- Email: [email protected]
Associations between marital and educational status and risk of completed suicide in Hungary
Journal of Affective Disorders 190 (2016) 777–783
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
Research report
Associations between marital and educational status and risk of completed suicide in Hungary Lajos Bálint a,n, Péter Osváth b, Zoltán Rihmer c,d, Péter Döme c,d a
Demographic Research Institute of the Hungarian Central Statistical Office, Buday László u. 1-3, 1204 Budapest, Hungary Department of Psychiatry and Psychotherapy, Faculty of Medicine, University of Pécs, Rét u. 2, 7623 Pécs, Hungary c Department of Clinical and Theoretical Mental Health, Semmelweis University, Faculty of Medicine, Kútvölgyi út 4, 1125 Budapest, Hungary d National Institute of Psychiatry and Addictions, Laboratory for Suicide Research and Prevention, Lehel u. 59, 1135 Budapest, Hungary b
art ic l e i nf o
a b s t r a c t
Article history: Received 24 June 2015 Received in revised form 13 October 2015 Accepted 9 November 2015 Available online 12 November 2015
Background: Suicide rates in Hungary are notoriously high. According to the literature, marital and educational status are associated with suicidal behaviour and these associations are somewhat influenced by gender. Since in Hungary these associations have not yet been investigated by means of largescale multivariate epidemiological studies we aimed to investigate these in the current paper. Method: Census data on marital and educational status, age and gender from 1980, 1990, 2001 and 2011 were used for the general population. Corresponding data from the same years for suicide victims derived from the Hungarian Demographic Register. Suicide victims younger than 20 years were excluded. Negative binomial regression analyses were used to reveal the effects of the above variables on suicide. All statistical procedures were conducted using Stata 12 software (StataCorp. 2011). Results: Female gender, young age, higher educational attainment and marriage were significantly associated with decreased risks of suicide. Intriguingly, effects of educational and marital status on suicide were stronger in males. Limitations: Data on the length of the periods between changes in marital status and suicides were unavailable. Our four categories are not suitable to cover the whole gamut of marital statuses in a modern society (e.g. we did not have a specific category for people living in cohabitation). Ecological study design. Conclusion: We found that in Hungary between 1980 and 2011 the effects of some frequently investigated societal factors (e.g. educational and marital status) on suicide risk were very similar to those found in the majority of other countries. The effects of studied determinants of suicide have not changed dramatically over the past three decades in Hungary. & 2015 Published by Elsevier B.V.
Keywords: Suicide Marital status Marriage Educational level Gender differences
1. Introduction Currently – despite its almost monotonic and approximately 50% decrease from the mid-1980s – Hungary has the second highest suicide rate (after Lithuania) in the EU27 group of countries and the Hungarian suicide rate ranked in the global top 10 (WHO; Värnik, 2012; Rihmer et al., 2013). Furthermore, Hungary has had the world’s highest suicide rate averaged over the last 100 years (Szanto et al., 2007). It is well known that suicide is a multicausal event, so the probability of its occurrence is determined by several factors (Balint et al., 2014). According to the literature, marital status is one such factor. Enrico Morselli (1882), one of the most influential n
Corresponding author. E-mail address: balint@demografia.hu (L. Bálint).
http://dx.doi.org/10.1016/j.jad.2015.11.011 0165-0327/& 2015 Published by Elsevier B.V.
authors before Émile Durkheim, was the first who convincingly argued that proportionally fewer married than unmarried men commit suicide in Italy and France, but he found the opposite for women. He also demonstrated that widowhood (compared to married or unmarried status) was associated with the highest risk of suicide in both genders. In addition, he found that divorce is a potent suicide risk factor in both genders. Durkheim found that unmarried and widowed status are associated with a higher suicide risk than married status among those aged over 20. Furthermore, he demonstrated that widowed subjects generally are less likely to commit suicide than unmarried persons (Durkheim, 2005). However, based on his datasets, Durkheim was not able to determine clearly whether the above associations were stronger in males or in females (Durkheim, 2005). The majority of more recent investigations on this question have found – irrespective of data type (i.e. individual or aggregated) used – that widowed, divorced and, less obviously, unmarried statuses are associated with higher
778
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
suicide risk than married status. In addition, the majority of findings suggest that the protective effect of marriage is more pronounced among males and – in turn – the detrimental effects of widowed, divorced/separated and never-married status on risk of suicidal behaviour are less pronounced among females (Fukuchi et al., 2013; Masocco et al., 2010; Denney et al., 2009; Yamauchi et al., 2013; Liu et al., 2013; Yip and Thorburn, 2004; Smith et al., 1988; Wyder et al., 2009; Milner et al., 2013a; Barth et al., 2011; Corcoran and Nagar, 2010; Luoma and Pearson, 2002; Kposowa, 2000, 2003; Kõlves et al., 2010). The possible socio-psychological explanations for men’s higher vulnerability to committing suicide after a relationship breakdown (e.g. greater role inflexibility among males; men have fewer alternative intimate relationships than females; due to divorce men more frequently lose their close relationship with children and their home than females) are discussed in a recent paper by Scourfield and Evans (2015) and also by a paper by Kposowa (2003). Although it is a universal finding that ratios of suicide rates for divorced and married groups respectively are higher than one, there are significant differences between the exact values of these ratios across nations (e.g. in the 1960s for males the ratio was 6.2 in Denmark, while it was only 2.2 in France) (Stack and Scourfield, 2015). Intriguingly, the protective effect of living in marriage is not restricted to suicide mortality but also exists in regard to all-cause mortality, and the gender difference is also observable in that case (Gove 1973, Manzoli et al., 2007; Shor et al., 2012; Robards et al., 2012). Although some results indicate that – when compared to married status – the effect of single status (unmarried; widowed; divorced, separated) on higher risk of suicidal behaviour varies by age, it is hard to find a universal pattern in the findings of different workgroups (Masocco et al., 2010; Yip and Thorburn, 2004; Smith et al., 1988; Wyder et al., 2009; Corcoran and Nagar, 2010; Luoma and Pearson, 2002; Roškar et al., 2011). The association between the risk of suicidal behaviour and educational attainment is another intensively investigated area but the results are somewhat ambiguous. Although Durkheim found that people with higher education are more prone to commit suicide, he stated that this connection is not a causal one, since – in his interpretation – both suicide and the desire to be educated have a common root, namely the weakening cohesion of religious society (Durkheim, 2005). Morselli also found that suicide prevails among well-educated subjects (Morselli, 1882). Although some more recent studies also found that (groups of) subjects (or at least some of them) with higher school attainment have an increased risk of suicidal behaviour (e.g. Stack, 2000; Pompili et al., 2013) or that there is no relationship between the two variables (e.g. Qin et al., 2000; Lewis and Sloggett, 1998; Cubbin et al., 2000), the vast majority of recent investigations have found that the risk of suicidal behaviour is negatively correlated with higher educational attainment (Nock et al., 2008; Reques et al., 2014; Abel and Kruger, 2005; Ten Have et al., 2013; Flavio et al., 2013; Burrows and Laflamme, 2010; Beautrais, 2003; Denney et al., 2009; Kposowa, 2000; Mackenbach et al., 2015; Tjepkema et al., 2012; Lorant et al., 2005; Ishtiak-Ahmed et al., 2013; Lee et al., 2009; Milner et al., 2013a). A recent meta-analysis of studies on suicide by occupation also confirmed that the risk of suicide is greater in lower-skilled occupational groups (Milner et al., 2013b). Similarly to the case of marital status, some studies have found that the inverse relationship between educational attainment and suicide risk is less marked (Reques et al., 2014; Denney et al., 2009; Kposowa, 2000; Tjepkema et al., 2012; Lorant et al., 2005; Lee et al., 2009; Ross et al., 2012) or even the opposite among females (Strand et al., 2010; Mäki and Martikainen, 2009). An explanation for this gender difference may be that the advantages associated with educational attainment depend, in part, on the promise of occupational success in the future, a promise that may be fulfilled for males more
frequently than for females. Ultimately, the same level of education provides males with greater social integration than it does for females (Denney et al., 2009). An other possible explanation may be that some risk factors for suicidal behaviour (e.g. alcohol and drug addiction) are more common among men and people with low socioeconomic status (Lorant et al., 2005). Although several aspects of the Hungarian suicide scene have been investigated (cf. Rihmer et al. (2013)), according to our best knowledge only a few investigations (with quite different methodological approaches from those used in the current paper) have been conducted so far on the effects of marital status and educational attainment on suicide risk. One of those Hungarian investigations was a psychological autopsy study (n ¼194), the results of which confirmed the association between elevated suicide risk among those with low educational attainment and/or not living in marriage (Almasi et al., 2009). An other psychological autopsy study (n ¼100) did not demonstrate considerable associations between risk of suicide and marital status or educational attainment (Zonda, 2006). A short preliminary paper reported that in 1985 regional suicide rates contributed significantly to the multiple regression prediction of regional divorce rates (Nr. of regions ¼20) (Lester and Rihmer, 1997). Osváth et al. (2003) compared the characteristics of repeated suicide attempters with first suicide attempters. They reported that in the whole sample divorced (compared to married) status and low and middle (compared to high) educational statuses were significantly associated with higher risks of repeated suicide attempts (interestingly, divorced status was associated with higher risks for repeated suicide only among females and the effect of educational attainment was also more consistent among females). Considering that there is a paucity of large-scale epidemiological research on the relationship between marital status and educational attainment and the risk of completed suicide in Hungary we aimed primarily to investigate the effects of educational attainment and marital status on the risk of completed suicide and secondarily to assess whether the strength of the supposed effects have changed in the last three decades.
