Improvements to PTSD quality metrics with natural language processing.

Area of Science:

• Health Informatics
• Clinical Quality Measurement
• Natural Language Processing

Background:

• Healthcare quality measurement increasingly relies on structured electronic medical record (EMR) data.
• Structured data capture tools may not be feasible or acceptable in all clinical settings.
• Natural language processing (NLP) offers an alternative for extracting quality data from clinical notes.

Purpose of the Study:

• To assess the quality of care for posttraumatic stress disorder (PTSD) using structured EMR data alone.
• To evaluate the impact of supplementing structured EMR data with NLP-derived information on quality measurement.
• To improve the completeness of quality measurement in real-world clinical practice.

Main Methods:

• A retrospective analysis of 2,098,389 US Department of Veterans Affairs patients diagnosed with PTSD between 2000 and 2019.
• Measurement of evidence-based psychotherapy (EBP) delivery and measurement-based care (MBC) using structured EMR data.
• Recalculation of quality metrics incorporating NLP analysis of clinical note text.

Main Results:

• Using structured EMR data (2015-2019), 3.2% of eligible PTSD patients received EBP, and 48.1% received MBC.
• Supplementing with NLP-derived data increased EBP delivery estimates to 4.1% and MBC to 58.0%.
• NLP significantly enhanced the completeness of quality metrics for PTSD care.

Conclusions:

• Supplementing structured EMR data with NLP-derived data substantially improves healthcare quality measurement.
• NLP can bridge documentation gaps when structured data capture tools are unavailable or have barriers.
• This combined approach enhances the accuracy and completeness of quality assessment in clinical practice.