Prevention of alcohol-related suicide: a rapid review

Abstract

Suicide remains a leading cause of death worldwide, with an estimated 700,000 suicide deaths per year. The World Health Organization identifies reducing alcohol use as one component of comprehensive approaches to suicide prevention. This paper conducted a rapid review of the evidence on alcohol-related suicide prevention interventions. PubMed, Embase and Web of Science were searched for articles related to alcohol, suicide, prevention, and policies, published between 1990 and 2020. 5293 articles were identified; after deduplication, 2567 studies were screened at the title and abstract level. 402 articles underwent full-text review. 69 articles were ultimately included and underwent data extraction. Interventions were categorized as policy interventions, community-based interventions, and clinical interventions. While there is evidence that policy interventions targeting alcohol may be associated with lower suicide rates, more evidence using stronger study designs is needed. The evidence for community interventions was mixed and supported the need for further research on these types of interventions. Pharmaceutical and therapy-based clinical interventions also showed some promise, with more research needed. Overall, despite evidence of alcohol’s role in suicide attempts and deaths, few interventions have been developed with the purpose of addressing alcohol-related suicide. More research is needed to identify effective interventions to prevent alcohol-related suicide.

Keywords:

• Suicide
• alcohol
• policies
• prevention
• community
• clinical

Introduction

Suicide is a leading cause of death around the world. The World Health Organization (2021) reported that the unadjusted global mortality rate from suicide was 9.2 deaths per 100,000 in 2019; in Canada, it was 10.7 per 100,000 (Government of Canada, 2020) and in the United States it was 14.5 (Center for Disease Control & Prevention, n.d.). At the population level, suicide deaths are only the most serious of outcomes on the suicide and self-injury pyramid – for every suicide death there are many more hospitalizations for intentional injury, and even more suicide attempts that are not captured through routine surveillance (World Health Organization, 2014). Both outcomes have a significant burden on those who experience suicide-related behaviour and their loved ones, traumatic impact on first responders, and major costs to society (Jordan, 2017; Kinchin & Doran, 2018). Suicide behavior is highly gendered, and while ratios vary by global region, males are more likely to die by suicide (World Health Organization, 2014).

There are many proximal and distal factors that contribute to suicide, and they vary by region, community, and population. One significant risk factor is alcohol use, and reducing alcohol consumption is recognized by the World Health Organization (2014) as an integral component of comprehensive suicide prevention. Alcohol is estimated to be a direct contributing factor in a substantial portion of suicide deaths, with one-third of suicide decedents in the United States testing positive for serum alcohol (Kaplan et al., 2020). Alcohol may be a means to suicide itself, through alcohol poisoning, or in overdose in combination with other drugs. In addition to acute alcohol use at the time of the attempt, alcohol may contribute to suicide via alcohol use disorder and the wide-ranging impacts of living with this disorder. The stimulating effects of alcohol may result in impulsivity, disinhibition or aggression, while the depressant effects may include feelings of sadness or despair (Hendler et al., 2011; Sher, 2006). Finally, alcohol use may be associated with mental illness, and thus contribute to suicide behavior through the shared associations between mental illness, alcohol use and suicide (World Health Organization, 2014).

Cherpitel et al. (2004) reviewed 16 papers focusing on acute alcohol use and suicidal behavior. They found a wide range of alcohol-positive results, from 10 to 69% for suicide deaths and 10 to 73% for suicide attempts. A meta-analysis of the acute use of alcohol and risk of suicide attempt by Borges et al. (2017) based on seven papers found a common odds ratio of 6.97 for any acute use of alcohol and suicide attempt. Canadian research by Orpana et al. (2021), conducted a narrative synopsis of papers published between 1998 and 2018 reporting that ecological studies support the association between alcohol sales, annual per capita consumption, and suicide rates.

Despite the body of evidence, the association between alcohol and suicide is largely overlooked in suicide prevention programs and policies by clinicians, researchers and policy-makers. Giesbrecht et al. (2017) found reference to alcohol in only 5 of 53 reviews focused on suicide prevention that were published between 2011 and 2017. Furthermore, in their systematic review of suicide prevention strategies, van der Feltz-Cornelis et al. (2011) made no reference to alcohol. Witt and Lubman (2018) note that alcohol is absent in the discussion of the prevention of suicide.

Nevertheless, several reviews focusing specifically on alcohol and suicide prevention have been conducted. Xuan et al. (2016) examined 17 studies in the English language literature published between 1999 and 2014 that focused on alcohol policies and suicide deaths. Common alcohol policies include control of physical availability of alcohol, pricing of alcohol, drink-driving countermeasures, and, in some cases, multiple measures. While there was some inconsistency in the outcomes, they found general support for the protective effect of restrictive alcohol policies on preventing suicide and decreasing the level of alcohol involvement among suicide decedents.

Kolves et al. (2020) published a systematic review on the impact of alcohol policies on suicidal behavior. From the 19 papers in their analysis, covering policies implemented between the early 1900s and 2008, they concluded that studies suggest alcohol policies may contribute to reducing suicide at the population level. However, the studies applied a variety of methods, generated a range of effect sizes, and the risk of bias for many studies was unclear.

Hurzeler et al. (2021) report on a systematic review of randomized controlled trials to prevent both suicidal behavior and alcohol use among populations with alcohol use disorders or problems. Six studies were included, and the authors concluded that some psychotherapies might hold promise in reducing suicidal behavior, but that the current evidence is inconclusive.

A recent paper published by Witt et al. (2021), after the cut-off date of or our search, involves a systematic review and meta-analysis of the effect of alcohol-related psychological interventions on suicidal ideation and behaviour. They analyzed 11 studies. The authors note that there was no apparent effect on suicide ideation or suicide deaths.

Our contribution builds on this literature focusing on the relationship between alcohol and suicide. Specifically, we conducted a rapid review of alcohol-related suicide prevention policies and interventions, including, but not limited to, alcohol policies. Although previous reviews exist on the prevention of alcohol-related suicide, such reviews have been limited in scope to examining specific interventions. To our knowledge, our study is the first of its kind to synthesize all the available evidence on interventions to prevent alcohol-related suicide into a single review. Moreover, while previous studies were limited to specific interventions, our study employs less strict inclusion criteria than previous reviews, allowing the detection of studies with varying designs and settings. We considered both attempted suicide and suicide deaths as our outcomes. Given the high proportion of suicide deaths internationally that involve alcohol, combined with the low attention to alcohol as a contributing cause (Witt & Lubman, 2018), it is timely to take stock of what is currently known about how to prevent alcohol-related suicide by various interventions, policies or strategies.

Materials & Methods

Registration and Protocol

This review was registered with PROSPERO (ID: CRD42021231112) and is available at: https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=231112. Contrary to the original protocol, risk of bias assessment was not undertaken due to the unexpected volume and variability of studies included. In addition, after screening and data extraction, we removed studies screened in with the following criterion: “including alcohol in the outcome (e.g. alcohol-use was a secondary outcome of the study in addition to suicide).” Studies with alcohol-related suicide deaths and/or attempts as an outcome, etc. were still included.

Search Strategy

PubMed, Web of Science Core Collection, and Embase were searched in Fall 2020 for relevant articles. The search strategy used search terms related to alcohol, suicide, prevention and policies, and only included studies published in English, from 1990 to 2020. A primary search for general articles discussing alcohol and suicide prevention was conducted, followed by searches specific to alcohol policies and suicide. The search terms used for each search are shown in Table 1. Articles were imported into Covidence for deduplication, followed by the title and abstract screening.

Table 1. Search strategy.

Main search
General search	((alcohol OR ethanol) AND suicide AND  (policy OR policies OR intervention OR prevent*))
Alcohol policy searches
Alcohol taxation & pricing	((alcohol OR ethanol) AND suicide AND (pricing OR price* OR tax OR taxation OR taxes))
Minimum legal drinking age	((“Minimum Legal Drinking Age” OR “Drinking Age”) AND suicide)
Alcohol availability	(((alcohol AND outlet*) OR (“Alcohol availability”) OR (“availability of alcohol”)) AND suicide)
Alcohol marketing	((“alcohol marketing” OR “alcohol advertising”) AND suicide)
Alcohol control	(“alcohol control” AND suicide)
Wet & dry policies	(wet AND dry AND suicide)

Preliminary Screening & Full Text Review

Identified articles from database searches (n = 5293) were imported for title/abstract screening. After duplicates (n = 2726) were removed, n = 2567 studies were independently screened by one reviewer against the inclusion and exclusion criteria (see Table 2). In addition, n = 27 articles identified from previous reviews or reference sections of relevant articles were imported into the full-text review. A total of n = 402 studies underwent full-text review by one reviewer. In a small number of cases, a reviewer sought the opinion of a 2^nd reviewer if it was unclear whether the study met the inclusion criteria. A total of n = 69 publications meeting the inclusion criteria underwent data extraction and were ultimately included in our review. Figure 1 displays the PRISMA flow chart for this study.

Figure 1. PRISMA flow chart for this rapid review. *Studies where alcohol use was only a secondary outcome of the study and was not part of the intervention, population or the suicide outcome (e.g. alcohol-related suicide), were screened out after data extraction. **Other reasons for exclusion included: Study protocol without data (n = 3), article not written in English (n = 1), intervention is not relevant (n = 1), study was retracted (n = 1), unable to differentiate alcohol from other drugs (n = 1).

Table 2. Inclusion and exclusion criteria.

Inclusion criteria

Exclusion criteria

Original peer reviewed research A policy, community intervention, clinical intervention or other prevention strategy is evaluated Alcohol is included in the design of the study through one of the following: (a) Including alcohol in the intervention (e.g. an alcohol policy) (b) Including alcohol in the outcome (e.g. alcohol use was a secondary outcome of the study in addition to suicide*; alcohol-related suicide deaths and/or attempts as an outcome, etc.) (c) Including alcohol in the study population (e.g. study population was patients with alcohol use disorder, etc.) Includes any quantitative measure of suicide death or attempt as the outcome variable Written in English Study design is one of the following (a) A randomized controlled experiment (b) A quasi-experimental design (c) Ecological study Published between 1990 and Fall 2020

Review, systematic review, meta-analysis commentary, case study, letter, news article No intervention, policy or structural-level change was evaluated Alcohol is not included in the intervention, in any outcome, nor is it a defining feature of the study population. Studies where alcohol is only measured at baseline and is not a defining feature of the study population were excluded. Does not include a quantitative measure of suicide death or attempt as an outcome. Studies that only described deliberate self-harm or parasuicide, without specifying suicidal intent, were excluded. Not written in English Study design does not provide comparisons pre/post intervention, between intervention groups or between populations Published before Jan 1, 1990

*Studies screened in with this criterion were ultimately excluded after data extraction was completed.

Data Extraction

Data extraction for the 69 publications included was conducted in Microsoft Excel. The following variables were extracted independently by one reviewer: Geography, period start/end, study design, data source(s), specific population(s) [including age, sex, race/ethnicity], intervention(s), sample size, main outcome(s), populations and/or types of suicide most likely to benefit from the intervention. After data extraction, we grouped studies according to intervention types broadly accepted in public health – policy interventions, community-based interventions, and clinical interventions, which target structural, community and individual levels respectively. We then summarized findings within each intervention type.

Results

Overview

We examined the full texts of 69 publications that met our criteria. We grouped studies according to intervention type, and then again, within intervention type when a diverse range of interventions was present. The results are presented in three main groups: those focusing on policies (49 publications which include policy interventions applied at a structural level, including alcohol control campaigns, local restrictions, pricing and taxation, physical availability, and minimum legal drinking age; also see Appendix A); those focusing on community-based interventions (7 publications which include an intervention implemented at a group level or in the community, such as group counselling or education; also see Appendix B), and those focusing on clinical interventions (13 publications which include an intervention that involves one-to-one care, such as medication or individual counselling; also see Appendix C). Studies were also grouped into subgroups according to their best fit, although this grouping was imperfect, as some studies had multiple foci.

Policy Interventions

Alcohol campaigns and multiple policy interventions

Several papers focused on population-level interventions, such as prohibition and alcohol regulation in the USA, and anti-alcohol campaigns in the former USSR. Wasserman (1992) concluded that a decline in alcohol consumption related to prohibition in the US was associated with a decline in suicide deaths (estimated OLS coefficient = 2.25, p < 0.05). A recent analysis by Law and Marks (2020) focused on the fraction of the state population that lived in an area where alcohol was prohibited (i.e. a dry area) per year. Their estimates suggest that prohibition was associated with a reduction in suicide deaths by 2.5–3.7 per 100,000 (p < 0.10). Markowitz et al. (2003) estimated the impact of alcohol policies on youth suicide in the US. They concluded that a 10% increase in excise tax on beer was associated with a statistically significant 2.4–5% decrease in male suicide deaths, and a 10% increase in alcohol outlets was associated with a statistically significant, 1% increase in male suicide deaths among those aged 15–19. They also concluded that 0.08 BAC and zero tolerance drunk-driving laws were associated with a decrease in male suicide deaths by 6.5% and 7.7% respectively for those aged 20–24. Finally, zero tolerance laws were also associated with a 10.9% reduction in female suicide deaths among those aged 15–19. Choi et al. (2020) conducted a longitudinal, ecological study of US states from 2012 to 2016 and examined the effects of an index of alcohol regulation, including laws related to alcohol availability, taxation, and mandatory server training, on firearm-related suicide mortality. They found that greater alcohol regulation was associated with a significant decrease in firearm-related suicide mortality rates (β = −0.43 to −0.44 (depending on model), p < 0.001).

Nemtsov (2003) examined a Russian campaign to control alcohol, beginning in 1985 which included a 62% cut-back in authorized sales of liquor and powerful anti-alcohol pressure from the authorities. Following the campaign, there was a sharp decline in suicide deaths, from 37.9 to 23.1 per 100,000, a 39.1% decrease. Wasserman et al. (1994) focused on the same intervention in USSR republics with high alcohol poisoning mortality and found that the male suicide death rate in 1986 (after the campaign) was 41% lower than that in 1984 (before the campaign), with notable regional variation. A study focusing on suicide deaths among females in USSR republics (Wasserman et al., 1998) found that after restrictive alcohol policies were enacted, female suicide deaths declined by approximately 20% in Slavic and Baltic republics. Similar general conclusions for both males and females were drawn by Razvodovsky (2007, 2009) who focused on Belarus, and Värnik et al. (2007) in their examination of Estonia. Specifically, Värnik et al. (2007) noted a sharp decline in blood alcohol concentration (BAC) positive suicide deaths, a 39.2% decline (p = 0.012) for males and a 41.4% decline (p = 0.075) for females, during the campaign in Estonia.

Sauliune et al. (2012) examined policy developments in Lithuania from 2007 to 2009 including increased alcohol taxes, reduced selling times, and stronger controls on drunk driving. Nevertheless, there was a 16.3% increase in Years of Potential Lives Lost (YPLL) due to alcohol-related suicide among males, possibly influenced by the economic recession and the rise in unemployment. Pridemore et al. (2013) examined the effect of laws implemented in 2006 that regulated the production and sale of alcohol products in Russia. They found a statistically significant and permanent 9.2% reduction (Ln ω₀ = −0.096; p = 0.01) in the monthly number of male suicide deaths. They estimated that the policy saved about 4,000 male lives per year that would otherwise be lost to suicide.

In a study in Slovenia, Pridemore and Snowden (2009) found a 10% statistically significant (ω₀ = −3.62; p = 0.033) decrease in suicide deaths among males after the government made 18 the minimum legal drinking age and limited when and where alcohol could be purchased.

A retrospective study by Zupanc et al. (2013) focused on the same 2003 Act in Slovenia which included the restriction of alcohol advertising and sales at given locations (e.g. educational institutions), as well as the prohibition of selling alcohol to those under 18 or intoxicated individuals. They found no significant differences in the proportion of suicide deaths that were BAC-positive in the three periods studied 2000–2002, 2003, 2004–2007 (i.e. before, during and after the act’s implementation). However, the authors found that the mean BAC level of suicide decedents was significantly higher in the period before the act’s implementation (2000–2002) than after (2004–2007) (p < 0.05).

Lester (1993) examined states in the U.S. according to differences in their level of restrictions on access to alcohol as of 1973, and found that restrictions, such as high state alcohol taxes, illegal alcohol seizures, dry counties, and decreased alcohol outlets, were significantly associated with lower suicide mortality rates (r = − 0.30, p < 0.05).

In contrast to the literature from the United States, former USSR, Russia and Eastern Europe, which points to a general association of lower suicide mortality with more restrictive alcohol policies, Andreasson et al. (2006) considered the large number of changes in Sweden’s alcohol management since joining the European Union that increased access to alcohol, including the elimination of alcohol monopolies and removing limits on private imports. They concluded that suicide deaths show a decreasing trend instead of the expected increase.

Local restrictions on alcohol sales

Three papers were identified that focused on local alcohol restrictions in Alaskan settings. Berman et al. (2000) examined 135 Alaskan Native villages, grouped according to the level of restrictions on access to alcohol, with “dry” restrictions allowing no alcohol sales, “damp” allowing limited sales, and “wet” having the most access. The authors compared those with a “dry” or “damp” policy at some point between 1980 and 1993, to those that remained “wet” during this time period. The authors found that less restrictive measures (i.e. “damp” measures) were associated with a significant, 56 per 100,000, decrease in suicide deaths (p < 0.05) compared to “wet” measures. Wood and Gruenewald (2006) focused on Alaskan Native villages between 1991 and 2000 and examined the effects of the level of local prohibition. Both fatal and non-fatal self-harm injury rates were not significantly different for “wet” versus “dry” isolated Alaska Native villages. A subsequent study by Berman (2014) examined local alcohol control in 178 small Alaskan Native communities between 1980 and 2007. Suicide death rates among males aged 15–34 were higher in communities prohibiting alcohol importation, but the effect was non-significant after controlling for other factors. Finally, Joubert (1994) considered wet and dry counties in Alabama during 1978–1988 and found that dry counties had higher mean suicide rates than wet counties (p < 0.05).

Alcohol pricing/taxation

Six studies examined the impact of alcohol pricing and/or taxation on suicide. Skog (1993) found a 19% decline in suicide deaths after a dramatic increase in alcohol prices in Denmark during 1916–1917. In particular, there was a 56% decline among suicide deaths where “alcohol abuse” was recorded (“alcoholic” suicide deaths). In Russia between 2000 and 2015, Razvodovsky (2019) found a strong negative association between retail prices of all alcoholic beverages and suicide death rates among both sexes (correlations ranging from r = −0.92 to −0.96, all at a significance of p < 0.000). Male suicide deaths decreased by a factor of 2.3 from 70.3 to 30.0 per 100,000, and female suicide deaths by a factor of 1.9 from 10.6 to 5.7 per 100,000.

Sloan et al. (1994) examined the effect of alcohol prices in the USA between 1982 and 1988 and found alcohol prices had a statistically significant, negative association with suicide deaths with time and state variables included (regression coefficient: −0.039 to −0.040 (depending on model), p < 0.05). In contrast, Son and Topyan (2011) concluded that the effects of spirits and beer excise taxes in the United States from 1995 to 2004 on suicide mortality were statistically insignificant; however, the coefficient for real wine tax was negative and statistically significant (regression coefficient: −1.220, p < 0.05).

Two studies found a positive relationship between alcohol pricing/taxation and suicide. Yamasaki et al. (2005) found that male suicide deaths in Switzerland increased between 1965 and 1994 which coincided with a higher tax on alcohol (partial regression coefficient: 0.042, p < 0.001). Furthermore, Puljula et al. (2013) reported a 50% decline in alcohol-related suicide deaths specifically among those with traumatic brain injury after a reduction of alcohol taxes in a region of Finland.

