Cognitive screening for dementia and delirium in sub-Saharan Africa :development and validation of cognitive screening tools for identification of dementia and delirium in older adults by non-specialist health workers in sub-Saharan Africa

Abstract

Dementia and delirium are recognised to be common inter-related disorders affecting older people worldwide, the majority of whom live in low and middle income countries (LMICs). In sub-Saharan Africa (SSA), research data on dementia and delirium in older adults are currently very few, despite evidence of a growing older population. A major factor limiting research is the lack of appropriately validated cognitive screening instruments. Existing instruments appear educationally biased in SSA, where illiteracy amongst older people remains high. Specialist clinicians are few, and human resources generally limited, resulting in a need for brief tools suitable for use by non-specialists with limited time. The IDEA six-item screen was developed from cognitive screening data collected during a community based door-to-door prevalence study of dementia conducted in Tanzania in 2010. This work aimed to formally validate the IDEA screen for identification of dementia and delirium in older adults in a variety of clinical settings in Tanzania. Additional aims were to identify potential educational bias, consider feasibility of use of the IDEA screen by non-specialist clinicians and consider utility and effectiveness of additional screening items to improve diagnostic accuracy in some settings. Validation studies were conducted in the Kilimanjaro region of Tanzania as follows. 1) Consecutive admissions to medical wards of a Government hospital aged 60 and over (n=97); 2) A randomised sample of individuals aged 60 and over attending a free-of-charge Government outpatient clinic (n=108); 3) A randomised rural community sample aged 60 and over presenting for dementia screening (n=466), and 4) A consecutive sample of 507 individuals aged 60 and over admitted to medical wards of a tertiary referral hospital, supported by detailed informant interview and follow-up where necessary (against blinded consensus DSM-IV dementia and DSM-5 delirium criteria). A culturally appropriate assessment of Instrumental Activities of Daily Living (IDEA-IADL) was developed at a workshop for primary healthcare workers. Validation against DSM-IV dementia criteria took place in a community sample (n=417) of adults aged 60 and over presenting for dementia screening, used alone and in addition to the IDEA six-item screen. Screening was conducted by trained research nurses, occupational therapists and clinical officers in hospital settings, and by nurses and primary health care workers in rural settings. Diagnostic accuracy of the IDEA six-item screen for major cognitive impairment was high in the pilot hospital settings (IP AUC 0.917, OPD AUC 0.919). Accuracy in the lower prevalence community setting was lower (AUC 0.846), but improved by the addition of the IDEA-IADL (AUC 0.896). In the tertiary referral hospital sample AUC was 0.874 for major cognitive impairment (DSM-IV delirium or Identification of dementia in sub-Saharan Africa Doctoral Statement - 3 - DSM-IV dementia) and 0.866 for delirium, but a substantial number of participants were unable to complete the screen (83/507) and the IDEA did not differentiate dementia and delirium. A combined tool developed through modelling outperformed the IDEA (AUC 0.94 95% CI 0.92-0.97) and had the advantage of being much shorter, and being possible in most individuals regardless of illness severity. These studies conclude that the IDEA six-item screen is a brief and culturally appropriate cognitive screening tool with a high degree of diagnostic accuracy for identification of dementia in clinical and community settings in Tanzania. Use of the screen by non-specialist health workers is feasible, and the screen does not appear educationally biased. Diagnostic accuracy is improved by additional use of a functional assessment tool, the IDEA-IADL in the community. In hospital settings with a relatively high prevalence of delirium, the IDEA six item screen is accurate in identification of major cognitive impairment (dementia or delirium) but cannot differentiate these conditions. An alternative screening method for identification of delirium in this setting is presented. Further validation work in other settings in sub-Saharan Africa is needed. Since the proportion of individuals aged 50 and over living with HIV in SSA is projected to increase from one in seven to one in four by 2030, assessment of clinical utility in HIV-associated neurocognitive impairment is also needed, and this is the focus of ongoing work.