would be able to figure the costs of intervening at points in the production process that would be the most effective in making the product safer.

Professor Helen Jensen

"We think this model will be helpful for the industry whether it's the packing plant or the Pork Board," said Professor Jensen, an Iowa State University (ISU) professor of economics. She has collaborated with Scott Hurd, an associate professor of veterinary diagnostic and production animal medicine who is spending most of 2008 on leave from ISU as deputy undersecretary for food safety at the US Department of Agriculture. They are pursuing the research with support from the Food Safety Consortium.

Professor Jensen said in discussing the subject with the pork industry, the key question has been whether to invest resources in the farm or in the slaughterhouse. "Then if we do invest resources in the slaughterhouse or on the farm, what's the gain we're going to get? Gain in this situation is measured by reduction in the number of human cases of illness," she said.

It is unusual to carry the research all the way to measuring the number of human illnesses. Professor Jensen noted that US research has not explored that angle to the extent that Europeans have done. Data are limited on salmonella in the United States for anyone seeking to find out about its sero-prevalence - the number of persons who test positive for a disease stemming from the bacterium.

"We're mostly depending on one study done here at Iowa State which has found that the really large swine farms have a somewhat higher sero-prevalence than everybody else," Professor Jensen said. "Now we're converting our data by size categories so that we can put them into our model and say, 'If that's true for the whole the whole US and we apply those data across all herds in the US, how is that going to work out as far as the number of contaminated carcasses?'"

The epidemiological model is being integrated with a multi-market economic model that evaluates producer and processor behaviour and the economic effect on the pork industry.

One significant question to follow is what interventions can be made during the production process that will have an effect on curbing human illness. It is not an easy matter to resolve.

"There's a tendency to think that there's a nice linear relationship - every reduction of salmonella on the pig farm is going to reduce human health risk by the same portion. And that's just not true, for a number of reasons," Professor Jensen said.

Professor Hurd analysed data from Denmark, which has kept more extensive records of the cause-and-effect situation than has the United States. In Denmark, the data show that pathogen reduction on the farm had little impact on human health risk. But pathogen reduction in the slaughterhouse did make a difference.

"The Danish study showed the most cost-effective place to invest resources was in the slaughterhouse because you get a better return on your investment when return is measured in terms of human health cases," Professor Jensen explained. "How many human cases of salmonella can we reduce per dollar invested?"