Abstract
Background:
Mental health practices consider unemployment as a suicide main risk factor, based on models built for populations of developed countries.
Aim:
This study assesses the association between the occupational risk factor and suicidal behaviors, by considering sex, age and temporal distribution, in a Latin American Andean population from north westernmost Argentina.
Methods:
Data include 481 suicide attempt cases assisted by emergency service psychologists at a head hospital in the Jujuy Province, northern Argentina, during two biennials. General categories and specific types of occupational situation, sex, age and temporal distribution were analyzed.
Results:
83.58% of cases corresponded to the with occupation category, but without occupation characterized male cases (p-value = .01). The type no referred occupation (19.15%) mainly contributed to this association (p-value = .02). Unemployed only represented 1.6%. These features revealed independent from the biennial period of assistance (p-value = .96 (general), p-value = .86 (specific)). Associations by age ranges did not seem to be specific of suicidal attempts.
Conclusion:
The present contribution provides an occupational characterization of suicide attempts in an Andean population from north westernmost Argentina. Unemployment is not associated with suicidal behavior in this population, suggesting that dissimilar patterns underlie suicidal behavior of populations from developed and developing regions.
Introduction
Many research works in the last decades have converged in considering unemployment as a factor underlying suicidal behavior (Dos Santos, Tavares, & Barros, 2016; Garcy & Vågerö, 2012; Iglesias García et al., 2017; Johansson & Sundquist, 1997; Kposowa, 2001; Mäki & Martikainen, 2012; Morrell, Taylor, Quine, & Kerr, 1993; Preti & Miotto, 1999; Rivera, Casal, & Currais, 2017; among others). Despite the large amount of reported evidences, the nature of such association remains a challenging issue (Platt, 1984, 1986). Some studies report the existence of a causal association between unemployment and suicide (Blakely, Collings, & Atkinson, 2003; Mäki & Martikainen, 2012), whereas there are arguments against its casual interpretation (Laanani, Ghos, Jougla, & Rey, 2015), and others authors have instead found an association between unemployment and mental disorders (Beautrais, Joyce, & Mulder, 1998; Mortensen, Agerbo, Erikson, Qin, & Westergaard Nielsen, 2000). The psychological distress that emerges from unemployment has been also reported as being a mediator of the association between unemployment and suicide (Li, Page, Martin, & Taylor, 2011). However, the role of mental health in the association between unemployment and suicidal behavior remains unclear (Lester & Yang, 2003).
The relationship between unemployment and suicide is revealed to be complex, and being also ruled by additional key factors, such as the economic cycle. However, specialists have dissimilar opinions concerning the influence of economic growths and recessions on the considered phenomena. For example, a significant positive association between unemployment and suicide has been found during periods of economic stability in Spain, which become − surprisingly − less consistent during the recent financial crisis (Iglesias García et al., 2017). Conversely, other studies have found a positive association between the Spanish economic crises of 2008, 2009 and 2012, and suicide rates (Ruiz Perez, Rodriguez Barranco, Rojas Garcia, & Mendoza Garcia, 2017).
All in all, and despite some discrepancies, there seems to be a general agreement among most researchers that suicidal behavior associates with unemployment. Studies supporting such results have been conducted in developed countries, where people are less exposed to unemployment and economic restrictions than inhabitants from developing regions. Suicidal behavior and unemployment have been poorly assessed by research works in developing regions, and in particular, in South American Andean populations. Most health care practices in developing parts of the globe arise on models that have been proposed by studies conducted in developed, foreign countries. These models are often not replicated or tested prior to their adoption and might reveal as having a variable adaptation to local particularities, especially in mental health practices. For example, negative associations have been reported between unemployment and poverty rates, and suicide mortality in Colombia, which seems to indicate that factors underlying suicide in such developing country might be indeed different from those regions having high incomes (Dávila Cervantes & Pardo Montaño, 2017).
It is widely accepted that suicide prevention strategies are required to be based on the typical features of suicidal behavior in the target population. This research focuses on inhabitants of the north westernmost Andean region of Argentina (Figure 1). The analyzed population corresponds to persons living in the province of Jujuy (ca. 673,307 inhabitants; Instituto Nacional de Estadística y Censos (INDEC), 2010), which has high rates of suicide mortality compared with the rest of the country. For example, during 2016 and 2017, suicide mortality rates (×100,000 inhabitants) in Jujuy were of 12.90 and 9.68, respectively, which were 44.19% and 24.48% higher than the national reported rates of 7.20 and 7.31, respectively (suicide mortality rates calculated based on Dirección de Estadísticas e Información en Salud (DEIS), n.d.; INDEC, 2019). Little is known about the typical characteristics of suicidal behavior in these populations. Therefore, mental health practices consider risk factors that have been defined for foreign populations. For instance, in developed countries, unemployment has been reported as showing differential suicidal risk according to sex (Coope et al., 2014; Garcy & Vågerö, 2012, 2013) and age (Coope et al., 2014; Pritchard, 1992). However, given that the prevalence and characteristics of suicidal behavior vary widely among different communities, demographic groups and over time (World Health Organization (WHO), 2014), it is possible that the extrapolation of foreign characterizations does not adequately represent the own features of the local populations. The aim of this study is to analyze whether occupational aspects might be considered as defining risk factors for suicide attempts in Andean populations of developing countries of South America, and particularly, in north westernmost Argentina, taking into account, in turn, the variations according to sex, age and time.

Location map of the studied site.
Materials and methods
Field of study and data sources
The field of study involves the Emergency Service of the head public (i.e. free of care charges) hospital (Pablo Soria Hospital, PSH) of Jujuy, northwestern Argentina (Figure 1). The PSH is a third level of complexity, polyvalent institution of health care that assists patients of both sexes older than 15 years. Until July 2012, the Emergency Service of the PSH has been offering the unique public mental assistance of emergency for inhabitants in that province. This means that all cases needing emergency assistance for mental health–related problems must be referred to this Emergency Service. Only since August 2012, an additional public hospital (San Roque Hospital) in Jujuy started to offer emergency assistance in mental health. Data were collected in the context of a larger study on suicidal emergencies, which covered two biennial periods chosen by convenience (i.e. data availability) sampling: 2007–2008 and 2011–2012 (López Steinmetz, 2015, 2017a, 2017b). Data sources correspond to written records of psychological assistance in the PSH Emergency Service (Guidelines of the written records of psychological assistance is available in Supplemental material). The considered data include the totality cases (n = 481) of psychological assistance corresponding to suicide attempt during the considered period. All these cases met the diagnostic criteria for intentional self-harm (codes X60–X84), according to the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10; WHO, 1995).
Variables or factors
Data analysis was retrospective and was conducted by considering two categories and five types of occupational situation (Figure 2).

Flowchart showing the categories and types of occupational situation considered in this study.
Occupational situation: general and specific
The general occupational situation groups cases into the following dichotomous categories: with occupation and without occupation. The category with occupation includes two types of specific occupational situation: student and working (the latter includes homekeeper). The category without occupation includes three types of specific occupational situation: no referred occupation, unemployed and retired. The type no referred occupation corresponds to egosyntonic cases of unemployment, and not studying. Egosyntonic refers to an attribute that the person considers acceptable or consistent with his or her self-image (Rosenthal, 2003). The type no referred occupation is different from the unemployed type, which corresponds to cases perceiving unemployment as a problematic situation causing uncomfortable feelings. Finally, the type named as retired included retired properly and pensioned.
Given that sex and age are usually described as main demographic risk factors for suicidal behaviors (Hawton, 2000; Polewka et al., 2004; Zhang, McKeown, Hussey, Thompson, & Woods, 2005), and specific demographic groups whose suicide rates were most affected by unemployment has been reported (Garcy & Vågerö, 2012, 2013; Coope et al., 2014; Pritchard, 1992), we test associations between sex, age, and occupational situation in the considered data set. The analysis of occupational situations considers these variables in the following categories:
Sex is analyzed in dichotomous terms (female and male).
Age is analyzed in 10-year segments (total range = from 15 to 74 years old).
In addition, the case distribution of suicide attempt in both sex group has been assessed according to occupational situation (general and specific) and the biennial periods of assistance (2007–2008 and 2011–2012). These analyses have been conducted in order to test whether the occupational features of the considered sample are variables changing through time (i.e. they are associated to one particular biennial period) or if, in the opposite, occupational features remain unchanged, which would be suggesting that occupational characteristics of the suicide attempt are not time-dependent.
Data analysis
Data analysis was performed by employing the R Studio version 3.1.1 (2014-07-10). Statistical assessment includes descriptive and association analyses. Descriptive analysis reports the mode, ratio and percentages. Associations have been assessed by employing the chi-square test (χ2). The Yates continuity correction was applied when necessary. The correction procedure (when applicable) for frequency distributions larger than 2 × 2 has consisted of grouping categories, types and/or adjacent intervals. The statistical significance level was set at p-value ⩽ .05.
Approval of bioethics
The protocol of this research has been approved by the Bioethics Committee of the province of Jujuy (dictum of 8 January 2012).
Results
General occupational situation
The distribution of cases according to the general occupational situation and sex is summarized in Table 1. Most cases of suicide attempt (83.58%) in the both sex groups corresponded to the category with occupation. However, the ratio between with occupation and without occupation was 6.71:1 in females and 3.59:1 in males. The ratio by sexes (female/male) was 1.73:1 for category with occupation and 0.93:1 for without occupation. Differences between categories of general occupational situation and sexes were statistically significant (χ2 = 5.89, df = 1, p-value = .01), but the strength of the association was only V = .12. The category with occupation characterized female cases, whereas male cases were characterized by without occupation.
Suicide attempt cases according to general occupational situation and sex.
Case distributions according to general occupational situation, and age, for the both sex groups and for each sex separately, are summarized in Table 2. The mode by age was 19.50 years in both with and without occupation. In the four lower age ranges (i.e. from 15 to 54 years), there were more cases with occupation than without occupation, whereas in the two higher age ranges (i.e. from 55 to 74 years), the opposite was observed. Differences between the categories of general occupational situation and age ranges is revealed to be statistically significant in the both sex groups (χ2 = 44.37, df = 4, p-value < .001), in females (χ2 = 30.35, df = 4, p-value < .001) and in males (χ2 = 16.62, df = 4, p-value = .002) when analyzed separately (grouped intervals for the chi-square test in the both sex groups and in each sex: 55–74 years). The strength of the associations was V = .30 (the both sex groups), V = .32 (females) and V = .30 (males). The main differences between the both sex groups and in each sex showed that the category with occupation was characterized by cases of 15–24 and 45–54 years old, whereas without occupation was characterized by cases of 55–64 and 65–74 years old.
Suicide attempt cases according to general occupational situation, sex and age range.
Grouped intervals for computing the chi-square test in both sex groups and in each sex: 55–74 years old.
Specific occupational situation
The distribution of cases according to specific occupational situation and sex is summarized in Table 3. The mode corresponded to the student type in females and to the working type in males. The differences between types of specific occupational situation and sex were statistically significant (χ2 = 10.11, df = 3, p-value = .02) (grouped types for the chi-square test: no referred occupation-retired). The strength of the association was V = .14. The main differences revealed that females were characterized by the student type, whereas males were characterized by the no referred occupation type.
Suicide attempt cases according to specific occupational situation and sex.
Grouped types for computing the chi-square test in both sex groups and in each sex: no referred occupation-retired.
The distribution of cases according to specific occupational situation, and age, for the both sex group is summarized in Table 4. The differences between types of specific occupational situation and age ranges were statistically significant (χ2 = 48.12, df = 4, p-value < .001) (grouped types and intervals for the chi-square test: student-working; no referred occupation-retired; 15–34, 35–54, 55–74 years old). The strength of the association was V = .56. The main differences found revealed that the student type was characterized by cases in the range 15–24 years, the working type was characterized by cases of 25–34, 35–44 and 45–54 years old, the no referred occupation type was characterized by cases of 55–64 years old and the retired type was consistent with cases of 65–74 years old.
Suicide attempt cases (both sex groups) according to specific occupational situation and age range.
Grouped intervals and types for computing the chi-square test: 15–34, 35–54, 55–74 years old; student-working; no referred occupation-retired.
Occupational situation and biennial period of assistance
No statistical association has been found between the general occupational situation and the biennium of assistance (χ2 = 0.002, df = 1, p-value = .96), nor for the specific occupational situation and the biennium of assistance (χ2 = 0.75, df = 3, p-value = .86) (grouped types for the chi-square test in specific occupational situation: no referred occupation-retired).
Discussion
The majority of cases (83.58%) of suicide attempt correspond to the with occupation category. The differences by sex reveal that the with occupation type characterizes female cases, whereas without occupation type characterizes male cases. Female involves 63.43% of the total with occupation cases (n = 402), which is consistent with the prevalence of the with occupation category (over the without occupation characterizing the male group) as a feature describing the majoritarian situation of suicide attempting persons.
In male cases, the no referred occupation type represents 87.80% of the entire without occupation category in this sex (n = 41) (Table 3). According to this, there would be a low influence (1.60%) of unemployment on the development of suicide attempts in males. However, it is important to note that this pattern concerning egosyntonic male cases under a without occupation situation reveals an opposite tendency relative to previous studies (Blakely et al., 2003; Dos Santos et al., 2016; Garcy & Vågerö, 2012; Iglesias García et al., 2017; Johansson & Sundquist, 1997; Kposowa, 2001; Mäki & Martikainen, 2012; Morrell et al., 1993; Platt, 1984, 1986; Preti & Miotto, 1999; Rivera et al., 2017). The meaning of such a representativeness of the no referred occupation type (19.15%) in male cases would be alternatively understood in terms of shifting patterns in psychological perceptions under unemployment. Some previous studies suggested that the association between suicidal behavior and unemployment would be influenced by the duration of unemployment (Garcy & Vågerö, 2012; Mäki & Martikainen, 2012; Milner, Page, & LaMontagne, 2013). The intensity of job search, psychological stress and work centrality would fluctuate through time, reaching its highest intensity during the 2 or 3 months next to the onset of unemployment, and would then gradually decrease (Kulik, 2001). Consequently, the male no referred occupation cases reported in this study would have corresponded originally to the unemployed type, shifting to the egosyntonic type under a long-term exposure to unemployment. The analysis of the duration of unemployment in further studies may be important in another sense. As previously reported, the contribution attributable to mental disorders in the development of suicide attempts in this Andean region would be strikingly low (López Steinmetz, 2017a). Nevertheless, symptoms of mental illness due to unemployment have been reported as being stable during the second year of exposure to unemployment, but being followed by an increase in distress under persistent unemployment (Paul & Moser, 2009). Therefore, the role of mental disorders in the postulated association between unemployment and suicidal behavior must be evaluated, since it might also be a time-dependent attribute in the target population. Additional studies are required in order to setting reliably these patterns.
Results emerged from our study show that in female cases, the association with the with occupation category mainly corresponds to the student type (44.37%). It is important to note that this pattern is consistent with the prevalent age of suicide attempts (15–24 years; López Steinmetz, 2017b). Similar to the male pattern, unemployment only represents 1.71% of suicide attempts in females. Consequently, these features would be revealing that (a) there is no clear association between unemployment and suicidal behavior in the considered Andean population, and (b) there would be a dissimilar pattern underlying suicidal behaviors of populations of developed and developing countries.
Some studies that have been conducted in developed countries have found no association between unemployment and suicidal behavior (Furness, Khan, & Pickens, 1985; Nielsen, Wang, & Brille Brahe, 1990; Platt, Micciolo, & Tansella, 1992; Saurina, Bragulat, Saez, & López Casanovas, 2013), and a negative association has been even reported in Colombia (Dávila Cervantes & Pardo Montaño, 2017). Unfortunately, the lack of historical records and previous studies characterizing local features difficults deeper assessments of temporal shiftings on the link between unemployment and suicide attempts. Our temporal analysis of data indicates that suicide attempts would be independent from time, as scores are consistent in the two considered biennial periods.
Aspects such as the quality of the employment and its stability/instability are determining features of psychological discomfort (Sverke, Hellgren, & Näswall, 2002 and references therein) and could reveal as relevant factors in the development of suicidal behavior according to the high prevalence of cases of the with occupation category sourced from our study. However, recent preliminary data (year 2019) show no association between job instability and psychological discomfort in inhabitants from Córdoba, a major city (>1,500,000 inhabitants) in the center of Argentina (López Steinmetz, Hill, Rivas Ruzo, & Caracciolo, unpublished report). However, it is important to notice the contrasting cultural characteristic of all these populations; Córdoba is a multicultural city, whereas the population in Jujuy is largely composed of Andean people (i.e. persons descendent of who were once part of the Tawantinsuyu, the Inca Empire in the Central Andes; see García Moritán & Cruz, 2011). Despite that these aspects lead to suppose that the relevance of job instability in mental health-related problems, as it is suicide, may be not so clear, a recent report of the Organization for Economic Cooperation and Development (OECD) states out that it is instability in employment, rather than unemployment itself, the factor that would be associated to high rates of suicide in the youngest population (Kim & Cho, 2017). It seems necessary to replicate this study on suicide attempts because Argentina currently undergoing a major economic crisis that is deeply impacting on employment (i.e. rising unemployment, job instability, underemployment and so on).
The associations between the occupational situation (general and specific) and the age of the patients did not seem to reveal specific features of suicide attempts, but, on the contrary, these features are more likely defining elements of the general population (e.g. the associations: 15–24 years old and the student type; 65–74 years old and the retired type; and among others). Additional local studies would be necessary to test if the greatest suicidal risk is actually associated with certain types of occupation or profession. In this sense, studies have reported that high suicide rates particularly affect certain occupations or professions (Boxer, Burnett, & Swanson, 1995; Milner, Spittal, Pirkis, & LaMontagne, 2013) and have suggested that there would be a stepwise gradient in risk, with the lowest skilled occupations being at a greater risk of suicide than the highest skill-level group. However, the highest rates of suicide in certain occupational groups could be the result of a complex interaction between factors of the work environment, such as work stress and access to suicidal methods, and other risk factors, such as age and the presence of mental disorders (Boxer et al., 1995). However, more recent findings suggest that, with the exception of physicians and nurses, most of the suicide over-risk associated with particular occupations would be understood in terms of the social and economic characteristics of the people carrying these occupations (Agerbo, Gunnell, Bonde, Mortensen, & Nordentoft, 2007).
The reported findings provide the first assessment of the employment/unemployment relative to suicide attempts in an Andean, Latin American region. Results enhance the need of avoiding practices based on the direct extrapolation of models built on social context from developed countries, and the necessity of revisiting mental health practices in the evaluation and assistance of suicide attempts in developing countries. Our findings from north westernmost Argentina might be applicable in other Andean Latin American countries because the Andean idiosyncratic and cultural features are closer to those of Latin American developing countries, rather than to those of developed regions. Despite that chief urban centers of Argentina comprise multiethnic and multicultural populations, the Jujuy province is mainly composed of Andean people and is characterized by negative net migration rates (Ministerio del Interior, Obras Públicas y Vivienda, 2019). According to our results, aspects ruling suicidal behavior in developed regions would have a minor relevance for inhabitants of developing countries. In developed countries, getting an employment would be likely perceived as a normal situation and thus unemployment majorly would implicate a personal failure that might conduce, in certain cases, to attempting suicide. On the contrary, high rates of unemployment that are recurrent or last for decades in developing regions would influence the collective and individual perception of local inhabitants. In this context, there would be a rupture between suicide attempts and not getting a job. However, the job quality and the economic perspectives reveal instead as likely ruling factors of suicide attempt that might supplant unemployment. Finally, it is important to insist on the fact that further research works are needed to complement these findings and in-depth analysis.
Conclusion
Unemployment is not associated with suicidal behavior in Andean populations from the Jujuy province in north westernmost Argentina. These findings suggest dissimilar patterns underlying suicidal behavior of populations from developed and developing regions. The quality of employment and the economic perspectives should be further assessed since, rather than unemployment, these could be relevant factors in the development of the local suicidal behaviors.
Supplemental Material
Lopez_Steinmetz_et_al_Suppl_Material_IJSP_Rev1 – Supplemental material for Is unemployment less important than expected for suicide attempted in developing regions? Occupational profile of suicide attempts in Jujuy, north westernmost Argentina
Supplemental material, Lopez_Steinmetz_et_al_Suppl_Material_IJSP_Rev1 for Is unemployment less important than expected for suicide attempted in developing regions? Occupational profile of suicide attempts in Jujuy, north westernmost Argentina by Lorena Cecilia López Steinmetz, Romina Lucrecia López Steinmetz and Juan Carlos Godoy in International Journal of Social Psychiatry
Footnotes
Conflict of interest
The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This research was carried out with a doctoral research grant awarded by the National Council of Scientific and Technical Research.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
