Introduction

Dementia is often thought of as a condition of old age and although most cases are found in the elderly a significant number of people develop symptoms of dementia at a younger age. Patients with onset of symptoms below a certain age (usually set arbitrarily at 65) are said to suffer from “early onset dementia” or “presenile dementia”. The causes and classification of dementia in this age group are the same as in the more elderly population in that Alzheimer’s disease, vascular dementia, Lewy body dementia and frontotemporal dementia can all be recognised.

Study Design

We summarise the findings of studies reporting the prevalence of early onset dementia. We included studies that had determined prevalence rates of dementia in patients less than 65 years of age. Using a Medline and Embase search we found a number of studies using the search terms “Dementia/Prevalence/Epidemiology” or “Early onset dementia/Incidence/Prevalence/Epidemiology.” We followed this with a hand search of the references of these studies as well as any knowledge of any studies by the authors. To be included in the review studies needed to specify prevalence of dementia in subjects aged 65 or younger either looking specifically at this younger age group or as a easily identifiable subgroup of a larger study population. Papers that included the younger age groups but could not be easily determined from older ages were excluded. Those reporting only on incidence were also excluded. The initial database search produced 9 references, 5 of which were included in the review. A further 5 papers were identified by hand-searching the references of publications in the initial database search.

Results

The methodology and geography of the papers found reporting prevalence are summarised in table 1. Their key findings are summarised in tables 2a-2d which also give a breakdown of the prevalence in different “pre-senile” age groups, of different sub-types of dementia and any gender differences where given.

Table 1: Methods of studies giving prevalence of early onset dementia

A Known disease specific genetic mutation, neuropathological results from cerebral biopsy, or autopsy (top level diagnosis).

NINCDS/ADRDA criteria for Alzheimers, NINDS/AIREN criteria for vascular dementia, Lund and Manchester criteria for Lewy body and frontotemporal dementia, DSM-IV for alcohol related dementia (level 2 diagnosis).

DSM-IV criteria but not for one particular category (level 3).

Table 2a: Results summary of studies giving prevalence of early onset dementia

Table 2b: Results summary of studies giving prevalence of early onset Alzheimer’s dementia

Table 2c: Results summary of studies giving prevalence of early onset Fronto-temporal dementia

Table 2d: Results summary of studies giving prevalence of early onset vascular dementia

Study types, geography and methods

Most of the studies we found that reported on prevalence were performed in Western Europe with a spread between the UK (3 papers)1,2,3, Scandinavia (2)4,5 and mainland Europe (3).6,7,8,9 One paper reported on a study in Rochester, Minnesota.10

There was heterogeneity in both study design and in how cases were identified. In the case of rare diseases the usual methods of the field study rapidly reach their limits, since even expensive examinations of extensive samples of the population permit only unreliable frequency estimates due to the low number of illness cases which can be identified in the process. Thus in the three largest prevalence studies with a total of more than 13,000 persons under 65, for example, only a total of 29 dementing illnesses could be diagnosed, among them 8 cases of primary degenerative dementia.4,6,9 In order nevertheless to be able to estimate the illness burden, in countries with well developed care systems one resorts to identifying rare illnesses through contact with therapy centres. These can be termed registry studies. Contrary to the cases of late life, where frequently no clinical diagnosis is made and medical care of the demented is restricted to that of the GP, this appears to be a suitable method for the investigation of early onset dementia, since almost all afflicted are diagnosed at some time or other in the course of the illness by a specialist and avail themselves of the services of psychiatrists/neurologists, outpatient departments or specialised hospitals for diagnosis and treatment. Thus five of the studies were cross-sectional studies 1,2,7,8,10 and relied on various computer databases and coding systems to identify the majority of cases. Some of them went further by enquiring to a variety of sources in the community (such as GPs, nursing homes, CPNs, clinical psychologists and social services departments) that may be involved with and be aware of patients with dementia to identify them to the study groups. The Pitea Valley study recruited cases prospectively as they attended neurology clinics5 and Newen et al’s study in Northern England retrospectively reviewed case notes of patients who had had a diagnosis of dementia queried.3

Types of dementia

The majority of the papers looked at all cases of dementia whatever the exact aetiology or classification although many did give a breakdown of the prevalence of different subgroups of dementia. One study looked only at Alzheimer’s dementia8 and another looked purely at fronto-temporal dementia.7 Four of the papers we found were looking specifically for cases of dementia in people aged less than 65,1,2,3,8 whereas the remainder included people of all ages but included subgroup analyses allowing calculation of prevalence for those with “presenile” disease.4,5,6,7,9,10 In terms of definition of “presenile” this varied between papers. Most used a cut off of 65 years with either onset or diagnosis before this age required to be included. Campion’s study used a cut-off of 61 years.8