2. Data and methods 2.1. Data Census data stratified by age, gender, marital and educational status for the years 1980, 1990, 2001 and 2011 were used for the general population. Corresponding data from the same years for suicide victims derived from the Demographic Register of the Hungarian Central Statistical Office (“DEMO”). Accordingly, the latter database consisted routinely collected data. It follows from the foregoing that we used “unlinked data”, which may be the source of the so-called “numerator-denominator bias” (Shkolnikov et al., 2007). “Marital status” and “educational attainment” variables had four (married; unmarried; divorced; widowed) and three (not higher than vocational school (i.e. graduated from a school ranked not higher than levels 2C or 3C according to the International Standard Classification of Education (ISCED) system); graduated from schools awarding a secondary school leaving certificate, the prerequisite for entry into colleges and universities (ISCED3A); graduated from university or college (ISCED5A)) categories, respectively (for the Hungarian education system see www.cedefop. europa.eu/files/4103_EN.pdf and http://www.ofi.hu/isced-az-okta tas-egyseges-nemzetkozi-osztalyozasi-rendszere). Although age is presented as a five-category variable (aged 20–29, 30–39, 40–49, 50–59, 60r) in Table 1, it was used as a continuous variable in the regression models. Our data were stratified in an identical manner in every census year, which enabled us to ascertain some temporal
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
779
Table 1 Number of suicide cases and age-standardized suicide rates (per 100,000 capita per year) by marital status and educational attainment of the population above 20 years of age in Hungary in 1980, 1990, 2001 and 2011. Variables
1980
1990
Male
Female
Cases Rates
420 554 610 668 1001
2001
Male
Female
2011
Male
Female
Male
Female
Cases Rates Cases Rates Cases Rates Cases Rates
Cases Rates
Cases Rates
Cases Rates
63.6 68.2 89.2 119.8 127.1
96 163 229 264 680
15.0 20.1 32.3 40.8 58.0
249 563 640 496 958
37.7 69.3 93.6 89.0 121.7
64 126 199 174 557
10.0 15.6 28.1 26.9 47.5
196 313 600 459 662
24.0 47.5 80.2 74.6 82.3
39 55 126 136 327
5.0 8.5 16.0 19.5 25.6
122 220 336 474 547
19.4 27.6 51.3 69.8 60.0
11 46 83 120 261
1.8 5.9 12.6 15.8 18.4
2074 471 333 375
81.3 107.9 136.1 290.5
613 120 159 540
25.6 51.9 45.5 58.5
1664 465 432 345
63.4 113.9 146.2 170.8
420 83 144 433
19.6 35.6 38.3 34.1
1057 441 453 279
43.1 75.3 127.6 151.2
217 68 119 279
9.3 17.4 23.8 21.0
695 429 394 181
31.6 54.2 83.6 272.0
178 49 124 170
8.4 11.1 16.9 15.0
2911 215
116.8 39.3
1212 159
39.2 31.1
2536 273
102.1 51.2
892 173
28.2 31.1
1799 331
78.23 41.08
531 126
18.19 9.62
1299 294
62.05 30.01
352 119
14.73 8.73
127 4 3257
34.8
61 5 1437
23.5
97 4 2910
25.7
55 1 1121
21.7
100 15 2245
20.80
26 1 684
6.17
106 115 1814
16.93
50 40 561
7.23
n
Age group 20–29 30–39 40–49 50–59 60-x Marital status Married Unmarried Divorced Widowed Educational attainment Not higher than vocational school Secondary grammar (or secondary vocational) school Graduated from university or college Missing cases Total n
95.1
34.8
85.2
27.1
63.0
15.1
45.9
11.3
Age specific rates.
properties of the explanatory variables (in other words the stability of effects of explanatory variables on the risk of suicide). Data of suicide victims derived from the “DEMO” were almost complete in the first three census years investigated (i.e. 1980, 1990 and 2001) but there were missing data for educational attainment in 155 cases (6.5%) from the year 2011 (however, data on marital status and the gender of suicide victims were complete also for 2011) (Table 1). We excluded suicide victims younger than 20 years of age, since it is reasonable to presume that neither educational level nor marital status should be considered final in most of the members of that young age group. In accordance with previous literature (e.g. Vyssoki et al., 2014), an individual whose International Classification of Diseases-9 (ICD-9) or ICD-10 code in the paragraph “cause of death” was one of the following was considered a suicide completer: ICD-9: E950 to E959; ICD-10: X60 to X84 and Y87.0 (cases were coded by ICD-9 criteria in 1980 and 1990 and by ICD10 criteria in 2001 and 2011). 2.2. Data analysis and statistical methods By taking the age composition of the Hungarian population in 1980 as a standard, age-standardized suicide rates presented in Table 1 were calculated using the direct method. As a first approach to estimate incidence rate ratios (IRRs) and their 95% confidence intervals (CIs), Poisson regression models with the number of completed suicide cases as the response variable were fitted to the data with a canonic log link function and the corresponding stratum-specific population as an offset variable. Due to the limited number of cells the effect of age was estimated as a continuous variable (single parameter). Poisson regression was employed for every census year, both for the total population and separately for males and females. The most educated (‘graduated from university or college’) and married were chosen as reference categories in each model. Furthermore, in models for the total population female gender was an additional reference category. The main assumption for the Poisson model is that the conditional variance is equal to the conditional mean. If the equidispersion assumption is violated the model does not fit well.
Dispersion was assessed with Pearson statistic divided by the model degrees of freedom (Hilbe, 2011). Then score test was performed to evaluate whether overdispersion is present in the model (Dean and Lawless, 1989). Because, using the above methods, we found remarkable overdispersion in each model the extra variation was taken into account employing negative binomial regression. All statistical procedures were conducted using Stata 12 software (StataCorp. 2011).
3. Results Table 1 shows the Hungarian suicide numbers and rates by gender, age, marital status and educational attainment in 1980, 1990, 2001 and 2011 (the number of cases with incomplete data by gender are also presented in this table). In the four years investigated, suicide rates for both genders were highest in 1980 and – after a monotonic decrease – were lowest in 2011. This finding is consistent with the results of several previous studies on the Hungarian suicide rate (e.g. Rihmer et al., 2013). The results of negative binomial regression analyses on the effects of marital status, educational attainment, age and gender on suicide risk are presented for different years in Tables 2–5. In the total population, married status was associated with a significantly decreased risk of suicide compared to other marital statuses (i.e. unmarried, divorced or widowed) in all years investigated (Tables 2–5). Similarly, in male and female subpopulations unmarried/divorced/widowed statuses (compared to married status) pose an extra risk of suicide (of the 24 comparisons, only 2 were non-significant (2011: unmarried females vs. married females; 2011: widowed females vs. married females)) (Tables 2– 5). So we found that suicide risks for unmarried/divorced/widowed males (compared to married males) were significantly higher in each comparison while the same is not entirely true for females. In addition, in the majority of cases (i.e. in 8 out of 12 cases) the incidence rate ratios of unmarried/divorced/widowed males were higher than those of unmarried/divorced/widowed females, respectively (Tables 2–5). Some authors have suggested that in the past decades – as divorce has become more frequent in modern societies – the stigma associated with it has decreased,
780
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
Table 2 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 1980. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC
Total
Male
IRR
95% CI
1.214nnn 1.000
1.144–1.288
2.526nnn 1.000
2.150–2.966
1.721nnn 2.084nnn 2.637nnn 1.000 1.118 2.306nnn 0.17 347.31 5.94
IRR
Female 95% CI
IRR
95% CI
1.146nnn
1.085–1.211
1.303nnn
1.194–1.421
1.388–2.134 1.698–2.558 2.064–3.368
1.259n 1.793nnn 2.249nnn
1.030–1.538 1.457–2.208 1.766– 2.864
2.160nnn 2.152nnn 2.791nnn
1.571–2.969 1.587–2.919 1.972–3.950
0.875–1.427 1.853–2.871
1.093 3.120nnn 0.19 173.46 6.05
0.842–1.418 2.484–3.918
1.093 1.670nn 0.17 160.71 5.62
0.763–1.566 1.204–2.316
n
po 0.05. p o0.01. nnn p o0.001. nn
leading ultimately to a decrease in the number of stigma-driven suicide cases (Stack and Scourfield, 2015). Although, in the period investigated, a continuously growing proportion of marriages ended in divorce in Hungary (total divorce rates were 0.29 in 1980 and 0.46 in 2011), we were not able to demonstrate that the suicide risk of divorced (compared to married) subjects had decreased (Bukodi et al., 2005; Földházi, 2012; KSH,online). In fact, contrary to our expectations, when compared to married subjects, the risks of suicide among divorced subjects in 1990, 2001 and 2011 were higher in the total population and also in the male subpopulation than in 1980 (in the female subpopulation the situation was the same for the years 1990 and 2001, but not for 2011) (Tables 2–5). Some results have indicated that the recency of changes in marital status, i.e. the length of time that has elapsed since divorce or being widowed, has a more pronounced effect on suicide risk than marital status per se (Stack and Scourfield, 2015; Ajdacic-Gross et al. 2008; Batterham et al., 2014). Unfortunately, our dataset did not allow us to test this phenomenon. Generally speaking higher educational levels were associated
with reduced suicide risks in our total sample, as well in the male and female subsamples. Accordingly, in all the years studied (1980, 1990, 2001 and 2011), the following association was observable in the total population and in the male subpopulation: the suicide risk of the least educated group (“not higher than vocational school”) was significantly higher than the corresponding risk for the most-educated group (“graduated from university or college”) (for females the same association was observable in all the studied years but the association did not reach the level of significance in 1990) (Tables 2–5). As discussed in the “Introduction” section, some studies have found that the inverse relationship between educational attainment and suicide risk is less marked or even the opposite among females. A somewhat similar phenomenon was also observable in our dataset, since (1) among females there were 3 years (1980, 1990 and 2011) when significant differences were not observable between the suicide risks of the groups “completed general secondary school” and “graduated from university or college” (i.e. the reference category), while there was only one such year (1980) in the case of the male subpopulation; (2) compared to
Table 3 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 1990. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC †
po 0.1. p o0.001.
nnn
Total
Male
Female
IRR
95% CI
IRR
95% CI
IRR
95% CI
1.301nnn 1.000
1.225–1.381
1.272nnn
1.192–1.357
1.333nnn
1.230–1.445
3.226nnn 1.000
2.758–3.774
1.921nnn 2.780nnn 2.686nnn 1.000
1.562–2.361 2.292–3.371 2.117–3.407
1.842nnn 2.954nnn 2.729nnn
1.480–2.293 2.388–3.653 2.051–3.632
1.736nnn 2.302nnn 2.290nnn
1.279–2.356 1.769–2.996 1.706–3.073
1.738nnn 2.391nnn 0.22 324.14 5.55
1.359–2.223 1.905–3.001
1.911nnn 3.528nnn 0.23 168.77 5.89
1.436–2.543 2.724–4.568
1.371† 1.316† 0.20 139.68 4.923
0.974–1.928 0.960–1.803
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
781
Table 4 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 2001. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC
Total
Male
Female
IRR
95% CI
IRR
95% CI
IRR
95% CI
1.225nnn 1.000
1.137–1.320
1.207nnn
1.098–1.326
1.260nnn
1.119–1.418
4.749nnn 1.000
3.948–5.713
1.704nnn 2.944nnn 2.638nnn 1.000
1.326–2.190 2.345–3.697 1.989–3.499
1.634nn 3.117nnn 3.041nnn
1.209–2.208 2.342–4.148 2.090–4.424
1.855nn 2.671nnn 2.191nnn
1.204–2.858 1.856–3.844 1.436–3.343
1.890nnn 3.099nnn 0.220 311.43 5.34
1.430–2.499 2.384–4.028
1.856nnn 3.297nnn 0.17 178.15 6.20
1.329–2.590 2.415–4.502
1.886n 2.743nnn 0.14 131.48 4.65
1.157–3.074 1.722–4.370
n
po 0.05. p o0.01. nnn p o 0.001. nn
the reference group, the suicide risks of the least educated subjects (i.e. members of the “not higher than vocational school” group) were more pronounced in males than in females in each year investigated; accordingly, for the least educated males IRRs were above 3.1 in each year, but for the least educated females IRRs were lower than 2.75 in each year; (3) as we discussed previously, among females there was a year (1990) when there was no significant difference in the risk of suicide between the least- and the most-educated groups (while there was not such a year in the male group) (Tables 2–5). An additional result may also be discerned from the data: looking at the year representing the socialist era (1980) and the year immediately following the transition (1990) grouped together, we can see that the suicide risk of the least-educated group in the total population was lower than in 2001 and 2011 (IRRs were 2.3; 2.4; 3.1 and 3.0, respectively) (Tables 2–5). It seems that this change in the total population during the three decades investigated was driven primarily by females (IRRs in the four years were 1.7; 1.3; 2.7 and 2.1,
respectively) since in males IRRs were roughly the same in the years of socialism (3.1 and 3.5) and in the years after transition (3.3 and 3.5). Our results suggest that in Hungary total, male, and female suicide rates increase with age (Tables 2–5). Moreover – regardless of gender – the effect of age was relatively stable over the period studied. According to our results, the effect of gender was highly significant in all years. We found that male gender carries a higher risk of suicide than female (Tables 2–5). Controlling for age, marital status and educational level, the risk of taking their own lives was about 4–5 times higher for males than for females in 2001 and in 2011. This effect was somewhat smaller in 1980 and in 1990, probably due to the much larger decrease in the standardized suicide rate of females than of males between 1980 and 2011 (in this period the suicide rate decreased by two-thirds in females, while it only halved in males) (Table 1).
