Bulletin of the World Health Organization

Prevalences of dementia and cognitive impairment among older people in sub-Saharan Africa: a systematic review

Angelique Mavrodaris a, John Powell b & Margaret Thorogood c

a. Warwick Medical School, University of Warwick, Gibbet Hill Road, Coventry, CV4 7A, England.
b. University of Oxford, Oxford, England.
c. University of the Witwatersrand, Johannesburg, South Africa.

Correspondence to Angelique Mavrodaris (e-mail: A.Mavrodaris@warwick.ac.uk).

(Submitted: 15 February 2013 – Revised version received: 21 June 2013 – Accepted: 22 June 2013 – Published online: 20 August 2013.)

Bulletin of the World Health Organization 2013;91:773-783. doi: http://dx.doi.org/10.2471/BLT.13.118422


The prevalence of age-related health problems is becoming an important public health concern as proportions of older individuals in populations worldwide grow.1 Dementia is one of the major causes of disability in older people.2 It is a complex syndrome characterized by global and irreversible cognitive decline that is severe enough to undermine daily functioning.3 Dementia is a chronic illness that arises from an interplay of genetic, environmental and behavioural factors, with severe adverse influences on social and physical activities and quality of life. In cognitive impairment and mild cognitive impairment, the cognitive deficit is less severe than in dementia and normal daily function and independence are generally maintained. It is a chronic condition that is a precursor to dementia in up to one third of cases.3

Cognitive impairment and dementia are increasing globally and are predicted to increase proportionately more in developing regions.1,46 Projections indicate that by 2050 the number of individuals older than 60 years will be approximately 2 billion and will account for 22% of the world’s population. Four fifths of the people older than 60 years will be living in developing countries in Africa, Asia or Latin America.7 It is estimated that 35.6 million people are currently living with dementia worldwide and that the number will nearly double every 20 years, reaching 115.4 million in 2050, with the majority living in developing countries.8 Of the total number of people with dementia worldwide, 57.7% lived in developing countries in 2010 and a proportionate increase to 70.5% by 2050 is anticipated.8 Consequently, the health and social burden of cognitive impairment and dementia will rise dramatically in these regions.9 Developing countries, including those in sub-Saharan Africa, are currently undergoing a demographic and epidemiological transition and the impact of population ageing in sub-Saharan Africa will increasingly augment the burden of noncommunicable and degenerative diseases in this region.

Few studies to determine the prevalence of dementia have been conducted in sub-Saharan Africa.10,11 Early studies12 showed a lower prevalence than in Europe11,13 and the United States of America, where approximately 6.2% and 8%, respectively, of people aged 65 years or older are reported to have dementia.1416 The aim of this study was to perform a systematic review of the literature to obtain the best available estimates of the prevalences of cognitive impairment and dementia in sub-Saharan Africa and to identify gaps in current research.


A systematic review of studies reporting the prevalences of dementia and cognitive impairment among older black Africans in sub-Saharan Africa countries was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.17

Search strategy

We searched the following databases between February and May 2011: PubMed (search range, 1950 to week 18 of 2011); the Web of Knowledge, which includes the Web of Science Conference Proceedings Citation Index (1970 to week 18 of 2011), the BIOSIS Citation Index (1969 to week 18 of 2011), Journal Citation Reports (1997–2008) and CAB Abstracts (1973 to week 18 of 2011); EMBASE (1947 to week 18 of 2011); ASSIA (1987 to week 18 of 2011) and PsycNET (1894 to week 18 of 2011). The search terms used included variations across the following broad terms: epidemiology, prevalence, incidence, cognition, cognitive impairment, dementia, Alzheimer disease, sub-Saharan Africa, names of individual sub-Saharan African countries, elderly and older people. Searches incorporated MeSH (medical subject heading) terms or equivalents, exploded terms and text word terms (Appendix A, available at: http://www2.warwick.ac.uk/fac/med/research/hscience/research/dementia/). We checked the reference lists of retrieved articles for further relevant studies.

Study selection

Studies were included if they satisfied all of the following criteria: (i) the study had a cohort, case–control or cross-sectional study design and reported population-level data, defined as data from population surveys or patients identified in primary or community care samples; (ii) the study was limited to black African adults who were older than 50 years or described as “elderly” or “old” or, if the age range of subjects was broader, reported data separately for those older than 50 years; (iii) the study population resided in sub-Saharan Africa or, if the geographic area of residence was broader, reported data separately for those residing in sub-Saharan Africa; and (iv) the study reported at least one measure of cognitive impairment or clinical outcomes relevant to cognitive decline (i.e. dementia, Alzheimer disease, amnesia and/or confusion) and at least one measure of the frequency of cognitive impairment (i.e. prevalence and/or incidence).

