A Systematic Review of Research on Autism Spectrum Disorders in Sub-Saharan Africa

The burden of autism spectrum disorders (ASDs) in sub-Saharan Africa (SSA) is not well known. We carried out a systematic review of the literature to identify published work from SSA. We have systematically searched four databases, namely, Medline, PsycINFO, CINAHL, and Child Development & Adolescent Studies, through EBSCO and identified studies from across SSA. Based on predefined inclusion criteria, 47 studies were included in this review. Most of the identified studies (74%) were conducted in only 2 African countries, that is, South Africa and Nigeria. Additionally, most of these studies (83%) were carried out in the last decade. These studies had four major themes: development of measurement tools of ASD in Africa, examining the prevalence of ASD, identifying risk factors and risk markers, and examining psychosocial issues. We identified only a single population level study aimed at documenting the prevalence of ASD and could not identify a single case-control study aimed at examining a comprehensive set of potential risk factors. All intervention studies were based on very small sample sizes. Put together, our findings suggest that current evidence base is too scanty to provide the required information to plan adequately for effective intervention strategies for children with ASD in Africa.


Introduction
Autism spectrum disorders (ASDs) are a neurodevelopmental syndrome with growing global health concern. This syndrome is characterized by deficits in social and communication skills and restricted and repetitive behaviour; and these adversely impact quality of life of those affected as well as their families [1]. Globally, one in every 160 persons is estimated to live with ASD, contributing to 7.6 million disability life adjusted years [2]. However, this burden is currently underestimated since prevalence of ASD in the African region and other low or middle income regions is still unclear [2][3][4]. One study, for example, that involved two North African countries documented a high frequency of ASD at 11.5% and 33.6% among African children with developmental disorders [5]. Other studies conducted among children of African descent have reported high occurrence of ASD [6][7][8] although their representativeness is questionable [9]. Similarly, studies on ASD document a large burden of nonverbal ASD cases (50−71%) and over 60% comorbid intellectual disability among African children with ASD [10,11]. These and other distinctive traits of ASD in Africa such as a potential infectious aetiology, late diagnosis, and poor management [2,3,12] accentuate the need for more research focus and public health response in this region.
Having gone past the prior dialogue questioning the universality of ASD [13], growing interest in ASD in Africa is currently documented by the increasing number of scientific studies on this condition in the continent [4,9]. There have been a few reviews synthesizing the data on ASD from Africa [4,9]; however, most of them were performed years ago, used a single bibliographic search, or did not utilize a systematic review. Although these earlier reviews provide interesting insights, there is a need to update and synthesize the most recent empirical evidence so as to identify research gaps and potential points of interventions. Of major interest is a clearer understanding of the current direction of ASD research in Africa (e.g., the focus on risk factors, prevalence, or interventions), identifying where in Africa ASD research is emanating as well as the key findings from ASD research in the African region. This current systematic review builds upon this backdrop by exploring ASD research in the sub-Saharan African (SSA) region over the past 50 years. We hope that this systematic review will avail relevant evidence to support and guide research, intervention, and policy on ASD, especially in the SSA region.

Search Strategy.
Guidelines for preferred reporting items for systematic reviews and meta-analyses (PRISMA) were utilized [58]. We searched four databases, that is, Medline , through EBSCO. The search terms used were "Autism" OR "Autistic" OR "Pervasive" AND "Africa". Our database search was restricted to peer-reviewed articles and excluded dissertations. We further searched the reference lists of retrieved articles as well as the Google Scholar database for other potentially relevant studies that may have been missed from the systematic database search.

Criteria for Inclusion and Exclusion.
We set out the following inclusion criteria: (1) The study must be an empirical study on ASD conducted among humans.
(2) ASD must be the main condition of study.
(3) The study populations need to be from SSA and the study needs to be carried out in SSA.
We excluded studies that (i) were not empirical, (ii) were conducted in countries other than those from SSA, and (iii) did not consider ASD as the main disorder of interest.

