Downloaded from [Link] on October 26, 2017 - Published by [Link].

com

694 J Epidemiol Community Health 1999;53:694–701

Suicide and unemployment in Italy, 1982–1994

Antonio Preti, Paola Miotto

Abstract leading to suicidal ideation.2–4 In suicide, men-

Objective—To investigate whether either tal disorders indeed seem to be the determi-
the condition of being unemployed, or nant element.5–7 Negative life events are
changes in unemployment rates are asso- thought to act as favouring or precipitating fac-
ciated with suicide risk. tors: adverse financial circumstances and social
Design—Administrative data for suicide isolation have been found to be significant pre-
according to occupational status have dictors of suicide risk, and evidence both
been analysed considering three employ- among younger and older groups indicates that
ment categories: employed, seeking new stressful experiences, such as the loss of a loved
job (unemployed), seeking first job (never one or interpersonal conflict, can often precede
employed). Comparison of suicide rates suicide attempts.7–9
by economic position and correlation Special attention is often paid to the stress
between suicide and unemployment rates caused by working roles, especially when access
have been made. to such a role is rendered problematic by eco-
Subjects and settings—20 457 deaths by nomic crisis.10–12 Unemployment, in particular,
suicide registered in Italy among eco- is thought to be an important risk factor for
nomically active people from 1982 to 1994. suicide.13 14 Individual level cross sectional
Main outcome measures—Change over studies have generally found that people who
time in suicide rates by economic posi- committed suicide were more likely to be
tion; coeYcient of aggravation according jobless when they died than were people who
to occupational status. died from other causes.13 14 Conversely, aggre-
Results—Suicide rates among the unem- gate level studies have not consistently shown
ployed are clearly and constantly higher an increased incidence of suicide in areas of
than those among the employed: up to higher unemployment.13 14 Time series analyses
three times higher among men, and twice have found strong aggregate level correlations
as high among women. Among the unem- between unemployment and suicide, particu-
ployed a clear and significant rise in larly among young adults.13 15 16
suicide rates in both sexes took place over Many studies indicate a role for unemploy-
the study period; suicide rates among the ment as a precipitating, rather than causative,
employed showed a less marked increase. factor in suicidal crisis. Its contribution to the
The rise in suicide rates was accompanied precipitation of suicide is thought to work on
by a concurrent rise in unemployment two levels: both by reducing financial
rate percentage. Men seem to be aVected availability—which makes relational networks
most by this change in unemployment rate less supportive—and by provoking the loss of
percentage; women are subject to less evi- the social role and self esteem that a job
dent influences and variations. confers.11 12 15 17 18 The link between unemploy-
Conclusion—DiVerent suicidal behaviour ment and suicide risk, however, is far from
trends among unemployed compared with clearly defined. Three pathways from unem-
employed people indicate that unemploy- ployment to suicide have been suggested.19 20
ment (and above all the prospect of not The “vulnerability model” indicates that un-
having access to a working role) acts as a employment, by limiting access to supportive
contributing factor for suicide. Unem- resources, may increase the impact of stressful
ployment, even if symptomatic of a mental life events, thus increasing the risk of suicide.
disorder, should therefore always be taken The “indirect causative model” suggests that
into consideration as a risk factor for sui- unemployment, through the relationship diY-
cide: the potentially lethal consequences culties or financial problems it can provoke,
of its negative influence on both self may favour the occurrence of events that in
esteem and the ability to use supportive turn precipitate suicide. A third model pro-
CMG, Psychiatry
Branch, via networks in a eYcient way is an element to poses a non-causal link between unemploy-
Costantinopoli 42, which great attention should be paid. ment and suicide, both of which could be taken
09129 Cagliari, Italy (J Epidemiol Community Health 1999;53:694–701) as resulting from a third factor, such as belong-
A Preti ing to a disadvantaged social class, which in
itself would increase the risk of suicide and
Alcohol and Drug In Italy and in other Western countries there unemployment.21
Dependence Unit,
Conegliano, Italy
has been a progressive rise in suicide rates for All these models share a process of “social
P Miotto both sexes over the past 20 years, especially causation”, which supposes that unemploy-
among young people.1–3 The main factors that ment, whether directly or indirectly, is damag-
Correspondence to: have been suggested to explain this phenom- ing to well being and health. A diVerent set of
Dr A Preti.
enon are the increase in life stress (particularly explanations, grouped under the “health selec-
Accepted for publication for younger people), and the increase in the tion” hypothesis, asserts that poorer health
9 March 1999 incidence of medical and psychiatric illnesses itself, including poorer mental health, increases
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

