On the efficiency of targeted clinical trials