Background

Mental health disorders account for at least 14% of the global burden of disease [1], underlining its importance as a global public health concern. Three-quarters of people with mental health problems live in low- and middle-income countries (LMICs) [2, 3], where 75–85% with severe mental health concerns receive little or no treatment [4]. This large treatment gap [5] has multiple underlying causes, including a scarcity in the number of trained mental health professionals [6], lack of advocacy and awareness, as well as the associated stigma of mental illness. When comparing mental health professional availability between high- and LMICs, Europe on average has nine psychiatrists for every 100,000 of the population, while Africa has 0.05 psychiatrists per 100,000 [7]. In addition to the limited mental health services in LMICs, there is also a lack of contextually relevant, rigorous, mental health research in many of these geographical areas [8].

Mental health problems represent the greatest burden of disease among children and adolescents [9]. Ninety percent of children and adolescents live in LMICs, yet only 10% of all child and adolescent mental health (CAMH) research has been conducted in LMICs [8]. In sub-Saharan Africa, projected to be home to 40% of the world’s children by 2025, there are only about 60 qualified child and adolescent psychiatrists and very limited CAMH services. Given the rising incidence of mental health disorders and concomitant resource deficiencies, the treatment gap in Africa is widening [10, 11].

Identification, evaluation and implementation of simple, short and freely-available screening tools for CAMH difficulties may offer a powerful strategy to close the treatment gap, as it may enable the identification of children and adolescents in need of next-step evaluation and treatment. Screening tools might be particularly useful in primary care [12] and in educational settings. Currently a number of screening tools, including the Child Behaviour Checklist [13], SNAP-IV [14], Conners’ ADHD Rating Scales [15] and Social Communication Questionnaire [16] are used in Africa, but it is not clear where or how consistently these tools are used, or to what extent these have been evaluated for their reliability, validity and cultural appropriateness for the diverse populations and contexts in Africa [17].

In CAMH settings around the globe, the Strengths and Difficulties Questionnaire (SDQ) is widely used as a behavioural screening tool, as it has several advantages. It is relatively short, allows for rapid administration, measures both mental health difficulties and competencies, and can be administered by a non-professional with minimal training [18, 19]. The SDQ was developed by Goodman and colleagues [20] in the UK as an open-access, downloadable screening tool, available as a self-report (SDQ-S), parent/caregiver (SDQ-P) and teacher report version (SDQ-T). Goodman [18] recommended the optimal use of the instrument to be a multi-informant tool, with triangulation between parent, teacher and (where appropriate) adolescent self-report. Authorized translations of the tool are available on the SDQ website (http://www.sdqinfo.org) in 83 languages. A strict process of translation, back-translation and authorisation of the tools is maintained by the authors. These processes were designed to ensure the availability of the SDQ in a number of languages whilst maintaining the integrity of the instrument.

The SDQ consists of 25 items to assess a range of ‘strengths’ and ‘difficulties’ as behavioural markers of potential mental health problems. The items contribute to five subscales of five items each with a minimum score of 0 (lowest score) to 10 (highest score): conduct problems, hyperactivity/inattention, emotional symptoms, peer problems, and prosocial behaviour. The sum of the first four subscales generates a total difficulties score, which can range from 0 to 40. From the total difficulties and subscale scores, cut-off scores for clinical ‘caseness’ can be generated. The top 10% of scores based on UK population norms were used to define the ‘abnormal’ range, the next 10% as the ‘borderline’ range, and the remaining 80% of scores as the ‘normal range’ [20]. A higher total difficulty score indicates a greater likelihood of significant problems. Four of the five subscales are scored in a similar way with higher scores indicating more difficulties. The prosocial subscale provides a reverse score where higher scores indicate more prosocial behaviours or strengths.

The reliability and validity of the SDQ has been examined in a number of studies across Europe [20], Asia, Australia and South America [21, 22], but with little or no reference to the use of the SDQ on the African continent. Given the growing awareness of the CAMH needs in Africa [11], it seemed timely to establish the landscape of all research ever performed in Africa that used any versions of the SDQ. We therefore set out to conduct a comprehensive scoping review of the SDQ aimed to describe the use of the SDQ, and to examine the reliability and validity of the SDQ for local use in Africa.

