The epidemiology of psychiatric disorders in Africa: a scoping review

Abstract

This scoping review of population-based epidemiological studies was done to provide background information on the prevalences and distribution of psychiatric disorders in Africa for calls to broaden diversity in psychiatric genetic studies. We searched PubMed, EMBASE, and Web of Science to retrieve relevant literature in English, French, and Portuguese from Jan 1, 1984, to Aug 18, 2020. In 36 studies from 12 African countries, the lifetime prevalence ranged from 3·3% to 9·8% for mood disorders, from 5·7% to 15·8% for anxiety disorders, from 3·7% to 13·3% for substance use disorders, and from 1·0% to 4·4% for psychotic disorders. Although the prevalence of mood and anxiety disorders appears to be lower than that observed in research outside the continent, we identified similar distributions by gender, although not by age or urbanicity. This review reveals gaps in epidemiological research on psychiatric disorders and opportunities to leverage existing epidemiological and genetic research within Africa to advance our understanding of psychiatric disorders. Studies that are methodologically comparable but diverse in geographical context are needed to advance psychiatric epidemiology and provide a foundation for understanding environmental risk in genetic studies of diverse populations globally.

Introduction

Studying the epidemiology of psychiatric disorders in Africa should lead to a deeper understanding of the frequency, patterns, and distributions of these disorders, and risk factors in populations residing in Africa, and facilitate global cross-population comparisons. Although it is widely presumed that for psychiatric disorders with high heritability the prevalence is roughly constant worldwide, psychiatric disorders with low heritability are likely to interact with modifying social and environmental risk factors that could influence their incidence and prevalence across populations. The past few years have seen several calls to broaden the scope of ancestral and geographical diversity in large-scale genetic studies.^1,2 The Psychiatric Genomics Consortium, the largest international consortium to study the genetic architecture of 11 psychiatric disorders, was formed in 2007 to launch a global effort to obtain the necessary sample sizes for genome-wide association studies. Although by 2020 the Psychiatric Genomics Consortium had published several landmark studies on the genetic architecture of major psychiatric disorders, these studies were predominantly done in populations of Asian or European ancestry and outside of Africa.³ Research on the genetics of schizophrenia and bipolar disorder led by African and international researchers in Africa has been emerging, but it remains under-represented compared with studies in Europe, Asian, or American countries.^4,5 Examining populations of similar ancestry that are under-represented in existing genetics studies could enable the exploration of the role of environment and culture separately from the role of genetic factors in the contribution to psychiatric disorders.

The originators and participants who provided data for the largest Psychiatric Genomics Consortium publications noted that it was essential that diverse populations and locations become better represented.¹ 71% of the individuals in the study by Peterson and colleagues had been recruited from the USA, UK, and Iceland.¹ Broadening of diversity would “improve the effectiveness of genomic medicine by expanding the scope of known human genomic variations and bolstering our understanding of disease etiology”.¹

Data on the epidemiology of psychiatric disorders has flourished since Freedman’s introduction of the Epidemiologic Catchment Areas study in 1984.⁶ Similar to genetic studies, these landmark epidemiological studies focused primarily, but not exclusively, on populations of European descent in European or North American countries. The extent to which the prevalence and distribution of psychiatric disorders is comparable with populations in other settings is largely unknown. By contrast, large population-based epidemiological studies in Africa have been essential to understanding HIV and neurological disorders such as epilepsy, among others.^7,8 Epidemiological studies on disorders that differ in prevalence between African and non-African populations have advanced our understanding of the causes of these high-burden disorders, which could serve as a model for furthering knowledge on the epidemiology of psychiatric disorders in Africa.⁷ It was within this context that this review on the epidemiology of psychiatric disorders in Africa was undertaken. We are ultimately interested in exploring the similarity or difference in syndrome constructs globally, potential environmental factors that modify prevalence, and the delineation of genetic and environmental factors that contribute to risk across populations.

The purpose of this scoping review of population-based studies is to identify sources of evidence and gaps in existing research on the epidemiology of psychiatric disorders in Africa.⁹ The objective is to obtain information on prevalence and distributions of psychiatric disorders using clinical criteria and methods similar to those used in ongoing genetic studies. To guide the scoping review, we focused on the following research questions: how has the prevalence of psychiatric disorders in African countries been estimated across population-based studies? And what is the nature, range, and extent of the literature on the epidemiology of psychiatric disorders in Africa? This review aimed to explore the prevalence of psychiatric disorders based on geography (ie, studies done in African countries) as opposed to populations of African ancestry.

Methods

Rationale

We did a scoping review to identify key concepts, sources of evidence, and gaps in existing research on the epidemiology of psychiatric disorders in Africa. Scoping reviews are appropriate for assessments in areas in which the range of existing literature on a particular topic is initially uncertain. Accordingly, we have followed the methodological framework by Arksey and O’Malley.¹⁰

Search strategy and selection criteria

We searched PubMed, EMBASE, and Web of Science to retrieve relevant literature in English, French, and Portuguese from Jan 1, 1984, to Aug 18, 2020. We chose 1984 as this was the year that the first epidemiological studies using clinical criteria were published,¹¹ and we were interested in diagnoses that had been used in current genetic studies worldwide.

The search strategy included (1) terms relevant to psychiatric disorders; (2) terms for each country in Africa; and (3) terms related to prevalence and population-based surveys. Terms within each set were grouped using Boolean OR operators and the three sets were combined with AND operators. Terms related to psychiatric disorders were as follows: (psychiatric and disorder*) OR (mental and disorder*) OR depress* OR anxiety OR “post-traumatic stress” OR PTSD OR panic OR (eating and disorder*) OR schizo* OR psychosis OR bipolar OR (mood and disorder*) OR ([substance OR alcohol OR drug] AND [disorder* OR abuse OR dependence]) OR (personality and disorder*) OR ADHD OR (emotional and problem*) OR (behavioral and problem*). Terms for each country in Africa were as follows: Africa OR sub-Saharan Africa OR Algeria OR Egypt OR Libya OR Morocco OR Tunisia OR Cameroon OR Central African Republic OR Chad OR Congo OR Democratic Republic of the Congo OR Equatorial Guinea OR Gabon OR Burundi OR Djibouti OR Eritrea OR Ethiopia OR Kenya OR Rwanda OR Somalia OR Sudan OR South Sudan OR Tanzania OR Uganda OR Angola OR Botswana OR Lesotho OR Mozambique OR Ivory Coast OR Namibia OR South Africa OR Swaziland OR Zambia OR Zimbabwe OR Benin OR Burkina Faso OR Côte d’Ivoire OR Gambia OR Ghana OR Guinea OR Guinea-Bissau OR Liberia OR Mali OR Malawi OR Mauritania OR Mauritius OR Eswatini OR Madagascar OR Niger OR Nigeria OR Senegal OR Sierra Leone OR Togo. Terms related to prevalence and population-based surveys were as follows: prevalence and (“general population” OR community OR population OR epidemiolog*).

