Pancreas Cancer Learning Health Network Established to Share Best Practice Across 14 Centers and Improve Patient Outcomes

Abstract

Pancreas cancer (PC) survival is among the lowest of all malignancies. Clinical trials have failed to significantly improve outcomes. Individual and institutional biases in care result in significant variation in practice, further hindering progress. Learning health networks (LHNs) prospectively collect real world data across centers and test improvements that can rapidly be expanded across centers if deemed successful. Herein, we report preliminary progress from the Pancreas Cancer Canopy Cancer Collective (PC-CCC), the first oncology LHN, established to improve duration and quality of survival in PC. In 2019, we established the PC-CCC with six care centers who engaged in a collaborative design process to create a set of improvement aims, change ideas, and outcome measures. Center team members receive training and coaching in collaborative quality improvement methods, applied to local improvement efforts. Eight more centers joined in 2021, and a shared Canopy outcomes database was built and implemented to inform center-specific and network-wide improvement efforts and allow the LHN to undertake research using real-world data. Current improvement efforts are focused on proactively screening new PC patients for: (1) Clinical trials, (2) pancreas enzymes, (3) palliative care needs, and (4) goals of care conversations. Currently, 14 care centers are active participants in the PC-CCC LHN. Data on a total of 2,002 PC patients are available to date. At presentation to the care center, most patients are female (51%) and have biopsy proven PC (83.9%). Average age is 68 years, and presenting disease status is metastatic (14.5%), resectable (11.4%), locally advanced (10.9%), borderline resectable (8.1%), or not yet staged (40%). For those who received radiation, 75.8% received stereotactic body radiation therapy. Among patients whose chemotherapy regimen was documented, most received 5-fluorouracilbased treatment (52%). Descriptive follow up data (including treatment and outcomes) are being actively updated, to be reported at time of presentation. Creation of a cancer LHN for PC is feasible and has set the stage for improving patient and provider outcomes through iterative community-building, continuous improvement, and sharing of data and multidisciplinary best practices. Additionally, the data obtained from the CCC database can rapidly inform the network how variation in clinical practice across centers can influence outcomes.