32 Labour market, work environment and suicide

Work is an important sphere of human life. Besides economic subsistence, it also furnishes workers with social status and influences their life conditions in a profound manner. Social class at large seems to be connected with suicidality, but studies on the effects of specific occupations have produced few lasting results, perhaps due to the different societies investigated. In addition, lack of adequate data and problems with the methods chosen cause problems in the estimates of suicide mortality by class or occupation. However, it seems that the most vulnerable position is that of those who do not work at all. There is abundant empirical evidence of a surplus risk for suicide among the unemployed, but the causal nature of this relationship still needs clarification. Globally, the labour markets differ greatly, and so does their connection with suicide. Labour-market oriented suicide prevention issues concern unemployment policies, reduction of work-related access to means of suicide, and the use of the workplace as a base for suicide prevention.

Work, the activity on which human beings base their subsistence, is one of the central parts of human life. In a modern society, work is not performed for the needs of the workers and their immediate families only, but for a larger group of potentially interested persons, economically speaking, for a market. At the same time, work itself is a market commodity, bought and sold on the ‘labour market’.

The market, and the positions existing on it, varies between societies and over time. Agricultural, industrial, and post-industrial societies all have their specific divisions of labour, with specific positions and labour markets. Seen from this perspective, the sheer number of potential positions to be occupied varies greatly, as does the degree of spatial and chronological separation between work and home.

The seller side of the labour market comprises actively working persons, employed by others or by themselves. Since complex tasks require experience and/or education for their correct performance, an individual's general position on the labour market tends to be stable, reflecting their human capital. Not everyone performs paid work, however, their not doing so is still often related to the labour market, as in the case of the unemployed (persons who would like to work but cannot find employment), the sick (persons judged incapable of working), students (persons presumably in preparation for a position on the labour market), and the retired (mostly persons who have worked). Thus, a majority of any population is in one way or another connected to a labour market and its order, in which the positions are often more or less hierarchically related to each other.

Individuals’ positions on the labour market determine, in broad terms, what kind of physical environment they will spend the major part of their days in, and what kind of stresses their bodies will thereby be exposed to. The same applies to the mental environment, although the variation between specific places of work may be larger in that regard. Moreover, different work situations offer different possibilities to stress relief such as, for example, independent decision-making (Karasek and Theorell 1990). In addition to the mental stimuli provided by the work itself, one must take into account the social work environment consisting of persons close to oneself in terms of location, activities, and status. This is likely to be the second most salient social environment in a working individual's life beside family, and like family, its members are rarely chosen by the individual.

An individual's position on the labour market is a strong social marker, and it is often used as an indicator of one's overall standing in society. Besides that, it is also a major determinant of one's life chances in general through the economic, sociopsychological, and physical conditions it implies. Work is not only a source of material subsistence, but a gateway to many other things, and one of the main ties between the individual and the larger society. This in turn is one important rationale behind the concept of social class, a group of people defined by its similar position on the labour market, and who is often supposed to share other significant characteristics as well. Within each class, there is then a number of occupations, each with its own qualities and conditions.

The various positions on the labour market are intimately related to the health of the individuals occupying them (Marmot 2004). Mental health is no exception. Since suicidal behaviour is partly socially conditioned, it would seem logical that different positions on the labour market would also imply different frequencies of suicide, and so they do, to a certain extent. The influences related to the organization of labour can be divided into (a) the effects of different class (and occupational) positions; (b) the effects of immediate work environments; and (c) the effects of a lack of position on the labour market.

Existing models of suicidal behaviours do not usually put directly work-related issues in central positions. However, both stress and object loss are important constitutive parts of many models (see Maris 1997; Wasserman 2001), and these can certainly be related to work.

