Potential for bias in studies on efficacy of prophylactic surgery for BRCA1 and BRCA2 mutation.