Annelotte van Bommel

90 Because patient (age, SES) and tumor factors (tumor and nodal stage, multifocality, grade) significantly affected IBR rates (data not shown) 10 , these factors were included in the multivariable model to correct for confounding ( Table 3 ). The multivariable model demonstrated that patients who underwent a mastectomy at a cancer specific hospital had a higher chance of receiving IBR (OR=13.39, 95%CI: 9.76–18.38) compared to patients who received amastectomy at a district hospital. As for DCIS, invasive breast cancer patients who were treated at a teaching hospital did not have a significantly higher chance of receiving IBR (OR=0.97, 95%CI: 0.83–1.14) compared to patients treated at a district hospital. University hospitals demonstrated to perform significantly less IBR compared to district hospitals (OR=0.65, 95% CI: 0.45–0.95). Also, the number of weekly MDT meetings positively affected the rate of IBR. Hospitals having one or two MDT meetings per week (OR=0.74, 95%CI: 0.61–0.89 and OR=0.66, 95%CI: 0.54–0.82, respectively) performed significantly less IBR compared to hospitals that organizedmore than twoMDTmeetings per week. The percentage of patients receiving IBR increased with an increasing number of plastic surgeons practicing in that specific hospital. Hospitals with 0.5–2.5 plastic surgeons per 100 new diagnoses of breast cancer performed 5-fold more IBR (OR=5.55, 95%CI: 3.04–10.11) and hospitals with more than 2.5 plastic surgeons performed almost twelve-foldmore IBR (OR=12.33, 95%CI: 6.03–25.21) compared to hospitals with less than 0.5 plastic surgeons per 100 diagnoses of breast cancer. The number of breast surgeons did not affect IBR rates. The structural attendance of a plastic surgeon at the weeklyMDTmeeting was strongly associated with performingmore IBR compared to MDT meetings with no or incidental plastic surgeon attendance (OR=2.91 95%CI: 2.39–3.54). In Figure 2 , the variation between hospitals in the use of IBR after mastectomy for invasive breast cancer in the Netherlands is demonstrated. Case-mix adjustments for patient and tumor factors, significantly affecting the use of IBRwere performed. Adjustments for hospital organizational factors were performed, due to the characteristics of a multilevel analysis. Adjusted data demonstrated a decrease in hospital variation in the use of IBR from 0–62% to 0–18%.