Multimodal Assessment and Prognosis in Disorders of Consciousness After Severe Brain Injury (CyDoC-MAP)

Overview

After a severe brain injury some patients survive but cannot communicate, and deciding early which of them are likely to recover consciousness remains one of the hardest problems in neurocritical care. Bedside examination alone misclassifies a substantial proportion of these patients. CyDoC-MAP is a single-centre, prospective, observational cohort study conducted at Nicosia General Hospital, the sole trauma referral centre for Cyprus. It enrols patients aged 16 years or older who are intubated within 24 hours of a moderate-to-severe traumatic brain injury or a haemorrhagic stroke (intracerebral or subarachnoid haemorrhage), and who are subsequently classified as being in a vegetative state / unresponsive wakefulness syndrome (VS/UWS) or a minimally conscious state (MCS) on the Coma Recovery Scale-Revised (CRS-R). Four assessment modalities are recorded: (1) the CRS-R, performed at least twice with an interval of at least 48 hours; (2) serum neuron-specific enolase (NSE) sampled within 24 hours of intubation; (3) the bispectral index (BIS), recorded at least twice with an interval of at least 48 hours, after five minutes of standardised noxious and auditory stimulation; and (4) in the traumatic subgroup only, 1.5 T magnetic resonance imaging performed 7-28 days after injury and graded 1-4 by lesion depth by two independent raters. Level of consciousness is reassessed with the CRS-R 6 to 12 months after the index event, and the total CRS-R score (0-23) at that reassessment is the primary outcome. The primary aim is to estimate the strength of the association between the bispectral index recorded on the ward and that later CRS-R score, and to quantify what the bispectral index adds beyond the baseline clinical assessment. The number of eligible patients at a single national centre does not support the development of a prognostic model; the study is designed to produce effect-size estimates with confidence intervals that will inform a subsequent multicentre study. The study is purely observational. No intervention is administered, no study procedure alters clinical management, and transfer to rehabilitation is never delayed for research purposes.

Background. Functional neuroimaging has shown that a proportion of patients who meet the clinical criteria for a vegetative state retain detectable, volitional brain responses to command - the phenomenon of cognitive motor dissociation (CMD). A recent multicentre study reported CMD in approximately one quarter of behaviourally unresponsive patients, with marked variation between centres. Cyprus has no established diagnostic pathway for disorders of consciousness and no locally available specialised rehabilitation service for this population. Design. Single-centre, prospective, observational cohort study. No randomisation and no intervention. Patients are identified in the intensive care unit and formally entered into the study once they are transferred to the Neurosurgical or Neurological ward and a disorder of consciousness is documented with the CRS-R. The single exception to this sequence is the NSE biomarker, for which venous blood is drawn within 24 hours of intubation. Assessment schedule. Step 1 - Enrolment and screening: patients aged 16 years or older with traumatic brain injury or haemorrhagic stroke, intubated for a disorder of consciousness within 24 hours of the event; exclusion criteria applied. Step 2 - Acute phase (ICU, within 24 hours of intubation): venous sample for serum neuron-specific enolase. Step 3 - Ward baseline: CRS-R and BIS, each at least twice with an interval of at least 48 hours; classification as VS/UWS or MCS. BIS is recorded after five minutes of noxious and auditory stimulation to avoid falsely low values due to drowsiness; the highest value is recorded. The CRS-R is administered by the study neurologist or speech and language therapist, with attention to sedation, sepsis, subclinical seizures and environmental confounders; the highest score is recorded. Step 4 - Subacute phase (7-28 days after injury, traumatic subgroup only): 1.5 T MRI with T2/FLAIR, DWI, T2\*GRE and SWI sequences; lesions graded 1-4 by depth (cortex / white matter; corpus callosum, basal ganglia, thalamus; brainstem; substantia nigra / mesencephalic tegmentum) by two independent raters, with disagreements resolved by consensus. The timing allows postoperative changes to resolve and intracranial pressure to stabilise. Step 5 - Follow-up (6-12 months after the index event): reassessment with the CRS-R; outcome classified as VS/UWS, MCS or eMCS. Statistical analysis. The analysis is pre-specified in a signed Statistical Analysis Plan (SAP-01) dated before the enrolment of the first patient. The primary outcome is the total CRS-R score (0-23) at 6-12 months, treated as a continuous variable; the VS/UWS-MCS-eMCS classification is retained as a secondary outcome. The highest bispectral index value is designated in advance as the primary prognostic marker, because it is the novel element of the study and the only marker whose measurement is guaranteed in every enrolled patient. The single hypothesis test is the Spearman rank correlation between the highest bispectral index and the total CRS-R score at follow-up, reported with a 95% confidence interval, alongside a simple linear regression estimating the slope. One secondary pre-specified analysis adds the baseline total CRS-R as a second predictor and reports the incremental explained variance; no further predictor enters that model. Serum neuron-specific enolase, MRI lesion depth grade, age, pre-intubation Glasgow Coma Scale and intensive care length of stay are analysed on an exploratory basis only, without correction for multiple comparisons and without conclusions. The anatomical distribution of MRI lesions is presented descriptively, by site and side, with no significance testing. Inter-rater reliability of the MRI grading is expressed as weighted Kappa. Development of a multivariable prognostic model, stepwise or automated variable selection, searching for an optimal bispectral index cut-off, and dichotomisation of continuous variables are excluded at the outset: with 30-40 patients and an expected 10-18 outcome events, current sample-size criteria for prediction-model development require approximately 320-385 patients. The study is therefore reported as a study of prognostic associations and feasibility, not as prognostic model development. Ethics and data protection. Approved unanimously by the Cyprus National Bioethics Committee (protocol EEBK/EP 2025/47) on 10 July 2026. Written informed consent is obtained from the next of kin. Data are pseudonymised at the point of collection; no names or direct identifiers are held in the study database. Data handling complies with the EU General Data Protection Regulation, and records are retained for eight years from the start of the study and are then destroyed.