Centre for Pesticide Suicide Prevention — General Support

Published: December 2017

Note: This page summarizes the rationale behind a GiveWell Incubation Grant to the Centre for Pesticide Suicide Prevention (CPSP). CPSP staff reviewed this page prior to publication.

We have had conversations about updates on this grant in June 2020, May 2020, January 2020, October 2019, August 2019, June 2019, January 2019, September 2018, June 2018, February 2018, and November 2017.

The problem

The World Health Organization (WHO) estimates that 800,000 people die due to suicide each year.1 The most recent comprehensive systematic review we identified in our research suggests that ~14%-20% of global suicides between 2010 and 2014 were due to deliberate ingestion of pesticides, although we have not investigated the individual studies in this review.2 A WHO report claims that pesticide suicide is particularly prevalent in rural populations, where agricultural workers have easy access to pesticides.3

The intervention

CPSP plans to collect data in India and Nepal on which pesticides are most commonly used in suicide attempts, and are most likely to result in death. CPSP plans to use this data to help the governments of India and Nepal decide which pesticides to ban, with the intention of reducing suicide rates.

Evidence of effectiveness

The strongest evidence we have found that pesticide bans reduce suicides is from two observational time series analyses in Sri Lanka, described in more detail below. We do not think the evidence is conclusive (it is probable that other factors partially contributed to the decline in suicides), but believe it is more likely than not that the decline in suicide rates in Sri Lanka was primarily caused by the pesticide bans, as the banned pesticides were commonly used in self-harm and had a high case fatality relative to other methods of self-poisoning.

Evidence from Sri Lanka

We found two observational time series analyses in Sri Lanka, which find a substantial decline in suicide rates following pesticide bans aimed at reducing suicides.4

• Gunnell et al. 2007 finds a 50% decline in the suicide rate between 1995 and 2005.5 The decline followed bans on pesticides commonly used in suicide (methamidophos and monocrotophos in 1995, endosulfan in 1998)6 and appears to be driven by a decline in suicide by self-poisoning.7
• Knipe et al. 2017 finds a 21% decline in the suicide rate between 2011 and 2015.8 The decline followed phased bans of pesticides commonly used in suicide (paraquat, dimethoate, fenthion) between 2008 and 20119 and appears to be driven by a decline in suicide by self-poisoning.10

We note two major limitations of these studies, which limit causal inference:

• They use time series data, with no comparison group. It is possible that the decline in pesticide suicides was caused by events other than the pesticide ban.11
• They use national level suicide data of uncertain quality.12

However, we find the case that the decline in suicides was primarily caused by the pesticide bans reasonably compelling. Two analyses of Sri Lankan hospital records for poisoning patients indicate that (i) the banned pesticides were among the most commonly used in suicide and (ii) self-poisoning with the banned pesticides was more likely to result in death than self-poisoning with other pesticides.13

The organization

Track record

CPSP is a new organization, started by Michael Eddleston, a professor of Clinical Toxicology at the University of Edinburgh whose research focus is self-harm through poisoning.14 Professor Eddleston spent four years in Sri Lanka as a Wellcome Trust Fellow in the 2000s researching pesticide suicide. The Sri Lankan pesticide registrar at the time has said that Professor Eddleston's work was important for deciding which pesticides to ban.15 We discuss the evidence that pesticide bans in Sri Lanka caused a reduction in suicide rates above.

Activities

CPSP expects to conduct activities to meet the following milestones:16

• December 1, 2017:
  • Completed recruitment of in-country staff and established an office alongside the pesticide registrar in Nepal and India.
  • Identified 5-10 sentinel hospitals in Nepal and 20-30 sentinel hospitals in India from which to collect medical records, including (i) number of pesticide suicides (ii) which pesticides are most commonly used in suicide attempts (iii) case fatality from self-poisoning with different types of pesticide.
  • Signed Memorandum of Understanding for joint working with the Food and Agriculture Organization of the United Nations (FAO), which also works on pesticide regulation.
• April 1, 2018:
  • Completed training of data collection staff to collect medical records.
  • Begin collection of medical records from sentinel hospitals.
• September 1, 2018:
  • Completed collection of first six months of medical records and started analysis to determine which pesticides are commonly used in attempted suicide and have high case fatality.
  • Start discussing which pesticides should be banned with pesticide registrars and FAO.
• December 1, 2018:
  • Completed analysis of medical records and decided which pesticides to recommend for regulation.
  • Begin drafting legislation to ban selected pesticides.
• March 1, 2019:
  • Legislation to ban selected pesticides implemented.