We aimed to identify studies that reported on the prevalence of psychiatric disorders in a sample that was expected to be representative of the characteristics of the general population within the setting in which the study was done. We were interested in exploring the study designs used to estimate prevalence and, therefore, did not restrict the results by sampling method (eg, probability samples). Publications were eligible if they were quantitative community-based and population-based studies published from 1984 onwards and reported prevalence estimates of one or more psychiatric disorders in the general population in an African country. Prevalence estimates were described as point prevalence (ie, current), period prevalence (ie, past year), or lifetime prevalence (ie, ever). Studies that assessed psychiatric disorders but did not report one of these three prevalences were excluded. We did not exclude any studies based on sample size. Given the focus of this review was to study the prevalence of psychiatric disorders in Africa (ie, geographical context), we excluded studies reporting on the prevalence of psychiatric disorders in populations of African ancestry outside of Africa.

Reference lists of eligible studies were reviewed to identify additional studies.¹² Titles and abstracts were screened using the inclusion criteria, after which full articles were retrieved. Studies with irrelevant titles were excluded, as were commentaries, conference abstracts, editorials, intervention studies, and theoretical papers. We also excluded studies on specific subpopulations (eg, participants with medical conditions, treated samples, and pregnant women). Potentially relevant full-text studies were then evaluated against review eligibility criteria by two authors (TY and MCG), and any discrepancies were resolved by consensus, including review by the senior author (MMW).

Data extraction

We extracted data on the last name of the first author, year of publication, study location, year of data collection, study sample, age range of the study population, sample size, the diagnostic tool, and prevalence of psychiatric disorders. For instances in which the same parent study data were used in more than one publication, we put the reports together and considered the papers as one study, even if the final analytic sample size differed between studies.

Results

The search yielded 20075 records and 9798 records were left after duplicates and completely irrelevant records were removed (figure 1). We screened 9512 titles followed by 1103 abstracts, which resulted in 105 articles assessed for eligibility. 58 full-text articles were excluded because they recruited samples that were not representative of the general population (n=33 records), did not describe which type of prevalence estimate the study was reporting (n=18 records), and were not a community population-based study (n=5 records). One was a conference abstract, and one study of geriatric depression in Tanzania was classified as awaiting classification because the full text was not available and thus could not be fully evaluated against our eligibility criteria.¹³ 47 articles representing 36 unique studies were included in this Scoping Review. Included studies were published between 1996 and 2020, with half published after 2008 (n=28 articles).

Figure 1: Screening and selection of articles.

Setting and population characteristics

36 studies estimated the prevalence of one or more psychiatric disorders in the general population in 12 countries in north, east, west, and southern Africa (table 1; appendix). 24 studies were done in five of these countries: South Africa (n=8; 22%),^14–28 Ethiopia (n=5; 14%),^29–37 Kenya (n=5; 14%),^38–42 Nigeria (n=3; 9%),^{18–20,43–46} and Uganda (n=3; 9%).^47–49 Remaining studies were done in Benin (n=1; 3%),⁵⁰ Burkina Faso (n=2; 6%),^51,52 Egypt (n=2; 6%),^53,54 Ghana (n=1; 3%),²⁶ Morocco (n=2; 6%),^55,56 Mozambique (n=2; 6%),^57–59 and Rwanda (n=2; 6%).^60,61 Studies were designed to be nationally representative (n=10),^{15–17,21–23,25–27,52–55} or representative of populations in rural regions (n=14),^{28,29,33–35,37–41,47–50,58–60} a peri-urban community (n=1),²⁴ or a region that included a mix of rural, urban, and peri-urban communities (n=11).^{14,18–20,30–32,36,42–46, 51,56,57,61} 12 studies aimed to estimate the prevalence of psychiatric disorders among adults (age ≥18 years).^{14,17–20,25,27, 36,37,43,44,46–48,50,52,53,60,61} 17 studies aimed to estimate the prevalence of psychiatric disorders among adolescents and adults usually defined as 15 years or older.^{15,16,21,23,24,29–35,38–40,51,54–57} Three studies focused on either young children,⁴¹ or children and adolescents.^42,49 Two studies were restricted to older adults.^22,26,45 One study was restricted to female heads of household.^58,59 Recruitment and study interviews were completed at participants’ homes in all studies with one exception. One study that focused on estimating the prevalence of substance use disorder avoided doing study interviews within the household to prevent underreporting of substance use by participants when in close proximity to their family.⁵⁴

Table 1:

Characteristics of included studies

	Country and year	Sample	Age (years)	Diagnostic interview (sample size)
Adewuya et al (2018 and 2020)43,44	Nigeria 2015	Multistage sampling; representative of Lagos state (urban or peri-urban); LSMHS	18–40	DSM-IV; PHQ-9; AUDIT-C; MINI 5-0 for 2018;43 DSM-5; MINI 5·0 for 2020;44 (n=11 246)
Andersson et al (2018)14	South Africa 2012	Multistage sampling; representative of the Eastern Cape Province (urban or peri-urban)	18–40	DSM-IV; MINI (n=977)
Audet et al (2018)58 and Wainberg et al (2018)59	Mozambique 2014	Multistage sampling; representative of female heads of household in 14 rural districts in central Mozambique (rural)	16–62	DSM-5; PHQ-8; AUDIT (n=3892)
Bolton et al (2002)60	Rwanda 1999	Random sample of households; representative of Kanzenze Commune (rural)	18 and older	DSM-IV; DHSCL (n=368)
Bolton et al (2004)47	Uganda 2000	Systematic random sample of households; representative of Masaka and Rakai districts (rural)	18 and older	DSM-IV; DHSCL; WHODAS-II (n=587)
Department of Health, Medical Research Council (2007)16	South Africa 2003	Multistage sampling; representative of South Africa (national); South Africa Demographic and Health Survey 2003	15 and older	CAGE questionnaire (n=10 214)
Duthe et al (2016)51	Burkina Faso 2010	Multistage sampling; representative of Ouagadougou (peri-urban); Ouagadougou Health and Demographic Surveillance System	15 and older	DSM-IV; MINI (n=2187)
Fedaku et al (2004)29	Ethiopia 1998	Complete coverage of Zeway Islands (rural)	16 and older	ICD-10; CIDI; SCAN (n=1691)
Gedif et al (2019)37	Ethiopia 2018	Multistage sampling; representative of Mandura Woreda (rural)	18 and older	ASSIST (n=1588)
Ghanem et al (2009)53	Egypt 2003	Purposive sampling of sites selected to represent different socioeconomic, cultural, and geographical characteristics in Egypt (national)	8–64	MINI-Plus (n=14 640)
Gureje et al (2007)45	Nigeria 2003–04	Multistage sampling; representative of Yoruba-speaking areas of Nigeria (urban, peri-urban, or rural); Ibadan Study of Aging	65 and older	DSM-IV; CIDI (n=2152)
Gureje et al (2006)46 and Kessler et al (2005,93 2009,18 200719)	Nigeria 2001–03	Multistage sampling; representative of Yoruba-speaking areas of Nigeria (urban, peri-urban, or rural); Nigerian	18 and older	DSM-IV; CIDI (n=4985);46 (n=2143);93 (n=6752);18 (n=1203)19
Hamdi et al (2013)54	Egypt 2005–06	Purposive stratified sampling of individuals in eight governates (national)	15 and older	ICD-10; ASI (n=44 000)
Herman et al (2009),17 Kessler et al (2015, 2009, 2007),18–20 Stein et al (2008),25 Tomlinson et al (2009)27	South Africa 2002–04	Multistage sampling; representative of South Africa (national); SASH	18 and older	DSM-IV; CIDI 3·0 (n=4351)
Hunduma et al (2017)36	Ethiopia 2016	Multistage sampling; representative of Harari People Regional State (urban or rural)	18 and older	SRQ (n=901)
Jenkins et al (2012)38	Kenya 2000	Multistage sampling; representative of Maseno, Kisumu District, Nyanza Province (rural)	16–65	ICD-10; CIS-R; PSQ (n=876)
Jenkins et al (2015)39	Kenya 2004–13	Multistage sampling; representative of Maseno, Kisumu District, Nyanza Province (rural)	16–65	ICD-10; CIS-R (n=1158)
Kadri et al (2007)56	Morocco 1994	Multistage sampling; representative of eight prefectorats of Casablanca (urban)	15 and older	DSM-IV; MINI (n=800)
Kadri et al (2010)55	Morocco 2004–05	Multistage sampling; representative of Morocco (national)	15 and older	DSM-IV; MINI (n=5498)
Kariuki et al (2017)41	Kenya	Simple random sample of parent-children dyads; representative of Kilifi county (rural); Kilifi Health and Demographic Surveillance System	1–6	DSM-IV; CBCL (n=3273)
Kebede et al (1999)30–32	Ethiopia 1994	Multistage sampling; representative of Addis Ababa (urban)	15 and older	ICD-10; CIDI; SRQ (n = 1420)
Kebede et al (2003, 2005)33,34 and Negash et al (2005)35	Ethiopia 1998–2001	Complete coverage of Butajira, Ethiopia (rural)	15–49	ICD-10; CIDI; SCAN (n=68 378);33,34 (n=68 491)35
Kinyanda et al (2013)49	Uganda	Multistage sampling; representative of Lira, Tororo, Kaberamaido, and Gulu districts (rural)	3–19	DSM-IV; MINI; SDQ (n=1587)
Kwobah et al (2017)40	Kenya 2015–16	Random sample of individuals; representative of Kosirai division, Nandi County (rural)	15 and older	DSM-5; MINI-7 (n=420)
Magai et al (2018)42	Kenya	Multistage sampling; representative of Kiambu and Nyeri County (urban or rural)	6–18	CBCL; YSR (n=1022)
Nalwadda et al (2018)48	Uganda 2013	Random sample of households; representative of Kamuli District (rural)	18 and older	AUDIT (n=351)
Ouedraogo et al (2019)52	Burkina Faso	Multistage sampling; representative of Burkina Faso (national)	18 and older	MINI (n=2587)
Parry et al (1998)21	South Africa 1998	Multistage sampling; representative of South Africa (national); First South African Demographic and Health Survey	15 and older	CAGE questionnaire (n=13 826)
Patel et al (2007)57	Mozambique 2003	Multistage sampling; representative of Maputo City (urban) and a rural community	17 and older	SI and SSI (urban n=1796; rural n=943)
Peltzer et al (2011)23	South Africa 2008	Multistage sampling; representative of South Africa (national); South African National HIV Incidence, Behaviour and Communication survey	15 and older	AUDIT (n=15 828)
Peltzer et al (2013)22 and Thapa et al (2014)26	South Africa 2007–09	Multistage sampling; representative of South Africa; National population-based study—(SAGE)	50 and older	ICD-10; CIDI (n=3840);22 and NIAAA or a diary recording alcohol use (n=3668)26
Rumble et al (1996)28	South Africa 1992	Multistage sampling; representative of Mamre village (rural)	15 and older	SRQ; SE-CATEGO (n=560)
Smit et al (2006)24	South Africa 2002	Random sample of households; representative of a township outside of Cape Town (peri-urban)	15 and older	DSM-IV; CES-D; AUDIT; LEC (n=645)
Thapa et al (2014)26	Ghana 2007–09	Multistage sampling; representative of Ghana; National population-based study—(SAGE)	50 and older	NIAAA; a diary recording alcohol use (n=4289)
Tognon-Tchégnonsi et al (2020)50	Benin 2013	Sampling of households along a randomly selected direction in Tourou community (rural)	18 and older	DSM-IV; CIDI (n=603)
Umubyeyi et al (2014)61	Rwanda 2011–12	Multistage sampling; representative of eight districts in the Southern Province (primarily rural)	20–35	DSM-IV; MINI 5·0 (n=917)

AUDIT=Alcohol Use Disorder Identification Test. ASI=Addiction Severity Index. CES-D=Center for Epidemiologic Study for Depression Scale. CIDI=Composite International Diagnostic Interview. CIS-R=Clinical Interview Schedule—Revised. DHSCL=Depression Section of the Hopkins Symptom Checklist. DSM=Diagnostic and Statistical Manual of Mental Disorders. ICD-10=International Classification of Disease, 10th revision. LEC=Life Event Check-list. LSMHS=Lagos State Mental Health Survey. MINI=Mini-International Neuropsychiatric Interview. NIAAA=National Institute on Alcohol Abuse and Alcoholism. PHQ=Patient Health Questionnaire. PSQ=Psychosis Screening Questionnaire. SAGE=Study of Global Ageing and Adult Health. SASH=South African Stress and Health Study. SCAN=Schedule for Clinical Assessment in Neuropsychiatry. SI=Structured Interview. SQ=Structured Questionnaire. SRQ=Self-Reporting Questionnaire. SSI=Semi-structured Interview. WHODAS-II=WHO Disability Assessment Schedule 2.0.

Sampling procedures

Participants were sampled using multistage cluster probability sampling procedures (n=25 studies),^{14–23,25–28,30–32,36–39,42,43,45,46,49,51,52,55–59,61} simple or systematic random samples of households or individuals (n=6 studies),^{24,40,41,47,48,60} enrolling all eligible people in the sampling frame (n=2 studies),^29,33–35 selecting a random direction from the centre of the village and approaching all households in that direction to identify eligible participants (n=1 study),⁵⁰ or using a non-probability sampling method (n=2 studies).^53,54 Sample sizes ranged from 351 to 68 491 individuals,^33,35,48 with a median sample size of 1769 individuals. In total, included studies enrolled 236 104 people.

