How many premature deaths from pesticide suicide have occurred since the agricultural Green Revolution?

Introduction: The agricultural Green Revolution in the 1950s and 60s is thought to have averted many deaths from famine. However, it also introduced highly hazardous pesticides such as parathion and endrin into poor rural communities that were totally unequipped to store or use them safely. Pesticide self-poisoning rapidly became one of the two most common global means of suicide. Thus far, no attempt has been made to enumerate the total number of deaths that have occurred subse- quent to the Green Revolution. Objective: To calculate plausible estimates for the total global number of pesticide suicides that have occurred since 1960. Methods: We performed a literature review on Medline and Embase databases to July 2019 to find papers that reported national or global numbers of pesticide suicides. We restricted our search to papers published in English. We used the search terms: pesticide) OR insecticide) OR paraquat) OR organophosphate) OR organophosphorus) OR agrochemical) AND suicide) OR ““ ) OR deliberate) AND poison in all fields. These searches identified 2,144 papers; a further 8 citations were added through the searching of reference lists and our own paper collections. 2,136 papers were excluded as they contained no data on pesticide suicide, or were case reports, case series, or related to specific socio-demographic groups, or were non-human studies. This left 16 papers giving country specific or global pesticide suicide data. Long-term national trend in pesticide suicides: We found studies from one low- and middle-income country (Sri Lanka) that recorded long-term trends in suicide throughout the Green Revolution. These data showed a steady increase in suicides from 1960 to the early 1970s, with a more rapid increase from 1979 to 1984. The number of suicides plateaued until 1995, when they started a steady almost linear decrease that has continued at least until 2015. We used the Sri Lankan epidemiology as a model of the incidence of pesticide suicides in other low- and middle-income countries. Data from Bangladesh suggested that the decrease might have started in 2002. Estimating global numbers since 1960: Starting from a conservative estimate of zero deaths in 1960, the best estimate of the total global burden of pesticide suicides from 1960 to 2018 is 14,272,105 or 14,936,000 (depending on whether a fall in incidence began in 1995 or 2002), with a plausible range of 9,859,667 to 17,303,333 deaths. These are likely underestimates because suicide is illegal in many countries, and most pesticide suicides occur in poor rural areas without effective death registration systems. Conclusions: Pesticide self-poisoning has been a major clinical and public health problem in rural Asia for decades, while being long ignored. Most pesticide suicides are relatively impulsive with little planning: in the absence of highly hazardous pesticides, many people would have survived their suicidal impulse, gone on to find support amongst family, community, and health services, and lived a full life. Pesticide suicides must therefore be considered a category 4 occupation condition following Schilling ’ s classification - if they had not been brought into rural communities for agricultural use, pesticide suicides would not have occurred. Preventing these deaths should be a global public health priority.


