1. Background The healthcare, societal and economic impact from paediatric traumatic brain injury (pTBI) is one of the greatest unmet needs with estimates suggesting upto 2/3rd of survivors developing life-long neurological deficits. Due to ethical and logistical challenges related to paediatric research, there is limited age-appropriate evidence for children which perpetuates the funding gaps further confounding the lack of evidence. However, extrapolation of adult TBI research evidence to children is inaccurate and inappropriate given the differences in mechanisms and patterns of injury, pathophysiological responses and neurological recovery from it, in the context of developmental trajectory; hence, robust evidence is required to improve outcomes from pTBI as well as facilitate research collaborations.
The prognostic models used for adult TBI research (IMPACT, CRASH)5,6 and imaging criteria (Marshall and Rotterdam scores), have been derived and validated from analysis of large international datasets which have undergone further validation in multiple prospective studies; the wide use of these prognostic models across neurotrauma research highlights the relative simplicity and the variables used for prediction making them applicable to both low and high resource set-ups.
This has facilitated international collaborative research in adult TBI; however, no such models exist for pTBI with most pTBI research continuing to use adult prognostic models. Though the variables used for these models show association with outcome in pTBI as well, there are multiple issues with this approach with the key difficulty being the age variable. In the adult models, the younger age is expected to be associated with better outcome; for CRASH, the model is likely weighed heavily towards non-age variables as the younger patients are expected to do better with patients between 18-40 lumped as \<40 years and for IMPACT, slightly better which accounts for 14 years and above.
However, the balance between neuroplasticity and neurodevelopmental trajectory in paediatric age group is difficult to predict with evidence suggesting worse neuro-developmental outcomes after TBI in younger children. Hence the adult models can either over- or under-predict neurological outcomes in pTBI and have never been validated in pTBI datasets.
2. Rationale \& Theoretical Framework A validated predictive model for pTBI would be the first step towards collaborative research and identifying age-appropriate benchmarks for pTBI research studies as well as to complement an individual child's clinical assessment, treatment decisions, informing families and resource allocation. Given that the amount of data required to create pTBI predictive model is difficult to collect and the reasonable validity of adult prognostic models in pTBI, albeit in small single-centre studies, investigators propose to create paediatric modification to the adult models and identify a robust pTBI predictive model for improved classification of injury severity to predict disease trajectory and outcome as well as stratification of patients for interventional research and benchmarking in pTBI to help with appropriate resource allocation for neuro-interventions towards improved outcomes.
3. Research Question/Aims and Objectives:
To create and validate a paediatric specific prognostic model for outcomes following moderate to severe TBI for use in both resource-limited and resource-rich environments. The study hypothesis is that the performance of the current adult TBI prognostic models in predicting outcomes following pTBI can be improved by introducing age-appropriate modifications to the existing models.
4. Study Design/ Methods:
The study will use combined retrospective pTBI datasets (moderate to severe TBI in children \< 18 years).
5. Sample \& Statistical Methods:
There is no defined sample size for the study. Investigators aim to collect datasets which contain the variables defining injury severity from children with moderate to severe TBI upto the age of 18 years. There are no exclusion criteria for the study. All the datasets will either be from a prospective study (ethics approved with valid consent) and/or datasets created for audit/institutional purposes with appropriate approvals from local/national research committees and contain anonymised non-identifiable information.
