From a starting point of 299 ideas, we selected the top 33 for further research through an idea sort. This Idea Prioritization Report is the second stage of our research process, and ranks the 33 ideas from most to least promising. Now, the top 7 ideas will move to the next stage: 80-hour reports, which will be released in the coming weeks.
To rank the 33 ideas, we spent two hours evaluating each idea using cost-effective analysis (CEA). We selected this methodology for two reasons. Firstly, we expect that different interventions within the space of mental health and subjective well-being will have very wide ranges of cost-effectiveness. Knowing these ranges will be helpful as we continue assessing the interventions. Secondly, we believe that CEA can offer information that is difficult to obtain from other sources. Few expert models focus on the cost-effectiveness of mental health/happiness initiatives. Those that do, consider it only in terms of willingness to pay for quality-adjusted life years (QALYs). As a result, we decided to use the satisfaction with life scale (SWLS) rather than QALYs in our CEA. The table below shows the ranking of all 33 ideas. The two-hour CEA report for each idea is linked in the score column. Some ideas appear more than once, where we evaluated the cost-effectiveness of more than one method of delivery (e.g. promoting gratitude journals through books vs an app). Ideas in bold are our top priorities for research and will be assessed in more depth this year. Below the table, we describe each idea and provide summaries of the main factors that contributed to its place in the ranking.
Problem | Idea | Area/Sub Areas | CEA Percentile |
Anemia and live birth deaths | Iron fortification | Fortification (another Fortify Health) | |
General mental illness | Guided self-help | Guided workbook or online | |
Mundane, sub-optimal happiness | Policy | ||
Perinatal depression | Standard or peer delivery | ||
Depression | Distribute antidepressants such as SSRIs | Mostly examined paroxetine but other SSRIs may be viable | |
Suicide | Training gatekeepers | General population or target universities | 0.10 |
Pain | Policy | ||
Mundane, sub-optimal happiness | Gratitude journals | App with conditional cash transfer | |
Mundane, sub-optimal happiness | Gratitude journals | Distribute books with conditional cash transfer | |
Post-traumatic stress disorder | Trauma clinics | Narrative exposure therapy | |
Depression and anxiety | Standard, or utilize the model for other group | ||
General mental illness | Increase mental health funding | Policy | |
General mental illness | Capacity building | Training psychologists | |
Depression | Standard | ||
Mundane, sub-optimal happiness | Recommendations for companies based on subjective well-being research | Standard | |
Depression, anxiety, and post-traumatic stress disorder | Standard | ||
General mental illness | Acceptance and commitment therapy (ACT) | Standard | 0.39 |
General mental illness | The Biaber Project | Standard | |
Depression | Group interpersonal psychotherapy (IPT-G) | Standard | |
Suicide | Google ads redirect method | Standard | |
Anemia and live birth deaths | Iron supplementation | Supplementation for pregnant women | |
Depression and anxiety | Computerized cognitive behavioral therapy (CBT) | Develop an app | |
Mundane, sub-optimal happiness | Positive education | Classes | 0.54 |
General mental illness | Improve school-based mental health services | SHINE | 0.56 |
Depression and anxiety | Group cognitive behavioral therapy | Standard | |
Anxiety | Australian FRIENDS program | Standard | 0.62 |
Bipolar disorder | Psychoeducation | Standard | |
Depression and anxiety | Another Friendship Bench | Standard | 0.67 |
Mundane, sub-optimal happiness | Gratitude journals | Distribute books | |
Depression and anxiety | Mindfulness-based cognitive therapy (MBCT) | Standard | 0.72 |
Depression and anxiety | Peer support groups | Standard | |
Insomnia | Cognitive behavioral therapy for insomnia (CBT-I) | Standard | |
Mundane, sub-optimal happiness | Wellness app | Develop an app | |
Alcoholism | Alcoholism self-help groups | Standard | 0.82 |
Mundane, sub-optimal happines | Universal basic income (UBI) | Standard | 0.85 |
Depression and anxiety | Computerized cognitive behavioral therapy (CBT) | Pay per use | |
Malaria | Trial malaria nets (Another AMF) | Standard | |
Bipolar disorder | Lithium treatment | Standard | |
Mundane, sub-optimal happiness | Wellness app | Pay per use | |
Alcoholism | Regulate the availability and demand for alcohol | Policy |
IRON FORTIFICATION (ANOTHER FORTIFY HEALTH)
We examined iron fortification to combat mineral deficiency as provided by Fortify Health. This intervention performed very well on the raw CEA due to the incredibly low costs of fortification and the high rate of anemia in India. We were unable to find a SWLS estimate for the effect of anemia, but even with a conservative estimate, Fortify Health ranks among our most cost-effective interventions. The only possible issue seems to be whether it can be achieved at a scale large enough to offset operating costs. This intervention is nonetheless in the 90th percentile because of the counterfactuals: it is unlikely to add anything beyond Fortify Health’s current work. The first 80-hour report will examine Fortify Health as if it did not exist to establish a benchmark for other charities that aim to improve subjective well-being.
