Arts & Neuroscience
To determine whether neural responses to music and language stimuli identify consciousness in patients with severe traumatic brain injury (TBI). We hypothesized that neural responses to music stimuli identify consciousness more reliably than do responses to language stimuli.
Prospective, observational EEG study of consecutively enrolled patients with severe TBI.
Setting : Intensive care unit (ICU) at a Level 1 Trauma Center.
Participants (or Animals, Specimens, Cadavers) : We enrolled 14 patients (18-65 years old; 8 conscious on behavioral assessment, 3 conscious on task-based functional MRI assessment, 3 unconscious) and 16 age- and sex-matched healthy subjects.
Interventions : Not applicable.
Main Outcome Measure(s) : We evaluated the specificity and sensitivity of the average strength (STR) and global efficiency (GE) of EEG α-band functional networks for classifying consciousness. A significant neural response was defined as being within one standard deviation of the healthy control mean. As a secondary outcome measure, we also compared areas under receiver operator characteristic curves (AUC). These curves were generated after calculating the sensitivities and specificities at several significance cutoffs.
Results : Both music metrics had 100% specificity (3/3 unconscious patients correctly classified), but low sensitivities (STR: 2/11, GE: 3/11 conscious patients correctly classified). Both language metrics yielded a lower specificity of 2/3, but higher sensitivities (STR: 4/11, GE: 5/11). Both language metrics resulted in a higher AUC (0.64) than both music metrics (0.52).
Conclusions : Language, rather than music, is a more discriminative stimulus paradigm for classifying patients who are conscious.