An evaluation of methods of estimating premorbid ability

• Main Author: Moore, John Webster
• Published: University of Edinburgh 1999
• Subjects: 155
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.657966

The detection and quantification of acquired intellectual impairment is an activity of cardinal importance in contemporary clinical neuropsychology. This is complicated by inter-individual and intra-individual variability, whereby simply comparing a patient's test performance with the relevant normal values would not provide information about any change in functioning that may have taken place. A major aim of the present study was to evaluate regression models for the estimation of premorbid ability. The use of the demographic variables, a present ability measure (the NART) and the combination of these variables was evaluated. The vast majority of previous research on the estimation of premorbid ability has used summary IQs as the criterion variables. In the present study the ability of these methods to estimate individual WAIS-R subtest scores was evaluated. The models were tested in a large sample (N=245) of healthy participants which was broadly representative of the adult UK population in terms of the distributions of age, sex, and social class. Thus this study replicates and extends previous work in the US in which demographically based regression models were developed from the WAIS-R standardisation sample. In the present study it was found that the models combining NART and demographic variables were most successful in estimating variance in subtest performance, estimating from 30% to 71% and from 28% to 52% of the variance in verbal and non-verbal subtest performance respectively. Models estimating from demographic information alone and from NART alone are also presented and evaluated for particular clinical circumstances where either NART or demographic information is not available. For comparative purposes models estimating WAIS-R factor scores and summary IQs from NART and demographic variables were also constructed. It was found that the models combining NART and demographic information were most successful in estimating variance in indices of WAIS-R performance. They accounted for 68%, 27% and 41% of the variance in V, PO and A/C, and for 71%, 38% and 65% of the variance in VIQ, PIQ and RSIQ.