Physical availability of alcohol

Eleven studies observed the relationship between the physical availability of alcohol and suicide. Ramstedt (2002) found that hospitalizations due to attempted suicide stopped increasing, a non-significant change, following medium strength beer in Sweden being taxed as strong beer and no longer being allowed for sale in grocery stores. In Iceland, permitting the sale of strong beer in 1989, which resulted in decreased consumption of spirits, was associated with a statistically significant decline in total (15.08 to 11.98 per 100,000 per year; one-tailed p < 0.05) and female (7.40 to 4.44; p < 0.05) suicide deaths, but not for male deaths alone (Lester, 1999a).

Lester (1995) assessed an index of availability of alcohol, and the association between suicide rates and alcohol consumption for each US state. Alcohol consumption was positively related to higher availability. Although alcohol consumption was positively and significantly associated with suicide rates, the effect of alcohol availability was non-significant. In an examination of six US states where wine was removed from sale in state monopolies, Lester (1999b) found an increase in suicide deaths in four states (ranging from 0.52 to 1.50 mean increase per 100,000 per year) but two had a decrease in their suicide rate (ranging from −0.44 to −1.56).

Zalcman and Mann (2007) found that privatization of alcohol sales in Alberta, which involved a transition from government-run retail stores to privately owned alcohol retail outlets, with a higher density of outlets and longer hours of sale, was associated with a statistically significant, abrupt, and continuing increase in both male and female suicide mortality rates (p < 0.01, see Appendix A for coefficients).

Johnson et al. (2009) examined 581 zip codes in California between 1995 and 2000. The density of both local bar and off-premise alcohol outlets was significantly associated with increased suicide mortality rates (b = 7.600, p = 0.016; b = 3.547, p = 0.022, respectively), while only the density of local bars was significantly associated with increased suicide attempts (b = 8.157, p = 0.004). Both suicide deaths and suicide attempts were lower in zip code areas with greater local restaurant densities (b = −3.348, p = 0.003; b = −2.205, p = 0.029, respectively). Escobedo and Ortiz (2002) found that with a one unit increase in the rate of liquor outlet density, the rate of suicide deaths per 100,000 increased significantly (p = 0.03) by 0.23 in New Mexico from 1990 to 1994. A study by Giesbrecht et al. (2015) of 14 US states, using data from 2003 to 2011, found that greater on- and off-premise alcohol outlet density per capita was related to the presence of alcohol in male suicide decedents (AOR = 1.05 and 1.08, respectively, p < 0.01). They also found that on-premise density was positively associated with a BAC level of 0.08 or greater among male suicide decedents (AOR = 1.05, p < 0.01).

Malaga et al. (2012), focused on two districts in Peru: La Victoria had a reduction in hours of sale during certain days of the week, while the neighboring Cercado district did not. In La Victoria suicide deaths declined by 35%, whereas Cercado experienced a 13% increase in suicide deaths, but the difference was non-significant. Wilkins et al. (2019) focused on Australia, Canada and the US and concluded that alcohol outlet density was not significantly associated with suicide deaths. Similarly, Branas et al. (2011) found that proximity to an on-premise or off-premise alcohol outlet in Philadelphia was not significantly related to intentional self-inflicted gun injuries or deaths.

Minimum legal drinking age and other under-age drinking policies

A number of studies focused on the minimum legal drinking age (MLDA). In this group, we also included a study by Carpenter (2004) which examined state-level variation in the adoption of zero tolerance (i.e. zero blood alcohol content) drunk driving laws for drivers under 21 years. Two main approaches to assessing MLDA effects are evident: either the impact on suicide is assessed when a jurisdiction changes the legal drinking age, or data on suicide is linked to when persons in a jurisdiction reach the legal drinking age.

Callaghan, Sanches, Gatley, et al. (2013) found a significant 51.8% increase (p = 0.01) in suicide deaths related to alcohol among youth who crossed the MLDA threshold in Ontario (i.e. age 19). Among young women, the increase was 91.8% (p < 0.01). Callaghan, Sanches, & Gatley (2013) also compared in-patient admissions for self-inflicted injuries before and after the MLDA in Canadian provinces, excluding Quebec. They report that the rate of admissions for self-inflicted injuries rose approximately 10% at the MLDA threshold in contrast to the pre-MLDA baseline rate (p = 0.029).

Comparing across states, Carpenter and Dobkin (2011) found that with an increase in the share of youth legally allowed to drink, there was a statistically significant 10% increase in suicide deaths (p < 0.01). Their regression-discontinuity model also demonstrated a 20.3% increase in suicide deaths at the MLDA of 21 (p < 0.01).

An earlier study by Carpenter (2004) considered state-level variation in the adoption of drunk driving laws that set strict legal blood alcohol limits for drivers under age 21 (i.e. zero tolerance laws). The author found that slightly older males were unaffected by these laws and there was no consistent effect for female suicide deaths. However, there was a 10.3% (p < 0.10) and 7.7% (p < 0.05) decline in suicide deaths among males aged 15–17 and 18–20 years respectively.

Jones et al. (1992) compared the impact of various legal drinking ages across 50 US states and the District of Columbia on fatal injuries among adolescents and young adults. Among persons of a given age who could drink legally, the suicide rate was 9.7% greater than among persons of the same age who could not drink legally (OR = 1.097; p < 0.05). Birckmayer and Hemenway (1999) assessed the impact of minimum drinking age laws on suicide deaths among youth aged 15–23. They concluded, holding other factors constant, that states with a lower MLDA had 8% higher suicide rates among those aged 18 to 20 (p < 0.01).

Carpenter and Dobkin (2009) report the suicide death rate increased by more than 16% at age 21, the MLDA in the US. Grucza et al. (2012) examined exposure to being legally able to drink. Women exposed to a lower MLDA had a significant 12% increase in suicide deaths during 1990–2004 (OR = 1.12, p = 0.0003). The effect sizes of suicide mortality based on MLDA exposure were stronger for those born after 1960 and significant for women.

However, in a study of five states (New Jersey, New York, Arizona, Wisconsin and Texas), Carpenter and Dobkin (2017) found no significant increase in suicide attempts between 1990 and 2010 related to the MLDA. Matsubayashi and Yoshikawa (2018) report on the impact of an MLDA of 20 in Japan and found that while the rate of suicide mortality did increase between the age of 18 and 22, it was uninterrupted at the 20-year age point.

Modelling studies

Modelling studies can be used to synthesize evidence to estimate the impact of interventions on health outcomes in a population. Two modeling studies were identified that focused on alcohol policies and suicide. Holder et al. (1995) considered Norway and Sweden and the partial or complete elimination of the alcohol monopoly, and a low to substantial drop in alcohol prices as a result of private competition. Depending on the scenario, they projected an increase of 6% to 57% in suicide deaths. Norstrom et al. (2010) examined the potential effects of replacing the retail alcohol monopoly with a private license system in Sweden. The authors analyzed two possible outcomes: (1) replacing the current alcohol retail monopoly with private licensed stores that specialize in alcohol sales, or (2) making all alcohol available in grocery stores. For scenario one they expected an increase in suicide of 15.0% for men and 11.9% for women, and for scenario two they expected an increase of 36.3% for men and 28.1% for women.

Community Based Interventions

Seven papers were identified that considered community-based interventions. Two focused on American Indian/Alaska Native populations. Kelley et al. (2017) assessed peer recovery support and found at the six-month follow-up, there was no significant difference in suicide attempts or binge drinking compared to baseline; however, a significant decline in alcohol use in the past 30 days was noted. Cwik et al. (2016) assessed the impact of the White Mountain Apache Suicide Surveillance and Prevention System, a multi-faceted suicide prevention program that included dissemination of resources as well as educational workshops such as Applied Suicide Intervention Skills training. Programming also included brief interventions for youth who were in the surveillance system due to binge drinking. They reported that the age-adjusted suicide death rate dropped by 38.3% between 2001–2006 and 2007–2012, and suicide attempts decreased by 53.3% from 75 (in 2007) to 35 (in 2012) individuals, of which about one-half were using alcohol at the time of the attempt.

Brown and Block (2001) implemented Project Chrysalis, a comprehensive school-based program that provided a supportive network of services among four cohorts of adolescent females with a history of trauma. The authors reported a reduction in suicidal behaviors, including suicide attempts in both the intervention and control groups, with a greater difference in the intervention group (significance unreported). Furthermore, those who attended support groups had decreased alcohol use.

Fekkes et al. (2016) conducted a cluster randomized controlled trial of students in the Dutch Skills for Life program. Although significantly fewer students in the experimental group reported using alcohol compared to the control group; there were no significant differences in suicide attempts.

In Brazil, Machado et al. (2018) assessed the impact of a network of community mental health care units (CAPS), which provides resources for persons with psychiatric disorders including alcohol problems. During the five-year study period, there was no significant difference in suicide death rates in cities with or without CAPS; however, increased CAPS coverage was associated with significant decreases in hospitalizations for alcohol problems (RR = 0.88, p < 0.05) and suicide attempts (RR = 0.89, p < 0.05).

Finally, Mann et al. (2006) considered Alcoholics Anonymous (AA) membership in Ontario between 1968 and 1991. They found that both total and female suicide rates were negatively related to AA membership (b = −0.002, p < 0.05; b = −0.003, p < 0.10, respectively). Mann et al. (2008) drew similar conclusions in their study of AA membership and suicide mortality in Manitoba between 1976 and 1997 (see Appendix B).

Clinical Interventions

Twelve clinical studies (13 publications) were identified in our review. Most of these studies (n = 9) were randomized control trials (RCTs).

Interventions focusing on suicidal populations with alcohol use

Three studies in total focused specifically on treating people experiencing suicidality who were also using alcohol. McManama O'Brien et al. (2018) developed the Alcohol and Suicide Intervention for Suicidal Teens (ASIST), which involves individual and family sessions addressing the alcohol/suicide relationship via motivational interviewing. In their sample of 50 adolescents, they found no significant differences in suicide attempts, days of alcohol use and quantity of alcohol use between the control and intervention groups.

