Perhaps that was a p-value typo of 0.0501, which would be more consistent with the other results reported, including the p-value for that variable from the pooled studies. If a typo, it would be a classic. Unlikely, I suppose.