Reliability of a computerized neuropsychological test
Resch, Jacob Earl
MetadataShow full item record
Since 2005, computerized platforms have been scrutinized for lack of psychometric data to support their use in the clinical setting. Poor to moderate reliability coefficients have been reported for commercially available computerized platforms. The purpose of this study was to replicate and extend the research addressing the reliability of computerized neuropsychological tests while controlling for effort, time of day, and mood state. One-hundred and fifty two (N=152) healthy, college aged students were recruited to participate in this study. Each participant completed the Profile of Mood States - Brief Form (POMS-B), Green’s Word Memory Test (WMT), and the ImPACT computerized concussion assessment test at three clinically relevant time points: baseline, day 45 and day 50. ImPACT calculates six composite scores utilized to determine cognitive decline post-concussion. The POMS-B is a measure of six mood states and a composite total mood disturbance score and was utilized to measure mood state at the time of each session. Green’s WMT was utilized to assess effort at each time point. Any baseline data which was considered invalid by ImPACT or was incomplete was removed from subsequent data analysis. Forty-five (n=45) and one hundred and eight (n=108) participants were included in studies one and two, respectively. For study one, intraclass correlation (ICC) values were calculated for each ImPACT composite score. The ICC values for each composite score were slightly higher than previously reported in the literature at the same clinically relevant time points but still fell below what is considered clinically acceptable. For the second study, although one or more mood states were significantly correlated with one or more ImPACT composite scores, correlations coefficients were weak to moderate and did not appear to influence test performance. Values for the Green’s WMT exceeded suggested criteria for good effort and improved over time. The test-retest reliability coefficients reported in this study are slightly higher than those previously reported. Differences may be due to the delivery of one computerized test at each time point. The added evidence that mood state does not influence test performance leads the authors to believe that the poor to moderate reliability coefficients are due to systematic rather than random error.