Methods to validate structured electronic health record data from emergency departments (Preprint)
JMIR Medical Informatics · 2026-08-27
DOI: 10.2196/110624 ↗At a glance
Journal article in JMIR Medical Informatics, published online on 2026-08-27. Written by Erika M. Brown and 15 co-authors. 26 references.
Abstract (original language)
BACKGROUND Electronic health records (EHRs) can help emergency departments conduct disease surveillance, evaluate and improve patient care, as well as accelerate health services research and system learning. However, emergency providers face unique workflow and time constraints that make documentation prone to data errors, inconsistencies, and omissions. Comprehensive validation is required to ensure resulting data is suitable for research use. OBJECTIVE We applied six data quality dimensions to evaluate and refine electronically extracted emergency department EHR data. METHODS We extracted EHR data from twelve British Columbia hospitals from July 2025 to March 2026. We used computational and manual assessments to evaluate the conformance, completeness, plausibility, correctness, concordance, and currency of 37 key variables. Computational assessments measured variable standardization, unexpected missingness, distribution, and relationships between related variables. Research assistants completed independent and blinded manual record reviews to create the criterion standard in which electronically extracted variables were evaluated using kappa scores, correlation coefficients, and mean similarity rates. Physicians provided consultations throughout. RESULTS We verified the data of 8,323 patients who made 34,035 emergency department visits. We identified 35 data issues total from all assessments; 18 related to the source data and 17 related to data extraction. Three variables had poor coverage (missingness >80%) and were omitted from the dataset. Timestamps did not always reflect the sequence of care. Obvious data entry errors were rare ( 95%) agreement with manually extracted variables. CONCLUSIONS Iterative data quality assessments in six data quality dimensions allowed us to improve the extraction process and provided an understanding of data accuracy and limitations. Manual chart reviews and physician consultations were critical for this process and identified important limitations of the extracted data. Future standardized assessment frameworks should account for medical specialty, workflow, and documentation practices. CLINICALTRIAL N/A
Cite
APA
Brown, E. M., Oh, R. D., Yeom, D. S., Niosco, C., Ho, V., Hau, J. P., Randhawa, L., Wang, J., Mah, C., Pavlovich, S., Lau, E., Cragg, A., Pilarinos, A., Matthews, C., Rosychuk, R. J., & Hohl, C. M. (2026). Methods to validate structured electronic health record data from emergency departments (Preprint). JMIR Medical Informatics. https://doi.org/10.2196/110624
Vancouver
Brown EM, Oh RD, Yeom DS, Niosco C, Ho V, Hau JP, et al. Methods to validate structured electronic health record data from emergency departments (Preprint). JMIR Medical Informatics. 2026. doi:10.2196/110624
BibTeX
@article{Brown2026,
author = {Brown, Erika M. and Oh, Rachel Da-Eun and Yeom, David Seonguk and Niosco, Chris and Ho, Vi and Hau, Jeffrey P. and Randhawa, Livleen and Wang, Justin and Mah, Christopher and Pavlovich, Sasha and Lau, Erica and Cragg, Amber and Pilarinos, Andreas and Matthews, Chenoa and Rosychuk, Rhonda J. and Hohl, Corinne M.},
title = {Methods to validate structured electronic health record data from emergency departments (Preprint)},
journal = {JMIR Medical Informatics},
year = {2026},
doi = {10.2196/110624}
}
EUVIMED