Introduction
The Green Revolution in the 1950s and 60s introduced new farming techniques and high-yield varieties of crop plants, particularly wheat and rice, into low-and middle-income countries [1,2]. Yields increased markedly over the following decades, permitting food production to keep pace with population growth and reduce the risk of famine. However, the Green Revolution had downsides, with the new crop varieties being dependent on pesticides and fertilisers, producing harmful effects on human health and environment [1,3,4]. Subsequently, intensified agricultural pesticide use has resulted from a shift from subsistence farming to cash-crop and monocrop farming, a need to increase yield per area of land, a limited focus upon developing alternatives to pesticides use, and an overall increasing focus upon input dependent agriculture [5,6]. Alternative farming approaches without use of pesticides lack the heavy marketing and lobbying support provided by the agrochemical industry [7,8].
The introduction of highly hazardous pesticides [9] such as the organophosphorus (OP) and organochlorine insecticides parathion (E605) and endrin, respectively, [10][11][12][13][14] into poor rural households that were totally unequipped to use or store them safely had immediate repercussions, with deaths soon being reported from unintentional occupational poisoning and intentional (suicidal) self-poisoning [13,[15][16][17][18]. Self-poisoning with non-pesticide products [19] was relatively safe before the arrival of highly hazardous pesticides. Unfortunately, the easy availability of these pesticides markedly increased the lethality associated with self-poisoning, changing non-fatal to fatal poisoning, and causing a rapid increase in both pesticide and total suicides, as best illustrated by Sri Lanka [20,21].
Pesticide suicides are likely to have killed many millions of people over the last 50-60 years. In 1990, based on Jeyaratnam's work [22], the World Health Organisation (WHO) estimated that 200,000 pesticide suicides occurred each year (while at the same time acknowledging the paucity of data) [23].
Pesticide self-poisoning has been a major clinical and public health problem in low-and middle-income countries for decades while being long ignored [18,24]. Unfortunately, it is difficult to estimate the global number of pesticide suicides largely because most occur in poor rural regions, with weak death certification and reporting systems [25,26]. This problem has been exacerbated by the illegality of suicide in many countries, hindering reporting of suicides and recognition of the scale of the problem [27].
More recently, we have performed systematic reviews to estimate plausible estimates, taking account of underestimation of pesticide suicides in India, of 372,000 deaths/year in the 1990s [28] and 168,000 deaths/year in 2010-2014 [29]. However, no attempt has been made to enumerate the total number of deaths that have occurred subsequent to the Green Revolution.

Objective
To calculate plausible estimates for the total global number of pesticide suicides that have occurred since 1960.

Methods
We performed a literature review on Medline and Embase databases to July 2019 to find papers that reported national or global numbers of pesticide suicides. We restricted our search to papers published in English. We used the search terms: pesticide) OR insecticide) OR paraquat) OR organophosphate) OR organophosphorus) OR agrochemical) AND suicide) OR "") OR deliberate) AND poison in all fields. These searches identified 2144 papers; a further 8 citations were added through the searching of reference lists and our own paper collections. 2136 papers were excluded as they contained no data on pesticide suicide, they were case reports, case series, or related to specific socio-demographic groups, or were non-human studies. This left 16 papers giving country specific or global pesticide suicide data.

Long-term national trend in pesticide suicides
From a review of Medline and Embase databases from inception, we identified three studies that estimated the global burden of pesticide suicides [16,28,29]. Our previous studies from one low-and middle-income country (Sri Lanka) recorded long-term trends in suicide throughout the Green Revolution [20,21,30,31]. These data showed a steady increase in suicides from 1960 to the early 1970s, with a more rapid increase from 1979 to 1984 when pesticide imports were liberalised by a new government [20,32]. The number of suicides plateaued until 1995, when they started a steady almost linear decrease that has continued at least until 2015. The rapid increase in suicides after 1960 was almost completely due to poisoning and most likely pesticides [20]. We used the Sri Lankan epidemiology as a model of the incidence of pesticide suicides in all low-and middleincome countries.
We found no publications from low-and middle-income countries other than Sri Lanka that reported the timing of the increase in suicide incidence after the Green Revolution. However, data from China [11,33], India [10,13,34], Sri Lanka [34][35][36], United Kingdom [37], Israel [38], Denmark [39] and Finland [40] indicate that both unintentional and intentional pesticide poisoning with parathion and endrin were already problems by the end of the 1950s, sufficient to cause Finland to ban parathion in 1960 [40] and Denmark to restrict its availability in 1961 [39]. Unfortunately, despite serious concerns expressed in both India [13] and Sri Lanka [36] about the dangers of making these highly hazardous pesticides freely available over the counter to small-scale farmers in low-and middle-income countries, no such country banned their use at this time. Such a ban would have saved many lives.
One paper reported the timing of reductions in pesticide suicides in Bangladesh as starting in 2002 [41]. Chinese papers were unable to show a start date for the reduction in cases, since they started in 2006 when the incidence was already falling [42,43].

