Abstract
The Kimberley Region of Australia is an exceptional environment. It is three times the size of England but is inhabited by only 33,000 people. Its Aboriginal population speak in 30 different languages with English often as a third or fourth language, with low levels of literacy. They live in remote rural communities and small towns. In the Wet, the consistent temperature is around 38° to 40°C. Kimberley people are generally representative of more remote and distant-rural Aboriginal people who make up some 30% of the total Aboriginal population. However, around 70% of Australian Aboriginal people live in non-remote cities, urban settings or rural towns; the large majority of these are English speaking and literate with at least basic schooling. Aboriginal Australians, both remote and urban, face profound socio-economic and educational disadvantage, trauma and stress compared to the non-Indigenous population. From Australian Bureau of Statistics data, health outcomes are equally poor and life expectancy reduced in both remote and urban settings and poorer than those in other Indigenous cultures within developed nations.
In a previous paper, Smith and her colleagues found the prevalence of dementia by DSM-IV criteria to be 27% in the Aboriginal elderly aged over 65 years [1]. This estimate is five times higher than the overall Australian population after age standardisation. In this issue, they now report their findings on risk factors.
The study is a remarkable achievement in its own right. There must have been many difficulties in obtaining the findings. Each of these difficulties raises questions when it comes to their interpretation. Accurate case ascertainment required firstly the identification of the denominator of elderly individuals in communities where a person's age is often not known, conducting an appropriate clinical examination, and then obtaining clinically relevant information from informants. The latter might often be reluctant to portray older individuals as impaired in behaviour. There is then the crucial matter of applying diagnostic criteria that are valid for this group of Aboriginal elderly. Importantly, the authors have previously validated their screening instrument (the KICA Cog) for this remote Aboriginal population with low literacy and English often a second language, using clinical assessment by geriatricians and psychiatrists experienced in dementia diagnosis and they have used this methodology in the current study [2].
When it comes to measuring exposure to risk factors, the authors have appropriately limited this initial survey to key demographic, lifestyle and clinical factors using culturally adapted instruments. The study has therefore been a major undertaking in the epidemiology of dementia. We now examine the significance of the findings both on prevalence and risk factors. Firstly, the strikingly high prevalence estimate calls for interpretation. The study was apparently not constrained by having any hypotheses before data collection started. It is unknown whether the investigators expected to find higher or lower rates than in elderly Caucasians. Cases were ascertained in a two-phase design using their KIKA screening instrument. This was followed by reaching a diagnosis through clinical consensus, made by two specialists applying DSM-IV criteria. These criteria are now considered by some to have limitations for non-western populations because of the primacy accorded to memory impairment and their lack of operational definition. According to Prince et al. [3], this leads to much scope for unreliability. The alternative 10/66 criteria developed by Prince and his colleagues led to higher rates than DSM-IV in all of the seven low or middle income countries collated by Llibre Rodriguez et al: India, China, Cuba, the Dominican Republic, Venezuela, Mexico and Peru [4]. So the Kimberley finding, already high, may really be an underestimate of the true prevalence. Here it has to be pointed out that since prevalence is the product of incidence and duration of disease, a high prevalence could conceivably be due to low mortality after developing dementia. But it is improbable this would apply in the Kimberley elderly.
Next, the high prevalence is at odds with what has consistently been found in other populations with low literacy and a non-western life-style. In their influential paper on the global prevalence of dementia, Ferri et al. [5] reported rates of 1.6% in Africa, 1.9% in India and 2.7% in Indonesia, Thailand and Sri Lanka for individuals aged over 60 years. In the Ballibgarh study in rural India, where 73% were illiterate, Chandra et al. reported a rate of 1.4% in those over 65 years [6]. Hendrie et al. [7] found considerably lower rates in sub-Saharan Africa than most other world regions. The Kimberley finding of 27% is at least ten times higher than these non-Caucasian populations. This calls for an explanation of what might be raising the rate so markedly. It remains a possibility that it is due to eliciting erroneous clinical information in the first place, or to the use of inappropriate criteria. But such sources of error would have to be very powerful to account for the findings. If the estimate is in fact valid, the next task is to identify which of the risk factors, including but not limited to those examined, are proving detrimental to brain function.
For the risk factors, a number of methodological points need to be identified. Firstly, the analysis has perforce been carried out on prevalence data, not incidence. In an ideal situation the latter are always superior because it helps avoid confounding by factors related to survival after developing the disease. That is unlikely to be applicable here. Secondly, the number of cases is modest in size. Thirdly, the investigators did not differentiate the type of dementia, so that any one risk factor may not apply as much to one as to another type. Then we note that no information has been provided on family history. Importantly, the Apolipoprotein E status of the cases and those without dementia is not known. It has been suggested [8] that exposure of the ε4 allele to contemporary environmental conditions, including diet, could render it a susceptibility allele for both Alzheimer's disease and cardiovascular disease. That is, there might be gene- environment interaction.
Of the environmental risk factors identified, four merit particular attention, having both a high prevalence in this population and being potentially modifiable: no formal education (40%, OR 2.7 [1.1–6.7]); current smoker (35%, OR 4.5 [1.1–18.6]); previous stroke (9% OR 17.9 [5.9–48.7]); and head injury (51%, OR 4.0 [1.7–9.4]). A reduced level of education is a recognised risk factor for late life dementia [9]. Reduced access to education is well documented for both urban and remote Aboriginal people. This finding further supports the need for a major social and financial investment in education for Aboriginal children. Cigarette use as a risk factor for dementia is also unsurprising and has a potential link with the finding of previous stroke as a potent risk factor for dementia in this group. This is important for Aboriginal people as both remote and urban populations smoke at around twice the rate of non-Indigenous Australians; vascular risk factors and vascular disease including stroke, are major causes of premature mortality in mid-life in Aboriginal Australians and are major known, potentially remediable risk factors for late life dementia in non-Indigenous populations [10]. With their current increasing survival combined with poor mid-life vascular disease control, these risks are highly likely to apply to Aboriginal Australians. A further significant finding in this population is the association of dementia with head injury involving loss of consciousness (defined as ‘knocked out’) which was found in 51% of the study population. Traumatic brain injury with loss of consciousness has long been discussed as a potential risk factor for dementia in epidemiological studies; however, its prevalence in non-Indigenous populations has been low [11]. The role of head injury as a potential causal factor for the high rates of non-specified dementia found in this study certainly merits further study and intervention.
Ernest Gruenberg [12] said that the ultimate service of epidemiology is prevention. The need for prevention of dementia in the Aboriginal elderly is manifestly clear. Their numbers are likely to increase markedly, particularly if health care interventions in younger individuals are successful in coming decades. There is high exposure to established risk factors. In this matter it would be highly desirable in the first instance to attempt replication of the high prevalence rate in other Indigenous populations in Australia and in particular in urban/regional Aboriginal people who form a large majority and have equally poor health status as remote. At the same time, action is needed right away to reduce the risk factors identified by Smith and her colleagues.