This sounds like one of those papers that got published with just barely "significant statistical evidence" but never had follow-up to verify, meaning you can't really draw conclusions for it.