After legislation passes, CPSP would train government staff to monitor the effectiveness of pesticide bans on reducing suicide rates. CPSP may also raise additional external funding to run a cluster randomized controlled trial in India on the effect of reducing availability of particular pesticides on suicide rates.17

Budget

CPSP plans to use this grant to cover the below program costs for two years:18

• Core international staff and administration: $459,317
• Program in India: $270,000
• Program in Nepal: $138,000
• International travel: $146,000
• Consultants: $145,600
• Training: $56,000

An additional $121,492 (equal to 10% of the above program costs) will cover costs to the University of Edinburgh, in which CPSP will be housed.

Cost-effectiveness

We constructed a cost-effectiveness model for CPSP’s activities.19 We roughly estimate that, in expectation, CPSP’s activities will be ~9x as cost-effective as cash transfers to people living in extreme poverty. This estimate depends on inputs for a number of parameters which are particularly subjective or difficult to estimate.

The inputs into this cost-effectiveness analysis which we are most uncertain about are:

• The probability that pesticide bans are implemented as a result of CPSP’s work in India and Nepal.
• Government costs of implementing pesticide bans.
• The discount we apply to the costs incurred by governments to implement a pesticide ban.
• The value of preventing a suicide through means restriction, relative to preventing a child's death from malaria.

Room for more funding

The grant is intended to fully cover operational costs for two years,20 and is the first significant funding CPSP has received.

Professor Eddleston told us that he has previously found it difficult to raise funds for CPSP,21 so we think it likely would not have attracted funding from elsewhere without this grant.

Risks of the grant

We see a number of potential risks to the success of this grant:

• Strength of evidence. The case for pesticide bans reducing suicides relies on a limited number of observational studies. It is possible that the reduction in suicide in Sri Lanka was, at least partially, caused by factors other than pesticide bans, and/or that pesticide bans in India and Nepal would not result in similarly large declines in suicides.
• Agricultural production. Banning commonly used pesticides may have a negative impact on agricultural yields, or increase input costs for farmers. A time-series analysis of crop yields in Sri Lanka found no noticeable reduction in net agricultural production in Sri Lanka following pesticide bans22 and we have not seen evidence that banning 79 different pesticides in Mozambique (for reasons unrelated to suicide)23 caused a decline in agricultural production. However, we believe it is unlikely that small decreases in net agricultural production would have been detected.
• Influencing policy. The success of CPSP's work will depend on the engagement of policymakers, which is difficult to predict. We have seen an email from the Ministry of Agriculture in Nepal asking for CPSP’s assistance,24 but remain uncertain whether CPSP will be able to translate its research into policy change.
• New organization. As CPSP is a new organization, we have not yet seen evidence that it will be able to effectively implement the activities described above.

Plan for follow up

We expect to have conversations with CPSP every few months to check in on progress, with a major check-in 12 months after the initial grant is received.

