Another Grower Prevents an Outbreak

AnotherGrower Prevents an Outbreak


International Specialty Supply

November 12, 2001

Wereceived a call from a grower on the US East Coast who had a confirmedpositive for Salmonella.  His test was done at 48 hours and was a compositeof water taken from three Rota-Techs.  Alfalfa, broccoli and cabbagesprouts were being produced in those drums.  He just used up hisnext to last bag of alfalfa.  He had been using the broccoli for acouple months and the cabbage for a year.  He bought his alfalfa seed from ISS,and we had sampled every bag and tested it for pathogens using thesprout production method.  He bought his cabbage and broccoli seed fromanother supplier who said they "tested it" but didn't describe how andprovided no documentation.  The grower sanitizes his seed using the20,000-ppm calcium hypochlorite.  He was using the hold and release programrecommended by the FDA, so no product had left his building. 

We maynever know if the salmonella came from the seed, a person, or theenvironment.  For sake of discussion let's assume that it came from theseed.  But if it came from the seed, which seed did it come from?  

If thelot of seed was quite contaminated, it probably came from the broccoli orcabbage.  If it were in the alfalfa, it probably would have been detectedwhen each bag of the entire lot of 880 bags were sampled and tested by ISS. 

If thecontamination were localized to one bag of seed, it could have been thebroccoli, cabbage, or alfalfa.  If there is light contamination, say, 4 contaminatedseeds per kilogram, the odds of finding it if it is only in one bag isonly 9.5% (by taking a 25 gram sample).  It needs to be in 7 bags toincrease the odds of capture (for detection) to 50%, and in 47 bags toincrease the probability of capture to 99%.  So when you are testing880 bags, it only needs to be in 5% of the bags, at low levels, to geta 99% probability of capture. 

Inorder to have a 50% probability of capture in one bag the seed wouldneed to be contaminated at the rate of 28 seeds per kilogram, and in orderincrease to probability of capture in one bag to 99% there would need to be 185contaminated seeds per kilogram, or about 1 in every 2,700 seeds.  Theother alternative is to increase the sample size from 25 grams per bag to 175grams.

Inother words, if only a few bags of a lot are lightly contaminated, there is noreliable way of detecting it other than testing larger quantities of seed. This is impractical for most seed companies because the seed is not usable afterit is tested.  

 Thismeans that even though seed has been properly sampled and tested for pathogens,there may be times when it escapes detection.  If it is contaminated,it is likely to be either extremely lightly contaminated (one contaminatedseed per 4 bags) or have light contamination localized in a few bags. 

Thechlorine kill step would probably be effective against the extremely lightlycontaminated seeds.  If it is localized to one bag, then fortunately theother 879 bags are not contaminated so there would not be a major outbreak. Just like sanitizing or post testing, the system is not perfect, but it isa very important risk reduction step. 

 Inthe case of the seed used by the grower who prevented the outbreak, he does notknow which seed contained the pathogen.  If the contamination came from thebroccoli or cabbage, the industry may have a major problem.  If it camefrom the alfalfa seed, it probably doesn't.