49 An economic perspective on suicide across the five continents

In considering suicide prevention measures it is important to consider potential economic risk factors, as well as the costs and consequences of suicide. We provide a brief overview of some areas where economics has played a role in the analysis of suicide and suicide prevention strategies. Evidence on the wide-ranging socio-economic costs and consequences of suicide is provided, as well as a reflection on the development of economic theories on individual motivations for suicide. The evidence from econometric models at both cross-country and single country levels on the links between suicide and socio-economic risk factors such as poverty and unemployment are reviewed. Cost-effectiveness is increasingly used as part of decision-making processes in health and other sectors. In respect of suicide prevention, such evidence remains limited. The present review, nonetheless, suggests that suicide prevention measures may be highly cost-effective. Incorporating economic analyses into future effectiveness studies is likely to help strengthen the case for investment in suicide prevention. There is also scope to look at the economic implications of interventions already shown to be effective.

The impact of suicide is profound. In addition to the emotional suffering, caused by suicide and attempted suicide, the cost to the nation and/or region in the case of suicide or attempted suicide is increasingly recognized as a significant burden. While the avoidance of just one such tragedy provides a powerful motivation for action, the economic perspective may well be one additional incentive that hopefully will act as catalyst for noteworthy preventive measures. Policy-makers in health and other sectors are continually faced with many competing claims for resources, and economics is concerned with the optimal use of scarce resources. Investing in measures to reduce suicide is unlikely to be a cost-free exercise. A better understanding of the socio-economic risk factors and costs of suicide may provide valuable information to policy-makers considering whether or not to increase investment in suicide prevention strategies and programmes. Of course this is of little use unless effective strategies are available, but assuming that they are, then it is also helpful to look at their cost-effectiveness.

This chapter provides a brief overview of some of the different areas where economic evidence and approaches can play a role. We begin by looking at the economic impact of suicide and by identifying its many costs and consequences. To provide effective suicide prevention interventions, it is also critical to have a better understanding of what the individual motivations for suicide might be. We look at what economic theory can contribute to this discussion. We also review the econometric work that has been undertaken to identify whether socio-economic factors such as poverty and unemployment are linked to suicide. Finally, we reflect on what is known about the cost-effectiveness of different measures to counter suicide, and how such information may be used to inform policy and practice across the five continents.

As part of any process, considering whether or not to invest in suicide prevention, it is important to have information on the costs of not taking action; i.e. the costs of suicide. These are substantial impacts on many sectors of society and are long lasting. Most obviously, such costs include what economists call direct costs, including the demands placed on the emergency services, as well as use of potential life-saving interventions, police investigations and funerals. For those individuals who survive lengthy physical and psychological rehabilitation may follow (Meneghel et al. 1996; O'Sullivan et al. 1999).

There are also two kinds of indirect costs. Lost output costs refer to the fact that as a result of premature death, individuals lose the opportunity to contribute to the economy, whether this is through paid work, voluntary activities, or family responsibilities such as looking after one's children or parents. The most fundamental impact of suicide, of course, is the loss of the opportunity to experience all that life holds. The pain and grief caused to family and friends, as well as ensuing health consequences, can also be immense and take many years to subside. Suicide can be stigmatizing for families, particularly in cultures where suicide may be regarded as sinful or where it is illegal. Conversely, some individuals who complete suicide may believe this to be of benefit (or negative cost) to others; for instance it may be seen as a way of avoiding shame for failure or becoming a burden to family. All of these very personal, and often culturally driven impacts, are collectively known by economists as ‘intangible costs’ because they are often hidden and difficult to evaluate. Increasingly, however, studies try and place a value on some of these complex costs.

