Background
Mental health concerns are a leading cause of disability worldwide, as well as one of the largest sources of economic burden [
1]. This burden of mental disorders is especially high among youth in low- and middle-income countries (LMICs) [
2]. As such, developing and disseminating evidence-based interventions for youth mental health has emerged as an urgent global health priority over the last decade [
3]. Even though there has been increased research dedicated to developing and testing the efficacy of youth mental health interventions in LMICs, very few of these attempts have included economic evaluations of these interventions. Indeed, such economic evaluations of evidence-based interventions are extremely rare globally. One recent systematic review of economic evaluations revealed only one evaluation of an intervention for youth depression, and zero for anxiety [
4].
How then do decision makers choose which interventions to prioritize in LMICs where chronic government under-investment in mental healthcare persists [
5,
6]? Economic evaluations of youth mental interventions, in addition to efficacy examinations, are needed to inform important policy, practice, and research decisions by policymakers, practitioners, researchers, and other mental health professionals interested in investing limited resources wisely. It seems that efforts to expand help-seeking options for youth mental health in LMICs are handicapped insofar as research efforts emphasize efficacy evaluation without robust economic evaluations.
One approach to reducing the youth mental health treatment gap in LMICs involves the development of brief and scalable youth mental health interventions. The work on brief and scalable interventions is inspired by three very promising ideas: First, simple intervention strategies that focus on improving broader human functioning, rather the reduction of psychopathology, and cultivating individual character strengths can be effective for many youth mental health problems [
7‐
9]. These interventions, which are sometimes called “wise” interventions or “character strength” interventions [
10], tend to focus on single and simple human attributes, like “growth mindset”, and show promise as effective interventions for youth mental health problems [
11]. Second, task-shifting to lay providers with minimal to no formal mental health training can be an effective avenue for expanding the presently limited mental health caregiving workforce. Lay providers have been shown to be capable of effectively delivering a wide array of mental health interventions across diverse settings in LMICs [
12,
13]. Task-shifting has indeed emerged as a WHO-recommended approach to mental health caregiving in LMICs [
14]. Finally, a community-orientation to help–seeking can be effective for tackling societal stigma, getting buy–in, and mobilizing existing resources and infrastructure for mental healthcare [
15,
16]. Indeed, community–delivered interventions are often found to be feasible, acceptable, and effective for youth mental health problems in LMICs [
17].
The
Shamiri (Kiswahili for “thrive”) intervention, a brief group-based intervention for adolescent depression and anxiety symptoms, was developed using these three ideas [
18]. First, the intervention consists of three simple intervention strategies drawn from the “wise” intervention literature: (1) growth mindset, which teaches youth that their personal attributes and characteristics are malleable and can change through effort, (2) gratitude, which encourages them to notice and appreciate things in their lives, and (3) value affirmations, which teaches them to take value-aligned actions [
18,
19]. Second, the intervention is delivered by lay providers aged 18 to 22 after only 10 hours of training [
18‐
20]. Third, the intervention is group-based and is delivered in schools, i.e., a community-based setting, as an afterschool program. It is delivered in only four one hour sessions across four weeks [
21]. The efficacy of the Shamiri intervention has been tested in at least two randomized controlled trials (RCTs) including in a recent RCT which found that youth assigned to the
Shamiri condition experienced greater reductions in symptoms of depression and anxiety than those assigned to an active study-skills control condition [
18,
21].
The RCTs revealed that
Shamiri’s efficacy was similar to those of traditional evidence-based youth mental health interventions, but
Shamiri differed in its
low-intensity low-touch intervention approach. Most evidence-based interventions for youth mental health are resource intensive: they are delivered by expert caregivers (with a master’s or doctoral-level background), use a one-on-one format, are implemented in office settings, and last 12–16 weeks [
22]. It seems that
Shamiri’s low-intensity approach (use of lay providers, community–based delivery, and brevity) may yield significant cost-savings compared to traditional delivery systems. An economic evaluation of the
Shamiri intervention can allow us to test this premise. Thus, we conducted an economic evaluation to investigate whether
Shamiri is a cost-effective way to promote mental health for youths in Kenya. Assessing the cost-effectiveness of
Shamiri can facilitate its comparison with other mental health interventions that have been subject to economic evaluation. Comparisons of that kind could guide such policy decisions as which mental health programs to invest in, in LMICs [
23]. In line with best practices [
4,
24‐
26], costs were estimated from multiple perspectives, and sensitivity analyses were performed to address sources of uncertainty regarding clinically meaningful improvements.
