We write to correct some misconceptions evident in the recent editorial by Ferrier & Thompson (Reference Ferrier and Thompson2002). Previously, we reported impairment in accuracy measures on recognition memory tasks and increased latencies on executive tasks in patients with bipolar disorder in remission (Reference Rubinsztein, Michael and PaykelRubinsztein et al, 2000). Ferrier & Thompson (Reference Ferrier and Thompson2002) argue that the cognitive impairment observed in our study may have been confounded by the effects of ‘residual’ symptoms. As yet there is no generally accepted ‘cut-off’ for what constitutes remission. We devised rigorous criteria to define remission based on a patient's own view of his or her illness, that of their psychiatrist and a structured interview. We excluded patients with scores of ≥8 on both the Hamilton Rating Scale for Depression (HRSD) and Young Mania Scale (YMS). These rating scales were devised to rate symptom severity in patients with an affective disorder and not for use in normal control subjects. Our average reported score on the HRSD was 2.1 (s.e.m.=0.5) and on the YMS it was 0.8 (s.e.m.=0.4). Thus, very few residual symptoms were evident and these scores certainly do not support any concern that patients had residual depression or mania.

Although the rationale for using such scales in controls is dubious, for the sake of argument we have reanalysed our data reported in Rubinsztein et al (Reference Rubinsztein, Michael and Paykel2000) using a partial correlation analysis, as in Clark et al (2002), to control for differences observed on the HRSD (we did not rate control subjects using the mania scale) on the tests that showed significant impairment by analysis of variance (ANOVA). We still find significant impairment on both the visual recognition memory tasks and on latency measures from the one-touch Tower of London planning task (see Table 1).

Table 1 Results of partial correlation analysis on tests in which ANOVAs were significant

	Dependent variable	Partial correlation coefficients	P
Pattern recognition memory	Proportion correct	0.41	0.02
Spatial recognition memory	Proportion correct	0.31	0.07
Delayed matching to sample	Proportion correct	0.35	0.04
One-touch Tower of London	Response time	-0.42	0.02
Affective shifting task	Response time	0.04	0.81

These findings suggest that there are trait impairments in accuracy of visual recognition memory and slower responses on a planning task in bipolar remission. Importantly, impairments of memory and learning have been consistently observed in a number of other recent studies where rigorous diagnostic criteria for remission were applied (e.g. Reference Van Gorp, Altshuler and ThebergeVan Gorp et al, 1998; Reference Krabbendam, Honig and WiersmaKrabbendam et al, 2000; Reference Cavanagh, van Beck and MuirCavanagh et al, 2002) as well as in a recent unpublished study (L. Clark, personal communication, 2002) that showed that verbal recall was still impaired following partial correlation for residual symptoms. The presence of significant impairments on executive tasks in bipolar remission has been more variable and may depend on clinical factors or the specific neuropsychological test paradigm employed. The precise functional significance of the cognitive impairment in bipolar remission needs to be examined further but may well impact on response to psychological and drug treatments. Cognitive symptoms could in fact be among the most sensitive indicators of incomplete remission.