Surprisingly, there are few cost estimates for suicide compared with those for other causes of premature death, such as illness or unintentional injury. Moreover, those available do not always include all costs, and values can vary considerably, but all indicate that preventing suicide would avert enormous costs to society. One of the earliest estimates from the Canadian province of New Brunswick, despite not including intangible costs, still estimated the cost of each suicide to be CAN$1,019,210 (Clayton and Barceló 2000). In the United States, the total costs of suicides for those aged 20 or less in 1996 were estimated to be more than US$15.5 billion, 75 per cent of which were intangible costs and 18 per cent indirect costs, due to the loss of opportunity to contribute to the economy (Cox and Miller 1999). The costs of the lost opportunity alone to earn a paid wage, as a result of all suicides in the US in 2002, have also been estimated at US$13 billion (Knox and Caine 2005).

One Australian study estimated the total costs of suicide or self-harm events in the state of Victoria at A$428 million in 1994, 25 per cent of these costs were for direct health sector and other sector costs (Watson and Ozanne-Smith 1997). There are also cost-of-illness studies for different mental disorders that include suicide. In England, the costs of depression in 2000 (excluding the value of intangible costs) included £562 million for premature mortality, much of which was due to suicide (Thomas and Morris 2003).

Several studies, using similar methods, include all three types of cost (see Table 49.1). In New Zealand, costs were estimated at £1,158,768 (NZ$3,094,243) per suicide (O'Dea and Tucker 2005) while in Ireland, the corresponding figure was some £1,402,438 (€1,982,667) (Kennelly et al. 2005). In Scotland, the average cost was £1.29 million per completed suicide (Platt et al. 2006).

From Table 49.1, it is clear that the direct costs of suicide account for just a tiny fraction of total cost. The lifetime lost opportunity to contribute to the economic output of a nation, as well as the additional value placed by society on the loss of life, are the main contributors. Given these high costs, the potential economic benefit, if the number of suicides can be reduced, may be substantial. In the Scottish case, for instance, a 1 per cent reduction in the number of suicides on average might avoid costs of up to £10.7 million over the lifetimes of these individuals. This is dependent on the availability and cost-effectiveness of interventions.

Having illustrated the socio-economic costs of suicide, how and what can economics contribute to our understanding of individual

motivations to contemplate suicide? A better understanding of these motivations may help us to invest in interventions and strategies that address the underlying factors influencing motivation.

Durkheim, in his classic work, looked at the links between the structures and roles that individuals play in society and their risk of suicide, including some discussion of the role of income and changes in economic circumstances (Durkheim 1897). For instance, he suggested that a lifetime of poverty might be a protective factor against suicide, as the poor would have time to learn to live without material resources. In contrast, those individuals who experienced a downturn in their fortunes as a result of unemployment would be at higher risk of suicide. Similarly, he argued that suicides might increase in times of economic prosperity, when the demand for consumer goods might become insatiable, leading to anomie or social fragmentation, where social norms do not correspond with an individual's perception of the meaning of life.

This theory has been refined and/or challenged ever since. Ginsberg, for instance, writing in 1966, suggested that suicide would only increase in times of economic prosperity. During recessions, economic aspirations would decrease at a greater rate than economic growth, thus, individuals would expect less than what they would actually receive (Ginsberg 1980). Henry and Short, in contrast, suggested that suicides would only rise in times of recession, due to aggression and frustration being released when individuals realize that they cannot meet all their goals (Henry and Short 1954).

Despite this long-standing interest in the links between broad socio-economic factors and suicide by sociologists and others, it was not until the early 1970s that economists began to turn their attention to trying to explain the individual rationale behind suicide. While the number of economic papers on suicide is probably still too few to talk about any standard or basic model, the closest that any has come to attaining such status is the Lifetime Utility Model (Hamermesh and Soss 1974). This begins by noting Schopenhauer's famous statement that ‘as soon as the terrors of life reach the point at which they outweigh the terrors of death, a man will put an end to his life’ (Schopenhauer 1974, p. 310). Lifetime utility is determined by age and income and by a ‘taste for living or distaste for suicide’. The basic assumption is that an individual kills himself when the total value of remaining lifetime reaches zero (or below some threshold required to continue to live).