Table 5 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 2011. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC †
p o0.1. p o0.01. p o 0.001.
nn
nnn
Total
Male
Female
IRR
95% CI
IRR
95% CI
IRR
95% CI
1.330nnn 1.000
1.239–1.428
1.278nnn
1.177–1.389
1.387nnn
1.229–1.566
4.243nnn 1.000
3.589–5.017
1.346nn 2.509nnn 2.149nnn 1.000
1.082–1.673 2.043–3.082 1.633–2.828
1.493nn 2.765nnn 2.885nnn
1.161–1.919 2.151–3.554 2.006–4.148
1.028 2.144nnn 1.372†
0.702–1.506 1.614–2.848 0.978–1.925
1.593nnn 2.985nnn 0.25 294.64 5.06
1.252–2.027 2.385–3.737
1.748nnn 3.485nnn 0.21 168.11 5.870
1.305–2.342 2.660–4.567
1.280 2.141nnn 0.21 119.05 4.24
0.894–1.833 1.526–3.003
782
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
4. Discussion Suicide mortality in Hungary was unexpectedly high in the previous century, and the Hungarian suicide rate is still the second highest in the European Union (EU27) (Szanto et al., 2007; WHO). Despite the disappointing features of the Hungarian suicide scene, according to our best knowledge, there is a lack of large-scale multivariate epidemiological research on the relationship between marital status and educational attainment and the risk of completed suicide in Hungary. The longitudinal observation of the effects of various suicideinfluencing factors is substantiated at least by two facts. Firstly, annual suicide rates in Hungary showed an almost monotonic decrease from the mid-1980s. Secondly, relevant political (i.e. the fall of the communist regime) and economic (i.e. the transition from a centrally planned economy to a market economy) changes occurred in Hungary in the era investigated (1980–2011). Furthermore, intensive societal fluctuations also occurred during the observation period, such as the change in proportions of preponderant family structure types. Accordingly, the proportion of marriages decreased from 80% to 65% between 1990 and 2011. On the other hand, the proportion of cohabitations and single-parent families (together) increased by 15% in the same period (KSH, 2014). Similar changes (e.g. the gradual decline in marriages and the increase in cohabitations) are typical in all regions of Europe (Sobotka and Toulemon, 2008). Our main findings based on Hungarian data are consistent with the majority of previous results from other countries (see the discussion in the “Introduction” section). Accordingly, we found that (1) in each population investigated (i.e. total; male; female) married status, in the great majority of cases, was associated with a significantly decreased risk of suicide compared to other marital statuses (i.e. unmarried, divorced or widowed) (Tables 2–5). In line with the results of several studies from other countries (e.g. Fukuchi et al., 2013; Kõlves et al., 2010; Kposowa, 2000, 2003; Masocco et al., 2010) we found weaker effects of marital status on suicide risk among females compared to males. However, our results do not confirm (and, indeed, rather disprove) the theory that the suicide risk of divorced individuals had decreased as divorce has become more frequent (Stack and Scourfield, 2015); (2) generally speaking higher educational levels were associated with reduced suicide risks in our total sample, as well in the male and female subsamples. This finding is concordant with the majority of results from other countries (Nock et al., 2008; Reques et al., 2014; Abel and Kruger, 2005; Ten Have et al., 2013; Flavio et al., 2013; Burrows and Laflamme, 2010; Beautrais, 2003; Denney et al., 2009; Kposowa, 2000; Mackenbach et al., 2015; Tjepkema et al., 2012; Lorant et al., 2005; Ishtiak-Ahmed et al., 2013; Lee et al., 2009; Milner et al., 2013a). Results of some previous studies have suggested that the inverse relationship between educational attainment and suicide risk is less marked (or even the opposite) among females (Reques et al., 2014; Denney et al., 2009; Kposowa, 2000; Tjepkema et al., 2012; Lorant et al., 2005; Lee et al., 2009; Ross et al., 2012; Strand et al., 2010; Mäki and Martikainen, 2009). In line with the results of the above studies, our findings also indicate a weaker association between educational attainment and suicide risk among females than among males. Our finding about the growing difference between suicide risks of the highest and the lowest educated groups before and after the political transition is consistent with the results of a previous Hungarian study on mortality due to external causes (a category which consists, inter alia, of suicide) based on data from 1986 to 2005 (Hablicsek and Kovács, 2007). One of the several possible explanations for this finding may be that the egalitarian approach of the socialist health care system (e.g. free health care provision – of the same quality – became a civil right after 1975) was better able to balance the
detrimental effects of low education on suicide risk (Horváth, 2011); (3) in Hungary total, male, and female suicide rates increase with age; and (4) male gender carries a higher risk of completed suicide than female. The latter two findings are also in line with the majority of results from developed countries (Rihmer et al., 2013; Hawton and van Heeringen, 2009). There are some limitations to the present study: (1) data on the length of the period between the change in marital status and suicide were not available, so we were not able to test whether recent changes in marital status have different effects on suicide risk from distant ones; (2) our four categories are not suitable to cover the whole spectrum of marital statuses of a modern society (e.g. we did not have specific categories for people who are separated (but not divorced by law) or for those who are living in cohabitation). Of course acquiring data on the quality of marriages was beyond our power; (3) as we have already mentioned before in the “Data and methods” section there were some missing data for the year 2011; (4) the use of “unlinked data” may be the source of the so-called “numerator-denominator bias” (Shkolnikov et al., 2007); (5) ecological study design. In conclusion, we found that in Hungary – similarly to the great majority of previous results from other countries – female gender, higher educational attainment, lower age and living in marriage were significantly associated with decreased risks of completed suicide in the period between 1980 and 2011. The effects of investigated socio-economic determinants of suicide have not changed dramatically over the decades of our observation period.
Acknowledgement Peter Dome is recipient of the János Bolyai Postdoctoral Fellowship of the Hungarian Academy of Sciences.
References Abel, E.L., Kruger, M.L., 2005. Educational attainment and suicide rates in the United States. Psychol. Rep. 97, 25–28. Ajdacic-Gross, V., Ring, M., Gadola, E., Lauber, C., Bopp, M., Gutzwiller, F., Rössler, W., 2008. Suicide after bereavement: an overlooked problem. Psychol. Med. 38, 673–676. Almasi, K., Belso, N., Kapur, N., Webb, R., Cooper, J., Hadley, S., Kerfoot, M., Dunn, G., Sotonyi, P., Rihmer, Z., Appleby, L., 2009. Risk factors for suicide in Hungary: a case-control study. BMC Psychiatry 9, 45. Balint, L., Dome, P., Daroczi, G., Gonda, X., Rihmer, Z., 2014. Investigation of the marked and long-standing spatial inhomogeneity of the Hungarian suicide rate: a spatial regression approach. J. Affect. Disord. 155, 180–185. Barth, A., Sögner, L., Gnambs, T., Kundi, M., Reiner, A., Winker, R., 2011. Socioeconomic factors and suicide: an analysis of 18 industrialized countries for the years 1983 through 2007. J. Occup. Environ. Med. 53, 313–317. Batterham, P.J., Fairweather-Schmidt, A.K., Butterworth, P., Calear, A.L., Mackinnon, A.J., Christensen, H., 2014. Temporal effects of separation on suicidal thoughts and behaviours. Soc. Sci. Med. 111, 58–63. Beautrais, A.L., 2003. Suicide and serious suicide attempts in youth: a multiplegroup comparison study. Am. J. Psychiatry 160, 1093–1099. Nők és férfiak Magyarországon, 2004. In: Bukodi, E., Mészárosné Halász, J., Polónyi, K., Tallér, A. (Eds.), Ifjúsági, Családügyi, Szociális és Esélyegyenlőségi Minisztérium. KSH, Budapest, p. 16. Burrows, S., Laflamme, L., 2010. Socioeconomic disparities and attempted suicide: state of knowledge and implications for research and prevention. Int. J. Inj. Control Saf. Promot. 17, 23–40. Corcoran, P., Nagar, A., 2010. Suicide and marital status in Northern Ireland. Soc. Psychiatry Psychiatr. Epidemiol. 45, 795–800. Cubbin, C., LeClere, F.B., Smith, G.S., 2000. Socioeconomic status and the occurrence of fatal and nonfatal injury in the United States. Am. J. Public Health 90, 70–77. Dean, C., Lawless, J.F., 1989. Tests for detecting overdispersion in Poisson regression models. J. Am. Stat. Assoc. 84, 467–472. Denney, J.T., Rogers, R.G., Krueger, P.M., Wadsworth, T., 2009. Adult suicide mortality in the united states: marital status, family size, socioeconomic status, and differences by sex. Soc. Sci. Q. 90, 1167. Durkheim, É., 2005. Suicide. A study in Sociology. Routledge, London and New York, first published 1897. Flavio, M., Martin, E., Pascal, B., Stephanie, C., Gabriela, S., Merle, K., Anita, R.R., 2013. Suicide attempts in the county of Basel: results from the WHO/EURO Multicentre Study on Suicidal Behaviour. Swiss Med. Wkly. 143, w13759.
