Leveraging real-world data sources for clinical oncology research: A review of studies published in ASCO publications

We used the ASCO Editorial Manager to identify original reports published between January 2018 to December 2022 that leveraged commonly used RWD in the US. We included all journals in ASCO Publications (Journal of Clinical Oncology [JCO], Precision Oncology [PO], Oncology Practice [OP], Clinical Cancer Informatics [CCI], and Global Oncology [GO]). The RWD assessed were SEER, Medicare, NCDB, Flatiron, CancerLinQ, IQVIA, Optum, MarketScan, MEPS, and NHANES. The data were identified by specifically referencing the source within the abstract. Five independent, trained reviewers abstracted clinical, epidemiologic, and reporting data from the manuscripts. For this study, we focused analysis on STROBE methodology reporting criteria (items 4-12).

One hundred and seventy-four publications were included in our study (57% OP, 18% CCI, 14% JCO, 5.8% PO, 5.2% GO). The most common RWD used were Medicare (49%), SEER (40%), and NCDB (14%). Nearly one-quarter of the publications used more than one of the databases evaluated, and one-third included a secondary RWD source and/or linked patients with an external data source (e.g., Google Maps, electronic health records). The most common malignancy types evaluated in the manuscripts were breast (33%), lung (25%) and colorectal (21%). Study design, setting, and study variables had high reporting (>90%), while descriptions about study size derivation, the rationale for quantitative analyses, and the approach to missing data were less well reported (<78%).

RWD are published within ASCO Publications and are commonly combined with multiple data sources. Though some STROBE methodologic reporting criteria are regularly addressed, we identified criteria that were not well-reported. Further research is needed to assess changes in the use of RWD and whether lower reporting of methodologic information may impact the replicability and interpretation of study findings.