From evidence-based guidelines to practice: Using natural language processing to identify impactful interventions in...

Area of Science:

• Gerontology
• Healthcare Quality Improvement
• Nursing Informatics

Background:

• Falls are a significant risk in long-term care (LTC), leading to injury and reduced quality of life.
• Effective implementation of evidence-based guidelines is essential for mitigating falls in LTC settings.
• The Registered Nurses' Association of Ontario (RNAO) provides a Best Practice Guideline (BPG) for falls prevention.

Purpose of the Study:

• To identify and characterize implementation strategies for falls prevention in LTC homes.
• To analyze the uptake of the RNAO's 'Preventing Falls and Reducing Injury from Falls (2017)' BPG.
• To leverage natural language processing (NLP) for analyzing implementation data.

Main Methods:

• A retrospective quality improvement study analyzed 153 narrative reports from 63 Ontario LTC homes (2017-2022).
• An NLP pipeline, including RAKE keyword extraction and sentiment analysis, processed free-text data.
• Themes were refined through expert consensus and inter-rater reliability assessed using Cohen's kappa.

Main Results:

• Ten key themes of implementation strategies were identified, including falls risk assessment, prevention strategies, and post-fall assessments.
• Commonly implemented strategies reflect significant adoption of BPG recommendations.
• Sentiment analysis revealed predominantly neutral to positive implementation experiences, with noted gaps in documentation and policy-level strategies.

Conclusions:

• A core set of falls prevention strategies are frequently implemented in LTC.
• Standardizing risk screening, strengthening post-fall evaluations, improving documentation, and enhancing staff education are recommended.
• Targeted strategies can improve the translation of evidence-based guidelines into practice for better falls prevention.