Epidemiology of neurodegenerative diseases in sub-Saharan Africa: a systematic review

Background Sub-Saharan African (SSA) countries are experiencing rapid transitions with increased life expectancy. As a result the burden of age-related conditions such as neurodegenerative diseases might be increasing. We conducted a systematic review of published studies on common neurodegenerative diseases, and HIV-related neurocognitive impairment in SSA, in order to identify research gaps and inform prevention and control solutions. Methods We searched MEDLINE via PubMed, ‘Banque de Données de Santé Publique’ and the database of the ‘Institut d’Epidemiologie Neurologique et de Neurologie Tropicale’ from inception to February 2013 for published original studies from SSA on neurodegenerative diseases and HIV-related neurocognitive impairment. Screening and data extraction were conducted by two investigators. Bibliographies and citations of eligible studies were investigated. Results In all 144 publications reporting on dementia (n = 49 publications, mainly Alzheimer disease), Parkinsonism (PD, n = 20), HIV-related neurocognitive impairment (n = 47), Huntington disease (HD, n = 19), amyotrophic lateral sclerosis (ALS, n = 15), cerebellar degeneration (n = 4) and Lewy body dementia (n = 1). Of these studies, largely based on prevalent cases from retrospective data on urban populations, half originated from Nigeria and South Africa. The prevalence of dementia (Alzheimer disease) varied between <1% and 10.1% (0.7% and 5.6%) in population-based studies and from <1% to 47.8% in hospital-based studies. Incidence of dementia (Alzheimer disease) ranged from 8.7 to 21.8/1000/year (9.5 to 11.1), and major risk factors were advanced age and female sex. HIV-related neurocognitive impairment’s prevalence (all from hospital-based studies) ranged from <1% to 80%. Population-based prevalence of PD and ALS varied from 10 to 235/100,000, and from 5 to 15/100,000 respectively while that for Huntington disease was 3.5/100,000. Equivalent figures for hospital based studies were the following: PD (0.41 to 7.2%), ALS (0.2 to 8.0/1000), and HD (0.2/100,000 to 46.0/100,000). Conclusions The body of literature on neurodegenerative disorders in SSA is large with regard to dementia and HIV-related neurocognitive disorders but limited for other neurodegenerative disorders. Shortcomings include few population-based studies, heterogeneous diagnostic criteria and uneven representation of countries on the continent. There are important knowledge gaps that need urgent action, in order to prepare the sub-continent for the anticipated local surge in neurodegenerative diseases.

SSA population (estimated total 2.074 billion) will be aged 60 years and above, which in absolute number will translate into four times the 2010 estimates, and correspond approximately to 156.7 million people [2].
Population ageing is considered a global public health success, but also brings about new health challenges in the form of chronic diseases including cardiovascular diseases, cancers, as well as neurodegenerative disorders. A characterization and updated picture of the latter conditions in SSA is particularly important in view of a) the ongoing demographic transition and the resulting surge in the prevalence of neurodegenerative diseases in SSA; b) the successful roll-out of antiretroviral therapies in the region and the potential, yet unknown impact of long-term survival with HIV infection and related treatments on the occurrence of neurodegenerative disorders [3]; and c) lastly, the need for reliable data for health service planning. Recently, there have been efforts to summarize existing data for conditions like Parkinson disease (PD) [4,5] dementia [6,7] or amyotrophic lateral sclerosis [8], but not for other common neurodegenerative disorders, while there are suggestions of possible African distinctiveness in their occurrence and features [9].
We systematically reviewed the published literature on common neurodegenerative disorders and HIV-related neurocognitive impairment among sub-Saharan Africans, with the objective of describing their main features as well as clinical and public health implications.

Data sources
We searched MEDLINE via PubMed, and the French database 'Banque des Données en Santé Publique' (BDSP www.bdsp.ehesp.fr) for articles published until February 2013. In addition we searched the database of the 'Institut d'Epidemiologie Neurologique et de Neurologie Tropicale' (IENNT). We used a combination of relevant terms to search (in English for PubMed and in French for BDSP and IENNT), which are presented in Additional file 1 (except for IENNT searches for which we used 'neuroepidemiologie' and other themes referring to neurodegenerative diseases). Two evaluators (AL and JBE) independently identified articles and sequentially (titles, abstracts, and then full texts) screened them for inclusion ( Figure 2). For articles without abstracts or without enough information in the abstract to make a decision, the full text, and where necessary supplemental materials, were reviewed before a decision was made. We supplemented the electronic searches by scanning the references lists of relevant publications, and identifying their citations through the ISI Web of Science, and by hand-searching all issues of the African Journal of Neurological Sciences. Disagreements were solved by consensus or review by a third investigator (APK).

Study selection
We included studies conducted in a country of the SSA region ( Figure 1) that reported on the following neurodegenerative diseases among adults: Alzheimer's disease, fronto-temporal dementia, Lewy body dementia, vascular dementia, cortico-basal degeneration, multi system atrophy, Parkinson's disease (PD), amyotrophic lateral sclerosis (ALS), Huntington disease, cerebellar degeneration, and HIV-related neurocognitive impairment. We made no restriction by study design. We excluded duplicate publications, review articles, studies conducted exclusively in pediatric populations, studies conducted exclusively on migrant populations of African descent living out of the continent. Figure 2 shows the study selection process. We provide a rigorous appraisal of the overall data and the epidemiological studies in particular, and make recommendations regarding future approaches to measurement, notwithstanding the challenges involved in such undertakings.

Data extraction, assessment, and synthesis
Two reviewers (AL and JBE) independently conducted the data extraction from included studies. We extracted data on study settings, design, population characteristics, measures of disease occurrence (incidence and/or prevalence), and risk factors for the various conditions examined. Given the diversity of neurodegenerative pathologies and the heterogeneity of populations assessed, we did not use a particular framework for the assessment of the quality of studies. However, whenever population-based studies and hospital-based studies had been conducted for a condition, we relied more on the conclusions of population-based studies to address relevant questions, and appropriately reported the results. We conducted a narrative synthesis of the evidence.

Results
The study selection process is shown in Figure 2. A total of 4049 citations were identified through MEDLINE, the IENNT database and BDSP searches; 337 abstracts were evaluated in detail and 214 full-text publications reviewed. The final selection included 144 publications reporting on Parkinsonism (20 studies), dementia (49 publications), HIV-related neurocognitive impairment (47 publications), Huntington disease (19 studies), amyotrophic lateral sclerosis (15 studies), cerebellar degeneration (4 studies) and Lewy body dementia (1 study). These studies were published between 1955 and 2012, with about 50% conducted in only two countries: Nigeria and South Africa.

Parkinson disease, other Lewy body diseases and fronto-temporal dementia
Twenty studies reported on Parkinsonism (Table 1), including five community-based and sixteen hospitalbased. Four were case-control in design and all the others were cross-sectional studies, including reviews of medical records. These studies were conducted in seven countries including Nigeria (ten studies), South Africa (four studies), Tanzania (two studies), Ethiopia, Ghana, Cameroon and Zimbabwe (one studies each). The number of participants with PD ranged from two to 32 and the prevalence from ten to 235/100,000 in community-based studies. The number of participants with Parkinsonism ranged from four to 397, and the prevalence of Parkinsonism varied from 0.41 to 7.2% of neurological admissions/consultations in hospital-based studies. The proportion of men among those with PD ranged from 53 to 100%, and age ranged from 30 to >100 years. Age at the clinical onset of the disease ranged from 17 to 90 years. The clinical types of the disease were largely dominated by Parkinson disease (38 to 100%).
The most commonly used tool to diagnose PD was the UKPDS Brain bank criteria and population-based (hospital-based) prevalence for the studies that applied those criteria ranged from 40 to 235/100,000 (11 to 69.4/1,000 neurological consultations). In general risk factors were not investigated across studies, although one study found that 38% of patients with Parkinsonism had atherosclerosis and 8% had encephalitis [18].
We found three cases of Lewy body dementia in a retrospective study in Nigeria, and one case in a retrospective study in Senegal representing respectively 1.2/ 100,000 of admission over a period of 10 years [30] and 7.5/1000 of participants in a specialized memory clinic [31].
Risk factors for dementia were reported in 14 publications. The following were associated with an increased risk of dementia: age (twelve publications), female sex (five publications), low body mass index (three publications), anxiety/depression (three publications), hypertension (three publications), social isolation (two publications), lifetime history of alcohol consumption, elevated total-or LDL cholesterol in those without Apo E ε4 (one publication), low socio-economic status, history of stroke and family history of dementia (one publication). The following characteristics were inversely associated with dementia: living with others, use of non-steroidal antiinflammatory drugs and absence of Apo E ε2. Some risk factors were more strongly related to the disease. These include age, which increased the risk of dementia by five to 16% across groups [34,43], but this effect was much higher after the age of 60 years, more than 100% increase risk especially after the age of 75 [46,50,51,55,66,67]. Female sex, low level of education (<6 years), rural residence and family history increased the risk of dementia by >100% [34,43,46,55,56,66].

HIV-related neurocognitive impairment
Fifty-one hospital-based studies (47 publications) reported on HIV-related neurocognitive impairment (Table 3), of which ten were case-control, six cohort and 31 crosssectional. These studies were conducted in 14 countries including South Africa (14 studies), Uganda (eight studies), Nigeria (six studies), Zambia and Kenya (four studies each), Cameroon and Democratic republic of Congo (three studies each) Ethiopia and Malawi (two studies each), Central African Republic, Botswana, Guinea Bissau, Tanzania and Zimbabwe (one study each). A total of 33 out of the 47 selected publications were published during the last 5 years and only 7 before 2000. The absolute number of participants with HIV-related dementia ranged from 0 to 396, with a prevalence ranging from 0% to 80%.
The diagnostic tools used to identify HIV-related dementia were variable, making comparison between studies less reliable. However, the International HIV Dementia Scale (IHDS) [89,95,97,105,[107][108][109][110]112,113,120,121] and the Sloan Memorial Kettering scale [86,89,90,98] were frequently used. Studies that used the IHDS reported a prevalence ranging from 21.1 to 80%. The mean/median age of participants ranged from 31 to 40 years for those with HIV-related dementia, and men represented 25% to 56% of this group. In the nine studies that investigated etiological factors, the identified determinants of HIVrelated dementia were: low level of CD4 count (four studies), low level of education, and advanced age (three studies), comorbid psychiatric conditions (two studies each), advance clinical stage (two studies), male sex, HIVsubtype and duration of disease (one study each). The most commonly reported risk factors of HIV associated dementia were the level of CD4 count [89,97,112,120,121] and the clinical stage of disease [97,121].

Amyotrophic lateral sclerosis and cerebellar degeneration
Fifteen studies (12 retrospective, 2 cross-sectional and 1 case-series) ( Table 4) including 13 hospital and two community-based studies on amyotrophic lateral sclerosis (ALS) have been conducted in 9 SSA countries including Nigeria (four studies), Senegal (three studies), Ethiopia (2 studies), Zimbabwe, Kenya, South Africa, Sudan, Cameroon and Ivory coast (one study each). The number of participants with ALS ranged from two to 73. Two community-based studies provided a prevalence of 15/100,000 and 5/100,000 respectively in Nigeria [19] and in Ethiopia [122]. Five hospital-based studies provided prevalence figures: between 0.2 and 8.0/1000 of all neurologic consultation/admission [16,21,[122][123][124][125][126]. The method of ascertainment of ALS was variable across studies, but electromyography was done in four of the fifteen studies included [125][126][127][128][129]. The proportion of men among those with ALS was 57.6 to 100%. The age of those with ALS ranged from 12 to 84 years. When provided, the age at the clinical onset of ALS ranged from 12 to 71 years and the time to diagnosis from 3 months to more than 15 years. In general, risk factors for ALS were not investigated across studies.
One retrospective study in Nigeria reported on two cases (a 32 year old male and a 42 year old female) of cerebellar degeneration among 2 · 1 million admissions over a period of 25 year [14]. One study in Rwanda reported on a family of 33 members, with 15 (including Table 3 Overview of studies on HIV-related dementia and risk factors in sub-Saharan         eight men, age at onset 12-49 years) having type 2 spino-cerebellar ataxia [134]. A study in Mauritania reported on 12 cases of cerebellar degeneration-based on clinical criteria, including 9 familial cases (including 7 men, aged 3 to 29 years) and 3 apparently sporadic cases (all men, aged 8 to 50 years) [135]. Another clinic-based study of paraplegia in Senegal reported on 7 cases of spino-cerebellar degeneration among 6100 neurological admissions [123].

Huntington disease
Nineteen studies (four community-based studies and 15 hospital-based) investigated Huntington disease; including 8 cross-sectional studies (including reviews of medical records), 10 case series (two to 13 patients), and one case report ( Table 5). The studies were conducted in nine countries: South Africa (nine studies), Zimbabwe and Tanzania (two studies each), Nigeria, Mauritius Island, Senegal, Sudan, Togo and Burkina Faso (one study each). The diagnostic of Huntington disease was mostly clinical, based on a constellation of probing clinical elements; however genetic testing was carried out in five studies [136][137][138][139][140]. The absolute number of participants with Huntington disease ranged from one to 481. Only one community-based study provided a prevalence estimate of 3.5/100,000 in South-Africa [141]. The hospital-based prevalence of Huntington disease when reported ranged from 0.2/100,000 to 46.0/ 100,000 [138,[142][143][144][145][146]. No study reported data on the incidence of Huntington disease. Among those with the disease, males represented 42 to 100%, and age varied from <9 years to 80 years. When provided, the age at the clinical onset of the disease ranged from less than one year to 58 years. In general, antecedent risk factors for Huntington disease were not investigated across studies except for a positive family history reported in 58.3 to 100% of cases.

Discussion
This review represents an unprecedented effort to summarize epidemiological data on neurodegenerative diseases in SSA. However, this being a large diverse multicultural and multiethnic region, it is difficult to reliably quantify and compare the burden of neurodegenerative disorders across countries. Although mostly based on prevalent cases and on retrospective data, from studies that have essentially included urban populations, findings summarized in the current review are very informative. The most widely investigated and prevalent neurodegenerative condition appeared to be dementia with most cases being of Alzheimer disease type. Major risk factors of AD include an advanced age (higher after the age of 60), female sex, a low schooling (less than 6 year of education), family background and rural residence. Unlike North America, Australia, Europe, and Japan where several population-based studies have been conducted on dementia, good quality epidemiological studies (prospective, population-based, using standardized criteria) are scanty in SSA, with methodological issues hampering any meaningful comparison with other regions of the world. The reported prevalence in one collaborative good quality study in Nigeria about 20 years ago among those aged >60 years was 2.3%. This was lower than the reported prevalence in developing countries, but within the range of reports from developing countries in Asia and Latin America where reported prevalence range from 1.9 to 3.8% [155]. The anticipated ageing of the population (which is the main driver of dementia figures) in Africa may translate in a higher prevalence and absolute number of people living with dementia as observed in other developing regions. However, caution is needed when interpreting findings from studies conducted in different settings by different investigators. Our overview tends to suggest that the projected increase in the prevalence of dementia in SSA is likely, based on the comparison of findings from three recent studies with those from the study above conducted in Nigeria 20 years ago [55][56][57]. Furthermore, with the large scale implementation of antiretroviral therapy and related improved survival, it is expected that the number of patients with the diagnosis of HIV-related neurocognitive impairment may increase as suggested by the increasing number of related-publications. Such trends will need to be confirmed by large scale prospective observational studies which will also assess the putative accelerating effect of HIV-related neurocognitive impairment on other types of prevalent dementia and neurodegeneration.
For Parkinsonism, the wide prevalence range observed both in population and hospital-based studies might also be a consequence of differences in methodologies for case ascertainment, diagnostic criteria, or age distributions of the study populations. These heterogeneities in PD prevalence are not unique to SSA as these have also been observed in Europe where prevalence of PD ranged from 66 to 12,500/100,000 [156]. There have been provisional set of minimal scientific criteria for conducting epidemiological studies on PD which, when adopted at a large scale will improve comparison within SSA and between SSA and other regions of the world [156]. Prevalence rates reported in population-based studies in the continent are limited to two studies and cases were ascertained through screening and neurological exam in one study, thus making any comparison with other region difficult. In ALS and Huntington disease, the picture is less clear as the majority of studies were hospitalbased, retrospective in nature, with a final diagnosis not always based on pathology or genetics and the risk  factors not properly assessed; thus making comparisons and inferences inaccurate. For these two conditions therefore, important gaps remain to be filled, without which the issues of prevention and control will not be efficiently addressed in the African context. The comparatively higher number of population-based investigations of dementia relative to other neurodegenerative conditions in SSA, may at least in part be explained by the availability of standardized and widely accepted screening and diagnostic tools/criteria which facilitate epidemiological studies of dementia [157] as compared with other conditions where existing tools have not always been validated in different settings and therefore remain unpopular [158,159], or which, by the virtue of their low prevalence makes any assessment in the general population difficult and very expensive. There are context-specific challenges to obtaining key epidemiological data on neurodegenerative conditions in SSA including the low level of patient education, the need to accurately translate available screening and diagnostic tools to local languages, limited number of scientists and clinicians in neurosciences, and competing health interest in the setting of limited financial resources [5,16].