Internal forecasts

For this grant, we are recording the following forecasts:

Confidence	Prediction	By Time
Elie: 67%; James: 65%	We recommend a second grant to CPSP	July 1, 2019
Elie: 33%; James: 55%	Conditional on CPSP entering Nepal, Nepal will pass legislation to ban at least one of the three pesticides most commonly used in suicide.	July 1, 2020
Elie: 5%; James: 10%	Conditional on CPSP entering India, India will pass legislation to ban at least one of the three pesticides most commonly used in suicide.	July 1, 2020
Elie: 15%; James: 35%	Conditional on CPSP entering India, a state in India will pass legislation to ban at least one of the three pesticides most commonly used in suicide.	July 1, 2020
James: 85% for each	Conditional on legislation to ban pesticides used in suicide being passed in Nepal / India, the pesticide suicide rate is lower in the combined three years after the ban than it is in the three years before the ban (as determined by national level data or representative survey data)	3 years after legislation passed
James: 40% for each	Conditional on legislation to ban pesticides used in suicide being passed in Nepal / India, the pesticide suicide rate is >10% lower in the combined three years after the ban than it is in the three years before the ban (as determined by national level data or representative survey data)	3 years after legislation passed
James: 20% for each	Conditional on legislation to ban pesticides used in suicide being passed in Nepal / India, the pesticide suicide rate is >15% lower in the combined three years after the ban than it is in the three years before the ban (as determined by national level data or representative survey data)	3 years after legislation passed

Our process

Professor Eddleston contacted one of our researchers (James Snowden) about funding in 2016, when James was working at the Centre for Effective Altruism. We continued to investigate the grant when James started working for GiveWell in March 2017. We had two additional conversations with Professor Eddleston and Dr. Leah Utyasheva (CPSP’s Policy Director) before deciding to make this grant. We also spent significant time constructing a cost-effectiveness model, and reviewing the evidence that pesticide bans caused a decline in suicide rates in Sri Lanka.

Sources

Document	Source
Centre for Effective Altruism, interview with Gamini Manuweera, November 23, 2016	Source (archive)
Cha et al. 2016	Source (archive)
Chang et al. 2012	Source (archive)
CPSP Budget, June 2017	Source
CPSP Milestones Gantt chart, June 2017	Source
Dawson et al. 2010	Source (archive)
FAO undated press release	Source (archive)
GiveWell's non-verbatim summary of a conversation with Professor Michael Eddleston and Leah Utyasheva, May 18, 2017	Source
GiveWell's non-verbatim summary of a conversation with Professor Michael Eddleston and Leah Utyasheva, May 3, 2017	Source
GiveWell’s cost-effectiveness analysis of Centre for Pesticide Suicide Prevention	Source
Global Health Academy - Professor Michael Eddleston	Source
Gunnell et al. 2007	Source (archive)
Knipe et al. 2017	Source (archive)
Manuweera et al. 2008	Source (archive)
Mew et al. 2017	Source (archive)
Pearson et al. 2015	Source (archive)
Unpublished email from Professor Eddleston on June 4, 2017	Unpublished
van der Hoek & Konradsen 2006	Source (archive)
WHO 2014	Source (archive)
WHO suicide fact sheet	Source (archive)

• 1

  “Close to 800 000 people die due to suicide every year.” WHO suicide fact sheet

• 2

  "We identified data from 108 countries (102 from WHO data, 6 from the literature). A conservative estimate based on these data indicates that there were approximately 110,000 pesticide self-poisoning deaths each year from 2010 to 2014, comprising 13.7% of all global suicides. A sensitivity analysis accounting for under-reporting of suicides in India resulted in an increased estimate of 168,000 pesticide self-poisoning deaths annually, that is, 19.7% of global suicides." Mew et al. 2017, Pg 93.

• 3

  “One of the key methods of suicide in LMICs [low- and middle-income countries], particularly in countries with a high proportion of rural residents engaged in small-scale agriculture, is pesticide self-poisoning.” WHO 2014, Pg 24.

• 4
  • "We obtained data on the number of suicides in Sri Lanka from the following sources: (i) total suicides and suicide rates between 1940 and 1975 from two previous analyses of suicide in Sri Lanka both based on data from Sri Lanka’s Registrar General; (ii) age-, sex- and method-specific suicide data from 1975 to 2005 from the Department of Police, Division of Statistics, Sri Lanka." Gunnell et al. 2007, Pg 2.
  • "Suicide data were obtained from the Department of Police, Division of Statistics, Sri Lanka. Age, sex and method-specific data were only available for 1978, 1980, 1982–1985 and 1989–2015. The data are available for selected years online (http://www.police.lk/index.php/crime-trends), and missing years are available upon request directly from the Department of Police. In order to investigate trends, we restricted this analysis to the period with continuous method-specific suicide data available (1989–2015)." Knipe et al. 2017 Pgs 2-3.

