ObjectivesTo establish an appropriate data governance mode in according with the database status of clinical study.MethodsForty-six doctors of different seniority with clinical research experience from six hospitals in Beijing were selected by stratified purposeful sampling and semi-structured interview and were used to understand the status and shortcomings of data acquisition and storage in clinical research. The data resource of current clinical studies were summarized and the main target of data governance and the characteristics of clinical study data were explored to establish the domains of clinical study data governance to construct the framework of clinical research data governance.ResultsCurrently, the data sources of clinical studies were diverse, including real-world data from various medical and health records, data collected independently for clinical studies and numerous other sources. However, since collecting the data from electronic medical records was difficult for numerous reasons, a large number of researchers still collected research data by hand writing and stored it insecurely. In addition, the combination of electronic information from multiple sources was difficult. Building ALCOA+CCEA standard clinical research data management system based on clinical research data governance was urgent. Data governance includes data architecture, data model, data standards, data quality, master data, timeliness management, metadata and data security, while life cycle management and data insight were not essential parts.ConclusionsBased on the real-world data resources, domains of data governance in clinical study should include data architecture, data model, data standards, data quality, master data, timeliness management, metadata and data security.
ObjectiveTo construct a strategy for classification of clinical research data security for real-world research, based on the features of clinical research data.MethodsBased on the laws, regulations, and data security classification method in relevant fields, the clinical research data was classified into five security levels. Then, the method was gradually perfected through three times of revisions, which followed the advice from experts who were experienced in many relative areas, such as clinical medicine, clinical research methodology, clinical research management, ethics, genetics and public health data application and management.ResultsExperts’ opinions gradually converged through several times of consultation. The clinical research data was finally classified into five security levels with explicit definition and security policy for each security level. Thirty-three data categories, which covered demographic information, clinical examination, diagnosis, treatment information, genetic information, health economics information, medical data and information on research processes that have been published, were included in the five security levels.ConclusionsSince there is an increasing trend of data scale and the data security classification and management are necessary to ensure the data security and appropriately utilization of data. The method of clinical research data classification proposed in this paper can provide beneficial references for the further improvement of data security in the future.