Needs in terms of epidemiological data
In order to improve the knowledge base of each of the neurodegenerative conditions addressed in this review, two main types of epidemiological studies appear necessary and feasible in SSA. A population-based prevalence and incidence study including both urban and rural populations, in order to capture the real variability in socioeconomic status and possibility in other factors that may exist in the population. Such a study may serve a dual purpose, providing information on disease rate and identification of key risk factors, as it would permit to establish the sequence of events. Given that such an undertaking could be planned beforehand, it offers the possibility of addressing multiple questions and/or diseases at a reduced cost. Inclusion of a large enough but manageable number of participants would be necessary to ensure adequate precision around the estimates generated. As many patients with possible neurodegenerative conditions would be tempted to consult traditional healers rather than accessing health facilities in SSA, special efforts would be required to ensure that these people are captured by such a study. Also, ascertaining cases of neurodegenerative conditions in a populationbased sample may be costly and logistically challenging, particularly with regard to the asymptomatic or mildly symptomatic nature of early stages of some of the diseases, and the lack of validated instruments and appropriate expertise.
A second type of epidemiological study is a multicenter, hospital-based, registry investigation. The latter has several advantages over a single large-scale cohort study. Large numbers of cases could potentially be collected over a relatively short period of time, with the possibility of comparing resources and outcomes within and across countries. However, the major limitations of this approach include the costs associated with the effort and infrastructure for coordination and communication between centers, as well as data capture and ongoing monitoring and quality control. In addition, there are biases inherent to any such hospital-based study, especially given that in SSA there is major access and cost barriers to care, with a sizeable proportion of patients with neurodegenerative conditions who are never seen by health care providers thus limiting the scope of registries. The degree of such selection bias is likely to vary considerably across centers, affecting both case mix and outcomes. The approach would therefore not provide a study population fully representative of incident cases and the natural history of disease and its management.
For both types of studies, the definition of the pool of people 'at-risk' population could be challenging in the SSA context, given the lack of formal census of the population in many countries; thus making reliable estimation of the effect of individual risk factors difficult. Other methodological issues relate to the assessment of the outcome in a reliable fashion in the African context as discussed above. Hence, a combination of the aforementioned study approaches would probably overcome some of their respective limitations and improve the quality of estimates generated.
The challenges to performing high quality incidence and prevalence studies of neurodegenerative diseases are well known [159]. Cases of most neurodegenerative conditions are difficult to define and ascertain reliably in population-based sample, and there are problems in relating events and the effects of different exposures to defined 'at-risk' populations. With the ageing of the population in SSA, the importance of HIV/AIDS, as well as the surge in risk factors such as hypertension and diabetes that have been linked to dementia [157,160,161] and possibly to Parkinson diseases [162,163], the importance of neurodegenerative disorders would considerably increase over time. Indeed, by 2025, the numbers of people aged 60 years and over will more than double in many countries [164]. With this rapid demographic and nutritional transition, neurodegenerative conditions would become an important public health problem in SSA. Critical investments are therefore necessary to improve surveillance and programrelevant research to provide an evidence base for policy development and effective control and prevention of neurodegenerative diseases. Precise identification of risk factors other than ageing would allow proper prevention effort spanning from primordial to secondary and event tertiary prevention, given that most of those conditions are associated with higher levels of disability and increased risk of death. Community-based risk factor control, combined with high risk approaches and realignment of health systems to incorporate the chronic management of neurodegenerative diseases are needed.

Strengths and limitations of the review
Our review is the first of its kind on neurodegenerative conditions in SSA. It is more up-to-date and broader than previous attempts to summarize evidence on single diseases in this setting [4][5][6][7][8]. By systematically assessing all published articles on these conditions, we aimed to draw the attention on the importance of the conditions in the region, and identify the research priorities. A limitation of this review is inherent to the limitations of the individual studies included. We relied on clinic-based studies where necessary in this systematic review; but such studies have limitations, particularly with regard to the generalization of their results data. However, we have tried to convey a clear understanding of the current burden and risk factors of each condition by examining all published papers across a broad range of clinical, biology, public health, and psychosocial literature, incorporating various types of evidence. By the nature of the disease, the age range for participants in studies on ALS and HIV-related neurocognitive impairment extended to the pediatric age for some studies. It is of note that large number of studies are realized in hospital in Africa, often published in local journals or reported in thesis. It the absence of straightforward strategies for capturing this sort of evidence in a systematic way, we did not account for them, which may have lowered the number of results found in some countries. Finally, the many sources of heterogeneity precluded any meaningful assessed of the quality of the included studies.

Conclusion
This review summarizes the body of literature on neurodegenerative disorders in SSA, which is large with regard to Dementia and HIV-related neurocognitive disorders but limited for other neurodegenerative disorders. In addition, it emphasizes some of the challenges in conducting good quality, population-based studies on the continent including the lack of standardized criteria for some neurodegenerative disorders, with most studies limited to few regions/countries on the continent. Highquality prospective cohort studies, which would use internationally-validated criteria, wide catchment areas in several geographic regions, and adjust for the projected ageing of the continent population, by compensating for the imprecise nature of the available data, will help map the epidemiology of neurodegenerative diseases in SSA and improve comparisons with the rest of the world.

Additional file
Additional file 1: Search terms and strategies.