<p>EHR data are widely used in clinical research; however, their temporal stability remains poorly understood. This study aimed to evaluate temporal variability in EHR data arising from differences in extraction time points. We conducted a retrospective, descriptive, observational study using diagnostic data from a multicenter integrated EHR database in Japan. The target period was 2022, and diagnosis records with a start date within this period were extracted at three time points: October 2023, October 2024, and October 2025. Diagnoses were mapped to the three-character level of the International Classification of Diseases, 10th Revision (ICD-10). Changes in diagnosis code distributions were quantified using the Jensen–Shannon distance (JSD). Changes in record counts per ICD-10 code between 2023 and 2025 were also evaluated. Five facilities were included in the analysis. In the integrated dataset, variations in total records and unique patients were minimal (total records: +0.06% in 2024 and + 0.05% in 2025; unique patients: +0.06% in 2024 and + 0.02% in 2025). JSD values for ICD-10 distributions were non-zero across all facilities, indicating that diagnosis code distributions differed across extraction time points even for the same target period. Across all facilities excluding Facility C, JSD was 0.0022 in 2024 and 0.0026 in 2025 relative to the 2023 baseline. In contrast, Facility C exhibited a markedly larger JSD in 2024 (0.0310), which persisted at a similar level in 2025 (0.0317). At the code level, H61 showed the largest absolute change (− 266 records; −19.98%). Overall, acute conditions tended to decrease, whereas chronic conditions increased across extraction time points. EHR data exhibit temporal variability across extraction time points, even for the same target period. These findings highlight the importance of documenting extraction timing and dataset version, and suggest the need for study designs that explicitly account for such variability.</p>