GUIDED SELF-HELP
Guided self-help can be delivered using physical or digital workbooks and remote assistance. It can target common mental illnesses using a variety of therapy techniques. Our CEA predominantly drew on studies examining acceptance and commitment therapy as these were most common within the 2-hour search time. Workbooks performed well: they are cheap to distribute, and the time cost is incurred mostly by the patient. The effect size of treatment was also larger than initially expected, although this may reduce outside of clinical studies where the adherence rate will likely be lower. The cost of providing assistance is also uncertain and could increase upon further examination.
POSITIVE EDUCATION: POLICY
Positive education involves teaching people techniques to cope with everyday stressors and increase their appreciation of life. Campaigning for policy change seemed like the most promising sub-approach for this intervention. Policy interventions tend to perform well in CEA, as they can leverage a larger amount of funding and affect a large population. However, the overall probability of success of any individual bill is hard to determine. The prior rate for private members’ bills is quite high, and obtaining legislation on the agenda seems more likely, thanks to the All-Party Parliamentary Group (APPG) on Mental Health. However, this estimate is likely to change during the final report. The counterfactual cost of government funding was also not modeled at this stage due to time constraints - this could reduce the intervention’s cost-effectiveness. Positive education will be examined after guided self-help, as the Happier Lives Institute is currently researching this intervention. We will delay our own research until the completion of their report.
THINKING HEALTHY PROGRAMME
This intervention would task shift talking therapy based on cognitive and behavioral techniques to community health workers for perinatal depression. It performed very well in our analysis due to the low reported costs on the Mental Health Innovation Network and the increased recovery rate from the intervention (about 30%). If this is accurate, Thinking Healthy is a very promising program but further investigation is likely to raise the estimated cost. Another potential issue is that Thinking Healthy has seen some interest from the World Health Organization so may already be scaled up sufficiently.
DISTRIBUTE ANTIDEPRESSANTS SUCH AS SSRIS
Selective serotonin reuptake inhibitors (SSRIs) are a class of drugs used to treat major depressive disorder. This intervention would distribute SSRIs to people diagnosed with depression. SSRIs are widely used and well-evidenced, and can be surprisingly cheap in lower-middle-income countries - something that we could not account for in the short CEA. Possible weaknesses of this intervention are side effects, which were not modeled, and the effects of prolonged usage. Logistical costs may be higher than estimated, which could also alter the cost-effectiveness of this intervention.
TRAINING GATEKEEPERS
Training gatekeepers (e.g. community workers, police, teachers) could enable early identification of priority disorders, provision of low-intensity psychosocial support, and improvements in referral pathways. The main outcome for this intervention is the reduction in suicide rate. Impact was much higher than we had anticipated - studies found in the 2-hour time cap suggest that such programs can significantly reduce suicide rates. For the CEA, it was necessary to convert deaths avoided to an estimate of the SWLS value of a life. This involves substantial uncertainty about the true neutral point on the SWLS. For example, Russian students have a mean SWLS of about 17.7. Going by the SWLS neutral point of 20 would lead to the counterintuitive conclusion that preventing deaths is morally bad.
