Would need to see the full article to find out the size of their sample, but their confidence intervals are quite tight for a study of this nature, and they add in the caveat of further research being needed. This should be a least a warning correlation for cell phone users, and will make me rethink some of my usage habits.