As regards the hierarchical, more fixed, and partly inherited aspect of the organization of labour in society, the social classes, many studies have at least indirectly investigated the relationship between class and suicide. In contemporary industrialized countries, clear signs of a ‘class ladder’ in regard to suicide exist among men. Those with low income or education tend to have higher rates of suicide than those with higher income or education, and those in manual work higher than those in non-manual work; even lower-ranking office clerks often have higher suicide rates than those in the higher echelons (Stack 1982, 2000; Platt and Hawton 2000; Blakely et al. 2002; Qin et al. 2003; Kwan et al. 2005; Kalediene et al. 2006; Kim et al. 2006; Stark et al. 2006). The differences in suicide rates between the lowest and highest positions are typically two- or threefold. Mäki and Martikainen (2007) have calculated that the class difference in suicide mortality contributed 0.6 years (or 10 per cent) to the general difference in the life expectancy (at age 25) between manual workers and upper-echelon non-manual workers in Finland in 1991–2000.

However, the steepness, and sometimes even the existence, of the class ladder depends on the method used to determine the suicide risk, the specific groups studied, the time of the study, and other factors controlled for (Stack 2001; Lostao et al. 2006). Comparisons between groups are sometimes made using proportional mortality ratios (PMR), which relate the mortality from suicide to that from all causes. This tends to highlight suicides in groups with lower overall mortality (Kelly et al. 1995; Platt and Hawton 2000; Stack 2001), and the results must be interpreted accordingly. To take an example: in Japan, standardized all-cause mortality among miners in the ages of 20–64 was five times higher than that among managers in 1985 (Fujioka et al. 2002). Thus, five miners' suicides would be needed to produce the same proportional measure of ‘suicide risk’ than one managerial suicide. However, there is no obvious reason why suicide mortality and mortality from other causes should be related to each other.

Stack (2001) found in the USA that the higher risks for suicide among manual workers were no longer significantly higher when the demographic differences between occupational groups were controlled for. Class differences are often closely intertwined with those in other matters: lower-class men may, for example, have an elevated risk for divorce due to their low, work-based social position and the disadvantages it entails. Should a suicide in this situation be attributed to the divorce or the social position? Moreover, many occupational groups have their characteristic demographics which partly shape the prevailing work atmosphere, perhaps influencing even mental health and suicidality. The meaning and implications of the controls used in analyses is sometimes uncertain (Agerbo et al. 2007a).

The empirical evidence concerning the impact of social class on women's suicide is more mixed. Some studies find relations more or less similar to those prevailing among men (Pensola and Martikainen 2003; Steenland et al. 2003; Kim et al. 2006), while others would seem to indicate a surplus risk for suicide among women in higher classes (Kelly et al. 1995), and still others find either inconsistent relations (Blakely et al. 2002; Lorant et al. 2005), or none at all (Kwan et al. 2005). This may reflect women's various—and changing—roles in relation to the labour market in different societies, but also the fact that their general socio-economic position is less often defined by their own occupation or income than men's, which makes individual-based class definitions more insecure.

However, seen in a larger perspective, social class seems to be connected with suicide not mainly through the economic and status positions it implies, but through its cultural aspect, the specific ways of thinking and acting which may be connected with certain classes. Rather than having any absolute effect on suicidal behaviour, class membership ‘organizes’ it much in the same way as it patterns other behaviours. Thus, the relation between class and suicide must be understood in its specific context.

To take an example, higher classes were over-represented among suicides in Czarist Russia a century ago, while the opposite is true today (Mäkinen 2002, 2006). The reason for this shift can hardly be economic or status-related, since those differences between the different classes in Russia remain similar today. However, the individualism and secularism which separated the educated higher classes from others in the past and ‘enabled’ them culturally to commit suicide, have now spread to the whole society, turning the class ladder upside down. In France, the lower rates of suicide in poor areas led Emile Durkheim in 1897 to his famous hypothesis of the suicide-protective effect of poverty. However, later developments have shown that it did not stand the test of time (Rehkopf and Buka 2006).

