Transcription Error Rates in Retrospective Chart Reviews

Area of Science:

• Health Informatics
• Clinical Data Management
• Orthopedic Research

Background:

• Electronic health record (EHR) systems enhance access to structured clinical data.
• Manual chart review remains a primary method for data collection.
• The accuracy and efficiency of manual review compared to electronic queries are not well-established.

Purpose of the Study:

• To compare the accuracy and efficiency of manual chart review versus electronic data warehouse queries for retrieving clinical data.
• To evaluate the types and rates of errors associated with each data collection method.

Main Methods:

• A manual chart review of EHRs was conducted for 100 inpatient venous thromboembolic events post-total joint arthroplasty.
• A separate electronic data query was performed using the same criteria.
• Data sets were compared algorithmically to identify and categorize discrepancies (random vs. systematic errors).

Main Results:

• Manual review had an average transcription error rate of 9.19% per patient encounter and 11.04% per data variable.
• Electronic data query completion time was 58 seconds, compared to 915 minutes for manual review.
• Systematic errors constituted 7.41% of discrepancies, with random errors at 5.79% per encounter.

Conclusions:

• Manual chart review is prone to significant transcription and systematic errors, potentially impacting study validity.
• Computer-based data queries significantly enhance the speed, reliability, and reproducibility of clinical data retrieval.
• Adopting electronic queries facilitates more efficient, data-driven decision-making in healthcare settings.