Data Extraction and Analysis.
One data extraction sheet was used to summarize the data in Microsoft Excel spreadsheet (version 2013) on the general characteristics of the studies and their key findings. These characteristics of interest included (i) author, (ii) year of publication, (iii) country where the study was done, (iv) sample description, and (v) key findings. A narrative synthesis was used to summarize the findings of eligible studies included in this systematic review.

Summary of Study Characteristics.
We identified a total of 341 potentially eligible studies of which 47 fulfilled the criteria. Figure 1 presents the flow chart on the number of identified abstracts, reasons for exclusion, and articles that were further considered. Table 1 presents a summary of the characteristics of the eligible studies and their key findings. These eligible studies were conducted in very few African countries (around ten). Most of these studies were from South Africa ( = 25, 51%) and Nigeria ( = 11, 23%). Most of the studies ( = 38, 83%) were carried out in the past decade indicating increased interest in the area.

Screening and Diagnosis of ASD in Africa.
There are few studies that have attempted to validate screening and diagnostic measures for use in SSA. A recent study from Uganda adapted and extended the Ten-Question Questionnaire (TQQ) into a 23-item questionnaire and evaluated the sensitivity and specificity of this new tool to identify ASD alongside general "disability" [17]. It was reported that the 23item questionnaire was modestly successful in identifying a subgroup of children at high risk of being diagnosed as having ASD. A study from Tanzania evaluated the potential use of the Childhood Autism Rating Scale to perform a structured observation to diagnose ASD [16]. In this study, the Childhood Autism Rating Scale was culturally adapted for use in Tanzania. Some of the cultural adaptations included ensuring that the play interactions, materials used, and social routines used to probe the child's behaviour were familiar to the children. Following these adaptations, the authors reported excellent discriminative validity and acceptable levels of sensitivity and specificity. Two recent studies in South Africa have also examined and evaluated the cultural adaptability of ASD measures in their context [14,15]. The study by Smith and colleagues evaluated the cultural appropriateness of the Autism Diagnostic Observation Schedule-2 (ADOS-2) [15]. Participants were requested to evaluate the cultural appropriateness of the materials and procedures for administering the ADOS. They reported that most of the social interaction demands, materials, and activities were appropriate for use in the urban samples from Cape Town. However, potential linguistic and semantic biases were observed and therefore guidelines for using ADOS in their setting were developed.

Prevalence of ASD.
A few studies have attempted to estimate the burden of ASD in SSA [17,19]. However, most of these studies used convenience sampling with data largely from hospital and specialist units for children with special needs. For instance, a study by Lagunju and colleagues [18] recruited 2320 patients at a paediatric neurological clinic. After a systematic screening, 54 of the 2320 patients were diagnosed with ASD, with estimated prevalence of 2.3%. Additionally, it was noted that parents reported a deviation in development at a mean age of 22.1 months, and they received a diagnosis at a mean age of 44.7 months. Among those with ASD, approximately 75.5% presented with associated neurological comorbidities. Only one community-based study was identified, in which 1169 Ugandan children aged 2-9 years were surveyed in the Kampala District (half urban and half rural) and eight children had a positive diagnosis of ASD. The authors reported unadjusted prevalence for ASD of 6.8/1000 [17]. All the studies reported higher prevalence among boys compared to girls. Bakare et al. [19] in a study from Nigeria reported an ASD ratio of 4 : 1 for boys and girls, respectively.

Risk Markers and Risk Factors for ASD.
We did not identify any case-control study that examined a comprehensive set of risk factors for ASD in SSA. Small studies have identified specific genetic risk markers and nongenetic risk factors for British children (e.g., higher prevalence among boys and its existence across a wide range of IQ). Additionally, the authors also observed marked differences in the prevalence of certain symptoms such as lower rates in occurrence of ritualistic and repetitive behaviour.

