fbpx

Friendship Bench Comprehensive Summary

Last updated: November, 2023

Problem

Depression is strongly linked with lower subjective wellbeing (Clark et al., 2017), but mental health services are significantly underfunded in low- and middle-income countries (LMICs)1Mental and addictive disorders form between 7% and 13% of the global disease burden (Vigo et al., 2019) and their relative share has grown in recent years (Rehm & Shield, 2019), but only receive 1% of governmental health budgets in LMICs (Vigo et al., 2019) and 0.3% of health-directed international assistance (Liese et al., 2019). . As many as 94% of people with depression in LMICs do not receive adequate treatment (Thornicroft et al., 2017).

Intervention

Friendship Bench’s standard programme consists of 6 sessions of individual problem solving therapy (PST), followed by optional group support sessions with others who have finished Friendship Bench counselling. The sessions are facilitated by trained lay health workers, and aim to help individuals identify problems and build a positive attitude toward resolving them. For participants who received at least one session, the average attendance is 2 sessions.

Organisation

The Friendship Bench is an NGO that treats people for depression with problem-solving therapy (PST), primarily in Zimbabwe.

Evaluation

Methods

Because improving one person’s mental health can benefit others who are close to that person, we also estimated the spillover effects on household members. Thus, the total effect was the impact on recipients and household members over time.

Impact

We estimate that Friendship Bench benefits the individual recipient of their intervention by 0.91 WELLBYs. We also estimate, based on a household size of 3 non-recipients and household spillover rate of 16%, that Friendship Bench has a total household effect of 1.34 WELLBYs.

However, we think that effects reported in psychotherapy studies are inflated by 10% due to a statistical issue known as ‘range restriction’ 2‘Range restriction’ may occur in psychotherapy trials that select participants based on a threshold of mental health conditions, as this restricts the distribution of scores, which then reduces the variance in the groups, which then inflates effect size estimates., so we add a final discount of 10%, resulting in a final estimated total household effect of 1.21 WELLBYs per treatment.

Cost

We currently estimate Friendship Bench has a cost of $21 per person treated.

Cost-effectiveness

The total cost-effectiveness of Friendship Bench is $21 / 1.21 WELLBYs = $17 per WELLBY. This means for every $1,000 donated to Friendship Bench, the organisation provides 58 WELLBYs.

To quantify the statistical uncertainty in our results, we simulated Friendship Bench’s cost-effectiveness thousands of times by varying key parameters 3 These include: the initial effect, the duration, the annual decay rate, the household spillover, the discount (if any) for study quality, and the cost.
to see how the results might differ across a range of reasonable inputs (e.g., Monte Carlo simulations)4 Monte Carlo simulations allow us to treat inputs in a cost-effectiveness analysis (CEA)—often merely stated as point estimates—as distribution. Thereby, this allows us to communicate a range of probable values (i.e., uncertainty around the point estimates). See our cost-effectiveness methodology for more detail.. This is illustrated in Figure 1 below.

Figure 1. Density plot of the quantified uncertainty around Friendship Bench’s cost-effectiveness

Note. The diamond represents the central estimate of cost-effectiveness (i.e., the point estimate). The shaded area is a probability density distribution and the solid whiskers represent the 95% confidence interval.

Quality of evidence

We think the quality of evidence supporting the effect of the Friendship Bench is moderate. The primary reasons for this are5 This is the same rating, and these are the same reasons we give StrongMinds, as the evidence base is the same for both. :

  • The effect of psychotherapy is based on a large number of studies (77 RCTs) and participants (n = 28,491), and this evidence is of moderate quality.
  • However, the evidence specifically for Friendship Bench is relatively slim (3 RCTs).
  • There are only 5 RCTs for household spillover effects in LMICs (compared to the many more RCTs for the individual effects, see above). This increases our uncertainty about the overall effect.

Depth of analysis

Moderate. We have completed an in-depth analysis on the effect of psychotherapy in general, but we’ve only completed an initial evaluation of Friendship Bench’s programme.

Funding need

We haven’t evaluated this in depth yet.

Conclusion

Our cost-effectiveness estimates suggest that the Friendship Bench may be more cost-effective than StrongMinds. However, we have conducted a relatively shallow evaluation of Friendship Bench’s programme, and we consider ourselves in a relatively early stage of our process to thoroughly understand Friendship Bench’s programme, track record, and future projects. Thus, we think the Friendship Bench is a promising charity, but we refrain from giving it a top recommendation until we have completed further research.

FAQs

Our reports so far

We first identified that mental health was a key global priority in our Mental health cause area report. We then evaluated the impact of psychotherapy interventions in the Psychotherapy cost-effectiveness analysis. We first assessed the cost-effectiveness of Friendship Bench in our substantial update to our psychotherapy analysis in November 2023.

Endnotes

  • 1
    Mental and addictive disorders form between 7% and 13% of the global disease burden (Vigo et al., 2019) and their relative share has grown in recent years (Rehm & Shield, 2019), but only receive 1% of governmental health budgets in LMICs (Vigo et al., 2019) and 0.3% of health-directed international assistance (Liese et al., 2019). 
  • 2
    ‘Range restriction’ may occur in psychotherapy trials that select participants based on a threshold of mental health conditions, as this restricts the distribution of scores, which then reduces the variance in the groups, which then inflates effect size estimates.
  • 3
     These include: the initial effect, the duration, the annual decay rate, the household spillover, the discount (if any) for study quality, and the cost.
  • 4
     Monte Carlo simulations allow us to treat inputs in a cost-effectiveness analysis (CEA)—often merely stated as point estimates—as distribution. Thereby, this allows us to communicate a range of probable values (i.e., uncertainty around the point estimates). See our cost-effectiveness methodology for more detail.
  • 5
     This is the same rating, and these are the same reasons we give StrongMinds, as the evidence base is the same for both.