Assessing the Ability of Hospital Sites From a European Real-World Network to Support Oncology Evidence Generation

OBJECTIVES: We conducted a series of data assessments across multiple tumor types (endometrial, gastric, ovarian and breast) to check feasibility of real-world (RW) data to support evidence generation. This is the first step in setting up an evidence platform that could efficiently generate steady streams of RW insights across these tumor types.

METHODS: Hospital sites from an IQVIA RW network were invited to participate in this data assessment to ascertain in-depth information on patient counts, and data availability, quality and access.

RESULTS: Data assessment was completed at 16 sites in France, Germany, Israel, Italy, Portugal, Spain and the UK representing a total 13,895 endometrial, 8,681 gastric, 7,283 ovarian, and 71,905 breast cancer patients from 2018.

Data on tumor characteristics (e.g., stage, ECOG, histology), treatment (start/end dates) and outcomes were captured consistently across sites. Data on established biomarkers (e.g., HER2) were captured, however emerging biomarkers not currently included in clinical guidelines were poorly captured.

Some country level differences were observed (e.g., PD-L1 appears to be tested more widely in France while Ki-67 is less commonly tested for breast cancer in the UK). Differences in data format were also observed with France, Germany, and UK sites having a greater proportion of data in a structured format (i.e., in database or electronic medical record) compared to other countries.

CONCLUSIONS: Our analysis highlights the importance of conducting fit-for-purpose data assessments to assess data availability and quality prior to the design of RW evidence studies. It also emphasizes the need to consider RW clinical practices particularly relating to biomarker testing for precision medicine during study design (tissue sample re-testing may be necessary if a novel biomarker is to be investigated).

Moreover, the data format should also be considered in the set-up of RW studies, given its implications on cohort identification approach, data accuracy and site resourcing requirements.