Authors' reply

We are pleased that Biswas & Chakrabarti highlight the strengths of our study design and large sample size, and consider our work a significant contribution to understanding psychological factors in bipolar disorder. We agree that it is important to consider potential confounders and therefore examined the effects of differences in illness duration and severity. Although there were some differences between our two patient groups on measures of illness severity and a small number of modest correlations between illness severity and cognitive style, covarying for these measures had no effect on our finding that those with bipolar disorder have fragile self-esteem and dysfunctional beliefs similar to those of people with unipolar disorder. We have not been able to examine the possible effects of pharmacoprophylaxis on cognitive style, but agree that this could be a target for future research.

We do not think we would have found differences in cognitive style between participants with bipolar and unipolar disorder if we had used a longer version of the Dysfunctional Attitudes Scale (DAS). The 24-item version used in our study was factor-analytically derived from the longer version and has improved robustness (Power et al, 1994). The ‘need for achievement’ and ‘dependency’ sub-scales of the 24-item DAS comprise items from the ‘perfectionism’ and ‘need for approval’ subscales of the 40-item DAS respectively.

We hope that future studies of cognitive style in people with mood disorder will build on the strengths of our study by using prospective longitudinal designs, systematically ascertained samples and perhaps implicit measures which cover other potentially interesting and clinically relevant cognitive traits such as goal attainment, attributions, self-representations and novelty-seeking.