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783 Földházi, E., 2012. Divorce. In: Zsolt, Spéder, Péter, Őri (Eds.), Demographic Portrait of Hungary. The Report on the Condition of the Hungarian Population. Demographic Research Institute, HCSO, Budapest, pp. 21–30. Fukuchi, N., Kakizaki, M., Sugawara, Y., Tanji, F., Watanabe, I., Fukao, A., Tsuji, I., 2013. Association of marital status with the incidence of suicide: a populationbased Cohort Study in Japan (Miyagi cohort study). J. Affect. Disord. 150, 879–885. Gove, W.R., 1973. Sex, marital status, and mortality. Am. J. Sociol. 79, 45–67. Hablicsek, L., Kovács, K., 2007. Az életkilátások Differenciálódása Iskolázottság Szerint, 1986–2005. KSH Népességtudományi Kutatóintézet, Budapest, pp. 89–93. Hawton, K., van Heeringen, K., 2009. Suicide. Lancet 373, 1372–1381. Hilbe, J.M., 2011. Negative Binomial Regression, Second edition. Cambridge University Press, Cambridge, UK. Horváth, G.K., 2011. Magyarország társadalomtörténete a szocialista korban. Ishtiak-Ahmed, K., Perski, A., Mittendorfer-Rutz, E., 2013. Predictors of suicidal behaviour in 36,304 individuals sickness absent due to stress-related mental disorders – a Swedish register linkage cohort study. BMC Public Health 13, 492. Kõlves, K., Ide, N., De Leo, D., 2010. Suicidal ideation and behaviour in the aftermath of marital separation: gender differences. J. Affect. Disord. 120, 48–53. Kposowa, A.J., 2000. Marital status and suicide in the National Longitudinal Mortality Study. J. Epidemiol. Community Health 54, 254–261. Kposowa, A.J., 2003. Divorce and suicide risk. J. Epidemiol. Community Health 57, 993. KSH, 2014. Családtípusok* és párkapcsolati formák változása a népszámlálási adatok tükrében. Stat Tükör. VII/115. (In Hungarian). 〈www.ksh.hu/docs/hun/ xftp/stattukor/csaladtipusok.pdf〉. KSH online. 〈https://www.ksh.hu/thm/2/indi2_1_3.html〉. Lee, W.Y., Khang, Y.H., Noh, M., Ryu, J.I., Son, M., Hong, Y.P., 2009. Trends in educational differentials in suicide mortality between 1993–2006 in Korea. Yonsei Med. J. 50, 482–492. Lester, D., Rihmer, Z., 1997. The regional variation of divorce in Hungary. J. Divorce Remarriage 26, 83–85. Lewis, G., Sloggett, A., 1998. Suicide, deprivation, and unemployment: record linkage study. BMJ 317, 1283–1286. Liu, Y., Zhang, Y., Cho, Y.T., Obayashi, Y., Arai, A., Tamashiro, H., 2013. Gender differences of suicide in Japan, 1947–2010. J. Affect. Disord. 151, 325–330. Lorant, V., Kunst, A.E., Huisman, M., Costa, G., Mackenbach, J., EU Working Group on Socio-Economic Inequalities in Health, 2005. Socio-economic inequalities in suicide: a European comparative study. Br. J. Psychiatry 187, 49–54. Luoma, J.B., Pearson, J.L., 2002. Suicide and marital status in the United States, 1991–1996: is widowhood a risk factor? Am. J. Public Health 92, 1518–1522. Mackenbach, J.P., Kulhánová, I., Bopp, M., Deboosere, P., Eikemo, T.A., Hoffmann, R., Kulik, M.C., Leinsalu, M., Martikainen, P., Menvielle, G., Regidor, E., Wojtyniak, B., Ostergren, O., Lundberg, O., for the EURO-GBD-SE Consortium, 2015. Variations in the relation between education and cause-specific mortality in 19 European populations: A test of the “fundamental causes” theory of social inequalities in health. Soc. Sci. Med. 127, 51–62. Mäki, N., Martikainen, P., 2009. The role of socioeconomic indicators on non-alcohol and alcohol-associated suicide mortality among women in Finland. A register-based follow-up study of 12 million person-years. Soc. Sci. Med. 68, 2161–2169. Manzoli, L., Villari, P., M Pirone, G., Boccia, A., 2007. Marital status and mortality in the elderly: a systematic review and meta-analysis. Soc. Sci. Med. 64, 77–94. Masocco, M., Pompili, M., Vanacore, N., Innamorati, M., Lester, D., Girardi, P., Tatarelli, R., Vichi, M., 2010. Completed suicide and marital status according to the Italian region of origin. Psychiatr. Q. 81, 57–71. Milner, A., Hjelmeland, H., Arensman, E., De Leo, D., 2013a. Social-environmental factors and suicide mortality: a narrative review of over 200 articles. Sociol. Mind 3, 137–148. Milner, A., Spittal, M.J., Pirkis, J., LaMontagne, A.D., 2013b. Suicide by occupation: systematic review and meta-analysis. Br. J. Psychiatry 203, 409–416. Morselli, H., 1882. Suicide: An Essay on Comparative Moral Statistics. D. Appleton and company, New York. Nock, M.K., Borges, G., Bromet, E.J., Alonso, J., Angermeyer, M., Beautrais, A., Bruffaerts, R., Chiu, W.T., de Girolamo, G., Gluzman, S., de Graaf, R., Gureje, O., Haro, J.M., Huang, Y., Karam, E., Kessler, R.C., Lepine, J.P., Levinson, D., Medina-Mora, M.E., Ono, Y., Posada-Villa, J., Williams, D., 2008. Cross-national prevalence and
783
risk factors for suicidal ideation, plans and attempts. Br. J. Psychiatry 192, 98–105. Osváth, P., Kelemen, G., Erdös, M.B., Vörös, V., Fekete, S., 2003. The main factors of repetition: review of some results of the Pecs Center in the WHO/EURO Multicentre Study on Suicidal Behaviour. Crisis 24, 151–154. Pompili, M., Vichi, M., Qin, P., Innamorati, M., De Leo, D., Girardi, P., 2013. Does the level of education influence completed suicide? A nationwide register study. J. Affect. Disord. 147, 437–440. Qin, P., Agerbo, E., Westergård-Nielsen, N., Eriksson, T., Mortensen, P.B., 2000. Gender differences in risk factors for suicide in Denmark. Br. J. Psychiatry 177, 546–550. Reques, L., Giráldez-García, C., Miqueleiz, E., Belza, M.J., Regidor, E., 2014. Educational differences in mortality and the relative importance of different causes of death: a 7-year follow-up study of Spanish adults. J. Epidemiol. Community Health 68, 1151–1160. Rihmer, Z., Gonda, X., Kapitany, B., Dome, P., 2013. Suicide in Hungary-epidemiological and clinical perspectives. Ann. Gen. Psychiatry 12, 21. Robards, J., Evandrou, M., Falkingham, J., Vlachantoni, A., 2012. Marital status, health and mortality. Maturitas 73, 295–299. Ross, C.E., Masters, R.K., Hummer, R.A., 2012. Education and the gender gaps in health and mortality. Demography 49, 1157–1183. Roškar, S., Podlesek, A., Kuzmanić, M., Demšar, L.O., Zaletel, M., Marušič, A., 2011. Suicide risk and its relationship to change in marital status. Crisis 32, 24–30. Scourfield, J., Evans, R., 2015. Why might men be more at risk of suicide after a relationship breakdown? Sociological insights. Am. J. Mens Health 9, 380–384. Shkolnikov, V.M., Jasilionis, D., Andreev, E.M., Jdanov, D.A., Stankuniene, V., Ambrozaitiene, D., 2007. Linked versus unlinked estimates of mortality and length of lifeby education and marital status: evidence from the first record linkage study in Lithuania. Soc. Sci. Med. 64, 1392–1406. Shor, E., Roelfs, D.J., Curreli, M., Clemow, L., Burg, M.M., Schwartz, J.E., 2012. Widowhood and mortality: a meta-analysis and meta-regression. Demography 49, 575–606. Smith, J.C., Mercy, J.A., Conn, J.M., 1988. Marital status and the risk of suicide. Am. J. Public Health 78, 78–80. Sobotka, T., Toulemon, L., 2008. Overview chapter 4: changing family and partnership behaviour: Common trends and persistent diversity across europe. Demogr. Res. 19, 85–138. Stack, S., 2000. Suicide: a 15-year review of the sociological literature. Part II: modernization and social integration perspectives. Suicide Life Threat. Behav. 30, 163–176. Stack, S., Scourfield, J., 2015. Recency of divorce, depression, and suicide risk. J. Fam. Issues 36, 695–715. Strand, B.H., Grøholt, E.K., Steingrímsdóttir, O.A., Blakely, T., Graff-Iversen, S., Naess, Ø., 2010. Educational inequalities in mortality over four decades in Norway: prospective study of middle aged men and women followed for cause specific mortality, 1960–2000. BMJ 340, c654. Szanto, K., Kalmar, S., Hendin, H., Rihmer, Z., Mann, J.J., 2007. A suicide prevention program in a region with a very high suicide rate. Arch. Gen. Psychiatry 64, 914–920. Ten Have, M., van Dorsselaer, S., de Graaf, R., 2013. Prevalence and risk factors for first onset of suicidal behaviors in the Netherlands mental health survey and incidence study-2. J. Affect. Disord. 147, 205–211. Tjepkema, M., Wilkins, R., Long, A., 2012. Cause-specific mortality by education in Canada: a 16-year follow-up study. Health Rep. 23, 23–31. Värnik, P., 2012. Suicide in the world. Int. J. Env. Res. Public Health 9, 760–771. Vyssoki, B., Kapusta, N.D., Praschak-Rieder, N., Dorffner, G., Willeit, M., 2014. Direct effect of sunshine on suicide. JAMA Psychiatry 71, 1231–1237. WHO. 〈http://data.euro.who.int/hfadb/〉 (data retrieved in September 2014). Wyder, M., Ward, P., De Leo, D., 2009. Separation as a suicide risk factor. J. Affect. Disord. 116, 208–213. Yamauchi, T., Fujita, T., Tachimori, H., Takeshima, T., Inagaki, M., Sudo, A., 2013. Ageadjusted relative suicide risk by marital and employment status over the past 25 years in Japan. J. Public Health 35, 49–56. Yip, P.S., Thorburn, J., 2004. Marital status and the risk of suicide: experience from England and Wales, 1982–1996. Psychol. Rep. 94, 401–407. Zonda, T., 2006. One-hundred cases of suicide in Budapest: a case-controlled psychological autopsy study. Crisis 27, 125–129.
