Is it possible to automatically assess pretreatment digital rectal examination documentation using natural language...

Area of Science:

• Medical Informatics
• Oncology
• Health Services Research

Background:

• Provider-documented pretreatment digital rectal examination (DRE) is a key quality metric in prostate cancer care.
• Assessing adherence to this metric often relies on manual chart review, which is time-consuming and resource-intensive.
• Electronic health records (EHRs) contain vast amounts of unstructured clinical notes that could be leveraged for automated quality assessment.

Purpose of the Study:

• To develop and validate a natural language processing (NLP) framework for the automated assessment of provider-documented pretreatment DRE quality.
• To evaluate the feasibility of using NLP to identify quality metric performance within routine clinical documentation.
• To identify patient and clinical factors associated with the performance of the DRE quality metric.

Main Methods:

• A cohort of 7215 prostate cancer patients was identified from an EHR-based data warehouse (2005-2017).
• A previously developed NLP pipeline was utilized to classify DRE assessment from clinical notes as documented, deferred, or refused.
• The study analyzed DRE documentation within 6 months prior to treatment initiation.

Main Results:

• Of 7215 patients, 82.6% had a DRE documented in the EHR. Only 51.9% had a DRE documented within 6 months before treatment.
• Patients with private insurance were more likely to meet the DRE quality metric compared to those with Medicaid or Medicare.
• Patients receiving chemotherapy, radiation therapy, or surgery as first-line treatment were more likely to have a timely DRE documented.

Conclusions:

• Natural language processing (NLP) provides a feasible and accurate method for automatically assessing quality metrics like pretreatment DRE documentation in EHRs.
• Automated assessment using NLP can be integrated into current clinician workflows, improving efficiency and data capture.
• EHR unstructured data, when processed with NLP, offers valuable insights for quality improvement initiatives in cancer care.