Estimation of premorbid intelligence and executive cognitive functions with lexical reading tasks
Introduction: Estimation of premorbid function is essential to accurate assessment of cognitive impairments in clinical neuropsychology and behavioral neurology, and has numerous research applications. However, current methods are rudimentary and imperfect. We explored how lexical tasks can be best used to accurately and precisely estimate intelligence and executive functions.Methods: We studied lexical word pronunciation, lexical decision, and stem-completion naming in the estimation of cognitive ability, in samples of healthy adults (n = 143), and patients with cognitive impairment due to neurological illness (n =15). Cognitive assessments included intelligence (WAIS-IV), episodic memory, and eight tests of executive functioning, including Theory of Mind.Results: When examined at the group level, single word pronunciation was particularly robust in the presence of cognitive impairment in patients with dementia. However, as a case series, patients showed idiosyncratic patterns of preservation of lexical skills including on tests of pronunciation, lexical decision and stem-completion naming. All of these tasks were highly correlated with IQ scores in a non-clinical sample, suggesting that they could be used as estimators of premorbid intelligence. Simulated impairments in non-clinical controls suggested that the median score from the three different tasks had the highest correlation with, and provided the most accurate and precise estimates of, intelligence, and was also the least sensitive to impairment. Finally, we show that these methods also predict executive functions, in particular, strong correlations were found for proverb interpretation, phonemic/semantic alternating verbal fluency, and working memory span performance. Conclusions: Several lexical tasks are potentially useful in predication of pre-illness cognitive ability in patients with neurological or psychiatric illness. However, due to the heterogeneity of impairments between patients, estimation of premorbid levels could be improved by the use of the median estimated values from multiple tests. This could potentially improve diagnostic accuracy and quantification of neuropsychological impairments.