Methods

The methodological framework for scoping reviews [23, 24] was followed. This included identifying the research question, searching for relevant studies, selecting studies, charting and summarizing the data, and reporting the results. The review objective, inclusion criteria and study methods for this scoping review were specified in advance. Inclusion criteria were (i) any of the versions of the SDQ was used in the study, (ii) the study took place in Africa, (iii) the article was data-driven (i.e. not a review paper), and (iv) the article had been peer-reviewed. We also expressly included studies performed in any language and with no time limit since the development of the SDQ. Given that this was a comprehensive scoping review and not a systematic review, no articles were excluded on the basis of any quality criteria.

Search strategy

A literature search to identify studies that used the SDQ to evaluate CAMH in Africa since its development in 1997, was conducted until December 2016. Online databases Ebscohost (Africa Wide Information, Medline, PsycINFO) and PubMed were searched with no date limitations, or language restrictions for any African study involving use of the SDQ, in December 2016 and a follow-up search was conducted in April 2017. A general search of Google Scholar was also conducted. The databases were searched using the following keywords: ‘Strengths and Difficulties Questionnaire’, ‘SDQ’, ‘Africa’, ‘children’, ‘adolescents’, and ‘mental health’. An additional search was conducted which included the use of ‘reliability’ and ‘validity’. Searches were not restricted to any search date, but included all available published studies. An additional search was conducted using key words ‘Strengths and Difficulties Questionnaire’, ‘SDQ’ combined with ‘Africa’, ‘adolescents’ and ‘mental health’. The terms ‘reliability’ and ‘validity’ were also used in addition to the combination of the above terms used. Additionally, examination of relevant bibliographies provided further references for review. Titles and abstracts were examined using the inclusion criteria, after which full articles were retrieved.

The initial online search produced 216 articles. Titles were screened for eligibility and 99 articles were identified. A further 37 articles were generated from reference lists and other sources, producing a total of 136 articles. Duplicates were removed, reducing the sample to 91 articles. The abstracts of these were then reviewed to confirm study location and use of the SDQ, producing a total of 72 articles. The 19 articles excluded in the abstract screening phase included review papers, those not conducted in Africa, and articles that used tools with the same acronym as the SDQ (e.g. self-description questionnaire). Eighteen of these were found to be review articles (not apparent from abstract review), reports, one study on African populations living outside of Africa, and presentations using the SDQ in Africa, and were thus excluded. Two independent reviewers assessed the articles for eligibility (NH, ELD). Disagreements were resolved by consensus between the reviewers and, in cases of an impasse, the two senior authors (PJdV, MSK) made the final decision. Figure 1 outlines the process involved in the literature review and final selection of articles. The final sample consisted of 54 articles included in the review (Table 1).

Fig. 1
figure 1

Schematic representation of the literature search process

Full size image
Table 1 Strengths and difficulties questionnaire in Africa: data extraction
Full size table

Data extraction

Full PDF versions of all included articles were collated by the first author and analysed by the first and second author to extract relevant information for the review. A summary table was generated (Table 1) which included the following fields: author, year, study location, participant age, study samples, aim of the paper, theme of study, SDQ report version used, SDQ language version used, translation process, clinical cut-off scores used, comparison to UK norms, internal consistency of the SDQ, as well as the results of each study.

Data analysis

The data gathered and extracted from the articles were analysed using thematic analysis. In keeping with the aim of the review to examine the use and cultural appropriateness of the SDQ in Africa, the deductive themes extracted included specific examination of the location of studies, languages used, instrument translations, cultural comments about use, and psychometric properties of the instrument. Other inductive themes which emerged during extracting and synthesis of data, were research theme of use, versions and subscales used. All findings are summarised in Table 1.

