Figure 1.

The life cycle of secondary use of clinical data for research. It spans data generation, data transformation, data reuse, and optional data quality reporting and feedback provision to the data generation and data transformation processes. Data quality problems can be introduced at each phase in this life cycle. Knowledge of data quality (represented by the orange dotted lines) gets lost at each step along the way. Different stakeholders were only included at specific phases without much collaboration: e.g. clinicians are usually in the data generation step, software engineers or technicians are involved in data transformation processes without knowledge of intended uses of the data, and researchers are only involved at the data reuse phase without knowledge of data provenance. Heterogeneous data reporting occurs occasionally at the last step, with limited feedback loops established between the last step and the first two steps in the data life cycle.