No language restrictions were applied. All titles identified were screened independently for potential relevance by at least two authors and abstracts were obtained if the title was considered relevant by at least one author. Abstracts were screened in the same way. Finally, full texts of the remaining papers were assessed for eligibility and when two authors disagreed on inclusion a decision was made in discussion by all three authors.

Data extraction and analysis

Data were extracted from each paper independently by two authors, using a data collection form developed by the authors and tested on a sample of papers (Appendix B, available at: http://www2.warwick.ac.uk/fac/med/research/hscience/research/dementia/). Details about the study design, participant characteristics, analysis methods and findings, and authors’ conclusions were entered into the form.

For studies in which several related papers were identified, the papers were evaluated collectively as one study for data extraction.12,15,16,1823 In one paper, two studies were reported, and we considered these studies separately.24 Because of considerable heterogeneity in methods (sampling and study design), population samples (age, health, socioeconomic status, geographical region and cultural background) and screening or diagnostic measures, we did not pool the data but report our findings in narrative form.

Two authors independently evaluated the quality of the studies, using a standardized checklist. The following factors were assessed: whether the sample was representative of the target population, whether screening and diagnostic tools produced reliable and valid measures of psychiatric outcome and key concepts and whether the analysis accounted for special features of the sampling design and included confidence intervals.25


Study characteristics

We retrieved 281 references from PubMed, 1081 from the Web of Knowledge, 55 from CAB Abstracts, 1192 from EMBASE, 43 from ASSIA and 4 from PsycNET. After removal of duplicates, 2320 references remained. Initial screening yielded 86 papers for full-text review and one additional reference was identified from searching reference lists of included papers. (Fig. 1). One additional reference was identified while searching reference lists of included papers. Our systematic review yielded 19 papers associated with 11 studies.

Fig. 1. Flow diagram for a systematic review of the literature to select studies evaluating the prevalences of dementia and cognitive impairment in sub-Saharan Africa
Fig. 1. Flow diagram for a systematic review of the literature to select studies evaluating the prevalences of dementia and cognitive impairment in sub-Saharan Africa

The studies reported findings for 10 413 participants. All studies were cross-sectional and were published between 1995 and 2011, with approximately two thirds published between 2005 and 2011 (Table 1). The studies were conducted in Benin, Botswana, the Central African Republic, the Congo and Nigeria; most occurred in Nigeria, in Ibadan, Dunukofia, Jos, Uwan and Zaria. Five studies included participants from both rural and urban settings, one study included only urban participants and five studies included only rural participants. All studies included participants aged 65 years and older; two studies also included participants aged 60–64 years. The mean age of the participants ranged between 72.0 and 76.1 years. All studies included both female and male participants and 10 studies reported prevalence data stratified by sex.

Three studies reported the prevalences of both cognitive impairment and dementia, with data for 1518 participants analysed.24,26 Five studies reported only the prevalence of dementia12,15,16,1823,28,30,31,33 and three reported only the prevalence of cognitive impairment, with data for 5164 and 4097 participants, respectively, analysed.27,29,32

Table 2 shows the cognitive screening tools and diagnostic processes used in the studies and indicators of the methodological quality are shown in Table 3. Only three studies documented evidence that confirmed the reliability of the tools.12,15,16,1823,30,32 Three studies did not report confidence intervals (or information necessary to calculate them)27,29,32 and one did not report a confidence interval but provided enough information for this to be calculated.33 One study did not document sufficient information regarding appropriate probability sampling.29 One study documented uncertainties regarding cultural validity of the screening tools.27

Dementia prevalence


A study from Djidja, a rural area, included 502 participants and reported a prevalence of 2.6% (Table 1).26 The Community Screening Interview for Dementia (CSI-D) was used for detection of dementia. Diagnosis was confirmed by clinical assessment and participant-based interview; clinical dementia was diagnosed according to DSM-IV criteria and probable or possible Alzheimer disease was based on National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer Disease and Related Disorders Association II (NINCDS-ADRDA II) criteria. Analysis on the basis of dementia subtype revealed that the prevalence of Alzheimer disease was 2.2% and accounted for 53.8% of dementia diagnoses in this population, while vascular dementia contributed 7.7%. A further 30% of dementia cases were documented as possible dementia.

