Open access Research BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. Socioeconomic inequalities in suicide mortality in Barcelona during the economic crisis (2006–2016): a time trend study Natalia López-Contreras,1,2 Maica Rodríguez-Sanz,1,3,4,5 Ana Novoa,2,5 Carme Borrell,1,3,5,6 Jordi Medallo Muñiz,7 Mercè Gotsens2,3 To cite: López-Contreras N, Abstract Rodríguez-Sanz M, Novoa A, Objectives This study aimed to analyse trends in Strengths and limitations of this study et al. Socioeconomic socioeconomic inequalities in suicide mortality in inequalities in suicide mortality ►► Data were obtained from the Judicial Mortality Barcelona before and after the start of the economic in Barcelona during the Registry of Barcelona, which is up-to-date and pre- crisis that started at the end of 2008, including both economic crisis (2006–2016): vents under-reporting of suicide as a cause of death. a time trend study. BMJ Open individual factors and contextual factors of the deceased’s ►► The analysis includes socioeconomic variables, both 2019;9:e028267. doi:10.1136/ neighbourhood of residence. social and material, which proved to be very useful bmjopen-2018-028267 Design This is a trend study of three time periods: pre- for suicide surveillance during the economic crisis. economic crisis (2006–2008), early crisis (2009–2012) ►► Prepublication history and ►► The low number of suicides in Barcelona, especially and late crisis (2013–2016). additional material for this among women, hindered detection of significant as- Setting Total Barcelona residents between 2006 and paper are available online. To sociations and stratified analyses. Protected by copyright. view these files, please visit 2016 (≥25 years of age) and death data derived from the ►► A change in Barcelona’s territorial divisions in 2009 the journal online (http://​dx.​doi.​ Judicial Mortality Registry of Barcelona. hampered collection of contextual data for the time org/​10.​1136/​bmjopen-​2018-​ Participants 996 deaths by suicide between 2006 and periods studied. 028267). 2016 were analysed. Primary and secondary outcome measures The Received 29 November 2018 outcomes were age-standardised suicide mortality Revised 17 July 2019 rates and the associations (relative and absolute risk) that experienced by western countries since Accepted 22 July 2019 between suicide mortality and individual and contextual 2008. characteristics for the three time periods. Like other health issues, suicide is influ- Results From 2006 to 2008, men with a lower enced by various social determinants. We educational level were more likely to commit suicide than have summarised these determinants in a better educated men (relative risk (RR)=1.46; 95% CI conceptual framework (figure 1), drawing 1.11 to 1.91). This difference disappeared after the onset on information previously presented by the of the crisis. We found no differences among women. From 2013 to 2016, suicide risk increased among men Commission on Reducing Social Inequalities in living in neighbourhoods with higher unemployment levels Spain (Comisión para Reducir las Desigual- (RR=1.57; 95% CI 1.09 to 2.25) and among women living dades Sociales en Salud en España).2 These in neighbourhoods with a higher proportion of elderly determinants can be classified as either struc- people living alone (RR=2.13; 95% CI 1.15 to 3.93). tural or intermediary. In terms of structural Conclusions We observed risks for suicide among determinants, both the socioeconomic and men living in neighbourhoods of Barcelona with higher the political features of a country can influ- unemployment levels and among women living in ence suicide mortality rates. Because macro- neighbourhoods with a higher proportion of elderly people economic and welfare state policies depend living alone. Inequalities in suicide mortality according to on the government in power, as well as on © Author(s) (or their educational level tended to disappear during the crisis political traditions, austerity measures and employer(s)) 2019. Re-use among men. Thus, it is important to continue to monitor suicide determinants especially in times of economic reduced social benefits implemented during permitted under CC BY-NC. No commercial re-use. See rights crisis. an economic crisis tend to increase suicide and permissions. Published by rates.3 In addition, social cohesion is known BMJ. to be a protective factor against suicide, with For numbered affiliations see Introduction previous studies showing that high social trust end of article. Suicide is a major problem for public health. (as part of social capital)4 5 and high electoral Correspondence to Suicide rates are often considered as a participation (voter turnout) are associated Natalia López-Contreras; measure of a country’s emotional well-being,1 with fewer suicides.6 7 Intermediary determi- ​ext_​nlopez@​aspb.