Estimating global numbers since 1960
Conservatively, bearing in mind the evidence that pesticide poisoning was already a problem in the late 1950s, we took 1960 as the approximate beginning of the Green Revolution, when pesticides became widely available for self-poisoning and suicide in small rural communities, consistent with the Sri Lankan data. The number of pesticide suicides in that year was therefore set at zero.
In Sri Lanka, deaths increased until 1984 and remained steady until 1995 [20]. In Bangladesh deaths remained steady probably until 2002 [41]. Gunnell and colleagues [28] estimated that the global number of pesticide suicides for the 1990s to be 372,000 (plausible range 347,000 to 439,000), which accounted for likely undercounting of pesticide suicides in India (as subsequently proven in the Million Death [44] and Global Burden of Disease studies [26]). To account for the number of suicides from 1960 to 1984, we presumed that they increased steadily over the period and took half the estimated number of suicides in 1984 (372,000/ 2 ¼ 186,000) as the annual number of pesticide suicides for these 24 years -a total of 4,464,000 (plausible range 4,164,000 to 5,268,000) global pesticide suicides from December 1960 to December 1984 (Table 1).
Using the Gunnell et al. [28] estimate for the static phase from Jan 1985 to December 1995 (Sri Lankan data) or December 2002 (Bangladeshi data), a total of 4,092,000 (3,817,000 to 4,829,000) or 6,696,000 (6,246,000 to 7,902,000) people, respectively, died from pesticide suicide during this period.
Mew and colleagues [29,44] estimated that 168,000 pesticide suicides occurred in 2014 (including an estimate of missing Indian pesticide suicides). To calculate the number of pesticide suicides between the two estimated time points for the start of the fall in incidence (January 1996 and January 2003) and 2014, we took the midpoint (270,000) of the peak estimate 372,000 and the 2014 estimate 168,000 and calculated the number for these 19 and 12 year periods: 5,130,000 and 3,240,000 respectively.
Estimates for January 2015 to December 2018 were calculated from extending the line linking 1996 or 2003 to 2014, providing estimates for these four years of 586,105 (557,895 to 596,632) and 536,000 (491,333 to 552,667) deaths, respectively.
Putting together these three periods of increase, stability, and reduction (1960-1984, 1985-1995 or 2002, and 1996  In 1990, WHO estimated the number of global pesticide suicide deaths in the 1980s to be 200,000, although without providing supporting data [23]. Using this value for 1984, rather than the value of 372,000, we calculated lower estimates of 12,208,105 and 12,872,000 depending on whether the fall began in 1995 or 2002. Outer estimates for pesticide deaths over the 59-year period using this WHO estimate were 9,859,667 to 14,435,333.

Discussion
The number of pesticide suicide deaths has fallen over the last 10-20 years, largely driven by a reduction in China [42] which was previously responsible for around half of the world's pesticide suicides [45,46]. This reduction has likely occurred due to movement of people from rural areas (where highly hazardous pesticides are widely available) to cities (where such pesticides are used infrequently), to mechanisation of agriculture, and to pesticide regulations which have banned highly hazardous pesticides [42].
Importantly, pesticide bans have not been associated with reduced agricultural yield [31,41,47,48]. Continued global regulation of pesticides to remove highly hazardous pesticides from poor smallholder farms in low-and middle-income countries, that are totally unable to use them safely, will further and rapidly reduce pesticide and total suicide rates [49].
In summary, we have here attempted to calculate the number of pesticide suicide deaths that have occurred since highly hazardous pesticides were introduced into rural agriculture of low-and middle-income countries in the 1950s/ 60s. We believe that a plausible estimate is at least 14 million deaths (with outer estimates of 9.9 and 17 million using WHO and Gunnell et al. [28] values for 1984, respectively) which is close to the total number of lives lost during WWI. It is also close to half the number of deaths that have occurred from human immunodeficiency virus infection (35 million) yet has comparatively received almost no public health attention. Many pesticide suicide attempts are impulsive with little planning [50,51]. Many deaths have been of young people with little desire to die but who died because the poisons at hand in their rural communities were highly hazardous pesticides for which no effective treatment is available [14,23,52]. Similar patients poisoning themselves in high-income countries would have survived due to the relatively low toxicity of available poisons [53].
This large number is likely to be an underestimate because of the registration and legal issues faced recording suicide deaths [25]. Studies in India in particular have reported very high rates of pesticide suicides in rural community verbal autopsy studies, rates far higher than the official statistics [25,54,55]. The global estimates proposed by Gunnell et al. [28] and Mew et al. [29] attempted to account for these high unofficial rates by increasing the estimate for India, following Table 1. Estimated pesticide deaths from 1960 to 2018 (using the Gunnel et al. [28] and Mew et al. [29] estimates, and 1995 vs 2002 for the end of the stable phase).

