Statistics and the storyline

The line between simplification and obfuscation

Carl Zimmer is a respected science journalist, whom I have rarely known to fall prey to hyperbole. So, when I called him out on justifying the term “bombing” animals with antibiotics during the course of husbandry, he cited this number:

That 30 million pound figure is in fact cited by Maryn McKenna in the linked article, which I then traced to a figure from the Pew Health Charitable Trusts Health Initiatives, claiming to compare antibiotic usage between animals and humans, with linked sources. Looks legitimate enough, but good science reporting should dig deeper. Let’s look at some of the problems with this summary:

  1. It reports it as total weight (in pounds) used in animals as almost 4 times higher than used in humans in 2011. It doesn’t say if this is global, but most such figures tend to be USA centric. But it also doesn’t account for the distribution in usage; we certainly produce far more weight in farmed animals than in humans, far higher than 4-fold to sustain the human population. Normalized to per unit weight, the usage will turn out to be in fact lower in animal husbandry than in humans. But even this is a misleading figure — the use in humans is focused on a small group that are ill, while in certain cases, such as in growth promotion, small amounts of antibiotics are administered to entire populations of animals.
  2. It falsely renders equivalent all antibiotics. A kilogram of penicillin has very different usage, properties, solubility, toxicity, from of a kilogram of cephalosporin. Summarizing it by weight conceals the important distinction that the chemical properties of the compounds are significantly different.
  3. Another problem with lumping together antibiotics also hides the fact that the different classes of antibiotics used in the farm overlap in a minor fashion with those used in the clinic. I am unable to access the IMS Health Inc. data source used for the human usage, but based on a CHRT report from BlueCross/BlueShield in Michigan in 2009, the most heavily used classes of antibiotics are fluoroquinolones and macrolides. In contrast, animal husbandry relies more heavily on ionophores and tetracyclines, with fluoroquinolones barely registering as a blip in the report. This is consistent with the US Farmers and Ranchers Alliance research that no direct link has been detected between farm antibiotic uses and clinical resistance. That’s not so say that it’s not possible, and certainly biologists intuit that this is an eventuality — but the evidence isn’t there yet. We shouldn’t fall prey to confirmation bias.

This is by no means a denial that antibiotic resistance is a burgeoning probem in the clinic. For any new antimicrobial, resistance will evolve; resistance mechanisms long predate the time when humans started leveraging antibiotics as drugs. But I am calling for more responsible science reporting on this issue. The mainstream press is already shrill with panic, fingering and condemning antibiotic “overuse” in farms, painting images of animals injected with antibiotics en masse, to the point that the general populace are willing to pay more for “antibiotic-free” meat — a marketing term that is not scientifically validated.

Our science journalists come by their reputation by more cogent, more critical analysis of the facts, not succumbing to the common story, beholden to the objective standards of scientific temperance. They shouldn’t fan the fear.

Although I suppose if you make your living off the specter of “superbugs”, it’s a framing conundrum.