Esposito-Smythers et al. (2011) compared an integrated outpatient cognitive behavioral intervention (I-CBT) to enhanced treatment as usual (E-TAU) in n = 40 suicidal adolescents with an alcohol or cannabis use disorder. The I-CBT group had significantly fewer suicide attempts (5.3% vs 35.3%, p = 0.023) and heavy drinking days compared to E-TAU, but there were no differences in the number of drinking days. Morley et al. (2014) compared an opportunistic cognitive behavioral intervention (OCB) to treatment as usual in a sample of 185 adults with alcohol and/or drug misuse and either a history of suicide attempt or current suicidal ideation. Regardless of the treatment group, they found significant decreases in suicide-related outcomes, including attempts (6-month reattempt rate of 2.7%), and a significant decrease in alcohol consumption over time.

Interventions focused on treating alcohol use disorder (AUD)/related conditions

Six studies in total examined interventions focused on treating AUD and/or related conditions, with suicide attempt or death as non-primary outcomes. Five used therapy-based interventions, while one studied pharmaceutical interventions.

Therapy-Based interventions

Nadkarni et al. (2017) recruited a sample of 378 adult men in India scoring 12–19 on the Alcohol Use Disorders Identification Test (AUDIT). Their RCT compared enhanced usual care (EUC) to EUC plus counselling for alcohol problems (CAP), which uses motivational interviewing. Although the EUC plus CAP group had a significantly greater proportion of individuals in remission and individuals with no drinking in the past 14 days, compared to EUC alone, there were no significant differences in suicide attempts between the two groups. Kristenson et al. (2002) compared brief intervention to a control in their sample of 667 middle-aged men with gamma-glutamyl transferase levels in the highest decentile. Of the alcohol-related deaths, there was one suicide death, and it was in the intervention group.

Timko et al. (2019) compared enhanced telephone monitoring (ETM), which links patients to addiction treatment, plus usual care (UC), versus UC alone among 298 psychiatric patients undergoing detoxification for alcohol and/or opioid dependence. At 3 months, the ETM group had significantly fewer number of drinking days and lower addiction severity, but these effects lost significance at the 6-month time point. There were no significant differences in the number of suicide attempts at any time point in the study.

Research by Gregory et al. (2008, 2010) compared dynamic deconstructive psychotherapy (DDP) to treatment as usual (TAU)/optimized community care (OCC) in adults with both borderline personality disorder and active “alcohol abuse” or dependence. This research found that suicide attempts were significantly less frequent in the DDP group compared to the TAU/OCC group during the 6–12 month treatment period (p = 0.013). In fact, throughout the study period, there were no reported suicide attempts in the DDP group. There was one suicide death in the TAU group at 9 months. The authors also found significant improvements in alcohol-related outcomes in the DDP group (namely “alcohol misuse” and heavy drinking behavior).

van den Bosch et al. (2005) compared Dialectical Behavior Therapy (DBT) vs TAU in a sample of 58 women with borderline personality disorder, clinically referred from addiction treatment and/or psychiatric services. A significantly greater decrease in alcohol consumption, sustained for the first 6 months post-treatment cessation, was seen in the DBT group; however, no statistically significant difference in suicide attempt frequency was detected between groups.

Pharmaceutical interventions

Laaksonen et al. (2008) completed an RCT comparing Naltrexone, Disulfiram or Acamprosate in 243 alcohol-dependent patients. In the 52-week study period, only one suicide death occurred; it was in the Acamprosate group. Disulfiram was significantly better than Naltrexone and Acamprosate in a variety of alcohol outcomes from week 1 to 12; however, at weeks 13–52, some significance was lost.

Ecological studies examining a clinical intervention

Three studies took an ecological approach to examine a clinical intervention. Moustgaard et al. (2014) examined the correlations between antidepressant sales and both alcohol and non-alcohol-related suicide deaths. Sales of antidepressants were not associated with alcohol-related suicide deaths among males or females. Nettelbladt et al. (2007) examined the effects of the introduction of tricyclic antidepressants in 1962 in Sweden. Over the course of the period studied, a shift from depressive disorders to other psychiatric disorders and AUD (alone, not comorbid) associated with suicide occurred. Bellanger et al. (2007) found alcohol withdrawal prescription rates were positively correlated with suicide deaths in females in France.

Discussion

Our analysis of 69 publications exploring the prevention of alcohol-related suicide provides some insights into the existing evidence, quality of data and underdeveloped topics in the field. Our findings come from many different study designs, including ecological studies, community-based interventions, clinical interventions and RCTs. Over half of the studies pertained to alcohol policies and their impact on suicide deaths or attempts; the remaining were clinical or community-based interventions.

Policy interventions

There is evidence that population-level interventions such as the prohibition in the United States (e.g. Wasserman, 1992) or an alcohol control campaign in Russia (Nemtsov, 2003; Wasserman et al., 1994, 1998) are associated with reduced suicide rates. However, “wet” versus “dry” regional or local policies were not necessarily linked with suicide rates (e.g. Berman, 2014). This discrepancy may be explained by consumers in “dry” areas travelling to nearby “wet” counties or villages to purchase alcohol, which is only possible in situations of local (as opposed to national) alcohol restrictions (Joubert, 1994). The discrepancy may also be because other factors, not analyzed, had an impact, such as variations in disposable income, baseline drinking habits or demographic characteristics. Alcohol taxes or prices were generally inversely related to suicide rates (Razvodovsky, 2019; Skog, 1993; Sloan et al., 1994); however, this was not always the case (Puljula et al., 2013). Inconsistent results may be due to variations in disposable income and social status.

A number of studies assessed the impact of physical availability of alcohol on suicide. A general conclusion is that higher physical availability of alcohol, including higher alcohol outlet density, is associated with higher suicide rates, including alcohol-positive suicide deaths (Escobedo & Ortiz, 2002; Giesbrecht et al., 2015; Johnson et al., 2009; Zalcman & Mann, 2007).

The studies on minimum legal drinking age (MLDA) focused either on when a jurisdiction raised or lowered the MLDA, or what happened to suicide rates when a cohort passed the legal age to consume alcohol. In both cases, most of the evidence indicates that higher MLDAs appear to be associated with fewer suicide deaths, and when measured, alcohol-related suicide deaths (Callaghan et al., 2013; 2013). Studies using models to predict the effects of alcohol policies on suicide found that both a lower price on alcohol as well as greater access via privatization of alcohol sales was linked with a projected increase in suicide mortality (Norstrom et al., 2010).

There is also recent research, which was published after we concluded our search, which is highly relevant. Coleman et al. (2021) found that in the United States more restrictive alcohol and firearm policies were associated with lower rates and odds of suicide death involving alcohol and firearms.

While these policy changes can be considered an intervention, they were generally not specifically implemented to reduce suicide deaths. Furthermore, there was not always a comparison or control population; although in many cases, time series analyses were used in order to facilitate a comparison of suicide deaths before and after a policy was implemented. Relatively few studies had biological data on the presence of alcohol at the time of death (e.g. Giesbrecht et al., 2015; Razvodovsky, 2007; Zupanc et al., 2013) or alcohol dependence history. Thus, it is not feasible, in many studies, to assess if the alcohol control policy had a greater impact on former or current heavy drinkers who died by suicide, than on light drinkers or abstainers.

In some cases, alcohol policies did not have the expected effect. For example, in Lithuania, during the period of a comprehensive alcohol control policy, years of potential life lost (YPLL) per 100,000 due to alcohol-related suicide increased by 16.3% among males while the YPLL rate was relatively stable for females (Sauliune et al., 2012). This increase may represent confounding by gendered effects of economic conditions and other factors.

There are various reasons why an alcohol policy is implemented. A reduction in suicide deaths may be a public health benefit not necessarily planned for by the architects of these policy modifications. However, these studies provide a strong rationale for considering population-level alcohol policies as an important resource for preventing suicide and other harms related to alcohol consumption.

Community interventions

There were several papers that focused on community interventions in the United States and documented alcohol use among those who attempted or died by suicide. Two studies focused on Indigenous populations. While one study found some degree of impact on suicide deaths from their intervention (e.g. Cwik et al., 2016) the other did not (e.g. Kelley et al., 2017). This may be due to the fact that suicide is a relatively rare outcome. Studies may be underpowered, and subsequent systematic reviews and meta-analyses may allow for the pooling of results to detect significant effects. Other community-based interventions showed reductions in suicidal behavior and, in one study, alcohol use as well (Brown & Block, 2001; Fekkes et al., 2016).

Very few community-based studies were identified, and only two were RCTs. Alcohol use was noted, but it was typically not a central focus. Further work seems warranted, including, for example, research assessing the impact of multi-dimensional and more intensive interventions. Studies should include an adequate comparison group and follow-up with multiple data points, and with both self-report and biological data.

Clinical interventions

Although our review found several studies examining clinical interventions to prevent alcohol-related suicide, studies were often limited by small sample sizes. Because both suicide attempts and suicide deaths are relatively rare outcomes, small sample sizes can make it difficult to detect differences among treatment groups.

Clinical interventions that directly address the relationship between alcohol and suicide seem to be particularly effective in the prevention of suicide. Two studies that used cognitive behavioral interventions in high suicide risk populations with AUD/related conditions demonstrated significant decreases in suicide attempts and certain alcohol measures compared to treatment as usual (Esposito-Smythers et al., 2011; Morley et al., 2014). In addition, we detected one intervention that addressed the relationship between alcohol and suicide that was not restricted to a population with alcohol use disorder. McManama O'Brien et al. (2018), found non-significant decreases in suicide attempts for suicidal adolescents receiving the Alcohol and Suicide Intervention for Suicidal Teens (n = 25), compared to a control (n = 25). Their results suggest that interventions targeting the relationship between alcohol and suicide among those who drink alcohol and have suicidality may hold promise in preventing suicide attempts, but larger, high-powered studies are needed.

