Wednesday, September 22, 2010

To test, or not to test – that is the question (with apologies to Shakespeare).


I had intended to write my next blog article on the latest information on Campylobacter in New Zealand.  However, in view of the continuing Salmonella-in-eggs problem in the USA and yet another recall in New York State of ground beef owing to the detection of Salmonella, I decided to write about product testing and the alternatives.

It had been suggested that the feed was the source of Salmonella that infected the Wright County Egg Co.  Reporting in the Wall Street Journal last week, Alicia Mundy and Bill Tomson wrote that the Food and Drug Administration had failed to detect Salmonella at the premises of the suppliers of the feed: 


Officials warned that the investigation was continuing, but from my point of view, the fact that Salmonella was detected in the Wright County Egg Co. farm(s) on 426 occasions from 2008 to present is a pointer to the farms at least having some responsibility.

Testing of products, be they eggs, steaks or lettuces, can potentially detect Salmonella and other pathogenic bacteria, such as Escherichia coli O157:H7, allowing us to reject the lot.  Unfortunately, testing doesn’t guarantee that we will detect Salmonella.  In all sampling plans that I am aware of, detection of Salmonella in the sample is cause for rejection.  Under the ICMSF 2-class attribute sampling plan nomenclature, ‘n’ is the number of samples to be taken from a lot, ‘c’ is the number of failures permitted and ‘m’ is the threshold above which the sample fails. Thus for Salmonella testing, c = 0 and m = 0.  What number of samples should we take?

Statistics show us that when the frequency of contamination in the lot is 1 in 1000, setting c = 0 we would need to examine 2995 samples to achieve 95% confidence in detecting the defective sample; that is, we would still have a 5% chance of accepting a contaminated lot.

Put another way, extensive testing is prohibitively expensive, and many sampling plans call for only 5 samples to be taken from a lot. Even with 40% of the lot being defective, we would still accept the lot about 8% of the time.

So, testing seems a bit unreliable.  Is there a better way?  Well, yes there is.  It’s called HACCP (Hazard Analysis Critical Control Points) and it came out of the space programme, developed initially as the Modes of Failure Programme by the Natick laboratory and then by The Pilsbury Corporation.  Without going into detail on this occasion, it involves looking at the ingredients, the process and the processing facilities and looking for every way the food might become unsafe.  These aspects are eliminated where possible and those hazards that cannot be eliminated are controlled by keeping tight control over Critical Control Points, such as cooking operations.

The hazard analysis can then be used to develop a Food Safety Programme for the factory.  The factory runs the process according to this programme and monitors the CCPs and keeps accurate records of the entire process.  There is then no need for end product testing – if the process is under control,  safe food will be produced.  Regulatory monitoring becomes a matter of checking the records and auditing the programme on a regular basis.  This is a really good system and in a perfect world, it would work perfectly.

The problem with this approach is that the regulators need to have enough staff to do the inspections and the companies running the factory have to obey the rules, keep accurate records and come clean the minute something goes wrong.  Companies must want to produce safe food, rather than just make money.

It is very sad that in many of the recent food poisoning outbreaks, it has eventually been shown that companies falsified records, while supposedly independent inspections were at best inadequate and at worst negligent and corrupt.  

In this environment, the regulators need more staff, better legislation and more powers to shut down unsafe operations.  I have watched in horror as one outbreak after another has resulted in severe injury to some consumers, massive food recalls and in some cases permanent shutdown of factories and wondered, “Where will it all end?  Will there be any food manufacturers left in business?  Will testing requirements become so onerous that food becomes ridiculously expensive?”  

I don’t think there is an easy solution, but until the cowboys have all gone, I see no alternative to better legislation, more truly independent inspection, more frequent testing of product and full traceability of raw materials and finished products.

No comments:

Post a Comment

Comments on this blog are welcome, as are questions and suggestions for further articles. Comments are moderated to reduce the incidence of spam. If your comment includes a link to a commercial site, it will normally be rejected.