There are differences in suicidal behaviour between the members of different occupational groups. Although Stack (2001) concludes, on the basis of studies in the USA and in England, that ‘most occupations tend neither to drive people to suicide nor to offer protection against suicide’, there are exceptions. However, the question has been studied in different circumstances and using different methodologies, and the studies have often been limited by the small numbers of suicide cases or otherwise insufficient data.

Studies analysing many occupational groups at once are not very numerous. In Stack's analysis in 21 US states in 1990, 12 occupations out of the 32 studied showed a significantly higher risk (PMR) for suicidal death, among them dentists, physicians, artists, and carpenters (Stack 2001). On the other hand, elementary school teachers, postal workers, and clerks were at a significantly lower risk. In Agerbo et al.'s (2007a) study of Denmark in 1991–1997, high suicide rates were observed among medical doctors, nurses, plant and machine operators, and drivers and mobile-plant operators, among others. Architects, engineers, and technicians were found to be at lowest risk.

Kposowa (1999), studying industrial and occupational groups in the USA in 1979–1989, found that persons employed in the mining industry had the highest relative suicide risk (RR), 4.39 times greater than that of the comparison group of finance, insurance, and real estate employees. Persons employed in construction and in business and repair services were also at a very high risk. When the material was divided according to the occupational positions within industries, only laborers differed significantly (with a twofold surplus risk) from the reference group. Burnley (1995) on the other hand, in a study performed in New South Wales in Australia in 1985–1991, found that farmers, transport and production workers and labourers had high suicide mortality, while miners and quarrymen showed significantly lower rates. Kagamimori et al. (2004) found similar results in their study of Japan between 1965–1995.

Why would there be a relationship between one's occupation and the propensity for suicide? Different explanations may apply in different cases. First, there may be a relatively direct relation between occupation-related stress and suicide in cases where very stressful situations are regularly encountered in one's work, or where a more permanent stressful situation, such as one's income being dependent on customers, is a part of the conditions. Labovitz and Hagedorn (1971), comparing client-dependent occupations with others in the USA, found a 60 per cent higher suicide rate among persons in such occupations.

There are also more temporary situations, where some occupational groups—for example, over-indebted Southern Indian cotton farmers (Stone 2002)—encounter grave crises and may then respond suicidally. Farmers are a group with a very specific, ‘lifestyle’ work whose suicide risks are often discussed when different crises afflict agriculture. However, it has also been found that the rationalization of agriculture has not influenced the suicide rates in Europe discernibly during the post-war period (Mäkinen and Stickley 2006), possibly because of the existence of sufficient alternatives.

Secondly, some occupations provide knowledge of the means of suicide, and sometimes also a privileged access to them. Doctors and care personnel (Wasserman 1992; Kelly et al. 1995; Hawton et al. 2000; Stack 2001, 2004; Hawton et al. 2004; Stark et al. 2006; Agerbo et al. 2007a) are probably those most studied in this regard. Stack (2004) estimates the suicide risk of US physicians to be 2.45 times higher than the average of the working-age population, after controlling for demographic factors, while the surplus risk for dentists (Stack 2001) was calculated to 5.43. Access to means of suicide is the most frequently quoted cause in the discussions of health care personnel suicide. However, other factors such as work-related stress and client dependency should not be dismissed.

Military personnel (Desjeux et al. 2004; Mahon et al. 2005), policemen (Schmidtke et al. 1999), and even the large agricultural populations of many developing countries (Nwosu and Odesanmi 2001; Khan 2002; CDC 2004; Recena et al. 2006) are also thus endangered, although research on military or police suicide does not always witness of higher rates among these groups (Stack and Kelley 1994; Loo 2003; Aasland et al. 2005; see also Chapter 36 in this book). Agricultural workers with access to highly toxic pesticides constitute the numerically largest identifiable group in the world at high risk for occupation-related suicide. However, there may also be cases of beneficial non-access to means of suicide. Schwartz (2006), for example, ascribes the low suicide rate of American college students to the fact that firearms have been banned from the campuses.