​cat especially in times of economic crisis such as nants, which are very important for suicide López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 1 Open access BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. Figure 1 Conceptual framework of the determinants of inequalities in suicide. This figure was adapted from the conceptual framework of the determinants of health inequalities developed by the Commission on Reducing Social Inequalities in Spain (Comisión para Reducir las Desiguald ades Sociales en Salud en España).2 Protected by copyright. risk, can be divided into material and social resources. crisis, including both individual and contextual factors of Higher suicide rates are known to occur in areas with the deceased’s neighbourhood of residence. greater unemployment8 or less social or family support, as measured by the proportion of people living alone.9 More- over, mental illness is also a risk factor for suicide and is Methods involved in a high percentage of cases.10 Because all these Design and information sources determinants change according to axes of inequality, they We performed a trend study of the period 2006–2016, influence suicide in a non-uniform manner. For example, including all Barcelona residents aged ≥25 years. We particular groups are more affected, such as men,11 both gathered information from the Judicial Mortality Registry young and old adults,8 12 and individuals with a poorer (maintained by the Catalan Institute of Legal Medicine socioeconomic position—whether it be in terms of educa- and Forensic Sciences and the Municipal Population tion, income or occupation.12–16 Census of Barcelona) and the Drug Addiction Infor- While many studies have analysed trends in suicide mation System maintained by Barcelona Public Health mortality during economic crises,17 few have exam- Agency, the city’s Municipal Population Census and the ined concurrent trends in socioeconomic inequalities Department of Statistics of Barcelona City Council. in suicide mortality during these periods. These latter Variables and indicators studies have observed greater suicide mortality among The dependent variable was suicide mortality: codes men from a manual social class18 or with a lower level of X60-X84 of the International Classification of Diseases education.19 However, a more recent study conducted in 10th revision. Spain showed that the trend in educational inequalities Individual explanatory variables were (1) educational in suicide mortality was stable among men before and level (‘primary education or lower’ or ‘secondary educa- during the latest economic recession.20 Thus, further tion or higher’) as an indicator of individual socio- studies are required to investigate the impact of the economic position, (2) country of birth (‘Spain’ or economic crisis on social inequalities in suicide mortality, ‘elsewhere’), (3) neighbourhood of residence and (4) using updated data after the economic crisis by gender year of death (‘pre-economic crisis’ (2006–2008), ‘early and taking into account many aspects presented in our crisis’ (2009–2012) or ‘late crisis’ (2013–2016)). We used conceptual framework. gender as a stratification variable and age as an adjust- The objective of our study was to analyse trends in ment variable. For all independent variables, the propor- socioeconomic inequalities in suicide mortality in Barce- tion of missing data was <11%. An analysis of missing data lona before and after the onset of the recent economic was carried out. We observed a random distribution of 2 López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 Open access BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. the independent variables among people older than 25 elderly living alone; model 4: % voter turnout; model 5: years, with the education level and the address assigned, drug use index). We evaluated the model fit by using the and those who did not have this information. deviance and likelihood ratio between nested models (p Contextual explanatory variables related to the neigh- value). We then estimated the interaction of contextual bourhood of residence were chosen based on the concep- explanatory variables and time periods for each variable tual framework (figure 1): (1) level of unemployment (in using the time period as a categorical variable in the terms of the average annual percentage of individuals model (presenting the p value of the interaction). enrolled in the Employment Service of Catalonia) among individuals aged 16–64 years in 2011, (2) the percentage of individuals aged ≥65 years who lived alone in 2008, (3) Results the percentage of individuals who participated in general Between 2006 