Discussion
We aimed to identify the cost, effectiveness, and cost-effectiveness of
Shamiri, a school-based intervention program delivered by lay providers in Kenya. At treatment termination, a greater proportion of students in the control condition reported reductions in anxiety according to a standard definition of change. Both definitions of clinically meaningful change have discriminated between treatment responders and non-responders for depression and anxiety symptoms, although some consider the RCSC a more valid indicator of clinical improvement [
37,
39]. However, a greater percentage of
Shamiri students, relative to students in a control group, experienced clinically significant reductions in depression and anxiety symptoms at a 7-month follow-up. Results of a sensitivity analysis suggest that it cost between $48.28 and $172.72 to help one student in
Shamiri, relative to the control, achieve reliable and clinically significant change in depression and anxiety. Our study should be interpreted alongside its limitations. For example, we estimated costs retroactively, which is susceptible to error. To minimize error, we worked closely with study administrators to identify intervention resources and used gold-standard value-estimation methodology. Although retroactive resource identification and cost estimation is a common practice, prospective estimations are more reliable. Additionally, lay counselors were present during administration of baseline and follow-up assessments, which could introduce presentation bias.
Our economic evaluation is a rare contribution to the youth global mental health intervention literature and showcases how low-intensity interventions can have meaningful mental health impacts. With budgets for mental health decreasing and the need for prevention and treatment expected to increase, economic evaluations of mental health interventions are needed to guide mental health investment decisions. Sources of cost-savings included delivering the intervention in a group format at schools and training lay providers to deliver the intervention. Our results suggest that lay providers, after screening and training, can effectively deliver a mental health intervention. Of note,
Shamiri condition was not superior to the control condition for the RCSC criteria until the 7-month follow-up. When using the RCSC criteria at 7-month follow-up, 43% (vs. 29%) of
Shamiri students met such criteria for depression symptom reduction and 54% (vs. 42%) met the criteria for anxiety symptom reduction. This “sleeper effect” serves as a useful reminder that clinical benefits are not always apparent immediately after treatment and reiterates the importance of using longitudinal research designs in clinical studies. The effectiveness of
Shamiri lay providers is promising for the future of global mental health, especially in light of the shortage of mental health providers with post-baccalaureate degrees in LMICs [
44].
Shamiri’s group and school-based delivery system also makes it more scalable than the traditional one-on-one psychotherapy delivery system; we believe that
Shamiri has the potential to increase access to mental health treatment, particularly in resource-restricted LMICs.
Purchasing power and cost of living differ between the U.S. and Kenya and contribute to differences in hourly wages and costs of goods/services. Relative to the U.S., it is between 52 to 71% less expensive to live in Kenya (Expatisan.com; Livingcost.org). This cost-of-living discrepancy between the U.S and Kenya is reflected in the different salaries for psychologists based in the U.S. versus Kenya: The U.S. psychologist was estimated to make $50/hour, whereas the psychologist in Kenya was estimated to make $3.67/hour. Additionally, the U.S. dollar goes much farther in Kenya than it does in the U.S. One could purchase $43.80 worth of goods/services in Kenya for every $1 spent in the U.S. for the same goods/services [
45]. Put another way, $1 could be exchanged for 43.8 × more goods and services in Kenya compared to the U.S. Making cross-country comparisons in cost-effectiveness can be problematic if differences in economic conditions are not acknowledged [
46]. To maximize generalizability,
Shamiri should be compared to mental health interventions delivered in countries with similar purchasing power and cost of living. Economic evaluations of mental health interventions in LMICs are needed so that more accurate conclusions can be generated about cost-effectiveness.
Also of note, costs associated with transportation time alone for supervisors and lay providers accounted for nearly half (44.7%) of total costs in our base case scenario. Due to the rural geography of Kenya, supervisors and providers spent more time traveling to and from the school sites than they did delivering the intervention. When adding the out-of-pocket transit fees for both supervisors and lay providers, that proportion increased to 53.7% of base case delivery costs. Although total delivery costs for
Shamiri were quite low, future researchers and providers would be wise to acknowledge that a significant amount of resources for in-person interventions based in LMICs with a similar geographic makeup may go to transporting providers and support personnel to intervention sites. Of note, mobile phone supervision has shown promise as a feasible, cost-saving alternative to in-person supervision for lay providers in Kenya [
47].