Thirdly, there is the selection of people into occupations—for example, character traits necessary or desirable for certain work may also be more prevalent among persons who are more prone to commit suicide. An example of such a connection is Wasserman's (1992) hypothesis that persons with depressive symptoms would more often be found among psychiatrists. Also, socially downwards-spiralling life careers often lead to certain kinds of work, of necessity rather than free will, which might then show high suicide rates. The high rates observed among unskilled workers may be partly due to such processes. To this category of causes could also in part be included what Stack (2001) calls ‘demographics’, i.e. the fact that different occupational groups are differently composed as regards sex, age, ethnicity, marital state, and other properties. Here, the class aspect of occupation is clearly visible. According to Agerbo et al. (2007a), class is the main factor behind the observed differences in suicidality between occupations.

Finally, more refined theories may attribute the high suicide rates in certain occupations to the fact that their members are exposed to some more general circumstances that are thought to generate suicide. People in commercial professions were, for example, especially exposed to anomie in Durkheim's classical theory, while those whose work did not comply with their other roles were supposed to suffer from status incompatibility (‘role strain’) according to Gibbs and Martin (1964). Such generalizations provide alternatives to the classification principles besides stress, access, and selection.

The everyday work environment, connected with but not determined by the previous aspects of one's labour market position, also seems to exercise influence on suicidal behaviour. Theoretically, that influence is of a different order—a work environment situation is in most cases changeable, while occupation may not be, and class is even more permanent (Erikson and Goldthorpe 1992). In addition, the working conditions themselves may change rapidly due to new technologies and organisational changes, the effects being different for different groups involved (Smith 1997; Wikman 2004). At the same time, the effects—including the health effects—of class are partly mediated through the quality of one's working environment (Borg and Kristensen 2000), and the same is probably true of occupations (Arnetz 2001).

Work environment is usually ascribed a more active role in the causation of suicide than class or occupation, and it also affects the individual's life in a more direct manner. Its main potential influences can be divided into the more permanent, related to the character of the work itself, the less permanent, pertaining to the actual workload, and those related to the social relations at the place of work.

In the Swedish Five County Study, Karasek and Theorell (1990) found that a work situation with few possibilities to learn new things was associated with a surplus risk for suicide or attempt of 2.44 among men, while hectic and monotonous work more than doubled the risk for women. Feskanich et al. (2002) found an association between both very high and very low occupational stress and suicide among US nurses, especially when combined with domestic stress. Even knowledge of dangerous substances present at the place of work could be counted into these more permanent stress factors—the Chernobyl nuclear accident clean-up workers are a salient example (Rahu et al. 2006).

In a study of Japanese work-related suicide cases, Amagasa et al. (2005) noted the heavy workloads and long working hours relevant in the context. Even positive processes such as company expansion can lead to negative outcomes (Westerlund et al. 2004). The very importance of work and the hierarchic nature of its organization can also make the workplace a potential source of deep conflict. The style of conflict resolution at the workplace has been linked to mental well-being in general (Hyde et al. 2006). In extreme cases the conflicts may result in bullying, which is undoubtedly psychologically detrimental but has so far been little researched in relation to adult workplaces (Leymann 1990).

A parent's work conditions may also affect the children: Aleck et al. (2006) report finding a relationship between fathers' adverse psychosocial work conditions and suicides (and suicide attempts) among their children in a cohort of Canadian sawmill workers.

In line with the classic theory of Durkheim, work can be understood as one of those important ties that integrate individuals into the larger society and regulate their everyday life, providing them with realistic means to fulfil some of their basic needs. Accordingly, just having a position in the labour market would in itself seem to be very important.

Not everyone works in a modern society. Even in the most employment-friendly societies, such as Denmark, ‘only’ 75 per cent of the working-age population is likely to be gainfully employed. The growing complexity of modern societies produces different groups of adults who for different reasons are permanently or temporarily outside the labour market. In addition to the traditional housewives, the severely ill and handicapped, conscript soldiers, and prisoners, we also have students, retired persons, unemployed persons, those on a disability pension, welfare recipients, asylum applicants, and others—there is little that is common for all these groups, except their not performing paid labour at the moment. Global comparisons are also more difficult here than in the case of social classes or occupations—the categories are very differently present in different societies, and those based on legal definitions are also often differently defined.

