Use of Mendelian randomisation to assess potential benefit of clinical intervention