Objective: To review the literature on the reliability of the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT).
Design: Systematic review of the relevant literature in PubMed, CINAHL, and PSYCHINFO. Studies were evaluated using the STROBE instrument and custom developed items.
Results: Search yielded 5 943 articles. Ten studies met the inclusion criteria and were reviewed. With the exception of processing speed, all composite scores consistently exhibited poor to moderate reliability (ie, intraclass correlation coefficient <0.80). When considering 2 time points, participants who were misclassified as experiencing a "reliable change" in any score ranged between 5% and 26% for verbal memory, 2.2% and 19.6% for visual memory, 4% and 24% for processing speed, and 4% and 23.2% for reaction time.
Conclusions: The Pearson r correlation coefficient and average measures intraclass correlation coefficient may be inappropriately utilized to examine the reliability of ImPACT scores. Given the poor to moderate reliability of most ImPACT scores, clinicians should be cautious when ImPACT is used as a criterion for medical clearance to return to play after concussion. Because of its widespread use in concussion-related clinical research, researchers must exercise due diligence when utilizing ImPACT to evaluate outcomes after concussion or to validate other outcome measures.