Let's Eliminate Salmonella. No, wait, let's not.

Over at Hacker News, there was a link to this great Forbes article talking about the differences in regulation in the treatment of eggs. Apparently in the US, eggs are forced to be washed, while in the EU eggs are forced to not be washed. This also relates to the fact that US eggs are stored in the fridge, while EU eggs tend to be left at room temperature.

The whole thing is presented as a kind of 'duelling regulations' thing - in the end, it looks like there's odd biological reasons why being washed or not can impact how you choose to store them, and the chances of disease.

And then, buried at the end of page three, comes this gem:

Since the late 1990’s British farmers have been vaccinating hens against salmonella following a crisis that sickened thousands of people who had consumed infected eggs. Amazingly, this measure has virtually wiped out the health threat in Britain. In 1997, there were 14,771 reported cases of salmonella poisoning there, by 2009 this had dropped to just 581 cases. About 90 percent of British eggs now come from vaccinated hens – it’s required for producers who want to belong to the Lion scheme. The remaining 10 percent come from very small farmers who don’t sell to major retailers.

In contrast, there is no such requirement for commercial hens in the US. Consequently, according to FDA data, there are about 142,000 illnesses every year caused by consuming eggs contaminated by the most common strain of salmonella. Only about one-third of farmers here choose to inoculate their flocks. Farmers cite cost as the main reason not to opt for vaccination –FDA estimates say it would cost about 14 cents a bird. The average hen produces about 260 eggs over the course of her lifetime.

Wait, what? You mean that for 0.05 cents per egg, you can virtually eliminate salmonella poisoning? And this isn't being done in the US, because US farmers have correctly estimated that ignorant consumers aren't savvy enough to insist on this purchase?

Talk about burying the lead.

Wow. That sounds pretty outrageous. I read this piece, and my instinct was the think that the British policy of vaccinating hens sounds like a no-brainer.

But then again, we wouldn't be economists if we didn't shut up and multiply.

Let's assume that the entire reduction in salmonella comes from this policy:  14,771 - 581 = 14,190 cases of salmonella avoided by vaccinating hens.

The cost per egg as we noted is 0.05 cents (14 cents per hen, divided by 260 eggs per hen).

So how many eggs are consumed in Britain each year?

According to this estimate, almost 11 billion. That sounds ridiculously large, until you realise that with a population of 63.182 million, this amounts to a consumption of 174 eggs per person per year, or roughly one egg every two days.

Let's go with that number.

So the total cost of the policy each year is thus roughly 11,000,000,000 *$0.14/260 = $592 million.

This implies a shadow cost of each case of Salmonella equal to ($592 million / 14,190), or $41,741.

Put that way, it seems like more of an arguable proposition. Maybe we should cancel the policy?

Not so fast! Do you know how much weight I should place on your hunch about the value of salmonella? Zero! Shut up and keep multiplying!

According to this PubMed article there were 1316 salmonella-related deaths between 1990 and 2006. The paper abstract reports the mortality per person-year, but what we want to know is the mortality per salmonella case. According to the original article, there are about 142,000 salmonella cases per year in the US. Assuming a constant number of infections over the years, this gives us a probability of death conditional on salmonella poisoning of 1316/(142,000*17) = 0.000545. Using a statistical value of human life of about $7 million, this gives an expected mortality cost per salmonella case of $3816.

Do you value the pain and suffering of a non-fatal case of salmonella at $37,925? I sure don't. If you paid me 38 grand and guaranteed it wouldn't kill me, I'd be pretty keen to sign up for a case of salmonella.

Put differently, the implied cost of human life in the salmonella reduction program is ($41,741 / 0.000545) = $76.57 million, ignoring any value placed on pain and suffering for non-lethal cases.

In other words, despite the intuitive appeal of getting rid of salmonella just by vaccinating chickens, as a society we'd probably be better off spending the money on road safety, medical research, or something else with a lower cost of saving each life.

The interesting thing is that when I started out writing this blog post, my initial reaction was that it was amazing that the US wasn't requiring chicken vaccinations, and the hard numbers changed my mind.

Sometimes the best treatment is to do nothing. Long live NPV!