Diagnostic Efficiency of ImPACT and CogSport in Concussed Rugby Union Players Who Have Not Undergone Baseline Neurocognitive Testing

Abstract

The current study endeavored to replicate the approach to sports-related concussion management adopted by some community-based sporting organizations by examining the diagnostic efficiency of CogSport and ImPACT in athletes without baseline test data and assessed only once postinjury. Recently concussed nonelite-level rugby union players (N = 51) were tested within 72 hours of sustaining a concussion and were compared to nonconcussed matched controls (N = 41). Demographic information and history of recent concussion were also collected. Logistic regression analysis and receiver-operating characteristic curve analysis were conducted to evaluate the most accurate scores at identifying group membership. Overall, the classification accuracy of CogSport and ImPACT represented only very minimal improvements over a demographic-only (age, estimated premorbid Full-Scale IQ, and number of previous concussions) model. Positive predictive power and negative predictive power of composite scores were modest. The ImPACT postconcussion symptoms total (severity) score was most accurate at classifying concussed athletes. Where neuropsychological tests are utilized on a single occasion postconcussion and in the absence of baseline testing, postinjury testing does not improve diagnostic utility beyond the demographic model. These results do not validate this approach as a useful method of managing sports-related concussion.

Key words:

• CogSport
• computerized neuropsychological testing
• ImPACT
• sports-related concussion

ACKNOWLEDGEMENTS

We would like to thank Professor Mark R. Stevenson and Stephanie J. Hollis from The George Institute, The University of Sydney, for access to a proportion of rugby players from their mild traumatic brain injury study. We would also like to thank Dr. David Darby from CogState Ltd., Carlton, Victoria, for the gratis use of CogSport for our study.

Notes

Note. Due to some evidence of skewness within variables in each group, analyses were rerun using nonparametric Mann-Whitney U tests with no differences in results. CONC = Concussion; CTRL = Control; FSIQ = Full-Scale IQ.

No. = Number; CONC = Concussed Group; CTRL = Control Group.

Note. Due to some evidence of skewness within variables in each group, analyses were rerun using nonparametric Mann-Whitney U tests with no differences in results. The ImPACT data represented in this table are raw scores. Memory composite scores are presented as percentage correct; visual motor and reaction time are presented as time in seconds; impulse control and total symptoms are presented as a cumulative total. The Processing Speed scores are standard scores. A negative effect size score represents inferior performance by the concussion group; conversely a positive score represents a superior performance by the concussion group. SD = Standard Deviation; PSI = Processing Speed Index.

*Significant at the .05 level.

CI = Confidence Interval; B = logit coefficient; SE = Standard Error; LB = Lower Bound; UB = Upper Bound; FSIQ = Full-Scale IQ; No. = number.

***p < .001.

Sens = Sensitivity; Spec = Specificity; PPP = Positive Predictive Power; NPP = Negative Predictive Power.