Founded in 2016, the Clinical Pathways Forum is a community of pathways professionals now totaling over 12 institutions from across the United States and Canada who are utilizing clinical pathways in their practices and institutions to improve cancer care. Forum leader Mishellene McKinney, MHA, RN, OCN, organizes quarterly conference calls with Forum members to facilitate discussion of shared experiences and lessons learned regarding pathway use as clinical pathways become more prevalent and evolve to meet the needs of value-based health care systems and reimbursement models.
The Clinical Pathways Forum is a community of pathways professionals founded in 2016. As of November 2018, the Forum has grown to include over 12 institutions from across the United States and Canada who are using clinical pathways in their practices and institutions to improve cancer care. The mission of the group is to facilitate a knowledge exchange for overcoming the challenges of developing, implementing, and measuring clinical pathways in order to demonstrate the value of standardizing clinical care. The main activity of the Forum is to schedule time quarterly for conference calls to share experiences and lessons learned using clinical pathways. In an effort to increase discussion and collaboration between other organizations using clinical pathways, the Forum publishes highlights from each of the Forum conference calls that occur throughout the year in the Journal of Clinical Pathways (JCP). This second installment summarizes the speakers and discussion from the January 2019 call. Health care professionals from across the continuum of care are encouraged to join in these collaborative discussions—Forum organizer information is located at the end of this article.
The January 2019 Clinical Pathways Forum Call
During the Clinical Pathways Forum call on January 15, 2019, we had the opportunity to speak with Ruchit Kumbhani, MPH, about the program that the University of Chicago (UC) Medicine has developed to support value-based care. As the director of the Cancer Network and Program Operations, Mr Kumbhani is responsible for business development and strategic planning as well as value-based care initiatives within UC’s cancer network, which includes the Oncology Care Model (OCM) and payer contracts. Also featured this Forum call was data scientist David Hughes, who discussed the experiences and challenges of developing a data science program to support clinical pathway metrics at Seattle Cancer Care Alliance (SCCA). Mr Hughes is the cofounder of the Clinical Pathways Forum, is the former Associate Director of Clinical Pathways at SCCA, and is currently a data scientist and machine learning engineer at SCCA.
Merging Clinical Pathways and Analytics at UC
Mr Kumbhani discussed how UC’s clinical pathways program partnered with their data and analytics team to develop dashboards and other analytic tools to improve operations and quality tracking. A strong partnership between oncologists, supportive care physicians, pharmacists, nurses, administration and analysts has helped to inform UC’s pathways and dashboards. UC is leading the way by looking beyond medical oncology pathways to help reduce the overall cost of care and improve outcomes for patients.
UC is an OCM participant and uses Via Oncology pathways for medical oncology. Via provides quarterly reports to UC on pathway usage and concordance are presented routinely at disease meetings. To establish oversight, UC schedules a biweekly improvement council wherein a dedicated analyst reviews the Cancer Service Line data and helps to build dashboards in Tableau, a data visualization program, to drive operational change and drive quality improvement.
While medical oncology pathways are an important component in the value-based care at UC, the Service Line felt it was important to go beyond just medical oncology treatment data. They began to create clinical pathways for indications such as malignant bowel obstructions and febrile neutropenia with the goal of decreasing length of stay and improving overall outcomes of care. These pathways were created in AgileMD—an electronic-health-record-integrated web and mobile application that gives clinicians access to guidelines, protocols, and scoring tools—and developed by multidisciplinary teams, including nurses, pharmacists, dieticians, medical oncologists, and surgeons. The internally developed pathways were created with corresponding order sets in EPIC for ease of use and adherence measurement, which is then reported in the dashboards.
Dashboard development. UC has a process to identify the denominator for patients who have a pathway diagnosis and then tracks how many of those patients the pathway was activated for. The focus has been on identifying opportunities for patients by disease site and determining how to best engage clinicians in that discussion. Mr Kumbhani and his team took all of the patients in their payer contracts, particularly OCM, married the cost data with the clinical and staging data with the patients, and then determined which patients are matched by stage, metastatic disease, molecular phenotype, and subtype. This data is then used to analyze variations above and below the target.
The patient risk score: identifying patients to proactively manage care. UC developed a patient risk score to proactively identify high-risk patients, so that resources could be immediately deployed to augment what the traditional care team has available. Through literature review and internal research, the team at UC determined what they felt were the key correlators to high-risk patients who may require a higher level of care management support. The framework was then tested and built into EPIC to provide a real-time ability for providers to look on a dashboard to see how their patients are stratified. The risk score provides insight into the variables that make their patients high risk, which prompts providers to consider care management enrollment.
The Value of a Data Science Platform for Clinical Pathways: Experiences & Challenges
Mr Hughes provided an overview of the SCCA Clinical Pathways Program and how data science functions within and supports their clinical pathways and reporting processes. SCCA has 27 internally developed pathways that cover 100% of medical oncology, 74% of radiation oncology, and 85% of surgical oncology practice. Using a development lifecycle that begins with evidence-based content development, workflow integration, reporting, and continues with ongoing maintenance, pathways are created by multidisciplinary workgroup teams who identify the most effective and least toxic treatments based on National Comprehensive Cancer Network guidelines. Pathways are initially built in Visio and imported into LucidChart, which produces a graph API. The Graph API is then parsed and loaded into Neo4j GraphDB.
SCCA uses dashboards to report pathways coverage and concordance to the disease teams and administrators. The denominator for coverage is defined as any analytic case with a pathway diagnosis. The dashboard reports coverage metrics at the patient, disease group, and pathway levels and reports concordance metrics at the patient, provider, disease group, and pathway levels.
Generating “data epiphanies” from complex pathways data. The data science team at SCCA is dedicated to surfacing clinical intelligence derived from innovative reasoning about complex data. Through the use of Natural Language Processing (NLP) and predictive modelling, the data science team is able to generate “data epiphanies” about clinical pathways. SCCA uses OncNLP, which is an oncology-specific data science platform for analyzing large amounts of human-generated natural language. Compared with a human abstractor reviewing over 1700 elements such as pathology reports and lab data from 625 patients in 59.8 hours, data science tools like NLP can read and analyze the same data in about 5 minutes.
Using the metaphor that pathways are a program and pathways and physicians are programmers, the SCCA team is working to create a programming language and environment for pathways. In addition, the pathway data and logic flow can be graphically modelled to create a representation of clinical intelligence that can illustrate the various treatment points along a patient’s care trajectory (Image 1). Using tools such as ImmunoSEQ, the data science team at SCCA has enabled data epiphanies regarding the relationships between t cells, their encoded amino acids, and the corresponding gene families, genes, and alleles. They have also created complex algorithms based upon clinical trial eligibility criteria to help match patients to the most appropriate clinical trials.