and 2016, 1178 residents of Barcelona elections in 2011 and (4) the index of high-risk drug aged >25 years committed suicide. In 996 of these indi- use in 2015 (a synthetic index that evaluates drug use viduals, we were able to retrieve information regarding by combining five indicators: the incidence of treatment their educational levels and neighbourhoods of residence initiation for substance abuse, the incidence of over- through the Judicial Mortality Registry. These individuals dose-induced mortality according to place of residence formed the final study sample and constituted the target and place of death, the incidence of drug-related hospital population for the statistical analyses. In total, 69.1% emergencies and the incidence of syringes found in the (n=688) of the study population were men. streets). All the contextual explanatory variables related The ASMR decreased among men (p=0.022) during to the neighbourhood of residence were categorised in the study period (table 1), while rates among women quartiles. remained stable. In all three periods, the suicide mortality rate was higher among men than women. Patient and public involvement Suicide mortality rate was higher among men born in This study analysed the Judicial Mortality Registry data- Spain compared with those born elsewhere. This differ- base. Therefore, we did not inform participants of the ence decreased with time because rates among men born research question, outcome measures or results. No outside of Spain increased during the last period of the Protected by copyright. participants were involved in the study, including its study (p=0.019). By educational level, the rate of suicide design, recruitment or conduct. There was no patient between 2006 and 2008 was higher among men with a adviser for the contributors’ statement. lower level of education. However, in this group of men, the rate decreased in the other time periods (p=0.002), Data analysis while men with a higher educational level had stable rate All analyses were carried out using the STATA V.13 over time. programme and were stratified by gender. Among women, suicide rates were highest during the First, we directly calculated age-standardised suicide pre-crisis period (2006–2008) among those born outside mortality rates (ASMRs) according to gender, educational Spain, but the number of suicides was very low in this level and country of birth for each time period using the group. In this group of women, the rate decreased in the study population described above. And we conducted other time periods (p=0.025), while the rate was higher Poisson regression models adjusted for age using the vari- for women born in Spain. By educational level, women able period in continuous to estimate trends across time. with a higher level of education had higher suicide Second, we calculated ASMRs for the neighbourhoods mortality rates in all three periods, and this rate remained represented in the quartile maps. We calculated the stable over time among in this group, as well as in women Spearman correlation between suicide mortality rates with a lower level of education. and the contextual variables. Analysis of the distribution of suicide mortality in the Finally, due to the hierarchical structure of the data, we various neighbourhoods of Barcelona (figure 2) revealed conducted a multilevel analysis by fitting a Poisson regres- no clear patterns for men or women in any of the time sion model with a random neighbourhood effect (random periods. Table 2 shows the Spearman correlations between constant) to control for variability between neighbour- suicide mortality rates and the contextual explanatory hoods. As we found no significant variability, we decided variables, and shows that suicide mortality in men was to fit Poisson regression models including individual and positively correlated with the index of high-risk drug use, contextual neighbourhood variables to estimate relative and negatively correlated with voter turnout. In contrast, risk (RR) and absolute risk (AR) of suicide mortality and suicide mortality among women was positively correlated 95% CI. Initially, we fitted a model (model 1) containing with voter turnout and with the proportion of elderly the individual variables and a term representing the people living alone. Unemployment was not significantly interaction between educational level and time period. correlated with the suicide rate in either men or women, In subsequent models, we added a different contextual although the correlation in men tended to be opposite variable and a term representing the interaction between to that in women. Online supplementary figure 1 shows this variable and the time period, thereby generating four the distribution in quartiles across the neighbourhoods