ACCESS TO PAIN RELIEF
Lobbying for increased access to or deregulation of opioids can provide pain relief to those experiencing severe health-related suffering and chronic pain. Both of these conditions have large detrimental effects to life satisfaction and have simple solutions which are not being implemented. As a policy intervention, a charity campaigning for the wider use of painkillers could leverage a much larger budget by campaigning against fear of opioids. As mentioned for positive education policy, counterfactual sources of government funding were not considered, which would reduce the cost-effectiveness of this intervention. A large amount of uncertainty still remains around probability of success and what percentage of a country/district’s burden of pain would be relieved.
GRATITUDE JOURNALS: APP AND CONDITIONAL CASH TRANSFER
This intervention would develop a gratitude journaling and exercise app incentivized by a conditional cash transfer (CCT). This was the best performing gratitude journal intervention, as the estimated adherence rate was highest. Although the effect of gratitude journaling seems to be small compared to other interventions, it can be incentivized quite cheaply. That some gratitude apps already exist could diminish the counterfactual impact. The main concerns with this intervention are whether it could be easily gamed and whether it could be deployed in a lower- or middle-income country.
GRATITUDE JOURNALS: BOOKS AND CONDITIONAL CASH TRANSFER
The second-best performing gratitude intervention also uses a CCT to increase adherence. With a smaller number of participants, this gratitude intervention performs better than the app, as it avoids the large fixed cost of developing the app. However, it will probably be more difficult to confirm participation and delivery of the CCT.
TRAUMA CLINICS
Following the model created by Peter C. Alderman Trauma Clinics, a charity focused on this intervention would establish clinics in public-private partnerships with governments and train indigenous caregivers to deliver evidence-based therapy for post-traumatic stress disorder. This partnership shifts the cost onto local governments, significantly reducing the cost for the charity. As such, the cost per patient treated is quite low for the charity. The therapy used by Peter C. Alderman Trauma Clinics seemed less promising than narrative and exposure therapy, which was used in the model instead. The main concerns with this intervention are whether the model used by Alderman could be employed and whether counterfactual use of government funds is too high. If full costs were covered by the charity, the increased cost per patient would diminish its cost-effectiveness. It may be possible to establish clinics in refugee camps with a high prevalence of PTSD to somewhat counteract this increased cost.
WOMEN’S CIRCLES
From MHIN: “Co-designed, community-led Women Circles nurture the innate strength of women and communities through a series of group activities that promote well-being, reduce stress, strengthen relationships, and support women in becoming agents of change.” As an idea, women’s circles has the least robust model in our top 8. However, even when adjusted towards the mean estimate for interventions, it seems promising. Within the 2-hour time cap, results were only found with the Mental Health Continuum Short Form, which means that a lot of uncertainty remains around this model. This uncertainty made conversion into SWLS very difficult, and instead we went with a conservative educated guess. Similar models have effects resembling the estimate in the model, but the effect on satisfaction with life is still imprecise.
INCREASE MENTAL HEALTH FUNDING
This is our most straightforward policy intervention, aiming to pressure governments to provide more funding to treat mental illness. The cost-effectiveness of this intervention varies greatly depending on factors like the estimated probability of success, $ per SWLS for mental health spending vs counterfactuals, and the increase in funding if successful. In the UK, mental health budgets are set to increase by 16.4% by 2023. This may in part be attributed to the efforts of Rethink Mental Illness, so this figure was used. However, the charity’s influence here is unclear. Greater increases could potentially be achieved in countries with smaller budgets like India, which spends much less on mental health.
CAPACITY BUILDING
This intervention would provide training, scholarships, or other incentives to train psychologists in areas with very low capacity. If the model is correct, this is highly cost-effective and would be in the top 4. However, it has been discounted heavily due to the large amount of uncertainty in the model and surrounding counterfactuals. The raw model attributes to the intervention the full value of a career’s worth of work, but brain drain, counterfactuals, and career changes make this unlikely. The lifetime salary cost was also not accounted for, which (although not paid for by the charity) could displace funds from other valuable projects.