Assessment of psychiatric disorders

Case definitions for psychiatric disorders were most commonly informed by DSM-IV criteria (n=16 studies)^{14,17–20,24,25,27,35,41,43,45–47,49–51,55,56,60,61} followed by ICD-10 (n=6 studies),^{22,26,30–34,38,39,54} DSM-5,^40,44,58,59 or a combination of ICD-10 and DSM-IV (n=1 study).²⁹ Psychiatric disorders were measured using diagnostic interviews in 21 studies using the WHO World Mental Health Composite International Diagnostic Interview,^{17–20,22,25–27,29–35,45,46,50,62} the Mini International Neuropsychiatric Interview (MINI),^{14,40,44,49,51–53,55,56,61,63} and the Clinical Interview Schedule-Revised.^38,39,64 Five of these studies combined the use of a diagnostic interview with screening tools,^{30–32,38,43,44,49} which included the Patient Health Questionnaire (PHQ-9),⁶⁵ the Generalised Anxiety Disorder-7 (GAD-7),⁶⁶ the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C),⁶⁷ the Self Reporting Questionnaire,⁶⁸ and the Strengths and Difficulties Questionnaire (SDQ).⁶⁹ 13 studies used screening tools to estimate the point, period, or lifetime prevalence of probable cases of psychiatric disorders using the Hopkins Symptom Checklist in combination with a locally developed functional impairment measure;^47,60,70,71 the Center for Epidemiologic Studies Depression Scale;^24,72 PHQ-8;⁵⁸ the Self Reporting Questionnaire (point prevalence);³⁶ the Child Behaviour Checklist (point prevalence);^41,42,73 the Youth Self Report (point prevalence);^42,74 the AUDIT-C (point prevalence);^43,44 the AUDIT (point and period prevalence);^{23,24,48,59,75} the Cut-Annoyed-Guilty-Eye (CAGE) questionnaire (lifetime prevalence);^15,16,21,76 the Alcohol, Smoking, and Substance Involvement Screening Test (point prevalence);^37,77 adapted Addiction Severity Index (lifetime prevalence);^54,78 and the Life Events Checklist⁷⁹ in combination with the Harvard Trauma Questionnaire.^24,80

Other methods included having participants maintain a diary recording their alcohol consumption during the past week and applying the National Institute on Alcohol Abuse and Alcoholism cut-offs to determine harmful alcohol use during the past year,²⁶ and locally developed questions and vignettes as part of structured and semi-structured interviews to increase community-based case identification of lifetime psychosis and intellectual disability.⁵⁷ Nine studies reported using measures that had been adequately validated in the study population or country.^{18–20,24,40,43,44,47,55–57,60,62}

The prevalence of mood and anxiety disorders

22 studies estimated the prevalence of one or more mood disorders. The point, period, and lifetime prevalence of any mood disorder ranged from 3·8% to 6·4% for point, from 1·1% to 4·9% for period, and from 3·3% to 9·8% for lifetime prevalence.^{17–19,25,46,53} The point, period, and lifetime prevalence of major depressive disorders ranged from 2·0% to 33·2% for point, from 1·1% to 7·1% for period, and from 0·3% to 26·2% for lifetime prevalence (table 2, figure 2).^{17,20,22,24,25,27,28,33,40,43,45–47,52,53,58,60} Studies restricted to older adults and those that used screening tools to ascertain cases tended to report higher prevalence estimates than studies with younger populations and those that used diagnostic interviews. The prevalence of depression was consistently higher among females than males (prevalence ratio 1·1–2·2) and lowest in studies done in Nigeria. The point and lifetime prevalence of bipolar disorder ranged from 0·1% to 3·2% for point, and from 0·0% to 5·2% for lifetime prevalence. Studies reporting the prevalence of bipolar disorder by gender produced inconsistent findings. The point and lifetime prevalence of bipolar disorder were twice as high for point and six times higher for lifetime prevalence among females in Ethiopia’s capital city Addis Ababa,³¹ whereas the lifetime prevalence of bipolar disorder was twice as high among males than females in a study done in a rural setting in Ethiopia.³⁵ Other studies examined the prevalence of major depressive episodes,^{31,49–52,55,61} affective problems,⁴¹ recurrent depressive episodes,^31,52 persistent mood disorder,³¹ and dysthymia.^{46,49,50,52,53}

Table 2:

Estimated prevalence of mood disorders by country in Africa

	Study	Survey year	Sample size	Age (years)	Diagnostic criteria	Measure	Mood disorder	Major depressive disorder	Depressive episode	Dysthymia	Bipolar disorder
Point prevalence
Benin	Tognon-Tchégnonsi et al (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	··	32·3	··	··
Burkina Faso	Duthe et al (2016)51	2010	2187	15 and older	DSM-IV	MINI	··	··	4·3	··	··
Burkina Faso	Ouedraogo et al (2019)52	··	2587	18 and older	··	MINI	··	5·1	11·6	10·4	··
Egypt	Ghanem et al (2009)53	2003	14 640	18–64	··	MINI-Plus	6·4	2·7	··	1·0	··
Ethiopia	Fedaku et al (2004)29	1998	1691	16 and older	ICD-10	CIDI or SCAN	··	··	··	··	1·8
Ethiopia	Kebede et al (1999)31	1994	1420	15 and older	ICD-10	CIDI or SRQ	3·8	··	2·1	··	0·1
Kenya	Jenkins et al (2012)38	2000	876	16–65	ICD-10	CIS-R	··	··	0·7	··	··
Kenya	Jenkins et al (2015)39	2004–13	1158	16–65	ICD-10	CIS-R	··	··	0·9	··	··
Morocco	Kadri et al (2010)55	2004–05	5498	15 and older	DSM-IV	MINI	··	26·5	··	··	3·2
Mozambique	Audet et al (20l8)58	2014	3892	16–62	DSM-IV	PHQ-8	··	14·0	··	··	··
Nigeria	Adewuya et al (2018)43	2015	11 246	18–40	DSM-IV	MINI-5, PHQ-9	··	5·5	··	··	··
Rwanda	Bolton et al (2002)60	1999	368	18 and older	DSM-IV	DHSCL	··	15·5	··	··	··
Rwanda	Umubyeyi et al (2014)61	2011–12	917	20–35	DSM-IV	MINI-5	··	··	19·6	··	··
South Africa	Rumble et al (1996)28	1992	560	15 and older	··	SRQ PSE-CATEGO	··	2·0	··	··	··
South Africa	Smit et al (2006)24	2002	645	15 and older	DSM-IV	CES-D	··	33·2	··	··	··
Uganda	Bolton et al (2004)47	2000	587	18 and older	DSM-IV	DHSCL, WHODAS	··	21·0	··	··	··
Uganda	Kinyanda et al (2013)49	··	1587	3–19	DSM-IV	MINI-KID	··	··	7·6	··	··
Period prevalence
Benin	Tognon-Tchégnonsi et al (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	··	11·6	··	··
Nigeria	Gureje et al (2007)45	2003–04	2152	65 and older	DSM-IV	CIDI	··	7·1	··	··	··
Nigeria	Gureje et al (2006);46 Kessler etal (2007),19 (2009),18 (2015)20	2001–03	4985–6752	18 and older	DSM-IV	CIDI	1·1–1·3	1·1	··	0·1	0·0
South Africa	Herman et al (2009);17 Kessler et al (2007),19 (2009),18 (2015);20 Stein et al (2008);25 Tomlinson et al (2009)27	2002–04	4351	18 and older	DSM-IV	CIDI	4·9	4·9	··	··	··
South Africa	Peltzer et al (2013)22	2007–09	3840	50 and older	ICD-10	CIDI	··	4·0	··	··	··
Lifetime prevalence
Ethiopia	Kebede et al (1999)31	1994	1420	15 and older	ICD-10	CIDI or SRQ	5·0	··	··	··	0·3
Ethiopia	Kebede et al (2005);33 Negash et al (2005)35	1998–2001	68 478–68 491	15–49	ICD-10 and DSM-IV	CIDI or SCAN	··	0·3	··	··	0·5
Kenya	Kwobah et al (2017)40	2015–16	420	15 and older	DSM-5	MINI	··	12·6	··	··	5·2
Nigeria	Gureje et al (2007)45	2003–04	2152	65 and older	DSM-IV	CIDI	··	26·2	··	··	··
Nigeria	Gureje et al (2006);46 Kessler et al (2007);19 (2009);18 201520	2001–03	4985–6752	18 and older	DSM-IV	CIDI	3·3–4·1	3·1	··	0·2	0·0
South Africa	Herman et al (2009);17 Kessler et al (2007),19 (2009),18 (2015);20 Stein etal (2008);25 Tomlinson et al (2009)27	2002–04	4351	18 and older	DSM-IV	CIDI	9·8	9·7–9·8	··	··	··

