A Data-Driven Analytical Framework To Learn And Improve Clinical Workflow In Radiation Oncology

Abstract

Radiation treatment planning is a complex process with multiple, dependent steps involving an interdisciplinary patient care team. We have previously implemented an interactive, web-based dashboard, which integrates with the departmental electronic medical record system (EMR), and provides real-time monitoring and visualization of the treatment planning workflow and resource utilization. In addition, the dashboard implements a standardized, integrated framework to acquire and analyze data for clinical workflow evaluation. We present this framework and the results of data-driven analyses to learn and evaluate clinical practice performance in an effort to improve clinical efficiency and assess the impact of potential workflow changes. Process maps and flowcharts were created to model the treatment planning workflow. The flowcharts described: (1) standardized activities associated with a patient’s care path for distinct treatment modalities from the time of CT simulation to treatment, (2) activity order and timeline, (3) activity ownership, and (4) staff interaction. Three treatment modalities, 3D, intensity modulated radiotherapy (IMRT), and stereotactic body radiotherapy (SBRT), were modeled. For each modality, an ideal timeline with a duration of 6 days for 3D and IMRT and 9 days for SBRT, and amenable to future refinement, was formulated based on perceived timelines, staffing levels and departmental throughput goals. Care team physicians, dosimetrists, physicists and therapists were trained in the use of the EMR to record activity status, which was also visualizable on the web-based dashboard. A quantitative representation of how different care path activities unfold was obtained from queryable timestamps in the EMR. Staff-reported actual completion times for the care path activities were computed and compared to their ideal timeline. We collected and analyzed data for 133 new patient treatments (72 3D, 55 IMRT, 6 SBRT) and 793 associated care path activities. The mean (standard deviation) of reported actual completion times in working days for the main activities are listed in the table below. On average, 3D, IMRT and SBRT treatments required 4.9, 6.6 and 9.2 days, respectively, from CT simulation to treatment. Actual total completion times for 3D treatments were significantly better than ideal (p < 0.05), but not for IMRT and SBRT treatments. The framework developed allows for informed, data-driven decisions regarding clinical workflow management and the impact of changes on existing workflow as we seek to optimize clinical efficiency and incorporate new interventions due to new treatments into clinical practice.Abstract 3799; TableCT importPhysician contour/beam placementTreatment planningPhysician plan approvalIMRT QAPre-treatment physics chart review3D0.1 (0.2)0.9 (1.0)4.2 (2.3)4.0 (2.3)4.9 (2.7)IMRT0.1 (0.3)1.8 (1.1)6.3 (2.2)5.8 (2.3)6.5 (2.3)6.6 (2.1)SBRT0.03 (0.02)2.6 (2.3)3.0 (2.2)8.7 (2.5)7.3 (2.5)9.2 (2.7) Open table in a new tab