This is the first study which aims to identify the cross-country economic burden of suicide in general population by estimating the production losses associated with suicide deaths in 28 EU countries. By applying a HCA model (designed [15] and used previously for alcohol-attributable mortality [16]) and Eurostat data, I obtained highly comparable estimates which reflect the specificity of economic burden experienced in particular countries.
The total production losses attributable to suicides at working age in the EU in 2015 were €8.46 billion, being a result of 39,265 suicides and 745 thousand years of productive life lost. A notable geographical difference was observed among the member states both in terms of the suicide rates and cost itself. All in all, the countries experiencing the highest relative burden of suicides were three Baltic states (Estonia, Lithuania, Latvia), Poland and Finland. On the other hand, the ones with the lowest detrimental effect of suicides were Cyprus, Greece, Italy, Spain and Bulgaria. Therefore, the countries located in the southern regions of Europe experienced minor health and economic burden of suicide compared to those situated in the north. This clearly reflects the geographical distribution of suicide mortality, particularly low death rates in the Mediterranean states and higher rates in Scandinavia and the Baltic post-Soviet republics. However, this mortality pattern is strengthened in terms of production losses because the NE states generally have higher employment rates (Denmark, Sweden, Latvia) and/or longer time of economic activity (Estonia, Ireland, Sweden) and/or higher prospects of future economic growth (Lithuania, Latvia, Estonia) than the SE countries. On the other hand, the greatest relative economic burden (measured by the share of GDP lost) in the CEE countries arises mainly from high suicide rates among men and dynamic projected economic growth which make future losses greater.
The results show that the cost of suicide is mainly driven by males’ death and this obviously reflects the sex disparities in suicide incidence [42, 43]. As much as 89.8% of production losses were due to suicides of men in CEE and this share was 71.2% in Sweden, a country with the most even gender distribution of cost. Considering the age distribution of suicide cost, the highest burden was identified for those at the ages 30–39 and 40–49 years. The age distribution of losses was similar for both sexes with a slightly higher magnitude of cost shares among adolescent women and young men (20–29 years).
The stability of results was tested using a one-way deterministic sensitivity analysis. The results changed notably with no discounting applied and this can be explained by the fact that suicide mortality is disproportionally high among relatively young people. Another scenario which varied estimates markedly was the use of general population employment rates instead of rates for depression population. It should be emphasized that this choice regarding employment rate is crucial for the indirect cost estimates. Yet, I decided to opt for the conservative scenario of a heavily reduced employment rate because the previous adjustment in this respect might be understated. For most other parameter changes used in the sensitivity analysis, the results’ variation was lower.
Because this is the first cross-country study on the cost of suicide mortality in general population (there is one recent international research but it focuses on the youth suicides solely [13]), the findings of this research could only be compared to previous estimates from single states. The recent results from Spain [12] identified the production losses of €380.5 million in 2013 (in 5% DR scenario) which is a close value to my estimate of €398.9 million in 2015. This similarity is irrespective of some important differences in the methods and data sources used; e.g. for the productivity measure, the Spanish study applied gross salary while mine–GDP adjusted with 0.65 coefficient. The study from Poland [9] reported the indirect cost of completed suicides as equivalent to €694.5 million in 2012 (using 5% DR) and this value is, again, comparable to my estimate of €751.5 million in 2015 for this country. This similarity might be surprising regarding the fact that my estimates are adjusted for lower employment rates among those who committed suicide. According to the conservative adjustment used here, the employment rates in the populations under study are 33.2 percentage points lower in Spain and 31.7 percentage points lower in Poland as compared to general employment rates. Therefore, this similarity of results seems to be coincidental. However, I argue that using employment rates for general population in a country bias the results upward because those committing suicides are possibly less likely to work. Finally, the Irish study identified the cost of lost market output attributable to suicides in 2001 and 2002 to be €205.9 million and €192.8 million, respectively [10]. In my estimates, the value of market production lost in Ireland was €213.7 million in 2015 and the difference between the results of the two studies is low regarding a 13–14 years divergence of the time settings. The methodological discrepancies seem to explain this difference as the Irish estimates are based on employment rates revised downward by only 2.25%, while here this adjustment is of greater magnitude (27.1%).
The major advantage of this research is the fact that it is the first study providing internationally comparable data on the production losses of suicide deaths, a major cost component of the total economic burden of suicides. Moreover, country-specific measures regarding labour market characteristics (real-life data on market entry and exit) and future economic growth are used and this reflects real differences among the countries investigated. With the use of homogenous data for these indicators, the peculiarities among the 28 states could not be unravelled.
On the other hand, the caveats of this analysis should also be pointed to. First, it only provides estimates on the burden of completed suicides in terms of potential production lost, while it does not investigate health care, medicolegal and intangible costs. However, the cost of mortality is a major cost driver (in several studies reaching more than 90% of the total cost [4, 5, 6, 7]) of the suicide-attributable economic burden; therefore, my estimates provide useful insight into the topic. Second, although the model uses homogenous data from a set of countries, some of the measures employed do not represent the reality in a perfect manner; e.g. the employment rates of those suffering from depression possibly deviate from the real employment among suicide population. Yet, it seemed to be the best choice available to account for different employment engagement of suicide population in particular countries. Moreover, such potential inaccuracies resulting from the use of proxy input data has been addressed by the sensitivity analysis. Third, the use of the HCA for the assessment of production losses is criticized for inflating the real economic burden [44, 45]; still, it is the most common method for assessment of indirect cost and the alternative of the friction cost method has its own weaknesses [46]. The detailed discussion on pros and cons of the two methods is beyond the scope of this paper (and can be found elsewhere [24]) but the choice of HCM can be justified by the fact that also the previous studies used it and this allows for direct comparison between the cost estimates. Fourth, the estimates from this research might be biased downward because of under-reporting suicides [7, 47]; however, it is not clear how serious this issue is and under-reporting data is critically missing in international, comparable context [48]. Moreover, this is just one of the problems concerning validity and reliability of the suicide statistics. The cross-country study using WHO European mortality data shows that the quality of suicide statistics varies notably between countries while the reasons for this variability are numerous [49] making the magnitude of the figures’ uncertainty difficult to assess. All and all, until now the ambiguous quality of suicides statistics appears to be an unsolved problem [50] and I was not able to address it in a reasonable way. Finally, the analysis only accounts for the cost borne in formal economies and does not include the burden experienced in terms of household activities undone or unregistered production lost because of suicides. Unfortunately, with the data presently at hand, this shortcoming could not have been addressed.