Friday, April 27, 2012

Am I Missing Something Here?

With the recent discovery of a cow in California found to have "Mad Cow Disease" (BSE—bovine spongiform encephalopathy), facts about the disease, and its potential for being transmitted to humans, have been repeated in the news.

Here's something I didn't know: According to Marion Nestle (note: those are two separate links in case you want to click through), professor of nutrition, food studies, and public health at New York University and author of many books on food and public policy, some 34 million cows are slaughtered in the U.S. every year. Of these 34 million, only 40,000 are randomly selected and tested for the presence of BSE. If my math is correct, that means one cow out of every 850 is tested for the presence of the disease.

To be clear: Those numbers don't say that the chances of encountering BSE in beef are 1:850. It says that one in 850 cows is tested. The chances of consuming tainted beef are much lower given the rare occurrence of finding afflicted cows.

But still, the fact that a cow was found with BSE means that the disease is out there. With a testing ratio of 1:850 I would not, were I a meat-eater, be feeling very confident about the purity of the "other 849" that aren't tested—especially given the death sentence represented by this disease for those who develop the human version of BSE (Creutzfeldt-Jakob disease) from eating tainted beef.

In light of the discovery of BSE in the afflicted cow, the USDA issued this statement:

"The beef and dairy in the American food supply is safe and USDA remains confident in the health of U.S. cattle. The systems and safeguards in place to protect animal and human health worked as planned to identify this case quickly, and will ensure that it presents no risk to the food supply or to human health. USDA has no reason to believe that any other U.S. animals are currently affected, but we will remain vigilant and committed to the safeguards in place."

But—does testing one in every 850 cows justify the confidence expressed in the above statement? How would I feel about flying cross-country on an airplane if only one out of every 850 planes was given even cursory pre-flight maintenance and checks before taking off? Or only one in 850 was given the routine periodic detailed checks that planes undergo?

Knowing that this disease exists in the meat supply (remember: an afflicted cow was just discovered), I wouldn't be feeling too good about my chances of avoiding it if I included beef in my diet given a 1:850 testing ratio.

Am I missing something? In case I'm not, I'm sticking with my plants-only diet.


  1. Dad, I think what you are missing is the power of random sampling to capture characteristics of populations. We gain surprisingly accurate insights into the political views of all adults in the US by sampling just several thousand.

    A recent Gallup poll with 95% confidence and 3% error sampled 1534 to get an indication of what the roughly 227.5 million adults think about Obama's job approval. But the MATH works.

    Generally in scientific literature I think 1% and 5% error is acceptable for statistical tests. If we want to be within a 1% margin of error with 99% confidence (setting a very high standard) of bovines with MCD we only need to randomly sample 16581 out of 34000000. If the testers are testing 40,000 they are working well beyond scientific standards for accuracy and taking surprising care. And well they should given the seriousness of the disease.

    So, while I'd agree that caution is warranted I bet your risk of getting in a car wreck, plane wreck, or heck maybe even being robbed could be greater than randomly purchasing an infected piece of meat. But, just like roaches-if you find one there are bound to be more. I guess that's why beef importers are already freezing orders.

  2. That's very helpful -- I think I understand and agree with the power of random sampling in polling and other applications. But when it comes to life and death issues like tainted meat, it seems the sample needs to be larger. When writing the post, I thought about other life/death samples that you mention like plane and car crashes, lightning strikes, etc. -- and we take those chances in the course of daily events without a second thought. And I guess it applies to meat as well -- just a built-in risk. And the people who have actually died from BSE is infinitesimally small relative to the number of cattle slaughtered. One's risk of dying from BSE is WAY smaller than dying in a car or plane crash, etc. Still, if one infected cow is not tested, potentially scores or hundreds of people could be affected from the meat. It just seems the sample:universe ratio is awfully small for such a deadly possibility. Again, ALL airplanes are checked for malfunctions, not just one out of 850.

    Anyway, thanks for the sampling/margin of error info -- good input.

    1. Good point about ALL planes being checked for malfunctions. I wonder what the cost of checking every piece of meat, every other, or even one out of ten would do to the cost of meat in general? And I wonder what meat eaters would say about the cost/benefit of such increased testing?

      We should also include all other relevant regulations and procedures that are put in place in addition to the testing of cows for BSE that reduce the risk of any cow getting the disease. It's not as if the BSE sampling is the only measure in place to prevent someone purchasing contaminated meat. These regulations probably involve cleanliness standards, feed standards, quality of life standards for cows, medicines given to cows, safety measures for killing, handling, transporting, packaging, etc.

      Given that so few die from BSE maybe our conclusion should be that we have a really good system in place. We should be PROUD not WORRIED?

      I'm teaching inductive logic in the fall and will have to reeducate myself on much of this. The world of risky inference and the mathematical procedures used to minimize risk in scientific testing always end up surprising me.


    2. All things being equal (which they're not) the market could determine the cost/benefit ratio of increased checking up and down the chain. Given peoples' love of beef, it's not likely they would put up with the increased costs -- most would rather "pay their money and take their chances."

      I think the scarcity of BSE-caused deaths is less related to checking (again, NOT checking 849 out of 850 is not very thorough checking) than it is to other regulations -- specifically, eliminating the cannibalism that was present in the animal food chain -- feeding animal parts to other animals as part of their manufactured feed. The spine/brain/nerve tissue that contains BSE agents that used to routinely be fed back to cattle as processed feed allowed one infected cow to infect many others. Eliminating that source of contamination has probably drastically reduced the chances of any cattle being infected. But still, the one found in California -- one out of 850 -- means that the disease is still out there. Kudos for eliminating the possibility of contaminated feed are due to "the system." Still -- I'm glad I don't eat meat. One less "chance" to worry about.