Abstract
Background: As the availability of Electronic Health Records (EHR) has become more ubiquitous, data derived from the EHR has increased in parallel. Data retrieval was originally limited by the need for manual extraction; today, many vendors offer ways to quickly extract large volumes of electronically recorded data, opening doors to the assessment of large study populations. In 2019, we identified where errors can exist within data electronically extracted from the chart and identified several approaches that investigators should address in order to validate a subset of extracted variables and to determine their fitness for research inclusion. A review of twenty recently published manuscripts of retrospective research utilizing electronically extracted health record data showed only two (10%) documented data validation procedures by the researchers in the methods section, and only one (5%) documented data quality measures used by the large, federated, de-identified database from which their data was retrieved. We present an updated review of current practices in the use of the EHR for research, highlighting emerging challenges and the importance of transparent reporting of data validation and quality assessment methods.
Conclusion: The expansion of ICD codes and creation of large, federated databases have added additional complexities to the use of EHR for research purposes. Validation of electronically extracted data from health records should be required and the details of the validation processes clearly documented in published methods. When large de-identified databases are utilized, the platform’s intrinsic data-quality assessment processes should be identified and reported. Currently, few manuscripts address these practices.
Keywords
Hypoglycemia, Gestational diabetes, Diabetic complications, Data, Validation, Research, Extraction, Electronic health records