Prevalence of Sensitive Terms in Clinical Notes Using Natural Language Processing Techniques: Observational Study.

Area of Science:

• Health Informatics
• Medical Privacy
• Clinical Documentation

Background:

• The 21st Century Cures Act mandates increased electronic health information sharing, raising concerns about patient confidentiality.
• The prevalence and characteristics of sensitive terms within clinical notes remain largely unknown.

Purpose of the Study:

• To define sensitive terms indicative of private patient information.
• To determine the prevalence and characteristics of provider notes containing these sensitive terms.

Main Methods:

• A list of 781 sensitive terms was generated using keyword expansion.
• A comprehensive search of pediatric patient notes (ages 0-21) from 2019 was conducted for direct string matches.
• Prevalence and characteristics of notes with sensitive terms were analyzed across clinical encounters and patient demographics.

Main Results:

• Sensitive terms were found in notes across all pediatric age groups and clinical contexts.
• Mental health terms were most frequent overall, while substance abuse and reproductive health terms were common in very young children (0-3 years).
• History and physical notes, ambulatory progress notes, pain management, and child abuse clinics showed the highest prevalence of sensitive terms.

Conclusions:

• The presence of sensitive terms is not confined to adolescents, specific note types, or particular medical specialties.
• The widespread nature of sensitive terms across all pediatric ages and clinical settings presents significant challenges for protecting patient and caregiver privacy.
• Information-blocking regulations necessitate advanced strategies to safeguard sensitive clinical information.