Of all the ‘outsider’ groups mentioned above, persons retired due to old age and students are the largest ones in contemporary Western societies. Persons older than 65 years typically constitute up to a fifth of the adult population (15+ years) in economically developed countries, while students of working age constitute roughly one-tenth of it. Depending on the society, housewives or disability pensioners may also be numerous, as may the unemployed in hard times. In Europe, Poland and Slovakia have lately shown unemployment rates around 15 per cent, corresponding to approximately 10 per cent of the entire working-age population. Together all these groups may easily constitute half of the adult population of any industrialized society.

The unemployed are probably the most researched group from the point of view of suicide. However, members of the other groups outside the labour market may live in equally detrimental situations.

Persons who commit suicide are generally less often employed than others (Heikkinen et al. 1995b; Agerbo 2005). A two-to-fourfold surplus risk for suicide has often been observed among the unemployed in different parts of the world, mostly even after controlling for other variables, in individual-level studies based on cases from national mortality registers (Iversen et al. 1987; Platt et al. 1992; Andrian 1996;, Kposowa 2001; Blakely et al. 2002, 2003; Qin et al. 2003), and even using twin registers (Voss et al. 2004). The risk is often assessed by comparing with the fully employed, or with some other average value. This, along with the different properties controlled for, may account for differing results. Despite the individual surplus risk, Qin et al. (2003) calculated, on the basis of Danish register data, a population attributable risk (PAR) of only 2.8 per cent to unemployment. Others (Platt et al. 1992: 4.5–9.6 per cent) have presented somewhat higher figures in a situation with higher unemployment.

The risk for suicide when unemployed is generally thought to be less for women than for men (Platt 1984; Platt et al. 1992; Qin et al. 2003), but this finding is not consistent (see Blakely 2003). Kposowa (2001) in her follow-up study points out that the surplus risk among unemployed women lasts for a longer time and even increases with time. Other studies suggest that the time between the onset of unemployment and the suicidal act is generally relatively short (Blakely et al. 2003).

It could seem that the suicide rates of the elderly in general would be influenced by factors other than work or retirement, since the form of the age curve of suicide seems to depend on the society in question. However, Girard (1993) argues that especially the propensity of men's suicides in achievement-oriented societies to increase towards old age witnesses of the fact that the status risks of life in these grow towards old age. Retirement, especially for men, seems to constitute a risky life event (Qin et al. 2000; Harwood et al. 2006). In Qin et al.'s study (2003), the suicide PAR measure for being an age pensioner was 10.2 per cent. However, since being an old age pensioner is almost equal with having reached certain age, that risk also contains the risk of being old.

Currently, more information is needed concerning disability pensioners, who seem to show very high rates of suicide in the places where they have been researched. Agerbo et al. (2007b) found in Denmark that their ‘rest’ group, consisting of persons neither employed nor unemployed, had a surplus suicide risk of 4–6 times among women, and 4–8 among men. In Qin et al.'s study (2003), the PAR measure for being on a disability pension was 3.2 per cent. It is possible, depending on the society in question, that this group contains many members whose reasons for being on a disability pension are connected to a high risk of suicide, such as severe mental illness or substance abuse (Karlsson et al. 2007).

There are also groups outside the regular labour market, such as students and homemakers, who do not generally show elevated rates of suicide (Niemi and Lönnqvist 1993; Schwartz 2006). Thus the risk for suicide due to being outside the labour market seems to depend on the position occupied by the individual. Compared with many other categories, students, housewives and also motivated conscript soldiers might be able to construct more positive life-roles out of their social positions, which also entail work, although usually unpaid.

