RESUMO
PURPOSE: We examined the relationship between comorbid medical conditions and changes in cognition over the course of rehabilitation following acquired brain injury. In particular, we compared outcomes between traumatic brain injury (TBI) and non-TBI using a retrospective inpatient rehabilitation dataset. We hypothesized that differences by diagnosis would be minimized among subgroups of patients with common comorbid medical conditions. MATERIALS AND METHODS: We used the Functional Independence Measure (FIM)-cognition subscale to index changes in cognition over rehabilitation. A decision tree classifier determined the top 10 comorbid conditions that maximally differentiated TBI and non-TBI. Ten subsets of patients were identified by matching on these conditions, in rank order. Data from these subsets were submitted to repeated-measures logistic regression to establish the minimum degree of commonality in comorbid conditions that would produce similar cognitive rehabilitation, regardless of etiology. RESULTS: The TBI group demonstrated a greater increase in ordinal scores over time relative to non-TBI, across all subscales of the FIM-cognition. When both groups were matched on the top 3 symptoms, there were no significant group differences in rehabilitation trajectory in problem-solving and memory domains (Cohen's d range: 0.2-0.4). CONCLUSION: Comorbid medical conditions explain differences in cognitive rehabilitation trajectories following acquired brain injury beyond etiology.