The role of alcohol use in pesticide suicide and self-harm: a scoping review

This scoping review followed the Joanna Briggs Institute guidance on scoping reviews [23]. As the literature in this field was expected to be limited and diverse, a scoping review was considered appropriate. No protocol was registered. The search strategy was developed with support from a university librarian and used a combination of terms related to pesticides, alcohol use, self-harm, and suicide were adapted for each database (Supplementary Table 1) and citation linking was conducted for all included articles. Supplementary searches were undertaken using Google Scholar and websites of relevant global organisations, including WHO, the Food and Agriculture Organization, United Nations Office for Project Services, and Pesticide Action Network International. All searches were conducted on 3 March 2022.

Articles were included if they: (i) focussed on pesticide suicide and/or self-harm, and (ii) assessed involvement of alcohol in relation to pesticide suicide and/or self-harm (i.e. not just any method of suicide), (iii) were published in any language, (iv) were published since 2001 (as since then several countries around the world have banned highly hazardous pesticides [HHPs], which led to increased attention to risk factors for pesticide self-harm and suicide), and (v) were empirical articles, editorials, commentaries or reviews published in peer-reviewed journals; or research reports; government reports; book chapters; or conference abstracts. Articles were excluded if they: (i) had a broader focus on suicide generally (e.g. Widger [24]), providing few examples or cases and with no detailed discussion or analysis of alcohol’s role in pesticide self-harm, (ii) did not specify the method of self-harm or suicide and where associations with alcohol use was not assessed specifically for pesticide self-harm or suicide.

All articles were screened for inclusion based on title and abstract by LS and JBS. Full-text screening was performed by the same 2 researchers for 20 articles to assess level of agreement. The first ten articles yielded discrepancies in four articles; after discussion, the second ten reached full consensus (i.e. inclusion and exclusion criteria were clarified). LS screened all full-text records and extracted information using a pre-determined data extraction form: (i) year of publication, (ii) study location, (iii) study design, (iv) how alcohol use was assessed, (v) how alcohol use was involved in pesticide suicide and/or self-harm, (vi) key findings, and (vii) if and how any alcohol interventions were discussed as a strategy for suicide and self-harm prevention. For articles in languages other than English [25], Google Translate was used to screen full-text articles. One article in Spanish was extracted by LS who has an independent level of proficiency, though supported with Google Translate, recognised as a valid method [26].

Descriptive information was summarised in table format. The process of exploring common themes across included articles was iterative and informed the synthesis of findings, in addition to the pre-determined aspects of the data extraction form. LS developed the key themes, which were discussed with JBS and the wider research team. Thematic analysis [27] was used to synthesise the findings with a combination of deductive (pre-determined themes based on existing knowledge which were acute/chronic alcohol use and recognition of interventions) and inductive approaches (toxicological effects, gender and harm to others). The role of alcohol use was assessed as acute (as determined by self-report, clinical observation, or BAC), chronic (evidence of harmful, hazardous or dependent alcohol use, or overall alcohol use, as per self-report, clinical observation or assessment using diagnostic criteria such as ICD-10), or as harm to others (impact on family members or others from an individual’s alcohol use).

Following screening of 1281 records, 52 articles were included (Fig. 1). The articles covered a broad range of study designs and publication types. Most studies were quantitative, including case reports (n = 24), cohort (n = 9), case series (n = 7), cross-sectional (n = 5), case–control (n = 3), case series (n = 2), and before-and-after (n = 1) (Table 1). Of the remaining articles, three were qualitative (interviews and focus group discussions) and four were reviews.

Just under half of articles were from HICs (n = 24, 18 were case reports), followed by lower–middle (n = 22, four were case reports), upper–middle-income countries (n = 2, both case reports), and two had a global or regional (Asia) focus. Notably, the 20 epidemiological studies were from only 5 countries: Sri Lanka (n = 10), India (n = 4), Korea (n = 4), Taiwan (n = 2), and Spain (n = 1). Most studies related to hospital settings (n = 37), followed by community (n = 7) and autopsy (n = 6). In the 45 empirical studies that  assessed alcohol use in relation to pesticide self-poisoning just over half (n = 26) described the method for assessing alcohol use. One article was published in Spanish [28] and the remaining articles in English. Characteristics of all included case reports are summarised in Table 2 and characteristics of all other included studies in Table 3.

