Examination of Reaction Time Deficits Following Concussion: A Systematic Review and Meta-analysis

Abstract

Reaction time (RT) deficits are reported following concussion, but it is unknown when these deficits normalize to pre-injury status. It is also unclear how factors such as RT measurement technique and participant characteristics influence post-concussion RT. The purpose of this systematic review and meta-analysis was to (1) characterize acute post-concussion (0-3days) RT impairments, (2) examine RT recovery over time, and (3) explore moderating factors related to acute RT impairment following concussion. Database searches (PubMed, CINAHL, EBSCOhost) were conducted according to PRISMA guidelines for articles published in English from January 2002 to March 2019. Studies compared baseline-to-post-injury RT within individuals (within-subject) and/or RT in concussed individuals to non-concussed controls (between-subject). Sixty studies met inclusion criteria, reporting on a total of 9688 participants with 214 discrete RT effects (Hedges' d; between-subject: N = 29, k = 129; within-subject: N = 42, k = 85). Of the 214 effects, 93 occurred in the acute (0-3days) post-injury timeframe (k = 47 between-subject). Numerous demographic [sex, age, concussion history, population type (athlete, military, and general population), athlete level (high school, college), and sport], and method-based (RT test and measure type, computerized neurocognitive testing platform, concussion definition, and time post-injury) moderators were examined for mean effect influence. Mixed-effects multi-level modeling with restricted-maximum-likelihood estimation was used to account for nested effects and high heterogeneity for the pooled effect size (D+). Significant medium-magnitude RT deficits were observed acutely for between- (D+ = -0.7279, 95% CI -0.9919, -0.4639, I2 = 88.66, p < 0.0001) and within-subject (D+ = -0.7472, 95% CI -0.9089, -0.5855, I2 = 89.21, p < 0.0001) effect models. RT deficits were present at the sub-acute and intermediate-term timeframes for between-subject effects (sub-acute: D+ = -0.5655, 95% CI -0.6958, -0.4352, p < 0.0001; intermediate-term: D+ = -0.3219, 95% CI -0.5988, -0.0450, p = 0.0245). No significant RT mean effect was observed for the between-subject model at the long-term timeframe, indicating RT recovery among concussed participants relative to controls (D+ = 0.3505, 95% CI -0.4787, 1.1797, p = 0.3639). Sex was a significant moderator for between-subject effects, with every 1% male sample size increase demonstrating -0.0171 (95% CI -0.0312, -0.0029, p = 0.0193) larger RT deficits. Within-subject effect models resulted in RT measure type (simple: [D+ = -0.9826] vs. mixed: [D+ = -0.6557], p = 0.0438) and computerized neurocognitive testing platforms (ANAM: [D+ = -0.3735] vs. HeadMinder CRI: [D+ = -1.4799] vs. ImPACT: [D+ = -0.6749], p = 0.0004) having significantly different RT-deficit magnitudes. No other moderators produced significantly different RT-deficit magnitudes (between-subject: [p ≥ 0.0763], within-subject: [p ≥ 0.1723]). Robust RT deficits were observed acutely following concussion. Minimal magnitude differences were noted when comparing between- and within-subject effects, suggesting that pre-injury baselines may not add clinical value in determining post-injury RT impairment. RT deficits persisted up till the intermediate-term (21-59days post-injury) timeframe and indicate lingering deficits exist. Mean effect size differences were observed between RT measure types and computerized neurocognitive testing platforms; however, all categories displayed negative effects consistent with impaired RT following concussion. Clinical interpretation suggests that measuring RT post-concussion is more important than considering the RT method employed so long as reliable and valid tools are used. PROSPERO Registration #CRD42019119323.