Bad FDA reporting reminds us self reported data is tricky

I’m a dog person and a nerd. No surprise I eventually come across curiosities like the FDA’s page on pet food recalls. Normally, these are boring: Food brand so-and-so has toxic levels of vitamin D, etc. Sometimes, they do cool investigations on Chinese jerky treats.

A recent investigation is different though. Apparently, there’s been an explosion of reports to the FDA on a condition called Canine Dilated Cardiomyopathy (CDM):

Note that these are self-reported to the FDA. So either there’s an explosion in a cause of the disease, or there’s an explosion in likelihood to self report for some reason.

The FDA also finds that the more expensive “grain-free” foods tend to be over-reported:

It’s probably not the food’s fault

It also finds that Golden retrievers are are more likely to be reported, but note that:

FDA has observed a reporting bias for breeds like Golden Retrievers due to breed-specific social media groups and activities that have raised awareness of the issue in these communities and urged owners and vets to submit reports to FDA. Then they detail all the tests they did on the reported foods, and find nothing anomalous.

At this point, the most likely cause of the explosion in reports is evident: social media panic. The fact that certain foods are over-represented is simply showing the correlation between people who are active on pet social groups and buying expensive dog food.

Of course, when I saw this FDA report online, people put it another way: “the grain-free dog food is causing horrible disease!” which it probably doesn’t, but if it did it surely not what this FDA investigation is showing.