of different models (model 2: % unemployment; model 3% Barcelona of average annual percentage of unemployed López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 3 4 Open access Table 1 Age-standardised suicide mortality rates (ASMRs)* stratified by gender, country of birth and educational level, Barcelona, 2006–2016 2006–2008 2009–2012 2013–2016 n Population ASMR (95% CI) n Population ASMR (95% CI) n Population ASMR (95% CI) P value Men Total 221 1 762 191 13.87 (12.04 to 15.90) 225 2 379 238 9.89 (8.62 to 11.29) 242 2 350 383 10.59 (9.28 to 12.02) 0.022 Country of birth  Spain 199 1 402 158 14.73 (12.71 to 16.98) 201 1 815 824 11.01 (9.53 to 12.66) 204 1 771 794 11.26 (9.75 to 12.94) 0.721  Elsewhere 22 360 033 6.14 (3.00 to 13.65) 24 563 414 7.26 (3.07 to 14.62) 38 578 589 8.71 (4.39 to 15.53) 0.019 Level of education  Primary or lower 96 533 198 16.69 (13.34 to 20.63) 69 659 573 8.95 (6.84 to 11.52) 63 500 705 9.34 (6.89 to 12.43) 0.002  Secondary or higher 125 1 228 993 11.71 (9.51 to 14.28) 156 1 719 665 9.43 (7.9 to 11.19) 179 1 849 678 10.09 (8.55 to 11.84) 0.577 Women Total 86 2 026 864 4.20 (3.36 to 5.19) 100 2 733 362 3.61 (2.94 to 4.40) 122 2 720 035 4.22 (3.49 to 5.04) 0.665 Country of birth  Spain 73 1 683 270 4.16 (3.24 to 5.25) 97 2 171 574 4.41 (3.55 to 5.41) 113 2 102 161 4.78 (3.90 to 5.80) 0.168  Elsewhere 13 343 594 6.57 (2.7 to 13.14) 3 561 788 0.46 (0.09 to 2.52) 9 617 874 2.04 (0.64 to 4.88) 0.025 Level of education  Primary or lower 32 753 070 3.74 (2.24 to 5.87) 33 908 527 2.99 (1.79 to 4.70) 42 733 161 3.85 (2.14 to 6.40) 0.333  Secondary or higher 54 1 273 794 4.47 (3.28 to 5.98) 67 1 824 835 3.76 (2.87 to 4.85) 80 1 986 874 4.27 (3.34 to 5.39) 0.661 P value=to compare age-standardised suicide mortality rates trends. *Rate per 100 000 inhabitants. López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 Protected by copyright. BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. Open access BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. Figure 2 Distribution in quartiles of age-standardised mortality rates in men during the periods (A) 2006–2008, (B) 2009–2012, (C) 2013–2016, and in women during the periods (D) 2006–2008, (E) 2009–2012, (F) 2013–2016. individuals for 2011, the percentage of people aged >65 risk of suicide during the early crisis (RR=1.56; 95% CI Protected by copyright. years living alone in 2008, the percentage of participation 1.08 to 2.25), those living in areas with the highest unem- in the general election in 2011 and the index of high-risk ployment had greater risk during the late crisis period drug use in 2015. (RR=1.57; 95% CI 1.09 to 2.25). During the pre-crisis Table 3 summarises the associations between suicide period, men living in neighbourhoods in the first quar- mortality in men and the individual and contextual char- tile of voter turned out to be more likely to commit acteristics for the three time periods studied. During suicide (RR=1.68; 95% CI 1.12 to 2.52). Nonetheless, the pre-crisis period, less well-educated men were more men living in neighbourhoods in the third quartile areas likely to commit suicide than those with a higher level of were less likely to commit suicide during the late crisis education (RR=1.46; 95% CI 1.11 to 1.91), whereas the period (RR=0.69; 95% CI 0.49 to 0.98), with significant level of risk in each group was similar during the early variation in the RR over time. In addition, the index of (RR=0.96; 95% CI 0.72 to 1.28) and late (RR=1.04; 95% high-risk drug use was a risk factor during the pre-crisis CI 0.77 to 1.40) crisis periods. While men living in areas period among men living in fourth quartile neighbour- with high unemployment (third quartile) had greater hoods (RR=1.53; 95% CI 1.04 to 2.24), and in the early Table 2 Spearman correlations between suicide mortality and the contextual variables during the three time periods, Barcelona, 2006–2016 ASMR 2006–2008 ASMR 2009–2012 ASMR 2013–2016 Men  % unemployment, 2011 0.01 0.11 0.22  % elderly people living alone, 2008 0.10 0.02 0.12  % voter turnout, 2011 −0.20 0.02 −0.28*  Index of high-risk drug use, 2015 0.13 0.02 0.25* Women  % unemployment, 2011 −0.12 −0.12 −0.19  % elderly people living alone, 2008 0.26* 0.16 0.39*  % voter turnout, 2011 0.19 0.02 0.23*  Index of high-risk drug use, 2015 −0.17 −0.04 −0.14 *Significant correlation, p<0.05. ASMR, age-standardised suicide mortality rate. López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 5 6 Table 3 Association between suicide mortality among men and individual and contextual factors for each time period, Barcelona, 2006–2016 Model 1 Model 2 Model 3 Model 4 Model 5 RR RR RR RR RR Men P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3 P interaction Open access Level of education  