Discovery of data quality issues in electronic health records: profound consequences for critical care medicine applications – a systematized review
Authors' conclusion
Data quality issues in critical care EHRs create cascading effects that compromise ML model reliability, clinical decision-making, and patient safety.
Study context
- Population
- Studies of EHR data quality issues in critical care settingsstudies selected
- Participants
- 29 studies
- Outcome
- EHR data quality issues in critical care and their impact on machine learning (ML) model performance and patient safety
Methodological detailsexpert-based questions, 281 relevant studies identified, Searched MEDLINE, Embase, IEEE Xplore, ACM Digital Library, CINAHL, Google Scholar, DBLP, Web of Science, Cochrane Library
- Expert-based questions
- 281 relevant studies identified
- Searched MEDLINE, Embase, IEEE Xplore, ACM Digital Library, CINAHL, Google Scholar, DBLP, Web of Science, Cochrane Library
Main finding
As reported in the abstract
- Missing data rates for some variables
- >80%
- Predictive features that were missingness indicators rather than actual values
- 40%
- EHR-related medication errors among all ICU medication errors (one-third with life-threatening potential)
- 34%
- Copy-paste prevalence in residents' progress notes
- 82%
- Sepsis detection model AUC drop at external validation
- 0.76 to 0.63
- Vital sign quality degrading at % of average LOS
- 60-75%
Original publication
Discovery of data quality issues in electronic health records: profound consequences for critical care medicine applications – a systematized review. Crit Care. Published 8 January 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →