A recent review of the prevalence of dementia in all world regions revealed a particular paucity of data from the Middle Eastern region (Ferri et al., Reference Ferri2005). Iran is one of the largest countries in this region, with a population of 70 million, of whom 5.2% are aged 65 years and over and where life expectancy is already 71 years (www.amar.ir – Iranian Government Statistics site). Our best estimates suggest that there may be nearly 200,000 people living with dementia in Iran at the moment (Ferri et al., Reference Ferri2005). The 10/66 Dementia Research Group developed and validated a cultural- and education-fair dementia diagnosis in 26 centers worldwide (Prince et al., Reference Prince, Acosta, Chiu, Scazufca and Varghese2003). We have now assessed the feasibility and validity of a Farsi version of the 10/66 dementia diagnostic system in Ekbatan, Tehran, Iran.

Sixty people with dementia and 60 similarly aged controls who were free of dementia were selected from registers held by the Iran Alzheimer's Association (IAA) and invited to participate. Those with dementia had already sought help from the IAA; the diagnosis (according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994) criteria) was reconfirmed by an independent clinician. Controls were selected from a partial register of older residents, military veterans' associations, and volunteer organizations. Another clinician (NN) administered the interviews comprising the Geriatric Mental Status (GMS; Copeland et al., Reference Copeland, Dewey and Griffith-Jones1986), the CERAD 10 word list learning with delayed recall (Ganguli et al., Reference Ganguli, Chandra and Gilbey1996), the Community Screening Instrument for Dementia (CSI-D, informant and participant versions; Hall et al., Reference Hall1993). All instruments were translated to Farsi, back translated to English and assessed for acceptability. Informed consent was obtained for all participants and the study approved by the ethics committee of the Memory and Behavioral Neurology Department (MBND) at Tehran University of Medical Sciences. Nearly all interviews were conducted privately at the IAA, and the interviewer (NN) was masked to case/control status where possible. For each cognitive test we calculated the area under the curve (ROC) and their specificity and sensitivity at the optimum cut-off point. We also calculated the sensitivity, specificity and agreement (kappa) of GMS/AGECAT dementia, probable dementia using the CSI-D cognitive test and cognitive test and informant interview combined, and the 10/66 Dementia algorithm that incorporates information from GMS, CSI-D and CERAD 120 word list against the gold standard of the clinician's DSM-IV dementia diagnosis.

The age distribution of dementia cases (mean 74.9 years, SD 6.4, range: 65–89) and controls (mean 72.0 years, SD 5.6, range: 65–87) was similar. By design, the male to female ratio was 1:1 for both cases and controls. The area under the ROC curve was high for all the cognitive measures (between 0.95 and 0.99), with the CSI-D DFSCORE (combining information from the cognitive test and the informant report) performing best of all (see Table S1 available online as supplementary material attached to the electronic version of this letter at www.journals.cambridege.org/jid_IPG). The optimum cut-off points were similar to those previously recommended for the CSI-D (Hall et al., Reference Hall1993) and to those identified in the previous 10/66 international pilot study. The 10/66 diagnostic algorithm performed better than GMS diagnostic assessments, with a kappa of 0.97 and sensitivity and specificity of 98.3%.

The study is limited by the purposive selection of participants, and the difficulty of maintaining masking to case status. Furthermore, controls needed to be healthy enough to travel to the AAI centre for assessment. Hence, the very favorable sensitivities and specificities recorded in this study may not be achievable in population surveys. Nevertheless, this pilot study suggests that the 10/66 assessment protocol is feasible and practical, and that the 10/66 dementia diagnosis is likely to be as valid in this setting as in the many other regions and cultures in which it has already been tested. The Farsi version of these measures will be made available to download from the 10/66 study website at www.alz.co.uk/1066.