Impact of Alcohol Policies on Suicidal Behavior: A Systematic Literature Review

Alcohol consumption has been found to be related to suicidal behavior at the individual and population level, but there is lack of literature reviews on the effect of alcohol policies on suicidal behavior. Therefore, the aim of the current study is to conduct a systematic literature review of the impact of alcohol policies at the population level on suicidal behavior and ideation. We searched the Cochrane CENTRAL, Cochrane DARE, EMBASE, Medline, ProQuest, PsycINFO, PubMed, SCOPUS, and Web of Science electronic databases in March 2019. Papers analyzing alcohol policies limiting alcohol use and studying suicidal behaviors as an outcome measure were included; we identified 19 papers. Although the methods and effect sizes varied substantially in the studies, reducing alcohol often led to reduction in suicidal behavior. Ecological-level studies predominantly investigated the effect of restrictions on alcohol availability and increased cost of alcohol, and the majority presented a reduction in suicides across Western and Eastern Europe, as well as the US. The majority of studies were rated as unclear risk of bias for a number of domains due to a lack of clear reporting. Policies targeting harmful alcohol consumption may contribute towards a reduction in suicidal behavior at the population level.


Introduction
Alcohol use, both abuse and acute consumption, has been found to be an important contributing factor for suicidal behavior at the individual level [1][2][3][4]. A number of potential mechanisms have been proposed, including increases in impulsivity, aggression, depressive and suicidal thoughts, and feelings of hopelessness, especially if people are predisposed or have depression. In addition to the link on individual level, there is a relationship between alcohol consumption at the population (ecological) level and national suicide rates [1]. It has been suggested that the aggregated-level link between alcohol consumption and suicide depends on a drinking culture; the relationship is strong in predominantly spirit-consuming countries with binge-drinking and intoxication-oriented drinking patterns, rather than in wine-cultures such as Southern Europe [5,6], as has been further evidenced in a recent systematic literature review [1]. Overall, spirits consumption have been linked to higher levels of aggression, emotional responses, and confusion than other alcoholic beverages; therefore, potentially strengthening the relationship between acute alcohol intoxication with distress and negative emotions through constricted thinking and impulsivity to suicide [7].
Considering individual and aggregate level links between alcohol and suicidal behaviors, it would be logical to expect that alcohol policies limiting alcohol use in the population should have the potential to prevent also suicidal behavior [8]. However, there is limited discourse about the topic in suicide prevention. For example, Witt and Lubman [9] highlighted inadequate attention of alcohol and other drug use in Australian suicide prevention strategies. There is also a lack of systematic reviews on the impact of alcohol restrictions on suicidal behavior at the individual and aggregated levels. It has been also noted that participants with alcohol or substance abuse are not included in intervention studies or systematic reviews [9]. Nevertheless, we identified a critical review of alcohol policies, which was limited to suicide only within a restricted timeframe (i.e., 1999-2014) and did not follow PRISMA guidelines or estimate risk of bias [10]. Therefore, we aim to fill the gap and conduct a systematic literature review of the impact of alcohol policies at the population level on suicidal behavior by also addressing the limitations of the earlier critical review.

Materials and Methods
This systematic review followed the guidance contained in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [11].

Search Strategy
A comprehensive search of the Cochrane CENTRAL, Cochrane DARE, EMBASE, Medline, ProQuest, PsycINFO, PubMed, SCOPUS, and Web of Science electronic databases was conducted for English-language papers without date restriction until 19 March 2019 according to the protocol (Table 1).

Inclusion and Exclusion Criteria
Inclusion criteria required studies to report data on suicide and self-harm (encompassing both non-suicidal self-injury [NSSI] and/or suicide attempt).
Studies were excluded if they measured associations between alcohol use and suicidal outcomes without evaluating the effect of a specific policy [12][13][14][15]. Studies were also excluded to avoid duplication where the sample either partially or fully overlapped with that of other studies included in the review [16,17]. This exclusion criterion particularly affected studies on the effect of the anti-alcohol Perestroika campaign in individual states (countries) of the former Union of Soviet Socialist Republics (USSR) [18,19].
Studies were selected for inclusion using a two-stage process: (1) titles and abstracts were screened independently by KW and RW; (2) full texts were reviewed against the inclusion and exclusion criteria to determine eligibility by KW, RW, and KC independently. Disagreements were resolved through consensus discussions with KK.

Data Extraction and Synthesis
For each study, we extracted information on: (1) study information; (2) methodological characteristics, and (3) details of the intervention-alcohol policy approach. In categorizing the alcohol policies, we followed the World Health Organization's (WHO) recommended target areas for policy action at the national level, which have been proposed in WHO's global strategy to reduce the harmful use of alcohol [20].
Given that we anticipated few studies would report sufficient numerical data to enable meta-analysis, particularly with regards to mortality of suicide in the control (or historical) comparator condition, we elected to undertake a systematic review of results from these studies.

Risk of Bias Assessment
Risk of bias in the included studies was assessed using the Risk Of Bias In Non-randomized Studies of Interventions-of Exposures (ROBINS-E) tool [21].

Results
We located 10,881 records. A further 55 relevant studies were identified through ancestry-searching. Following de-duplication, 6519 individual records remained, with 6397 records being excluded after title and abstract screening. The remaining 122 full-text records were assessed for eligibility with 19 papers remaining in the final review ( Figure 1).
Nineteen papers investigated the effect of changes to alcohol policies on suicidal behavior. All studies were 'natural experiments' analyzing the impact of changes in alcohol policies on suicide rates using mainly time-series analysis with a before and after or a quasi-experimental design on the population-level ( Table 2). The majority of studies analyzed suicide mortality; only one used self-harm admissions to the general hospital as an outcome measure. Nine papers examined the effects of alcohol availability [22][23][24][25][26][27][28][29][30], seven changes to alcohol pricing [28,29,[31][32][33][34][35], three changes to drink-driving countermeasures [28,29,36] and four investigated the effects associated with change in alcohol policy including multiple measures [37][38][39][40]. Two papers analyzed different measures separately and are therefore included [28,29], whilst two other papers, despite reporting on the same intervention over similar time periods, presented data for different eligible outcomes (i.e., suicide [22] and non-fatal self-harm [26]). The majority of these studies were conducted in the US [22][23][24]26,28,29,31,35,36], followed by the USSR [39,40], Canada [30], Denmark [33], Lithuania [38], Russia [32], Slovenia [37], Switzerland [34], Sweden [27], and the UK [25]. Two papers about the same policy from the USSR were included as they were separated by gender [39,40]. Different data presentation and analysis methods did not enable meta-analysis and a narrative analysis following the WHO recommended target areas for policy action at the national level [20] was conducted.

Alcohol Availability
There were a variety of different policy components examined that specifically addressed the impact on suicidal behavior associated with various restrictions on alcohol availability in the form of enforcing minimum legal drinking age (MLDA), dram shop laws, restrictions on hours of trading, privatization, outlets, and complete alcohol bans.
A US study examining the MLDA on suicide rates compared states with a younger MLDA of 18 years to those with an older age requirement of 20-21 years [23]. The study compared youth suicide across the 48 states (1970-1990); states with younger MLDAs had 8% higher suicide rates amongst 18to 20-year olds and 6% higher rates in 21-to 23-year olds, even following adjustment for a number of indicators of socioeconomic disparity [23]. No significant effects were found for adolescents below the MLDA.
Findings from studies examining varying degrees of alcohol bans have been mixed. A study from the US found that implementing a 'dry' law (i.e., prohibiting the sale and importation of alcohol in the community) was associated with increased suicide rates [24]. However, another analysis using more complex modelling showed that a higher proportion of dry counties is associated with the lower level of suicides in males aged 20 to 24 [28]. Two more recent studies investigated the effect of differing restriction policies implemented across a number of Native Alaskan communities. Communities selected the option of adopting either: a 'dry law', a 'damp law' (i.e., less restrictive controls on the sale and/or importation of alcohol in the community) or a 'wet law' (i.e., no restrictions are placed on the sale and/or importation of alcohol in the community) [22]. Implementing either law was associated with a 10.3% reduction in suicide rates; however, the effect was greatest following the introduction of a 'damp law' as compared to a 'dry law' (63.0% versus 4.9%) [22]. There was no significant difference between communities in rates of hospitalized self-harm cases [26].
A study investigating the effect of liberalization of alcohol licensing laws in the form of extending trading hours for bars and public houses in Scotland found an increase in hospitalizations for self-poisoning with co-ingested alcohol in both genders [25]. Privatization of the alcohol stores in Canada saw an increase in male suicides [30]. However, an analysis of dram shop law in the US showed no effect on suicides in the age group of 25-64 years [29].

