To identify principal components of a battery of cognitive assessments that best capture the clinical state of mTBI patients across multiple time points and can be used to inform selection of outcome measures in future mTBI research studies.

A current difficulty in the design of mTBI clinical trials is the wide overlap in cognitive domains assessed, use of various assessments to evaluate a single domain, and the large variation in assessments. There is a need for statistically driven data reduction strategies to identify the domains that most accurately capture the clinical state of mTBI patients.

325 patients and 152 controls were enrolled after mTBI. Patients were enrolled at either Encounter 1 (<72 hours post-injury) or Encounter 2 (5-10 days), and returned for Encounter 3 (12-16 days) and Encounter 4 (83-97 days). At each encounter, subjects underwent a neuropsychological assessment battery (RPQ, BSI, TMT, HVLT, WAIS/WISC). A covariance matrix was calculated from clinical assessments. PCA was used as a data reduction strategy to elucidate the core dimensions of the mTBI sequelae.

In mTBI patients, there was a high degree of covariance among symptoms inventories. Among neurocognitive assessments, TMT correlated significantly with WAIS/WISC and to a lesser degree total recall from HVLT. In the PCA, the primary component correlated mostly with TMT-B (0.6), delayed recall and retention (0.5 each), and BSI (0.4), and explained 89.7% of the variance in the data across visits. The remaining components explained 10.3% of the variance.

In our population of mTBI patients, 89.7% of the variance was explained by a primary component which remained consistent across all encounters. Clinical and neurocognitive metrics showed a significant degree of correlation suggesting that neurocognitive batteries could be simplified. The results of the study are useful for future clinical trial design of similarly stratified mTBI subjects.