Seed Sampling of Blended Lot

Seed Sampling of Blended Lot


International Specialty Supply

June 5, 2002

Dear Bob,

“How reliable is your seed sampling program if the lot has been blended with other lots?”

Dear Grower,

The statistical reliability is only as good as the probability of detection of the smallest lot that is blended in. For example, suppose you have a full load of alfalfa seed with 880 fifty-pound bags. Let’s also say that the shipment is blended from 3 lots of 1,000, 15,000, and 28,000 lbs. It does not matter if the lots are blended because the probabilities average out. If there is a pathogen in the seed at the rate of 4 cfu per kg or higher, your probability of capture for each is:

1,000 lbs = 20 bags sampled x 20 grams per sample = 400g sampled. According to our charts, the probability of capture is 79.81%

15,000 lbs = 300 bags sampled x 20 grams per sample = 6.0 kg sampled. The probability of capture is 99.99999999%

28,000 lbs = 560 bags sampled x 20 grams per sample = 11.2 kg sampled. The probability of capture is virtually 100%.

In this case, even though you sampled and tested a total of 17.6 kg (39 lbs) of seed, you would need to figure the sample is only 79.81% reliable because of the small sublot within the lot. You are mixing the 1,000 lbs of seed with the other 43,000lbs of seed. This reduces a 4 cfu per kg contaminated lot to 0.091cfu/kg over 44,000 pounds of seed. On the other hand, you are increasing the amount of seed sampled from 440g to 17.6 kg. If you plug in the numbers, you find out that it comes out to be exactly 79.81%. A statistical probability chart for seed sampling is located on our website under ISS Seed Safety.

I will be the first to admit that this method is not perfect. An example often brought up is,”What if a mouse gets into one corner of one bag?”

Careful inspection of the shipment is critical to the integrity of the sampling and testing program. If there is a mouse in a bag, there is probably a hole in the bag where the mouse entered. And if the seed was stored or shipped in a rodent infested area,the bag inspection and bag swab tests should reveal the problem.

If a mouse got into a bag or in to the seed before the seed was bagged, and was only in one bag, the probability of detection is very low. But it is only one bag, not 880. If your seed sanitation does not completely kill the pathogens in the seed, hopefully your post testing will detect it. If the pathogens make it past all three of these hurdles, there could be a small, localized, outbreak. Not a national outbreak, involving many growers, that lasts for months before the source is detected. No doubt this is an effective risk reduction step and probably the most effective of the three.

None of the three methods employed to alleviate pathogen problems in sprouts is 100% reliable. Post testing is not completely reliable, and when using chlorine, unless you kill every pathogen, the re-growth is back to about what it would have been if you had not sanitized.

There are a lot of researchers and regulators who read the SproutNet so I’m going to through out the following hypothesis and see if it sticks:

There is an inverse relationship between seed sampling & testing, and sanitation. That is, it stands to reason that the more contaminated the seed is, the less likely you are to be able to kill all of the pathogens using seed sanitation. Yet, the more contaminated the seed is, the easier it is to detect a pathogen using seed sampling and testing.

To rely just on chlorine is folly. A truckload of seed, contaminated at the rate of 4 cfu/kg, has roughly 72,575 individual pathogen cells total. the FDA recommended chlorine soak gives a 2.5 – 3 log reduction. A three log reduction would bring the count down to 72.5 viable pathogen cells in the truckload, which will multiply during sprouting. In other words, unless the pathogens are detected in post-testing, there will still be an outbreak. Not every batch of sprouts will be contaminated, just the theoretical 72.5 batches. But if the lot is contaminated at 4 cfu/kg, the probability of detection using proper seed sampling and testing is well over 99.9999%.

On the other hand, seed that is very lightly contaminated, say at the rate of .01 cfu/kg (that’s 200 cells per truckload), has a probability of detection of less than 3% using our method of seed sampling and testing. But the chlorine would theoretically reduce the cells to 1/5 of one cell, for an 80% chance of eliminating every single cell.

The two methods, seed sampling and testing along with the FDA recommended chlorine soak, compliment each other well to virtually eliminate the risk of salmonella or E.coli 0157:H7 making it from the seed to the consumer.