In contrast, there is less evidence to support interventions that primarily address alcohol use, with suicide attempts or deaths as non-primary outcomes (or as an adverse event). Of all studies in this category, only one had a significant decrease in suicide attempts (Gregory et al., 2008, 2010) and it was among those with active “alcohol abuse”/dependence and borderline personality disorder (BPD). BPD is associated with a very high rate of suicide attempts (3 lifetime suicide attempts on average) and suicide deaths (up to 10% of BPD patients die by suicide) (Paris, 2019), which may explain the significant findings of this study. In the other studies that primarily addressed alcohol use, measures of alcohol use demonstrated significant reductions while measures of suicide were non-significantly reduced in treatment groups. Our review did not locate any studies that found equal or increased suicide measures in the treatment groups. This suggests that these studies may be under-powered to detect differences in suicide-related outcomes, and future studies should use greater sample sizes.

Finally, ecological studies of population levels of an indicator of a clinical intervention, namely antidepressant use, found that in general, antidepressant sales and usage were not associated with changes in alcohol-related suicide deaths (Moustgaard et al., 2014; Nettelbladt et al., 2007). Bellanger et al. (2007) found a positive association between alcohol withdrawal prescription rates and suicide deaths in females in France. Such studies are difficult to interpret as it is not clear whether medication prescription rates are a proxy for disease prevalence or the proportion of those being treated for a disease.

Limitations

Our review has several limitations. Because only one reviewer was involved in each stage of the process, our rapid review is not as rigorous as a systematic review. However, the authors erred on the side of inclusion in their selection of abstracts. They also communicated on potential candidate articles if there was uncertainty. Another limitation is the limited time period of the review; we focused on publications from 1990 to the Fall of 2020, and therefore, our review, while broad in scope, did not include publications over thirty years ago. Some of these were captured in previously published reviews (e.g. Hurzeler et al., 2021; Kolves et al., 2020; Xuan et al., 2016;) and, as noted in our analysis, there has been a substantial body of evidence accumulated on this topic in the most recent 30-year period. Because of a large number of studies included, we did not conduct a risk of bias assessment. Results should be interpreted with this in mind, and future systematic reviews on specific types of interventions should be conducted, including the risk of bias assessment on included studies. We did not stratify study results by gender; however, both alcohol use and suicide are highly gendered behaviours. Future systematic reviews should stratify results by gender and sex, when possible. Finally, we acknowledge the challenge with classifying deaths as suicide, as different countries and jurisdictions may have varying guidelines about how coroners and medical examiners can determine a death to be by suicide.

Implications for prevention and research

Alcohol is involved in suicide attempts and death via various pathways, including access to alcohol, alcohol dependence, use in combination with other drugs, and binge drinking. This is supported by our review findings, in that in studies where alcohol use measures were available, the direction of alcohol use and suicide-related outcomes were most often in concordance (e.g. Värnik et al., 2007; Gregory et al., 2008, 2010, etc.). In order to prevent alcohol-related suicide, alcohol needs to be considered as a proximal and distal risk factor. At the international, national and local levels, wide-reaching, multi-level, effective programs hold promise in preventing alcohol-related suicide.

It would be very beneficial to implement comprehensive drug and alcohol screening among suicide decedents. This may allow governments, non-governmental organizations, and researchers to track alcohol and other drug use at the time of death and ensure that it is feasible to keep track of any trends in relation to these factors and suicide deaths. It is worth noting, however, that blood alcohol concentration levels do not necessarily reflect a decedent’s mental status at the time of death, due to complicating factors such as tolerance seen among certain alcohol consumers.

This review has shown that alcohol control policies can effectively reduce suicide-related harms due to their wide scope and potency. Governments may be reluctant to implement the most effective alcohol controls (e.g. Giesbrecht et al., 2016; Vallance et al., 2021), such as increased taxation or decreased physical availability, because these may be unpopular with the electorate. However, as the evidence grows on the chronic, acute and social harms from alcohol, there may be further public support for evidence-based actions.

Furthermore, every time there is a change in a local, regional or national alcohol policy, this provides an opportunity for a high-quality natural experiment to assess the impact on suicide and other harms. If feasible, these projects should seek alcohol-related data on the decedents, such as treatment history for alcohol use disorder and BAC level at the time of death; variables that were largely ignored, or not available, in many of the policy studies included in our review.

In addition to policies, community-based initiatives and clinical interventions oriented to alcohol-related suicide show potential promise as key suicide prevention strategies. However, such interventions have not been adequately explored and most clinical interventions included alcohol as a secondary consideration, not as an integral part of a prevention strategy. Future studies would benefit from quasi-experimental and RCT designs, with sufficient sample size to detect differences between groups.

At the individual level there appear to be several intervention pathways to consider to prevent alcohol-related suicide deaths. Our review suggests that clinical interventions targeting the interaction between alcohol use and suicide are most likely to be effective. Such interventions could be completed through either one-on-one or group counselling. Screening, brief intervention and referral (SBIR) interventions could be considered to detect high-risk alcohol use among those experiencing suicidality. Likewise, a suicide prevention module in SBIR would make counselling on suicide prevention available to heavy drinkers. Such screening could allow healthcare professionals to identify individuals with high-risk alcohol use and suicidal ideation much earlier and provide them with the appropriate care (such as the synergistic interventions described above) to reduce their alcohol use as well as their suicide risk at an earlier stage.

Furthermore, community and clinical interventions for suicidal populations should include alcohol use education as a component of the intervention, even if the participants do not engage in heavy drinking (e.g. McManama O'Brien et al., 2018). Finally, policies that reduce access to alcohol are an important step to reducing means to suicide, use of alcohol just prior to the event, and the rate of alcohol dependence in a population.

Countries should consider addressing alcohol as both a proximal and distal contributing factor in national suicide prevention strategies (Conner et al., 2014). State, provincial and local governments and organizations can use national recognition of alcohol as a risk factor to guide more local interventions. Such a strategy should support further research at the policy, community and clinical levels to assess which interventions are most effective given the needs of the population. Tracking the intervention's effectiveness could, in turn, promote further policy support.

Overall, our review indicates that despite significant evidence of alcohol’s distal and proximal role in suicide attempts and deaths, there are very few suicide prevention strategies and interventions that intentionally and thoughtfully address alcohol use. Future suicide prevention initiatives could address this either through the target population (e.g. people experiencing alcohol use disorder, high-risk drinkers, etc.) or the intervention itself (i.e. addressing alcohol use as a risk factor in patients at high risk of suicide). If the suicide prevention community is to make strides in preventing suicide deaths or attempts, suicide prevention strategies must address the critically important risk factor of alcohol use.

Acknowledgements

We wish to thank Eva Fernandes for her contributions to the revision of this manuscript. Short sections of this paper were presented at the Alcohol Use & Suicide Webinar, November 12, 2020, organized by the Mental Health Commission of Canada, and at the Issues of Substance Conference, Canadian Centre on Substance Abuse, November 23–25, 2021. Work on this manuscript was made possible by funding received from the Public Health Agency of Canada (PHAC).

Disclosure statement

The authors report no conflict of interest.

Additional information

Funding

Funding for this study was received from the Public Health Agency of Canada, contract no. 4500414332

Notes

1 Refers to the number of participants randomized in randomized controlled trials. Sample sizes not reported for non-RCT studies.

Appendix A.

Studies focusing on alcohol policy interventions – summary of characteristics, intervention and outcome

 