Another strength of our study was the 7-month time horizon. To put this time horizon in perspective, 16 of 42 (38%) of economic evaluations of anxiety treatment had a time horizon of less than 7 months [
48]. It is important to look at outcomes beyond treatment termination because clinical improvements are not always immediately apparent or may continue to improve over time. The relationship between costs and effectiveness over time can be difficult to quantify; however, it is essential to make this quantification interpretable to clinicians and administrators to guide implementation and dissemination decisions. Our effectiveness-cost ratios, a dimensional categorization, mirrored results from the parent RCT, which suggested that adolescents in
Shamiri achieved greater reductions in anxiety and depression symptoms over time compared to the control [
21]. In other words,
Shamiri students experienced greater reductions in GAD-7 and PHQ-8 symptoms per dollar spent.
Categorical approaches, such as estimated incidence rates, can also illustrate the relationships between intervention costs and clinical outcomes. Although these approaches de-emphasize individual differences and, as a result, are less precise, some clinicians may find it useful to dichotomize treatment outcomes as a “success” or “failure.” We used multiple definitions of clinically significant change, which favored different conditions at different time points. For example, at treatment termination, students in the control condition fared better than students in
Shamiri for the standard definition of change for anxiety. Experts note that pressure to academically succeed may create anxiety for adolescent Kenyans [
49]. This observation was consistent with our baseline findings for the sample, which showed that youths, on average, reported anxiety in the moderate range. It could also explain why a greater proportion of students in the study skills active control condition experienced certain reductions in anxiety at termination: Learning study skills may have targeted and helped assuage their main source of anxiety.
We found that
Shamiri was superior to control for the RCSC definition of change for depression and anxiety at 7-month follow-up. Of note, the RCSC is a more valid indicator of treatment effectiveness than the standard definition of change [
37]. Because costs between the two conditions did not differ, differences in cost per NNT are driven entirely by differences in effectiveness. If two treatments are equally effective, the treatment with the lower cost per NNT could be considered more cost-effective if the cost is within a decision-maker’s willingness-to-pay threshold. A recent systematic review of nine youth mental health economic evaluations included two studies that delivered school-based interventions [
4]:
Not on Tobacco [
50] targeted smoking and
Planet Health [
43] targeted disordered eating.
Not on Tobacco [
50] cost $33.63 per “treated” (i.e., quit smoking) student in 2000 USD ($52.11 per student, in 2021 USD), and 10 students were estimated to quit per 100 students treated.
Planet Health [
43] cost $46,803 in 2010 USD ($56,500 in 2021 USD) to prevent one case of bulimia. However, all of these studies were conducted in the U.S. Compared to other youth interventions, of our statistically significant findings, our highest estimated cost per “treated” student (i.e., using the RCSC criteria) is acceptable at a cost of $172.72 per treated student in the U.S., but may be above a LMIC’s willingness-to-pay threshold due to differences in purchasing power. A recent economic evaluation of the
Shamiri digital intervention in Kenya by Wasil and colleagues [
34] estimated that it cost $25-$35 per student achieving clinically meaningful change, thus reiterating the cost-saving potential of interventions delivered in a single session and without an in-person provider. Regardless of treatment status, it cost between $6.80 and $20.81 to deliver
Shamiri to each student. This is similar to the $22.40 (2020 USD) cost per participant to provide usual care plus adjunctive depression services (psychoeducation, antidepressant pharmacotherapy, and home-based follow-up) in a primary care center in Nepal [
51], which has a similar purchasing power (34.3) to Kenya (43.8) [
45]. Of note, in the context of different cost methodologies and outcome metrics across studies, direct comparisons of cost-effectiveness are difficult. Given the dearth of cost data on youth mental health interventions in real-world settings, it is challenging for cost-effectiveness data to guide real-world decisions. We encourage applied clinical researchers to consider reporting delivery costs in their research. Yates’ (2020) [
26] instructional guide for cost-inclusive research is a useful resource.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.