Main findings

Suicide rates have increased dramatically over the past 45 years despite prevention efforts. ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas40} There is marked geographic variability in suicide rates, with the highest rates being found in Eastern Europe and some US states and the lowest in Muslim and Latin American countries. ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas40,Reference Tondo, Albert and Baldessarini41} So far, this variability in suicide rates has not been satisfactorily explained. This variability is particularly true for the continent of Europe. Reasons for these great differences between national/regional suicide rates have not been fully explained yet. Geographic (latitude, longitude, altitude) climatic, dietary, genetic, economic, religious and other sociocultural differences can be taken into account. However, differences in psychiatric morbidity, the accuracy of the registration of suicide, the stigma associated with mental illness and suicide (possibly influencing help-seeking behaviour and reporting rates), the availability of lethal methods, and the availability of the social and healthcare systems should be considered. ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas40,Reference Rihmer, Kantor, Rihmer and Seregi42}

The current study analysed the correlation between suicide rates in 29 European countries and economic indices. Although a correlation is evident, the temporal relationships do not support a direct link between the economic crisis and any change in the suicide rates. It also found the following results.

1. (a) There was a nadir for suicide rates across Europe around the year 2007. This synchronisation cannot be considered to be random; instead some common aetiology should have influenced the rates across the continent.

2. (b) Even in countries with more or less stable rates, the specific nadir was evident.

3. (c) The nadir occurred more than a year before the onset of the economic crisis and the subsequent increase in suicide rates also occurred several months before the crisis.

4. (d) There was a strong correlation of suicide rates with all economic indices except GDP per capita in men but only with unemployment in females. For the vast majority of countries the suicide rates correlated strongly with GDP per capita and rather weakly with unemployment and the other indices. This correlation was strong within each country and less strong across countries.

5. (e) In Eurozone countries the correlation with GDP per capita was the dominant pattern. In the rest of EU countries the correlation with unemployment was also strong. In countries outside the EU, the correlation of suicide rates with economic variables was weak.

6. (f) The combined data from European countries for the years 2000-2010 (Table 1 and online Table DS1 and online Fig. DS1) suggest that the average suicide rate in the European region is similar to that reported for the USA.

Fig. 3 Patterns of change in suicide data in European countries in relation to the 2007 nadir.

Pattern A: the declining suicide rate is followed by a temporal increase after 2007 and then stabilises; in pattern B the declining suicide rate is interrupted by a temporal increase after 2007 and then the decline continues; in pattern C the declining suicide rate is reversed after 2007.

Several conclusions can be made on the basis of the above. First, these data clearly dispute the assumption that specifically changes in unemployment have a direct effect on suicide rates. The temporal sequence and correlation of events (suicide rise first, economic recession follows, synchronisation of suicide rate changes across the continent) suggests there is a close relationship between the economic environment and suicide rates; however this relationship is not that of a direct cause and effect. The findings reported by the present study concerning Europe are similar to those reported for the USA. In spite of claims that the rise in unemployment caused a rise in the suicide rate in the USA, ^{Reference Reeves, Stuckler, McKee, Gunnell, Chang and Basu34} a closer look at the data revealed that suicides rose first and unemployment followed. ^{Reference Fountoulakis, Koupidis, Siamouli, Grammatikopoulos and Theodorakis39} One could argue that those people who are going to lose their jobs are stressed months before this happens, but ‘fear’ of unemployment is quite different from unemployment per se, especially since such an assumption suggests that employed people take their own life before they become unemployed.

Second, it is certain that an ‘event’ occurred in the European continent after 2005 and had a profound effect on suicide rates, probably by halting the ongoing reduction in suicide rates. The strong correlation of these rates with GDP per capita and the weaker but still important correlation with the other economic indices suggest that this event was probably the prodromal phase of the economic crisis. Unfortunately, no data concerning this ‘prodrome’ can be found in the literature.

It is important however to place a question mark on the nature of the 2007 nadir in the suicide rates. This nadir was evident as a sharp decline from previous year even in countries with otherwise low and stable suicide rates until that time (such as Greece) or in countries with a robust increasing trend (such as Slovakia).

Third, an important and robust finding of the current study was the strong correlation of suicide rates with GDP per capita, although such an effect was not detected by the regression analysis. This correlation was of the magnitude of 0.80 and was found in the vast majority of countries with independent calculations. The correlation was strong within countries and weaker across countries (and also in the regression models) suggesting that the GDP is related to fluctuations in the suicide rates but may not be with the absolute baseline value. Unemployment was more important for countries outside the Eurozone area. This pattern is rather difficult to interpret.

