Absolutely. If A and B are significantly correlated, either A causes B, or B causes A or X causes both.

The article itself, citing the study, presents some possible mechanisms through which such a reduction could be explained.