CIDI=Composite International Diagnostic Interview. CIS-R=Clinical Interview Schedule—Revised. DHSCL=Depression Section of the Hopkins Symptom Checklist. MINI=Mini-International Neuropsychiatric Interview. PHQ=Patient Health Questionnaire. PSE-CATEGO=Present State Examination computerised diagnostic system. PSQ=Psychosis Screening Questionnaire. SCAN=Schedule for Clinical Assessment in Neuropsychiatry. SRQ=Self-Reporting Questionnaire. SSI=Semi-structured Interview. WHODAS-II=WHOS Disability Assessment Schedule 2.0.

Figure 2: Range of prevalence estimates by disorder as compared with the World Mental Health Survey estimates¹⁹.

Range of point, period, and lifetime prevalence estimates for studies done in Africa. The shapes display lifetime prevalence estimates from selected countries included in the World Mental Health Survey initiative.

18 studies estimated the prevalence of anxiety and stress-related disorders. The period and lifetime prevalence of anxiety disorders ranged from 4·1% to 8·1% for period, and from 5·7% to 15·8% for lifetime prevalence (table 3). 12 of these studies reported the point prevalence of generalised anxiety disorder,^{17,25,38,39,43,46,50,52,53,55,56,61} which ranged from 0·9% to 36·5% (table 3). The point prevalence of generalised anxiety disorder, often referred to as the presence of symptoms during the past 6 months, is consistent with the DSM-IV definition of current generalised anxiety disorder. Only one study reported a point prevalence of generalised anxiety disorder greater than 10%, which the authors attributed to the high prevalence of violence and history of trauma in post-genocide Rwanda. Three studies reported the past-year and lifetime prevalence of generalised anxiety disorder and estimated that 0·0–1·4% of the target population met the criteria for generalised anxiety disorder in the past year, and 0·1–4·0% in their lifetime. The prevalence of generalised anxiety disorder was consistently higher among females than males. Among children, the estimated point prevalence of anxiety problems was 12·6%.⁴¹ Other anxiety disorders included panic disorder,^{17,25,38,39,46,50,52,53,55,56} agoraphobia,^{17,25,38,39,46,52,53,55,56} social phobia,^{17,25,38,39,46,50,52,53,55,56} specific phobia,^{30,38,39,46,53} obsessive-compulsive disorder,^{38,39,46,50,53,55,56} neurosis,^28,30,53 and posttraumatic stress disorder.^{17,24,25,38–40,46,50,52,53,55,56,61}

Table 3:

Estimated prevalence of anxiety disorders by country in Africa

	Study	Survey year	Sample size	Age (years)	Diagnostic criteria	Measure	Anxiety disorders	Generalised anxiety disorder	Panic disorder	Agoraphobia	Obsessive-compulsive disorder	Social phobia	Specific phobia	Post-traumatic stress disorder
Point prevalence
Benin	Tognon-Tchégnonsi et al (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	9·6	3·5	··	20·7	14·3	··	11·1
Burkina Faso	Ouedraogo et al (2019)52	··	2587	18 and older	··	MINI	··	4·0	5·0	3·6	··	2·7	··	4·1
Egypt	Ghanem et al (2009)53	2003	14 640	18–64	··	MINI-Plus	4·8	0·9	0·7	··	0·7	0·2	1·4	0·11
Kenya	Jenkins et al (2012)38	2000	876	16–65	ICD-10	CIS-R	··	1·6	2·6	··	0·2	··	0·3	··
Kenya	Jenkins et al (2015)39	2004–13	1158	16–65	ICD-10	CIS-R	··	0·7	3·1	··	1·4	··	0·4	··
Morocco	Kadri et al (2007)56	1994	800	15–80	DSM-IV	MINI	··	4·3	2·0	7·6	6·4	3·4	14·3	3·4
	Kadri et al (2010)55	2004–05	5498	15 and older	DSM-IV	MINI	··	9·3	6·6	9·4	6·6	6·3	··	2·1
Nigeria	Adewuya et al (2018)43	2015	11 246	18–40	DSM-IV	MINI-5	··	3·5	··	··	··	··	··	··
Rwanda	Umubyeyi et al (2014)61	2011–12	917	20–35	DSM-IV	MINI-5	··	36·5	··	··	··	··	··	13·6
South Africa	Smit et al (2006)24	2002	645	15 and older	DSM-IV	LEC, HTQ	··	··	··	··	··	··	··	14·9
Period prevalence
Nigeria	Gureje et al (2006)46	2001–03	4985–6752	18 and older	DSM-IV	CIDI	41	0·0	0·1	0·2	0·1	0·3	3·5	0·0
South Africa	Herman et al (2009)17	2002–04	4351	18 and older	DSM-IV	CIDI	8·1	1·4	0·8	4·8	··	1·9	··	0·6
Lifetime prevalence
Benin	Tognon-Tchégnonsi et al (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	··	9·0	··	··	··	··	··
Kenya	Kwobah et al (2017)40	2015–16	420	15 and older	DSM-5	MINI	15·7	··	··	··	··	··	··	4·5
Morocco	Kadri et al (2007)56	1994	800	15–80	DSM-IV	MINI	··	··	2·3	8·4	··	··	··	··
Nigeria	Gureje et al (2006)46	2001–03	4985–6752	18 and older	DSM-IV	CIDI	5·7	0·1	0·2	0·4	0·1	0·3	5·4	0·0
South Africa	Herman et al (2009)17	2002–04	4351	18 and older	DSM-IV	CIDI	15·8	2·7	1·2	9·8	··	2·8	··	2·3

CIDI=Composite International Diagnostic Interview. CIS-R=Clinical Interview Schedule—Revised. HTQ=Harvard Trauma Questionnaire. LEC=Life Event Check-list. MINI=Mini-International Neuropsychiatric Interview.