Results

The review identified 54 studies [25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78], conducted in 12 African countries (Table 1). Below we summarise data based on (i) sociodemographic descriptors, (ii) SDQ versions, (iii) SDQ triangulation, (iv) tool properties and validation, (v) translation, back-translation and authorization, and (vi) the purpose of using the SDQ in Africa.

Sociodemographic descriptors

The 12 countries represented in the review included South Africa (n = 21 plus an additional 3 from two-country studies between South Africa and Malawi), Nigeria (n = 6 plus an additional study from a three-country study between Nigeria, Angola and Ghana), Ghana (n = 5, plus the additional three-country study with Nigeria and Angola), Uganda (n = 5), Tanzania (n = 3, plus an additional three-country study between Tanzania, Kenya and Ethiopia), Kenya (n = 2, plus the additional three-country study), Zambia (n = 3), Democratic Republic of Congo (n = 2), Angola (n = 1, based on the three-country study with Nigeria and Ghana), Egypt (n = 1), Ethiopia (n = 1, based on the three-country study with Tanzania and Kenya), and Malawi (n = 1, plus the additional 3 two-country studies with South Africa) (Fig. 2). As listed above, five studies represented samples from two countries, namely South Africa and Malawi in three studies, and two three-country studies, namely Tanzania, Kenya and Ethiopia as well as Nigeria, Angola and Ghana. All studies included male and female participants who ranged in age between 31 months and 24 years. Interestingly, 16 studies (29.6%) administered the self-report version to participants as young as six and as old as 24 years, despite the SDQ administration guidelines recommending the self-report version for 11–17 year olds [26, 30, 34, 37,38,39, 41, 42, 54,55,56, 62, 64, 66, 70, 75].

Fig. 2
figure 2

Geographical location of SDQ studies in Africa

Full size image

SDQ versions used

Of the 54 studies, four (7.4%) used all three versions (parent, teacher and self-report) of the SDQ [44, 49, 50, 57] (see Fig. 3). Seven studies (12.9%) employed 2 SDQ versions [29, 32, 40, 42, 43, 47, 65]. Thirty-one studies (57.4%) used only one version of the SDQ, of which 22 used the SDQ-S [26, 30, 34, 35, 37,38,39, 41, 45, 46, 53, 55, 56, 58, 60, 62, 64, 69,70,71, 75, 78], four the SDQ-T [25, 27, 36, 59] and five the SDQ-P [33, 48, 61, 67, 68] (see Table 2 for summary of results by SDQ versions). In the remaining 12 articles (22.2%), three used only the peer and conduct problem subscales from the SDQ self-report version [28, 31, 77], one used only the peer and emotional problem subscales [66], while a further two articles used only the conduct subscale of the self-report SDQ [52, 54]. In addition, the self-report prosocial subscale was used together with the parent prosocial subscale [76] and the teacher total difficulty score [63]. Three studies did not specify which version of the SDQ was used [72,73,74]. One study used an unauthorised modified 10-item version of the SDQ-P [51] which assessed total difficulties. Overall, the SDQ self-report version was most frequently used (32 studies, 56.1%), followed by the parent report (15 studies, 26.3%), and the teacher report version (10 studies, 17.5%).

Fig. 3
figure 3

SDQ versions used in studies from Africa

Full size image
Table 2 Summary of results by SDQ version (SDQ-S, SDQ-P and SDQ-T) used
Full size table

SDQ triangulation

Triangulation of SDQ data through the use of multiple informants is recommended by the developers [79]. Only four (7.4%) [44, 49, 50, 57] of the 54 studies in the review made use of all versions (parent, teacher and self-report) of the SDQ to satisfy triangulation of the screening tool. All four studies took place within South Africa, three made use of the Sesotho versions of the SDQ (parent, teacher and self-report), while the other study did not specify the language version used. The remaining 50 studies in the review did not make use of triangulation.