The GDP was first developed in 1934 by Simon Kuznets ^{Reference Kuznets43} and the danger of using it as a measure of welfare was pointed out by the author. It was established as the main tool to measure a country’s economy in 1944 after the Bretton Woods conference. In spite of the early warnings, the GDP per capita is often used as a measurement of the standard of living. This is based on the assumption that all citizens would benefit from the economic growth of their country. However, essentially it is not such an indicator and it is not a measure of personal income, since it does not take into consideration the inequalities within a given country. Nevertheless, a similar index, the gross national income (GNI) per capita is included as a significant contributor to the Human Development Index (HDI). ⁴⁴ In this paper, all the economic indices checked by the authors (Gini index, private debt % of GDP, HDI) did not manifest any pattern that would relate them to the change in the suicide rates trends after 2007.

One previous study reported that cross-sectionally the GDP per capita had a tendency to correlate with suicide rates in males (r = 0.3, P = 0.06) but not in females (r = 0.2, P = 0.9). ^{Reference Sher45} Our results are in accord with this. However, the relationship of GDP with suicidality seems to be more complex and it might follow an inverted U-shaped curve, with suicide trends declining after peaking at a certain threshold of economic development. ^{Reference Moniruzzaman and Andersson46} At low GDP levels, increases in GDP are associated with increases in suicide rates, but once a given threshold of economic development is reached, further increases in GDP do not correlate with further increases in suicide rates. ^{Reference Voracek47}

Overall, there was a worldwide increase in the purchasing power parity-adjusted GDP per capita over the past 3 decades, and during the same time period the suicide rates have increased in developing Latin American and Caribbean countries and in several high-income Asian countries (such as Japan and South Korea), and have decreased in the majority of European countries and Canada. ^{Reference Blasco-Fontecilla, Perez-Rodriguez, Garcia-Nieto, Fernandez-Navarro, Galfalvy and de Leon48} Suicide rates in India were also positively correlated with GDP rates although the quality of the data is low and should be interpreted with caution. ^{Reference Soman, Safraj, Kutty, Vijayakumar and Ajayan49-Reference Jacob, Sharan, Mirza, Garrido-Cumbrera, Seedat and Mari53} Concerning Japan and South Korea, they are both high-income countries with universal health coverage but with the private health sector prevailing over public health resources. In South Korea, although life expectancy has rapidly increased, mortality as a result of suicide increased, particularly among men aged 30 years or older. ^{Reference Yang, Khang, Harper, Davey Smith, Leon and Lynch54}

It seems that economic growth is not invariably followed by a decrease in suicide rates, especially because income inequality causes them to increase. ^{Reference Chen, Choi and Sawada55} Also if economic growth is not accompanied by adequate infrastructures for mental health services, suicide rates might trend up. ^{Reference Jacob, Sharan, Mirza, Garrido-Cumbrera, Seedat and Mari53}

Suicide rates and mental illness

When studying the possible causes of suicide, one should have in mind that suicide is probably the end result of an interaction between many different risk factors with mental disorders being the decisive one. ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas40,Reference Qin, Agerbo and Mortensen56} It is solidly proven that over 90% of people who die from suicide have a mental illness. Mood disorders are found in 80-85% ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas40,Reference Rihmer, Kantor, Rihmer and Seregi42,Reference Gray and Otto57} and schizophrenia in 9-13% of people dying each year because of suicide. ^{Reference Meltzer58} The significance of mental disorders may be much smaller in low- and middle-income countries. ^{Reference Manoranjitham, Rajkumar, Thangadurai, Prasad, Jayakaran and Jacob59} Other risk factors in the field of psychiatry also exist, including personality disorders and substance and alcohol dependence, ^{Reference Oquendo, Bongiovi-Garcia, Galfalvy, Goldberg, Grunebaum and Burke60,Reference Comtois, Russo, Roy-Byrne and Ries61} and a family history of suicide. ^{Reference Oquendo, Bongiovi-Garcia, Galfalvy, Goldberg, Grunebaum and Burke60,Reference Cavazzoni, Grof, Duffy, Grof, Muller-Oerlinghausen and Berghofer62,Reference Hawton, Sutton, Haw, Sinclair and Harriss63} Ethnic group, ^{Reference Kupfer, Frank, Grochocinski, Houck and Brown64} problematic coping skills ^{Reference Johnson, Lydiard, Morton, Laird, Steele and Kellner65} and environmental variables such as recent psychosocial stress ^{Reference Leverich, Altshuler, Frye, Suppes, Keck and McElroy66,Reference Leverich, McElroy, Suppes, Keck, Denicoff and Nolen67} and occupational problems or interpersonal problems with spouses or romantic partners ^{Reference Tsai, Lee and Chen68} also constitute risk factors. The availability and access to lethal means (such as firearms) might be of importance. ^{Reference Rihmer, Kantor, Rihmer and Seregi42} Theoretically, any intervention that helps reducing these risk factors could ultimately reduce the suicide rate; however this has not been solidly proven for most of these variables. ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas40} Unfortunately, research on suicide is limited by the fact that the majority of suicide victims die at the first attempt. ^{Reference Rihmer, Belso and Kiss69,Reference Isometsa, Henriksson, Aro, Heikkinen, Kuoppasalmi and Lonnqvist70}