Limitations
Our review has several clear limitations, most importantly the lack of accurate annual global data for deaths from pesticide poisoning. Instead, we have relied on two systematic reviews of the literature for the number of cases at the peak of the problem in 1990-2007 and 2010-2014. These provide us with plausible estimates for these two time points, allowing an estimate to be calculated for the total number of deaths since 1960.
Reasonably accurate national data from low-and middleincome countries on the epidemiology of pesticide suicides from before 1980 was only available from Sri Lanka. However, the Sri Lanka data provided a basic structure for the estimatean increase from zero when pesticides were introduced into agriculture, a peak point followed by a period of stability, then a reduction due to pesticide regulation and perhaps migration from countryside to cities. Pesticide suicides were first reported before 1960 but the numbers were likely to have been very small compared to our overall estimate. It is possible that pesticide suicides increased more slowly in other countries, but we have no data to support this idea. Data from China and India -the two countries where most pesticide suicides occur -clearly show that poisoning was a significant problem by the end of the 1950s. The duration of the stable phrase is unclear, and so we have used two estimates -one from Sri Lanka and one from Bangladesh. To estimate the numbers of cases from 2015-2018, we have proposed that the number of cases has continued to fall steadily after 2014. This is likely conservative since there has been only a very modest fall if any in India, where the majority of global pesticide suicides now occur [26,29].
Both Sri Lanka and Bangladesh, like many others, first used pesticides in the 1950s, with increasing use and novel pesticide classes over subsequent decades [57]. Review of data from the Food and Agricultural Organisation of the United Nations (FAO; www.fao.org/faostat/) indicate that Sri Lanka's pattern of pesticide use (stable from 1990, then declining from 2010) resembles India, Sub-Saharan Africa, Central America & Caribbean, and Central Asia while Bangladesh's pattern (slow increase from 1990, more marked increase from 1998) is similar to low-and middle-income countries in South America, Northern Africa and East Asia. This suggests that the two countries are reasonably representative of global pesticide use in low-and middle-income countries. However, we acknowledge that the population with easy access to highly hazardous pesticides for self-harm will be mostly determined by the number engaged in smallscale agriculture, who keep pesticides in their homes, rather than overall national pesticide use.
We did not attempt to enumerate deaths from accidental or occupational poisoning, although these were estimated to number at least 20,000 deaths per year in the 1980s [23]. There will likely have been many tens of thousands of nonsuicide deaths from pesticide poisoning worldwide since 1960. Similarly, this paper does not capture the health impacts of environmental exposure to pesticides.

Conclusions
In conclusion, pesticide suicide rates increased dramatically once the Green Revolution brought highly hazardous pesticides into poor rural households resulting in an estimated fourteen million premature deaths. The number is now falling worldwide but still accounts for around 150,000 deaths each year. Pesticide suicides must be considered as a category 4 occupation condition following Schilling's classification [58,59] -if pesticides had not been brought into rural communities and houses for agricultural use, pesticide suicides would not have occurred. Effective prevention will require better data and should rely on pesticide regulation to remove all highly hazardous pesticides from small scale farming [60], since improved pesticide storage is unlikely to be effective [61].