Table

Author, year	Location, year(s)	Suicide attempt or death; Role of alcohol	Design	Study population	Intervention	Main outcomes [both suicide and alcohol outcomes, if relevant]
ALCOHOL CAMPAIGNS & MULTIPLE INTERVENTIONS
Wasserman, 1992	US, 1910–1920	Deaths; Intervention and outcome	Time series	General population	Prohibition	Decrease in access to alcohol and associated decline in alcohol consumption was associated with a reduction in national suicide deaths (estimated OLS coefficient = 2.25, p < 0.05).
Law and Marks (2020)	US, 1900–1920	Deaths; Intervention and outcome	Quasi-experiment with regression analysis	General population	Prohibition	Prohibition was associated with significantly fewer deaths due to alcoholism; however, deaths rates due to suicide did not fall significantly as state-years became drier. Prohibition was estimated to reduce suicide deaths by 2.5–3.7 per 100,000 (p < 0.10).
Markowitz et al., 2003	US, 1976–1999	Deaths; intervention	Negative binomial regression	General population aged 10–24 years	A variety of laws were examined including beer taxation, percent of the population living in dry counties (for beer), alcohol outlet density, and drunk-driving prevention laws.	A 10% increase in tax on beer was associated with a statistically significant 2.4–5% decrease in male suicide deaths (depending on age group). A 10% increase in alcohol availability was associated with a statistically significant 1% increase in male suicide deaths among those aged 15–19 years. Among males aged 20–24, 0.08 BAC and zero-tolerance drunk driving laws were associated with a significant decrease in suicide deaths by 6.5% and 7.7% respectively. Zero tolerance laws were also associated with a 10.9% decrease in female suicide deaths among those aged 15–19.
Choi et al., 2020	US, 2012–2016	Deaths; intervention	Longitudinal, ecological	General population	Index of alcohol regulation, including laws related to alcohol availability, taxation, and mandatory server training.	More alcohol regulation was associated with significantly fewer firearm-related suicide deaths (β = −0.43 to −0.44 (depending on model), p < 0.001).
Nemtsov, 2003	Russia, 1965–1999	Deaths; intervention and outcome	Time series	General population aged 15 to 85+	Campaign launched in June 1985, that had a 62% cut-back in authorized sale of liquor and anti-alcoholic pressure from authorities.	Suicide deaths declined sharply from 37.9 to 23.1 per 100,000 (a 39.1% decrease) from 1984 to 1986, after the initiation of an anti-alcohol campaign. There was also a decline in alcohol consumption. After this time period, suicide deaths increased slowly as alcohol consumption increased.
Wasserman et al., 1994	USSR, 1984–1990	Deaths; intervention and outcome	Time series	General population of males	Restrictive alcohol policy enacted in June 1985 during perestroika. Decrease in alcohol production and retail outlets for alcohol.	Both alcohol consumption and suicide mortality declined (e.g. in 1986 the suicide rate was 41% lower than that in 1984), with notable regional variation. In most republics, more than 50% of suicide deaths were attributable to alcohol.
Wasserman et al., 1998	USSR, 1984–1990	Deaths; Intervention and outcome	Time series	General population of females	Restrictive alcohol policy enacted in June 1985 during perestroika. Decrease in alcohol production and retail outlets for alcohol.	Suicide deaths among females decreased by approximately 20% in the study period. In the whole USSR, alcohol explained on average 27% of the suicide deaths.
Razvodovsky, 2007, 2009	Belarus, 1979–2007	Death; intervention, outcome	Time series (ARIMA)	General population	Anti-alcohol campaign of 1985–1988 limiting availability and manufacturing of alcohol.	From 1984 to 1986, BAC-positive suicide deaths declined by 54.2%, while BAC-negative suicide deaths declined by 7.1%. There was a statistically significant association between alcohol poisoning mortality and suicide mortality.
Vaarnik et al. (2007)	Estonia, 1981–1992	Deaths; intervention, outcome	Retrospective study	General population	Anti-alcohol campaign of 1985, including restrictions on production, sales, raising taxes, propaganda, and penalties.	During the anti-alcohol campaign, there was a 39.2% (p = 0.012) and 41.4% (p = 0.075) decline in BAC positive suicide deaths among males and females respectively. During the same time, the annual per capita alcohol consumption declined from 10.9 to 6.6.
Sauliune et al., 2012	Lithuania, 2006–2009	Deaths; intervention, outcome	Time series	General population aged 15–64 years	Multiple changes in alcohol policy from 2007–2009, including banning daytime TV and radio alcohol advertisements, raising excise taxes, reducing selling hours, and greater controls on drunk driving.	The Years of Potential Life Lost as a result of alcohol-related suicide increased by 16.3% among males, despite the campaign. The overall suicide rate also tended to increase during the 2006–2009 study period.
Pridemore et al., 2013	Russia, 2000–2010	Deaths; intervention	Time series (ARIMA)	General Population aged 15+ years	January 2006 law regulating production and sale of alcohol products, with resultant increasing retail prices.	The ARIMA analysis revealed a statistically significant, immediate and permanent, 9.2% decrease (Ln ω0 = −0.096; p = 0.01) in male suicide deaths due to the policy implementation, with no significant effect on female suicide deaths.
Pridemore & Snowden, 2009	Slovenia, 1997–2005	Deaths; intervention	Interrupted time series (ARIMA)	General population	January 2003 Act which set a minimum legal drinking age of 18, limited where alcohol could be sold, limited hours of sale, mandated workplaces be alcohol free and no longer allowed alcohol in vending machines.	The 2003 policy had no significant association with suicide deaths in females, but it was associated with a significant, 10% reduction (ω0 = −3.62; p = 0.033) in suicide deaths among males compared to preintervention levels.
Zupanc et. al., 2013	Central and Northwest Slovenia, 2000–2007	Deaths; intervention, outcome	Time series	Suicide decedents who underwent autopsy	January 2003 Act, including the prohibition of selling of alcohol to and drinking of alcohol among intoxicated individuals and those under age 18. The Act also included restrictions on alcohol availability and advertising at certain locations (e.g. educational institutions).	There were no significant differences in the proportion of suicide deaths that were BAC-positive in the three periods studied 2000–2002, 2003, 2004–2007 (i.e. before, during and after the act’s implementation). However, the mean BAC level of suicide decedents was significantly (p < 0.05) higher in the period before the act’s implementation (2000–2002) than after (2004–2007).
Lester 1993	US, 1973	Deaths; intervention	Ecological study	General population	State variation in alcohol availability measures including proportion of the population living under prohibition, state taxes on alcohol, retail outlets per capita, minimum purchasing age, and other measures.	Correlation analysis showed suicide rates were significantly and negatively associated with restrictions on alcohol (r = −0.30, p < 0.05). However, a measure of liberal availability (as opposed to restriction) of alcohol was not as strongly associated with suicide rates.
Andreasson et al., 2006	Sweden, 1994–2002	Deaths; intervention, outcome	Time series (ARIMA)	General population	Changes in Sweden’s alcohol policy since joining EU. Monopolies on alcohol were eliminated in 1995, and limits on the private import of alcohol were essentially removed in 2004.	Over the course of the period studied, suicide deaths showed a decrease, while estimated total alcohol consumption showed an increase.
LOCAL RESTRICTIONS ON ALCOHOL SALES
Berman et al., 2000	US (Alaska), 1980–1993	Deaths; intervention	Quasi-experimental	Alaskan Natives	Comparing villages that adopted a state local option law restricting alcohol (i.e. changed from being “wet” to being “dry” or “damp”) compared to villages that did not adopt such law (i.e. remained “wet”) during the study period.	There was no decrease in suicide deaths in areas with more restrictive measures (i.e. “dry” communities). However, there was a statistically significant decrease in suicide deaths in villages adopting less restrictive alcohol controls (i.e. “damp” communities), compared to “wet” villages (a 56 per 100,000 decrease, p < 0.05).
Wood & Gruenewald, 2006	US (Alaska), 1991–2000	Deaths and attempts; intervention	Retrospective study	General population aged 15+ years	Alaska Native Village-level variation in local alcohol prohibition. 68 villages were always “dry,” 53 always “wet,” and 11 changed policies at least once during the study period.	There was no significant difference in the age-adjusted rate of intentional self-harm injury for wet versus dry Alaska Native villages.
Berman, 2014	US (Alaska), 1980–2007	Deaths; intervention	Poisson regression	Alaska Native males aged 15–34 years	Variation in local alcohol control, including dry status and other local alcohol regulations, among 178 rural Alaska communities.	Suicide rates were higher in communities that prohibited the importation of alcohol, but the effect was not significant after controlling for other community-based factors.
Joubert, 1994	US (Alabama), 1978–1988	Deaths; intervention	Quasi-experimental	General population	Local prohibition of alcohol, comparing “dry” to “wet” counties.	Dry counties had significantly higher mean suicide rates than wet counties (p < 0.05).
ALCOHOL PRICING/TAXATION
Skog, 1993	Denmark, 1911–1924	Deaths; intervention, outcome	Time series	General population	Increase in alcohol taxation in 1916–1917 due to shortage of alcohol in World War I.	Suicide deaths declined by 19% overall from 1911–1915 to 1916–1920. This decline was particularly profound (56% decline) for suicide deaths where “alcohol abuse” was recorded (“alcoholic” suicide deaths). There was no notable change for suicide deaths not involving “alcohol abuse.” There was also a decrease in per capita alcohol consumption, with a r = 0.95 correlation between alcohol consumption and “alcoholic” suicide deaths. For every litre that per capita alcohol consumption decreased, total suicide mortality decreased 2.5%.
Razvodovsky, 2019	Russia, 2000–2015	Deaths; intervention	Time series	General population	Between 2000 and 2015, the mean retail price of vodka increased by 6.7 times, wine by 3 times, and beer by 3 times.	During the study period, the male suicide rate decreased by a factor of 2.3 from 70.3 to 30.0 per 100,000; the female rate decreased by a factor of 1.9, from 10.6 to 5.7 per 100,000. There were significant negative correlations between the price of all types of alcohol beverages and suicide deaths in both males and females (correlations ranging from r = −0.92 to −0.96, all at a significance of p < 0.000), with suicide deaths being more sensitive to vodka pricing than beer or wine pricing.
Sloan et al., 1994	US, 1982–1988	Deaths; intervention, outcome	Time series	General population aged 25–64 years	State- and year-based variation in alcohol pricing.	Alcohol prices had a statistically significant, negative association with suicide deaths when time and state variables were included (regression coefficient: −0.039 to −0.040 (depending on model), p < 0.05).
Son & Topyan, 2011	US, 1995–2004	Deaths; intervention, outcome	Time series	General population aged 25–64 years	State-level excise tases on wine, beer, and spirits.	The associations between excise taxes on spirits and beer and suicide rate were not statistically significant. However, the association between excise taxes on wine and suicide rates was negative and statistically significant (regression coefficient: −1.220, p < 0.05). During the study period, real tax rates for alcohol declined significantly, with overall alcohol consumption per person increasing by 2.4% (from 3.40 to 3.48 gallons).
Yamasaki et al., 2005	Switzerland, 1965–1994	Deaths; intervention	Time series	General population	Changes in taxes on alcohol over time, as measured by the sum of specific taxes on alcohol products.	Taxes on alcohol were significantly and positively related to the male age-standardized suicide rate (partial regression coefficient: 0.042, p < 0.001), with no effect on female suicide rates.