Central African Republic and Congo

The dementia prevalence was 8.1% among 496 participants in a rural region of Bangui, Central African Republic, and 6.7% among 520 participants from an urban region in Brazzaville, Congo.24 Cognitive screening was performed using the CSI-D and 5-Word Test, adapted and presented in local languages. Diagnosis was confirmed by neurological examination and further psychometric testing in participants suspected of dementia (defined as a CSI-D score of < 25.5 of 30 or a 5-Word Test score of < 10 of 10).

Both studies measured the prevalence of dementia subtypes. According to findings of neurological examinations, 82.5% of participants with dementia in Bangui had probable Alzheimer disease and 17.5% had vascular dementia, while in Brazzaville the proportions were 68.6% and 31.4%, respectively.


Five studies reported prevalence data for dementia. A small study of 322 participants conducted in rural Zaria used a battery of cognitive tests similar to that used in the 1995 study from Ibadan described below and reported a prevalence of 2.8%.28 Two studies from Ibadan were of a similar size but conducted approximately 10 years apart. These reported prevalences of 2.3% in 1995 and 10.1% in 2006. Differences between the studies included use of a single validated screening tool, the adapted 10-Word Delay Recall Test (10-WDRT), to define probable dementia in the 2006 study30 and a combination of screening tools (including the 10-WDRT) and clinical assessment, to define dementia in the 1995 study.12,15,16,1823 The 2006 study also covered a wider region, including rural areas, while the earlier study was conducted in an inner-city region.

A study from Jos, in central Nigeria, reported a prevalence of 6.4% among 280 participants.31 This study used the CSI-D but there was no clinical assessment to confirm the diagnosis. A small study of 164 participants aged 60 years or older conducted in Uwan described common mental disorders. No cases of dementia were found. This study used the Self-Reporting Questionnaire (SRQ-24) and the Geriatric Mental State Schedule (GMS) as screening tools but the GMS had not been validated in Nigeria.33

Two studies examined subtypes of dementia. Yusuf et al. confirmed clinical cases of dementia based on the fulfilment of diagnostic criteria in both the International classification of diseases, tenth revision (ICD-10), and the Diagnostic and statistical manual of mental disorders, fourth revision (DSM-IV). DSM-IV criteria were then used to diagnose Alzheimer disease and ICD-10 criteria were used to diagnose vascular dementia.28 Clinical criteria were used to diagnose other subtypes. The prevalence of Alzheimer disease was 1.9%, which accounted for six of the nine participants with dementia; of the remaining three, two had vascular dementia and one had fronto-temporal dementia. Among patients with dementia in the study from Ibadan, Alzheimer disease was diagnosed in 64.3% and vascular dementia was diagnosed in 28.6%.12 Clinical diagnoses were confirmed using DSM-IV, ICD-10 and clinical criteria developed by the NINCDS-ADRDA II.

Cognitive impairment prevalence


In a study of 502 participants in the rural community of Djidja, researchers adapted the cognitive section of the CSI-D to allow for a lack of temporal orientation and a high rate of illiteracy, such that a score of 30 (rather than 33) was the highest possible (Table 1). The researchers also had considerable difficulties in identifying informants to complete the informant section of the CSI-D and therefore adapted the criteria for cognitive impairment. Cognitive impairment was defined as a poor performance in the CSI-D cognitive section (defined as a score of < 25.5 out of 30) or a poor performance on the 5-Word Test. Use of this adjusted instrument revealed a prevalence of cognitive impairment with no dementia (CIND) of 10.4%.26


Researchers screened 372 participants aged ≥ 60 years for several chronic diseases. Cognitive impairment was assessed using an adapted but non-validated version of the Mini–Mental State Examination (MMSE), which omitted sections that required writing or reading and adapted other sections to the local context. A 9% prevalence of cognitive impairment was reported without measures of uncertainty, with no distinction made between cognitive impairment with or without dementia.27

Central African Republic and Congo

Studies conducted in Bangui and Brazzaville reported CIND prevalences of 25% and 18.8%, respectively.24


A study from Dunukofia screened 914 participants with three memory tests (CSI-D and 10-word immediate and delayed recall). A total of 11.8% of participants had just one memory test indicating cognitive impairment, whereas cognitive impairment was detected in 9.9% by two or more memory tests.29

A study from Uwan, described above, used the SRQ-24 and the GMS to screen 164 participants older than 60 years and 34 (20.7%) reported forgetfulness, which may have been indicative of cognitive impairment.33

The Ibadan study group also studied cognitive impairment in 2000 in both rural and urban regions. They found a CIND prevalence of 6.3% among 2487 participants who were screened using the CSI-D tool and were assessed clinically. The sample used for this study included some of the participants in the 1995 study of dementia described above, as well as additional participants. The study also followed up 87 of the 152 patients with CIND and found that after 2 years 16.1% had developed dementia, 58.6% still had a diagnosis of CIND and 25.3% had normal cognitive function.32


To our knowledge, this is the first systematic review of the prevalence of dementia and cognitive impairment in older populations in sub-Saharan Africa. Eleven studies were identified and all had a cross-sectional design. Prevalences varied widely between countries and by publication date. The predominant factors associated with a higher prevalence of dementia were older age and female sex. Older age was strongly associated with dementia and cognitive impairment in all studies and similar associations have been reported globally.34 Female sex was a risk factor for both outcomes in five studies12,24,30,31 and has been confirmed in many international studies.34 The majority of studies were from Nigeria and western and central Africa, with only one from southern Africa and none from eastern Africa. We are aware of another study currently being conducted in a rural community of 2000 households in Bloemfontein, South Africa, but to date only a pilot study has been completed, with prevalence of 6.4% for dementia diagnosed by DSM-IV criteria.35

These studies have several limitations. Dementia involves a change in function and cognition but in a cross-sectional study such change is difficult to determine and assessment is reliant on the reports of individuals close to the individual. Survivor bias may have influenced prevalence data for both dementia and cognitive impairment, since individuals who are more vulnerable to cognitive impairment and poor health would be less likely to survive into old age in sub-Saharan Africa.

The generalizability of these studies is also limited. The more recent estimates of dementia prevalence remain lower than those reported in Europe (6.2%; 4% for Alzheimer disease) and the United States of America (8%; 6% for Alzheimer disease),11,1316 but this comparison is flawed because the age structure of the population older than 65 years in sub-Saharan Africa and Europe will differ. In sub-Saharan Africa, there is a higher proportion of individuals aged 66–80 years and a smaller proportion aged 81 years and older, leading to a lower overall prevalence of dementia, which is predominantly a disease of older people. Cognitive impairment prevalences of 3% to 19% among people older than 65 years in other parts of the world (range: 3–19%) are comparable with those reported by studies in our review.36 Studies of migrant African origin populations now living in Europe or North America have reported higher prevalences of cognitive impairment (range: 8–34%) but the majority of these studies included African-Caribbean participants or did not clearly distinguish the ethnic origins of study subjects. 37,38 Further research of first-generation sub-Saharan Africa migrant populations may provide more accurate data and insight into environmental factors associated with dementia.

There are challenges to assessing cognitive change and mental health in sub-Saharan Africa. Social impairment is not as highly associated with cognitive impairment in African communities, compared with communities in developed countries. The social functions and participation of elderly individuals in African culture differ from those found in Western cultures and factors associated with normal functioning in old people also differ.32 This may be because the DSM-IV and ICD-10 screening instruments require certain levels of social impairment to confirm clinical dementia and thus may lead to under-diagnosis in sub-Saharan Africa. The move from participant-based to informant-based cognitive tests has the potential to diminish the effects of education level and cultural biases because the informant reports direct changes in cognition on the basis of their knowledge of the subject’s prior cognitive and functional status. The CSI-D combines both approaches and incorporates a direct cognitive assessment and it has been accepted as a culturally adaptable screening tool in some African countries. Further development and refinement of the CSI-D is needed and future studies of prevalence should use this or some other validated screening tool. The 10/66 Dementia Research Group has developed evidence-based diagnostic tools that would facilitate inter-cultural research.39

Large cross-sectional surveys in both rural and urban populations from more countries in sub-Saharan Africa would aid health-care planning. Longitudinal studies examining incidence and predictors and factors modifying the rate and severity of cognitive decline specific to populations in this region are also needed. Qualitative analyses investigating cultural perceptions of dementia from the perspectives of the community, caregivers and individuals with dementia would offer insight into access issues and improvements in health care.

While still experiencing high mortality and fertility rates and heavily burdened by infectious disease, the population of sub-Saharan Africa is ageing and a rise in noncommunicable chronic diseases is inevitable. With no known cure or preventative intervention, dementia and cognitive impairment are set to be one of the biggest public health challenges in sub-Saharan Africa in the 21st century.


We thank Karen Rees for her assistance with statistical methods. Margaret Thorogood is also affiliated with Warwick Medical School, University of Warwick.


AM held an Academic Clinical Fellowship post which was funded by the United Kingdom of Great Britain and Northern Ireland National Institute for Health Research (NIHR). The views expressed are those of the authors and not necessarily those of the United Kingdom National Health Service, the NIHR or the United Kingdom Department of Health. The sponsor had no role in study design, data collection, data analysis, data interpretation or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Competing interests:

None declared.