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
Research report
Associations between marital and educational status and risk of completed suicide in Hungary Lajos Bálint a,n, Péter Osváth b, Zoltán Rihmer c,d, Péter Döme c,d a
Demographic Research Institute of the Hungarian Central Statistical Office, Buday László u. 1-3, 1204 Budapest, Hungary Department of Psychiatry and Psychotherapy, Faculty of Medicine, University of Pécs, Rét u. 2, 7623 Pécs, Hungary c Department of Clinical and Theoretical Mental Health, Semmelweis University, Faculty of Medicine, Kútvölgyi út 4, 1125 Budapest, Hungary d National Institute of Psychiatry and Addictions, Laboratory for Suicide Research and Prevention, Lehel u. 59, 1135 Budapest, Hungary b
art ic l e i nf o
a b s t r a c t
Article history: Received 24 June 2015 Received in revised form 13 October 2015 Accepted 9 November 2015 Available online 12 November 2015
Background: Suicide rates in Hungary are notoriously high. According to the literature, marital and educational status are associated with suicidal behaviour and these associations are somewhat influenced by gender. Since in Hungary these associations have not yet been investigated by means of largescale multivariate epidemiological studies we aimed to investigate these in the current paper. Method: Census data on marital and educational status, age and gender from 1980, 1990, 2001 and 2011 were used for the general population. Corresponding data from the same years for suicide victims derived from the Hungarian Demographic Register. Suicide victims younger than 20 years were excluded. Negative binomial regression analyses were used to reveal the effects of the above variables on suicide. All statistical procedures were conducted using Stata 12 software (StataCorp. 2011). Results: Female gender, young age, higher educational attainment and marriage were significantly associated with decreased risks of suicide. Intriguingly, effects of educational and marital status on suicide were stronger in males. Limitations: Data on the length of the periods between changes in marital status and suicides were unavailable. Our four categories are not suitable to cover the whole gamut of marital statuses in a modern society (e.g. we did not have a specific category for people living in cohabitation). Ecological study design. Conclusion: We found that in Hungary between 1980 and 2011 the effects of some frequently investigated societal factors (e.g. educational and marital status) on suicide risk were very similar to those found in the majority of other countries. The effects of studied determinants of suicide have not changed dramatically over the past three decades in Hungary. & 2015 Published by Elsevier B.V.
Keywords: Suicide Marital status Marriage Educational level Gender differences
1. Introduction Currently – despite its almost monotonic and approximately 50% decrease from the mid-1980s – Hungary has the second highest suicide rate (after Lithuania) in the EU27 group of countries and the Hungarian suicide rate ranked in the global top 10 (WHO; Värnik, 2012; Rihmer et al., 2013). Furthermore, Hungary has had the world’s highest suicide rate averaged over the last 100 years (Szanto et al., 2007). It is well known that suicide is a multicausal event, so the probability of its occurrence is determined by several factors (Balint et al., 2014). According to the literature, marital status is one such factor. Enrico Morselli (1882), one of the most influential n
Corresponding author. E-mail address: balint@demografia.hu (L. Bálint).
http://dx.doi.org/10.1016/j.jad.2015.11.011 0165-0327/& 2015 Published by Elsevier B.V.
authors before Émile Durkheim, was the first who convincingly argued that proportionally fewer married than unmarried men commit suicide in Italy and France, but he found the opposite for women. He also demonstrated that widowhood (compared to married or unmarried status) was associated with the highest risk of suicide in both genders. In addition, he found that divorce is a potent suicide risk factor in both genders. Durkheim found that unmarried and widowed status are associated with a higher suicide risk than married status among those aged over 20. Furthermore, he demonstrated that widowed subjects generally are less likely to commit suicide than unmarried persons (Durkheim, 2005). However, based on his datasets, Durkheim was not able to determine clearly whether the above associations were stronger in males or in females (Durkheim, 2005). The majority of more recent investigations on this question have found – irrespective of data type (i.e. individual or aggregated) used – that widowed, divorced and, less obviously, unmarried statuses are associated with higher
778
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
suicide risk than married status. In addition, the majority of findings suggest that the protective effect of marriage is more pronounced among males and – in turn – the detrimental effects of widowed, divorced/separated and never-married status on risk of suicidal behaviour are less pronounced among females (Fukuchi et al., 2013; Masocco et al., 2010; Denney et al., 2009; Yamauchi et al., 2013; Liu et al., 2013; Yip and Thorburn, 2004; Smith et al., 1988; Wyder et al., 2009; Milner et al., 2013a; Barth et al., 2011; Corcoran and Nagar, 2010; Luoma and Pearson, 2002; Kposowa, 2000, 2003; Kõlves et al., 2010). The possible socio-psychological explanations for men’s higher vulnerability to committing suicide after a relationship breakdown (e.g. greater role inflexibility among males; men have fewer alternative intimate relationships than females; due to divorce men more frequently lose their close relationship with children and their home than females) are discussed in a recent paper by Scourfield and Evans (2015) and also by a paper by Kposowa (2003). Although it is a universal finding that ratios of suicide rates for divorced and married groups respectively are higher than one, there are significant differences between the exact values of these ratios across nations (e.g. in the 1960s for males the ratio was 6.2 in Denmark, while it was only 2.2 in France) (Stack and Scourfield, 2015). Intriguingly, the protective effect of living in marriage is not restricted to suicide mortality but also exists in regard to all-cause mortality, and the gender difference is also observable in that case (Gove 1973, Manzoli et al., 2007; Shor et al., 2012; Robards et al., 2012). Although some results indicate that – when compared to married status – the effect of single status (unmarried; widowed; divorced, separated) on higher risk of suicidal behaviour varies by age, it is hard to find a universal pattern in the findings of different workgroups (Masocco et al., 2010; Yip and Thorburn, 2004; Smith et al., 1988; Wyder et al., 2009; Corcoran and Nagar, 2010; Luoma and Pearson, 2002; Roškar et al., 2011). The association between the risk of suicidal behaviour and educational attainment is another intensively investigated area but the results are somewhat ambiguous. Although Durkheim found that people with higher education are more prone to commit suicide, he stated that this connection is not a causal one, since – in his interpretation – both suicide and the desire to be educated have a common root, namely the weakening cohesion of religious society (Durkheim, 2005). Morselli also found that suicide prevails among well-educated subjects (Morselli, 1882). Although some more recent studies also found that (groups of) subjects (or at least some of them) with higher school attainment have an increased risk of suicidal behaviour (e.g. Stack, 2000; Pompili et al., 2013) or that there is no relationship between the two variables (e.g. Qin et al., 2000; Lewis and Sloggett, 1998; Cubbin et al., 2000), the vast majority of recent investigations have found that the risk of suicidal behaviour is negatively correlated with higher educational attainment (Nock et al., 2008; Reques et al., 2014; Abel and Kruger, 2005; Ten Have et al., 2013; Flavio et al., 2013; Burrows and Laflamme, 2010; Beautrais, 2003; Denney et al., 2009; Kposowa, 2000; Mackenbach et al., 2015; Tjepkema et al., 2012; Lorant et al., 2005; Ishtiak-Ahmed et al., 2013; Lee et al., 2009; Milner et al., 2013a). A recent meta-analysis of studies on suicide by occupation also confirmed that the risk of suicide is greater in lower-skilled occupational groups (Milner et al., 2013b). Similarly to the case of marital status, some studies have found that the inverse relationship between educational attainment and suicide risk is less marked (Reques et al., 2014; Denney et al., 2009; Kposowa, 2000; Tjepkema et al., 2012; Lorant et al., 2005; Lee et al., 2009; Ross et al., 2012) or even the opposite among females (Strand et al., 2010; Mäki and Martikainen, 2009). An explanation for this gender difference may be that the advantages associated with educational attainment depend, in part, on the promise of occupational success in the future, a promise that may be fulfilled for males more
frequently than for females. Ultimately, the same level of education provides males with greater social integration than it does for females (Denney et al., 2009). An other possible explanation may be that some risk factors for suicidal behaviour (e.g. alcohol and drug addiction) are more common among men and people with low socioeconomic status (Lorant et al., 2005). Although several aspects of the Hungarian suicide scene have been investigated (cf. Rihmer et al. (2013)), according to our best knowledge only a few investigations (with quite different methodological approaches from those used in the current paper) have been conducted so far on the effects of marital status and educational attainment on suicide risk. One of those Hungarian investigations was a psychological autopsy study (n ¼194), the results of which confirmed the association between elevated suicide risk among those with low educational attainment and/or not living in marriage (Almasi et al., 2009). An other psychological autopsy study (n ¼100) did not demonstrate considerable associations between risk of suicide and marital status or educational attainment (Zonda, 2006). A short preliminary paper reported that in 1985 regional suicide rates contributed significantly to the multiple regression prediction of regional divorce rates (Nr. of regions ¼20) (Lester and Rihmer, 1997). Osváth et al. (2003) compared the characteristics of repeated suicide attempters with first suicide attempters. They reported that in the whole sample divorced (compared to married) status and low and middle (compared to high) educational statuses were significantly associated with higher risks of repeated suicide attempts (interestingly, divorced status was associated with higher risks for repeated suicide only among females and the effect of educational attainment was also more consistent among females). Considering that there is a paucity of large-scale epidemiological research on the relationship between marital status and educational attainment and the risk of completed suicide in Hungary we aimed primarily to investigate the effects of educational attainment and marital status on the risk of completed suicide and secondarily to assess whether the strength of the supposed effects have changed in the last three decades.