• 5

  "In the 10 years since 1995, Sri Lanka’s suicide rates declined by 50%." Gunnell et al. 2007, Pg 2.

• 6
  • "From the late 1970s until the early 1990s, WHO Class I (‘extremely or highly toxic’); organophosphorus (OP) insecticides such as parathion, methyl parathion, monocrotophos, and methamidophos were the commonest poisons taken in fatal self-harm. The Registrar of Pesticides banned methyl parathion and parathion in 1984 and then phased out all the remaining Class I organophosphate pesticides gradually between 1991 and 1995, with a full ban on their [import] in July 1995." Gunnell et al. 2007 Pg 4.
  • "There appeared to be little impact on suicides of the bans placed on methyl parathion and parathion in 1984, other than perhaps a slowing of the rapid rise in suicide rate noticed between 1961 and 1983. However, the ban on class I pesticides (1995) and endosulfan (1998) were both followed by marked reductions in overall suicides." Gunnell et al. 2007, Pgs 4-5.

• 7
  • "The method-specific suicide data in Figure 4 indicate that the reduction in suicide occurred largely as a consequence of a reduction in self-poisoning and ‘other’ methods of suicide." Gunnell et al. 2007, Pg 5. See Figure 4.
  • Gunnell et al. 2007 gives four reasons for believing that suicides coded as 'other' methods mostly refer to pesticide self-poisoning (bold added):“We have four reasons to believe those coded as ‘other’ were mainly deaths from self-poisoning. First, in two years when the numbers of poisonings doubled (1983 and 1984) the number of suicide coded as ‘other’ halved, but there was no effect on the numbers of suicides using other specified methods (hanging etc.) (see Figure 2). Second, previous studies of suicide in Sri Lanka indicate that the principle method used for suicide is self-poisoning. In three population-based studies in different regions of Sri Lanka self-poisoning with pesticides accounted for over two thirds of suicides. Third, as noted above, pesticide suicides were categorized as a separate specific method of suicide from 1997 onwards and in that year 2589 pesticide suicide deaths were recorded. In the same year the total number of ‘poisoning’ suicides dropped from 1,954 in 1996 to 231; likewise the number of deaths coded as due to other methods declined from 3,818 in 1996 to 2027 in 1997 and 1397 in 1998 and continued to decline to 31 by 2005. At this time no other specific methods of suicide showed an increase in numbers suggesting most if not all the ‘transfer’ from ‘other means’ to specific means was to the pesticide poisoning category. Lastly, for the period covered by our analysis, the specific methods of suicide listed in the mortality data were self-poisoning (acetic acid), hanging, drowning, self-burning, firearms, self-piercing or jumping in front of a train—and these seven methods account for over 90% of suicide deaths in most countries. The ‘other’ category therefore is unlikely to consist of a different (unspecified) method. Based on these observations, in our analysis of method-specific suicide numbers, we therefore combined other and all self-poisoning categories into a single class—self-poisoning.” Gunnell et al. 2007, Pg 3.

• 8

  "Overall suicide mortality dropped by 21% between 2011 and 2015, from 18.3 to 14.3 per 100,000." Knipe et al. 2017, Pg 1.

• 9

  "Objective: To investigate the effect of 3-year phased bans of the pesticides dimethoate and fenthion in 2008–2010, and paraquat in 2009–2011, on suicide mortality in Sri Lanka." Knipe et al. 2017, Pg 1.