Primary or lower 1.46* 0.96 1.04 1.46* 0.89 0.97 1.46* 0.95 1.05 1.33* 0.95 0.97 1.36* 0.92 0.98  Secondary or higher 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.081 % unemployment  Q1 (least) 1.00 1.00 1.00  Q2 0.97 1.41 1.16 0.212  Q3 0.86 1.56* 1.03  Q4 (most) 1.04 1.36 1.57* % elderly living alone  Q1 (least) 1.00 1.00 1.00  Q2 1.10 0.94 0.95 0.749  Q3 0.92 0.85 0.94  Q4 (most) 1.30 0.88 1.22 % voter turnout  Q1 (least) 1.68* 0.84 1.29  Q2 1.20 1.36 1.22 0.014  Q3 1.30 0.96 0.69*  Q4 (most) 1.00 1.00 1.00 Drug use index  Q1 (least) 1.00 1.00 1.00  Q2 1.08 1.08 1.03 0.192  Q3 1.03 1.66* 1.29  Q4 (most) 1.53* 1.02 1.37 Deviance 3412.89 3398.70 3404.99 3385.90 3394.63 P value 0.116 0.5448 0.0014 0.0323 P1, pre-crisis period (2006–2008); P2, early crisis period (2009–2012); P3, late crisis period (2013–2016). RR for each variable in each time period was obtained from the combined effect of the variable coefficients and the interaction between the variable and the period. P value=likelihood ratio test to compare models with and without contextual explanatory variables. P interaction=likelihood ratio test to compare models with and without the interaction of contextual explanatory variables and time period variable. All models were adjusted by age and country of birth. Model 1: level of education; model 2: level of education and % unemployment; model 3: level of education and % elderly living alone; model 4: level of education and % voter turnout; model 5: level of education and drug use index. *p<0.05 RR, relative risk. López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 Protected by copyright. BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. Open access BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. crisis period among those living in third quartile neigh- (2013–2016) crisis periods, men living in areas with higher bourhoods (RR=1.66: 95% CI 1.16 to 2.37). unemployment were more likely to commit suicide. In Table 4 summarises the associations between suicide women, we found the opposite, with those living in neigh- mortality in women and the individual and contextual bourhoods with high unemployment being less likely to characteristics for the three time periods studied. Educa- commit suicide. tional level showed no statistically significant differences It is widely accepted that poor mental health is associ- in suicide risk for any of the periods. In contrast, unem- ated with suicide mortality.10 During the economic crisis ployment proved to be protective against suicide during in Spain, men experienced a general deterioration in the late crisis among women living in neighbourhoods in mental health,26 especially those from lower socioeco- the third quartile of unemployment (RR=0.48; 95% CI nomic classes. We observed a general correlation between 0.27 to 0.84). Finally, women living in neighbourhoods suicide and the indicator for substance consumption, with the highest proportion of elderly people living alone although we did not have access to the relevant indi- were generally more likely to commit suicide during the vidual information to properly support this assertion, so late crisis period (RR=2.13; 95% CI 1.15 to 3.93), although it may be susceptible to ecological fallacy. More specifi- this risk did not vary significantly over time. cally, we found that men living in neighbourhoods with In general, the results obtained from ARs have the problematic drug consumption had a higher risk of same pattern as those observed for the RRs in both men suicide (although the statistical significance of this factor and women. In men, the AR decreases in those with lower decreased in the late crisis period). levels of education and increases in men living in neigh- The material and social factors that influence suicide bourhoods with higher unemployment. In women, the mortality appear to differ between men and women, AR increases in women who live in neighbourhoods with which may be because traditional gender roles create more elderly people living alone (online supplementary more pressure on men in terms of income and unem- table 1). ployment. In contrast, job loss among women is culturally more accepted since women typically find psycholog- ical compensation in their family role.27 According to Discussion studies on masculinity, this could also be associated with Protected by copyright. In this study, we analysed trends in socioeconomic the fact that men are less likely to express their feelings inequalities in suicide mortality in Barcelona. We found and seek support in social networks when they need that after the onset of the economic crisis, inequalities in help concerning their mental health.28 In contrast, the suicide rates by educational level tended to disappear in increased