The Food Safety and Inspection Service has stated an intention to “drive down human illness rates” by reducing Salmonella prevalence in meat and poultry. The link between human illness and Salmonella in meat and poultry is not so clear, however. Food Safety and Inspection Service activities can reduce human illnesses only by the number of cases being generated by meat and poultry, and reductions in prevalence in raw meat and poultry will have no effect on human illness unless the method of measuring Salmonella prevalence is related to the risk of human illness. In fact, recent Salmonella statistics indicate that many countries that have reduced Salmonella in food animals have much higher levels of human salmonellosis than does the U.S.

A 2009 report by the National Research Council said that we don’t know how much salmonellosis is caused by poultry or by non-poultry sources. Studies in the U.S. have indicated that about 95 percent of human salmonellosis is foodborne, but estimates in other developed countries are as low as 55 percent. There are human, pet and environmental sources of salmonellosis in addition to foods. The proportion of cases caused by Food Safety and Inspection Service-regulated products is also unknown. In recent years there have been Salmonella outbreaks caused by tomatoes, peppers, peanut butter, puffed vegetable snacks, dog treats, reptiles and amphibians kept as pets, and frozen mice sold for feeding pet snakes. For these reasons, we don’t know how much human illness could be prevented by Food Safety and Inspection Service regulatory policy.

Risk factors unknown  

The components of risk within poultry are also unknown – the proportion of illnesses caused by undercooking or by cross-contamination, or how much risk is caused by the presence of Salmonella (+/-) or whether some risk is influenced by the numbers of Salmonella cells that are present on raw poultry. There are few studies that directly relate HACCP measures of Salmonella presence in processing plants to the risk of consumers getting salmonellosis from those products, but some modeling work has suggested that samples with higher numbers of Salmonella cells carry more risk of causing illness. Testing for Salmonella could be more rigorous or less, but which method would be best correlated with risk is unknown. The National Research Council concluded that it is difficult to find control strategies or to measure the effect of controls when so many epidemiological factors are unknown.

Limited data on success in control programs  

Recent scientific reviews have reported that only limited data exists for evaluating the success of national Salmonella control programs. In the almost 15 years since the introduction of the Food Safety and Inspection Service version of HACCP, there has been about a 50 percent decrease in the number of Salmonella-positive HACCP verification samples taken at processing plants. The National Antimicrobial Resistance Monitoring System samples retail chicken breasts and other meat samples for the presence of Salmonella, but monitoring started well after the introduction of Food Safety and Inspection Service HACCP and less data is available from the National Antimicrobial Resistance Monitoring System program. There has been no obvious change, however, in the number of human cases per 100,000 reported by the 50 state health departments (CDC’s Public Health Laboratory Information System) or by the more intensive surveillance conducted in all or part of 10 states (FoodNet) since before the beginning of HACCP. There is no significant correlation between Salmonella prevalence in yearly chicken HACCP samples and the National Antimicrobial Resistance Monitoring System retail chicken samples, and no correlation between the annual Public Health Laboratory Information System and FoodNet human cases, or between either of the chicken Salmonella prevalence figures and either of the human case estimates.


FoodNet also estimates the relative risk of salmonellosis based on Poisson regression analysis, but due to changes in the surveillance areas, some years with a significant calculated risk reduction compared to 1996-98 have more salmonellosis case per 100,000 in the raw numbers. If the success of Salmonella control programs is seen only in complex statistical analysis rather than in obvious reductions in human illness, then there has not been much improvement.

Given this track record, it seems unlikely that human salmonellosis can be expected to respond to the next reduction in HACCP Salmonella results from meat and poultry. Because of Salmonella sources that are not foodborne, other food sources that are not regulated by Food Safety and Inspection Service, and all of the variation and uncertainty in detection and monitoring systems, it would probably require a large and significant reduction in meat and poultry and in human cases for any changes to be detected and reliably attributed to Food Safety and Inspection Service activities.

International comparisons  

The accompanying table shows both Salmonella prevalence in chicken carcasses after processing and human salmonellosis cases in selected countries for 2008. The Scandinavian countries and New Zealand have implemented successful Salmonella control programs in poultry, but do not have lower illness rates in their populations compared to the U.S. or the Netherlands. Up to 80 percent of human cases in Sweden are attributed to international travel, but the Dutch also travel and apparently have a lower illness rate.

If rational prevention or control strategies can’t be designed based on biology and epidemiology, we are still relying on the “have to do something” strategy. Fifteen years into the Food Safety and Inspection Service version of HACCP, there is still no clear way to evaluate the ultimate success of Food Safety and Inspection Service control strategies for Salmonella in meat and poultry: their effect on the risk of human illness.