Annelotte van Bommel

62 MATERIALS AND METHODS Data source Data were derived from the NABON (National Breast Cancer Consultation Netherlands) Breast Cancer Audit (NBCA), 9 a continuous national multidisciplinary quality improvement project in which a wide range of variables concerning patient, diagnostics, and treatments are prospectively collected by the hospitals themselves or the Netherlands Cancer Registry. The NBCA contains data registered in all 92 hospitals performing breast cancer surgery in the Netherlands. 10 The information concerning individual patients and hospitals is de-identified for this study, allowing comparisons without identification. Study population Data from all female patients who underwent a mastectomy for either primary DCIS or nonmetastatic invasive breast cancer diagnosed between January 1, 2011 and December 31, 2013 were selected. Information available in the NBCA on patient characteristics (age) and tumor characteristics (TNM classification, histological subtype, grade, and receptor status) were extracted. Four types of IBR were defined: implant breast reconstruction (including tissue expander), autologous breast reconstruction, a combination of both, and reconstruction not otherwise specified. Statistical analyses Invasive breast cancer and DCIS patients were analyzed separately. Differences in the use of IBR between hospitals were compared using a funnel plot. Patient and tumor- specific factors potentially affecting the use of IBRwere compared between women with and without IBR. Subsequently, to investigate which factors were related to the use of IBR, univariate regression analyses were performed. Next, factors with p-values of <0.10 were included into multivariate regression analyses using an enter model. These multivariate regression analyses were used to identify independent factors determining the use of IBR, corrected for the other factors that were included into the model.