National Adult Reading Test performance in established dementia.

Demographic and neuropsychological test data from patients with dementia of Alzheimer or multi-infarct type were analysed in an examination of the validity of the National Adult Reading Test (NART) as an index of premorbid ability in established dementia. The discrepancy between premorbid IQ predicted firstly using the NART and secondly using demographic variables entered in a recognised regression equation was calculated. This discrepancy correlated strongly with measures of current level of functioning, even after exclusion of patients with apparent dysphasia or dyslexia. The results indicate that NART performance is clearly influenced by severity of dementia, and that use of the NART as a comparator in estimating the severity of established dementia will lead to systematic underestimation of severity as a function of that severity. Estimates of premorbid function based on demographic variables may be preferable in this context.