Scholarly article on topic 'Social capital and suicide in 11 European countries: an ecological analysis'

Social capital and suicide in 11 European countries: an ecological analysis Academic research paper on "Sociology"

0
0
Share paper
OECD Field of science
Keywords
{""}

Academic research paper on topic "Social capital and suicide in 11 European countries: an ecological analysis"

Soc Psychiat Epidemiol (2009) 44:971-977 DOI 10.1007/s00127-009-0018-4

ORIGINAL PAPER

Social capital and suicide in 11 European countries: an ecological analysis

Brendan D. Kelly • Mary Davoren ■ Aine Ni Mhaolain • Eugene G. Breen • Patricia Casey

Received: 4 September 2008/Accepted: 18 February 2009 / Published online: 11 March 2009 © Springer-Verlag 2009

Abstract

Background ''Social capital'' refers to the existence of voluntary community networks and relationships based on trust, and the use of these networks and relationships to enable positive social action. Social capital is positively associated with selected indices of mental health. Methods We performed an ecological investigation of the relationship between social trust (as one component of social capital) and national suicide rates in 11 European countries (n = 22,227).

Results There was an inverse relationship between social trust and national suicide rates (i.e. the higher the social trust, the lower was the suicide rate), after controlling for gender, age, marriage rates, standardised income and reported sadness. Conclusions Social capital may have a protective effect against suicide at the national level. Multi-level analysis, taking into account both group-level and individual-level variables, would help clarify this relationship further and guide appropriate interventions at both the group and individual levels.

Keywords Suicide • Social medicine • Community networks • Social capital

Background

The term ''social capital'' [1] refers to the existence of voluntary community networks and relationships based on

B. D. Kelly (El) • M. Davoren • A. N. Mhaolain • E. G. Breen • P. Casey

Department of Adult Psychiatry, University College Dublin, Mater Misericordiae University Hospital, 62/63 Eccles Street, Dublin 7, Ireland e-mail: brendankelly35@gmail.com

trust, and the use of these networks and relationships to enable positive social action. Examples of activities associated with high social capital in a society include volunteering for charities, engaging in ''community alert'' schemes, organising local football leagues and voting in local elections. There is growing evidence of a positive relationship between social capital [1] and good mental health [2-6]. There is, however, a paucity of research about social capital and suicide.

Kushner and Sterk [7] re-analysed Emil Durkheim's original data on suicide and social cohesion and concluded that the rates of suicide are often greatest among communities with high levels of social cohesion. Neeleman et al. [8] studied the rates of deliberate self-harm in 73 south London electoral wards and found that the rates of deliberate self-harm in minority ethnic groups relative to the white group were low in some areas and high in others. This suggests that the relationship between ethnicity and deliberate self-harm is both significant and complex, and may be affected by additional factors such as ethnic density, status integration and the extent to which membership of a given ethnic group offers psychological and social support to the individual, or increased levels of social capital to the community.

In this ecological study, we used data from the European Social Survey (ESS) to investigate the relationship between suicide rates and social trust (as a component of social capital) in 11 European countries.

Methods

The European Social Survey (ESS) is an academically driven social survey designed to study attitudes, beliefs and behaviour patterns in European countries [9]. Full ESS data

are available on an open-access basis (www. europeansocialsurvey.org). We used ESS data relating to 11 European countries: Switzerland, Germany, Spain, Finland, France, the UK, Norway, Poland, Portugal, Sweden and Slovakia. These ESS data were collected between 2003 and 2004. We stratified data by country and gender, and extracted mean age, proportion of persons who were married, mean income and reported sadness. Income was categorised into 12 categories (category 1 was less than Sterling £1,312 per year; category 12 was greater than £87,432). Income was standardised using purchasing power parity (for actual individual consumption) conversion rates (US $, 2003) from the Organisation for Economic Cooperation and Development (http://stats.oecd. org/wbos/Index.aspx?datasetcode=SNA_TABLE4). Sadness was measured by asking each participant how often they felt sad in the previous week (a score of 1 meant ''none or almost none of the time''; 4 meant ''all or almost all of the time'').

We used three questions from the ESS to assess perceptions of social trust as a measure of social capital, as previously described by Von dem Knesebeck et al. [10]. The three questions were:

• ''Generally speaking, would you say that most people can be trusted, or that you cannot be too careful in dealing with people? Please tell me on a score of 0-10, where 0 means you cannot be too careful and 10 means that most people can be trusted.''

