Monitoring cognitive changes: psychometric properties of six cognitive tests

Repeated neuropsychological assessments are often used to monitor change in cognitive functioning over time. Thus, knowledge about the reliability and stability of neuropsychological tests and the effects of age and IQ is of paramount importance. In this study we document, for six cognitive tests: test-retest reliabilities, practice effects, reliable change (RC) indices corrected for practice, and the impact of premorbid IQ and age.

A sample of 188 normal adults (aged 40-70 years) were administered, on two occasions, one or more of the following tests: the Graded Naming Test (GNT), the Silhouettes Test, two tests of verbal fluency, the Modified Wisconsin Card Sorting Test, and a new test of speed and attention (the Symbol Digit Test). There was a 1-month interval between assessments. At first assessment, all participants were administered the revised National Adult Reading Test (NART).

The test-retest reliability of the tests ranged from very good (the GNT and Silhouettes Test) to moderate (verbal fluency tests and Symbol Digit Test) and to poor (Modified Card Sorting Test). Significant, although modest, practice effects were found on all tests. RC indices were generally large except for the Graded Naming Test and the Silhouettes Test. Premorbid IQ scores significantly correlated with performance on all the tests, the exception being semantic fluency. Age only correlated with the Silhouettes Test and the new Symbol Digit Test. Neither NART IQ nor age correlated with practice effects.

The psychometric properties of the GNT and Silhouettes Test indicated that they are useful tools for monitoring even small cognitive changes. In contrast, the verbal fluency tests and the new Symbol Digit Test are only suitable for monitoring large changes in performance. The Modified Card Sorting Test is an unreliable tool for monitoring 'executive' functions.