Header graphic for print
Marler Blog Providing Commentary on Food Poisoning Outbreaks & Litigation

Inspector General Fong questions FSIS sampling for E. coli O157:H7 in beef trim

fong.bmpAccording to press reports, in a testimony before the food agency subcommittee of the House Committee on Appropriations Phyllis K. Fong, Inspector General of the Office of Inspector General’s (OIG) discussed its budget request and recent audit and investigative activities. In a focus on USDA, Fong questioned beef trim sampling practices, recommending that the agency redesign its methodology.

One such assessment audit focused on Food Safety Inspection Service (FSIS) sampling of beef trim for E. coli. Currently, inspectors take 60 samples from large lots of beef trim to test. “We found, however, that this procedure does not yield a statistical precision that is reasonable for food safety,” she said. “Although 60 samples may be adequate to detect widespread contamination, more are needed when E. coli is less prevalent.” FSIS’ current sampling methodology results in detection of E. coli less than half the time when it is present in one percent of a beef trim lot, Fong explained, adding, “Accordingly, we recommended that the agency place its testing process on sounder statistical ground by redesigning its sampling methodology to account for varying levels of contamination.”

Translation – sample more and more often. Read the rest of the Testimony.

Screen shot 2011-03-03 at 10.24.42 AM.png

  • Minkpuppy

    I agree that the N60 sampling is weak and the certificates of analysis that are issued from it aren’t really worth the paper they’re printed on. However, what would be an acceptable form of sampling that could really verify the safety of beef trimmings?

  • Damn good question – I am going to find an answer.

  • I did get this response:
    Please look at the draft guidance document of 2008. The five combo N=60 is really N=12, combos have no connection to each other, FSIS needs to move to single combo N=60, this would be equivalent of N=300 for the current five combo lot testing. It is only through repeated sampling that we can ensure the safety of ground beef, once at the slaughterhouse, then secondary sampling and testing of the trim at grinders and robust sampling of ground beef (small lots, N=60) of finished products. Also realize that about 20-30% of the meat comes from other countries where the USDA equivalency program is a sad joke, the products get tested less than 5% of the testing program in the US.

  • Minkpuppy

    So in other words, the inspectors and plant personnel are going to be expected to spend all day taking MORE samples at slaughter, at cut-up and again at the grinder. The grinders have a huge headache to deal with when you consider they’re mixing trimmings from different sources into one batch. How are they expected to determine which supplier was the source of contamination? Yeah, sure, we have to gather supplier information when we take the samples but if there’s 10 different suppliers for that lot, it doesn’t help narrow it down much.

    The suppliers are adamantly against the grinders doing their own testing of the trimmings before grinding so where does that leave us? Inspection at grinding plants focus on testing the final ground product because the trimmings have already been “tested and passed” before they hit the door.

    When are they going to realize that testing is only useful as a measure of the effectiveness of the controls in place and will not eliminate E. coli? Lets focus on reducing pathogens at the source ie. the live cattle. Once that is accomplished, the in-plant interventions will be much more successful.