Kay et al's paper (Reference Kay, Forster and Newens2000) on long-term survival in pre-senile dementia adds a useful and important contribution to this under-researched area. While acknowledging the difficulties faced in drawing valid conclusions from a non-neuropathologically confirmed study, there are several points of interest and concern not raised by the authors.

Pre-senile dementia is a heterogeneous group of disorders and the report that only 19 of 233 cases were not pre-senile dementia of Alzheimer type or pre-senile vascular dementia is a concern. The authors previously recognised that cases defined as Alzheimer's disease by clinical criteria alone may include conditions with non-Alzheimer type pathology, such as Pick's disease (Reference Newens, Forster and KayNewens et al, 1993), but felt this reflected only a small number of patients. However, recent evidence suggests that the frontotemporal dementia (FTD) may account for up to a quarter of patients presenting before the age of 65 (Reference Snowden, Neary and MannSnowden et al, 1996). Retrospective analysis of case notes using the NINCDS—ADRDA criteria (Reference McKhann, Drachman and FolsteinMcKhann et al, 1984) for diagnosing Alzheimer's disease may well include many FTDs, as the criteria for a diagnosis of probable Alzheimer's disease are also features of this subgroup.

The diagnosis was reportedly confirmed in a proportion as part of a case—control study, although there is a risk of selection bias by possible exclusion of the more behaviourally challenging uncooperative FTD patients.

The unrepresentative nature of the pre-senile vascular dementia group is acknowledged by the authors, and patients with mixed Alzheimer's and vascular pathology are also likely to be included in this vascular category. To date it is unclear as to the degree to which the two conditions coexist. As it is apparent that the Alzheimer's disease group may also be unrepresentative, the question begs to be asked, what groups are actually being compared? The overall suggestion that pre-senile Alzheimer's disease and vascular dementia have a similar prognosis needs to be taken in the context of these limitations and highlights the need for neuropathological studies in this area.