OBJECTIVES: The aim of the study was to identify real-world secondary data sources in oncology in China and assess their strengths and limitations for conducting observational research.

METHODS: A data landscaping exercise was conducted to identify existing oncology data sources which can be used to assess patient profile, treatment, and clinical outcomes of patients with cancer in China Mainland. Specifically, a targeted literature review of studies supported with supplementary web-based searches were performed. Searches of studies published between 2013 and 2023 were conducted in both English and Chinese search engines (i.e., PubMed, Wanfang Data, and CNKI). The strengths and limitations of the content and access of each identified data source were assessed based on publicly available information and direct contacts with data custodians. Data sources including data for only one tumor type were excluded.

RESULTS: Thirty-four real-world oncology data sources containing individual level patient data were identified and assessed. Seven of them curated data from multiple geographic regions in China: the National Cancer Registry, EMR data platforms of patients with cancer (e.g., National Anti-Tumor Drug Surveillance System and LinkDoc), and claims (e.g., China Health Insurance Association [CHIRA] data). The rest were regional data sources, including regional cancer registries, claims, and local EMRs (e.g., Tianjin city-wide EMR database). National oncology data sources generally had good representativeness of Chinese population, but some regional data had additional benefits such as longer follow-up time and more data points. The main limitation was that clinical outcome data like tumor response and community death were not routinely recorded. Access to such data for commercial use often requires administrative, ethical and/or regulatory approvals, and collaboration with academic or clinical institutes.

CONCLUSIONS: Real-world oncology data is emerging in China. Understanding the strength, limitations and practical requirements of each data source is crucial for appropriate study design and execution.