• ''Do you think that most people would try to take advantage of you if they got the chance, or would they try to be fair?''

• ''Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves?''

The three questions used bipolar 11 point scales; an overall score for social trust was calculated by summing the three items (i.e. the higher the score, the higher was the social trust).

We obtained national suicide rates from the World Health Organisation (WHO) (http://www.who.int/mental_ health/prevention/suicide/country_reports/en/index.html). The WHO suicide data (2002-2004) were selected to coincide as closely as possible with the ESS data (20032004).

We analysed data using the Statistical Package for the Social Sciences [11]. We analysed correlations using Pearson's test (r). We performed linear regression analysis with national suicide rate (per 100,000 population per year) as the dependent variable; the independent variables were age, proportion of persons who were married, standardised income, sadness and social trust.

Results

This analysis included 22,227 participants (10,443 male, 10,784 female) from 11 European countries (Table 1). The mean age ranged from 42.57 years (males, Poland) to 52.10 years (females, Portugal). The proportion of persons who were married ranged from 57% (males, Sweden) to 80% (females, Portugal). The mean standardised income ranged from US $327.29 (females, Portugal) to US $848.39 (males, UK). Mean scores for sadness ranged from 1.25 (males, Finland) to 1.94 (females, Slovakia). Social trust ranged from 12.18 (males, Poland) to 20.60 (females, Norway). National suicide rates ranged from 3.3 suicides per 100,000 population per year (females, UK) to 31.7 (males, Finland). None of these differences between groups was statistically significant (P > 0.05).

The proportion of persons who were married was positively correlated with age (Pearson's r = 0.528, P < 0.05) and sadness (r = 0.654, P < 0.01), and negatively correlated with standardised income (r = -0.477, P < 0.05) and suicide rate (r = -0.512, P < 0.05; Table 2). Social trust was positively correlated with standardised income (r = 0.755, P < 0.01) and negatively correlated with sadness (r = -0.713, P < 0.01). There was no statistically significant bi-variable correlation between social trust and suicide rate (r = -0.004, P > 0.05) (Fig 1). On multivariable analysis (adjusted r2 = 47.9%), the suicide rate was inversely associated with sadness (P = 0.004) and social capital (P = 0.013; Table 3).

Discussion

This ecological analysis showed an inverse relationship between social trust (as a component of social capital) and national suicide rate; i.e. the higher the social capital, the lower was the suicide rate. This relationship was not apparent on bi-variable testing, but apparent on multivariable testing. This suggests that at least one of the factors additionally controlled for in the multi-variable analysis (age, proportion of persons who were married, standardised income, sadness) acts as either a negative confounder or an effect-modifier in the relationship between social trust and suicide rate, at the national level.

In order to be a confounder, the relevant factor would have to be independently related to both variables of interest (social trust, suicide rate) and must not lie on a causal pathway between them. In our analysis, sadness is the only factor that is independently related to both social trust and suicide rate. But sadness does not fulfil the other criterion for being a negative confounder because it is conceivable that sadness lies on a causal pathway between

Table 1 Gender, age, marital status, income, sadness, social trust and suicide rates for males and females in 11 European countries Country Gender n Mean age Proportion of Mean Standardised Sadness in Social Suicide rate per

(years) married persons income categorya mean income (US $)b the past weekc trustd 100,000 population per year

Switzerland Male 815 49.39 0.71 8.79 817.90 1.41 17.67 23.7

Female 988 50.59 0.77 8.48 776.08 1.55 17.99 11.3

Germany Male 1,437 47.37 0.65 6.81 711.80 1.38 15.34 19.7

Female 1,479 48.15 0.73 6.37 633.35 1.59 15.49 6.6

Spain Male 902 44.82 0.61 6.23 754.06 1.51 14.89 12.6

Female 974 47.26 0.71 5.89 676.70 1.74 15.05 3.9

Finland Male 919 47.45 0.64 7.28 688.98 1.25 18.95 31.7

Female 977 49.95 0.70 6.75 604.88 1.33 19.67 9.4

France Male 930 47.50 0.67 6.93 733.20 1.45 14.69 27.5

Female 1,056 48.79 0.70 6.41 640.48 1.71 14.81 9.1

UK Male 1,079 48.82 0.67 7.55 848.39 1.49 16.82 10.8

Female 1,315 50.52 0.74 6.99 740.16 1.61 16.92 3.3

Norway Male 891 45.30 0.62 8.83 847.65 1.27 19.27 15.8

Female 859 46.50 0.67 8.48 799.04 1.39 20.60 7.3

Poland Male 815 42.57 0.67 3.58 360.94 1.57 12.18 27.9

Female 906 45.52 0.75 3.42 331.79 1.80 12.60 4.6

Portugal Male 863 49.49 0.73 4.95 451.07 1.67 12.92 17.5

Female 1,359 52.10 0.80 4.39 327.29 1.90 12.80 4.9

Sweden Male 951 46.21 0.57 7.64 758.67 1.27 18.67 19.5

Female 975 48.19 0.65 7.13 672.18 1.50 19.21 7.1

Slovakia Male 841 42.70 0.65 4.06 497.10 1.83 13.00 23.6

Female 896 44.07 0.74 3.74 420.14 1.94 13.23 3.6

Data on gender, age, marital status, income, sadness and social trust (as one component of social capital) are derived from the European Social Survey (ESS) [9]; data on national suicide rates are derived from the World Health Organisation (http://www.who. int/mental_health/prevention/suicide/country_reports/en/index.html)

a Income was classified into 12 categories (category 1 was less than Sterling £1,312 per year; category 12 was greater than £87,432) b Income was standardised using purchasing power parity (for actual individual consumption) conversion rates (US $, 2003) from the Organisation for Economic Cooperation and Development (http://stats.oecd.org/wbos/Index.aspx?datasetcode=SNA_TABLE4) c Sadness was measured by asking each participant how often they had felt sad in the previous week (a score of 1 meant "none or almost none of the time''; 4 meant "all or almost all of the time'') [9]

d We used three questions from the ESS to assess perceptions of social trust as one component of social capital, as previously described by Von dem Knesebeck et al. [10] (see Methods)

social trust and suicide rate; i.e. social trust at the national level could affect the levels of sadness at the national level, which in turn could affect the suicide rate at the national level. On this basis, it is not possible to conclude that sadness is a simple negative confounder of the relationship between social trust and suicide, although it still appears to be an effect-modifier, the precise significance of which requires further study. It is also noteworthy that while sadness has a negative correlation with social trust, it also has a negative correlation with suicide rate; i.e. the lower the sadness level, the higher is the suicide rate. The latter, counter-intuitive correlation merits closer examination both at the national level (to seek to explain it) and at the individual level (to see if it holds true at the individual level).

Our study has several strengths: we report data relating to 11 European countries; we controlled for multiple variables through stratification (gender) and multi-variable linear regression analysis (age, proportion of persons who are married, standardised income, sadness); and our final model accounted for 47.9% of variance in the national suicide rates. Our study also has several limitations: while we took into account certain independent variables (e.g. social trust, standardised income), we did not take into account some others (e.g. inequality); and while we controlled for "sadness" in the week prior to the ESS interview, we did not control for formal mental illness (e.g. rates of depressive disorder). Our study just looked at one component of social capital (social trust); further studies are needed to examine the other dimensions of

Table 2 Correlation matrix (Pearson's coefficients) for age, proportion of persons who were married, standardised income, sadness, social trust and suicide rates in 11 European countries

Mean Proportion of Standardised Sadnessb Social Suicide rate per 100,000

age married persons incomea trustc population per year

Mean age - - - - - -

Proportion of married persons 0.528* - - - - -

Standardised incomeb 0.199 -0.477* - - - -

Sadnessc -0.063 0.654** -0.713** - - -

Social trustd 0.276 -0.380 0.755** -0.779** - -

Suicide rate per 100,000 population -0.315 -0.512* 0.118 -0.475* -0.004 -

Data on age, marital status, income, sadness and social trust (as one component of social capital) are derived from the European Social Survey (ESS) [9]; data on national suicide rates are derived from the World Health Organisation (http://www.who.int/mental_ health/prevention/suicide/country_reports/en/index.html); Pearson's coefficients (r) are shown

