ISS Seed Screening Seen Theoretically As Several Log Reductions
Bob Rust discusses ISS seed screening processes, which are widely viewed to be the most advanced sprout seed screening capabilities in the sprouting industry.
ISS Seed Screening Seen Theoretically as Substantial Log Reduction
September 16, 2009
Bob Rust, reviewed/edited by Martin Cole
In his keynote address to the June 9, 2009 Scientific Session of the ISGA Convention in Chicago Dr. Martin Cole (Chairman of the International Commission on Microbiological Specifications for Foods (ICMSF) and Director, National Center for Food Safety and Technology (NCFST)), discussed the theoretical possibility that ISS Seed Screening protocol may be equivalent to a 2.34 log reduction or better.
Risk can never be completely eliminated, but it can be controlled or reduced to an appropriate level of protection (ALOP). The FDA stated in the White Papersthat commercial sprout producers need to implement a series of procedures that reduce the risk of contamination by five logarithms, or “log”. This appears to be their ALOP.
Log reductions in sprouts are based on either real or theoretical levels of contamination. In experiments with chlorine, for example, either naturally contaminated seed with known levels of contamination are used, or seed is inoculated with pathogens to a known level of contamination.
In practice the seed is usually not contaminated, so the chlorine is not reducing any pathogen level at all. But theoretically it would if contamination were present.
In the case of seed screening, one can calculate the probabilities of capturing a contaminated seed at theoretical levels of contamination. Conversely, you can also look at what contamination level you will find with a certain probability. For example, according to ISS’ calculations, if you sample 880 bags of seed at 25 grams per sample, 99% of the time you will capture a pathogen in seed that is contaminated at the rate of 1 contaminated seed per 5 kg. That is only 4606 contaminated seeds per truckload, or 1 contaminated seed per 4.78 kg. This is assuming non-homogeneous distribution throughout the lot but is assuming homogeneous distribution throughout those bags with contamination.
Dr. Cole calculated that sampling 25 grams from 120 bags of seed (3kg composite sample) would capture for detection with 95% certainty contamination as low as 1 cell per 4.57 kg. His program calculates probabilities of acceptance for materials with different microbial loads and population standard deviations. The pathogens are assumed to be log-normally distributed with a standard deviation of 0.8. This is a high standard deviation used when distribution is expected to be uneven throughout the population (seed lot).
“A standard deviation = 0.2 log10 CFU g-1 is used to describe a food in which microbes would be expected to be rather homogenously distributed within a batch (e.g., for liquid food with a high degree of mixing). A standard deviation of 0.4 log10 CFU g-1 is assumed for a food of intermediate homogeneity (e.g., ground beef) and a standard deviation = 0.80 log10 CFU g-1 for an inhomogenous food (e.g., solid food).” 1
For the purpose of illustration, Martin compared this to a worst case contamination scenario of 100 cfu per kilogram of seed. Sampling 25 grams from 120 bags with a standard deviation of 0.8 was 2.34 logs less than 100 cfu/kg, or a 2.34 log reduction.
120 x 25g samples (=3kg) sprouted and tested would be able to reject a lot with -3.66 log cfu/g, or 1 cell in 4.57kg (SD=0.8). Operating Character (OC) Curve is scaled to relate mean log count to the microbiological limit.
ISS often samples up to eight times the amount that Dr. Cole ran through his analysis. ISS asked Dr. Cole if he could run larger samples. Unfortunately, his program was only designed to go up to 160 samples so he ran a few examples at various sample sizes.
(1cell in 4,57kg)
(1cell in 6.3kg)
(1cell in 2.2kg)
(1cell in 7.1kg)
(1cell in 13.2kg)
(1cell in 20kg)
(1cell in 33.8kg)
(1cell in 40.7kg)
(1cell in 45.7kg)
Note that the three models of 30 kilograms show the importance of increasing the number of samples compared to the size of the samples. The 30kg from 120 samples gave a 0.13 better log reduction than 30 samples of 1kg.
ISS will typically take 880 samples of a full truckload of seed. Dr. Cole is working to change his modeling program to accommodate these numbers. ISS estimates they will be in the range of a 3.5 log reduction.
These theoretical log reductions are for capturing a pathogen. It does not take into account the very real possibility of a not detecting captured pathogens. To reduce this risk ISS’ Seed Screening Protocol now involves testing the sprouts in one lab as well as the spent irrigation water in a separate lab. This too is no guarantee that the seed is not contaminated. It is a risk reduction step that should be employed along with Good Manufacturing Practices including sanitizing the seed and an effective microbiological hold, test, and release program.
The NCFST will shortly be re-convening a task force on sprout safety in collaboration with FDA and the industry to further develop this approach and to evaluate it in the context of the original guidance given by FDA in 1999.
1(Relating microbiological criteria to food safety objectives and performance objectives; Food Control, The International Journal of HACCP and Food Technology, Vol 20, Issue 11, Nov 2009, page 970; M. van Schothorst, M.H. Zwietering, T. Ross, R.L. Buchanan, M.B. Cole, International Commission on Microbiological Specifications for Foods (ICMSF))