The structure of postconcussion symptoms on the Neurobehavioral Symptom Inventory: a comparison of alternative models.

OBJECTIVE: To evaluate and compare the existing Neurobehavioral Symptom Inventory factor structure models to determine which model provides the best overall fit for postconcussion symptoms and determine which model is useful across different samples (eg, with and without mild traumatic brain injury [TBI] history).
SETTING: N/A. PARTICIPANTS: A Florida National Guard sample (N = 3098) and a national Department of Veterans Affairs sample (N = 48,175).
DESIGN: Retrospective structural equation modeling was used to compare 16 alternative factor structure models. First, these 16 possible models were examined separately in both samples. Then, to determine whether the same factor structures applied across subsamples within these samples, the models were compared for those deployed and those not deployed in the Florida National Guard sample and between those with TBI confirmed on clinical evaluation and those who were determined not to have sustained a TBI within the Department of Veterans Affairs sample. MAIN MEASURES: Neurobehavioral Symptom Inventory.
RESULTS: A 4-factor model--vestibular, somatic, cognitive, and affective--had the best overall fit, after elimination of 2 items (ie, hearing problems and appetite disturbance), and was most applicable across samples.
CONCLUSIONS: These findings extend the findings of Meterko et al to other samples. Because findings were consistent across sample and subsamples, the current findings are applicable to both Department of Veteran Affairs and Department of Defense postdeployment medical evaluation settings.