Dhanarisi et al. [58] reviewed studies that reported on co-ingestion of alcohol and pesticides. Fourteen studies were included in this review and no difference was found in length of hospital stay or amount of pesticide ingested, in studies which measured these indicators. In the one study, Eddleston et al. [60] that measured concentration of pesticide (dimethoate) patients who consumed alcohol also had higher pesticide concentration. Meta-analytical results from this same review indicated that patients who co-ingested alcohol were more likely to require intubation (OR = 8.0, 95% CI 4.9–13.0, p < 0.0001) and more likely to die from pesticide self-poisoning (OR = 4.9, 95% CI 2.9–8.2, p < 0.0001) [58]. The authors noted that higher risk of death could be related to higher suicidality, underlying health conditions, or higher amount of pesticide consumed. While alcohol appears to have a contributory effect to fatal outcomes, the authors concluded that “the data presented are insufficient to conclude how this secondary contributory factor would be responsible for increased fatal outcomes”. Furthermore, they highlighted that chronic alcohol use was not reported in most studies, which could be a confounder in the association between acute alcohol use and pesticide poisoning [58]. In addition, in relation to the amount of pesticide ingested, a toxicology review by Eddleston et al. [62] suggested that acute alcohol intoxication can cause complications due to alcohol withdrawal and alcohol cardiomyopathy, which can increase complications of tachycardia in organophosphate (OP) poisoning, impacting patient management and risk of death. The combined effect of alcohol and pesticides could explain the higher mortality among middle-aged men than women [59] and increased risk of coma [62].

The majority of studies in this category were epidemiological (n = 21), and in 13 of these studies, the prevalence of alcohol co-ingestion was reported. The average proportion of patients with alcohol co-ingestion was 30%, ranging from 15% in 110 patients with bispyribac poisoning in 2 hospitals in Sri Lanka [63] to 68% of 91 ‘suicidal patients’ across 5 emergency centres in the Republic of Korea [69]. One study of patients presenting to a medical centre in Taiwan only included complex suicide cases (i.e. cases with at least one other means in addition to pesticides), of whom more than half (54%) had ‘alcohol intoxication’ (assessment method not reported) with no significant difference between those with and without previous events of self-harm (‘suicide attempts’) [64]. A Sri Lankan study across three hospitals was the only study that assessed BAC and found that more than half of patients (51%) had a BAC ≥ 0.05 mg/dL and a median of 0.15 mg/dL [60].

In just over half of these studies (n = 11/21), alcohol use was assessed in specific terms, while in the remaining studies, this was described without detail. For example, Kim et al. [66] noted that those ‘under the influence of alcohol’ had consumed ‘more pesticides’ and Venugopal [75] reported that ‘the mode of ingestion’ in 20% of patients was ‘with alcohol’. Weerasinghe et al. [79] found that 28% of customers who purchased pesticides for self-harm were ‘under the influence of alcohol’ (self-reported) at the time of purchase, compared to 0.5% of customers who bought pesticides for other purposes [79].

Of the 21 studies that reported epidemiological data, 6 reported alcohol measures by gender [28, 56, 57, 60, 73, 79]. In these six studies, men predominantly self-harmed and co-ingested alcohol. In four of these studies, all patients who had co-ingested alcohol were men [28, 56, 57, 79], and in one study, 97% were men [60]. Tu et al. [73] did not stratify co-ingestion of alcohol by gender but among patients who underwent psychiatric assessment, more than eight in ten of those with an AUD were male (84%). Among all men, prevalence of AUD was 19% compared to 7% among women (p < 0.001) [73].

Underlying chronic alcohol use was prevalent among self-harm cases but definitions and assessment methods varied. In a Sri Lankan study, among participants who underwent a mental health module, 10% had probable alcohol dependence (DSM-IV), with an OR for probable alcohol dependence of 5.26 (95% CI 1.06–26.11), compared to controls [74]. In a hospital-based study from Taiwan, 26% of assessed patients were reported to have an AUD (DSM), with no significant difference between those who had a first and subsequent self-harm event [64]. A similar proportion was observed in a sample of patients in a general medicine ward in Kerala, India, where 23% of patients presenting at a general medicine ward had alcohol dependence, as per ICD-10 [70]. In Taiwan, at a population level, reductions in suicides were found following a paraquat ban but this was not associated with patterns of drinking [55].

Of the six case reports where chronic use was mentioned, it was not clear whether alcohol was also implicated at the time of the event in half of these studies (n = 3/6). For example, Bilics et al. [32] noted that the patient had a history of “chronic alcoholism” via medical history but there was no indication of assessment of acute alcohol consumption at the time of self-harm.

