Hospital mortality and discharge destination prediction after fall-related traumatic brain injury in older adults
Related Experiment Videos
Biche Osong1, Megan Armstrong1, Hannah Williams1
1Center for Pediatric Trauma Research and Center for Injury Research and Policy, The Abigail Wexner Research Institute at Nationwide Children's Hospital, Columbus, Ohio, USA.
Objective:
Develop prediction models to support care for older adults with fall-related traumatic brain injury (TBI).
Summary Of Background:
Fall-related TBI is a leading cause of hospitalization and death among older adults. Prognostication remains challenging in this population because frailty and comorbidity burden greatly influence recovery trajectories. Tailored prognostic models to support individualized patient care remain limited.
Patients And Methods:
Data from 179,287 older adults (≥65 years) hospitalized with fall-related TBI in the NTDB (2017-2021) were analyzed. MARS models were developed to predict in-hospital mortality and discharge destination (home, rehabilitation, or care facility). Predictor selection was guided by clinicians' input, and model performance was evaluated using discrimination and calibration measures.
Results:
The mortality model showed high discrimination (AUC 0.88, 95% CI 0.87-0.89; balanced accuracy 79%). The discharge model demonstrated moderate-to-good discrimination across discharge destinations (AUCs 0.77-0.85; balanced accuracies 70%-77%). Both models showed good calibration and were implemented as interactive web-based tools.
Conclusion:
This study developed two prognostic models that provide individualized outcome probabilities to support prognostic communication and care planning for older adults hospitalized with fall-related TBI.