a Income was standardised using purchasing power parity (for actual individual consumption) conversion rates (US $, 2003) from the Organisation for Economic Cooperation and Development (http://stats.oecd.org/wbos/Index.aspx?datasetcode=SNA_TABLE4) b Sadness was measured by asking each participant how often they had felt sad in the previous week (a score of 1 meant "none or almost none of the time''; 4 meant "all or almost all of the time'') [9]

c We used three questions from the ESS to assess perceptions of social trust as one component of social capital, as previously described by Von dem Knesebeck et al. [10] (see Methods) * Statistically significant with P < 0.05 (two-tailed) ** Statistically significant with P < 0.01 (two-tailed)

social capital (e.g. voluntary work). In addition, social capital is a property of communities rather than individuals and is, therefore, an essentially ''ecological'' concept; as a result, all studies of social capital will be subject to the inherent strengths and limitations of the ecological study design.

We used ecological data from two different sources (ESS and WHO), which may have limited the comparability; i.e. whereas ESS data are based on samples from each country, WHO suicide rates relate to each entire country. It is noteworthy, however, that ESS samples were generally large (in this analysis, n = 22,227). Nonetheless, the validity of our findings is still dependent on the validity of the ESS methodology. While there have been several studies of the general validity of the ESS (e.g. pre-testing of the questionnaire using interaction analysis) [12], it remains possible that the validity of at least some of the questions varies among countries; the ongoing assessments of validity and continuous quality-improvement strategies within the ESS methodology are likely to minimise, but not entirely eliminate, this concern.

Notwithstanding these considerations, our data suggest that social trust (as one component of social capital) is negatively correlated with suicide rate, at the national level. This is consistent with some, but not all, of the existing literature in relation to social capital and mental health. Weitzman and Kawachi [2], for example, found that students on campuses with higher-than-average levels of social capital had a 26% lower individual risk for binge

drinking than their peers at other colleges. Boydell et al. [6] found an inverse association between perceived social cohesion and the incidence of psychosis in South London, while Rosenheck et al. [4] reported that areas with high social capital offered better housing to the homeless mentally ill. McCulloch [3] found that people in the lowest categories of social capital had increased psychiatric morbidity and Murray et al. [13] reported that social participation was positively associated with various indices of mental health, including positive affect and satisfaction with life.

The relationship between social capital and mental health is, however, likely to be complex: Mitchell and LaGory [14] found that high levels of ''bonding'' social capital were associated with increased levels of mental distress. McKenzie et al. [5] noted that people with psychoses in areas of high perceived community safety showed higher hospital readmission rates than those in areas of low perceived safety. Kushner and Sterk [7] re-analysed Durkheim's original data on suicide and social cohesion and concluded that the rates of suicide are often greatest among communities with high levels of social cohesion. Kusher and Sterk [7] draw particular attention to Durkheim's classification of military suicide as ''altruistic'' (e.g. sacrificing oneself for one's colleagues in battle) rather than ''fatalistic'', thus effectively eliminating this group of suicides from further consideration in his work. Rather than resembling Durkheim's ''altruistic'' suicide, however, military suicide more closely resembles

> 30.00 -<ij

g- 20.00 -

(5 10.00-a)

12.00 14.00 16.00 18.00 20.00 22.00

Social trust

Fig. 1 Scatter-plot of social trust and suicide rates in 11 European countries. This scatter plot charts suicide rate per 100,000 population per year (Y-axis) against social trust (as one component of social capital; X-axis). We used three questions from the ESS [9] to assess perceptions of social trust as one component of social capital, as previously described by Von dem Knesebeck et al. [10] (see Methods). Data on national suicide rates are derived from the World Health Organisation (http://www.who.int/mental_health/prevention/ suicide/country_reports/en/index.html). On bi-variable analysis, there was no statistically significant correlation between social trust and suicide rate (r = -0.004, P > 0.05; Table 2), but on multi-variable, linear regression analysis (after controlling for gender, age, proportion of persons who were married, standardised income and sadness), there was a statistically significant inverse relationship between social trust and suicide rate (P = 0.013; i.e. the higher the social trust, the lower was the suicide rate) (Table 3)

Durkheim's "fatalistic" suicide, occurring in a setting of high regulation, close interaction and, arguably, high social capital; i.e. within a close-knit community with considerable inter-dependence, shared goals and common activities.

