By prospectively verifying euthymia and controlling for the effect of residual affective symptoms Goswami et al (Reference Goswami, Sharma and Khastigir2006) make a significant contribution to the existing evidence on cognitive impairments in euthymic patients with bipolar disorder. However, they did not define euthymia and the diagnosis of DSM–IV bipolar I disorder, verification of euthymia and exclusion of current and past psychiatric illness or substance use disorders in patients and controls were made without structured assessments. Controls were relatives of participating patients. In addition, exclusion criteria make no mention of birth injuries, the handedness of patients and whether patients had received electroconvulsive therapy. All these factors influence results of tests for cognitive function (Reference Ferrier and ThompsonFerrier & Thompson, 2002).

As the Schedule for Assessment of Psychiatric Disability assesses marital and occupational functioning, details of the patients’ marital or occupational status would have helped to better interpret the data. There is also no mention of the duration of illness (in Table 1, p. 368, duration spent in episodes is erroneously labelled as duration of illness). This variable has implications for the generalisability of findings.

A measure of the reliability and validity of the modified Kolakowska battery is not provided. The use of more systematic and better-validated instruments such as the Cambridge Neurological Inventory (Reference Chen, Shapleske and LuqueChen et al, 1995) and more than one rater to reduce assessment bias would have allowed better characterisation of neurological soft signs. About 92% of healthy controls in the current study had neurological soft signs. This unusually high prevalence could reflect the inappropriateness of the control group.

In the Rey Auditory Verbal Learning Test, significantly lower scores on lists A1–A5 were taken to infer a reduction in verbal memory. However, there was no difference between patients and controls for lists A6 and A7. The percentage of words retained between trials A5 and A7 would provide a purer index of retention (Reference Thompson, Gallagher and HughesThompson et al, 2005) and would help to better interpret the data.

In the future, meta-analyses of existing data and studies involving assessment of cognitive function and neuroimaging in euthymic patients with bipolar disorder should help elucidate a profile of cognitive cognitive deficits and their underlying neurobiological bases.