Two articles mentioned harm to others from alcohol [67, 72]. Konradsen et al. [67] explored alcohol use related to pesticide self-poisoning in Sri Lanka and found that in 40% of 159 cases, ‘alcohol misuse’ or ‘addiction’ reportedly played a role in self-harm. In half of these cases, the person who was drinking self-harmed, while in the remaining cases, family members self-harmed due to the drinking from the father of the household [67]. This was related to domestic violence, adverse impact on disposable income, shame and embarrassment [67]. Similarly, Sørensen et al. [71], in their exploration of self-harm in Sri Lanka, described alcohol as a domestic problem that exacerbated other daily life stressors, leading to self-harm. One case described how “while drunk, the man blamed his partner for the daughter’s promiscuous behaviour noting how their relatives would speak badly about them. This turned into a violent fight followed by the woman ingesting pesticides.” (p.4) [72]. Both Konradsen et al. [67] and Sørensen et al. [72] highlighted issues of gender differences; self-harm events which either involved men who self-harmed or their significant others or family members, who self-harmed in response to their own/their family members’ alcohol use.

Just seven studies discussed the need for alcohol interventions as a strategy to prevent pesticide self-harm and suicide. In their study of pesticide vendors’ role in preventing pesticide self-harm in Sri Lanka, Weerasinghe et al. [77] found that the majority of vendors (84%) increased their knowledge of the importance of not selling pesticides to individuals who were under the influence of alcohol. The authors suggested that training of vendors could help reduce pesticide self-harm, which was a favoured intervention by the stakeholders [78]. Dhanarisi et al. [56], in a Sri Lankan context, called for public health campaigns to reduce alcohol use and increase awareness of negative effects on health from drinking. Eddleston et al. [59], also in Sri Lanka, acknowledged that reducing alcohol consumption is part of pesticide self-harm prevention, which required ‘community efforts’; however, this was challenging due to ‘political power’, ‘drinks industry’ and ‘illegal distilling of alcohol’. Prakruthi et al. [71] were more specific suggesting that interventions should include stress management, coping skills and treatment for alcohol dependence and depression. A case report by Fellmeth et al. [40] described a suicide of a couple in a refugee camp, in which the authors noted the need for early identification of alcohol dependence and mental disorders in these settings.

Our findings demonstrate the importance of alcohol consumption in pesticide self-harm and suicide. As Dhanairisi et al. highlighted in their systematic review/meta-analysis, few studies have assessed dose of pesticide and outcomes among patients with alcohol co-ingestion, and with varying results [58]. In a prospective case series from Sri Lanka, which found that the dose of profenofos was not associated with outcomes [56], we question whether alcohol was behaviourally mediated in relation to self-harm. More studies are needed to understand not only the toxicological effects but also the mechanisms of alcohol consumption in pesticide self-harm. This will help explain why risk of death is higher in patients with alcohol co-ingestion.

The broader range of harm from alcohol use, or misuse, was less frequently explored. The most detailed accounts came from Konradsen et al. [67] who described how self-poisoning in Sri Lanka had become a response to difficult situations and a powerful communication method. Similarly, Marecek and Senadheera [73] described this phenomena as ‘dialogue suicide’, as opposed to monologue suicides, characterised as being solitary and inward focussed acts—as often seen in HIC settings [80]. Similarly, other qualitative studies on the link between alcohol and self-harm in Sri Lanka found that alcohol played a direct role in men’s self-harm [72]. Women were indirectly influenced by someone else’s alcohol use and interpersonal conflict often led to self-harm through which women would seek to teach their husbands a lesson to enable the drinker to moderate their alcohol use [72]. Studies from Uganda and South Africa, which explored all methods of suicide, found both direct and indirect impacts from alcohol in suicide cases [81, 82], with early onset of alcohol use and current alcohol dependence being particularly important factors [82]. Importantly, in our review, included quantitative studies from hospital-based samples did not elucidate whether patients may have engaged in acts of pesticide self-poisoning due to someone else’s alcohol use.

This review identified papers from several countries but, as expected, many focussed on Sri Lanka, which has been the epicentre for suicide prevention research over several decades [83]. Here, research capacity has developed to carry out large-scale self-poisoning studies, including via international research collaborations [86, 87]. Specifically, numerous studies have been conducted on self-harm and suicide, including on the steep reductions in suicide rates following bans of several kinds of pesticides [83‐85]. However, more research is needed in other countries to explore the link between alcohol and pesticide self-poisoning. We identified only a few studies from India, and none from China, though these two countries account for more than four in ten (44%) of all suicides globally [1], with pesticides among the most common means [8, 87], and alcohol use an important risk factor [89, 90]. Furthermore, this review found just one included record, a case study [45], from the African region. This reflects scarcity of data on pesticide suicide in the African region [2], despite evidence suggesting pesticide poisoning is a common method of suicide [91]. While more research in other countries is needed, it is worth noting the challenges of measuring alcohol consumption [92], including in Indigenous peoples [93]. To address these challenges, in an Australian context, an interactive and visual tablet computer-based survey tool has been developed and validated to help Aboriginal and Torres Strait Islander peoples describe their alcohol consumption [94‐96]. Such a tool may have the utility to improve epidemiological data on alcohol consumption, self-harm, and suicide in other contexts.