Psychometric properties and validation

Twenty-six studies (48.1%) included an evaluation of some aspect of the psychometric properties of the SDQ, for instance, 25 (46.2%) reported the Cronbach alpha scores for internal consistency [25, 29, 33, 36, 39, 42, 45,46,47,48,49,50, 52,53,54,55, 57, 58, 60, 61, 66, 68, 70, 71, 75, 76]. Four studies (7.4%) also presented Cronbach alpha scores from previously published studies [38, 62, 64, 69]. The overall reported Cronbach alpha scores ranged from 0.18–0.89. For the SDQ-S Cronbach alpha scores were reported in 14 studies (25.9%) and ranged between 0.18 and 0.80. Sixteen studies (29.6%) did not report on the psychometric properties for the SDQ-S. For the SDQ-P Cronbach alpha scores were reported in 9 studies (16.7%) and ranged between 0.24 and 0.73. Six studies (11.1%) did not report on psychometric properties of the SDQ-P. SDQ-T scores ranged between 0.35 and 0.89 in the 6 studies (11.1%) that reported Cronbach alphas, while four studies (7.4%) did not report on internal consistency or other psychometric properties of the SDQ-T (see summary in Table 2). One study made use of inter-item reliability on the SDQ-S [45] and another generated a composite score for all subscales used on the SDQ-S [52]. Overall, twenty-four studies (44.4%) failed to report or examine Cronbach alpha scores. One study examined the factor structure of the SDQ among children in the Democratic Republic of Congo (DRC) using the SDQ-T, and the same study established clinical cut-off scores on the SDQ-T for the DRC population [25]. One study examined the construct validity of all three versions of the SDQ in a sample of Sesotho children and adolescents in South Africa, as well as their caregivers and teachers [50]. To do this, the SDQ was administered alongside the computerised diagnostic interview schedule for children (4th edition; CDISC-IV) to the sample and the subscales were matched with the CDISC-IV diagnostic groups based on the DSM-IV criteria [50]. The emotional problem subscale was matched to detect anxiety disorders and affective disorders on the CDISC-IV, while the conduct problem subscale was matched to detect oppositional-defiant disorder and conduct disorder. In addition, the hyperactivity/inattention subscale was matched to detect ADHD. The peer problem and prosocial behaviour scales were not matched as there were no criteria in the CDISC-IV to find matched equivalent diagnostic groups [50]. The study [50] suggested support for the SDQ-P version, but not for the SDQ-T and SDQ-S, and provided clinical cut-offs with some caution.

Translation, back-translation and authorization

With regards to translation and adaptation of the SDQ, 23 studies (42.6%) reported the process of translation and back-translation of the SDQ. Of these, eight translations were not listed as authorized on the SDQ website and have not been translated in consultation with the tool developers, namely Nyanja [29], Dholou [39], Sesotho [44, 49, 50, 52], Luganda [46], Swati [52], Shangaan [52, 63], Twi [53] and Ga [53]. Challenges raised in the process of translations and adaptions included absence of a linguistic equivalent in some of the local languages for words such as ‘fidget’ [80], ‘nervous’ or ‘fidgeting’ [35]. Of the studies in the review, the SDQ was available in seventeen different languages, while twenty studies did not specify language(s) used. Five studies, two in South Africa [26, 54] and one each in Ghana [53], Tanzania [66] and Nigeria [71] reported the use of interviewer support when using the self-report SDQ, due to low literacy levels. There are no authorised procedure at http://www.sdqinfo.com for interviewer administration of the SDQ. Of the 23 studies reporting on the translation and adaption of the SDQ, only seven provided evidence of the evaluation of translated words and its equivalences [35, 39, 40, 43, 49, 64, 66], evidence of translation team discussions regarding semantics was found in one study [46], and another used a qualitative approach to perform a cognitive review of the translated version of the SDQ [50].

In addition, three studies used English versions of the SDQ and thus did not require translation [34, 37, 78]. Of the remaining 51 studies, three used translated versions of the SDQ from the official site [25, 61, 71] and 27 did not discuss any process of translation. Elhamid and colleagues [32] did not discuss translation of the SDQ used but referenced the Arabic version that had been previously validated.

