I think CHA2DS2-VASC is more discriminative in risk stratifying patients for stroke risk. There is an increased trend to rely on it more in clinical practice.
I echo Imran and Zaher's comments. Choice depends upon your studys' aim and research questions. I'll prefer CHA2DS2-VASc if testing a new antiagregan therapy for atrial fibrillation and stroke risk amelioration.
The effect of CHA2DS2-VASc is to reclassify many of those at intermediate or low risk into a higher risk category (i.e. score of 2 or greater). It likely identifies those at truly low risk of stroke better than CHADS2. However the predictive power is only slightly better than CHADS2 and it has some quirks - like that women automatically have a score of 1, although there is no evidence that young (
As it has been recognised, CHA2DS2-VASc is the best scheme to identify those patients at very low risk. These patients will not obtain benefit to be under oral anticoagulation.