TASK SHIFTING TO SPIRITUAL LEADERS
A research project funded by the nonprofit Grand Challenges Canada task shifted culturally-adapted cognitive behavioral therapy (CBT) to spiritual leaders from Catholic, Protestant, and Vodou communities. Although data on cost was not available on the mental health innovation network (MHIN), the intervention becomes very cost-effective even with a large upfront cost if training is effective for several years. The issue with this intervention is that the number of spiritual leaders that can be trained per year may be quite limited. This depends on locating leaders (as well as how willing leaders are - reportedly quite high), and organizing training camps. The intervention would perform better if more leaders could be trained per year than estimated.
RECOMMENDATIONS FOR COMPANIES BASED ON SUBJECTIVE WELL-BEING RESEARCH
This would be a consultancy firm that aims to increase employee satisfaction and productivity through subjective well-being research. Although a viable model for a for-profit, this intervention was modeled as a nonprofit. If Charity Entrepreneurship subsequently examines for-profits, we may use an expected value model instead of a CEA. As a nonprofit, the bottleneck for cost-effectiveness is the number and size of companies a consultant could affect per year. We were unable to find good sources for this, so cost-effectiveness could be higher than estimated. A weighting toward the average of these interventions was used as there was still a lot of uncertainty surrounding this model.
MIND AND HEART
Psychosocial assistants would deliver cognitive processing therapy (CPT) for survivors of sexual violence. Therapy would involve 1 individual session and 11 group sessions with around 7 attendees. This model keeps costs down while also allowing individuals to gain an understanding of the therapy in their first session. The effect size is moderate, with a large proportion of depressive symptoms relieved. Three possible issues with this model are: the use of the depression effect to convert SWLS; that the price may have been overestimated, as rent was included; and any additional effects Mind and Heart has on symptoms of post-traumatic stress disorder.
ACCEPTANCE AND COMMITMENT THERAPY
Acceptance and commitment therapy (ACT) provided by a trained psychiatrist in a low- or middle-income country. This intervention was less promising due to the high cost of professional treatment time even in developing countries.
THE BIABER PROJECT
The Biaber Project is a program on the Mental Health Innovation Network (MHIN). It involves training specialist and non-specialist workers to detect and treat common mental disorders. No evaluation has taken place on the MHIN, so a lot of uncertainty remains concerning the number of health workers who can be trained per year in each ‘Hub’, as well as what effect on efficacy this has. Although it performed moderately well in the raw CEA, it seemed less promising when adjusted for priors.
GROUP INTERPERSONAL PSYCHOTHERAPY
Trained psychologists would deliver interpersonal psychotherapy (IPT-G). This intervention was less promising due to the high cost of professional treatment time even in a group setting.
GOOGLE ADS REDIRECT METHOD
Targeted online advertising would direct high suicide-risk search terms to suicide hotlines. The intervention itself seems quite cost-effective, especially if using Google’s nonprofit ad grants. However, overheads reduce the cost-effectiveness, as the intervention can only target a limited number of search terms. It could be run alongside another intervention (e.g. gatekeeper training) with ¼ of full-time staff working on it per week. However, the rate of search to click-to-call to suicides prevented is likely to change: further investigation may reveal it to be less promising.
IRON SUPPLEMENTATION
Iron supplementation can combat mineral deficiency in pregnant women, which can lead to live birth deaths. The main determining factors for this intervention were the cost of supplements and the low incidence rate for birth complications. Since complications are uncommon and it is necessary to blanket-target the population to prevent them, the hit rate (and thus cost-effectiveness) is reduced. However, a large percentage of the population are anemic and thus this effect on SWLS dominates. The two main concerns with the model are the lack of information on the SWLS effect of anemia (conservatively estimated to be 1 point on the SWLS) and the replacement rate for live birth deaths. The importance of the replacement rate depends on views on population ethics: various ethical positions draw different conclusions as to whether, if an individual is replaced, the death is as bad as if they were not.