While some authors discussed the process and implications of instrument translation in both their methods and discussion, this was not universal, and showed varying degrees of detail across studies. Seventeen studies mentioned that back-translations were performed. Four studies did not provide specific details about the translation process, but rather referenced that the process was undertaken according to published guidelines or standard procedures [45, 49, 50, 55].

There is a distinct lack of consistent examination and reporting of translation processes for this tool. Devries [46] and Cortina [42] represent good examples of processes that take into account cultural appropriateness and linguistic equivalence in the translations of the SDQ. The use of trained mental health professionals who are native speakers, independent forward and blind back-translations, collaboration with teachers and staff to assess the tool, as well as pilot testing of tools in small and then bigger samples, are all steps that can be taken to assess the SDQ in a setting before use. The translation process for studies that made use of the SDQ-S commonly reported the use of a translation and back-translation process (n = 15; 27.8%), while the studies using the SDQ-P (n = 7; 13.0%) and SDQ-T (n = 4; 7.4%) more frequently did not report on the translation process (see Table 2).

Purpose of SDQ use in Africa

Use of the SDQ in Africa fell into two broad categories, (i) assessing internalising and externalising problems among children and adolescents in Africa, and (ii) assessing mental health in the context of HIV/AIDS.

Internalising and externalising problems among children and adolescents in Africa

The SDQ was used to examine internalising and externalising problems of children and adolescents in Africa in 21 of 54 studies. Some of the internalising disorders included emotional problems [27, 32, 36, 51, 59, 62], anxiety disorders [40], depression [43, 60], psychological functioning and mental health in homeless youth [53], orphans and vulnerable children exposed to maltreatment and stigma [44, 47, 55, 57], war-abducted adolescents [30], psychological well-being as it relates to parental migration [56, 70] and lack of parenting support [30]. Externalising problems examined included hyperactivity/impulsivity [27], behavioural problems [27, 32, 36, 51, 59, 62], effects of corporal punishment on mental health and educational outcomes [46, 75], antisocial behaviour and substance use [52], and caregiver association with substance use and mental health [58].

Mental health difficulties in the presence of HIV/AIDS

The majority of articles (24 of 54) identified in this review used the SDQ to explore child and adolescent mental health in the context of HIV/AIDS (17 studies from South Africa, three from Ghana, two from Zambia, Kenya and Malawi, and one from Tanzania) and compared mental health and HIV associations between orphaned and non-orphaned children in South Africa [26, 28, 31, 54, 76, 77]. Peer problems were among the common mental health difficulties for children orphaned due to AIDS [26, 28, 31, 77], followed by posttraumatic stress disorder [26, 28] and conduct problems [28, 31]. Studies included examination of emotional and behavioural difficulties in HIV positive adolescents [29, 61, 64], the impact of parental HIV/AIDS status and death on the mental health of the child [34], psychosocial adjustment of children affected by HIV/AIDS [37], evaluation of community art therapy intervention on the mental health of children affected by HIV [38], randomised controlled trials pilot evaluation [45], caregiver social support [33], positive parenting [48], and in the provision and outcomes of community-based organizations for children and adolescents [54, 67, 73, 74].

Discussion

Given the divide between need for and access to CAMH services particularly in low- and middle-income settings such as in Africa, screening tools such as the SDQ that are simple, accessible and freely-available has the real potential to improve early identification and access to care in Africa and other LMIC settings. However, for clinically-meaningful implementation, it is essential that any instrument not designed in the context should be examined to ensure it has good psychometric properties (e.g. is reliable and valid in the new context), is culturally appropriate, and is used in adherence to the guidelines of the developers of the instrument. We therefore set out to explore the current knowledge-base about the use of the SDQ in Africa.

