Quantitative performance of E-Scribe warehouse in detecting quality issues with digital annotated ECG data from...

Area of Science:

• Cardiology
• Pharmacology
• Regulatory Science

Background:

• The US Food and Drug Administration (FDA) requires digital electrocardiograms (ECGs) in HL7 XML format for new drug applications.
• ECG quality is crucial for accurate preapproval review, assessed by the FDA's ECG warehouse.
• The E-Scribe Warehouse is a commercially available component of the FDA's system.

Purpose of the Study:

• To evaluate the performance of E-Scribe Warehouse algorithms in assessing ECG quality.
• To quantify ECG acquisition quality, adherence to QT annotation protocols, and T-wave signal strength.
• To determine the reliability of E-Scribe Warehouse in identifying quality issues impacting QTc assessment.

Main Methods:

• Two datasets were used: a "reference" set with manually adjusted QT annotations and a "test" set with introduced quality issues.
• E-Scribe Warehouse algorithms were used to quantify ECG acquisition quality, QT annotation adherence, and T-wave signal strength.
• Performance was evaluated by comparing the algorithm's detection of introduced errors against expected outcomes.

Main Results:

• The E-Scribe Warehouse accurately detected differences between the reference and test datasets, reflecting the introduced quality issues.
• It correctly identified ECGs with questionable T-wave signal strength when scores fell below a threshold.
• The system demonstrated satisfactory performance in identifying ECG quality problems affecting QTc assessment reliability.

Conclusions:

• The E-Scribe Warehouse shows satisfactory performance in evaluating digital ECG quality for regulatory submissions.
• Its algorithms can reliably detect issues that may compromise the accuracy of QTc interval measurements in clinical trials.
• This tool supports the FDA's goal of ensuring high-quality ECG data for drug review.