This mainstream economic or ‘rational choice’ approach sees an individual's stock of health as a form of human capital, with skills and potential to be productive. The individual inherits an initial stock and invests over a lifetime to produce a stock of health which depreciates with age. The decision to invest in health is affected by income, the cost of health inputs, the return to investment in human capital and the time horizon over which this investment may be recouped. The theory suggests that suicide rates would increase with age as the costs of maintaining health increase with age.

This lifetime utility model has been criticised on several grounds. First, given that many suicidal individuals have mental disorders, it would be wrong to assume individuals to act rationally. Indeed, Hamermesh and Soss, themselves, acknowledged that the ‘individual agony resulting in suicide does not stem solely from an economic calculation’ (1974, p. 97). They, nonetheless, argued that their model illustrated the apparent links between macroeconomic conditions and the risk of some suicides. Others might argue that the economic model does not take rationality far enough—some individuals will have different views of the act of suicide and place a greater value on life after death depending on religious and cultural values.

A rational individual should also take into account the possibility that human life might improve. Given that suicide is an irreversible act, this ‘option value’ of staying alive could be extremely large (Dixit and Pindyck 1994). Indeed, those individuals who have enjoyed better fortunes in life might be more likely to survive into old age and, thus, may be less likely to contemplate suicide. Individuals might also wait to see if their circumstances improve before contemplating suicide. This would be consistent with evidence indicating that suicide actions are higher among the long-term unemployed (Platt 1986).

It is important to note that economic theories and models can be modified. One model, incorporating this option value on the potential of future life, no longer predicted any direct relationship between suicide and age (Cutler et al. 2001), in consistency with the increased rates of suicide in young people seen across the world (Wasserman et al. 2005). Economic models can potentially take account of a myriad of potential risk factors and socio-economic contexts. For instance, relative income rather than absolute income may potentially have more effect on one's likelihood to commit suicide (Daly and Wilson 2006). The prior model is part of a rapidly growing body of literature in economics that suggests that interdependent preferences matter. It also harks back to some of the theories put forward by Durkheim and others about placing individual income into an appropriate context.

Game theory has also been applied by economists to estimate the trade-offs individuals may consider between the benefits of obtaining help for suicidal behaviour and the potential negative consequence of perhaps being involuntarily detained. In this model, health service professionals also have to weigh up the pros and cons of detaining individuals against providing community-based support, with a possibly higher risk of suicide. The likelihood that an individual will seek help will be dependent on the probability that support will remove suicidal tendencies with certainty (Yaniv 2001). From a policy perspective, this would suggest not only developing ever-more effective support and treatment, but engaging in campaigns to improve public awareness of the availability of such options.

Models can also be adapted to specific phenomena, such as Koo and Cox's model (2006), accounting for the rise in suicide among middle-aged men in Japan. Their model hypothesizes that any individual's human capital declines as a result of unemployment (because of the loss of on-the-job training etc.). During periods of sustained economic transition driven by technological shocks, the human capital of the middle-aged unemployed depreciates faster than that of younger workers, because they tend to be slower in adapting to new labour market conditions. This in turn (along with other factors such as a greater chance of divorce) drives down the expected relative lifetime utility of unemployed individuals, and thus increases the risk of suicide. From a policy perspective, if the model is validated, it might suggest that appropriate labour market interventions may include schemes to help re-skill older workers.

One feature of the economic approach that has not received as much attention is the way in which decreased utility or distress is produced. Lillard and Firestine's (2006) model focus on how a person chooses to manage such distress using what economists call ‘a production function’ approach. According to this model, the output, such as the emotional well-being that an individual enjoys, is dependent on certain inputs chosen or provided. These inputs might either be health-related goods (such as drugs, counselling) or non-market activities (relationships with friends or family). When an exogenous event, such as the onset of depression, substantially reduces the individual's emotional well-being, the mix of inputs in the production of emotional well-being needs to change. Their model predicts that a reduction in the availability of antidepressants (for instance as a result of an increase in price) should cause individuals to spend more time cultivating relationships with friends and family to bolster mental well-being.