risk of suicide among women may be likely men, while suicide risk tended to increase among men due to alterations in social factors than to alterations in living in neighbourhoods with higher unemployment. material factors. This is similar to a study conducted in Among women, we found no inequalities by educational Denmark29 and in our study, in that we observed a higher level, but observed higher suicide risk in the late crisis suicide risk among women living in neighbourhoods with period among those living in neighbourhoods with a more elderly people living alone. higher proportion of elderly people living alone. One of the limitations of our study is its low statistical Several studies have reported an association between power due to the small number of suicides in Barcelona, suicide mortality and lower educational levels,12–14 but few especially among women, which made it impossible to have evaluated the late effect of economic crises. Studies detect significant associations and to stratify the analyses in the USA21 and Asian countries19 found that inequality by other variables such as age. It was also fairly difficult by educational level remained stable or even increased to obtain contextual data for the time periods studied during economic recession. This contrasts with our because Barcelona underwent a change in its territorial results in that we found that suicide mortality decreased divisions in 2009. Nonetheless, we were able to collect among men with a lower educational level and remained indicators for most determinants that are relevant for stable among men with a higher level of education. This social inequalities in suicide. suggests that the economic crisis in Barcelona has had a The main strength of our study is that this study offers greater effect on men with a socioeconomic advantage, a conceptual framework that presents the social determi- possibly because labour market restructuring created nants most commonly related to suicide. It also offers an additional work-related stress and a feeling of job insecu- analysis by gender, which is relevant taking into account rity. In such a scenario, men with greater responsibilities the difference in the risk factors for each gender. On could experience a more negative response and a decline the other hand, we used data from the Judicial Mortality in mental health.22 This notion is supported by the results Registry of Barcelona, which is not only more up-to-date of a Norwegian study23 reporting that suicide rates were than the Mortality Registry but also avoids under-re- higher among women with a higher level of education. porting suicide as a cause of death.30 Furthermore, this There is abundant evidence on the association between registry allowed us to analyse social inequalities because unemployment levels and suicide, especially during times it includes socioeconomic variables (educational level, of economic crisis.8 24 25 Consistent with this evidence, country of birth and neighbourhood of residence), we found that during the early (2009–2012) and late and was thus very useful for suicide surveillance during López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 7 8 Table 4 Association between suicide mortality among women and individual and contextual factors for each time period, Barcelona, 2006–2016 Model 1 Model 2 Model 3 Model 4 Model 5 RR RR RR RR RR Women P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3 P interaction Open access Level of education  Primary or lower 0.71 0.70 0.98 0.78 0.73 1.07 0.71 0.71 1.02 0.74 0.69 1.03 0.70 0.68 0.99  Secondary or higher 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.407 % unemployment  Q1 (least) 1.00 1.00 1.00 0.809  Q2 0.85 1.01 0.81  Q3 0.74 0.67 0.48*  Q4 (most) 0.66 1.06 0.88 % elderly living alone  Q1 (least) 1.00 1.00 1.00 0.374  Q2 1.61 1.02 1.36  Q3 1.46 1.19 1.85  Q4 (most) 1.14 1.04 2.13* % voter turnout  Q1 (least) 0.75 1.26 0.76 0.973  Q2 0.90 0.93 0.80  Q3 0.89 1.03 0.80  Q4 (most) 1.00 1.00 1.00 Drug use index  Q1 (least) 1.00 1.00 1.00 0.791  Q2 1.09 0.82 1.00  Q3 1.14 0.75 0.82  Q4 (most) 1.09 1.35 0.93 Deviance 1899.06 1886.99 1887.73 1896.30 1894.52 P value 0.210 0.254 0.973 0.873 P1, pre-crisis period (2006–2008); P2, early crisis period (2009–2012); P3, late crisis period (2013–2016). RR for each variable in each time period was obtained from the combined effect of the variable coefficients and the interaction between the variable and the period. P value=likelihood ratio test to compare models with and without contextual explanatory variables. P interaction=likelihood ratio test to compare models with and without the interaction of contextual explanatory variables and time period variable. All models were adjusted by age and country of birth. Model 1: level