• 10

  "Overall suicide mortality dropped by 21% between 2011 and 2015, from 18.3 to 14.3 per 100,000. The decline in pesticide suicides during this same period was larger than for overall suicides: from 8.5 to 4.2 per 100,000, a 50% reduction. This was accompanied by a smaller concurrent rise in non-pesticide suicide mortality with a 2% increase (9.9 to 10.1 per 100,000)." Knipe et al. 2017, Pg 1.

• 11
  • "The second limitation of our analysis is that the data are ecological, thus limiting causal inference. Whilst the timing of reductions in suicide rates correspond[s] with pesticide related interventions and such downturns did not appear to be associated with changes in other risk factors for suicide, we cannot rule out the impact of other (unrecognised) factors in the absence of a control population. Our data are, however, supported by recent declines in the case fatality of pesticide poisoning in Sri Lanka. Furthermore, data from two other Asian countries where such pesticide regulation has not occurred—India and Korea—reveal differing trends in suicide. In India, Sri Lanka’s closest neighbour, self poisoning (mainly insecticides) accounts for 37% of suicides and there has been no equivalent decline in suicide rates. Suicide rates in India were 9.7 per 100,000 in 1995, 10.8 per 100,000 in 1998 and 10.3 per 100,000 in 2005. In Korea, rates of pesticide suicide doubled between 1991 and 2001." Gunnell et al. 2007, Pg. 7.
  • "Time series analyses of data from a single country, as we have here, must be interpreted with caution. There are trends over time in many things, and similar trends in two measures provide no convincing evidence of one causing the other, or of a common underlying cause." Knipe et al. 2017, Pg 9.

• 12
  • "An important weakness in our analysis is the poor quality of the method-specific suicide data. Over the period of time covered by the data, the categories used to identify the specific methods used for suicide changed. Furthermore, extensive use was made of the ‘other means’ of suicide category, despite the most commonly used methods of suicide in Sri Lanka (i.e. hanging, poisoning, drowning, burning, jumping in front of trains) being available as specific coding categories throughout the study period." Gunnell et al. 2007, Pg 7.
  • "A further limitation is that we may not be capturing the true incidence of suicide. There are no studies which have investigated the quality or reliability of Sri Lanka’s suicide data. Anecdotal evidence from our field research indicates that the quality of suicide data may vary by police departments, but we have no evidence to suggest that the data quality has changed over time." Knipe et al. 2017, Pg 9.

• 13
  • Dawson et al. 2010, a prospective cohort study of 9,302 pesticide poisoning patients in Sri Lankan hospitals between April 2002 and November 2008, found that dimethoate, paraquat and fenthion (the three pesticides banned between 2008 and 2011) were responsible for 47% of deaths from pesticide self-poisoning over the period, and were substantially more likely to result in death than the average for pesticide self-harm.
    • "Methods and Findings: We examined the case fatality of different agricultural pesticides in a prospective cohort of patients presenting with pesticide self-poisoning to two clinical trial centers from April 2002 to November 2008. Identification of the pesticide ingested was based on history or positive identification of the container. A single pesticide was ingested by 9,302 patients. A specific pesticide was identified in 7,461 patients; 1,841 ingested an unknown pesticide. In a subset of 808 patients, the history of ingestion was confirmed by laboratory analysis in 95% of patients. There was a large variation in case fatality between pesticides—from 0% to 42%. This marked variation in lethality was observed for compounds within the same chemical and/or WHO toxicity classification of pesticides and for those used for similar agricultural indications." Dawson et al. 2010, Pg 1.
    • "For those ingesting an identified pesticide, just three compounds with relatively high case fatality (paraquat, dimethoate, fenthion) were responsible for 17.6% of total admissions but 47% of the total deaths." Dawson et al. 2010, Pg 3.
    • "The overall mortality for pesticide self-poisoning (including patients who had ingested an unidentified pesticide) was 10.1% (CI [confidence interval] 9.5–10.8)." Dawson et al. 2010, Pg 3.
    • See Dawson et al. 2010, Table 2.
      • Paraquat: case fatality 42.7% (95% CI 38.6-46.9)
      • Dimethoate: case fatality 20.6% (95% CI 17.9-23.6)
      • Fenthion: case fatality 14.8% (95% CI 10.5-19.9)
  • van der Hoek & Konradsen 2006 is a retrospective analysis of 8,110 hospital admissions for acute poisoning in Sri Lanka between 1990 and 2002. It finds endosulfan had a higher case fatality (29.3% (see Table 2)) and caused more deaths than any other type of pesticide.
    • "Methods. Time series of incidence of acute poisoning based on retrospective in-patient records of six government hospitals in southern Sri Lanka from 1990 to 2002. Results. Data of 8,110 admissions for acute poisoning were available for analysis." van der Hoek & Konradsen 2006, Pg 225.
    • "Endosulfan was the single most important cause of death in this study population with extremely high case fatality ratio (Table 2), and it is therefore not surprising that the overall mortality rate of pesticide poisoning went down from 26.41 (95% CI, 23.77–29.05) per 100,000 over 1990–1998, to 16.34 (95% CI, 13.94–18.74) per 100,000 over 1999–2002." van der Hoek & Konradsen 2006, Pg 228.