Alcohol Pricing
Changes to policy that have resulted in price changes have been investigated for all alcohol beverages and specific beverage types. The introduction of a 2006 law regulating the production and sale of ethyl alcohol in Russia through taxation resulted in an immediate reduction in rates of suicide in males, but not females [32]. After a dramatic increase in alcohol taxation during World War I, alcohol consumption in Denmark decreased as did the number of suicides [33]. This effect was pronounced in suicides of people with alcohol dependence.
Alcohol pricing/taxation was also found to be negatively correlated with suicides in the US [29]. Changes in the pricing of specific beverage types have also been associated with changes in suicide rates in the US. An increase in beer excise was associated with a reduction in suicides in young males aged 10 to 24 years; the effect on female suicide rates was negligible [28]. Another study reported a negative correlation with wine excise, but not with beer and spirits in the age group of 25 to 64 years. However, a study including six states found that the effect of removal of state-based retail monopolies on the sales of wine produced mixed effects [31]. Whilst four states experienced an increase in suicide rates following the removal of these monopolies (Idaho, Iowa, Maine, West Virginia), two experienced a decrease (Montana, New Hampshire) [31].
An opposite effect has been found in male suicides in Switzerland, but not for females [34]. In addition, a study on the changes to alcohol law after Sweden's entry to the European Union in 1995 found that decreased pricing/taxation was associated with decreases in suicide rates, with a greater effect on males [27].

Drink-Driving Countermeasures
Three studies examined the effect of drink-driving countermeasure on suicide rates, all from the US. Two studies analyzed different blood alcohol concentration (BAC) limits for young drivers [28,36] and for all drivers in the US [28]. One study found that adoption of 'zero tolerance' laws was associated with a reduction in suicide rates for youth aged 15-24 years. However, the authors report that these reductions were meaningful for males between 15 and 17 (10.3%) and for males between 18-20 years (7.7%) [36]. No meaningful effects were found for females, or for older age groups. Yet, the other study noted some effect on teenage girls (negative correlation; [28]). Another study analyzing mandatory jail terms for drinking under influences (DUI) showed no impact on suicide [29].

Mixed Policies
The remaining studies looked at the effects of change on overall alcohol policies in different countries, which incorporated several different components, including marketing restrictions, nationwide awareness-raising activities, leadership, health services response, addressing informal and illicit production, drink driving countermeasures, as well as alcohol pricing and availability.
Two papers investigated the effect of the introduction of a strict alcohol policy in 1985, alongside social changes as a result of Perestroika, on suicide rates in males [39] and females [40] in the former USSR. Restrictions included a major propaganda campaign with anti-alcohol advertising, a decrease in alcohol production, a decrease in the number of retail outlets for the sale of alcohol, time limits on sales, punishing alcohol misuse, criminalizing the production of home-distilled alcohol, and improvements in treatment [19,39,40]. Suicide rates of both sexes were positively correlated with alcohol consumption, which declined by 31.8% for males [39] and 19.3% for females [40] after the restrictions in alcohol were introduced in 1985. Similarly, a Slovenian study found that following the introduction of the 'Act Restricting the Use of Alcohol' in 2003, suicide rates immediately decreased by 10% amongst men, but there was no change to rates in women [37]. The Act included several measures, such as introducing a MLDA, restrictions on alcohol advertising, and reducing trading hours.
A more recent study from Lithuania found the opposite relationship: male suicide rate increased by 14.3% between 2006 and 2009, following the implementation of multiple measures, including regulations of advertising and alcohol availability, increased taxation, and drink driving countermeasures [38].  There was no effect on female suicides. Introduction of a law regulating the production and sale of ethyl alcohol and alcohol-containing products to control the availability of alcohol, and to require registration of alcohol production and distribution facilities. There was significant negative association between wine tax and suicide rate, but no association with beer or spirits tax.

