Predictors Of Performance On The Impact Validity Indices At Baseline

Abstract Objective To circumvent “sandbagging”, the ImPACT, a widely used computerized program in sports-related concussion evaluations, contains five embedded validity indices (VIs). This study aimed to identify variables associated with invalid performance (i.e., scoring beyond the trigger-warning threshold) on at least one VI. Method Analysis derived from 1287 college students was conducted utilizing a binary logistic regression. The dependent variable (the five VIs) was dichotomized into “Fail” (n = 71; 76.1% male, 66.2% Caucasian) and “Pass” (n = 1216; 69.8% male, 78.4% Caucasian) groups. Any athlete with invalid performance on one or more VIs was classified into the “Fail” group; athletes with valid performance across all five VIs were classified into the “Pass” group. Independent variables included age, sex, concussion history, depression, estimated FSIQ, learning disability, ADHD, and “post-concussion” symptoms. A SMOTE was applied to balance the dataset and 1000 iterations of the regression were applied. Results Significant associations across 1000 iterations of logistic regressions were found: Concussion history (p < 0.05, 992/1000), estimated FSIQ (p < 0.05; 997/1000), and depressive symptomatology (p < 0.05; 1000/1000) were negatively associated with “Fail” group membership. Learning disability (p < 0.05; 1000/1000) and post-concussion symptoms (p < 0.05; 1000/1000) were positively associated with “Fail” group membership. No associations were identified with sex or ADHD and group membership. Conclusion Results suggest that those with fewer concussions, lower estimated FSIQ, fewer depressive symptoms, learning disability, and more severe post-concussion symptoms may be more likely to show invalid test performance (or trend toward sandbagging) at baseline. Results have important clinical implications for identifying premorbid characteristics that may be indicative of sandbagging.