The first to suggest that the suicide rate depends on socioeconomic driving forces was Enrico Agostino Morselli in 1882. ^{Reference Morselli71} Since then, suicide has usually been considered a social problem, and several risk factors have been related to suicidal behaviour, both at the subject level (microsocioeconomic factors) and at the state level (macrosocioeconomic level). ^{Reference Qin, Agerbo and Mortensen56,Reference Lester and Yang72}

Suicide prevention

An example of a successful programme aiming to reduce suicide rates is the National Service Framework target set in the UK in 1999, which aimed to reduce suicides by at least 20% by 2010. This target was achieved. The interventions included awareness campaigns and encouragement particularly to general practitioners to recognise depression and treat it early, especially with the use of selective serotonin reuptake inhibitors. However, in the USA and the Netherlands the official warning in 2003-4 was against prescribing selective serotonin reuptake inhibitors for young people, and this was followed by a reduction in prescribing and a detected increase in suicides in that age group. ^{Reference Gibbons, Brown, Hur, Marcus, Bhaumik and Erkens73} In Slovenia an increase in antidepressant use lead to a decrease in the suicide rate ^{Reference Korosec Jagodic, Rokavec, Agius and Pregelj74} and this seems to be true for Sweden, ^{Reference Isacsson, Holmgren, Osby and Ahlner75,Reference Isacsson and Ahlner76} Hungary, ^{Reference Dome, Kapitany, Ignits, Porkolab and Rihmer77} Europe as a whole ^{Reference Gusmao, Quintao, McDaid, Arensman, Van Audenhove and Coffey78} and all Organisation for Economic Co-operation and Development countries. ^{Reference Kamat, Edgar, Niblock, McDowell and Kelly79,Reference Ludwig, Marcotte and Norberg80} It seems that in spite of potential problems, the use of antidepressants overall decreases suicide rates. ^{Reference Ludwig and Marcotte81}

Most of the risk factors are likely to be dependent on the victim’s behaviour and thus do not constitute independent factors, ^{Reference Isometsa, Heikkinen, Henriksson, Aro and Lonnqvist82} however the recent economic crisis constitutes a stress factor that is independent of the behaviour of the person, although people with specific behaviours (such as great risk-taking entrepreneurs) are likely to be more vulnerable to the crisis. On the other hand, specific cultures (such as Hispanic people in the USA) seem to have some kind of protective effects against suicidal behaviour. ^{Reference Oquendo, Dragatsi, Harkavy-Friedman, Dervic, Currier and Burke83} Additional support for this comes from the conclusion of a recent review that only the creation of social support networks reduces suicidality and that other interventions are of unproven effectiveness. ^{Reference Fountoulakis, Gonda and Rihmer84-Reference Gonda, Fountoulakis, Kaprinis and Rihmer87} Although it has been suggested that a reduction in unemployment through governmental action should lead to a reduction in suicidality, ^{Reference Stuckler, Basu, Suhrcke, Coutts and McKee88} this remains an unproven theoretical suggestion, and is not supported by the results of the current study.

