I think the idea of observational epidemiology is 2 fold. In extremely rare and specific cases (like with smoking and small cell lung cancer) cohort studies show such extreme correlation with new or rare diseases that causation can be implied. This is most certainly not the case with this study.

In other cases observational studies are used to look for a strong correlation which can be used to justify a much more expensive interventional study which could actually be meaningful.

From my account of the paper it's not apparent that the correlation is even strong enough to justify proper research. Something along the lines of proving the correlation between eating ice cream and drowning.