Evaluation of the breast cancer care network within the Lazio Region (Central Italy).

Background And Objectives: A summary indicator for evaluating the breast cancer network has never been measured at the regional level. The aim is to design treemaps providing a summary description of hospitals (including breast units) and Local Health Units (LHUs) in terms of their levels of performance within the breast cancer network of the Lazio region (central Italy). The treemap structure has an intuitive design and displays information from both general and specific analyses.

Methods: Patients admitted to the regional hospitals for malignant breast cancer (MBC) surgery in 2010-2017 were selected in a population-based cohort study. These quality indicators were calculated based on the international guidelines (EUSOMA, ESMO) to assess the performance in terms of volume of activity, surgery procedure, post-surgery assistance and timeliness of medical therapy or radiotherapy beginning. The quality indicators were calculated using administrative health data systematically collected at the regional level and were included in the treemap to represent the surgery or the post-surgery areas of the breast cancer clinical pathway. In order to allow aggregation of scores for different indicators belonging to the same clinical area, up to five evaluation classes were defined using the "Jenks Natural Breaks" algorithm. A score and a colour were assigned to each clinical area based on the ranking of the indicators involved. The analyses were performed on an annual basis, by the LHU of residence and by the hospital which performed the surgical intervention.

Results: In 2017, 6218 surgical interventions for MBC were performed in the hospitals of Lazio. The results showed a continuous increase of the level of performance over the years. Hospitals showed higher variability in the levels of performance than the LHUs. 36% of the evaluated hospitals reached a high level of performance. An audit of the S. Filippo Neri breast unit revealed incorrect coding of the input data. For this reason, the score for the indicator for the volume of wards was re-calculated and re-evaluated, with a subsequent improvement of the level of performance. Most LHUs achieved at least an average overall level of performance, with 20% of the LHUs reaching a high level of performance.

Conclusions: This is the first attempt to apply the treemap logic to a single clinical network, in order to obtain a summary indicator for the evaluation of the breast cancer care network. Our results supply decision makers with a transparent instrument of governance for heterogeneous users, directing efforts improving and promoting equity of care. The treemaps could be reproduced and adapted for other local contexts, in order to limit inappropriateness and ensure uniform levels of breast cancer care within local areas. The next step is the evaluation of audit and feedback interventions to improve the quality of care and to guarantee homogeneous levels of care throughout the region.