of education; model 2: level of education and % unemployment; model 3: level of education and % elderly living alone; model 4: level of education and % voter turnout; model 5: level of education and drug use index. *p<0.05. RR, relative risk. López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 Protected by copyright. BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. Open access BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. the economic crisis. Likewise, the records to obtain the Provenance and peer review Not commissioned; externally peer reviewed. contextual variables have a tradition of being of good Data availability statement No data are available. quality and provide information at a small area level, Open access This is an open access article distributed in accordance with the which is not always available for this type of study. Simi- Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which larly, while most previous studies only analysed variables permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is related to material deprivation in suicide mortality, we properly cited, appropriate credit is given, any changes made indicated, and the use evaluated both social and material contextual indicators. is non-commercial. See: http://​creativecommons.​org/​licenses/​by-​nc/​4.​0/. . In conclusion, we have shown that inequalities in suicide mortality by educational level among men tended to disappear during the crisis, although men living in neigh- References bourhoods with higher unemployment may be more 1. Okada K, Samreth S. A study on the socio-economic determinants of suicide: evidence from 13 European OECD countries. J Socio Econ likely to commit suicide. In contrast, we found no associa- 2013;45:78–85. tion between suicide and educational level in women, but 2. Borrell C, Malmusi D, Artazcoz L, et al. Propuesta de politicas e intervenciones para reducir las desigualdades sociales en salud en suicide might be more likely in women living in neigh- España: a proposal of policies and interventions to reduce social bourhoods with a higher number of elderly people living inequalities in health in Spain. Gac Sanit 2012;26:182–9. 3. Kentikelenis A, Karanikolos M, Reeves A, et al. Greece’s health crisis: alone. Future studies should analyse the socioeconomic from austerity to denialism. The Lancet 2014;383:748–53. inequalities of suicide and further explore the causes of 4. Kelly BD, Davoren M, Mhaoláin AN, et al. Social capital and suicide in 11 european countries: an ecological analysis. Soc Psychiatry gender differences in this phenomenon. Since suicide is Psychiatr Epidemiol 2009;44:971–7. the endpoint of a process that can be treated previously 5. Smith NDL, Kawachi I. State-level social capital and suicide mortality at the population level, it is also important to continue in the 50 U.S. states. Soc Sci Med 2014;120:269–77. 6. Whitley E, Gunnell D, Dorling D, et al. Ecological study of social monitoring its social determinants, especially during fragmentation, poverty, and suicide. BMJ 1999;319:1034–7. periods of economic crisis. This could be accomplished 7. De Silva MJ, McKenzie K, Harpham T, et al. Social capital and mental illness: a systematic review. J Epidemiol Community Health by establishing alliances with health centres that deal 2005;59:619–27. with cases of attempted suicide. In this way, we could gain 8. Milner A, McClure R, De Leo D. Socio-Economic determinants of suicide: an ecological analysis of 35 countries. Soc Psychiatry Protected by copyright. further insight into the phenomenon and develop better Psychiatr Epidemiol 2012;47:19–27. strategies to improve the living conditions of people and 9. Middleton N, Whitley E, Frankel S, et al. Suicide risk in small areas in prevent unnecessary deaths. England and Wales, 1991-1993. Soc Psychiatry Psychiatr Epidemiol 2004;39:45–52. 10. Turecki G, Brent DA. Suicide and suicidal behaviour. The Lancet Author affiliations 2016;387:1227–39. 1 Department of Experimental and Health Sciences, Pompeu Fabra University, 11. World Health Organization. Preventing suicide: a global imperative. CMAJ 2014;143:609–10. Barcelona, Spain 2 12. Kim M-H, Jung-Choi K, Jun H-J, et al. Socioeconomic inequalities Servei de Sistemes d'Informació Sanitària, Agència de Salut Pública de Barcelona, in suicidal ideation, parasuicides, and completed suicides in South Barcelona, Spain Korea. Soc Sci Med 2010;70:1254–61. 3 Centro de Investigación Biomédica en Red de Epidemiología y Salud Pública 13. Lorant V, Kunst AE, Huisman M, et al. Socio-Economic inequalities (CIBERESP), Madrid, Spain in suicide: a european comparative study. Br J Psychiatry 4 2005;187:49–54. Àrea de Recerca, Docència i Comunicació, Agència de Salut Pública de Barcelona, 14. Li Z, Page A, Martin G, et al. Attributable risk of psychiatric and Barcelona, Spain socio-economic factors for suicide from individual-level, population- 5 Institut d’Investigació Biomèdica (IIB) de Sant Pau, Barcelona, Spain based studies: a systematic review. Soc Sci Med 2011;72:608–16. 