It’s always good to know that people are reading your work. Guardian columnist George Monbiot has clearly had a thorough look at the Adam Smith Institute’s report on chlorinated chicken, which I authored.
In that, I argue that the UK should consider accepting imports of US chlorine-washed chicken, which could lead to lower food prices and help to secure a better transatlantic trade deal. I draw on existing research from the European Food Safety Authority, the Institute of Environmental Science & Research, and the University of Maryland to show that chemically-rinsed birds are safe for consumption, and that these rinses are effective at disinfecting poultry.
Monbiot does not contest these headline claims. Rather, he disputes one use of World Health Organization figures.
These figures are by no means the crux of the report. Nor are they particularly controversial - the Codex Alimentarius Commission, the body which sets international food standards, has published guidance for chlorine-washing chicken that is safe for human consumption.
And Monbiot does not question the logic behind my use of the WHO data. If chemically-rinsed chicken is ineffective in controlling pathogen spread, then as I write in the report, “one would expect foodborne illnesses carried by chicken to be much more prevalent”.
Here’s what I wrote:
WHO figures reveal that salmonella and campylobacter infections in North American countries are not out of line with their European counterparts.
And here is what Monbiot has to say:
But [the Adam Smith Institute] says that figures from the World Health Organisation reveal that salmonella and campylobacter infections there are “not out of line” with rates in the European Union.
I checked the source: the WHO study the Adam Smith Institute cited. While the incidence of campylobacter is similar, it shows that the burdens of infection per head of population from the two species of salmonella it analyses – Salmonella typhi and Salmonella paratyphi – are, respectively, four times and five times higher in North America than in Europe. I cannot state that this is caused by chlorinated chicken, as the WHO doesn’t provide such detail. But I can state that the Adam Smith Institute’s claim is false.
There is something to what Monbiot is saying. When you compare developed Western Europe, where we use the farm-to-fork approach, to developed North America, where they mostly chlorine wash at the end, the rates of the two types of salmonella seem higher in the US.
But what Monbiot doesn’t report is the actual numbers. Salmonella Paratyphi A and Salmonella Typhi infections are so rare in both subregions that the difference Monbiot highlights is trivial in the context of total infections: 0.1 and 0.4 per 100,000 in North America versus 0.02 and 0.09 in Western Europe, respectively.
But even those average estimates are misleading: the 95% confidence intervals on those numbers all include zero, and indeed the rates of the opposite countries. That is: the stats are statistically insignificant from each other. When you drill down to two such specific sub-figures, relying on imperfect sources, you can’t draw a clear result.
It’s like rolling two dice three times and arguing the one with the higher number is loaded: you haven’t got enough data to make that conclusion.
Once you put those two specific infections together with all the diarrhoeal and infectious diseases we’d expect chemical washing to reduce, and weight by the disability-adjusted life years lost to them, you get a clearer comparison. The sum of DALYs lost to these poultry-related diseases per 100,000 in those North American countries is 9.5, compared with 10.11 in those European countries. As I wrote, “not out of line”.
This chart makes it clear: AMR A (North America) and EUR A (Western Europe) are barely different, although if we zoomed in massively we’d see the rate for all three hazards taken together is slightly higher in Europe. Guessing from the confidence intervals they give us, this difference might even be (marginally) significant.
If a cancer affects just 0.01% of the population then large differences in survival rates between countries do not much affect the relative burden of cancer overall between them. Whereas even small differences in survival rates for a cancer that affects 10% will make huge differences to the health of society overall. I think it’s this slip-up that Monbiot has made.
With more context and better understanding of the figures, it should be clear that our initial claim was correct and Monbiot’s “correction” misguided.