Monday, October 4, 2010

Unscrambling The Egg Mess: Playing The Odds

Waiter, I think that there's a Salmonella in my egg!

When the British prepared their egg safety plan and wanted to know what proportion of the country's eggs contained Salmonella, they sampled eggs and tested them in the lab. And, after several years, when they wanted to know what impact the program had on the incidence of Salmonella in eggs, they sampled more eggs and tested them.

Based on the analysis of egg samples in the lab, the UK Food Standards Agency was able to show that their intervention resulted in a three-fold improvement. The frequency of Salmonella-contaminated 6-packs of eggs dropped from 1 in 100 (1%) in 1996 to 1 in 290 (0.34%) in 2004.

When the Australians wanted to determine the status of Salmonella contamination in their domestic egg supply, they sampled and tested 30,000 eggs – 10,000 for exterior shell contamination and 20,000 for internal contamination. Not one of the 30,000 eggs was contaminated with Salmonella.

When the US Department of Agriculture wanted to determine the annual fraction of eggs contaminated with Salmonella Enteritidis in the United States, they decided that conducting real-life sampling and testing would be too expensive. So the agency put pen to paper – or electrons to computer drives – and developed a statistical model using data from the scientific literature.

The computer model came up with a contamination incidence estimate of anywhere from 1 in 12,000 eggs to 1 in 30,000 eggs. The "most likely" estimate was 1 Salmonella Enteritidis-contaminated egg in 20,000. That number, which was based on literature reports from the 1990s and was published in 2000, has been treated as a hard number ever since.

The USDA "researchers" tested their computer model against data obtained from a single unpublished random survey conducted at the University of California-Davis in 1998. The California survey tested 1416 pooled egg samples (20 eggs in each pool) and found one Salmonella Enteritidis-positive pool – assumed by the researchers to be equivalent to one positive egg out of 28,000.

Was the USDA's approach appropriate? This is what the Australian's had to say about the need to carry out real-world investigations:
"To provide a meaningful estimate of exposure, it is essential to have data on the prevalence of Salmonella contaminated eggs (internal and external contamination).

Directly comparing results from published surveys is difficult due to differences in sample sizes and methodologies. For example testing for surface contamination of egg shells can be undertaken by swabbing a section of the shell or by rinsing the entire shell surface. Egg contents can be sampled aseptically by separating the contents without contact with the shell surface. Alternatively Salmonella can be isolated by crushing the egg, allowing contact of the egg contents with shell, and isolating from the mixture."

After a number of delays, FDA at last has begun its first round of egg inspections under the newly implemented Egg Safety Rule. But without real baseline data to compare against, how will anyone be able to determine whether and to what extent the plan is working? By carrying out another hypothetical statistical analysis?

If you would like to receive automatic email alerts for all new articles posted on eFoodAlert, please click here or submit your request using the sidebar link. Please include "subscribe eFoodAlert" in the subject line.

1 comment:

  1. Wow! What a shocker! Thank you, Phyllis...


Note: Only a member of this blog may post a comment.