The global increases in alcohol use at the overall population level makes the need for further research in this area even greater. Projections suggest alcohol per capita will continue to rise in the SEAR, Western Pacific Region (WPR), Eastern Mediterranean Region (albeit a small increase from a low baseline) and Region of the Americas—leading to an overall increase in global alcohol per capita consumption. Past changes in alcohol use levels in SEAR and WPR have been driven by sharp increases in India and China [13], which along with knowledge of burden of suicide in these regions further emphasises the need to explore the combination of these two public health issues.

In this review, few studies specifically set out to study the role of alcohol use in pesticide self-harm and suicide. This might explain why descriptions of methods used to assess alcohol use and further details about context drinking were limited. We were particularly interested in exploring the role of gender, but of the 20 studies reporting on epidemiological data, only 6 reported on alcohol measures (acute or chronic alcohol use) by gender [28, 56, 57, 60, 73, 79]. Furthermore, the method used to assess alcohol use was lacking in some papers and there was no uniformity in the way alcohol was assessed. In several case studies, the patient or deceased was noted to have been under the influence of alcohol, though this was not verified through BAC testing. Similarly, there was no uniformity in the way chronic alcohol use was assessed, as impact of acute and chronic alcohol use has particular relevance for populations where drinking is impacted by context [97]. The risk of self-harm (‘suicide attempts’) at lower levels of alcohol use increases the risk sevenfold while heavy drinking increases the risk by 37 times [15]. Comparable measures of alcohol use that can be stratified may, therefore, be needed in the future research to assess risk from any alcohol use as well as magnitude of risk at different levels. Limitations on reporting of data on method of suicide [15] made it difficult to draw conclusions about alcohol use and lethality of method or added toxicological effects from co-ingestion. On average, in our review, alcohol was involved in one-third of the cases of pesticide self-harm or suicide, which reflects the overall prevalence of reported alcohol use in previous studies including all methods of suicide [98].

While few articles in this scoping review acknowledged how alcohol interventions at an individual- or population level can help to prevent pesticide self-poisoning, there is recognition within the wider suicide prevention field. Alcohol has been integrated in the WHO’s Live Life document which provides Member States with an implementation guide to suicide prevention [98]. In Sri Lanka, a policy document outlining recommendations for action in suicide prevention emphasises AUD, specifically, as a risk factor in suicidal behaviour [99] and the need to address this within a suicide prevention framework [100]. In addition to a suicide prevention approach, strengthened alcohol policy can impact on suicide rates. In a systematic review, Kӧlves et al. [101] showed that across studies from Europe (including Russia and USSR) and the US, there was evidence that suicide rates have changed alongside alcohol policy changes, particularly by changing availability of alcohol (restricting or increasing availability) and pricing policies [102]. These are known as the ‘best buys’ to reduce harmful use of alcohol, along with advertising restrictions [102], which should be adapted to national contexts [103]. Whereas stricter alcohol policy is needed in South-East Asia to achieve the 2030 SDG target of reducing alcohol per capital consumption by 10% [20], other interventions are needed to target the widespread use of illicit alcohol. In Sri Lanka, a community-based participatory intervention showed promising results for moderating alcohol use [104]. If proved effective at larger scale, such interventions might be relevant in other contexts too, as it has been highlighted that in Sri Lanka alcohol consumption has an important social role and abstinence-based interventions may therefore not be successful [24, 72]. These types of community-based approaches may be effective, in addition to population-level interventions, to address contextual factors such as gender-based violence, mental health issues, poverty, and other stressors that co-occur with harmful alcohol use.

Preventing suicide by pesticide self-poisoning can also be effectively reduced by banning HHPs [105]. In Bangladesh, Chowdhury et al. [106] showed that the reduction in suicide deaths following pesticide bans was not associated with population-level ‘alcohol misuse patterns’ and Gunnell et al. [85] also found that declining suicide rates in Sri Lanka were not related to a decline in alcohol sales (a proxy for population-level alcohol use). As these reductions were independent of overall alcohol consumption, it suggests that targeted alcohol interventions alongside bans of the most acutely toxic pesticides may be a way forward to reduce suicide and self-harm rates.

This scoping review was set up with input from a university librarian, to ensure the search strategy was comprehensive; however, the protocol was not registered prior to undertaking the study. The team involved in the review is multidisciplinary and represented a range of regions including South-East Asia, Africa, Europe, and Western Pacific. The review has some limitations that should be acknowledged. While the title and abstract screening was done by two researchers independently, only a subset of full-text articles were screened by two reviewers and the remaining articles were screened by one researcher. However, continuous discussion between two researchers ensured clarity on inclusion criteria. Articles of any language were eligible for inclusion in this review; however, the searches were only conducted in English language databases.