Surveillance of Health Care-Associated Violence Using Natural Language Processing.

Area of Science:

• Healthcare management
• Clinical informatics
• Patient safety

Background:

• Healthcare-associated violence (HAV) is increasing, driven by patient and family outbursts.
• HAV episodes are often under-reported, hindering effective intervention.
• A behavioral health crisis exacerbates these reporting challenges.

Purpose of the Study:

• To assess the feasibility of using nursing notes to identify under-reported HAV episodes.
• To evaluate the effectiveness of natural language processing (NLP) in detecting HAV.
• To improve the completeness of HAV reporting in healthcare settings.

Main Methods:

• Nursing notes from two hospitals were extracted for analysis.
• A natural language processing (NLP) assisted manual review process was employed.
• NLP models were trained on data from a pediatric tertiary care center and validated on a community hospital's data.

Main Results:

• The NLP model achieved 96.8% sensitivity and 47.1% specificity in identifying HAV episodes at the community hospital.
• 31 HAV episodes were identified, with 26 deemed reportable.
• Only 7 of the 26 reportable cases were initially self-reported, highlighting significant under-reporting.

Conclusions:

• NLP-assisted review is a feasible and effective method for surveillance of under-reported HAV.
• This approach can enhance the completeness of HAV reporting.
• The method is implementable even in resource-limited hospital settings.