COMPUTERIZED COGNITIVE BEHAVIORAL THERAPY: DEVELOP AN APP
Cognitive behavioral therapy (CBT) would be delivered through software, with or without telehealth assistance as needed. The main limiting factors for this intervention are the number of downloads and the download-to-active user ratio. Without a conditional cash transfer, active users per download are likely to be quite low. Even with 100,000 downloads per year, adherence rates may be low enough to make the intervention less cost-effective.
POSITIVE EDUCATION: CLASSES
Classes would equip people with techniques to cope with everyday stressors and increase their appreciation of life. This could be delivered in a high-income country (e.g. the nonprofit Action for Happiness) or in a developing nation. Our model uses an estimate for the cost of an hour of schooling to determine overall cost-effectiveness, so the intervention could be more competitive in a low-income setting.
IMPROVE SCHOOL-BASED MENTAL HEALTH SERVICES (SHINE)
This intervention would adapt and implement WHO’s School Mental Health Program through an online training platform and chat-bot for school teachers in Pakistan. Very little data was found on this intervention within the 2-hour time cap, so the CEA estimate is very weak. It may be reviewed again when the clinical trial for SHINE is completed in 2021. However, there is too little data at this time to make any good estimates.
GROUP COGNITIVE BEHAVIORAL THERAPY
Group cognitive behavioral therapy is delivered in a group setting to reduce the price per participant. The effect size does not seem to diminish significantly for this model of treatment: it is comparable to or slightly less than individual therapy. When modeling this intervention, we assumed that an initial individual introductory session is not required.
AUSTRALIAN FRIENDS PROGRAM
The Australian FRIENDS program is a classroom based structured cognitive behavioral therapy combined with a workbook. It teaches children a range of skills to identify and control their anxious feelings. This intervention would look more promising if the quoted staff student ratio of 1:7 could be increased and the effect of anxiety on SWLS is higher than estimated in this CEA.
PSYCHOEDUCATION
This therapeutic intervention for patients and their families would provide information and support to better understand and cope with bipolar disorder. The evidence found within the short search time suggests that psychoeducation has a very limited effect on SWLS for those with bipolar disorder. Even when also modeling the effect on caregivers, the overall effect is therefore quite limited. Although psychoeducation can be delivered in a relatively short time frame by lay health workers, the small effect size means that the cost is still too large.
(ANOTHER) FRIENDSHIP BENCH
Trained and supervised nonprofessional health workers stationed on a “Friendship Bench” deliver problem-solving therapy based on cognitive behavioral therapy (CBT) principles. In the raw CEA, the Friendship Bench performed very well and may be a good donation opportunity. However, when accounting for counterfactuals of copycat charities it seems unlikely to be an effective intervention. Instead, further support should be provided for the original organization to scale up.
GRATITUDE JOURNALS: DISTRIBUTING BOOKS
This intervention would distribute gratitude journals and exercises. This gratitude intervention performs poorly, as it does not increase adherence through a conditional cash transfer. If 100% adherence was achieved and only the marginal cost was accounted for, this becomes one of the most cost-effective interventions. However, at most 20% are expected to continue throughout the whole year, and there are likely a limited number of participants each year.
MINDFULNESS-BASED COGNITIVE THERAPY
Mindfulness-based cognitive therapy (MBCT) uses CBT in collaboration with mindfulness-based practice. This would be provided by a trained psychiatrist in a low- or middle-income country. This was less promising due to the high cost of professional treatment time even in developing countries.
PEER SUPPORT GROUPS
This intervention would facilitate peer support groups (predominantly for depression and anxiety) to enable people struggling with mental illness to provide each other with structured support. Although the effect on SWLS was moderate, the cost of running bi-weekly or weekly sessions is proportionally too high. Attendance across such a long time span is also likely to be low, reducing the intervention’s effect on the overall target group. If adherence rates were higher and the cost of rent could be avoided, this intervention would be moderately promising.