In a similar fashion, another economic model considers child well-being as an output of a production function where the inputs are parental time and market goods (Mathur and Freeman 2002). Parental decisions to devote more time to work (or spend less time with children as a result of divorce) have a negative effect on child well-being, although this may be outweighed by the additional income earned. It might also imply that greater contact with relatives, e.g. grandparents or others through child care and/or after-school activities, might be needed to help counter some of the negative effects that may increase the risk of suicide.

Although not the focus of our chapter, it is worth noting that in addition to theories on motivations for completed suicide, economists have also sought to use game theory to explain motivations for attempted suicides, some of which may be perceived to be a ‘cry for help’. One recent economic model assumes that such actions may be a rational attempt to elicit attention and care from family and friends. In the case where an individual assesses the potential additional care and long-term improved utility gained to outweigh the risk of death as well as the potential negative consequences of non-fatal but life-threatening self-harm events, the likelihood of such actions may increase (Marcotte 2003).

To what extent does empirical evidence support theories on the role of macro-economic factors and suicide?

Challenges remain in obtaining comprehensive individual-level data. Most empirical work to date is ecological in nature, making it impossible to definitively infer any individual behaviour from the aggregate data. The studies cited below at the cross-national, national and individual levels mainly report results that have reached conventional levels of significance in multivariable regression analysis. Many of the papers use fixed effects estimation techniques to control for differences in cross-country suicide cultures that are time invariant within countries. They also use time dummies that account for changes in suicide rates over time common to all countries.

A number of cross-country studies have looked at income levels as a risk factor for suicide in both developed and developing countries. The findings have been mixed: some report that there is a negative relationship between per capita income and male suicide rates (Neumayer 2003b; Helliwell 2004; Lin 2006). Another study states that per capita income has a non linear effect on male suicide rates, initially positive, but then negative for incomes above a threshold ($30,700) (Neumayer 2003a). The impacts of low income are particularly marked in transition countries in Eastern Europe, where the decline in gross domestic product (GDP) per capita and the employment/population ratio, accounted for nearly one third of the increase in male suicides in the 1990s (Brainerd 2001). On the other hand, an analysis of suicide rates in fifteen western European nations found that economic growth had a significant negative effect on both male and female suicide rates (Rodriguez-Andres 2005). Several studies did not find any significant association between suicide rates and income inequality (Helliwell 2004; Rodriguez-Andres 2005, 2006; Leigh and Jencks 2007).

Many studies report a positive relationship between unemployment and male suicide rates (Brainerd 2001; Neumayer 2003a). The rate of female participation in the labour force has also been shown to have a positive effect on male and female suicide rates (Neumayer 2003a; Rodriguez-Andres 2006). The evidence is again mixed with some studies not finding any association between unemployment and suicide rates (Helliwell 2004; Gerdtham and Ruhm 2006). A study of countries in western Europe found a positive effect for unemployment on suicide rates only for 45–64-year-old men, indicating the importance to policy-makers of looking at suicide by age group when considering potential suicide prevention measures (Rodriguez-Andres 2005).

Turning to single country analyses, again there is a mixed pattern of evidence on the links between macroeconomic factors and suicide. Some US studies suggest that state-level income is insignificant (Markowitz et al. 2003; Ludwig and Marcotte 2005; Cuellar and Markowitz 2007). Hamermesh and Soss, in contrast, reported a negative relationship between discounted permanent income and state suicide rates for men in the US (Hamermesh and Soss 1974). Another US study reported that county level median income also had a negative effect on youth suicide rates, although this result was sensitive to gun ownership (Cutler et al. 2001).

Not only income, but relative income, as some economic models contend, can be important, and this can differ by income group. In one recent US study, increased inequality between the average and the upper tail of the income distribution was linked to increased suicide rates, while increases in inequality between the average and the lower tail of the income distribution reduced suicide rates. This has been interpreted as meaning that individuals in the middle of the income distribution experience higher utility if they manage to keep up with those on higher incomes, while moving further ahead of those on lower incomes also increases their utility. If an analysis simply looked at the population as a whole, however, no relationship between income inequality and suicide would be observed (Daly and Wilson 2006).