In addition to individual-level studies, the covariation of a social environment with a higher or lower level of unemployment with suicide mortality has also been researched. The evidence on the question whether the level of unemployment in a social grouping (most often persons living in the same geographical area) is connected to that group's suicide rate has been generally negative in contemporary societies (Platt 1984; Zimmerman 1995; Mäkinen 1997; Platt and Hawton 2000), although there are exceptions (Martikainen et al. 2004), and some authors (Rehkopf and Buka 2006) suggest that the fault may lie in deficient study designs. The relationship between unemployment and suicide has also been researched with the help of time-series studies, where the rates of unemployment and suicide are related over time. Here the results are mixed: both significantly positive (Morrell et al. 1993; Gruenewald et al. 1995; Norström 1995; Gunnell et al. 1999) and non-significant relationships (Mäkelä 1996; Lucey et al. 2005) have been found.

Nevertheless, the exact effect of any social position on suicide is determined by the specific context of the society and time in question. While the effects of unemployment were hard to find in the unemployment crises of Scandinavia in the beginning of the 1990s (Hintikka et al. 1999; Hagquist et al. 2000), they seemed unquestionable in the same societies—and others—in the mass unemployment situations of the 1920s and 1930s (Norström 1995; Gunnell et al. 1999). The difference may lie in the far less dramatic social consequences of contemporary unemployment in welfare societies.

The ‘paradoxal’ theory, on the other hand, has stated that lower suicide rates among the unemployed should be expected in areas with high unemployment, since unemployment would be less of a social stigma in places and times where it is common. So far the evidence seems to speak against this theory (Martikainen et al. 2004; Ahs and Westerling 2006; Agerbo et al. 2007b).

The mechanisms thought to be responsible for the relationship between unemployment and suicide correspond to those proposed for the relationships between other work-related positions and suicide. Unemployment, a state of seeking but not currently having an important social position in life, is stressful and stigmatizing as such; furthermore, the blessings of work are missing, making the individual psychologically more vulnerable to any other adversities, and increasing the possibility of isolation, substance abuse, and general mental ill-health. All this can contribute to provoking emotional states conducive to suicidal behaviour.

Recent studies (Blakely et al. 2003; Gerdtham and Johannesson 2003) suggest that the weaker economic position per se does not cause the relationship between unemployment and suicide. Blakely et al. (2003) found instead that the main mechanism connecting unemployment to suicide is the fact that unemployment is likely to lead to worse mental health. In a Finnish study, Heikkinen et al. (1995a) found that unemployment, a common event preceding especially young men's suicide, was also associated with alcohol misuse.

However, selection, the possibility that some persons may end up both unemployed and suicidal due to a third factor, such as mental illness, is also generally thought to be relevant when discussing the relationship. The question is difficult to solve because prospective longitudinal studies, which could determine the correct chronological order of events, are very difficult to organize due to the long time intervals and the small number of cases involved. The question of causality between unemployment and suicide is being actively researched and discussed (see, for example, the August 2003 issue of the Journal of Epidemiology and Community Health).

Finally, unemployment is not only an individual issue but a macro-economic state of affairs. Norström (1995), interpreting the difference in the attributable risk for suicide due to unemployment between individual- and aggregate-level studies (the risk estimate in the latter being far higher), points out that unemployment in society has also indirect effects, affecting more persons than only the unemployed.

So far, this chapter has mostly considered suicide in relation to labour markets such as they appear in the countries of Europe, North America, and East Asia, where an overwhelming majority of the studies have been conducted. This is due to both the availability of statistics of suicide, and the register systems that allow us to track individuals' labour market status after their death. Without these, little can be said about differences in suicidality between social and occupational groups. Lacking the tools for more advanced studies, the sampling of consecutive cases at a morgue, for example, is a good alternative, but it can seldom produce so many cases that they could reveal more than the most obvious relations. Suicide waves pertaining to certain places and groups are reported, but their value for systematic research is at present uncertain.

