Apparently they have controlled for this obvious possibility:
In each of the three resulting models, our independent variable was the prevalence of T. gondii in a given country (the proportion of people who tested positive for T. gondii exposure) – indicated by published seroprevalence data.
(https://figshare.com/articles/Supplementary_Methods_and_Anal...)