We identified 54 peer-reviewed publications from 12 African countries, most from South Africa. The SDQ was typically used to investigate internalisation/externalization disorders in different clinical populations, including vulnerable populations such as orphans, children in war-torn areas, and migrants. Interestingly, the SDQ was most frequently used in the evaluation of children and adolescents affected by HIV/AIDS. Many different languages were used, but authorized SDQs in those languages were not always available on the official SDQinfo website. Authors frequently commented on challenges in the translation and back-translation of mental health terminology in African languages. Sixteen studies (29.6%) administered the SDQ to participants outside the intended age range, only 4 (7.4%) used triangulation of all versions to generate assessments, and 8 studies (14.8%) used only subscales of the SDQ. Where ‘caseness’ was defined in studies, UK cut-off scores were used in all but one of the studies. Only one study conducted a thorough psychometric validation of the SDQ, including examination of internal consistency, generating cut-off scores, and factor analysis [25].

The African continent is highly multi-cultural and multilingual. Screening tools such as the SDQ initially validated for a UK population, have been reported to have good psychometric properties in high-income settings [81], but, as shown in this scoping review, these findings have not been replicated in Africa. The results presented in the current review suggest that the SDQ has been used in several African countries among various groups of children and adolescents, without comprehensive validation for use in these settings. The many studies that have used the SDQ in Africa reporting limited or no validation therefore raises concerns about the robustness of findings reported in the African CAMH literature to date.

We were surprised to observe the application of the SDQ outside the scope and guidelines of the instrument. We strongly believe that all researchers who work in Africa have an ethical duty to ensure adherence to instrument guidelines and to work in collaboration with tool developers. Given the limited research resources available in Africa, we have to ensure high-quality research at all times.

An additional potential area of concern was the limited use of triangulation of measures in only four studies all conducted in South Africa. Whilst the philosophical position of the SDQ developers is to encourage triangulation of data in clinical practice, many different research questions could be answered without triangulation of data, for instance, investigation of the psychometric properties of a specific version of the SDQ [82]. We also acknowledge that the low rate of triangulated use of the SDQ may have been purely pragmatic, for instance, for the assessment of homeless youth, a self-report version may have been the only option for assessment. For this reason, we are cautious not to overinterpret the limited use of triangulation reported here. Further examination of triangulated data would, however, be of interest to examine measurement invariance between parent, teacher and self-report versions, and cross-culturally. To date, measurement invariance for the different versions of the SDQ have yielded conflicting, and ambiguous findings [83, 84].

Context-specific validation of the SDQ is crucial, as demonstrated in a recent South African study that evaluated the psychometric properties of the SDQ in 3451 adolescents aged 12–16 [82]. The results showed reasonable, yet variable, internal consistency, but identified significant gender-based differences in scores. More importantly, it showed a very unusual profile of ‘caseness’. Using UK cut-off values (designed to identify the top 10% of scorers), 26% of 12–16 year olds were found to be at high risk of emotional problems and 33.7% to be at high risk for peer problems (de Vries et al. 2017 [82], Table 3). Based on these observations, the authors raised the need for extreme caution in making cross-country comparisons. For intra-county clinical use, the study recommended the use of local cut-off scores to define clinical ‘caseness’ in any screening procedure. For example, a suggestion had been made to use the SDQ-S for mass screening as part of the South African Integrated School Health programme (personal communication, MSK). UK cut-off scores for caseness would have led to a two- to three-fold overidentification of young people ‘at risk’, which could have caused potential distress among young people, and could have placed the already very limited mental health services under overwhelming strain.

The scoping review and our recent findings [82] also raised the question about the cultural appropriateness and challenges with translation and back-translation of the SDQ to ensure valid use [85]. Africa has 2000–3000 languages, and only a handful of these have been included in SDQ translations. We suggest that it will be important to understand the local perceptions of mental health and the linguistic subtleties in the description of symptomatology, in order to evaluate how best to translate and validate instruments such as the SDQ [82, 86].

