Two Processing Methods for the Isolation of Salmonella From Naturally Contaminated Alfalfa Seeds

Thefollowing research was conducted in order for researchers to be ableto determine contamination levels in seed. Though it was not the intention of the research, the data also shows thatpathogens can be detected in seed with very high probabilities if enough samplesare tested.  The highlights and underlines are mine in order to make apoint at the end of the article.

Research Note: TwoProcessing Methods for the Isolation of Salmonella from Naturally ContaminatedAlfalfa Seeds

Journalof Food Protection: Vol. 64, No. 8, pp. 1240-1243.
Gregory B. Inami, Sue Ming C. Lee, Robin W. Hogue, and Rita A. Brenden
California Department of Health Services, Microbial Diseases Laboratory, 2151Berkeley Way, Berkeley, California 94704, USA

Abstract-Twoprocessing methods were examined for the recovery of Salmonella fromnaturally contaminated alfalfa seed. Seed samples, from each of threeinvestigations, were processed by sprouting and shredding before preenrichmentand culture. In lot A, Salmonella serotype Newport wasisolated from 3 of 30 sample units with the sprouting method and 2 of30 with the shredding method. In lot B, three serotypes invarious combinations were isolated from 10 of 30 sample units with thesprouting method and 9 of 30 with the shredding method. In lot C,Salmonella group C1 was isolated from 27 of 30 sampleunits with the sprouting method and 24 of 30 with the shredding method.Additionally, serotype Newport was found in one lot C sample unit. Usingshredded seed data, a most probable number (MPN) for Salmonellacontamination per lot was calculated. Serotype Newport was estimated at 0.07MPN/100 g in lot A; the concentration for three serotypeswas estimated to be 0.36 MPN/100 g in lot B; Salmonellagroup C1 was estimated at 1.8 MPN/100 g in lot C.Our success in isolating Salmonella from alfalfa seeds was likelyattributed to the volume of material tested and the quick acquisition of theseeds after the outbreak was identified. Shredding the seeds was easier andyielded definitive results more quickly than sprouting. 

My Comments on the Above Research

Youonly need to find one pathogen in a seed lot to determine it is unfit for humanconsumption.  In this study, theresearchers were able to determine pathogens 100% of the time. For instance, in lot A, they found pathogens in 3 of the 30 samples. This may sound like 10% of the time, but because they took 30 samples,they found it not just once, which would have rejected it, but three times.  They used two different methods for each lot, making thetotal samples 60.  If you pull onesample out of a lot and test it, odds are, you won't find it.  Pull 60 samples and the odds increase dramatically.

Lot A,  5 of 60 = REJECTED
Lot B, 19 of 60 = REJECTED
Lot C, 51 of 60 = REJECTED 

Usingthe formula, Probability = 1-(C/T)^N, I calculate the probability of findinga pathogen in each of these lots is:

Lot A (.07 MPN/100g) = 98.5%
Lot B (.36 MPN/100g) = 99.9999999%
Lot C (1.8 MPN/100g) = 99.9^30+%

Withsuch probabilities it is no wonder the researchers were able to find a pathogenin each lot more than once.

(Note: (C/T) = assumed ratio of clean seeds to total. N = the number of seeds sampled.)

Lot Ais the least contaminated with a most probable number of 0.7 cells per kilogram. A seed trier (at least the one we use), draws 25 grams. When we pull samples from every bag of a truckload we are pulling 880samples.  Statistically, the odds ofcapturing a pathogen in this lightly contaminated seed are in excess of99.9999%.   In the heavilycontaminated seed the probabilities are astronomical.

If youwould like a copy of the entire article please give your ISS or Sungardensalesperson a call.

Bob Sanderson's Comments on the Same Article

DearBob (Rust)

"Congratulations!Not only for your courage and successful outcome in this lawsuit, but wait aminute- since when are you getting listed in JFP article references?

I am quite impressed with the Inami et al research note in the August JFP. Forseveral reasons. One of course being that you are deservedly mentioned.

If I read Greg Unami's report right, he seemed to find that taking a 3Kg samplefrom each lot, they had no trouble finding contamination in the three lots theylooked at, even in one seed which was only contaminated at .7cfu/Kg, which ismuch lower than the 4 cfu/kg mentioned as being very low (and by implicationvery hard to detect) in the White Paper.

It looks as though they got a positive from this seed (Lot A; going on 5 yearsold) 3 times out of 30 using the sprouting method. To me, that means that any100 g sample of that seed, grown out and tested, would have a 10% likelihood ofbeing positive. Therefore, sampling and testing 30: 100 g samples, thelikelihood of detection becomes as close to 100% as death and taxes.

Using the formula P = 1-(C/T)^N, the probability of getting a positive from a100g sample of seed contaminated at .7cfu/kg would be: P = 1-(9993/10000)^100,which solves to P = 6.7%, rather than the 10% which they got using the Thomasformula. So the two are close, but my formula is a little more conservative.Well, I'm a conservative kind of person.

The contamination levels in the other two seed lots were estimated at 3.6 cfu/Kgand 18 cfu/Kg. Using my P formula, a 100 g sample of the first would have a P =1-(9964/10000)^100, or 30% likelihood of being positive. Inami got 10 out of 30,for 33%. The third seed lot probability would be P = 1-(982/1000)^100, or 84%.Inami got 27 out of 30, or 90%.

It seems that the whole emphasis of the paper is in terms of researchers beingable to determine contamination levels in seed, rather than as a method forsuppliers or growers to determine if seed is contaminated.

You know what gets my goat (along with about a thousand other things): If peopleget sick because a simple procedure is not being used because the FDA says itisn't any good, and meanwhile the grower failed to use another procedure whichthe FDA says they should use, chlorine, which is horrible and dangerous anddoesn't work very well, and someone's ass ends up in the fire, and the FDAissues another press release about how dangerous the product is and what a bunchof slobs the sprout growers are... that gets my goat. How's your goat?

I would think that if there is another outbreak, and it can be determined thatit could have been averted if Inami's sampling and testing had been done, thatthe victims might consider a class action lawsuit against the FDA, if suchthings are possible.

So take a few days off. You deserve it."

Bob Sanderson


Dear Bob S.

Personally,the broccoli lawsuit was enough for me this year.  However, I understandyour frustration that seed sampling has received far less attention than itdeserves.  Thanks for your astute observations and for keeping seedsampling in the forefront.

Note: I generally don't put peoples names with their comments.  In this case, Bobhas done a tremendous amount of work and it would be good for researchers tocontact him regarding seed sampling  ( Don't be concerned about sending emails to the SproutNet.  Your nameisn't likely to appear with the article unless I request  your permission.