Data-driven characterization of traumatic brain injury severity from clinical, neuroimaging, and blood-based indicators

Abstract

The conventional clinical approach to characterizing traumatic brain injuries (TBIs) as mild, moderate, or severe using the Glasgow Coma Scale (GCS) total score has well-known limitations, prompting calls for more sophisticated strategies to characterize TBI. Here, we use item response theory (IRT) to develop a novel method for quantifying TBI severity that incorporates neuroimaging and blood-based biomarkers along with clinical measures. Within the multicenter Transforming Research and Clinical Knowledge in TBI (TRACK-TBI) study sample (N = 2545), we show that a set of 23 clinical, head computed tomography (CT), and blood-based biomarker variables familiar to clinicians and researchers index a common latent continuum of TBI severity. We illustrate how IRT can be used to identify the relative value of these features to estimate an individual's position along the TBI severity continuum. Finally, we show that TBI severity scores generated using this novel IRT-based method incrementally predict functional outcome over classic clinical (mild, moderate, severe) or International Mission for Prognosis and Analysis of Clinical Trials in TBI (IMPACT) classification methods. Our findings directly inform ongoing international efforts to refine and deploy new pragmatic, empirically-supported strategies for characterizing TBI, while illustrating a strategy that may be useful to evolve staging systems for other diseases.

Figure 1

Test information curves for each measurement domain. Test information reflects the sum of information across all indicators (items) in each measurement domain. Higher information reflects greater measurement precision in characterizing and distinguishing persons at a given level of TBI severity.

Figure 2

Item information curves, stratified by measurement domain of the TBI severity indicators.

Figure 3

Association between novel item response theory-based traumatic brain injury (TBI) severity scores, other TBI severity classification schemes, and functional outcome. (a) Histograms and scatterplots of item response theory TBI severity score versus traditional TBI severity classification schemes. Scatterplot points (individual subjects) are lagged in the direction of the y-axis to facilitate visualization of the number of points along the x-axis. (b) Model Nagelkerke R2 for models predicting death and unfavorable outcome at 6 months post-injury from traditional 3-level GCS-based mild, moderate, and severe categories, IMPACT Core model scores, and IRT-based TBI severity scores, as well as multivariable models investigating the incremental predictive value of IRT-based TBI severity scores in combination with the other variables. Abbreviations: c-mTBI, complicated mild TBI; GCS, Glasgow Coma Scale score; IMPACT, International Mission for Prognosis and Analysis of Clinical Trials in TBI; Mod, Moderate; u-mTBI, uncomplicated mild TBI