When these, and other similar factors identified by Kushner and Sterk [7], are taken into account, Durkheim's data do not provide strong support for a simple negative correlation between social cohesion and suicide rate. At a conceptual level, it is useful to note that our analysis, consistent with Kushner and Sterk's re-analysis of Durkheim's data [7], did not identify a simple, significant bivariable correlation between social trust and suicide rate: we only identified the relationship between social trust and suicide rate on multi-variable analysis, which also provided evidence that this relationship is modified by ''sadness'', at the national level. It is important to note, however, that Durkheim's data were collected, analysed and interpreted at various different levels (individual, community and national), while our data were analysed at the national level: direct comparison of results from these differing levels of analysis run the risk of ecological fallacy.

The ecological fallacy occurs when conclusions obtained through the analysis of data at the level of groups

(e.g. the conclusions in this study) are applied directly at the level of the individual. In order to avoid this error, conclusions drawn from ecological studies should be (1) applied at the level at which analysis occurred (in this study, at the national level) and/or (2) used, mindfully, to suggest directions for future research at the individual level. Our findings, for example, suggest a relationship between social trust and lower suicide rates at the national level, but, because of the ecological nature of our analysis, our findings do not indicate whether or not the individuals who experience low social trust are the same as those who die by suicide.

Our findings do suggest, however, that it may be useful to conduct studies at the individual level to see if this is the case. It may, for example, prove to be the case that low social trust has a contextual effect; i.e. it is not the individuals experiencing low social trust who necessarily die by suicide themselves, but the presence of such individuals in a community is associated with an overall increase in the risk of suicide amongst all members of the community. If this were the case, one might expect significant variation in the rates of suicide and/or attempted suicide amongst communities, such as those reported in 73 south London electoral wards [8]; this merits closer study. In addition, the effect of social trust may, in turn, vary amongst communities; Fitzpatrick et al. [15], for example, provide that the psychological benefits of social capital may not apply equally to all groups within a given community (e.g. homeless individuals).

It is, again, important to bear in mind the ecological fallacy in the interpretation of these findings. A range of individual-level risk factors have already been identified in relation to both suicide and attempted suicide, and it is possible that social capital is related to some of these; the elucidation of these inter-relationships would require a multi-level analytic approach, which would take appropriate, simultaneous account of both group-level and individual-level risk factors. Our study suggests a relationship between social capital and sadness at the national level, but multi-level study is required to elucidate the inter-relationships between this kind of group-level factor and various relevant individual-level factors, such as age, gender, mental illness, substance misuse and physical illness [16-18].

Conclusions

Our analysis suggests that there is an inverse relationship between social trust (as one component of social capital) and suicide rate at the national level in Europe; i.e. the higher the level of social trust in a country, the lower s the suicide rate. This is an ecological finding that is applicable

Table 3 Multi-variable, linear regression analysis of age, age, proportion of persons who were married, standardised income, sadness and social trust in relation to suicide rates in 11 European countries

Variable b Standard 95% confidence t P

error intervals

Constant 163.217 35.722 87.490 238.944 4.569 0.000

Mean age -0.415 0.904 -2.332 1.501 -0.460 0.652

Proportion of married persons -7.436 53.024 -119.841 104.969 -0.140 0.890

Standardised incomea -0.004 0.014 -0.034 0.026 -0.289 0.777

Sadnessb -49.958 14.731 -81.187 -18.729 -3.391 0.004

Social trustc -2.777 0.999 -4.895 -0.658 -2.779 0.013

Data on age, marital status, income, sadness and social trust (as one component of social capital) are derived from the European Social Survey (ESS) [9]; data on national suicide rates are derived from the World Health Organisation (http://www.who.int/ mental_health/prevention/suicide/country_reports/en/index.html); suicide rate per 100,000 population per year is the dependent variable in this analysis; there were 22 groups in this analysis (one male group and one female group for each country); adjusted r2 for the model is 47.9% a Income was standardised using purchasing power parity (for actual individual consumption) conversion rates (US $, 2003) from the Organisation for Economic Cooperation and Development (http://stats.oecd.org/wbos/Index.aspx?datasetcode=SNA_TABLE4) b Sadness was measured by asking each participant how often they had felt sad in the previous week (a score of 1 meant "none or almost none of the time''; 4 meant "all or almost all of the time'') [9]

c We used three questions from the ESS to assess perceptions of social trust as one component of social capital, as previously described by Von dem Knesebeck et al. [10] (see Methods)

at the national level, but which also suggests a useful direction for further research at the individual level. Both quantitative and qualitative research methodologies would be helpful in determining the extent and nature of the relationship between social capital and suicide rates, and the precise role of sadness in modifying the relationship at the national level. Multi-level analysis, which would take simultaneous account of group-level and individual-level variables, would be especially useful in determining the appropriate mix of public health and individual-level interventions likely to assist in better understanding and addressing problems related to suicide at both the national and individual levels.