Risk factors and markers
8 Ezegwui [20] 2014 Nigeria 21 children with ASD Significant refractive error, mainly astigmatism, was noted in the children with autism. 9 Sharma [21] 2013 South Africa 136 children with ASD and 208 controls The aim of the study was to find the genetic association of intronic rs736707 and exonic rs362691 (single-nucleotide polymorphisms [SNPs] of the RELN gene) with autism in a SA population. A significant association of SNP rs736707, but not for SNP rs362691, with autism in the SA population was observed. 10 Arieff [22] 2010 South Africa 109 children with ASD Allele frequencies and genotypes of the South African autistic populations (African, mixed, and Caucasian) were compared with matching South African ethnic control populations. The study showed significant differences in allele and genotype frequencies of 5-HTTLPR polymorphic region and provides impetus for investigating the role of transmission of the L and S alleles in families with autism in South Africa.

11
Bakare [23] 2008 Nigeria One child with ASD Observed ASD in a child with oculocutaneous albinism. 12 Claassen [24] 2008 South Africa A pair of twin siblings (1 had infantile autism; the other is a control) Using data from a dizygotic set of twins, the author concluded that prenatal stress may be a salient contributor to the pathogenesis of autism. 13 Mankoski [11] 2006 Tanzania 20 children recruited from a special needs primary school and from families having a child with clinically diagnosed autism The study aimed at investigating the association between malaria and autism. Using a case series approach, the study observed 14 children who met the criteria for ASD among whom 3 had onset of autism after recovery from malaria and 4 cases had a temporal relationship between ASD and malaria though it seemed spurious, while in seven cases the onset of ASD was unrelated to malaria.

37
Bakare [48] 2009 Nigeria 134 health workers The study noted that a significant percentage of healthcare workers in Nigeria still held negative or false beliefs on the aetiology, treatability, and preventability of ASD, leading to the conclusion that part of the efforts to improve services for families of children with ASD need to focus on improving knowledge among healthcare providers. 38 Olivier [49] 2009 South Africa 8 parents of children with ASD The study reported that some of the key challenges faced by parents of children with ASD include the denial of the diagnosis, lacking proper guidance having received the diagnosis, and a lack of adequate parenting and coping skills among other challenges. One child with ASD This case study carried out a conversational analysis of the conversations between the participant and his/her coparticipants with the aim of understanding some of the potentially useful points of intervention to enhance communication skills among children with ASD. 42 Akande [53] 2000 South Africa 3 children with ASD This study investigated colour learning and observed the need for a highly individualized approach to teaching children with ASD since the authors observed a significant variability in the approach to learning. 43 Akande [54] 1999 South Africa 7 children with ASD This study evaluated the efficacy of the "self-monitoring intervention" and observed that the success rates were very similar to what has been reported in other parts of the world.
44 Khan [7] 1996 Zimbabwe 18 children The authors noted that, in the Zimbabwean context, the DSM III-R categories were supported by empirical evidence. However, they recommended the inclusion of characteristics such as abnormal responses to sensory stimuli and disturbances to cater for nonclassical autism. 45 Dhadphale [55] 1982 Kenya 3 children with infantile ASD The study noted that there were no differences in features of childhood ASD presented by the 3 Kenyan children when compared to what has been described in the west. 46 Noach [56] 1974 South Africa 8 (4 children with ASD and 4 typically developing children) Children with ASD were observed to present with impairments in concept formation. 47 Silver [57] 1970 South Africa One Child with ASD This study illustrates how operant conditioning can be used to teach a child with ASD a few words. Studies mainly focused on

Screening
In Africa but not SSA (n = 1) ASD [11,21]. Infectious diseases such as falciparum malaria have been suggested as possible antecedents to ASD [11], but the association has not been established. Of the six studies identified in this category, 4 explored possible genetic and biomedical factors [11,[20][21][22]. Three of the studies looked at potential genetic markers, the ones by Ezegwui et al. [20], Sharma et al. [21], and Arieff et al. [22]. The studies observed that certain genetic characteristics, for example, allele and genotype frequencies of 5-HTTLPR, were more likely to be associated with an increased risk of ASD. Psychosocial risk factors such as parental stress level have been associated with increased risk of ASD. Unfortunately, the study had a very limited sample size ( = 2) compromising the generalisability of these findings [24].

