Tracking persistent postoperative opioid use: a proof-of-concept study demonstrating a use case for natural language...

Area of Science:

• Medical Informatics
• Clinical Data Analysis
• Natural Language Processing

Background:

• Large language models (LLMs) and natural language processing (NLP) are increasingly relevant in healthcare.
• Identifying persistent postoperative opioid use is crucial for patient care and addressing the opioid crisis.
• Unstructured clinical text contains valuable data for understanding opioid use patterns.

Purpose of the Study:

• To demonstrate the capability of an NLP engine to identify persistent postoperative opioid use.
• To evaluate the accuracy of NLP in analyzing clinical notes for opioid use status after major spine surgery.

Main Methods:

• Extracted clinical documents from patients undergoing major spine surgery (July 2015-August 2021).
• Defined persistent postoperative opioid use as opioid use >= 3 months post-surgery.
• Applied an NLP engine to outpatient follow-up notes and compared results with manual clinician review.

Main Results:

• The study included 965 patients; 73.1% exhibited persistent opioid use.
• The NLP engine achieved 92.9% overall accuracy in determining opioid use status.
• The NLP engine correctly identified persistent opioid use in 95.6% and no persistent use in 86.1% of cases.

Conclusions:

• NLP is a viable tool for extracting and analyzing unstructured clinical data to identify persistent postoperative opioid use.
• Leveraging NLP can provide insights into the opioid crisis and enhance patient-level care.
• Further research is needed to integrate NLP into clinical decision support systems within healthcare settings.