Puljula et al., 2013	Finland (Northern Ostrobothia), 1999–2007	Deaths; intervention, outcome	Logistic regression	Individuals with acute moderate-to-severe traumatic brain injury and/or those who died on scene	Alcohol taxes were reduced by one third and EU traveller allowances increased in 2004.	Alcohol consumption increased 25%, however alcohol-related suicide deaths that involved traumatic brain injury were 50% less frequent in 2007 vs 1999.
PHYSICAL AVAILABILITY OF ALCOHOL
Ramstedt, 2002	Sweden, 1973–1981	Attempts; intervention, outcome	Time series (ARIMA)	General population aged 10+ years	From July 1977 medium beer was taxed as strong beer and only allowed for sale in 300 retail alcohol monopoly stores, with no one under 20 years permitted to purchase.	There was a non-significant, weak trend for hospitalizations due to attempted suicide to stop increasing as medium strength beer was removed. Hospitalizations due to alcoholism, alcohol psychosis or alcohol intoxication significantly declined only among those aged 10–19 years as medium strength beer was removed.
Lester (1999a)	Iceland, 1984–1994	Deaths; intervention	Ecological	General population	After March 1989, Iceland permitted strong beer to be sold. Other studies reported this policy change was associated with decreased consumption of spirits.	With the allowance of strong beer, the overall suicide rate had a statistically significant decline from 15.08 to 11.98 per 100,000 per year (one- tailed p < 0.05). The female suicide rate declined significantly from 7.40 to 4.44 (p < 0.05), but the decrease for the male suicide rate was non-significant.
Lester (1995)	US, 1974	Deaths; intervention, outcome	Ecological	General population	State-level variation in an index of alcohol availability. The index includes MLDA, limitations on off- and on-premise sales, outlet density, and other limitations on sales.	A positive association was noted between the index of alcohol availability and alcohol consumption. Suicide rates were strongly and positively correlated with alcohol consumption (Pearson r = 0.43, p < 0.01), and positively, but not significantly, associated with the alcohol availability index (Pearson r = 0.19).
Lester (1999b)	US (Idaho, Iowa, Maine, Montana, New Hampshire, and West Virginia), unspecified year	Deaths; intervention	Ecological	General population	Retail state monopolies on wine sales were removed, which was associated with increased wine consumption.	After the monopolies on wine sales were removed, four states had an increase in the suicide rate (mean increase per 100,000 per year of 1.24 in Iowa, 1.50 in Idaho, 1.06 in Maine, and 0.52 in West Virginia), while two had a decrease in the suicide rate (−0.44 in Montana and −1.56 in New Hampshire)
Zalcman & Mann, 2007	Alberta, 1976–1999	Deaths; intervention	Time series (ARIMA)	General population aged 15+ years	Privatization of alcohol sales with three stages. Stage 1 (1985): Privately owned wine stores were opened. Stage 2 (1989): Privately owned beer stores were opened. Spirits and wine were sold in hotels in rural areas. Stage 3 (1994): All liquor stores were privatized.	Stage 1 and 2 were associated with statistically significant (p < 0.01) increased suicide rates for both males (ω = 0.42 and ω = 0.16, respectively) and females (ω = 0.30 and ω = 0.41). Stage 3 was associated with a significant increase in the suicide rate for males (ω = 0.17, p < 0.01), but not females.
Johnson et al., 2009	US (California), 1995–2000	Deaths and attempts; intervention	Ecological	General population aged 15+ years	The number of active alcohol outlets per zip code, including off-premise outlets, restaurants, and bars/pubs.	Densities of local bars and off-premise outlets were significantly and positively related to suicide deaths (b = 7.600, p = 0.016; b = 3.547, p = 0.022, respectively), while local restaurant densities had a significant, negative association with suicide deaths (b= −3.348, p = 0.003). Densities of local bars were significantly and positively associated with suicide attempts (b = 8.157, p = 0.004), while local restaurants had a significant, negative association with suicide attempts (b= −2.205, p = 0.029). Local off-premise densities had no significant association with suicide attempts.
Escobedo & Ortiz, 2002	US (New Mexico), 1990–1994	Deaths; intervention, outcome	Linear regression	General population	Liquor outlet density.	There was a statistically significant, positive association between liquor outlet density and suicide rates (adjusted regression coefficient: 0.23 (p = 0.03)). Suicide rates were 80% and 60% higher for the second and third tertiles of alcohol outlet density, respectively, compared to the first tertile.
Giesbrecht et al., 2015	US (14 states), 2003–2011	Deaths; intervention, outcome	Multi-level logistic regression models	General population	County-level variations in on and off-premise alcohol outlet density.	Among male suicide decedents, both on- and off-premise alcohol outlet density were positively associated with blood alcohol content positivity (AOR = 1.05, p < 0.01 and AOR = 1.08, p < 0.01, respectively), while only on-premise alcohol outlet density was associated with BAC ≥ 0.08 (AOR = 1.05, p < 0.01) in adjusted models. There was also a significant interaction between off-premise alcohol outlet density and American Indian/Alaska Native race (AOR = 1.36, p < 0.01).
Malaga et al., 2012	Peru (Lima), 2005–2008	Deaths; intervention, outcome	Ecological	General population	Reduction in hours of sale of alcohol in La Victoria district in Lima in January 2007. A nearby district, Cercado de Lima, not undergoing such policy change, was used as the control group.	Suicide deaths in La Victoria declined by 35% compared to a 13% increase in Cercado, but this difference was non-significant. Despite the decline in suicide deaths in La Victoria, the proportion of suicide decedents positive for alcohol increased.
Wilkins et al., 2019	Australia, Canada & US, 2011	Deaths; Intervention	Ecological	General population	State/province-level variation in alcohol outlet density.	Alcohol outlet density was not significantly associated with suicide deaths.
Branas et al., 2011	US (Philadelphia), 2003–2006	Deaths and attempts; Intervention	Case-control	General population aged 21+ years	Density of on- and off-premise alcohol outlets.	There was no statistically significant relationship between alcohol outlet density and intentional self-inflicted gun injuries, including gun suicide deaths.
MINIMUM LEGAL DRINKING AGE (MLDA) AND OTHER UNDERAGE DRINKING POLICIES
Callaghan, Sanches, Gatley, et al., 2013	Canada (Ontario), 2002–2007	Deaths; intervention, outcome	Regression- discontinuity analysis	General population aged 16–22 years	Minimum legal drinking age of 19 in Ontario.	Compared to those slightly below the MLDA, individuals just older than the MLDA showed a significant, 51.8% (p = 0.01) increase in emergency department visits classified as suicide deaths related to alcohol. Among young women crossing the MLDA threshold, there was a statistically significant, 91.8% increase in such suicide deaths (p < 0.01).
Callaghan, Sanches, & Gatley, 2013	Canada (excluding Quebec), 1997–2007	Attempt; intervention	Regression-discontinuity analysis	General population aged 15–22 years	Minimum legal drinking age threshold for each jurisdiction.	At the MLDA threshold, the rate of hospital admission for self-inflicted injuries rose significantly by 9.6% (p = 0.029), compared to the hospitalization rate just before the MLDA.
Carpenter & Dobkin, 2011	US, 1975–2003	Deaths; intervention, outcome	Fixed effects and regression discontinuity	General population aged 15–29 years	Minimum legal drinking age.	The fixed effects model estimated that an increase in the proportion of young adults who can legally drink was associated with a statistically significant, 10% increase in suicide deaths (p < 0.01). The regression-discontinuity model estimated that suicide deaths increased by 20.3% at the MLDA of 21, which was statistically significant (p < 0.01). The proportion of young adults who could drink legally was also associated with significant increases in a variety of measures of alcohol consumption.
Carpenter, 2004	US, 1981–1998	Deaths; intervention	Negative binomial regression	General population aged 15–29 years	State-level variation in adoption of zero tolerance (i.e. zero blood alcohol content) drunk driving laws for drivers under 21 years.	Adoption of zero tolerance laws was associated with a 10.3% decline in male suicide deaths among those aged 15–17 (p < 0.10) and a 7.7% decline among males aged 18–20 years (p < 0.05). There was no significant effect noted for older males or female suicide deaths.
Jones et al., 1992	US, 1979–1984	Deaths; intervention	Logistic regression	General population aged 15–24 years	State-level variation in legal drinking age.	At a given age, suicide death rates were 9.7% higher among those who could legally drink as compared to those who could not. This difference was statistically significant (OR = 1.097; p < 0.05).
Birckmayer & Hemenway, 1999	US states (48 states), 1970–1990	Deaths; intervention	Time series	General population aged 15–23 years	Raising of the minimum legal drinking age.	When controlling for other factors, states with an MLDA of 18 years had an 8% higher suicide rate among 18–20 year-olds than states with a 21-year MLDA. This difference was significant (p < 0.01).
Carpenter & Dobkin, 2009	US, 1997–2004	Deaths; intervention, outcome	Regression discontinuity	General population aged 19–22 years	Minimum legal drinking age.	At the MLDA, age 21, the suicide death rate increased by more than 16%. Measures of alcohol consumption also increased at the MLDA, including a 21% increase in the number of recent drinking days.
Grucza et al., 2012	US, 1990–2004	Deaths; intervention	Quasi-experimental	General population of individuals born between 1949 and 1972	Minimum legal drinking age laws, specifically whether individuals were legally permitted to consume alcohol before age 21.	Being exposed to a lower MLDA was associated with a significant, 12% increased risk of suicide death among women (OR = 1.12, p = 0.0003). For men, there was borderline significance for a decreased risk of suicide death with a more permissive MLDA.
Carpenter & Dobkin, 2017	US (Arizona, New Jersey, New York, Texas, Wisconsin), 1990–2010	Attempts; intervention	Regression discontinuity	General population aged 19–22 years	Minimum legal drinking age.	There were no significant increases in self-inflicted injuries in relation to the MLDA.
Matsubayashi & Yoshikawa, 2018	Japan, 1974–2014	Deaths; intervention, outcome	Regression discontinuity	General population aged 18–21 years	Minimum legal drinking age of 20 years.	The rate of suicide deaths did increase between 18 and 22 years of age, but it was uninterrupted at 20-year age point.
MODELLING STUDIES
Holder et al., 1995	Sweden & Norway, n/a	Deaths; Intervention and outcome	Multiplicative modelling	General population	Partial or complete elimination of the national alcohol retail monopolies, as well as low to substantial drops in alcohol prices.	All scenarios examined were projected to lead to increased per capita levels of alcohol consumption. Suicide deaths in Sweden were projected to increase by 100 (9%) to 638 (57%) for men, and 30 (7%) to 170 (37%) for women at 1 and 5 L increases in per capita alcohol consumption, respectively. In Norway, the projected increase was from 50 (10%) to 280 (57%) deaths for men, and 10 (6%) to 40 (23%) deaths for women, for 1 and 5 L per capita increase in consumption, respectively.
Norstrom et al. (2010)	Sweden, n/a	Deaths; intervention and outcome	Multiplicative modelling	General population	The effects of two scenarios were projected. The first scenario involved replacing the present retail monopoly with private stores that primarily sell alcohol. The second involved making alcohol available in grocery stores.	Scenario 1 was projected to yield a 17% and scenario 2 a 37.4% increase in alcohol consumption, respectively. Under the conditions of the first scenario, there was a projected 15.0% increase for men and 11.9% increase for women in suicide deaths. Under the second scenario, there was a projected 36.3% increase for men, and 28.1% increase for women in suicide deaths.