The labour markets of the world are very different due to different natural environments, different development of the division of labour, and also different political steering and control systems modifying the markets in different societies. Moreover, they may also differ greatly within single countries, especially large ones, but even elsewhere as a result of the globalization process. Thus, the traditional division described below must be read with some caution, keeping in mind that even modern societies retain parts of the earlier ways of organization.

In those societies where market-based transactions have not fully replaced the earlier, gender- and kinship-based forms of labour division, the labour market is restricted by these factors; unemployment or retirement in the modern sense of the word do not really exist either. Where labour is performed as a part of socially and culturally ascribed role obligation, it is a part of individuals' set life careers. This organization may increase the risk for suicide greatly in those cases where the expected life career is not a desirable one. The situation of new brides in societies with arranged (or determined) marriages is an example of a such precarious situation, where one's kinship, social status, and labour follow a traditional ‘script’. Bleak prospects plague also people with a background in tribal societies (Bjerregaard and Lynge 2006; CDC 2007), whose traditional ways are often being threatened, rendering their customary work trivial or superfluous.

Industrial societies have replaced (and merged) the traditional forms of residence, labour, and thinking with modern urbanity, mobility, wage-labour, and secularization, all of which advance the role of an individual no more dependent from her kin and tradition. For the ‘rootless’ modern individual, ties to the greater society, such as work, become increasingly important, and losing one's work through unemployment or even retirement may be a difficult challenge. In western Europe, the change from agricultural to industrial society brought about a dramatic rise in suicide mortality during the nineteenth and early twentieth centuries (Cavan 1928); in other parts of the world similar processes have taken place later. It is, however, an open question whether this development is the fate of all societies or not.

In the most affluent, post-industrial societies, workers are typically more protected, and the detrimental effects of unemployment, for example, can to some extent be mitigated (cf. above). However, there are some visible trends on the labour market which affect everyone participating in the global economy, but which are thought to deteriorate the current work conditions, especially in these countries because they affect the attained job security. Globalization, the process of increasing communication over national borders, has since the 1990s reached a new phase due to the Internet and the flow of multinational information that it brings to those able to use it. Together with the worldwide movement towards ‘open trade’ it sharpens the competition on those markets where foreign goods and services can be brought in. Producers attempt to increase their own mobility in turn. In the process workplaces, still relatively immovable only half a century ago, are created, closed, and moved at an ever-increasing pace.

For the working population in economically advanced countries, the prospects are uncertain. They face increasing demands of ‘flexibility’, which increase their work-related stress and can in some cases make planned life outside work difficult. Their work may become a target for outsourcing, the move of specific tasks outside the mother company at home or abroad, and they may either become unemployed or have to accept new work conditions at a new employer. Another trend is the ‘outsourcing’ of even relatively regular work, thereby forcing employees to become small entrepreneurs, with the risks that this entails. All this insecurity has, of course, its psychological consequences (Benavides et al. 2000; Cheng et al. 2005), of which increasing suicidality can be one. However, research on these issues in relation to suicide is still largely absent.

Among those in the weakest positions on the labour market, insecurity is only a part of what is known as ‘underemployment’ of the working poor, including involuntary part-time work, poverty-level wages, and insecurity, which all erode the benefits of regular work, incomes, social status, and set daily routines (Dooley et al. 1996). Stark et al. (2006) speculate on the possibility that underemployment might be one cause for the over-representation of low-income persons among suicides.

It is difficult to say anything truly universal about the relationship between labour market positions and suicide, and very little globally oriented research has been undertaken. Unemployment, due to its generally negative nature and adverse financial consequences, may be considered a rather general risk factor for suicide, as the studies from many different countries (Koronfel 2002; Gururaj et al. 2004; Chen et al. 2006) show. Vijayakumar et al. (2005a) think that low socio-economic position might also be a universal risk factor.

