Prospective Assessment of Improvement and Decline in Global Functioning After Traumatic Brain Injury Using Multistate Markov Models

Nicola L de Souza; Cynthia L Beaulieu; Raj G Kumar; Kristen Dams-O'Connor

doi:10.1097/HTR.0000000000001163

Prospective Assessment of Improvement and Decline in Global Functioning After Traumatic Brain Injury Using Multistate Markov Models

J Head Trauma Rehabil. 2026 May 8. doi: 10.1097/HTR.0000000000001163. Online ahead of print.

Authors

Nicola L de Souza¹, Cynthia L Beaulieu, Raj G Kumar, Kristen Dams-O'Connor

Affiliation

¹ Author Affiliations:Traumatic Brain Injury and Concussion Center, Department of Neurology, University of Utah School of Medicine, Salt Lake City, Utah (de Souza); Department of Rehabilitation and Human Performance, Icahn School of Medicine at Mount Sinai, New York, New York (de Souza, Kumar, Dams-O'Connor); Department of Physical Medicine & Rehabilitation, The Ohio State University College of Medicine, Columbus, Ohio (Beaulieu); and Department of Neurology, Icahn School of Medicine at Mount Sinai, New York, New York (Dams-O'Connor).

PMID: 42098925
DOI: 10.1097/HTR.0000000000001163

Abstract

Objective: To describe transitions in global functional status across consecutive follow-up visits, including probabilities for improvement, decline, and stability, and to identify factors associated with transition during the first 10 years after traumatic brain injury (TBI).

Setting: TBI Model Systems National Database.

Participants: In total, 13,731 participants with TBI requiring inpatient rehabilitation who completed 2 follow-up assessments within 10 years post-injury, with at least 2 consecutive assessments at 1, 2, 5, or 10 years.

Design: Secondary data analysis.

Main measures: Global functional status measured by the Glasgow Outcome Scale-Extended, collapsed into 4 states: Good Recovery, Moderate Disability, Severe Disability/Vegetative State, and Death. Transition probabilities were estimated using multistate Markov models, and transition-specific hazard ratios were calculated for covariate effects.

Results: Probability of remaining in the same state was highest among those with Good Recovery at year 1, with 68.4% (95% CI: 55.1%-80.9%) remaining stable between years 1 and 2. Improvement was common among those initially in Moderate or Severe Disability states. Among participants with Moderate Disability at year 1, 36.6% (95% CI: 18.1%-48.7%) improved and 15.4% (95% CI: 10.1%-18.3%) declined by year 2. Among those with Severe Disability at year 1, 34.5% (95% CI: 22.8%-40.1%) improved to Moderate Disability, and 13.7% (95% CI: 6.7%-21.3%) improved 2 states to Good Recovery. Trends were similar between years 5-10. Risk factors for both decline and mortality included older age at injury. Non-White race, lower education, and public insurance predicted lower likelihood for improvement and greater likelihood of decline. Protective factors included female sex, higher education, private insurance, higher functioning at hospital discharge, and residing in states offering better long-term services and supports.

Conclusion: Recovery after TBI is dynamic, with higher probabilities of functional stability and improvement than decline over 10 years post-injury. Findings highlight modifiable factors with policy-level implications that may influence recovery trajectories.

Keywords: brain injuries; functional outcome; multistate model; traumatic.

Abstract

Grants and funding