Results of Studies

The studies’ differing designs and breakdown of different dementia sub-types makes direct comparison difficult. Looking firstly at “all dementia” prevalence ranges from 38 to 420 per 100,000 of the population. This variation is likely to be due to the differing mix of dementia types and the relatively small number of cases which can skew results and give broad confidence intervals as discussed above. The higher figure of 420 comes from a paper where the age range was narrower (55-64) thus excluding younger age groups with lower prevalence which would otherwise skew the results.6 Harvey’s paper gave a breakdown of prevalence in different age categories below the age of 65 that indicated a rise in prevalence as age approaches 65.1 This is to be expected and this rising prevalence is likely to form a continuum with prevalence figures in “senile” dementia of onset after 65 years. It therefore follows that if you look at just the upper end of the pre-senile age range, as did Ott and colleagues, you will calculate a higher prevalence as you are only looking at the upper end of a skewed population and excluding a younger susceptible population with a lower prevalence.6 It is therefore unhelpful to compare this figure directly with those from other studies with a larger age range.

Alzheimer’s disease prevalence ranges from 15.1 to 153 per 100,000 of the population although the higher figure comes from a study with only 4 prevalent cases so is subject to inaccuracies in estimates of prevalence as discussed above.6 Again those papers that included a slightly broader age range for pre-senile dementia quoted lower prevalences due to the effect of skewing by including younger age groups.

Those studies that identified fronto-temporal dementia gave figures ranging from 4.0 (in the study looking purely at FTD)7 to 15.4 per 100,000.1

Five of the eight papers commented on difference in prevalence rates between the genders. Ratnavelli and colleagues found a male preponderance in the incidence of FTD but not other dementia subtypes2 and Campion found Alzheimer’s disease prevalence was higher in women.8 Harvey et al found a slight but non-significant excess in cases in men compared to women.1 The other three studies that commented on this found no significant gender differences.3,6,10 The Rochester study found all cases of presenile Alzheimer’s disease were in women but as the total number was only three this difference is unlikely to be statistically significant.10

Discussion of study quality

There was great variation in the type of population included in addition to the way they were sampled. Only two studies were “national” in that they sampled people throughout the countries of Finland and The Netherlands.4,7 The Finnish study used a sample of the population distributed throughout 40 areas of the country that was specifically selected to represent the Finnish population aged over 30. The Dutch study used the population of the Netherlands as a whole. All the other studies were performed in either a particular city or region within a country. It could be hypothesised that the “sub-national” studies may be less likely to represent the population of a country as a whole as they will not take into account regional variations. However as the case numbers are small any differences are unlikely to be significant so the sample size and methods are more likely to have a greater influence on the quality of the results. Most of the studies included both rural and urban populations although four were mainly urban based.1,6,8,10 None of the studies stated they excluded subjects in institutions although many made no mention of this factor.

In terms of the sample size, two studies did not state this number.3,7 Of the others the population eligible for inclusion ranged from 8000 to 426,710 in the Finnish and Rouen studies respectively.4,8 However it should be noted that the figure quoted for the Rouen study is the entire population of all ages, many of whom would not be “at risk” of dementia, whereas the Finnish study limited itself to those aged 30 or over. In the two field studies the response rates were 97% and 78% respectively for the Finnish and Dutch studies respectively which would normally be expected to give reasonably representative results.4,6

Methods of case ascertainment varied between studies. Two of the three field studies used screening methods involving tests of memory and intellectual function at first to identify those who may have dementia. They were then assessed further by a combination of neuropsychological testing, neuroimaging, blood tests and functional assessment to determine firm cases. As with any screening test sensitivity is unlikely to be 100% so a few cases may have been missed. However the cross-sectional/registry based studies rely on the fact that subjects have been in contact with either medical or care giving organisations meaning cases may have been missed if they had not yet come to the attention of such services. As stated above, this is less likely with younger populations than older ones as people are more likely to seek assistance and investigation for symptoms of dementia if it occurs in a younger patient rather than an older patient in whom many may just view it as part of the ageing process. However it is likely that sensitivity for case identification is greater in the field studies which most likely reflects the large step up in quoted prevalence figures in these two studies compared to the registry based studies. This also suggests there may be a large number of cases that are not coming to the attention of the medical/caregiving services.

Discussion

Epidemiological data for prevalence rates for early onset dementia is sparse. The majority of studies are European. Early onset dementia remains a rare condition with relatively low case numbers. The wide variation in rates across studies may be due to differing study design (case attainment, and diagnostic criteria) in addition to the sparsity of prevalence cases, which necessitates the study of vast underlying populations in order to reach an accurate true estimation.