However, the consequences of seemingly similar positions are not identical in different societies, and their effects on suicide are not similar either. The perceived importance of work in life varies between different societies, too. Kölves et al.'s (2006) finding that suicide after a job loss was more common in Tallinn (Estonia) than in Frankfurt am Main (Germany) might be an illustration of this. Also, studies of individual motives for suicide show that these vary greatly between different cultures (Bhatia et al. 1987). All this once again underlines the importance of careful, nuanced analyses of the relations between different social status positions and suicide worldwide.

The effectiveness of suicide-preventive measures is sometimes questioned, and the relative rarity of completed suicides is a constant obstacle for their scientific evaluation (Goldney 2005). In the national plans for suicide prevention so far in force, features related to working life can be divided into three categories: work and unemployment policies, reduction of work-related access to means of suicide, and improved access to help for substance abuse and other mental problems through place of work (Taylor, Kingdom and Jenkins 1997; Singh and Jenkins 2000; Wasserman et al. 2002, 2004).

It is uncertain to what extent policies aimed at reducing unemployment should be motivated by the reduced suicide risk (Mäkinen 1999; Goldney 2005). However, De Leo and Russel (2004) are of the opinion that measures pertaining to social and economic policy in times of crisis might be effective in preventing especially younger persons' suicides. Morrell et al. (2007) found in Australia a statistically significant relation between the National Youth Suicide Prevention Strategy and the decline in suicide, while there was no relation between the prevention strategy and unemployment. The authors suggest that this means that young male suicide and unemployment are no more correlated. According to Preti (2003), work is suicide-preventive for ‘mentally suffering’ persons. Yet Agerbo (2005), on the basis of his Danish study showing less suicidality among the unemployed who were admitted to psychiatric care, compared to the employed, does not think that this would be effective. In fact, work-related stress could worsen the existing mental problems.

Preventive measures targeting the access to suicide methods are both easier to implement and more effective than many other approaches (Gunnell and Frankel 1994). An obvious goal here is the reduction of the number of suicides committed with the help of pesticides in the rural areas of developing countries (Gunnell and Eddleston 2003; Vijayakumar et al. 2005b). The World Health Organization (WHO) announced a global public health initiative in 2005 to come to terms with this problem (Bertolote et al. 2006).

The workplace can also be used as a platform for suicide prevention, where a large number of persons can be reached. At the same time, it is difficult due to the relative rarity of suicide and the ensuing difficulties in motivating employers to preventive action (Lewis et al. 1997). This may be easier with very large employers such as the US Air Force, whose suicide-preventive program, seeking to facilitate recognition and treatment of psycho-social problems in 1997–2002, was accompanied by a 33 per cent reduction in the rate of suicide compared to the period before the intervention (Knox et al. 2003).

However, programs designed to identify and help persons dependent on alcohol or drugs are, due to their immediate usefulness, more likely to meet acceptance. Programs aiming at an early detection of depression are also of interest in this context (Wasserman 2006). Platt and Hawton (2000) suggest more general health promotion initiatives targeted at the occupations at risk, such as farmers and medical personnel, designed to promote help-seeking, reduce organisational stressors, and facilitate stress management. The retraining programs at closing industrial sites could also be used to that end.

Suicide prevention in relation to the labour market can entail both prevention of the negative consequences connected with some occupations and working environments and the use of the workplace as a platform for general suicide-preventive measures. The current national programmes focus mostly on unemployment policies and restrictions to the means of suicide. The programmes in workplaces promoting early detection of mental problems, depression, and alcohol/drug abuse should be encouraged.

There are many needs left to be fulfilled in this area of research. The most acute of these is the lack of data on suicide, which still hampers research in most countries. More systematic, replicable research is needed on social classes and occupations, and especially on the relation between the working environment and suicide. While unemployment studies belong to the more developed parts of suicide research, more knowledge is needed concerning both the ‘underemployed’ and the many different groups outside the labour market. Indirectly, this should lead to better recognition of what factors are important in having work (or not) in different societies. Obviously, any work-related suicide prevention programmes would have to take into account the reasons why certain groups have elevated suicide rates in order to better target their actions.