Evidence-Based Recommendations for Geriatric Trauma Care: Systematic Review and AI-Assisted Consolidation of Clinical...

This systematic review provides a comprehensive overview of evidence-based clinical practice guidelines (CPGs) for geriatric trauma care through the utilization of a novel artificial intelligence (AI) methodology.

With a growing proportion of trauma cases involving geriatric patients, there is a need for tailored management considering their unique challenges related to multimorbidity and frailty. Evidence-based recommendations support clinical decision-making in this population.

MEDLINE Ovid and 8 guideline databases were searched on December 18, 2021, for CPGs published after January 1, 2016. Eligible CPGs were required to be published in German or English, have an evidence assessment (such as Grading of Recommendations, Assessment, Development and Evaluation (GRADE)), report at least 1 recommendation for older patients (≥60 years, frail, or with dementia/delirium) applicable in an acute inpatient trauma setting, and achieve an Appraisal of Guidelines for Research & Evaluation II (AGREE II) overall quality rating of at least 4 points rated by 2 independent experts. The extracted CPG recommendations underwent semantic analysis with AI-based natural language processing, followed by dimensional reduction and grouping via k-means clustering.

After screening, full-text review, and AGREE II appraisal, 95 CPGs out of 12,106 records were included, yielding 821 CPG recommendations relevant to geriatric trauma care. Most recommendations addressed dementia, delirium, adverse drug reactions, deprescribing, multimorbidity, frailty, and cardiovascular disease. The clustering analysis identified 27 clusters of semantically similar recommendations, and each cluster was consolidated into a single meta-recommendation.

Our systematic review provides a comprehensive summary of evidence-based CPG recommendations for geriatric trauma care, offering clinicians a solid foundation for managing this vulnerable patient population.