Valid but invalid: suboptimal impact© baseline performance in university athletes

Using Variance to Explore the Diagnostic Utility of Baseline Concussion Testing

The Graded Symptom Checklist (GSC), Standardized Assessment of Concussion (SAC), Balance Error Scoring System (BESS), and King-Devick Test (KDT) are considered important components of concussion assessment. Whether baseline testing improves the diagnostic utility of these tests remains unclear. We performed an observational cohort study to investigate the within-subject and between-subjects variability of these tests over repeated assessments during two football seasons to examine whether baseline testing reduces variability in test performance. Thirty-five male collegiate football players completed weekly clinical concussion assessments over two seasons. Within-subject (week-to-week) and between-subjects (player-to-player) variability for each test were compared using a bootstrap analysis. Within-subject and between-subjects proportions of overall variance for each test score were calculated. Mixed-model analyses were used to quantify practice effects resulting from repeated testing. For the GSC and BESS, within-subject and between-subjects variability did not significantly differ. For the KDT, the proportion of within-subject variance (20.2%) was significantly less than the between-subjects variance (79.8%). For SAC, however, the proportion of within-subject variance (66.8%) was significantly greater than the between-subjects variance (33.8%). A small, but significant, practice effect was observed for the BESS and KDT tests. When athletes are evaluated during a football season for concussion using the GSC, SAC, and BESS, comparing their scores to baseline performance is likely no more beneficial than comparing them to normative population data for identifying neurological changes associated with concussion. For the KDT, comparison to baseline testing is likely beneficial because of significantly higher between-subjects variability.