Background: Identifying separate dimensions of concussion symptoms may inform a precision medicine approach to treatment. It was previously reported that a bifactor model identified distinct acute postconcussion symptom dimensions.
Purpose: To replicate previous findings of a bifactor structure of concussion symptoms in the Concussion Assessment Research and Education (CARE) Consortium sample, examine measurement invariance from pre- to postinjury, and evaluate whether factors are associated with other clinical and biomarker measures.
Study design: Cohort study (Diagnosis); Level of evidence, 2.
Methods: Collegiate athletes were prospectively evaluated using the Sport Concussion Assessment Tool-3 (SCAT-3) during preseason (N = 31,557); 2789 were followed at <6 hours and 24 to 48 hours after concussion. Item-level SCAT-3 ratings were analyzed using exploratory and confirmatory factor analyses. Bifactor and higher-order models were compared for their fit and interpretability. Measurement invariance tested the stability of the identified factor structure across time. The association between factors and criterion measures (clinical and blood-based markers of concussion severity, symptom duration) was evaluated.
Results: The optimal structure for each time point was a 7-factor bifactor model: a General factor, on which all items loaded, and 6 specific factors-Vestibulo-ocular, Headache, Sensory, Fatigue, Cognitive, and Emotional. The model manifested strict invariance across the 2 postinjury time points but only configural invariance from baseline to postinjury. From <6 to 24-48 hours, some dimensions increased in severity (Sensory, Fatigue, Emotional), while others decreased (General, Headache, Vestibulo-ocular). The factors correlated with differing clinical and biomarker criterion measures and showed differing patterns of association with symptom duration at different time points.
Conclusion: Bifactor modeling supported the predominant unidimensionality of concussion symptoms while revealing multidimensional properties, including a large dominant General factor and 6 independent factors: Headache, Vestibulo-ocular, Sensory, Cognitive, Fatigue, and Emotional. Unlike the widely used SCAT-3 symptom severity score, which declines gradually after injury, the bifactor model revealed separable symptom dimensions that have distinct trajectories in the acute postinjury period and different patterns of association with other markers of injury severity and outcome.
Clinical relevance: The SCAT-3 total score remains a valuable, robust index of overall concussion symptom severity, and the specific factors identified may inform management strategies. Because some symptom dimensions continue to worsen in the first 24 to 48 hours after injury (ie, Sensory, Fatigue, Emotional), routine follow-up in this time frame may be valuable to ensure that symptoms are managed effectively.
Keywords: CARE Consortium; bifactor model; blood biomarkers; clinical phenotypes; invariance modeling; sport-related concussion.