Findings from other studies

In his seminal work in 1979, Brenner reported that for every 10% increase in unemployment there is an increase of 1.2% in total mortality, including an increase by 1.7% in suicidality. ^{Reference Brenner and Barrett89} Suicide rates are lower in Western high-income countries in comparison with low- and middle-income countries. ^{Reference Jacob90} The relationship of suicide rates with GDP suggests that suicide rates drop in times of economic expansion and increase in times of recession. ^{Reference Tapia Granados91} It has been argued that business cycles affect suicide rates in the USA with the overall suicide rate usually rising during recessions and falling during expansions. ^{Reference Luo, Florence, Quispe-Agnoli, Ouyang and Crosby25} In the past, economic crises have been correlated with increases in suicides, such as the Great depression, ^{Reference Swinscow21,Reference Morrell, Taylor, Quine and Kerr23,Reference Stuckler, Meissner, Fishback, Basu and McKee92,Reference Tapia Granados and Diez Roux93} the Russian crisis in the early 1990s ^{Reference Chang, Stuckler, Yip and Gunnell33} (although the data are not published reliably) and the Asian economic crisis in the late 1990s. ^{Reference Chang, Gunnell, Sterne, Lu and Cheng26,Reference Chang, Sterne, Huang, Chuang and Gunnell27} On the contrary, other authors suggested that actually recessions improve several health indicators. ^{94-Reference Ruhm96} Concerning the present economic crisis, it has been calculated that close to 5000 excess suicides occurred in the year 2009, with the increase consisting mainly of men of working age and with unemployment a direct causal factor. ^{Reference Chang, Stuckler, Yip and Gunnell33} A deterioration in mental health with increasing depression and anxiety rates has been reported after the economic crisis in Hong Kong, ^{Reference Lee, Guo, Tsang, Mak, Wu and Ng97} south Australia, ^{Reference Shi, Taylor, Goldney, Winefield, Gill and Tuckerman98} Greece, ^{Reference Economou, Madianos, Peppou, Patelakis and Stefanis99} UK ^{Reference Katikireddi, Niedzwiedz and Popham15} and Spain, ^{Reference Gili, Roca, Basu, McKee and Stuckler30} and the effect seemed more severe in population groups who experienced unstable employment or financial problems. ^{Reference Gili, Roca, Basu, McKee and Stuckler30,Reference Lee, Guo, Tsang, Mak, Wu and Ng97,Reference Shi, Taylor, Goldney, Winefield, Gill and Tuckerman98} However, the methodology of these studies cannot differentiate between general distress and clinical mood disorders and thus any link of these results with the suicide rates is problematic.

Variations between countries in suicide rates

There are considerable variations in the effect of the crisis on the suicide rate across countries and it is unclear whether these variations relate to the severity of the recession as well as to varying social support and labour market protections in different countries, as it has been previously suggested. ^{Reference Stuckler, Basu and McKee12,Reference Chang, Stuckler, Yip and Gunnell33,Reference Stuckler, Basu, Suhrcke, Coutts and McKee88} We were also unable to reproduce the finding that there are stronger associations between increases in national suicide rates and unemployment rates in countries with low baseline unemployment rates than in countries with high unemployment rates. ^{Reference Chang, Stuckler, Yip and Gunnell33} On the contrary, we found that the baseline suicide rate was correlated with the overall pattern of the curve of suicide rates after the nadir year.

Social capital

One area that deserves further research is that of the so called ‘social capital’. Overall, the results of the current study imply that there were profound changes within the society of all European countries that preceded the economic crisis, however they were probably related to it or they might even constitute part of its aetiopathogenesis. The ‘social capital’ theory refers to the importance for the community of building generalised trust and, at the same time, the importance of individual free choice, in order to create a more cohesive society. It could be defined as ‘the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition’. ^{Reference Portes100-Reference Ferragina102} A decline in social capital is considered to be at the core of modern socioeconomic problems. ^{Reference Jain, Goyal, Fox and Shrank103-Reference Putnam105} Social cohesion has already been recognised as a factor influencing the suicide rate. ^{Reference Balint, Dome, Daroczi, Gonda and Rihmer106}

Limitations

The current study has a number of limitations. It is an observational analysis based on aggregate data collected from national statistical agencies. Probably there are differences between countries both in the quality of the data as well as in the level of misclassification of suicide, and these could lead to potential bias between countries, ^{Reference Kapusta, Tran, Rockett, De Leo, Naylor and Niederkrotenthaler107} but it is not expected they had a significant impact on the results of the current study.

Cross-level bias and aggregation bias are typical of studies similar to the current one. ^{Reference Moniruzzaman and Andersson108} The effects observed at the aggregate level might be modulated by the ecological context at the level of the individual. ^{Reference Lucey, Corcoran, Keeley, Brophy, Arensman and Perry109} Also time series data are frequently non-stationary and vulnerable to random findings. ^{Reference Lucey, Corcoran, Keeley, Brophy, Arensman and Perry109} Another source of bias is possible registration bias concerning suicides between countries and over time, and also concerning the quality of the economical statistics.

Finally, an important fundamental problem is that it is probably too early to arrive at conclusions concerning the impact of the current economic crisis on health, mental health and the suicide rate in particular. It seems necessary to wait until data up to at least 2020 are gathered in order to have a complete picture.

The authors chose to publish the full data-set their analysis was based on in an online table as they strongly believe that this database should be publicly available. One of the major obstacles of the current study was the difficulty of gathering statistical data from the various countries, although these data should had been easily accessible to every citizen. The authors believe that the publication of this data-set so that anyone can perform further analysis is one of the major contributions of the current study.