I would like to see the same investigation in a different population. The difficulty is that we don't know how many different kinds of associations they looked

I would like to see the same investigation in a different population. The difficulty is that we don't know how many different kinds of associations they looked for in this population - they mention no link between PCBs, but that doesn't necessarily mean they only looked for an association between DDE and PCB. If, for example, they looked for an association between 100 compounds, the chances are pretty good that there would be some positive correlation from at least one of them. Doing the study in a different population, where you're only looking for this one particular association increases the likelihood it's not just chance.