Diagnostic accuracy of CompCog: reaction time as a screening measure for mild cognitive impairment

Abstract

Background Reaction time is affected under different neurological conditions but has not been much investigated considering all types of mild cognitive impairment (MCI). Objective This study investigated the diagnostic accuracy of CompCog, a computerized cognitive screening battery focusing on reaction time measurements. Methods A sample of 52 older adults underwent neuropsychological assessments, including CompCog, and medical appointments, to be classified as a control group or be diagnosed with MCI. The accuracy of CompCog for distinguishing between the two groups was calculated. Results The results from diagnostic accuracy analyses showed that the AUCs of ROC curves were as high as 0.915 (CI 0.837-0.993). The subtest with the highest sensitivity and specificity (choice reaction time subtest) had 91.7% sensitivity and 89.3% specificity. The logistic regression final model correctly classified 92.3% of individuals, with 92.9% specificity and 91.7% sensitivity, and included only four variables from different subtests. Conclusions In summary, the study showed that reaction time assessed through CompCog is a good screening measure to differentiate between normal aging and MCI. Reaction time measurements in milliseconds were more accurate than correct answers. This test can form part of routine clinical tests to achieve the objectives of screening for MCI, indicating further procedures for investigation and diagnosis and planning interventions.

Establishing Reference Values for a New Computerized Cognitive Function Test Program for Children

“…Measuring reaction time is a good screening tool for normal aging and MCI, and it is more accurate than measuring correct answers in adults aged >60 years of age [ 30 ]. Previous studies established a correlation between processing speed and overall cognitive performance [ 31 ].…”

A Perspective: Challenges in Dementia Research

“…This results in large test variances for individuals and reduces the precision of the assessment. A second solution is to start with a number of tests including quantitative psychophysical tests (that may be less dependent on education and language than traditional paper and pencil tests) [ 19 , 20 ] and use statistical techniques to find a limited combination of these test elements, which is minimally dependent on the factors that influence cognition in cognitively healthy people. This is not enough, however, because the ideal test should not only yield results that were similar in cognitively healthy people, but also be sensitive to the different phenotypes seen in cognitive disorders of differing etiology.…”