OBJECTIVES: Due to variations in treatment patterns and patient characteristics, practice setting has been considered a confounder in oncology real world evidence (RWE). A standard methodology for classifying oncology practices in the United States (US) to adequately adjust for confounding has not been established. Therefore, we evaluated classification schemes from RWE studies.

METHODS: Schemes for classifying oncology practice settings by academic status were identified and evaluated.

RESULTS: Schemes varied in complexity and design. The American Society for Clinical Oncology defined three categories: “academic”, “health system owned”, and “physician owned”. The Commission on Cancer (CoC) designated 10 types of cancer programs, with the four most common being “Comprehensive Community” (38% of facilities), “Community” (26%), “Integrated Network” (14%), and “Academic Comprehensive” (13%). The CoC scheme also included categories with National Cancer Institute (NCI) designation: “NCI-Comprehensive” (3%) and “NCI-Network” (1%). In contrast, researchers often applied simplified schemes, such as a binary variable for “academic” vs. “community”. The “academic” category was occasionally further segregated into those facilities with and without the NCI designation. Integrated health systems were usually classified as “community” but in some studies were considered “academic” or retained as a third category. “Community” cancer practices are heterogeneous with respect to clinical trial activity, patient volume, geography, and access to multidisciplinary care. Some “community” sites offer advanced diagnosis and treatment modalities including on-site next generation sequencing and cellular therapy administration. Recent trends in expanding academic affiliations has likewise obscured the definition of “academic” sites. Limited guidance exists for researchers to classify facilities in light of the complicated and evolving oncology ecosystem in the US.

CONCLUSIONS: Classifying practice settings is nuanced. RWE scientists should ensure that their study cohorts reflect the populations of interest. Further research is needed to understand how factors that vary by practice setting might influence treatment patterns and outcomes.