Early dementia diagnosis is currently actively promoted in developed countries, with a view to preparing those affected, and their carers, and to ensure timely access to support and care when it is needed (Department of Health, 2009
). In low and middle income countries (LMIC), dementia remains to a large extent a hidden problem. Although the symptoms and syndrome are widely recognized and named, it is considered to be a normal part of ageing, not a medical condition (Cohen, 1995
;Patel and Prince, 2001
; Shaji et al.
). Family members rarely seek help, and primary care doctors rarely come across cases (Patel and Prince, 2001
; Prince et al.
). The treatment gap in south India was recently estimated to be as high as 90% (Dias and Patel, 2009
). Nevertheless, dementia is an important source of carer strain (Prince et al.
). In LMIC, community-based career education and training interventions have recently been shown to be particularly effective in reducing career strain and depression (Dias et al.
; Gavrilova et al.
). The World Health Organization is preparing evidence based guidelines for management of dementia by non-specialists in LMIC with a view to scaling up treatment and reducing the treatment gap (World Health Organization, 2008
), and evidence-based packages of care have recently been proposed (Prince et al.
). Effective case identification by non-specialists is an essential element (Prince et al.
Studies in high income countries (HIC) show that only a fifth to a half of cases of dementia are routinely recognised and documented in primary care case note records; with a median proportion from six studies of 39% (Olafsdottir et al.
; Valcour et al.
; Lopponen et al.
; Boustani et al.
; Wilkins et al.
). However, evidence suggests that primary care physicians and nurses can, if specifically prompted to do so, make a dementia diagnosis with reasonable accuracy, using their knowledge of the patient, available case note information and their own routine assessments in the limited time available during a typical consultation (O'Connor et al.
; Cooper et al.
). Similarly, in LMIC community healthcare workers could, with a few hours training, identify dementia in the community with a positive predictive value of 66%, based solely upon their prior knowledge of older people from their routine outreach work (Shaji et al.
; Ramos-Cerqueira et al.
). The discrepancy between what non-specialists might, and do in practice achieve is explained partly by limited help seeking. It may also be that non-specialists either do not attend to dementia, or are not motivated to confirm and record the diagnosis when the possibility occurs to them.
Population screening for dementia is not considered cost-effective even in HIC (National Collaborating Centre for Mental Health, 2007
). However, indicated screening of primary care attendees, based upon prior suspicion of possible dementia, can promote case detection. Research in developed countries has highlighted the short period of time available for each consultation in primary care, and the need accordingly for very brief assessments, ideally taking 5 min or less to complete (Brodaty et al.
). Screening involves cognitive testing of the older person or informant interview for a history of cognitive and functional decline. Sometimes both approaches are combined in a single test. The Mini-Mental State Examination (Folstein et al.
) is widely used in HIC, and adapted versions have been developed for use in many LMIC (Ganguli et al.
; Xu et al.
; Castro-Costa et al.
). However, it takes 10 min to administer and is prone to educational and cultural bias (Black et al.
; Ng et al.
). A brief version of the MMSE, the ‘six item screener’ performed as well as the full MMSE in clinical and population samples in the USA (Callahan et al.
). The three tools that are brief enough, and at least as valid as the longer MMSE (General Practitioner Assessment of Cognition (GPCOG (Brodaty et al.
)), the Memory Impairment Screen (MIS (Buschke et al.
)) and Mini Cog (Borson et al.
)) have only been validated in HIC (Brodaty et al.
). Reviewing their content, none is suitable for use in low education LMIC settings. MIS requires reading ability, and GPCOG and Mini Cog include clock drawing tasks that are not generally feasible for those with less than 5 years education (Lessig et al.
). The adaptation required to make them suitable would, in effect, be equivalent to the development of a new assessment. The recently developed Vellore Screening Instrument for Dementia seems promising, but with 10 cognitive and 10 informant items it may be too long for routine use. Furthermore, its only community validation to date was on a sample of only 101 participants, three of whom were diagnosed with dementia (Stanley et al.
The Community Screening Instrument for Dementia (CSI ‘D’) (Hall et al.
) is by far the most extensively validated dementia screening assessment, across a variety of LMIC. It combines culture and education-fair cognitive testing of the participant (32 items) and an informant interview enquiring after the participant's daily functioning and general health (26 items) into a single predictive algorithm. It was developed and first validated among Cree American Indians (Hall et al.
; Hendrie et al.
), and further validated and used in population-based research among Nigerians in Ibadan and African-Americans in Indianapolis (Hendrie et al.
). It has also been validated in Jamaica and among white Canadians in Winnipeg, (Hall et al.
). The CSI-D test score distributions among those with dementia and controls, and the degree of discrimination provided were remarkably consistent across these five very different cultural settings (Hall et al.
). CSI ‘D’ was further validated in the community among 2885 persons aged 60 and over recruited in 25 centres in India, China and South East Asia, Latin America and the Caribbean and Africa, as part of the 10/66 Dementia Diagnosis Protocol (Prince et al.
). The inclusion of the informant interview significantly improved upon the predictive power of the CSI ‘D’ cognitive test component (Hall et al.
; Prince et al.
). However, requiring around 30 min to administer, it is too long for routine use in primary care. It could, in principle, be shortened. Adoption of a simple scoring method will also add to its value as a primary care screening assessment. The purpose of the secondary analyses presented in this paper, using existing data from the 10/66 Dementia Research Group pilot studies and population-based studies, is to explore the potential for deriving much briefer cognitive and informant scales from the full CSI-D and to carry out initial assessments of their likely validity.