2. Data and methods 2.1. Data Census data stratified by age, gender, marital and educational status for the years 1980, 1990, 2001 and 2011 were used for the general population. Corresponding data from the same years for suicide victims derived from the Demographic Register of the Hungarian Central Statistical Office (“DEMO”). Accordingly, the latter database consisted routinely collected data. It follows from the foregoing that we used “unlinked data”, which may be the source of the so-called “numerator-denominator bias” (Shkolnikov et al., 2007). “Marital status” and “educational attainment” variables had four (married; unmarried; divorced; widowed) and three (not higher than vocational school (i.e. graduated from a school ranked not higher than levels 2C or 3C according to the International Standard Classification of Education (ISCED) system); graduated from schools awarding a secondary school leaving certificate, the prerequisite for entry into colleges and universities (ISCED3A); graduated from university or college (ISCED5A)) categories, respectively (for the Hungarian education system see www.cedefop. europa.eu/files/4103_EN.pdf and http://www.ofi.hu/isced-az-okta tas-egyseges-nemzetkozi-osztalyozasi-rendszere). Although age is presented as a five-category variable (aged 20–29, 30–39, 40–49, 50–59, 60r) in Table 1, it was used as a continuous variable in the regression models. Our data were stratified in an identical manner in every census year, which enabled us to ascertain some temporal
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
779
Table 1 Number of suicide cases and age-standardized suicide rates (per 100,000 capita per year) by marital status and educational attainment of the population above 20 years of age in Hungary in 1980, 1990, 2001 and 2011. Variables
1980
1990
Male
Female
Cases Rates
420 554 610 668 1001
2001
Male
Female
2011
Male
Female
Male
Female
Cases Rates Cases Rates Cases Rates Cases Rates
Cases Rates
Cases Rates
Cases Rates
63.6 68.2 89.2 119.8 127.1
96 163 229 264 680
15.0 20.1 32.3 40.8 58.0
249 563 640 496 958
37.7 69.3 93.6 89.0 121.7
64 126 199 174 557
10.0 15.6 28.1 26.9 47.5
196 313 600 459 662
24.0 47.5 80.2 74.6 82.3
39 55 126 136 327
5.0 8.5 16.0 19.5 25.6
122 220 336 474 547
19.4 27.6 51.3 69.8 60.0
11 46 83 120 261
1.8 5.9 12.6 15.8 18.4
2074 471 333 375
81.3 107.9 136.1 290.5
613 120 159 540
25.6 51.9 45.5 58.5
1664 465 432 345
63.4 113.9 146.2 170.8
420 83 144 433
19.6 35.6 38.3 34.1
1057 441 453 279
43.1 75.3 127.6 151.2
217 68 119 279
9.3 17.4 23.8 21.0
695 429 394 181
31.6 54.2 83.6 272.0
178 49 124 170
8.4 11.1 16.9 15.0
2911 215
116.8 39.3
1212 159
39.2 31.1
2536 273
102.1 51.2
892 173
28.2 31.1
1799 331
78.23 41.08
531 126
18.19 9.62
1299 294
62.05 30.01
352 119
14.73 8.73
127 4 3257
34.8
61 5 1437
23.5
97 4 2910
25.7
55 1 1121
21.7
100 15 2245
20.80
26 1 684
6.17
106 115 1814
16.93
50 40 561
7.23
n
Age group 20–29 30–39 40–49 50–59 60-x Marital status Married Unmarried Divorced Widowed Educational attainment Not higher than vocational school Secondary grammar (or secondary vocational) school Graduated from university or college Missing cases Total n
95.1
34.8
85.2
27.1
63.0
15.1
45.9
11.3
Age specific rates.
properties of the explanatory variables (in other words the stability of effects of explanatory variables on the risk of suicide). Data of suicide victims derived from the “DEMO” were almost complete in the first three census years investigated (i.e. 1980, 1990 and 2001) but there were missing data for educational attainment in 155 cases (6.5%) from the year 2011 (however, data on marital status and the gender of suicide victims were complete also for 2011) (Table 1). We excluded suicide victims younger than 20 years of age, since it is reasonable to presume that neither educational level nor marital status should be considered final in most of the members of that young age group. In accordance with previous literature (e.g. Vyssoki et al., 2014), an individual whose International Classification of Diseases-9 (ICD-9) or ICD-10 code in the paragraph “cause of death” was one of the following was considered a suicide completer: ICD-9: E950 to E959; ICD-10: X60 to X84 and Y87.0 (cases were coded by ICD-9 criteria in 1980 and 1990 and by ICD10 criteria in 2001 and 2011). 2.2. Data analysis and statistical methods By taking the age composition of the Hungarian population in 1980 as a standard, age-standardized suicide rates presented in Table 1 were calculated using the direct method. As a first approach to estimate incidence rate ratios (IRRs) and their 95% confidence intervals (CIs), Poisson regression models with the number of completed suicide cases as the response variable were fitted to the data with a canonic log link function and the corresponding stratum-specific population as an offset variable. Due to the limited number of cells the effect of age was estimated as a continuous variable (single parameter). Poisson regression was employed for every census year, both for the total population and separately for males and females. The most educated (‘graduated from university or college’) and married were chosen as reference categories in each model. Furthermore, in models for the total population female gender was an additional reference category. The main assumption for the Poisson model is that the conditional variance is equal to the conditional mean. If the equidispersion assumption is violated the model does not fit well.
Dispersion was assessed with Pearson statistic divided by the model degrees of freedom (Hilbe, 2011). Then score test was performed to evaluate whether overdispersion is present in the model (Dean and Lawless, 1989). Because, using the above methods, we found remarkable overdispersion in each model the extra variation was taken into account employing negative binomial regression. All statistical procedures were conducted using Stata 12 software (StataCorp. 2011).
3. Results Table 1 shows the Hungarian suicide numbers and rates by gender, age, marital status and educational attainment in 1980, 1990, 2001 and 2011 (the number of cases with incomplete data by gender are also presented in this table). In the four years investigated, suicide rates for both genders were highest in 1980 and – after a monotonic decrease – were lowest in 2011. This finding is consistent with the results of several previous studies on the Hungarian suicide rate (e.g. Rihmer et al., 2013). The results of negative binomial regression analyses on the effects of marital status, educational attainment, age and gender on suicide risk are presented for different years in Tables 2–5. In the total population, married status was associated with a significantly decreased risk of suicide compared to other marital statuses (i.e. unmarried, divorced or widowed) in all years investigated (Tables 2–5). Similarly, in male and female subpopulations unmarried/divorced/widowed statuses (compared to married status) pose an extra risk of suicide (of the 24 comparisons, only 2 were non-significant (2011: unmarried females vs. married females; 2011: widowed females vs. married females)) (Tables 2– 5). So we found that suicide risks for unmarried/divorced/widowed males (compared to married males) were significantly higher in each comparison while the same is not entirely true for females. In addition, in the majority of cases (i.e. in 8 out of 12 cases) the incidence rate ratios of unmarried/divorced/widowed males were higher than those of unmarried/divorced/widowed females, respectively (Tables 2–5). Some authors have suggested that in the past decades – as divorce has become more frequent in modern societies – the stigma associated with it has decreased,
780
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
Table 2 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 1980. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC
Total
Male
IRR
95% CI
1.214nnn 1.000
1.144–1.288
2.526nnn 1.000
2.150–2.966
1.721nnn 2.084nnn 2.637nnn 1.000 1.118 2.306nnn 0.17 347.31 5.94
IRR
Female 95% CI
IRR
95% CI
1.146nnn
1.085–1.211
1.303nnn
1.194–1.421
1.388–2.134 1.698–2.558 2.064–3.368
1.259n 1.793nnn 2.249nnn
1.030–1.538 1.457–2.208 1.766– 2.864
2.160nnn 2.152nnn 2.791nnn
1.571–2.969 1.587–2.919 1.972–3.950
0.875–1.427 1.853–2.871
1.093 3.120nnn 0.19 173.46 6.05
0.842–1.418 2.484–3.918
1.093 1.670nn 0.17 160.71 5.62
0.763–1.566 1.204–2.316
n
po 0.05. p o0.01. nnn p o0.001. nn
leading ultimately to a decrease in the number of stigma-driven suicide cases (Stack and Scourfield, 2015). Although, in the period investigated, a continuously growing proportion of marriages ended in divorce in Hungary (total divorce rates were 0.29 in 1980 and 0.46 in 2011), we were not able to demonstrate that the suicide risk of divorced (compared to married) subjects had decreased (Bukodi et al., 2005; Földházi, 2012; KSH,online). In fact, contrary to our expectations, when compared to married subjects, the risks of suicide among divorced subjects in 1990, 2001 and 2011 were higher in the total population and also in the male subpopulation than in 1980 (in the female subpopulation the situation was the same for the years 1990 and 2001, but not for 2011) (Tables 2–5). Some results have indicated that the recency of changes in marital status, i.e. the length of time that has elapsed since divorce or being widowed, has a more pronounced effect on suicide risk than marital status per se (Stack and Scourfield, 2015; Ajdacic-Gross et al. 2008; Batterham et al., 2014). Unfortunately, our dataset did not allow us to test this phenomenon. Generally speaking higher educational levels were associated
with reduced suicide risks in our total sample, as well in the male and female subsamples. Accordingly, in all the years studied (1980, 1990, 2001 and 2011), the following association was observable in the total population and in the male subpopulation: the suicide risk of the least educated group (“not higher than vocational school”) was significantly higher than the corresponding risk for the most-educated group (“graduated from university or college”) (for females the same association was observable in all the studied years but the association did not reach the level of significance in 1990) (Tables 2–5). As discussed in the “Introduction” section, some studies have found that the inverse relationship between educational attainment and suicide risk is less marked or even the opposite among females. A somewhat similar phenomenon was also observable in our dataset, since (1) among females there were 3 years (1980, 1990 and 2011) when significant differences were not observable between the suicide risks of the groups “completed general secondary school” and “graduated from university or college” (i.e. the reference category), while there was only one such year (1980) in the case of the male subpopulation; (2) compared to
Table 3 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 1990. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC †
po 0.1. p o0.001.
nnn
Total
Male
Female
IRR
95% CI
IRR
95% CI
IRR
95% CI
1.301nnn 1.000
1.225–1.381
1.272nnn
1.192–1.357
1.333nnn
1.230–1.445
3.226nnn 1.000
2.758–3.774
1.921nnn 2.780nnn 2.686nnn 1.000
1.562–2.361 2.292–3.371 2.117–3.407
1.842nnn 2.954nnn 2.729nnn
1.480–2.293 2.388–3.653 2.051–3.632
1.736nnn 2.302nnn 2.290nnn
1.279–2.356 1.769–2.996 1.706–3.073
1.738nnn 2.391nnn 0.22 324.14 5.55
1.359–2.223 1.905–3.001
1.911nnn 3.528nnn 0.23 168.77 5.89
1.436–2.543 2.724–4.568
1.371† 1.316† 0.20 139.68 4.923
0.974–1.928 0.960–1.803
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
781
Table 4 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 2001. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC
Total
Male
Female
IRR
95% CI
IRR
95% CI
IRR
95% CI
1.225nnn 1.000
1.137–1.320
1.207nnn
1.098–1.326
1.260nnn
1.119–1.418
4.749nnn 1.000
3.948–5.713
1.704nnn 2.944nnn 2.638nnn 1.000
1.326–2.190 2.345–3.697 1.989–3.499
1.634nn 3.117nnn 3.041nnn
1.209–2.208 2.342–4.148 2.090–4.424
1.855nn 2.671nnn 2.191nnn
1.204–2.858 1.856–3.844 1.436–3.343
1.890nnn 3.099nnn 0.220 311.43 5.34
1.430–2.499 2.384–4.028
1.856nnn 3.297nnn 0.17 178.15 6.20
1.329–2.590 2.415–4.502
1.886n 2.743nnn 0.14 131.48 4.65
1.157–3.074 1.722–4.370
n
po 0.05. p o0.01. nnn p o 0.001. nn
the reference group, the suicide risks of the least educated subjects (i.e. members of the “not higher than vocational school” group) were more pronounced in males than in females in each year investigated; accordingly, for the least educated males IRRs were above 3.1 in each year, but for the least educated females IRRs were lower than 2.75 in each year; (3) as we discussed previously, among females there was a year (1990) when there was no significant difference in the risk of suicide between the least- and the most-educated groups (while there was not such a year in the male group) (Tables 2–5). An additional result may also be discerned from the data: looking at the year representing the socialist era (1980) and the year immediately following the transition (1990) grouped together, we can see that the suicide risk of the least-educated group in the total population was lower than in 2001 and 2011 (IRRs were 2.3; 2.4; 3.1 and 3.0, respectively) (Tables 2–5). It seems that this change in the total population during the three decades investigated was driven primarily by females (IRRs in the four years were 1.7; 1.3; 2.7 and 2.1,
respectively) since in males IRRs were roughly the same in the years of socialism (3.1 and 3.5) and in the years after transition (3.3 and 3.5). Our results suggest that in Hungary total, male, and female suicide rates increase with age (Tables 2–5). Moreover – regardless of gender – the effect of age was relatively stable over the period studied. According to our results, the effect of gender was highly significant in all years. We found that male gender carries a higher risk of suicide than female (Tables 2–5). Controlling for age, marital status and educational level, the risk of taking their own lives was about 4–5 times higher for males than for females in 2001 and in 2011. This effect was somewhat smaller in 1980 and in 1990, probably due to the much larger decrease in the standardized suicide rate of females than of males between 1980 and 2011 (in this period the suicide rate decreased by two-thirds in females, while it only halved in males) (Table 1).