A prior US study illustrates the importance not only of an individual's income, but also that of other family members. Increased parental income was found to be associated with declines in state suicide rates of 15–19-year-olds (Mathur and Freeman 2002). This outweighed the negative impact of reduced contact between parents and their children. The results are not confined to the US—two studies in Taiwan have reported a negative association between per capita income and regional suicide rates (Chuang and Huang 1997; Lin 2006).

A number of studies report that the male suicide rate increases with unemployment, e.g. Hamermesh and Soss (1974), in some instances this relationship is stronger for older age groups (Ruhm 2000). Unemployment has also been positively associated with the suicide rate for 15–19 year olds in two US studies examined (Mathur and Freeman 2002; Cuellar and Markowitz 2007). Increased rates of female labour force participation have been linked to increased rates of suicides in young people at state level in the US (Cutler et al. 2001) and also in Japan (Koo and Cox 2006). The risk of suicide has also been shown to decrease in US counties when the proportion of the population with a university education increases (Markowitz et al. 2003; Daly and Wilson 2006).

Another US study used data from the 3 (NLMS) to follow a cohort of five national samples over a 10-year period from 1979 to 1989. Again the results suggest that an unemployed person is at greater risk of suicide, and that the effect may diminish over time, but be more enduring in women (Kposowa 2001). This dataset was also used to analyse differences in risk of suicide by occupation, after controlling for socio-economic factors. This suggested that risks are significantly higher for some occupations such as mining, retail and construction, but great care is needed in interpreting such findings (Kposowa 1999).

The positive relationship between unemployment and suicide can also be seen elsewhere around the world, for instance in Denmark, Sweden, Taiwan and Japan (Qin et al. 2003; Dahlberg and Lundin 2005; Lin 2006; Koo and Cox 2006). Agerbo and colleagues, making use of the rich data from Denmark's health, social and economic registers, were able to analyse individual level data and also area level-based factors. Again this study found that suicide risk was associated with a lack of employment and low income, combined with living alone or being divorced. It is also suggested that the characteristics of geographical areas do not have much impact on suicide, and thus policy-makers should target their efforts on the characteristics of individuals, which will include socio-economic and cultural factors, rather than their area of residence alone (Agerbo et al. 2007a, b). These socio-economic and culture factors might include ethnicity and sexuality, although evidence from economic models linking them to suicide remains limited.

As with cross-country studies, at country level not all evidence points in the same direction; no significant relationships between unemployment and suicide were found in some US studies (Markowitz et al. 2003; Ludwig and Marcotte 2005), while in Germany, for example, the economic downturn was associated with decreasing rates of suicide (Neumayer 2004). Again this illustrates the need for careful consideration of local contexts—for instance, the extent to which the social welfare safety net may cushion the economic impacts of recession.

Many economic analyses have also considered social and behavioural factors as illustrated in one recent review (Rehkopf and Buka 2006). Consistent with results reported by others, divorce rates have been shown to have a positive effect on suicide rates in both developed and economic transition countries (Brainerd 2001; Neumayer 2003a, b; Helliwell 2004), while birth and marriage rates have a negative effect (Neumayer 2003a, b). Alcohol has also been associated with poor mental health and increased risk of suicide; evidence from econometric studies in the developed world indicates a positive effect of alcohol consumption on both male (Wasserman et al. 1994; Neumayer 2003a; Rodriguez-Andres 2005; Värnik et al. 2007) and female suicides (Wasserman et al. 1998; Brainerd 2001; Rodriguez-Andres 2006). Social networks and community well-being appear to be protective against suicide. One study found that two indicators of social capital—membership per capita in non-religious organizations and trust—were significantly associated with lower rates of suicide for both men and women. Belief in God was also found to be protective (Helliwell 2004). A later study using the same data also identified a positive relationship with poor subjective perceptions of health (Helliwell 2006). The links between suicide and the utilization of health care services, in particular antidepressants, have also been examined. While the impact on the use of selective serotonin re-uptake inhibitors (SSRIs) and suicide are complex and subject to debate (Simon et al. 2006), some studies indicate that the use of SSRIs may have a strong negative effect on suicide, with rates falling fastest in those countries that experienced the most rapid rate of growth in SSRI sales (Ludwig and Marcotte 2005).

