I haven't done any quantitative studies on it, but I imagine such a situation would indeed be quite rare. It depends, somewhat, on the phonological context of your language. Is hiatus avoidance something that there are synchronic phonological rules for? As in, do underlying sequences of vowels occur, which are then resolved somehow, or do they just happen to not occur for whatever reason? It could be, for instance, that your language lost some consonant specifically between two <a>, such as /h/ or /ŋ/.