Related Experiment Video
Updated: Aug 7, 2026
07:31
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Temporal Variability in Diagnosis Code Distributions Across Extraction Time Points in a Multicenter Integrated EHR
Kyunghee Lee1, Chihiro Kumagai2,3, Masato Komuro2,3
1Japan Institute for Health Security, 1-12-1 Toyama, Shinjuku-ku, Tokyo, 162-8655, Japan. ri.k@jihs.go.jp.
Journal of Medical Systems
|August 5, 2026
Summary
Electronic Health Record (EHR) data show temporal variability even when extracted at different times for the same period. Documenting extraction timing is crucial for clinical research accuracy.
Area of Science:
- Health Informatics
- Clinical Research Methodology
- Data Science
Background:
- Electronic Health Record (EHR) data are vital for clinical research.
- The temporal stability and variability of EHR data over time remain underexplored.
- Understanding data changes is essential for reliable research outcomes.
Purpose of the Study:
- To assess the temporal variability in EHR diagnostic data based on different extraction time points.
- To quantify changes in diagnosis code distributions and record counts.
- To highlight the impact of extraction timing on EHR data consistency.
Main Methods:
- Retrospective, descriptive, observational study using multicenter EHR data from Japan.
- Diagnosis records from 2022 were extracted at three time points: Oct 2023, Oct 2024, and Oct 2025.
- International Classification of Diseases, 10th Revision (ICD-10) codes were analyzed using Jensen-Shannon distance (JSD) and record counts.
Main Results:
- Minimal variation in total records and unique patients across extraction times.
- Non-zero JSD values indicated differing ICD-10 code distributions across extraction time points.
- Facility C showed significantly higher JSD compared to other facilities.
- H61 code had the largest absolute change; acute conditions decreased while chronic conditions increased.
Conclusions:
- EHR data exhibit temporal variability influenced by extraction time points, even for the same target period.
- Variability differs across healthcare facilities.
- Documenting extraction timing and dataset version is critical; study designs should account for temporal data variability.