At a country level, the US studies also indicate that divorce has a significant positive impact on suicide rates (Cutler et al. 2001; Mathur and Freeman 2002; Daly and Wilson 2006). In Japan, this relationship was only significant for male suicide, although there was weak support for a negative relationship between fertility and suicide rates, with a greater effect on women than on men (Koo and Cox 2006). In contrast, divorce had a significant negative effect on female suicide rates in a Swedish study (Dahlberg and Lundin 2005).

Most studies that looked at alcohol consumption at country level indicated that this was a risk factor, as for instance in Sweden (Dahlberg and Lundin 2005). Alcohol was shown to be a risk factor for 15–19-year-olds in one US study (Mathur and Freeman 2002), although this finding was contradicted by a more recent analysis, which indicated that beer consumption had a negative impact on suicide rates in this age group (Cuellar and Markowitz 2007). In another study of US states, a higher tax on beer was associated with lower male suicide rates—a 5.5 cent increase in the beer tax was estimated to save on average one male in the 15–19 and 20–24 age groups per state per year. The evidence on drink driving laws was mixed: in one study having only a limited effect on male suicides, while few regulatory measures had any impact on female suicide rates (Markowitz et al. 2003). Another US study did, however, find that the adoption of very tough underage drink driving laws were associated with reductions in the suicide deaths of 15–20-year-olds (Carpenter 2004). In the US, the consumption of SSRIs appeared to have a significant negative relationship with the state level suicide count 15–19-year-olds (Cuellar and Markowitz 2007), but in Sweden, no relationship could be found between the use of antidepressants and county suicide rates (Dahlberg and Lundin 2005).

We have briefly highlighted the importance of looking at the costs and consequences of suicide, and examined how economic models can help potentially identify areas where policy-makers may wish to focus suicide preventive efforts. At the start of this chapter, we emphasized the importance of taking action based on evidence not only of effectiveness, but also the undertaking of economic evaluation to assess cost-effectiveness, which is increasingly used as part of the process of evidence-based health policy making (Oliver and McDaid 2002). Below, a brief look at some different approaches and then a review of the extent to which economic evaluation has been used to date is given.

Economic evaluation, in essence, compares both the costs and effectiveness of two or more alternative uses of resources. Figure 49.1 illustrates the range of possible cost-effectiveness results, and the way in which they can help inform policy-making.

Possible combinations of outcomes and costs when comparing two measures to prevent suicide are shown in Figure 49.1. Point B indicates that the new measure is both more effective in preventing suicide and less costly than the existing measure. In these circumstances, the task for the decision-maker is quite straightforward: recommend wider use of the new suicide prevention measure. It may be the case that a new measure is in fact at point A: producing better outcomes but at a higher cost. The decision now is more complex because a trade-off is needed: are the better outcomes worth the higher costs? There is no right or wrong answer to this question—decision-makers must decide how much they are willing to invest for improved health outcomes. In the UK, for instance, interventions that cost no more than £30,000 per life year gained are usually funded without great debate. Economic evaluation is just one source of information; many other factors will rightly influence decisions, not the least of which will be equity considerations.

There are different methods of economic evaluation; all estimate costs in the same way but differ in how they treat outcomes (Drummond et al. 1997; Byford et al. 2003b; Hale et al. 2005). Cost-effectiveness analysis (CEA) measures outcomes using a natural measure, e.g. the number of suicides averted. Cost-utility analysis (CUA) uses a common health outcome measure, such as the Quality-adjusted life year (QALY) or the Disability-adjusted life year (DALY) (Murray and Lopez 1996). Both these measures can use values from the public or specific target groups for different health or disability states respectively. This has the advantage of allowing policy-makers to compare investments in suicide prevention with alternative health-promoting interventions. However, it can be difficult to make use of values for health states from individuals experiencing suicidal thoughts, who may rationally place (at least temporarily) a greater value on death compared to life (Chisholm et al. 2006).