Risk of Bias Assessment
Risk of bias assessment was conducted using ROBINS-E and is presented in Appendix A. The majority of studies were rated as unclear risk of bias for a number of domains due to a lack of clear reporting on exposure bias, confounding bias, baseline confounding, missing data, and selection bias. Few studies assessed and adjusted data where necessary, for temporality and seasonality, which has a major influence on suicide rates [42]. Even fewer adjusted for other influences on suicide rates, such as age, gender/sex, and socio-economic deprivation distributions. However, the assessment of risk of bias in these studies is complicated by the lack of clear guidance on evaluating bias in studies of exposures [43].

Discussion
This study systematically reviewed literature on the impact of alcohol policies on suicidal behavior and identified 19 relevant papers. The reviewed studies were 'natural experiments' analyzing mainly changes in alcohol policies and their effect on suicide rates using time-series analysis with a before and after design or a quasi-experimental design. Overall, the effect of societal changes in alcohol consumption through alcohol policies on suicidal behavior were studied: (1) by examining the effect of decreased access to alcohol (assumed to be associated with decreased alcohol consumption); and, (2) by examining the effect of increased access to alcohol (assumed to be associated with increased alcohol consumption). It is important to highlight the differences between these approaches because, while the assumed effect on alcohol consumption on both is clear and opposing, the underlying political purpose behind these two changes is vastly different. The intended effect of implementing more restrictive alcohol policies is to reduce alcohol-related harms in the community; conversely, relaxing alcohol laws are done for political/economic purposes.
The studies included here predominantly investigated the impact of restrictions on alcohol availability and increased cost of alcohol, and the majority of such studies found associations with reduced suicides across Western and Eastern Europe, as well as the US. Hence, while not specifically implemented as a suicide prevention strategy, the policy changes were associated with the intended effect of reducing a form of alcohol-related harm. Indeed, regulating pricing and availability of alcohol are considered as 'best buy' measures of an alcohol policy by the WHO, meaning they are effective, cost-effective, and feasible [20]. There were some studies that investigated changes in suicide rates associated with the introduction of an alcohol policy with multiple strategies. These studies are harder to disentangle with regards to individual strategies that may aid in reducing suicide rates. Two interventions were associated with reductions in suicide rates in the former USSR [39,40] and in Slovenia [37]. Despite analyses of the strict alcohol measures during Perestroika in the former USSR seeing over 30% decline in male suicide rates in Lithuania (as a part of the former USSR [39]), a more recent study from Lithuania found the opposite effect with increases in suicides between 2006 and 2009 after an anti-alcohol campaign [38]. However, while we need to consider the role of the hope-inspiring social and political climate at the time of Perestroika, [39,40], we cannot ignore the impact of the Global Financial Crisis (GFC) at the time of the more recent alcohol campaign in Lithuania. Rises in suicide rates were reported in many countries across the world between 2006 and 2009 [44]. Sauliune et al. [38] refer to increases in unemployment, which was not controlled for in the analysis.
The studies that examined the effect of increased alcohol availability and decreased cost did not yield as consistent a message. One Swedish study looked at a time where alcohol prices dropped after entering the EU and found a decrease in suicide rates [27]. When interpreting this study, we must consider wider societal changes associated with entering the EU that occurred alongside this increased consumption of alcohol, such as increased immigration and increased economic prosperity. Another study found mixed effects of decreases in wine production monopolies across six states in the US, with some states displaying an associated decrease in suicide rates and some an increase [31]. When assessing beverage-specific changes, alcohol cultures must be considered; in different societies, different types of alcohol are consumed in different patterns [5]. When interpreting the effects of decreased cost of wine on suicidal behavior, the specific culture surrounding wine consumption in each location must be considered. Nevertheless, liberalization of Scotland's liquor licensing laws was associated with an increase in hospitalized self-poisoning and an increased proportion of those admitted who had co-ingested alcohol at time of poisoning [25]. Similarly, in a three-stage privatization of alcohol sales in Alberta, Canada, each stage showed an increase in suicide rates, especially for males [30].
Across all studies, the anti-suicide effects associated with restricting alcohol use were predominant in males. This is unsurprising, given males are more likely than females to drink alcohol, develop alcohol dependence [20], and have positive BAC at time of suicide death and die by suicide [19,45]. Stronger effects in males also supports a potential causal link between the ecological associations found-if the associations between alcohol restrictions and suicide rates were spurious, we would not expect to see such a prominent difference between the sexes.
Young people are particularly susceptible to alcohol-related harm [20] and accordingly, youth suicides seem particularly amenable to alcohol policy changes such as drink-driving countermeasures and increasing the MLDA. However, studies have found significant increases in hospitalizations for both alcohol-use disorders and alcohol poisoning, as well as self-harm, as young people transition across the MLDA [46,47], suggesting that policies to increase the MLDA alone are unlikely to meaningfully reduce suicidal behavior across the age spectrum.
Worldwide, the incidence of both alcohol misuse and suicidal behavior [48] is higher amongst Indigenous peoples as compared to their non-Indigenous peers. Incorporating traditional beliefs into treatment may, therefore, represent an important first step in improving adherence and, through this, the effectiveness of treatments both for alcohol and other drug use problems [49] and suicidal behavior [50] within Indigenous populations.

