Department of Psychology, University of Rhode Island, Kingston, RI, USA.
Department of Athletic Training, Florida International University, Miami, FL, USA.
Arch Clin Neuropsychol. 2021 May 21;36(4):561-569. doi: 10.1093/arclin/acaa074.
The most commonly used computerized neurocognitive test in concussion assessment and management, Immediate Post-concussion and Cognitive Testing (ImPACT), has demonstrated varying and sometimes concerning levels of test-retest reliability. This study aimed to further examine ImPACT's psychometric qualities and whether ceiling effects may suppress its reliability.
A total of 300 consecutively selected ImPACT score reports for students attending a secondary school between 2010 and 2015 were reviewed. Test-retest reliabilities for composite scores and subscales were computed using Pearson product moment correlations (r) and intraclass correlation coefficients. To examine the potential influence of ceiling effects, we conducted frequency analyses of scores falling at, or near, the maximum possible score.
A total of 92 score reports met inclusion criteria. Test-retest reliabilities ranged from 0.42 to 0.69 for composite scores and 0.19 to 0.71 for subscales. Subscales comprising the Verbal and Visual Memory composites evidenced the most prominent ceiling effects.
Obtained test-retest reliabilities were consistent with a large segment of the literature and add to concerns about ImPACT's reliability. Furthermore, at least in a select sample, this study identified sizeable ceiling effects that likely diminish the reliability of some composite scores. To mitigate the risk of false-negative errors on post-injury testing, ImPACT users should be cognizant of such ceiling effects. Supplemental, or alternative, approaches to protect against underestimating baseline functioning also warrant consideration.
在脑震荡评估和管理中最常用的计算机化神经认知测试——即时后发性和认知测试（ImPACT），其测试-重测信度表现出不同程度的可靠性，有时甚至令人担忧。本研究旨在进一步研究 ImPACT 的心理测量学特性，以及是否存在天花板效应会降低其可靠性。
回顾了 2010 年至 2015 年期间一所中学连续选择的 300 名 ImPACT 评分报告。使用皮尔逊积矩相关系数（r）和组内相关系数计算复合评分和子量表的测试-重测信度。为了研究天花板效应的潜在影响，我们对接近或达到最高可能分数的分数进行了频率分析。
共有 92 份评分报告符合纳入标准。复合评分的测试-重测信度范围为 0.42 至 0.69，子量表的测试-重测信度范围为 0.19 至 0.71。包含在言语和视觉记忆复合量表中的子量表表现出最明显的天花板效应。
获得的测试-重测信度与大部分文献一致，并增加了对 ImPACT 可靠性的担忧。此外，在一个特定的样本中，本研究确定了相当大的天花板效应，这可能会降低某些复合评分的可靠性。为了降低受伤后测试中出现假阴性错误的风险，ImPACT 用户应该意识到这些天花板效应。还需要考虑使用补充或替代方法来防止低估基线功能。
