Annelotte van Bommel

138 SUPPLEMENTARY APPENDIX 1: POWER CALCULATION The power calculation for this questionnaire was based on the Breast-Q modules that were included in our questionnaire, since the questions for these modules have been validated. The analysis for these questions however will be conducted in a separatemanuscript. Based on Zhong et al (2011) 1 , an SD of about 20 was found for separate Breast-Q subscales. A difference of 10 points on each scale (0-100) was considered clinically relevant. With an alpha of 0.05 (double sided) and a power of 0.85, we needed 32 respondents in each group. Since we estimated we could invite a lot more patients, we chose a lower difference in points: h = 0,50  n = 32 (difference of 10 points) h = 0,40  n = 49 (difference of 6 points) h = 0,25  n = 126 (difference of 5 points) h = 0,20  n = 197 (difference of 4 points) Furthermore,weexpectedaresponseofabout25%.Therefore,wedecidedto invited500patientspergroup (500*0,25=125).Sincewewantedasampleofpatientsthatweretreated inalltypesofhospitalsandallregions in the Netherlands, we decided to include at least 20 different hospitals. This meant that 50 patients had to be selected per participating hospital. Wehypothesizethatselecting50patientsperhospital leavestherequiredworkloadperhospitalacceptable, since all patients are selected in deliberation with the physician and each hospital has to prepare letters for all selected patients, without receiving a financial incentive. Sincemorethan20hospitalscommittedtoourstudy,wedecidedto includethesehospitalsaswell;therefore, we were able to indicate differences of 4 points on the Breast-Q subscales.