Predictive models in traumatic brain injury (CanTBI)

Chercheur (es) Principal (aux) Hutchison, Jamie
Co-investigateur(s) Hutchison, Jamie
Panenka, William
Wheeler, Anne

Predictive models in traumatic brain injury (CanTBI)

Note: la description de ce programme de recherche est en anglais seulement.


Traumatic brain injury (TBI) continues to have a huge impact on Canadians: both patients and their families. It is responsible for a loss of quality of life and prolonged health care utilization yet there are no useful models that help clinicians predict long term outcomes. This gap impairs clinical decision making and accurate risk stratification in therapeutic trials.


In patients with TBI we propose: (1) To develop and validate prognostic models using novel molecular, electroencephalography (EEG) and magnetic resonance imaging (MRI) biomarkers and psychosocial and clinical risk factors to predict long term quality of life; and (2) To evaluate the added value of pre-injury health care utilization information from provincial and national databases with the risk factors from objective 1 to predict post-injury health care utilization.

Study design, patient population, sample size and number of centres: Prospective observational study of 600 children and adults with mild, moderate and severe TBI at 10 Canadian hospitals (5 pediatric and 5 adult).


Biobanked and data collected: We have collected biosamples, clinical data and personal health information for linkage to health care utilization databases in 200 research participants. Biosamples are stored in 4 TBI biobanks (Vancouver, Calgary, Toronto and Montreal). Data recorded includes pre- and peri-injury risk factors. The primary outcome is a quality of life score and secondary outcomes include scores for post-concussion syndrome, global neurological function, memory, executive function and psychiatric diagnoses measured out to 12 months after injury.

Discovery work to be accomplished with this application: Molecular biomarker: In the biosamples collected acutely, we have developed pipelines to discover blood proteins and metabolites which are most closely associated with long term outcomes. Neuroimaging biomarker: Diffusion tensor imaging (DTI) will be recorded and analyzed in 120 patients at 6 sites with research MRIs. The goal of this study is to determine if novel MRI DTI biomarkers improve our prognostic models. EEG biomarker: EEG prognostic indices will be recorded using our novel ‘Brainsview’ device. Pre- injury health care utilization factors: We will add pre-injury health care utilization factors into the predictive models with post-injury health care utilization as the outcome. We propose to develop (in N=200 participants) and validate (N=400 participants) predictive models for quality of life and health care utilization.

Knowledge translation, impact and stake holder engagement: We will conduct knowledge user workshops to ensure that the predictive models and outcomes are patient and family centered and useful to clinicians (physicians, psychologists, rehabilitation specialists), scientists and policy makers. Applying prognostic models in the clinical process will enhance patients’, families’ and providers’ ability to make the best clinical decisions and implement better acute and long term psychological and rehabilitation management strategies. These models will also be used to risk stratify participants in therapeutic trials. This study will also provide robust data describing health care utilization, before and after TBI, and predictive models integrating Canadian health care utilization data. These models will facilitate better planning and allocation of health care resources.