Time Is Money: Considerations for Measuring the Radiological Reading Time

Area of Science:

• Radiology and Medical Imaging
• Health Informatics
• Artificial Intelligence in Healthcare

Background:

• Radiology Information System (RIS) timestamps offer valuable data for workflow analysis.
• Current focus on artificial intelligence (AI) applications highlights the need for precise performance metrics.
• Understanding dictation time is crucial for operational efficiency and resource allocation.

Purpose of the Study:

• To evaluate timestamp-based radiological dictation time using historical RIS data.
• To introduce and apply timestamp modeling techniques for data analysis.
• To compare timestamp-derived dictation times with prospectively measured reporting times.

Main Methods:

• Analysis of 108,310 RIS timestamps from May 2010 to January 2021.
• Simulation techniques (1000 iterations) to minimize contextual outliers and model reporting times.
• Prospective measurement of 329 reporting times by radiologists with varying experience levels.

Main Results:

• A mean dictation time of 16.62 minutes was calculated from 106,127 included exams.
• Specific mean dictation times varied by body region (e.g., Head CT: 16.05 min, Abdominal CT: 17.92 min, CT Foot: 10.96 min).
• No significant difference was found between timestamp-derived and prospectively measured reporting times.

Conclusions:

• Timestamp analysis is a reliable method for assessing current radiological reporting practices.
• Body region and radiologist experience are significant confounders affecting dictation time.
• This methodology can support cost-benefit analyses for workflow optimizations, including AI integration.