Limitations and Future Directions
The studies included were 'natural experiments', utilizing mainly ecological level measures; therefore, they are vulnerable to the ecological fallacy. Notable differences in alcohol polices and their components limited quantitative synthesis, as numerical data on rates of suicidal behavior prior to the intervention period were frequently not reported. In addition, different types of analytical approaches were used with majority of the studies not adjusting for potential confounding factors (e.g., unemployment, income level). As a consequence, whilst our results point to the potential anti-suicide effect of policies to restrict alcohol use, particularly in males, further work is required to elucidate the mechanisms by which this effect may occur, and particularly the role that local alcohol consumption patterns may play.
Additionally, although we have categorized the intervention approaches adopted in the included studies according to the WHO's recommended guidelines [20], a number of studies were characterized by mixed interventions. This makes it difficult to establish which particular approach may be most effective in reducing rates of suicidal behavior and ideation at either the individual or population-level. The implementation of staged alcohol restriction policies with sufficient lag between each stage to assess suicide-related outcomes would help to identify approaches likely to be of greatest value in global suicide prevention efforts. However, given that over one-in-three coronial determinations for suicide deaths remain open beyond two years [51], the lag period required to ascertain the effect of staged interventions on suicide rates in particular would need to be considerable, highlighting the potential value of so-called 'real time' surveillance for these outcomes [52]. It is also important to note the impact of other societal changes coinciding with the campaigns (e.g., Global Financial Crisis) and their impacts, which were not controlled for. Another aspect to consider in future studies is the complex relationship between alcohol and other drugs in the suicidal process [53] and the need to analyze policies related other substances.
Finally, our review is limited by inclusion of English language literature and the studies included have been mainly conducted in Western settings, which limits the generalizability.

Conclusions
The studies included in the review predominantly investigated the effect of restrictions on alcohol availability and increased cost of alcohol, and majority found associations with reduced suicides across Western and Eastern Europe, as well as the US.
Author Contributions: K.K. and K.W. developed the study design; K.W. ran the searches; K.W., K.M.C., R.W., and K.K. screened the papers and extracted data; K.K., K.M.C., and K.W. drafted the manuscript; A.V., D.d.L. and R.W. provided feedback and revised the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This research did not receive any specific funding.

Conflicts of Interest:
The authors declare no conflict of interest. Quote: "To analyze the potential effect of alcohol on Alaska Native injury deaths, we first divide the study population into two groups. The experimental group consists of Alaska Natives living in Alaska communities that used the state local option to restrict alcohol (went 'dry' of 'damp') at some point between 1980 and 1993. The control group consists of communities that did not exercise the local option (remained 'wet') throughout the period 1980-93". (p. 313). Comment: Not reported; however, it is likely there are further differences between those communities that chose restrictive options versus those that remained 'wet'.

Appendix A
High risk of bias.
Quote: "To analyze the potential effect of alcohol on Alaska Native injury deaths, we first divide the study population into two groups.
The experimental group consists of Alaska Natives living in Alaska communities that used the state local option to restrict alcohol (went 'dry' of 'damp') at some point between 1980 and 1993. The control group consists of communities that did not exercise the local option (remained 'wet') throughout the period 1980-93". (p. 313). Comment: Not reported; however, it is likely there are further differences between those communities that chose restrictive options versus those that remained 'wet'. High risk of bias.