Brexit and the geography of depression: A reply to Liew et al. (2020)

Liew et al. (2020) recently published a paper in this journal that analyzed antidepressant prescription trends in the context of the 2016 Brexit referendum and the sociopolitical discord that followed. They present a novel finding that Leave-majority constituencies in England seemed more adversely affected by that discord than Remain-majority constituencies. I offer criticism of their findings and methodology. Using the complete set of available NHS prescription data shows that the trend the authors detect dates from at least mid-2010 and is not associated with the referendum. In terms of methodology, I critique the potential ecological fallacy and issues of false equivalence in their study design. The former stems from the inability to adequately control for demographic heterogeneity within constituencies, and the latter stems from the fact that the populations from which they draw their data are not equivalent in potentially important ways. Finally, I conclude that the key trend the authors detect seems to merely be a geographic artifact. The set of Remain-majority constituencies unintentionally oversamples the areas of England with the lowest rates of antidepressant prevalence, Greater London and the Southeast. Remain-majority constituencies outside of those two regions have roughly the same antidepressant prescription levels as Leave-majority constituencies in all of England. In itself, that is a troubling fact of social epidemiology, but Brexit is associated with it neither spatially nor temporally.