Dunn et al (Reference Dunn, Holmes and Mullee2005) identified from the General Practice Research Database in the UK all cases of dementia between 1992 and 2002 (n=9954) and compared the number of prescriptions of lithium for individuals with this diagnosis with a control group without dementia (n=9374). They found that more patients with dementia (n=47, 0.47%) than controls (n=40, 0.43%) were exposed to lithium. We feel that this finding does not allow conclusions to be drawn as to whether lithium protects against or confers a risk for dementia, because it has been shown that patients with dementia have an increased risk of developing mania and depression (Reference Nilsson, Kessing and SorensenNilsson et al, 2002) and are thus more likely to receive treatment, including lithium. Conversely, affective disorders themselves seem to increase the risk for dementia.

In a series of studies based on data from the Danish psychiatric case register, Kessing et al found that 14% of elderly patients with bipolar disorder and 16% with unipolar disorder developed dementia (MMSE<24) compared with 3.4% of age-matched controls (Reference KessingKessing, 1998; Reference Kessing, Olsen and MortensenKessing et al, 1999). Even within a younger sample of psychiatric patients (approximate mean age 50 years), Kessing et al (Reference Kessing, Olsen and Mortensen1999) reported that people with bipolar disorder had the highest risk of receiving a diagnosis of dementia, followed by those with unipolar affective disorder, schizophrenia and neuroses. Thus, if affective disorders do increase both the risk of dementia and the likelihood of receiving lithium treatment, then owing to the sampling method used by Dunn et al one could expect to find more lithium treatment among elderly people with dementia. Dunn et al discussed this alternative explanation of their findings as a ‘reverse causation’ possibility.

Dr Terao mentions the possible effects of lithium on GSK-3 beta. We recently investigated the effects of lithium on the transcriptional regulation of GSK-3 beta and found a significant reduction of its expression in primary cultures of rat hippocampal neurons as well as a reduction in regional intracerebral expression in lithium-treated adult rats and in leukocytes of elderly patients undergoing chronic lithium therapy for bipolar disorder (details available from the authors). These observations suggest a mechanism for GSK-3 beta inhibition by lithium, which may influence the formation of both amyloid plaques and neurofibrillary tangles, the two neuropathological hallmarks of Alzheimer's disease.

We think that it is important to investigate further the potential protective effect of lithium in Alzheimer's disease, as this could represent a low-cost universally available strategy to reduce the prevalence.