The prevalence of psychotic disorders

Nine studies estimated the prevalence of psychotic disorders. Four studies examined psychosis or psychotic disorders or syndromes generally,^40,52,53,57 three studies assessed schizophrenia,^28,29,34 and one study assessed both schizophrenia and schizoaffective disorder (table 4, figure 2).³² The point prevalence of psychosis in a nationally representative study in Egypt was 0·19%. The lifetime prevalence of psychosis was assessed in three different settings and ranged from 1% in a rural setting in Kenya, and from 1·6% in an urban setting in Mozambique, to 4·4% in a rural region of Mozambique. Two studies assessed psychotic syndrome; one of these studies done in Burkina Faso reported that 1·7% of the sample had isolated psychotic syndrome while the prevalence of recurrent psychotic syndrome was 4·1%.⁵² In a separate study done in Benin, the point prevalence of psychotic syndrome was 9·3% and the lifetime prevalence of psychotic syndrome was 30·2%.⁵⁰ Another study examined the presence of psychotic symptoms without applying a diagnostic algorithm and found that 8% of people aged 16–65 years in a rural region of Kenya displayed at least one symptom of psychosis in the past year. Three of four studies evaluating the prevalence of schizophrenia and schizoaffective disorder were done in Ethiopia. These studies found that 0·06–0·3% of the population met criteria for schizophrenia, whereas the lifetime prevalence ranged from 0·40% in the capital city to 0·47% in a rural region of Ethiopia. One study assessed the point prevalence of paranoid schizophrenia in a rural setting in South Africa, which was estimated to be 5%.²⁸ The only study evaluating schizoaffective disorder found that the point and lifetime prevalence of schizoaffective disorder in Addis Ababa, Ethiopia was 0·4% for point and 0·5% for lifetime prevalence. The prevalence of schizophrenia was comparable between males and females; however, males were 2·5–3·5 times as likely to meet the current or lifetime criteria for schizoaffective disorder.

Table 4:

Estimated prevalence of psychotic disorders by country in Africa

	Study	Survey year	Sample size	Age (years)	Diagnostic criteria	Measure	Psychosis or psychotic disorder	Schizophrenia	Schizoaffective disorder	Paranoid schizophrenia	Recurrent psychotic syndrome
Point prevalence
Benin	Tognon-Tchégnonsi et al (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	··	··	··	9·3
Burkina Faso	Ouedraogo et al (2019)52	··	2587	18 and older	··	MINI	··	··	··	··	··
Egypt	Ghanem et al (2009)53	2003	14 640	18–64	··	MINI-Plus	0·19	··	··	··	··
Ethiopia	Fedaku et al (2004)29	1998	1691	16 and older	ICD-10	CIDI or SCAN	··	0·06	··	··	··
Ethiopia	Kebede et al (1999)32	1994	1420	15 and older	ICD-10	CIDI or SRQ	··	0·30	0·40	··	··
Ethiopia	Kebede et al (2003)34	1998–2001	68 378	15–49	ICD-10 and DSM-IV	CIDI or SCAN	··	··	··	··	··
Kenya	Kwobah et al (2017)40	2015–16	420	15 and older	DSM-5	MINI	··	··	··	··	··
Mozambique	Patel et al (2007)57	2014	3892	16–62	DSM-IV	PHQ-8	··	··	··	··	··
South Africa	Rumble et al (1996)28	1992	560	15 and older	··	SRQ PSE-CATEGO	··	··	··	5·0	··
Lifetime prevalence
Benin	Tognon-Tchégnonsi etal (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	··	··	··	30·2
Burkina Faso	Ouedraogo et al (2019)52	··	2587	18 and older	··	MINI	··	··	··	··	4·1
Ethiopia	Kebede et al (1999)32	1994	1420	15 and older	ICD-10	CIDI or SRQ	··	0·40	0·50	··	··
Ethiopia	Kebede et al (2003)34	1998–2001	68 378	15–49	ICD-10 and DSM-IV	CIDI or SCAN	··	0·47	··	··	··
Kenya	Kwobah et al (2017)40	2015–16	420	15 and older	DSM-5	MINI	1·0	··	··	··	··
Mozambique	Patel et al (2007)57	2014	3892	16–62	DSM-IV	PHQ-8	1·6/4·4	··	··	··	··

CIDI=Composite International Diagnostic Interview. MINI=Mini-International Neuropsychiatric Interview. PHQ=Patient Health Questionnaire. PSE-CATEGO=Present State Examination computerised diagnostic system. SCAN=Schedule for Clinical Assessment in Neuropsychiatry. SRQ=Self-Reporting Questionnaire.

The prevalence of substance use disorders

19 studies estimated the prevalence of alcohol or other drug misuse problems and disorders. In samples that enrolled adolescents and adults, the past-year and lifetime prevalence of any substance use disorder (ie, alcohol or other drugs) ranged from 0·8% to 5·8% for past-year and from 3·7 to 13·4% for lifetime prevalence (table 5, figure 2). The case definitions for alcohol-related problems varied and included hazardous or harmful alcohol use,^{23,26,37,43,44,48,52,59} alcohol problems,^15,16,21 or alcohol use disorder (misuse or dependence).^{14,17,24,25,44,46,48,50,52,53,55} The point, period, and lifetime prevalence of alcohol use disorder ranged from 0·03% to 16·7% for point, from 0·5% to 35·5% for period, and from 1·2% to 14·0% for lifetime prevalence. The point, period and lifetime prevalence of non-alcohol substance use disorder ranged from 0·13% to 5·8% for point, from 0·2% to 1·5% for period, and from 1·0% to 4·5% for lifetime prevalence.^{17–19,25,40,43,44,46,50,52–55} For studies that assessed both alcohol use disorder and other drug use disorders, studies done in Benin, Burkina Faso, Nigeria, and South Africa found the prevalence of alcohol use disorder to be approximately 2–3 times higher than that for drug use disorders. In contrast, studies done in Egypt and Morocco found that other drug use disorders were up to twice as prevalent as alcohol use disorder. The prevalence of alcohol use disorder was 1·8–5·9 times more common among males compared with females, with the exception of Morocco, where the prevalence of alcohol use disorder was 14·5 times as prevalent among males than females. The prevalence of drug use disorder was consistently higher among males than females. Notably, in Morocco, the point prevalence of drug use disorder was 25 times higher among men (10%) than women (0·4%).⁵⁵

Table 5:

Estimated prevalence of substance use disorders in Africa

	Study	Survey year	Sample size	Age (years)	Diagnostic criteria	Measure	Hazardous alcohol use	Harmful or problematic alcohol use	Alcohol use disorder	Substance use disorder	Alcohol or other drug use disorder
Point prevalence
Benin	Tognon-Tchégnonsi et al (2020)50	2013	603	18 and older	DSM-IV	CIDI	··	··	4·0	1·3	··
Burkina Faso	Ouedraogo et al (2019)52	··	2587	18 and older	··	MINI	··	0·1	1·0	0·5	··
Egypt	Ghanem et al (2009)53	2003	14 640	18–64	··	MINI-Plus	··	··	0·03	0·13	··
Ethiopia	Gedif et al (2019)37	2018	1588	18 and older	··	ASSIST	25·8	··	··	··	··
Ghana	Thapa et al (2014)26	2007–09	4289	50 and older	··	NIAAA	··	7·0	··	··	··
Morocco	Kadri et al (2010)55	2004–05	5498	15 and older	DSM-IV	MINI	··	··	3·4	5·8	··
Mozambique	Wainberg et al (2018)57	2014	3892	16–62	DSM-IV	AUDIT	8·0	··	··	··	··
Nigeria	Adewuya et al (2018),43 (2020)44	2015	11 246	18–40	DSM-IV	MINI-5, PHQ-9	··	8·7	7·1	2·1	··
South Africa	Thapa et al (2014)26	2007–09	3668	50 and older	··	NIAAA	··	4·4	··	··	··
South Africa	Smit et al (2006)24	2002	645	15 and older	DSM-IV	AUDIT	··	··	16·7	··	··
Period prevalence
Nigeria	Gureje 2006;46 Kessler et al (2007),19 (2009),18 (2015)20	2001–03	2143–4985	18 and older	DSM-IV	CIDI	··	··	0·6	0·2	0·8
South Africa	Andersson et al (2018)14	2012	977	18–40	DSM-IV	MINI, SSI	··	··	35·5	··	··
South Africa	Herman et al (2009);17 Kessler et al (2007),19 (2009),19 (2015);20 Stein et al (2008)25	2002–04	4351	18 and older	DSM-IV	CIDI	··	··	5·7	1·5	5·8
Uganda	Nalwadda et al (2018)48	2018	351	18 and older	··	AUDIT	2·9	0·7	0·5	··	··
Lifetime prevalence
Egypt	Hamdi et al (2013)54	2006	44 000	15 and older	ICD-10	ASI	··	··	··	··	1·6
Kenya	Kwobah et al (2017)40	2015–16	420	15 and older	DSM-5	MINI	··	··	··	··	11·7
Kenya	Gureje 2006;46 Kessler et al (2007),19 (2009),18 (2015)20	2001–03	2143–4985	18 and older	DSM-IV	CIDI	··	··	3·0	1·0	3·7
South Africa	Medical Research Council (2007)16	2003	10 214	15 and older	··	CAGE	··	··	21M / 7F*	··	··
South Africa	Herman et al (2009);17 Kessler et al (2007),19 (2009),18 (2015);20 Stein et al (2008)25	2002–04	4351	18 and older	DSM-IV	CIDI	··	··	14·0	4·5	13·3
South Africa	Parry et al (1998)21	1998	13 826	15 and older	··	CAGE	··	··	28M / 10F*	··	··

ASSIST=the Alcohol, Smoking, and Substance Involvement Screening Test. CIDI=Composite International Diagnostic Interview. CIS-R=Clinical Interview Schedule—Revised. CAGE=Cut-Annoyed-Guilty-Eye questionnaire. MINI=Mini-International Neuropsychiatric Interview. NIAAA=National Institute on Alcohol Abuse and Alcoholism. PHQ=Patient Health Questionnaire.

^*Study only reported gender-stratified estimates (M: male; F: female).

The prevalence of other psychiatric disorders

Other psychiatric disorders assessed included somatoform disorders or problems (point prevalence 0·7–11·8%; lifetime prevalence 3·1%),^30,42,53 dissociative disorders (point prevalence 0·4%; lifetime prevalence 0·8%),³⁰ insomnia,⁵² internalising problems (point prevalence 19·3%),⁴² impulse control and externalising disorders (point prevalence of oppositional problems 2·3%, or externalising problems 10·2%; period prevalence of externalising disorders 0·1%; and lifetime prevalence of impulse control disorder 0·3%),^18,19,41,42 attention or attention-deficit hyperactivity problems (point prevalence 2·3–5·0%),^41,42 antisocial personality disorder (lifetime prevalence 3·1%),^40,50 eating disorders (lifetime prevalence 1·7%),^40,50 intellectual disability (lifetime prevalence 1·3–1·9%),⁵⁷ pervasive developmental problems (point prevalence 5·3%),⁴¹ seizure disorders (lifetime prevalence 1·6%),⁵⁷ and suicide risk and behaviours (lifetime prevalence 6·7–16%; point prevalence 4·2%, low risk–0·6% high risk).^40,52,61 Several studies identified high rates of comorbidity between variable combinations of psychiatric disorders.^{24,25,43,44,53}

Discussion

In this Scoping Review, we identified 36 studies estimating the prevalence of one or more psychiatric disorders in the general population in 12 African countries, half of which were published after 2008. Studies reported lifetime prevalence estimates of 3·3–9·8% for mood, from 5·7% to 15·8% for anxiety, from 3·7% to 13·3% for substance use, and from 1·0% to 4·4% for psychotic disorders. Epidemiological studies of psychiatric disorders in Africa use comparable diagnostic interviews and screening tools to those studies done outside of the continent. However, this Scoping Review revealed sources of methodological and clinical heterogeneity and gaps in research.

First, there are gaps in coverage of epidemiological estimates across countries, populations, and disorders. Only 12 African countries were represented in this Scoping Review, none of which were in central Africa. More than half of the population-based psychiatric epidemiology surveys in Africa were done in only five countries— Ethiopia, Kenya, Nigeria, South Africa, and Uganda. The included studies most frequently assessed mood and alcohol use disorders. Few studies examined psychotic disorders.

Second, studies estimating the point prevalence of psychiatric disorders—particularly mood, anxiety, and substance use disorders—displayed substantial meth odological variation in measurement approaches. Most studies used a diagnostic interview or screening tool, similar to those commonly used in other world regions, to measure psychiatric disorders. Studies that used screening tools often reported higher rates of psychiatric disorders than studies that utilised diagnostic interviews. For example, studies that estimated the prevalence of alcohol-related problems reported higher estimates when using the AUDIT or CAGE than the WHO World Mental Health Composite International Diagnostic Interview and MINI.

