The case for using local expertise

First, working with a team of local researchers and stakeholders ensures the use of culturally appropriate research methods. Data collection methods common in HIC, such as paper-and-pen questionnaires or online surveys, are less suitable for use in LMIC such as Ethiopia, owing to low literacy rates and a strong oral tradition. Previous research led by Ethiopian colleagues (e.g. Shibre et al, Reference Shibre, Alem and Tekle-Haimanot2006) indicated that completion of a questionnaire through a face-to-face interview with a lay interviewer is a culturally appropriate alternative. The local researchers in our team guided the design of these study aspects, including appropriate training for the lay interviewers (Tilahun et al, Reference Tilahun, Hanlon and Araya2017a, Reference Tilahun, Fekadu and Tekolab).

Similarly, local expert input is needed to ensure that questionnaires fit local cultural norms and context. For example, we found that the anchoring of items of a caregiver involvement measure, originally developed in the USA, required cultural adaptation. Whereas enthusiasm tends to be displayed openly in the USA, equivalent levels of involvement will typically be shown in less overt terms in Ethiopia. Thus, the anchors providing examples of what typifies a ‘very enthusiastic’ or a ‘not enthusiastic’ caregiver required adjustment.

Local perspectives may also prove essential when interpreting research findings. Our survey in caregivers of children with developmental disorders indicated that the practice of restraining a child is relatively common (Tilahun et al, Reference Tilahun, Hanlon and Fekadu2016). Subsequent discussion of this finding during a stakeholder workshop taught us that restraint is not typically used as punishment, but a last-resort measure to protect the child from harm when there is no one to look after the child (Tekola et al, Reference Tekola, Baheretibeb and Roth2016). Thus, these results should not be interpreted in terms of abuse, but rather in the light of lack of available support. Consequently, efforts to reduce restraining of children with developmental disorders are more likely to be successful if they focus on improving support and care systems for families, rather than an education intervention solely conveying the message that restraining is harmful.

Local input is also essential when designing interventions. When we developed the mental health module of an education programme for Ethiopian community health workers (Tilahun et al, Reference Tilahun, Fekadu and Tekola2017b), local experts indicated that we should not emphasise the strong genetic influences on conditions such as schizophrenia and autism. An emphasis on high heritability could have had the unintended consequence of replacing existing stigmatising beliefs in the community with stigma associated with a ‘genetic taint’.

Perhaps most importantly, local collaboration and engagement are essential to ensure that the intervention addresses a locally perceived need, suits the country's health and education system, and is supported by local stakeholders. For an intervention to be sustainable and have scope for scale-up, the programme needs to be ‘owned’ by local partners. In an ongoing project, we are adapting and evaluating a parent skills training (PST) programme, developed by the World Health Organization, for use in Ethiopia. We set up a PST adaptation advisory board, consisting of local experts and stakeholders, who over the past 2 years have provided invaluable feedback on the draft content and delivery of the programme. Without local input, the training could have inadvertently set wrong expectations (e.g. that the intervention may result in a cure; Tilahun et al, Reference Tilahun, Hanlon and Fekadu2016), glossed over pertinent local issues or used culturally inappropriate intervention strategies.

Finally, working with local universities contributes to capacity building for research in the country. Although global research has burgeoned, recent analyses (Kelaher et al, Reference Kelaher, Ng and Knight2016; Nielsen et al, Reference Nielsen, Haun and Kärtner2017) suggest that the scientific articles reporting on these studies are primarily written by researchers from HIC. HIC researchers should invest in collaborating with local researchers and the training of local students, making sure they get the opportunity to write papers and submit to international peer-reviewed journals. These researchers and students are likely to take up prestigious posts and be involved in training future cohorts of local experts, thereby allowing for future expansion in research and mental health services.

How to achieve local collaboration and engagement

In many LMIC, in the absence of comprehensive governmental services, grassroots NGOs have been founded to address pertinent needs. The last author of this paper (Z.Y.) established Ethiopia's first autism school; following her pioneering work, additional schools for children with developmental disorders have opened (Tekola et al, Reference Tekola, Baheretibeb and Roth2016). A useful way to initiate a local network is through these local organisations, because they will have wide-ranging practical experience and knowledge of the challenges faced by local families. Connections with researchers can be built through local universities. Local researchers do not (yet) have to be autism experts; a public health or mental health researcher will still be well qualified to give input in locally appropriate research methods and study designs. It is this mix of partners that is important: local stakeholders are experts in the perceived needs and practical considerations, but are less well placed to comment on research techniques; in turn, researchers may miss subtle issues with research instruments if they are not also scrutinised by stakeholders with first-hand experience with autism.

The optimal method for local engagement will depend on the culture and context. Low literacy rates and poor internet access make written consultations less suitable for use in Ethiopia. In our experience, face-to-face workshop-style stakeholder consultations work best (Tekola et al, Reference Tekola, Baheretibeb and Roth2016). Once a network is established, it is important to sustain it by keeping stakeholders involved in each step of the research process, from the study design phase, through regular progress updates, to presentation of preliminary findings and dissemination of the final research report.

Global benefits of local collaboration and engagement

Although the preceding paragraphs illustrate our experiences in one particular low-income country, these are likely to have relevance more widely. High levels of stigma (Divan et al, Reference Divan, Vajaratkar and Desai2012) and lack of culturally appropriate screening and diagnostic tools (Durkin et al, Reference Durkin, Elsabbagh and Barbaro2015) and interventions are prominent issues in most LMIC. Research in LMIC might also inform strategies to improve access to services and research in hard-to-reach groups within HIC. Ethnic minorities are underrepresented in research in HIC (West et al, Reference West, Travers and Kemper2016); autism tends to be underdiagnosed and diagnosed late in children of ethnic minorities, and parents of these children report receiving lower-quality healthcare (Magaña et al, Reference Magaña, Parish and Son2015). Research findings from LMIC might thus provide clues to how to better reach these under-served groups in HIC.

In fact, the call for inclusion of local partners in research equally applies to research in hard-to-reach groups within HIC. Researchers and students from ethnic minorities are underrepresented at research-intensive universities. The inclusion of such groups in research teams is more likely to lead to innovative research designs that will promote successful recruitment and retention of diverse participants in research studies.