Our findings highlighted a few additional important issues. Firstly, there has been a steadily growing body of research on CAMH in Africa, suggesting a heightened awareness amongst practitioners and researchers of the need to identify and document mental health conditions. In this review we identified a significant number of articles in 2016, suggesting a recent growth spurt in research. The review further highlighted that mental health challenges for African children and adolescents often occur within contexts of significant large-scale trauma, such as in situations of conflict, conflict-linked migration and the persistent and long-standing HIV/AIDS epidemic with the accompanying personal and societal devastation. Close to 18 million children and adolescents on the African continent have been orphaned due to HIV/AIDS [87], it is therefore not surprising that 27 studies in the review examined children and adolescents’ mental health in relation to HIV/AIDS. Some studies were set among homeless children and adolescents [88] and those exposed to war [43]. This is not necessarily the case in other regions in which the SDQ had been applied. This raises the empirical question about the extent to which the SDQ, developed in a high-income setting, typically applied in a stable, non-crisis environment could, even if applied with great due diligence, be appropriate for the identification of mental health challenges in such challenging contexts.

Lessons learnt for future SDQ research in Africa

One of the key findings from our review is the importance of ensuring that use of screening tools such as the SDQ should be used in adherence with developer guidelines and authorization. Where instruments are translated without authorization, used outside the prescribed age-ranges or where only subsections or items are used, there is a real risk of incorrect and potentially misleading findings.

Secondly, given the fact that only one study in the review period had performed a comprehensive evaluation of the psychometric properties of the SDQ in a specific African country, there is a clear need to see similar studies in other African countries. The study by de Vries [82] illustrated the potential value of such evaluation. We support the recommendation that UK cut-offs should not be used to determine ‘caseness’ in African countries and that further validation work is required to compare normative cut-off values with gold standard diagnostic instruments to establish the true sensitivity and specificity of the SDQ [82].

de Vries and colleagues [82] also recommended that the SDQ could be useful as an ‘in-country’ instrument, but that it should be used with great caution as a ‘cross-country’ comparative measurement tool. There is clearly an increasing realisation of the need for cross-cultural measurement instruments in CAMH [83, 84, 89]. However, we acknowledge that, to date, most CAMH screening tools have not been developed with a global user in mind. In an Africa setting this task is significant and a range of challenges will need to be considered from the subtle cultural, ethnic, pragmatics of language [82, 89] to the large number of languages and variability of literacy levels. de Vries [82] recommended qualitative exploration of the cultural use of mental health language in order to develop mental health measurements that can capture similar global concepts appropriately and adequately in local settings.

In terms of policy-making, it is important to emphasize that measurement instruments should be selected for inclusion in policies and implemented not based on evidence and use in high-income countries, but only when local evidence has been generated to support the safe and meaningful use of these tools.

Limitations

We acknowledge a number of limitations in this study. Only four databases were searched to identify articles for this review. It is therefore theoretically possible that some studies were not identified. However, the four databases included typically covers the significant majority of all peer-reviewed journals. In particular, we did not have any language exclusion in order to ensure that we were able to identify potential publications from French- and other African language-speaking sources. In terms of translation/back-translation and authorization of SDQs, it is possible that there may have been approved SDQ versions not yet included on the SDQinfo website. For instance, the most recent Afrikaans and isiXhosa SDQ-S used in de Vries et al. [82] have been authorised by Goodman but are not yet uploaded on SDQinfo. It is possible that some of the ‘languages’ and dialects used in some of the studies may not have required specific translation/back-translation and authorization.

Conclusion

The SDQ is an easy to use, reliable screening tool for early identification of mental health disorders in children and adolescents, and has been used in numerous countries and languages. The comprehensive scoping review of the SDQ in Africa showed that it has been used in about a quarter of African countries, and that it may be a useful screening instrument to identify children and adolescents at risk of mental health problems. However, the limited and variable psychometric knowledge about the SDQ in Africa suggests that the tool should be used with caution to define ‘caseness’, and that research teams across the continent should perform careful psychometric evaluation of the SDQ in their countries and languages. We suggest that, apart from standard psychometric evaluation, the multicultural and multilingual nature of Africa also necessitates careful cultural evaluation of the instruments to ensure equivalence for clinical and research use. It throws down the gauntlet for practitioners, researchers and academics who work in this field, to pay meticulously attention to the rigour with which these instruments are applied.