Annelotte van Bommel

169 General discussion and future perspectives institutions that adhere to a predefined quality level. The NBCA has proven its merit by objectifying improvements in the complete tumor excision rates of breast conserving surgery and in radiological and pathological work-up, which underscore the current quality of breast cancer care. 11 Furthermore, a more consistent use of radiotherapy boost was objectivated following publication of national guidelines 12 as well as a decrease in axillary surgery since 2011. 13 With respect to the identification of true outliers, reality is rather unruly. First, the NBCA usually publishes its annual quality indicator results in funnel plots that depict the indicator for every individual hospital with a certain patient volume within a funnel of 95% confidence intervals corresponding to the number of patients. In a country with approximately 90 hospitals, it is a statistical certainty that two or three hospitals will have divergent results suggesting underperformance, while their results are based on mere coincidence. To overcome this risk of incorrectly identifying underperformers as well as erroneously pointing out best practices, annual benchmark results can be merged by presenting institutional results over a longer time period. This results in tapering of the confidence interval which will enhance interpretation of an institution’s audit result. Second, the funnel shaped confident intervals remain at times difficult to explain in particular to the external stakeholders. Figure 2 shows the quality indicator “percentage of standardized pathology reports for patients with invasive breast cancer”. Hospital A (98%) has an unequivocal satisfactory outcome compared to the standard of 90%. Hospital B (83%) is below the standard, but still within the corresponding 95% confidence interval. But for the general public it is less evident that the performance of hospital D (73%) is worse than the performance of hospital C (68%). The provision of quality indicators with funnel plots gives a comprehensive view compared to percentages alone, but also demonstrate that the number of patients should be sufficient. A standardized pathology report addresses estrogen receptor-, progesterone receptor-, and HER2-status, malignancy grade, tumor size, margin involvement and number of positive lymph nodes (when sentinel node procedure or axillary node dissection was performed) 9