Policy Makers’ Tough Choices for Psychological Interventions in Global Mental Health

More governments around the world are recognizing the importance of mental health in their national health, social welfare, development, and education plans. At the 2019 second global interministerial meeting held in Amsterdam, the Netherlands, 35 countries and organizations signed a declaration to integrate mental health and psychosocial support services in crisis situations.¹ However, policy makers are left with the challenge of what psychological services to select and how to implement them. There is a growing range of psychological treatments, implementation modalities, and options to use nonspecialist paraprofessionals (eg, community health workers, teachers, peers). The study by Dorsey and colleagues² exemplifies the promise and challenge facing policy makers needing to select from among these options.

Dorsey and colleagues² tested trauma-focused cognitive behavioral therapy (TF-CBT) for youth delivered by nonspecialists in 2 sites (rural and urban) in 2 countries (Kenya and Tanzania). A year after the intervention, children receiving TF-CBT in rural and urban sites in Kenya showed sustained improvement compared with usual care. This was not the case in Tanzania where children in usual care and TF-CBT groups showed similar improvement; both usual care and TF-CBT in Tanzania were comparable with TF-CBT in Kenya.

As a policy maker, what should one draw from these findings? Policy makers would have an incomplete picture of TF-CBT had the investigators only conducted the study in 1 site. Building on prior TF-CBT studies in Zambia and the Democratic Republic of the Congo, the Kenya findings—without Tanzania’s results—could lead a policy maker in another African country to presume that the intervention is likely to be beneficial across the region. In contrast, only the Tanzania results would raise many questions about comparability with prior studies and whether or not to scale TF-CBT. Fortunately, multisite studies with the same inclusion criteria, evaluation tools, and study design create an opportunity to tease out reasons for possible differential effectiveness across context.

The findings by Dorsey et al² are similar to other recent psychological intervention studies with nonspecialists. The Thinking Healthy Programme for perinatal depression delivered by peer mothers in India showed significant improvement over treatment as usual (73% vs 60% remission),³ but this was not the case in Pakistan (49% vs 45% remission).⁴ Similarly, a classroom-based intervention for children in conflict settings was conducted in 5 countries, showing differences in effectiveness across sites.⁵

Multisite studies demonstrating differential outcomes can help policy makers decide if the intervention will work in their setting. However, for the transition from research to policy to be successful, key information is needed beyond effectiveness outcomes. We provide recommendations for researchers conducting multisite studies to foster the research-policy nexus.

Involvement of Policy Makers in Planning Multisite Studies

Prior to selecting the intervention and sites, researchers would increase the likelihood of future implementation and dissemination by including policy makers to understand local priorities and context. The context will influence choice of intervention, study design, and other elements influencing scale-up. Proving efficacy or effectiveness is not enough to convince a policy maker to take up results of research because the implementation approach needs to consider feasibility, affordability, cost-effectiveness, and promotion of human rights. If in the study, for example, there is a strong supervision model that cannot be ensured in real settings, it may dissuade policy makers. To engage policy makers, trust, effective collaboration, and incentives are needed. Trust is especially important when differences in outcomes across sites arise because lack of trust risks discounting outcomes from 1 or more sites.

Integrate Implementation Science Theory and Measures

Implementation science is crucial to inform how best an intervention fits in local systems.⁶ Implementation factors (adoption of the intervention, competency of paraprofessional counselors, fidelity to treatment guidelines, attendance of participants) can guide policy makers when deciding if and how to scale-up an intervention. In the trial by Dorsey et al,² there was significant turnover of the counselors in Tanzania, and there were differences in supervision approaches between the countries. Contextual factors such as working conditions, turnover, and burnout, particularly among paraprofessionals, can influence motivation to take up new interventions, thus affecting effectiveness and scalability. Greater or different incentives may be needed to sustain a workforce. If high turnover is inevitable, resources are needed for continuously training new counselors.

In addition, approaches are needed to assess counselors’ competency. Dorsey and colleagues² evaluated knowledge, adherence, and competency with the same approach across sites. There is now a push for standardized measures of competence across all psychological treatments delivered by nonspecialists by using observed structures clinical evaluations, such as the Enhancing Assessment of Common Therapeutic factors tool.⁷ The World Health Organization is developing the Ensuring Quality in Psychological Services platform to disseminate Enhancing Assessment of Common Therapeutic factors and similar tools. This will enable services to implement standardized competency assessments and improve target competencies to assure professionals and paraprofessionals have the skills needed for effective care.⁸

Comprehensively Document Usual Care and Incremental Cost-Effectiveness Across Sites

For policy makers, it is crucial to know if the new psychological intervention has value beyond what is currently available in their setting. Standard practices vary significantly across multisite studies. Therefore, control conditions should be as close as possible to current practice. For the Tanzania site, usual care produced comparable results with TF-CBT. When documenting usual care, it is important to consider religious-based and community-based services, such as traditional healing. Given high levels of other risk factors in Kenya, there may have been deterioration of local healing resources. Cost needs to be addressed in comparison with what is currently practiced. Attributes such as cost-effectiveness and reducing human rights violations can motivate changes in practice. It is helpful to document outcomes in other domains (eg, educational performance among youth, reduction in adolescent pregnancies, less drug use, greater future employment) to mobilize investment from other divisions or ministries.

Evaluate Moderators and Mediators Across Sites

For multisite studies to inform policy, a priori hypotheses are needed for potential differences. Dorsey and colleagues² suggest that higher levels of risk factors (trauma exposure, health problems, food scarcity) in the Kenya sites compared with Tanzania may have contributed to differences in outcomes. These factors are likely to vary between and within countries and thus can be included as moderators in statistical models. An individual patient data meta-analysis found that child psychosocial interventions had greater effectiveness with nondisplaced populations compared with displaced groups.⁹

Multisite studies also should include mediators to explore site differences. This can be done by measuring expected behavior change associated with treatment, such as using coping skills taught in the intervention. For the psychological intervention Problem Management Plus, a tool—the Reducing Tension Checklist—has been developed to measure beneficiary use of expected skills¹⁰; this can document differences across sites and presence of coping skills within usual care. Interventions for children also may be effective via changes in caregivers. Given the high burden of risk factors in Kenya, caregivers may have had higher levels of mental health problems and caregiving difficulties that improved through TF-CBT, whereas the Tanzanian caregivers may have had better baseline functioning and therefore better able to help their children recover from trauma in both the TF-CBT and usual care arms. Qualitative studies, forthcoming from Dorsey’s group, also can be helpful to generate new hypotheses for moderator and mediator effects.

The study by Dorsey and colleagues² is an important starting point for multisite psychological treatment studies to aid policy makers. A habit of using these approaches can build research capacity so that it becomes standard practice to collaborate with policy makers.