Cost–benefit analysis (CBA) places a monetary value on all outcomes. With two or more alternatives, the intervention with the greatest net benefit would be deemed the most efficient. This is useful given that the costs and consequences of suicide go well beyond the health care system. CBA needs to obtain monetary values from individuals on the intangible value of life, which again can be problematic when considering suicide (Healey and Chisholm 1999), but the technique is widely used for assessing the value of comparable interventions such as transport-related injury prevention programmes (Department of Transport 2004). Nonetheless, because all outcomes (health and non-health) can be considered and valued, CBA can facilitate the comparison between suicide prevention measures and alternative investment in any sector such as housing, education or defence, for decision-makers to more easily compare.

The idea of assessing the costs of suicide prevention programmes, as well as the economic value of suicides averted, is not new (Diggory 1969; Lum 1973). Despite this, and the attention currently given to suicide prevention in national mental health policies worldwide, there are few economic evaluations of population-based measures to tackle suicide (US Preventive Services Task Force 2004). This is unsurprising, given that there is still relatively little evidence on the effectiveness of such complex community-based interventions (Beautrais 2005; Mann et al. 2005).

One rare example of the evaluation of an area-based prevention strategy, albeit with many limitations, one of them being that it is highly context-specific, is a retrospective analysis of a suicide prevention programme targeted at members of a reservation-based Native American group in New Mexico (Zaloshnja et al. 2003). The intervention focused primarily on young people aged 15–19, with the whole community as a secondary target group. It included peer training of young people, postvention outreach, community education programmes, and suicide-risk screening within local health and social care programmes. The rate of suicidal acts and suicides in the eight years prior to the programme was compared with suicides over the subsequent eight years. Direct and indirect costs were estimated. Overall the programme was deemed to have cost savings of $1.7 million due to the marked decrease in the rate of suicide from 59 per 1000 in the 15–19 age group, to a much lower rate varying between 10 and 17 per 1000 in subsequent years. Benefits gained were 43 times greater than costs incurred, while the cost per QALY saved was just $419, a value in most jurisdictions that is considered to be highly cost-effective. These findings must, however, be treated cautiously, not least because of the lack of a comparator group and observed natural cyclical changes in suicide rates within this population group. Nonetheless, the study illustrates how economic analysis might be incorporated into the evaluation, and also how this may be done retrospectively making use of existing knowledge on the effectiveness of interventions.

Other powerful tools that economists can make use of include decision-analytical models. In the absence of long-term empirical data, these can be used to synthesise evidence on short-term costs and consequences of interventions and make long-term projections (Buxton et al. 1997). While rare for suicide prevention, modelling has been used in the US to estimate the potential cost-effectiveness of both a general suicide education and peer support group programme targeted at university students in Florida (de Castro et al. 2004). The model indicated that, because of the substantial lifetime costs avoided, both programmes would be highly cost-saving, with the peer support programme and the general education programme generating benefits 5.35 and 2.92 times, respectively, greater than costs. Another example is that of a regression model being used to assess the potential costs and consequences of suicide prevention centres in the US in the 1980s (Medoff 1986). These centres provided a 24-hour telephone service which would initiate crisis intervention services if appropriate. The model suggested that the value of human lives saved would be at least five times greater than the costs of running the centres.

Training to improve the recognition by primary care physicians of individuals at risk of suicide has also been highlighted as effective (Mann et al. 2005), but little economic analysis has been undertaken. One evaluation, set on the Swedish island of Gotland, looked at an educational programme for general practitioners to improve their ability to recognisze and treat the symptoms of depression (Rutz et al. 1992). Although there were methodological limitations in the assessment and attribution of the effectiveness of the data, the study suggests that the programme was cost-saving, in part because of the avoidance of costs associated with suicides averted. More recently, the costs of a training intervention in England that focused on the risk assessment and management of suicide was assessed alongside information on implementation and acquisition of skills and knowledge by front line health professionals. This study indicated that if a 2.5 per cent decrease in the suicide rate could be achieved, the cost per life year gained would be just £3391, a value considered to be highly cost effective in high income countries (Appleby et al. 2000).

