In January, the Food Safety and Inspection Service published a proposed rule, “Modernization of Poultry Slaughter Inspection,” with important changes in the rules of processing, including the intention of FSIS to drop the requirement that plants test chilled carcasses for E. coli. Individual establishments will be allowed to use “other, more relevant” but unspecified indicators of process control, with individual HACCP plans identifying what bacterial monitoring is to be done. Dropping the E. coli testing requirement is a sharp departure from the dogma of poultry HACCP as mandated by FSIS in the last 15 years.

The 1996 HACCP Final Rule said that experts agreed that E. coli was the best indicator of fecal contamination, the source of pathogens such as Salmonella. “There is a strong association of E. coli with the presence of enteric pathogens,” with “wide acceptance in the international scientific community of its use as an indicator of the potential presence of enteric pathogens.” The Final Rule said that E. coli was “the most effective measure of process control for enteric pathogens,” and thus plants were required to test for generic E. coli, with specified limits in a moving window of test results.

E. coli counts not seen as reliable indicator of Salmonella  

Even before publication of the HACCP Final Rule in 1996, however, there were warnings that E. coli counts were not a reliable indicator of Salmonella, despite the endorsement by FSIS. At the scientific meeting sponsored by FSIS in Philadelphia in 1995, poultry microbiologist Dr. Amy Waldroup said that her work had not found any correlation between different groups of microorganisms. After reviewing FSIS’s HACCP proposal before its final publication, USDA’s Office of Risk Assessment and Cost Benefit Analysis commented, “There is an insufficient scientific basis to rely on E. coli for process control verification to ensure Salmonella reduction.”

The office suggested that FSIS “conduct research to test the statistical limits of the hypothesis that E. coli process control verification will ensure pathogen reduction,” and that samples intended for Salmonella testing be assayed by FSIS personnel for E. coli as well. In 2004, the National Advisory Committee on Microbiological Criteria for Foods reported, “There are no data that support the use of index organisms for Salmonella on broilers.” The committee also recommended generation of data to relate indicators to pathogens and show that reductions in indicators lead to reductions in pathogens.

FSIS has shown great interest in whether E. coli can indicate the presence of pathogens such as Salmonella. In the "Risk-Based Inspection” documents published in 2008, there was extensive discussion of a joint research project conducted by FSIS and Agricultural Research Service to study the relationship between E. coli counts and the presence of Salmonella. The peer-reviewed publication that resulted from that study reported, however, that sorting groups of carcasses by E. coli counts did not produce statistically significant differences in Salmonella prevalence. The paper recommended additional surveys with more data than the 800 carcass rinses available in that study. Similarly, one of the goals of the 2012 Nationwide Raw Chicken Parts Microbiological Baseline Data Collection Program is to “compare prevalence and counts between pathogenic organisms and indicator organisms to determine relationships and associations.”


FSIS silence about data on E. coli-Salmonella relationship  

In fact, the further studies urged by the Office of Risk Assessment and Cost Benefit Analysis, the National Advisory Committee on Microbiological Criteria for Foods and FSIS have accumulated gradually since the FSIS version of HACCP has been the rule in poultry processing. Table 1 shows Baseline and other studies conducted by FSIS during the HACCP era, with a total of 14,097 chicken carcasses, 5,501 turkey carcasses, 102 goose carcasses, 285 ground chicken samples, and 296 ground turkey samples that have been collected by FSIS using various sampling methods for analysis of E. coli counts and Salmonella prevalence. In addition, Most Probable Number assays for Salmonella were carried out for almost 8,000 chicken carcasses and 4,000 turkey carcasses, allowing a more powerful analysis for those samples. The published Baseline studies and the HACCP/HIMP study are absolutely silent on the relationship between E. coli and Salmonella in this mountain of data.

FSIS should release the data to answer questions  

What was the likelihood of Salmonella presence when the E. coli count was above the 80th or 98th percentiles (control levels m and M in the E. coli criteria)? A single observation above 98th percentile was said to indicate that the plant was “out of control.” What was the correlation between E. coli counts and Salmonella MPN in samples where numbers can be compared directly? Was there seasonal variation in E. coli results to match the seasonal variation in Salmonella?

The Data Collection and Reports section of the FSIS website says, “To ensure that FSIS maintains a science-based and data-driven approach to food safety, the Agency conducts a robust analysis of available data.” FSIS should release the E. coli/Salmonella data immediately so that establishments can know whether it is worthwhile to continue using E. coli as an indicator of process control under the modernized rules.