Wirsch, A, Mungmode, A, Dawson, J, Rioles, N, Ebekozien, O

Introduction: The electronic medical record (EMR) is rich in valuable information, however extracting the data for real world insights is challenging and difficult for comparison studies. This study’s aim is to identify some of the common challenges and perspectives from 18 data teams participating in the T1D Exchange Quality Improvement Collaborative (T1DX-QI).

Method: T1DX-QI centers share patient-level, de-identified, individual-level EMR data with the coordinating center for real world studies, benchmarking, and quality improvement studies. Centers’ data teams liaise with T1DX-QI to map and extract over 120 pre-specified T1DX-QI data fields. T1DX-QI invited 22 centers that have recently completed the mapping process to participate in this assessment. Data teams responded to an online survey and participated in focus groups. T1DX-QI’s coordinating center reviewed data samples to identify common errors with significant deviation from the expected variable.

Results: Most centers (18 out of 22,82% response rate) participated in the study. The majority (61%) of the data teams conduct 6 or more EMR data extraction projects annually. Nearly half (45%) of centers reported that the diabetes data extraction and mapping process was about the same level of effort as other disease area extraction process. Most of these experienced centers thought the T1DX-QI mapping process was efficient (73%). The most common errors from the extraction process were incorrect values, incomplete documentation, missing variables, and incorrect formats. All errors were readily addressed in iterative review cycles and by implementing a comprehensive best practice checklist.

Conclusion: EMR data extraction for diabetes quality improvement and population health collaboration is feasible and achievable, as demonstrated by the T1DX-QI network.

Click here to read the full abstract