A ccurate and early clinical diagno sis of dementia with Lewy bodies (DLB) is essential for optimal management so that pitfalls such as severe neuroleptic sensitivity¹ can be avoided, a trial of cholinesterase treatment can be instigated,² and parkinsonian symptoms³ can be appropriately treated, weighing up the balance of psychiatric and motor symptoms and fall risk. In practice it has often been difficult to differentiate DLB from other forms of dementia, particularly Alzheimer's disease. Several previous studies have suggested that DLB and Alzheimer's disease can be differentiated on the basis of visual and visuospatial tasks.^4,5 The tasks most easily applicable to initial diagnosis are drawing tests (such as the clock drawing).⁶ However, this approach ignores what is one of the more striking features of DLB: fluctuating cognition. Operationalised criteria for the clinical diagnosis of DLB (which do consider a broad range of symptoms) have achieved high levels of specificity (> 90%)⁷; in many studies sensitivity has been as low as 50%.⁸ Probably the major obstacle to accurate diagnosis has been the difficulty of identifying fluctuating cognition; for example, two inter-rater reliability studies have indicated that agreement between expert clinicians is barely better than chance.^9,10

A body of work focusing on these issues has validated two clinical rating scales¹¹ and indicated that fluctuation of reaction time on attentional tasks¹² and variability in spectral frequency assessed by electroencephalography¹² can also be useful methods for identifying and quantifying fluctuating cognition, which seems to be closely linked to impairments of consciousness.

In the paper by Doubleday et al¹³ (this issue, pp602–607) tackles these issues in an innovative way. The authors developed a standardised qualitative instrument based on observations of the patient during the course of the testing session. Information is obtained regarding a number of helpful domains including general inattention, distraction, and intrusions. Of particular interest, general inattention, distractibility, and environmentally cued intrusions were all significantly more common in DLB than in Alzheimer's disease, the latter occurring only in patients with DLB. In addition to the potential diagnostic value, this may be helpful in facilitating an understanding of the types of processing errors that are prominent in these patients. Perseveration and mental set shifting were also significantly more impaired in patients with DLB and they were more likely to confabulate.

This assessment method has been used as part of the standard diagnostic protocol that has achieved a high level of accuracy in 200 people with dementia at postmortem examination, although it is unclear how many of these people had neuropathologically confirmed DLB. Further work is needed to examine the inter-rater reliability and concurrent validity of the instrument, and comparisons between DLB and other dementias would also be helpful. In particular, patients with vascular dementia often experience some degree of fluctuating cognition and may be more prone to difficulties with set shifting and perseveration. Diagnostic discrimination between DLB and vascular dementia can be difficult and it would be useful to know whether the proposed assessment method can also contribute to this diagnostic issue. Therefore, although the method needs to be developed a little more, the work presents a new and useful approach that is very clinically applicable and that may be of considerable value in the assessment of patients with DLB.