3071 — Predicting More and Less Autonomy with Expression 1-Year after Severe Brain Injury
Pape TL (Hines VA), Smith B
(Hines VA), Wei Y
(Hines VA), Guernon A
(Marianjoy Rehabilitation Hospital), Tang C
(Hines VA), Lundgren S
(Minneapolis VA), Picon L
(Tampa VA), Blahnik M
(Minneapolis VA), Kieffer K
(Tampa VA), Schettler S
Severe brain injury (BI) is a catastrophic event that evolves into a complex chronic condition requiring substantial caregiver support. While the majority of persons recovering consciousness require caregiver assistance, we have limited knowledge of factors promoting autonomy, how to intervene, and evidence for long-term planning. The objective of this analysis is to identify the strongest predictors of more or less autonomy with expressing needs and ideas 1-year after injury.
Data for a sample of persons (n = 70) who incurred prolonged unconsciousness were abstracted from an ongoing observational study. Autonomy 1-year after injury was measured using the Functional Independence Measures (FIM) and is dichotomized as less (1 – 3) and more (4 – 7). Seventy two predictors include measures of neurobehavioral functioning using the Disorders of Consciousness Scale (DOCS), co-existing conditions, interventions, injury, individual, and environmental characteristics. We used optimal data analyses to construct a hierarchical multi-attribute classification tree model.
The sample includes mostly young (average age = 36) men (64%) who recovered consciousness (72%) after an average of 126 days of unconsciousness. Two predictors entered the final model providing 92.3% overall classification accuracy; The second DOCS visual measure (p = 0.006) and the baseline DOCS visual measure (p = 0.002) with an optimal cut-point of 44.85. The second visual measure, obtained seven days after baseline DOCS which is obtained within 94 days of injury, was the strongest predictor. Leave-one-out analyses indicate that the model will correctly identify 82% of persons who will have more autonomy and 96% of persons who will have less autonomy with expressing needs and ideas 1-year after severe BI. The model captured 80% of the possible theoretical improvement in classification accuracy that exists after accounting for chance; a strong effect.
For persons experiencing prolonged disordered consciousness, more or less autonomy for expressing needs and ideas 1-year after severe BI can be predicted with at least 92% accuracy using DOCS visual measures obtained during acute rehabilitation.
While a larger sample is needed for more specific autonomy levels, these results demonstrate that acute rehabilitation measures can help care managers, families, and therapists develop long-term resource and therapy plans for individual patients.