Previous meta-analyses similarly found that the estimated prevalence of psychiatric disorders in countries outside Africa is greater for studies using screening tools than diagnostic interviews.⁸¹ Other sources of measurement error could be due to differential misclassification by cultural context. Previous analyses of the WHO World Mental Health Composite International Diagnostic Interview diagnoses from countries in Africa (eg, Nigeria and South Africa), New Zealand, and the USA suggest that these diagnostic interviews underestimate the prevalence of depression in African countries compared with high-income countries due to differential item performance and relevance across countries.⁸²

It is also possible that screening and diagnostic tools do not capture the symptoms or features of psychiatric disorders that present in specific cultural contexts. For example, observed variation in substance use disorder could be partially attributable to the application of tools that do not cover all types of substances or use local terms for different types of substances. Khat, a stimulant plant with amphetamine-like properties, is widely used and culturally accepted in many parts of east Africa, and has been associated with psychosis and psychotic symptoms.⁸³ Most epidemiological screening and diagnostic tools do not capture khat use and related problems, which might underestimate substance use disorder in regions where khat use is prevalent. Scales that have been adapted to assess khat use were not included in the studies we identified in this Scoping Review,⁸⁴ which suggests the need to broaden the scope of substances to include those relevant to local use. In general, accurate measurement and the training of experts in the development and adaptation of screening and diagnostic tools is one of the greatest challenges for doing psychiatric epidemiological studies in all parts of the world.

Capacity in psychometrics and measurement adaptation should be strengthened and can leverage the training and capacity that has been developed for other disorders, such as epilepsy and other neurological disorders.⁷ There is also a need to improve reporting of these estimates. For example, several studies and data points were excluded because we did not specify the type of prevalence reported.

Third, although this scoping review identified consistencies in the epidemiology of DSM and ICD psychiatric diagnoses with previous studies from countries outside of Africa, it is probable that applying these diagnostic tools excluded culturally-specific presentations of mental health problems, idioms of distress, or culture-bound syndromes.⁸⁵ The relevance of universal application of psychiatric diagnoses from high-income countries without cultural formulation and considerations is debated.⁸⁶ The absence of culture-bound syndromes and little representation of local measurement approaches could also reflect limitations in our review process. Notably, we restricted studies to those published in English, French, and Portuguese and did not search grey literature databases. The official languages in most African countries include Afroasiatic languages and Niger–Congo languages; thus, this scoping review might have missed eligible studies published in these common languages. Potential measurement error resulting from a shortage of cross-cultural validity, different assessment approaches (ie, diagnostic interviews vs screening tools), and unique presentations of psychiatric disorders and culture-bound syndromes suggest observed differences in the prevalence of psychiatric disorder could be due, in part, to methodological and measurement differences between studies.

Results from this scoping review revealed notable heterogeneity in prevalence estimates across studies, which could be due to the methodology, but also to differences in population characteristics and risk factors related to cultural norms and behaviours, social determinants of health, genetic differences, and geographical or contextual differences. Most psychiatric disorders, including those with high heritability, displayed comparable or lower estimated prevalence in studies done in Africa compared with large epidemiological studies done outside of Africa (figure 2).^{18,19,87–95} For example, the range in lifetime prevalence of mood disorders (3·3–9·8%) and anxiety disorders (5·7–15·8%) was lower than that reported in countries from other world regions, including Colombia (mood 14·6%; anxiety 25·3%), France (mood 21·0%; anxiety 22·3%), Lebanon (mood 12·6%; anxiety 16·7%), and the USA (mood 21·4%; anxiety 31·0%), but similar to that observed in Japan (mood 7·6%; anxiety 6·9%).^19,87–95 The lifetime prevalence estimates for bipolar disorder were consistent with estimates from countries outside of Africa. The lifetime prevalence of alcohol use disorder was comparable to other countries (Colombia 9·4%; France 7·1%; Japan 7·3%; Lebanon: 1·6%; and USA 13·8%). The lifetime prevalence of substance use disorder (1–4·5%) was higher than that reported for Japan (0·2%) and Lebanon (0·4%), but lower than the lifetime prevalence of substance use disorder in the USA (8·5%). Comparison with these other studies must be interpreted with caution given substantial clinical and methodological heterogeneity and the absence of meta-analytic summary estimates that preclude statistical comparison of these estimates across world regions.

It is possible that observed differences in the prevalence of psychiatric disorders are due to genetic and environmental risk or protective factors, but could also be explained by confounders, stigma, and different cultural perceptions of mental disorders that can prevent people from reporting psychiatric symptoms. For example, Africa has the youngest population globally, which could partly explain the lower observed general population prevalence estimates. Within studies done in Africa, we also observed substantial heterogeneity by region. The prevalence of alcohol use disorder was consistently lowest in north African countries and highest in South Africa. With the exception for the studies from South Africa, the prevalence of alcohol use disorder was similar to previous studies from low-income and middle-income countries, and lower than those reported from high-income countries.^89,96 The prevalence of drug use disorder in Morocco and South Africa was higher than most other low-income and middle-income countries, yet comparable to estimates from high-income countries.⁹⁷ Regional differences in the prevalence of substance use disorders can reflect the differences in the types of substances consumed and cultural norms related to substance use.

The relative prevalence of psychiatric disorders by demographic subgroups displayed some consistencies and deviations from patterns observed outside of Africa. Gender differences in the prevalence of psychiatric disorders in Africa were consistent with studies outside of Africa, whereby major depression and generalised anxiety disorders were more common among females, whereas substance use disorders were more prevalent among males.^{88,89,91,96,97} With regard to age, we found that the prevalence of emotional and behavioural problems among children was comparable to what has been observed outside of Africa,^42,98,99 yet studies restricted to older adults as compared with studies including samples across adulthood (≥18 years) produced higher period and lifetime prevalence estimates for major depressive disorder. This differs from epidemiological studies of depression in high-income settings where the morbidity and period prevalence are often highest in early and middle-adulthood.^19,93,100 A study of psychosis in Mozambique found a higher prevalence in rural settings compared with urban settings, which is in contrast to research from the USA and western European countries that have consistently found a higher prevalence of schizophrenia in urban settings as compared with rural settings.^101–103

To better understand the global epidemiology of psychiatric disorders and set the foundation for future genetic research, studies that are methodologically comparable, but diverse in population characteristics, culture, and context are required. Epidemiological and population-based research is needed to investigate factors that explain the observed heterogeneity of psychiatric disorders in Africa, and could improve our understanding of the epidemiology of psychiatric disorders by disentangling methodological explanations for these differences from meaningful risk and protective factors.

Strengthening the infrastructure and capacity for doing psychiatric epidemiological research in Africa will facilitate understanding of the role of genetic and environmental factors while also building equitable partnerships and ownership of psychiatric genetic epidemiology research among scientists in Africa. There is a rich history of genetics and epidemiological research on HIV/AIDS,^104,105 malaria,¹⁰⁶ and chronic and infectious diseases in Africa,¹⁰⁷ including one of the first large-scale genomic studies of HIV and tuberculosis coinfection,¹⁰⁸ which have substantially advanced the global knowledge of the cause of these diseases. Future research should now include efforts to understand and measure culture-bound syndromes, unique presentations of and measurement considerations for DSM and ICD psychiatric disorders, and relevant disorders that were not covered in this scoping review (eg, autism spectrum disorder, and epilepsy). Similar to the advances made in the epidemiology of other chronic and infectious diseases, investment in research on psychiatric disorders in Africa has the potential to develop a better understanding of the epidemiology and genetics of psychiatric disorder and ultimately to inform contextually relevant policies and practices aimed at reducing the burden of mental disorder and improving public mental health equity globally.