Assessing data quality for Healthcare Analytics: A tiered approach to secondary use of EHR data
There is no recording for this session. The presenters have opted to provide the slides for attendees. These can be downloaded below in the Session Files section - UCTech 2019 Planning Committee
As the use and availability of Electronic Health Record system have expanded so have opportunities to reuse the information they collect to improve clinical outcomes, reduce costs, improve health system operations and support a variety of research activities.
Although the opportunities for secondary reuse of EHR are immense, many questions quickly emerge concerning the quality of the data. There is often a mismatch between the data quality required when originally collecting data and the data quality needed later after data has been collected when critical analytic questions arise.
Some specific factors are identified; First, heterogeneous, complex and dynamic clinical workflows often create complex and messy data. Second, EHR systems are complex dynamic systems that evolve through time, thus the data often change through time. Third, concepts that analysts want to represent in their models are often represented by in numerous and incomplete database fields. Finally, these and other data issues are often undocumented—it is only by laborious analytical work and clinical ethnography that analysts can obtain the highest quality of data.
Many analytical teams are not fully aware of the challenges associated with identifying and repairing data quality issues. To achieve better project outcomes, we discuss a framework to categorize projects into data quality tiers as well as a process for validating data for re-use.
Tis session is intended for those interested in the challenges that aare encountered when extending the use of health data beyond the clinic. Familiarity with EHR systems, research questions and analytic methods is helpful but not required.