Predictors of outcome in prolonged posttraumatic disorders of consciousness and assessment of medication effects: A multicenter study.
OBJECTIVES: To develop predictive models of recovery from the vegetative state (VS) and minimally conscious state (MCS) after traumatic brain injury (TBI) and to gather preliminary evidence on the impact of various psychotropic medications on the recovery process to support future randomized controlled trials. Design Longitudinal observational cohort design, in which demographic information, injury and acute care history, neuroimaging data, and an initial Disability Rating Scale (DRS) score were collected at the time of study enrollment. Weekly follow-up data, consisting of DRS score, current psychoactive medications, and medical complications, were gathered until discharge from inpatient rehabilitation. SETTING: Seven acute inpatient rehabilitation facilities in the United States and Europe with specialized programs for treating patients in the VS and MCS. PARTICIPANTS: People with TBI (N=124) who were in the VS or MCS 4 to 16 weeks after injury. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: DRS score at 16 weeks after injury and time until commands were first followed (among those participants demonstrating no command following at study enrollment). Results DRS score at enrollment, time between injury and enrollment, and rate of DRS change during the first 2 weeks of poststudy observation were all highly predictive of both outcomes. No variables related to injury characteristics or lesions on neuroimaging were significant predictors. Of the psychoactive medications, amantadine hydrochloride was associated with greater recovery and dantrolene sodium was associated with less recovery, in terms of the DRS score at 16 weeks but not the time until commands were followed. More detailed analysis of the timing of functional improvement, with respect to the initiation of amantadine provided suggestive, but not definitive, evidence of the drug's causal role. CONCLUSIONS: These findings show the feasibility of improving outcome prediction from the VS and MCS using readily available clinical variables and provide suggestive evidence for the effects of amantadine and dantrolene, but these results require confirmation through randomized controlled trials.