Importance of quality control in ‘big data’: implications for statistical inference of electronic health records in clinical cardiology

Abstract

The implementation of clinical information systems across healthcare, and recent advances in technology have created unprecedented opportunities to study population health, basic science, and the effects of medical care at scale. Indeed, everyday healthcare comprises many natural experiments, through which we can explore the real-world evidence of clinical practice. In clinical settings, such information is captured within electronic health records (EHRs), which are recorded routinely as part of medical care and frequently form an essential part of care delivery. The secondary use of EHRs for research purposes has proliferated in recent years. In the context of clinical cardiology, disease/procedurespecific national registries prospectively collect patient-level data, which is advantageous from both a clinical perspective (through audit and feedback of performance) and a research perspective (through understanding epidemiological results at scale). Equally, basic science research has benefitted from the large-scale data arising from advances in imaging technology, in silico and stem cell models, and genome-wide association studies. However, the secondary use of large-scale data for research purposes is not without challenges. In this commentary article, we discuss some of the opportunities of using large-scale observational data for clinical cardiology, the importance of data quality in this context, and the implications data quality might have on downstream analyses and clinical applications.