• 14

  "Michael’s major research aim is to reduce deaths from pesticide and plant self-poisoning in rural Asia." Global Health Academy - Professor Michael Eddleston

• 15

  "Professor Eddleston had a huge impact. We had really bad data – no local information on which pesticides were being used for suicide. Professor Eddleston was very helpful in providing the facts for me to move forward. His research was on pesticide suicides and he collected lots of data on which pesticides were used and what were the best treatments." Centre for Effective Altruism, interview with Gamini Manuweera, November 23, 2016, Pgs 4-5.

• 16

  CPSP Milestones Gantt chart, June 2017

• 17

  “If CPSP identifies a few pesticides that are responsible for the majority of deaths from pesticide poisoning in India, it will try to conduct a cluster randomized controlled trial (RCT) to determine the impact of banning these pesticides on the rate of pesticide suicide and pesticide self-poisoning. The RCT would involve banning the relevant pesticides, or providing incentives for shop owners not to stock them, in half of 30-50 large areas and following up for 1, 2, or 3 years to track the number of deaths from pesticide poisoning in each area as well as the effect on both health and agricultural costs.” GiveWell's non-verbatim summary of a conversation with Professor Michael Eddleston and Leah Utyasheva, May 18, 2017, Pg 6.

• 18

  @CPSP budget, June 2017@

• 19

  GiveWell’s cost-effectiveness analysis of Centre for Pesticide Suicide Prevention

• 20

  CPSP Budget, June 2017

• 21

  “Professor Eddleston has been looking for funding for this work for about 15 years and has struggled to find funders:

  • He approached fundraising staff at the University of Edinburgh four or five years ago, but they were not able to find funding.
  • Professor Eddleston and Dr. Utyasheva have approached the United Kingdom's Department for International Development (DFID), but have not received a response yet.
  • It is possible that academic funders would be interested in funding the cluster RCT. Professor Eddleston wrote a funding proposal several years ago that he did not submit (because he submitted a different one instead), and could submit this now, though he expects that it would take at least a year for a successful proposal to receive funding.”

  GiveWell's non-verbatim summary of a conversation with Professor Michael Eddleston and Leah Utyasheva, May 3, 2017, Pg 8.

• 22

  "We found no drop in productivity in the years after the main bans were instituted (1995, 1998). We observed substantial annual fluctuation in estimated yields in all data sources, but these did not coincide with the bans and were no larger than the fluctuations in other countries. Also, there was no sudden change in costs of rice production coinciding with bans." Manuweera et al. 2008, Pg 492.

• 23

  "Mozambique has taken important regulatory measures to protect its people and the environment by cancelling the registration of 79 Highly Hazardous Pesticides. Working with the National Directorate of the Agrarian Services, in the Ministry of Agriculture, FAO helped bring the prohibition about and promote an ecosystem-based approach to pest and pesticide management." FAO undated press release

• 24

  Unpublished email from Professor Eddleston on June 4, 2017