Acknowledgments The authors are grateful for the comments of the members of the Outcomes of Depression International Network (ODIN) on an earlier version of this paper. This paper uses data from the ESS, an academically driven social survey designed to chart and explain the interaction between Europe's changing institutions and the attitudes, beliefs and behaviour patterns of its diverse populations. It is funded via the European Commission's 5th and 6th Framework Programmes, the European Science Foundation and national funding bodies in each country. The project is directed by a central co-ordinating team led by Roger Jowell at the Centre for Comparative Social Surveys, City University, London (9).

Conflict of interest statement None.

References

1. Baum F (2000) Social capital: is it good for your health? Issues for a public health agenda. J Epidemiol Community Health 53:195-196

2. Weitzman ER, Kawachi I (2000) Giving means receiving: the protective effect of social capital on binge drinking on college campuses. Am J Public Health 90:1936-1939

3. McCulloch A (2001) Social environments and health: a cross-sectional survey. BMJ 323:208-209

4. Rosenheck R, Morrissey J, Lam J, Calloway M, Stolar M, Johnsen M, Randolph F, Blasinsky M, Goldmann H (2001) Service delivery and community: social capital, service systems integration, and outcomes among homeless persons with severe mental illness. Health Serv Res 36:691-710

5. McKenzie K, Whitley R, Weich S (2002) Social capital and mental health. Br J Psychiatry 181:280-283

6. Boydell J, McKenzie K, van Os J, Murray R (2002) The social causes of schizophrenia: an investigation into the influence of social cohesion and social hostility. Schizophr Res 53(s):264

7. Kushner HI, Sterk CE (2005) The limits of social capital: Durkheim, suicide, and social cohesion. Am J Public Health 95:1139-1143

8. Neeleman J, Wilson-Jones C, Wessely S (2001) Ethnic density and deliberate self-harm: a small area study in southeast London. J Epidemiol Community Health 55:85-90

9. Jowell R, Team CentralCo-ordinating (2005) European Social Survey 2004/2005: Technical Report. Centre for Comparative Social Surveys, City University, London

10. Von dem Knesebeck O, Dragano N, Siegrist J (2005) Social capital and self-rated health in 21 European countries. GMS Psychosoc Med 2: Doc02. Available at: www.egms.de/en/ journals/psm/2005-2/psm000011.shtml

11. Inc SPSS (2003) SPSS 12.0 Base Users Guide. Prentice-Hall Regents, Upper Saddle River, New Jersey

12. Ongena Y (2003) Pre-testing the ESS-questionnaire using interaction analysis. Centre for Comparative Social Surveys, City University, London. http://www.europeansocialsurvey.org/index. php?option=com_content&task=view&id=62&Itemid=96

13. Murray G, Judd F, Jackson H, Fraser C, Komiti A, Pattison P, Wearing A, Robins G (2007) Ceremonies of the whole: does social participation moderate the mood consequences of neu-rotocism. Soc Psychiatry Psychiatr Epidemiol 42:173-180

14. Mitchell CU, LaGory M (2002) Social capital and mental distress in an impoverished community. City Community 1:199-222

15. Fitzpatrick KM, Irwin J, Lagory M, Ritchey F (2007) Just thinking about it: social capital and suicide ideation among homeless persons. J Health Psychol 12:750-760

16. Williams M (1997) Suicide and attempted suicide. Penguin Books, London

17. Rhodes AE, Bethell J, Spence J, Links PS, Streiner DL, Jaakkimainen RL (2008) Age-sex differences in medicinal

self-poisonings: a population-based study of deliberate intent and medical severity. Soc Psychiatry Psychiatr Epidemiol 43:642-652 18. Hidaka Y, Operario D, Takenaka M, Omori S, Ichikawa S, Shirasaka T (2008) Attempted suicide and associated risk factors among youth in urban Japan. Soc Psychiatry Psychiatr Epidemiol 43:752-757