There has also been some limited economic analysis of specific interventions to reduce the risk of suicide, including various safety measures (such as safety nets and barriers for bridges) and restriction of access to means such as firearms and poisons. Much more information is available, however, on a wide range of interventions intended to prevent unintentional injury which are also relevant to suicide, such as the installation of airbags in cars (Thompson et al. 2002). Thus, it is also important to look at the potential cost-effectiveness, in respect of suicide, of many measures to counter risky behaviours that may have many adverse health impacts. Alcohol control measures have been widely evaluated, although again these typically do not focus specifically on suicide. One recent study indicates that cost-effective measures are available right across the globe: in Europe the most cost-effective approach to reduce hazardous alcohol consumption is through taxation, whereas in parts of the world where heavy alcohol use is rare, e.g. in South East Asia, targeted interventions such as brief physician interventions or advertising bans appear to be more cost-effective (Chisholm et al. 2004).

For those already identified as being at high risk of suicide, there is some emerging evidence on the cost-effectiveness of different interventions. In England, a home-based social work intervention targeted at children, who had previously deliberately poisoned themselves, reported no statistically significant differences in overall health and social care costs between this treatment and routine care. However, it did record lower rates of suicidal ideation at 6-month follow-up in the subgroup of children without major depression, who received the social work intervention compared with those receiving routine care alone (Byford et al. 1999). Another study looked at the use of cognitive behavioural therapy (CBT) for people with a history of deliberate self harm in centres in both England and Scotland. It suggested that manual CBT was likely to be cost-effective in reducing the number of deliberate self-harm events, although it did not explicitly look at suicides averted (Byford et al. 2003a).

Some studies have also indicated that substantial costs can be avoided by providing treatment and support to people with depression (Wolfersdorf and Martinez 1998). While we do not propose here to enter into the debate about the relative merits of antidepressants as an effective suicide prevention intervention, it should be noted that such medications have been subject to much assessment of cost-effectiveness. Some of these evaluations include the economic impact of suicides averted (Barrett et al. 2005). Ludwig et al. (2007) estimate that an increase in SSRI sales of one pill per capita would reduce suicide mortality rates by around 5 per cent. Based on this the cost per statistical life saved from increasing SSRI used is around $20,000.

While economic evaluation has an important role to play, common sense needs to be applied to its use, as resources to undertake such evaluations are often scarce. Some interventions may be shown to be of sufficient effectiveness and have such low implementation costs that a comprehensive economic evaluation would add little to the case for investment. They are already clearly cost saving. This, for instance, might include simple, but effective, measures such as the erection of signs for support services in some areas known to be suicide black spots (King and Frost 2005).

Economic evidence can play an important role in helping to understand how different risk factors influence individual motivations for suicide. Econometric techniques, both across and within countries, linked to economy theories, can add knowledge to ecological and epidemiological work already undertaken. Providing information on the profound socio-economic costs of suicide, in particular, emphasizing that these impacts go way beyond immediate costs to health care and emergency services, can help to strengthen the case for taking action to reduce suicide. Ultimately, such a case may stand or fall on evidence not only on the availability of effective interventions, but also on cost-effective interventions. Such information is increasingly part of evidence-based policy-making, but its use in suicide prevention remains limited, and is focused on high income countries. Incorporating economic analyses into new evaluations is one step that might be taken, but much can also be done through retrospectively considering what the crude economic costs would be of implementing effective prevention strategies in different contexts, and what the potential economic benefits of predicted suicides averted as a result might be. The limited evidence available to date would suggest that investment in effective interventions is likely to be highly cost-effective compared with other uses of public resources. The challenge is to expand on, strengthen and communicate findings from this emerging evidence base.