Annelotte van Bommel

72 treated for invasive breast cancer supports this explanation because DCIS usually involves a larger area of the breast than invasive breast cancer. Similar findings were found in the study by Roder et al. 2 The uptake and variation of IBR can be only partly explained by the identified patient and tumor factors, suggesting that other factors contributed to the variation to a larger extent. Patient preferences may vary between institutions or regions. For example, the reported percentages of patients deciding not to undergo IBR varied between 17% and 62% in different regions of the United Kingdom. 1 An even more important role could be attributed to the personal perception, preferences, and beliefs of physicians considering patients eligible for IBR. 6,7,13,20 Hospital-related factors such as location in an urban environment or being a teaching hospital, high-volume breast cancer center, private hospital, or hospital with a plastic surgical department may all affect the rate of IBR. 8,14,21 Other organizational factors such as the length of the operation and availability of a plastic surgeon in the hospital may further challenge the frequency of IBR. Further research should focus on identifying these additional factors that may have contributed to the large variation found. A strength of the present study is that a national audit with 100% participation of all hospitals in the Netherlands provides a unique insight into the quality of breast cancer care delivered and the areas for improvement. An audit includes patients who are usually not included in clinical trials and reflects practice patterns in daily practice. Moreover, the availability of data at a hospital level enables nationwide hospital comparisons. A limitation of the present study is in the nature of a national audit itself. Registration bias may be present as the data were collected for a national audit. However, high rate of case ascertainment was found when the data was compared with that in the National Cancer Registry. Second, although many case-mix variables were available, there may have been unknown confounding variables that were not available in the data set and may have influenced variation in IBR between hospitals.