WELLNESS APP: DEVELOP AN APP
This would entail application-based positive education and experience sampling that uses machine learning monitoring to increase subjective well-being. This intervention faces a lot of the same issues as computerized CBT: a limited number of downloads combined with a low regular user rate makes the organization as a whole less cost-effective. The cost-effectiveness was also discounted because of counterfactuals: existing apps like SmartMood may be able to achieve sufficient coverage anyway.
COGNITIVE BEHAVIORAL THERAPY FOR INSOMNIA
Trained psychiatrists in low- or middle-income countries would provide CBT for insomnia. This intervention was less promising due to the high cost of professional treatment time even in developing countries. Additionally, the effect remains uncertain, as the effect of treatment was measured using the Insomnia Severity Index (ISI). Since a lot of uncertainty remains so the intervention was adjusted towards the mean for the ideas examined.
ALCOHOLISM SELF-HELP GROUPS
Structured peer support meetings such as Alcoholics Anonymous (AA) would aim to help alcoholics achieve sobriety. This intervention performed poorly as although the effect of increased sobriety on SWLS was moderate, costs were still too high. Conversion from sobriety to SLWS also necessitated a discount for increased uncertainty. This intervention could be cost-effective if the cost of rent could be avoided and the treatment effects were higher than expected.
UNIVERSAL BASIC INCOME (UBI)
This intervention would reduce economic insecurity by providing monthly unconditional cash transfers. The high cost of the cash transfer ($405) and the low effect on SWLS even across a household meant that this intervention performed badly. The effect of UBI vs unconditional cash transfer seems negligible, at least across the course of a year.
COMPUTERIZED COGNITIVE BEHAVIORAL THERAPY: PAY PER USE
Cognitive behavioral therapy (CBT) delivered through software, with or without telehealth assistance as needed. As with the app model, the main limiting factors for this intervention are the number of downloads and the download-to-active user ratio. Without a conditional cash transfer, active users per download are likely to be quite low. Even with 100,000 downloads per year, adherence rates will be low enough to make the intervention less cost-effective. This problem is compounded for this approach because the pay per use model increases costs for a large number of users.
MALARIA NETS (ANOTHER AGAINST MALARIA FOUNDATION)
This practice CEA trials how well the Against Malaria Foundation (AMF) would perform against our intervention list. The raw CEA performs well, as nets are very cheap. However, converting between deaths and SWLS was necessary and entails some philosophical difficulties. Even when discounting for this, AMF is still among our most promising interventions (although not when accounting for the counterfactuals of starting another AMF, which would be unlikely to add anything beyond AMF’s current work).
LITHIUM FOR BIPOLAR DISORDER
Lithium is a cheap medication able to help alleviate the effects of bipolar disorder. The medication has two effects: a reduction in the suicide rate for those with bipolar, and an increase in subjective well-being thanks to reduced symptoms. The latter is the least well-understood at this stage, and has been modeled with a weak method. This intervention did not perform well, as the effects seem quite moderate and a large amount of medication must be consumed per day.
WELLNESS APP: PAY PER USE
This would entail application-based positive education and experience sampling that uses machine learning monitoring to increase subjective well-being. As with the app model above, this intervention faces a lot of the same issues as computerized CBT: a limited number of downloads combined with a low regular user rate makes the organization as a whole less cost-effective. This problem is compounded for the pay per use approach, as it increases costs for a large number of users. The cost-effectiveness was further discounted for counterfactuals as existing apps like SmartMood may be able to achieve sufficient coverage anyway.
REGULATE THE AVAILABILITY AND DEMAND FOR ALCOHOL
This policy-focused charity would aim to increase regulations on alcohol to reduce consumption. This reduction could be achieved in several ways, such as restricting sales hours or location, setting a maximum proof, or - as modeled at this stage - increasing excise tax. The two main outcomes for this intervention are reduced morbidity and reduced rates of alcoholism. The reduction in morbidity is better evidenced than the effect on alcoholism, but this imbalance in confidence is less concerning as morbidity seems like the major issue here.
Comments