Two recent studies failed to establish a relationship between the duration of untreated psychosis (DUP) and cognitive deterioration in first-episode patients (Reference Barnes, Hutton and ChapmanBarnes et al, 2000; Reference Norman, Townsend and MallaNorman et al, 2001). Both studies used the premorbid IQ (estimated using the National Adult Reading Test (NART)) minus the current full-scale IQ (measured using the Wechsler Adult Intelligence Scale (WAIS)) to measure cognitive deterioration. The validity of this approach to assessing cognitive deficit is open to question.

We examined DUP and cognitive deterioration in 42 individuals (mean age 22.3 years; s.d.=4.1) with first-episode schizophrenia (Reference Amminger, Edwards and BrewerAmminger et al, 2002). The revised version of the NART and WAIS (WAIS—R) were administered at clinical stabilisation and we have since taken the opportunity to apply the NART IQ minus WAIS—R full-scale IQ approach. Current IQ was higher than the estimated premorbid IQ in 38.1% of cases, suggesting an IQ increase.

The NART has been validated in older samples. We were therefore interested in the relationship between age at admission and IQ measures. NART IQ, but not WAIS—R full-scale IQ, was positively correlated with age at admission in our sample, (r=0.331, P=0.032). The WAIS—R ‘vocabulary’ sub-test, suggested to be a better estimate of premorbid IQ than the NART (Reference Russell, Munro and JonesRussell et al, 2000), had also no relationship with age. It is possible that the NART underestimates premorbid IQ in young people with schizophrenia.

Age-standardised WAIS sub-tests are another method to estimate cognitive deterioration (Reference Bilder, Lipschutz-Broch and ReiterBilder et al, 1992). Performance on ‘information’ and ‘vocabulary’ sub-tests are relatively stable, whereas the ‘digit symbol’ sub-test is sensitive to brain insult. Bilderet al's (Reference Bilder, Lipschutz-Broch and Reiter1992) deterioration index (DI),

\batchmode \documentclass[fleqn,10pt, legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \[\ \mathrm{DI}=\frac{[\frac{(\mathrm{Information\ +\ Vocabulary})}{2}-\mathrm{Digit\ Symbol}]}{[\frac{\mathrm{(Information\ +\ Vocabulary})}{2}]}\ \] \end{document}

is based on the principle that a larger discrepancy between an individual's best and poorest performance on cognitive functions suggests cognitive loss. We found longer DUP, male gender, higher NART IQ and younger age at depression to be independent significant predictors of the DI (Reference Amminger, Edwards and BrewerAmminger et al, 2002).

A cross-sectional test score (e.g. low-average full-scale IQ) cannot indicate deterioration on its own. In the absence of longitudinal data, indices reflecting decline from premorbid levels of functioning are required. Limitations of proxy methods need to be considered and studies which aim to validate measures of cognitive deterioration should be pursued.