Table 5 Incidence rate ratios (IRR) for suicide using negative binomial regression in the year 2011. Variables
Age Gender (reference category ¼female) Male Marital status (reference category ¼married) Unmarried Divorced Widowed Educational attainment (reference category ¼graduated from university or college) Completed general secondary school Not higher than vocational school Pseudo R2 Log Likelihood AIC †
p o0.1. p o0.01. p o 0.001.
nn
nnn
Total
Male
Female
IRR
95% CI
IRR
95% CI
IRR
95% CI
1.330nnn 1.000
1.239–1.428
1.278nnn
1.177–1.389
1.387nnn
1.229–1.566
4.243nnn 1.000
3.589–5.017
1.346nn 2.509nnn 2.149nnn 1.000
1.082–1.673 2.043–3.082 1.633–2.828
1.493nn 2.765nnn 2.885nnn
1.161–1.919 2.151–3.554 2.006–4.148
1.028 2.144nnn 1.372†
0.702–1.506 1.614–2.848 0.978–1.925
1.593nnn 2.985nnn 0.25 294.64 5.06
1.252–2.027 2.385–3.737
1.748nnn 3.485nnn 0.21 168.11 5.870
1.305–2.342 2.660–4.567
1.280 2.141nnn 0.21 119.05 4.24
0.894–1.833 1.526–3.003
782
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783
4. Discussion Suicide mortality in Hungary was unexpectedly high in the previous century, and the Hungarian suicide rate is still the second highest in the European Union (EU27) (Szanto et al., 2007; WHO). Despite the disappointing features of the Hungarian suicide scene, according to our best knowledge, there is a lack of large-scale multivariate epidemiological research on the relationship between marital status and educational attainment and the risk of completed suicide in Hungary. The longitudinal observation of the effects of various suicideinfluencing factors is substantiated at least by two facts. Firstly, annual suicide rates in Hungary showed an almost monotonic decrease from the mid-1980s. Secondly, relevant political (i.e. the fall of the communist regime) and economic (i.e. the transition from a centrally planned economy to a market economy) changes occurred in Hungary in the era investigated (1980–2011). Furthermore, intensive societal fluctuations also occurred during the observation period, such as the change in proportions of preponderant family structure types. Accordingly, the proportion of marriages decreased from 80% to 65% between 1990 and 2011. On the other hand, the proportion of cohabitations and single-parent families (together) increased by 15% in the same period (KSH, 2014). Similar changes (e.g. the gradual decline in marriages and the increase in cohabitations) are typical in all regions of Europe (Sobotka and Toulemon, 2008). Our main findings based on Hungarian data are consistent with the majority of previous results from other countries (see the discussion in the “Introduction” section). Accordingly, we found that (1) in each population investigated (i.e. total; male; female) married status, in the great majority of cases, was associated with a significantly decreased risk of suicide compared to other marital statuses (i.e. unmarried, divorced or widowed) (Tables 2–5). In line with the results of several studies from other countries (e.g. Fukuchi et al., 2013; Kõlves et al., 2010; Kposowa, 2000, 2003; Masocco et al., 2010) we found weaker effects of marital status on suicide risk among females compared to males. However, our results do not confirm (and, indeed, rather disprove) the theory that the suicide risk of divorced individuals had decreased as divorce has become more frequent (Stack and Scourfield, 2015); (2) generally speaking higher educational levels were associated with reduced suicide risks in our total sample, as well in the male and female subsamples. This finding is concordant with the majority of results from other countries (Nock et al., 2008; Reques et al., 2014; Abel and Kruger, 2005; Ten Have et al., 2013; Flavio et al., 2013; Burrows and Laflamme, 2010; Beautrais, 2003; Denney et al., 2009; Kposowa, 2000; Mackenbach et al., 2015; Tjepkema et al., 2012; Lorant et al., 2005; Ishtiak-Ahmed et al., 2013; Lee et al., 2009; Milner et al., 2013a). Results of some previous studies have suggested that the inverse relationship between educational attainment and suicide risk is less marked (or even the opposite) among females (Reques et al., 2014; Denney et al., 2009; Kposowa, 2000; Tjepkema et al., 2012; Lorant et al., 2005; Lee et al., 2009; Ross et al., 2012; Strand et al., 2010; Mäki and Martikainen, 2009). In line with the results of the above studies, our findings also indicate a weaker association between educational attainment and suicide risk among females than among males. Our finding about the growing difference between suicide risks of the highest and the lowest educated groups before and after the political transition is consistent with the results of a previous Hungarian study on mortality due to external causes (a category which consists, inter alia, of suicide) based on data from 1986 to 2005 (Hablicsek and Kovács, 2007). One of the several possible explanations for this finding may be that the egalitarian approach of the socialist health care system (e.g. free health care provision – of the same quality – became a civil right after 1975) was better able to balance the
detrimental effects of low education on suicide risk (Horváth, 2011); (3) in Hungary total, male, and female suicide rates increase with age; and (4) male gender carries a higher risk of completed suicide than female. The latter two findings are also in line with the majority of results from developed countries (Rihmer et al., 2013; Hawton and van Heeringen, 2009). There are some limitations to the present study: (1) data on the length of the period between the change in marital status and suicide were not available, so we were not able to test whether recent changes in marital status have different effects on suicide risk from distant ones; (2) our four categories are not suitable to cover the whole spectrum of marital statuses of a modern society (e.g. we did not have specific categories for people who are separated (but not divorced by law) or for those who are living in cohabitation). Of course acquiring data on the quality of marriages was beyond our power; (3) as we have already mentioned before in the “Data and methods” section there were some missing data for the year 2011; (4) the use of “unlinked data” may be the source of the so-called “numerator-denominator bias” (Shkolnikov et al., 2007); (5) ecological study design. In conclusion, we found that in Hungary – similarly to the great majority of previous results from other countries – female gender, higher educational attainment, lower age and living in marriage were significantly associated with decreased risks of completed suicide in the period between 1980 and 2011. The effects of investigated socio-economic determinants of suicide have not changed dramatically over the decades of our observation period.
Acknowledgement Peter Dome is recipient of the János Bolyai Postdoctoral Fellowship of the Hungarian Academy of Sciences.
References Abel, E.L., Kruger, M.L., 2005. Educational attainment and suicide rates in the United States. Psychol. Rep. 97, 25–28. Ajdacic-Gross, V., Ring, M., Gadola, E., Lauber, C., Bopp, M., Gutzwiller, F., Rössler, W., 2008. Suicide after bereavement: an overlooked problem. Psychol. Med. 38, 673–676. Almasi, K., Belso, N., Kapur, N., Webb, R., Cooper, J., Hadley, S., Kerfoot, M., Dunn, G., Sotonyi, P., Rihmer, Z., Appleby, L., 2009. Risk factors for suicide in Hungary: a case-control study. BMC Psychiatry 9, 45. Balint, L., Dome, P., Daroczi, G., Gonda, X., Rihmer, Z., 2014. Investigation of the marked and long-standing spatial inhomogeneity of the Hungarian suicide rate: a spatial regression approach. J. Affect. Disord. 155, 180–185. Barth, A., Sögner, L., Gnambs, T., Kundi, M., Reiner, A., Winker, R., 2011. Socioeconomic factors and suicide: an analysis of 18 industrialized countries for the years 1983 through 2007. J. Occup. Environ. Med. 53, 313–317. Batterham, P.J., Fairweather-Schmidt, A.K., Butterworth, P., Calear, A.L., Mackinnon, A.J., Christensen, H., 2014. Temporal effects of separation on suicidal thoughts and behaviours. Soc. Sci. Med. 111, 58–63. Beautrais, A.L., 2003. Suicide and serious suicide attempts in youth: a multiplegroup comparison study. Am. J. Psychiatry 160, 1093–1099. Nők és férfiak Magyarországon, 2004. In: Bukodi, E., Mészárosné Halász, J., Polónyi, K., Tallér, A. (Eds.), Ifjúsági, Családügyi, Szociális és Esélyegyenlőségi Minisztérium. KSH, Budapest, p. 16. Burrows, S., Laflamme, L., 2010. Socioeconomic disparities and attempted suicide: state of knowledge and implications for research and prevention. Int. J. Inj. Control Saf. Promot. 17, 23–40. Corcoran, P., Nagar, A., 2010. Suicide and marital status in Northern Ireland. Soc. Psychiatry Psychiatr. Epidemiol. 45, 795–800. Cubbin, C., LeClere, F.B., Smith, G.S., 2000. Socioeconomic status and the occurrence of fatal and nonfatal injury in the United States. Am. J. Public Health 90, 70–77. Dean, C., Lawless, J.F., 1989. Tests for detecting overdispersion in Poisson regression models. J. Am. Stat. Assoc. 84, 467–472. Denney, J.T., Rogers, R.G., Krueger, P.M., Wadsworth, T., 2009. Adult suicide mortality in the united states: marital status, family size, socioeconomic status, and differences by sex. Soc. Sci. Q. 90, 1167. Durkheim, É., 2005. Suicide. A study in Sociology. Routledge, London and New York, first published 1897. Flavio, M., Martin, E., Pascal, B., Stephanie, C., Gabriela, S., Merle, K., Anita, R.R., 2013. Suicide attempts in the county of Basel: results from the WHO/EURO Multicentre Study on Suicidal Behaviour. Swiss Med. Wkly. 143, w13759.