6 Gerència, Agència de Salut Pública de Barcelona, Barcelona, Spain 15. Burrows S, Auger N, Gamache P, et al. Influence of social and 7 Clinical Service, Catalonian Institute of Legal Medicine and Forensic Science, material individual and area deprivation on suicide mortality among 2.7 million Canadians: a prospective study. BMC Public Health Barcelona, Spain 2011;11:577. 16. Milner A, Spittal MJ, Pirkis J, et al. Suicide by occupation: systematic Acknowledgements To María Dolores Martos Hinojosa for her invaluable review and meta-analysis. Br J Psychiatry 2013;203:409–16. collaboration codifying all the judicial deaths of the Registry of Judicial Mortality 17. Parmar D, Stavropoulou C, Ioannidis JPA. Health outcomes during of Barcelona.To the people of the Institut de Medicina Legal i Ciències Forenses de the 2008 financial crisis in Europe: systematic literature review. BMJ 2016;354:i4588–4588. bmj BMJ. Catalunya and Civil Registry of Barcelona who have made this study possible. 18. Valkonen T. Changes in socioeconomic inequalities in mortality Contributors MRS and MG made substantial contributions to the conception and during an economic boom and recession among middle-aged men design of this study. NLC, MRS and MG performed data analysis. NLC, MRS, MG, and women in Finland. Eur J Public Health 2000;10:274–80. 19. Lee WY, Khang Y-H, Noh M, et al. Trends in educational differentials CB and AN contributed to the interpretation of data. NLC was involved in drafting in suicide mortality between 1993-2006 in Korea. Yonsei Med J the manuscript, and MRS, MG, CB, AN and JM revised it critically for important 2009;50:482–92. intellectual content. All authors gave final approval of the version to be published. 20. Borrell C, Marí-Dell’Olmo M, Gotsens M, et al. Socioeconomic Funding statement This research was partially supported by the CIBER inequalities in suicide mortality before and after the economic recession in Spain. BMC Public Health 2017;17:1–8. Epidemiología y Salud Pública. 21. Harper S, Charters TJ, Strumpf EC, et al. Economic downturns and Map disclaimer The depiction of boundaries on the map(s) in this article do not suicide mortality in the USA, 1980-2010: observational study. Int J imply the expression of any opinion whatsoever on the part of BMJ (or any member Epidemiol 2015;44:956–66. of its group) concerning the legal status of any country, territory, jurisdiction or 22. Chan CH, Caine ED, You S, et al. Suicide rates among working-age adults in South Korea before and after the 2008 economic crisis. J area or of its authorities. The map(s) are provided without any warranty of any kind, Epidemiol Community Health 2014;68:246–52. either express or implied. 23. Strand BH, Grøholt E-K, Steingrímsdóttir OA, et al. Educational Competing interests None declared. inequalities in mortality over four decades in Norway: prospective study of middle aged men and women followed for cause specific Patient consent for publication Not required. mortality, 1960-2000. BMJ 2010;340. López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267 9 Open access BMJ Open: first published as 10.1136/bmjopen-2018-028267 on 26 August 2019. Downloaded from http://bmjopen.bmj.com/ on December 12, 2023 at Univesity Autonomous of Barcelona. 24. Catalano R, Goldman-Mellor S, Saxton K, et al. The health effects of 28. Möller-Leimkühler AM. The gender gap in suicide and premature economic decline. Annu Rev Public Health 2011;32:431–50. death or: why are men so vulnerable? Eur Arch Psychiatry Clin 25. Baumbach A, Gulis G. Impact of financial crisis on selected health Neurosci 2003;253:1–8. outcomes in Europe. Eur J Public Health 2014;24:399–403. 29. Agerbo E, Sterne JAC, Gunnell DJ. Combining individual and 26. Bartoll X, Palència L, Malmusi D, et al. The evolution of mental ecological data to determine compositional and contextual socio- health in Spain during the economic crisis. Eur J Public Health 2014;24:415–8. economic risk factors for suicide. Soc Sci Med 2007;64:451–61. 27. Artazcoz L, Benach J, Borrell C, et al. Unemployment and mental 30. Gotsens M, Marí-Dell'Olmo M, Rodríguez-Sanz M, et al. Validación health: understanding the interactions among gender, family roles, de la causa básica de defunción en las muertes que requieren and social class. Am J Public Health 2004;94:82–8. intervención medicolegal. Rev Esp Salud Publica 2011;85:163–74. Protected by copyright. 10 López-Contreras N, et al. BMJ Open 2019;9:e028267. doi:10.1136/bmjopen-2018-028267