Abstract:
Context: Neurocognitive testing is a recommended component of concussion assessment. However, use of these testing measures should be performed with caution, as the reliability of computerized neurocognitive assessment tests has been underexplored. Understanding the reliability of neurocognitive performance will ensure clinicians appropriately interpret test results. Objective: To determine the reliability of three computerized concussion assessment tests: Axon Concussion Assessment Test (Axon), CNS Vital Signs, and the Immediate Postconcussion Assessment and Cognitive Test (ImPACT) across two weeks. Design: Repeated Measures. Setting: Clinical research laboratory. Patients or Other Participants: Fifty healthy, physically active college-aged students (age: 21.4±1.5 years; 13 males, 37 females) participated. Those with a concussion history in the previous 6 months, learning disabilities, attention disorders, and psychiatric conditions were excluded. Intervention(s): Participants completed four separate testing sessions. The first two testing sessions were 24-48 hours apart. The last two testing sessions were two weeks later (15.2 ± 2.9 days), and these two sessions were also 24-48 hours apart. Participants completed two of the three computerized tests: one at the first and third session, and a different one at both the second and fourth session (Axon: n = 17; CNS Vital Signs: n = 16; ImPACT: n = 17). The testing order was counterbalanced across sessions. Main Outcome Measure(s): To determine practice effects across the two test sessions for each computerized outcome measure (scores for 5 Axon cognitive domains, 8 CNS Vital Signs domains and 5 ImPACT composite domains), a series of dependent t-tests were employed. Intraclass correlation coefficients (ICC2, 1) were calculated to determine the reliability of performance across two weeks for all outcome measures. Results: There were no significant differences across the two testing sessions on any Axon domains, any ImPACT composites, or 6-of-8 CNS Vital Signs domains (P > 0.05). Participants performed better at time 2 (111.56 ± 8.88) compared to time 1 (99.33 ± 20.25) on CNS Vital Signs processing speed (t1,15 = -2.88; P = 0.01). Participants performed better at time 1 (106.11 ± 11.19) compared to time 2 (96.56 ± 16.34) on CNS Vital Signs visual memory (t1,15 = 2.40; P = 0.03). The ICC2,1 values ranged as follows: Axon from 0.24 (Attention) to 0.71 (Processing Speed); CNS Vital Signs from 0.23 (Visual Memory) to 0.62 (Psychomotor Speed); and ImPACT from 0.09 (Verbal Memory) to 0.63 (Visual Motor Speed). Conclusions: Overall, performance on many computerized neurocognitive assessments seems to be stable over time. Some scores, such as visual and verbal memory, appear to be less reliable, and may lead to misinterpretation of test results and premature or delayed return to play decisions. Clinicians should be aware of the reliability of the specific measures they use in clinical practice in order to ensure proper interpretation of test results and appropriate management decisions.