Single-Task and Dual-Task Gait Performance After Sport-Related Concussion: A Machine Learning Statistical Approach.

Abstract

Background: This study evaluated 2 different dual-task (DT) conditions during tandem gait (TG) to predict sport-related concussion (SRC) diagnosis.

Hypothesis: The best (fastest) single-task (ST) gait will differ between groups (controls vs SRC; baseline vs SRC), with auditory pure switching task (APST) response rate being the most important behavioral variable to aid prediction of SRC.

Study design: Cohort design.

Level of evidence: Level 3.

Methods: A total of 409 National Collegiate Athletic Association Division I student-athlete controls and 21 team-physician-diagnosed SRC participated. All data were collected at preseason physicals (baseline) and within 7 days of injury for SRC. Each participant completed 3 conditions of TG in a pseudorandomized order: (1) ST, (2) DT with serial-7s (SS) subtractions, and (3) DT with APST. Outcomes of time-to-complete for TG and behavioral (eg, responses per second) for SS and APST were recorded for each trial.

Results: ST Trials 2 (P = 0.03) and 3 (P = 0.01) were significantly different between controls and SRC. ST Trial 3 (P = 0.04) was significantly different between baseline and SRC. Average responses per second for APST were significantly different between- (P < 0.01) and within- (P = 0.01) group.

Conclusion: The results suggest that ST is significantly slower after SRC. However, DT (both SS and APST) time-to-complete are also important variables when predicting the SRC diagnosis. It is advised that both ST and DT be administered when making clinical decisions regarding postural instability after SRC.

Clinical relevance: The best ST TG time to complete gait is an important objective marker of concussion while DT paradigms, specifically SS and APST, are highly variable. DT may be more useful for clinical observable signs of SRC. Both SS and APST have unique usefulness, but APST response rate per second can be relied upon numerically for clinical decisions.