Psychosocial Aspects of ASD.
This category presents the bulk of studies on ASD in Africa. These studies largely examined (a) awareness levels, (b) quality of services provided to children with ASD and caregiver challenges, and (c) sociocultural aspects around ASD, for example, explanatory models on the aetiology of ASD. For instance, most of the healthcare workers in Enugu, Nigeria, had limited knowledge of ASD and perceived the quality of healthcare provided to families of children with ASD as suboptimal [50,51]. Most of the studies on awareness were carried out in Nigeria by Bakare and colleagues [33,48,50,51]. One South African study observed that there were relatively few barriers to participating in ASD research and suggested that most of these barriers were poverty-related [42]. Additionally, a few very-small-scale studies (with participants ranging from 1 to 7) have looked at potential intervention strategies to enhance outcomes among children with ASD [59,60]. For instance, Travis and Geiger investigated the efficacy of the Picture Exchange Communication System (PECS) to improve communication skills in two South African children with ASD [59]. In Tanzania, Harrison and colleagues [27] recently developed an intervention to raise awareness and help caregivers learn some basic behavioural intervention strategies in two phases. In the first phase, 14 caregivers took part in a needs assessment session and an ASD knowledge intervention. In the second phase, 29 caregivers were involved in an intervention focusing on basic behavioural strategies such as parenting skill training, teaching of basic skills (e.g., making eye contact and imitating), and teaching of self-help skills (e.g., feeding). Initial evaluation indicated that despite a few challenges it was feasible to implement the intervention and almost all participating caregivers found it to be useful.

Discussion
We reviewed the published literature on ASD in sub-Saharan Africa. Our results indicate that there is very limited data from Africa compared to other parts of the world. We identified only a single population level study aimed at documenting the prevalence of ASD in Africa. Additionally, we could not identify a single case-control study aimed at examining a comprehensive set of potential risk factors in Africa. Also, the few intervention studies had very limited sample sizes, were largely cross-sectional, and lack any measurement or evaluation of long-term impact. Put together, our findings suggest that current evidence is too scanty to provide the required information to plan adequately for effective intervention strategies for children with ASD in SSA.
Notable is the fact that most of the identified studies arose from only two African countries, that is, South Africa and Nigeria. The lack of literature from other parts of Africa may be due to several factors. First, there may be a lack of expertise in other African countries; for instance, a recent conference report does document the large difference in number of qualified psychologists and psychiatrists in South Africa compared to other countries that had representatives [61]. Second, this may arise from the lack of resources to carry out research in this area. Another potential explanation is the lack of interest in ASD as a research topic in other parts of Africa. Without any evidence, we can only speculate on the potential reasons for this. However, we feel that the dearth of research from the other African countries reflects the interaction between the first two reasons mentioned earlier. This calls for an urgent need to develop capacity and interest in ASD research in other countries outside SA and Nigeria to expand the evidence base.
An important step towards having an adequate research framework is to have standardized tools for screening and diagnosing. Our research indicates that they are very few tools that have been validated for the African context. The use of tools from other continents does provide challenges for various reasons including methodological and resource availability [15,61]. These challenges are the impetus towards the development and validation of tools in the African context. However, some studies [16,17] present early, yet very crucial steps towards identification of potential screening and diagnostic tools for the African context. The need to invest more into this process cannot be overemphasized given the potential benefit in accessing tools that can contribute towards early identification of children who have ASD.
Our review indicates that a significant proportion of the studies were on psychosocial issues. A major focus of most studies on psychosocial aspects of ASD potentially reflects significant burden on quality of life that African communities are increasingly witnessing and appreciating as resultant from the ASD condition. Thus, there is growing interest in understanding better care and management practices to avert this burden.

Limitations.
We have systematically searched several databases and identified studies from across different settings in Africa. We did not look at grey literature and other more local and regional based databases; consequently, we may have missed out some articles. We, however, did carry out a search of references list to identify more studies. We hope this reduces the potential number of missed studies; however, we cannot completely exclude this possibility.

Conclusions.
Based on our review of published works, it can be said that there is a dearth of scientifically vigorous published work from sub-Saharan Africa making it difficult to estimate the burden of ASD in this population, identify risk factors, or even plan effective intervention strategies.