Appendix B.

Studies focusing on community-based interventions – summary of characteristics, intervention and outcome

 

Table

Author, year	Location, year(s)	Suicide attempt or death; Role of alcohol	Design	Study population	Intervention	Main outcomes [both suicide and alcohol outcomes, if relevant]
Kelley et al., 2017	US, 2013–2016	Attempts; population, intervention and outcome	Bivariate analyses	American Indian individuals (mostly from Northern Plains Tribes) with substance use, recruited from tribal chemical dependency programs, social service agencies, tribal health programs or through self-referral.	Transitional Recovery and Culture Program, which uses Peer Recovery Support (PRS) to enhance recovery from substance use and related disorders.	From baseline to 6 month follow up, there was no significant difference in suicide attempts not due to alcohol or drug use. There were significant decreases in past 30-day alcohol use.
Cwik et al., 2016	US (Arizona), 2001–2012	Attempts and deaths; intervention and outcome	Ecological	White Mountain Apache Tribe	White Mountain Apache Suicide Surveillance and Prevention System, a comprehensive suicide prevention intervention, including dissemination of information on resources and hotlines, suicide prevention conferences, and educational workshops such as the Applied Suicide Intervention Skills Training (ASIST) and life skills curriculum. A brief intervention program was also implemented for youth with suicidal ideation and attempts, as well as youth who were in the surveillance system for binge drinking.	The Apache suicide death rate declined from 40.0 to 24.7 per 100,000, a 38.3% decrease. Annual suicide attempts also declined from 75 to 35, a 53.3% decrease.
Brown & Block, 2001	US (Portland, Oregon), 1995–2000	Attempts; intervention and outcome	Randomized controlled trial	Adolescent females aged 14–17 years with a history of being a victim of sexual, emotional or physical abuse and having at least two other psychosocial risk factors	Project Chrysalis, a multi-faceted intervention that primarily consisted of school-based support groups facilitated by school counsellors and local therapists. The program also included educational sessions about alcohol and other drug prevention.	Both the intervention and control group had declines in suicide attempts at the first follow up point compared to baseline, however this difference was greater among those in the Project Chrysalis program. The significance of these differences was not reported. At the second follow up time point, there was a significant, negative association between time in case management and the likelihood of trying alcohol.
Fekkes et al., 2016	Netherlands; unspecified year	Attempts; intervention and outcome	Cluster randomized controlled trial	Students aged 13–16 years	Skills for Life program, a classroom-based, two-year intervention consisting of 25 lessons designed to enhance social, emotional and moral skills. As part of the intervention, skills are also developed for situations involving substance use.	Significantly fewer students in the experimental group reported using alcohol compared to the control group, but there were no significant differences in the risk of past year suicide attempts.
Machado et al., 2018	Brazil, 2008–2012	Attempts and deaths; intervention and outcome	Mixed ecological design	General population	Municipal coverage of a network of community mental health care units (CAPS) which are the primary health care resource for people with persistent, severe mental health disorders. One of the sub-types of CAPS, CAPS I, which are the smallest units, have a patient population that includes those with disorders related to alcohol use.	Increased CAPS coverage was associated with significantly fewer hospitalizations for both attempted suicide (RR = 0.89, p < 0.05) and alcohol problems (RR = 0.88, p < 0.05), but not for suicide deaths.
Mann et al., 2006	Canada (Ontario), 1968–1991	Deaths; intervention	Time series (ARIMA)	General population aged 15+ years	Alcoholics Anonymous (AA) membership per 100,000 adults aged 20 years and older	AA membership rates had a significant, negative association with overall (b = −0.002, p < 0.05) and female (b = −0.003, p < 0.10) suicide death rates.
Mann et al., 2008	Canada (Manitoba), 1976–1997	Deaths; intervention	Time series (ARIMA)	General population aged 15+ years	Alcoholics Anonymous (AA) membership per 100,000 adults aged 20 years and older	There was a significant, negative association between the rate of AA membership and the rate of suicide deaths overall (b = −0.04, p = 0.02) and among females (b = −0.02, p = 0.01).

Appendix C.

Studies focusing on clinical-based interventions – summary of characteristics, intervention and outcomes

 

 

 

 

 

 

Table

Author, year	Location, year(s)	Suicide attempt or death; R ole of alcohol	Design	Study Population	Intervention; sample size^1	Main outcomes [both suicide and alcohol outcomes, if relevant]
INTERVENTIONS FOCUSING ON SUICIDAL POPULATIONS WITH ALCOHOL USE
McManama O'Brien et al., 2018	US; 2014–2018	Attempt; population, intervention & outcome	RCT	Adolescents hospitalized following suicide attempt or ideation, endorsing past month alcohol use	Alcohol and Suicide Intervention for Suicidal Teens (ASIST) [individual and family counselling using motivational interviewing] vs control; n = 50	Suicide attempts were non-significantly less frequent in the intervention group compared to the control. Days of alcohol use and quantity of alcohol use were also non-significantly lower in the intervention group compared to the control group, after controlling for baseline.
Esposito-Smythers et al., 2011	US; 2004–2007	Attempt; population, intervention & outcome	RCT	Adolescents at risk of suicide with comorbid alcohol and/or cannabis use disorder	Integrated outpatient cognitive behavioral (I-CBT) intervention vs enhanced treatment as usual (E-TAU); n = 40	The I-CBT group had significantly fewer suicide attempts over 18 months (5.3% vs 35.3%, p = 0.023) and significantly fewer heavy drinking days relative to E-TAU.
Morley et al., 2014	Australia; unspecified	Attempt; population, intervention & outcome	RCT	Adults with alcohol or other drug misuse and a suicide attempt in the last 3 months or current suicidal ideation	Opportunistic cognitive behavioural (OCB) intervention vs treatment as usual (TAU); n = 185	Substantial reductions in suicide attempts occurred in both the OCB group and the TAU group. Alcohol consumption also significantly decreased across the 6-month study period regardless of the treatment group.
INTERVENTIONS FOCUSED ON TREATING ALCOHOL USE DISORDER (AUD)/RELATED CONDITIONS
Therapy-Based interventions
Nadkarni et al., 2017	India; 2013–2016	Attempt; population, intervention & outcome	RCT	Adult men scoring 12–19 on AUDIT	Enhanced usual care (EUC) plus Counselling for Alcohol Problems (CAP) vs EUC alone; n = 378	Suicide attempts were non-significantly lower in the EUC plus CAP group. EUC plus CAP had a significantly higher proportion of individuals with no drinking and higher remission rates. EUC plus CAP also had non-significantly higher mean alcohol consumed among drinkers.
Kristenson et al., 2002	Sweden; 1975–1991	Death; population, intervention & outcome	RCT	Males aged 45–49 with heavy alcohol consumption and gamma-glutamyltransferase levels in the highest decile	Brief intervention vs control; n = 667	There were significantly fewer alcohol-related deaths in the intervention group. Among all alcohol-related deaths, there was one suicide death and it occurred in the intervention group.
Timko et al., 2019	USA; Unspecified	Attempt; population, intervention & outcome	RCT	Patients entering a psychiatry unit undergoing detoxification for alcohol and/or opioid dependence	Enhanced Telephone Monitoring (ETM) [connects patients to addiction treatment and/or mutual help] plus usual care (UC) vs UC alone; n = 298	Suicide attempts were non-significantly lower in the ETM group at 3-month follow up, and non-significantly higher at 6-month follow up. Both number of drinking days and alcohol addiction severity index were significantly decreased in the ETM group compared to UC alone at 3-month follow up, but significance was lost at 6-month follow up.
Gregory et al., 2008, 2010	USA; 2004–2008	Attempt & death; population, intervention & outcome	RCT	Adults with Borderline Personality Disorder and active alcohol abuse or dependence	Dynamic deconstructive psychotherapy (DDP) vs treatment as usual/optimized community care (TAU/OCC); n = 30	Suicide attempts were significantly less frequent in the DDP group compared to the TAU/OCC group during the 6 to 12-month treatment period (p = 0.013). Throughout the study period there were no reported suicide attempts in the DDP group. One participant died by suicide in the TAU group at 9 months. The DDP group also had significant improvement in heavy drinking behavior over the 30-month study period, and significantly less heavy drinking than the TAU/OCC group at 12 months.
van den Bosch et al., 2005	Netherlands; Unspecified	Attempt; population & outcome	RCT	Women with borderline personality disorder who were clinically referred from addiction treatment and psychiatric services	52 weeks of Dialectical Behavioural Therapy (DBT) vs treatment as usual (TAU); n = 58	Suicide attempts were lower in the group receiving DBT (effect approaching significance, p = 0.064). The DBT group also had a significantly greater decrease in alcohol consumption, sustained for 6-months after treatment cessation, relative to TAU.
Pharmaceutical interventions
Laaksonen et al., 2008	Finland; 2000–2005	Death; population, intervention & outcome	RCT	Patients with alcohol dependence	Randomization to either take Naltrexone, Disulfiram or Acamprosate; n = 243	Only one suicide death was observed, and it was in the Acamprosate group. On multiple alcohol measures, disulfiram was significantly better than naltrexone and acamprosate during weeks 1–12; this significance only persisted for certain measures from weeks 13 to 52.
ECOLOGICAL STUDIES EXAMINING A CLINICAL INTERVENTION
Moustgaard et al., 2014	Finland; 1995–2007	Death; outcome	Longitudinal Study	General population (adults)	Antidepressant usage (assessed via sales)	Although the increased proportion of antidepressant users may have prevented non-alcohol related suicide deaths among men, alcohol-related suicide deaths were not related to regional antidepressant sales.
Nettelbladt et al., 2007	Sweden; 1947–1997	Death; outcome	Prospective Cohort Study	General population	Introduction of tricyclic antidepressants in 1962 in Sweden	As antidepressants were introduced over the course of the study period, a shift from depressive disorders to other psychiatric disorders and alcohol use disorder (alone, not comorbid) associated with suicide occurred.
Bellanger et al., 2007	France, 1996–1999	Death; intervention	Ecological	General population	Alcohol withdrawal drug prescription rates	The prescription of drugs for alcohol withdrawal was positively associated with suicide mortality in females.

¹ Refers to the number of participants randomized in randomized controlled trials. Sample sizes not reported for non-RCT studies.