The contribution of animal antibiotic use to antibiotic resistance in humans: Panel evidence from Denmark

Antibiotic use (ABU) in animals is postulated to be a major contributor to selection of antibiotic resistance (ABR) which subsequently causes infections in human populations. However, there are few quantifications of the size of this association. Denmark, as a country with high levels of pig production and good surveillance data, is an ideal case study for exploring this association. In this paper, we compile a dataset on ABU across several animal species and antibiotic classes, and data on the rate of antibiotic resistance (ABR) in humans across key pathogens, in Denmark over time (2010 - 2020). We run panel data regressions (fixed effects, random effects, first difference and pooled ordinary least squares) to test the association between the level of ABR in human infections and the level of ABU in animals. Between ABR in humans and ABU in animal species, we find a positive relationship for cattle, some evidence of a positive relationship for poultry and companion animals, and a negative relationship for fish, although the latter is likely driven by confounding factors. When lagging ABU by one year, the effect of ABU in cattle and companion animals remained similar, the effect of ABU in poultry fell in size, and ABU in fish was no longer significant, perhaps due to differences in life cycle length among animal species. Additional covariates were explored, including pet populations, agricultural production and GDP per capita (at purchasing power parity), but these results were limited by the statistical power of the dataset. Under all models, animal ABU determined only a minority of the change in human ABR levels in this context with adjusted R2 ranging from 0.19 to 0.44. This paper supports the role of animal ABU in determining human ABR levels but suggests that, despite comprising a large portion of systemwide ABU, it only explains a minority of the variation. This is likely driven in part by data limitations, and could also be due to a persistence of ABR once resistance has emerged, suggesting a significant role for socioeconomic and transmission factors in bringing ABR down to desirable levels. Highlights- We use panel regression to explore the link between animal ABU and human ABR in Denmark - ABU was linked to ABR in cattle, and potentially in poultry and companion animals - However, animal ABU did not appear to be the main determinant of human ABR - This could be due to poor data, or persistence of ABR after reaching a certain level - Animal ABU reductions alone may be insufficient to curb ABR