L. Bálint et al. / Journal of Affective Disorders 190 (2016) 777–783 Földházi, E., 2012. Divorce. In: Zsolt, Spéder, Péter, Őri (Eds.), Demographic Portrait of Hungary. The Report on the Condition of the Hungarian Population. Demographic Research Institute, HCSO, Budapest, pp. 21–30. Fukuchi, N., Kakizaki, M., Sugawara, Y., Tanji, F., Watanabe, I., Fukao, A., Tsuji, I., 2013. Association of marital status with the incidence of suicide: a populationbased Cohort Study in Japan (Miyagi cohort study). J. Affect. Disord. 150, 879–885. Gove, W.R., 1973. Sex, marital status, and mortality. Am. J. Sociol. 79, 45–67. Hablicsek, L., Kovács, K., 2007. Az életkilátások Differenciálódása Iskolázottság Szerint, 1986–2005. KSH Népességtudományi Kutatóintézet, Budapest, pp. 89–93. Hawton, K., van Heeringen, K., 2009. Suicide. Lancet 373, 1372–1381. Hilbe, J.M., 2011. Negative Binomial Regression, Second edition. Cambridge University Press, Cambridge, UK. Horváth, G.K., 2011. Magyarország társadalomtörténete a szocialista korban. Ishtiak-Ahmed, K., Perski, A., Mittendorfer-Rutz, E., 2013. Predictors of suicidal behaviour in 36,304 individuals sickness absent due to stress-related mental disorders – a Swedish register linkage cohort study. BMC Public Health 13, 492. Kõlves, K., Ide, N., De Leo, D., 2010. Suicidal ideation and behaviour in the aftermath of marital separation: gender differences. J. Affect. Disord. 120, 48–53. Kposowa, A.J., 2000. Marital status and suicide in the National Longitudinal Mortality Study. J. Epidemiol. Community Health 54, 254–261. Kposowa, A.J., 2003. Divorce and suicide risk. J. Epidemiol. Community Health 57, 993. KSH, 2014. Családtípusok* és párkapcsolati formák változása a népszámlálási adatok tükrében. Stat Tükör. VII/115. (In Hungarian). 〈www.ksh.hu/docs/hun/ xftp/stattukor/csaladtipusok.pdf〉. KSH online. 〈https://www.ksh.hu/thm/2/indi2_1_3.html〉. Lee, W.Y., Khang, Y.H., Noh, M., Ryu, J.I., Son, M., Hong, Y.P., 2009. Trends in educational differentials in suicide mortality between 1993–2006 in Korea. Yonsei Med. J. 50, 482–492. Lester, D., Rihmer, Z., 1997. The regional variation of divorce in Hungary. J. Divorce Remarriage 26, 83–85. Lewis, G., Sloggett, A., 1998. Suicide, deprivation, and unemployment: record linkage study. BMJ 317, 1283–1286. Liu, Y., Zhang, Y., Cho, Y.T., Obayashi, Y., Arai, A., Tamashiro, H., 2013. Gender differences of suicide in Japan, 1947–2010. J. Affect. Disord. 151, 325–330. Lorant, V., Kunst, A.E., Huisman, M., Costa, G., Mackenbach, J., EU Working Group on Socio-Economic Inequalities in Health, 2005. Socio-economic inequalities in suicide: a European comparative study. Br. J. Psychiatry 187, 49–54. Luoma, J.B., Pearson, J.L., 2002. Suicide and marital status in the United States, 1991–1996: is widowhood a risk factor? Am. J. Public Health 92, 1518–1522. Mackenbach, J.P., Kulhánová, I., Bopp, M., Deboosere, P., Eikemo, T.A., Hoffmann, R., Kulik, M.C., Leinsalu, M., Martikainen, P., Menvielle, G., Regidor, E., Wojtyniak, B., Ostergren, O., Lundberg, O., for the EURO-GBD-SE Consortium, 2015. Variations in the relation between education and cause-specific mortality in 19 European populations: A test of the “fundamental causes” theory of social inequalities in health. Soc. Sci. Med. 127, 51–62. Mäki, N., Martikainen, P., 2009. The role of socioeconomic indicators on non-alcohol and alcohol-associated suicide mortality among women in Finland. A register-based follow-up study of 12 million person-years. Soc. Sci. Med. 68, 2161–2169. Manzoli, L., Villari, P., M Pirone, G., Boccia, A., 2007. Marital status and mortality in the elderly: a systematic review and meta-analysis. Soc. Sci. Med. 64, 77–94. Masocco, M., Pompili, M., Vanacore, N., Innamorati, M., Lester, D., Girardi, P., Tatarelli, R., Vichi, M., 2010. Completed suicide and marital status according to the Italian region of origin. Psychiatr. Q. 81, 57–71. Milner, A., Hjelmeland, H., Arensman, E., De Leo, D., 2013a. Social-environmental factors and suicide mortality: a narrative review of over 200 articles. Sociol. Mind 3, 137–148. Milner, A., Spittal, M.J., Pirkis, J., LaMontagne, A.D., 2013b. Suicide by occupation: systematic review and meta-analysis. Br. J. Psychiatry 203, 409–416. Morselli, H., 1882. Suicide: An Essay on Comparative Moral Statistics. D. Appleton and company, New York. Nock, M.K., Borges, G., Bromet, E.J., Alonso, J., Angermeyer, M., Beautrais, A., Bruffaerts, R., Chiu, W.T., de Girolamo, G., Gluzman, S., de Graaf, R., Gureje, O., Haro, J.M., Huang, Y., Karam, E., Kessler, R.C., Lepine, J.P., Levinson, D., Medina-Mora, M.E., Ono, Y., Posada-Villa, J., Williams, D., 2008. Cross-national prevalence and
783
risk factors for suicidal ideation, plans and attempts. Br. J. Psychiatry 192, 98–105. Osváth, P., Kelemen, G., Erdös, M.B., Vörös, V., Fekete, S., 2003. The main factors of repetition: review of some results of the Pecs Center in the WHO/EURO Multicentre Study on Suicidal Behaviour. Crisis 24, 151–154. Pompili, M., Vichi, M., Qin, P., Innamorati, M., De Leo, D., Girardi, P., 2013. Does the level of education influence completed suicide? A nationwide register study. J. Affect. Disord. 147, 437–440. Qin, P., Agerbo, E., Westergård-Nielsen, N., Eriksson, T., Mortensen, P.B., 2000. Gender differences in risk factors for suicide in Denmark. Br. J. Psychiatry 177, 546–550. Reques, L., Giráldez-García, C., Miqueleiz, E., Belza, M.J., Regidor, E., 2014. Educational differences in mortality and the relative importance of different causes of death: a 7-year follow-up study of Spanish adults. J. Epidemiol. Community Health 68, 1151–1160. Rihmer, Z., Gonda, X., Kapitany, B., Dome, P., 2013. Suicide in Hungary-epidemiological and clinical perspectives. Ann. Gen. Psychiatry 12, 21. Robards, J., Evandrou, M., Falkingham, J., Vlachantoni, A., 2012. Marital status, health and mortality. Maturitas 73, 295–299. Ross, C.E., Masters, R.K., Hummer, R.A., 2012. Education and the gender gaps in health and mortality. Demography 49, 1157–1183. Roškar, S., Podlesek, A., Kuzmanić, M., Demšar, L.O., Zaletel, M., Marušič, A., 2011. Suicide risk and its relationship to change in marital status. Crisis 32, 24–30. Scourfield, J., Evans, R., 2015. Why might men be more at risk of suicide after a relationship breakdown? Sociological insights. Am. J. Mens Health 9, 380–384. Shkolnikov, V.M., Jasilionis, D., Andreev, E.M., Jdanov, D.A., Stankuniene, V., Ambrozaitiene, D., 2007. Linked versus unlinked estimates of mortality and length of lifeby education and marital status: evidence from the first record linkage study in Lithuania. Soc. Sci. Med. 64, 1392–1406. Shor, E., Roelfs, D.J., Curreli, M., Clemow, L., Burg, M.M., Schwartz, J.E., 2012. Widowhood and mortality: a meta-analysis and meta-regression. Demography 49, 575–606. Smith, J.C., Mercy, J.A., Conn, J.M., 1988. Marital status and the risk of suicide. Am. J. Public Health 78, 78–80. Sobotka, T., Toulemon, L., 2008. Overview chapter 4: changing family and partnership behaviour: Common trends and persistent diversity across europe. Demogr. Res. 19, 85–138. Stack, S., 2000. Suicide: a 15-year review of the sociological literature. Part II: modernization and social integration perspectives. Suicide Life Threat. Behav. 30, 163–176. Stack, S., Scourfield, J., 2015. Recency of divorce, depression, and suicide risk. J. Fam. Issues 36, 695–715. Strand, B.H., Grøholt, E.K., Steingrímsdóttir, O.A., Blakely, T., Graff-Iversen, S., Naess, Ø., 2010. Educational inequalities in mortality over four decades in Norway: prospective study of middle aged men and women followed for cause specific mortality, 1960–2000. BMJ 340, c654. Szanto, K., Kalmar, S., Hendin, H., Rihmer, Z., Mann, J.J., 2007. A suicide prevention program in a region with a very high suicide rate. Arch. Gen. Psychiatry 64, 914–920. Ten Have, M., van Dorsselaer, S., de Graaf, R., 2013. Prevalence and risk factors for first onset of suicidal behaviors in the Netherlands mental health survey and incidence study-2. J. Affect. Disord. 147, 205–211. Tjepkema, M., Wilkins, R., Long, A., 2012. Cause-specific mortality by education in Canada: a 16-year follow-up study. Health Rep. 23, 23–31. Värnik, P., 2012. Suicide in the world. Int. J. Env. Res. Public Health 9, 760–771. Vyssoki, B., Kapusta, N.D., Praschak-Rieder, N., Dorffner, G., Willeit, M., 2014. Direct effect of sunshine on suicide. JAMA Psychiatry 71, 1231–1237. WHO. 〈http://data.euro.who.int/hfadb/〉 (data retrieved in September 2014). Wyder, M., Ward, P., De Leo, D., 2009. Separation as a suicide risk factor. J. Affect. Disord. 116, 208–213. Yamauchi, T., Fujita, T., Tachimori, H., Takeshima, T., Inagaki, M., Sudo, A., 2013. Ageadjusted relative suicide risk by marital and employment status over the past 25 years in Japan. J. Public Health 35, 49–56. Yip, P.S., Thorburn, J., 2004. Marital status and the risk of suicide: experience from England and Wales, 1982–1996. Psychol. Rep. 94, 401–407. Zonda, T., 2006. One-hundred cases of suicide in Budapest: a case-controlled psychological autopsy study. Crisis 27, 125–129.