It is clinically important to differentiate dementia with Lewy bodies (DLB) from other types of neurodegenerative dementias because of the prospect of better response to drugs such as cholinesterase inhibitors as well as the risk of development of neuroleptic sensitivity. However, usefulness of the commonly used definition of DLB (Reference McKeith, Galasko and KosakaMcKeith et al, 1996) remains to be established in community and primary care settings. The findings of the Islington study (Reference Stevens, Livingston and KitchenStevens et al, 2002) suggest that DLB is a clinically significant type of dementia in the community.

We had an opportunity to look at the prevalence of DLB in a small but representative community sample of patients meeting DSM—IV criteria for dementia (American Psychiatric Association, 1994) in Kerala, a southern state of India. These cases were identified using a novel case-finding method (Reference Shaji, Arun Kishore and LalShaji et al, 2002). We identified 33 cases of dementia in the study area with a population of 1979 aged >60 years. A psychiatrist assessed all the subjects at their residence and assigned a DSM—IV diagnosis. Seventeen people with dementia (52%) met criteria for Alzheimer's disease, 12 (36%) were diagnosed as having vascular dementia, while in 4 (12%) the dementia was felt to be secondary to other conditions such as chronic obstructive pulmonary disease, cardiac failure, Parkinson's disease and subdural haematoma. Three patients (9%) met criteria for probable DLB and one (3%) met criteria for possible DLB when the consensus criteria (Reference McKeith, Galasko and KosakaMcKeith et al, 1996) were applied. The combined prevalence of DLB was therefore 12%. All these patients had earlier met the criteria for Alzheimer's disease as per DSM—IV.

On retrospective application of the separate ‘clinical’ diagnostic criteria used in the Islington study to our case records, only two of our patients met those criteria, reducing the proportion of patients with DLB to 6%. This is strikingly similar to the figure reported in the Islington study. One limitation of these ‘clinical’ criteria is that they are too restrictive a definition of DLB. If visual hallucinations are indeed such an important diagnostic marker, then the consensus criteria themselves could be modified by making visual hallucinations an essential symptom for the diagnosis of DLB. We need to examine whether such modifications will improve the recognition of DLB in clinical practice.