Suicide and unemployment 695

the risk of unemployment: thus, having a disor- The source of the data on occupational
der that implies a higher risk of suicide would status is work statistics compiled by ISTAT:
also lead to unemployment.21 22 According to data derive from a trimestral cross sectional
this interpretation of existing data, more people survey and are collected at municipal level.
at risk of suicide are unemployed because of Yearly National data represent the average of
their illness, so higher suicide rates among the the four trimesters of the year: appropriate
unemployed would be an artefact of selection integrations are carried out to correct the
bias. In fact, a higher mortality from causes changing composition of the population by
other than suicide is also seen among the gender and age. The economically active
unemployed: this excess mortality for all causes population comprises all those people aged 15
of death among the unemployed is often and over who, when asked for the survey,
considered proof of the “health selection” claimed to be employed or to be seeking a job.
hypothesis.23 24 All people who do not satisfy this definition
Some authors suggest that if selection were a were considered not to be part of the economi-
major factor explaining above average mor- cally active population. In particular, this
tality among the unemployed, then this excess second “inactive” group comprises those aged
whould be smaller when the general unemploy- under 15; those who, though aged 15 or above,
ment rate is high than when it is low.25 At a time answered the survey stating that they had no
of high unemployment, becoming and remain- job and were not seeking one; and those who
ing jobless can be assumed to be less reported to be one of the following: housewife;
dependent on individual characteristics that student; retired from work (for whatever
might also increase the risk of ill health, and reason); disabled; under military service. Eco-
more fit and healthy people can be assumed to nomically active people have been divided
join the ranks of the unemployed than in peri- between those who were employed and those
ods of low unemployment. The same reasoning who were unemployed. This second group is
could also be applied to suicide.16 further divided between people looking for a
Over the past 20 years Italy, more than other new job (those properly titled “unemployed”)
European countries, has suVered a diYcult and people seeking their first job (never
economic period with a progressive rise in employed). Unemployment rate percentage is
unemployment rates: Italy thus represents an the proportion of unemployed to economically
obvious location for the study of the relation active people (both employed and unem-
between unemployment and suicide. The ployed).
“health selection” hypothesis would lead you to The coeYcient of aggravation (COA) for
expect a weaker association between unem- suicide, calculated according to Stack,26 has
ployment and suicide in recent years: the been reported for each economic position to
reverse would be true if unemployment directly further aid interpretation of the data. The COA
or indirectly increased suicide risk by adding to is the ratio between the suicide rates in one
pre-existing health conditions. social group and in the general population, and
allows a measure of the relative risk in the
group under study.
Method Changes in rates over time were analysed
To evaluate the impact of occupational status using a test for trend. Linear regression analy-
on patterns of suicidal behaviour in Italy, the sis of suicide rates for economic position
yearly distribution of suicides by occupational (employed, unemployed, seeking their first job)
status between 1982 and 1994 has therefore over 13 consecutive years was performed. A
been analysed. As mortality is low among those two tailed t test was performed to test the
of working age, data must relate to a sizeable hypothesis of a significant slope. As an alterna-
population if appreciable diVerences are to be tive to the simple linear regression model, a
noted: therefore oYcial national statistics have quadratic model was examined, whereby a sec-
been used in this study, taking into account the ond explanatory variable is added, which is
limitations that derive from using this type of simply the square of the year. This model
data, exposed as they are to risks of under- assumes the possibility of a yearly change of
reporting and incompleteness. rate and assumes that suicide rates by eco-
The source of data for suicides was the Ital- nomic position follow a single smooth para-
ian National Institute for Statistical Analysis bolic trend instead of strictly linear trends.
(ISTAT). Classification of suicides was com- Ordinary least squares (OLS) regression has
pleted according to the Ninth revision of the been used to evaluate the influence of change
ICD. Data were collected from the police and over time of unemployment rates (independent
the carabinieri (military police), who compile variable) on suicide rates by economic position
detailed case reports based both on preliminary (dependent variable). OLS regression has also
death certificates completed by the examining been used to test the ecological (geographical)
doctor, and on extensive inquiries among key association between unemployment rate (inde-
informants and relevant witnesses. In this pendent variable) and suicide rates (dependent
study, data from 1982 to 1994 have been com- variable).
pared by taking into account people with a job,
regardless of the type of job, and people Results
without a job. Rates per 100 000 by economic From 1982 to 1994 a total of 20 457 deaths by
position have been calculated on the basis of suicide were registered among economically
the distribution of the Italian population active people in Italy. Of these persons, 13 276
according to occupational status. were employed men, 3401 unemployed men,
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

696 Preti, Miotto

Table 1 Suicide rates (per 100 000) in Italy by gender and economic position Table 2 Change over time of suicide rates (per 100 000)
(economically active only), 1982–1994 among the economically active (suicide rates over years)

In search of first Ratio unemployed/ Simple linear regression model

Employed Unemployed job employed
Slope (SE) t p
Men
1982 6.7 17.8 6.2 2.6 Employed men 0.19 (0.03) 5.24 <0.001
1983 5.6 15.9 4.8 2.8 Unemployed men
1984 6.5 19.2 6.9 2.9 seeking a new job 1.62 (0.26) 6.18 <0.001
1985 7.1 25.9 7.8 3.6 seeking their first job 2.40 (0.28) 8.56 <0.001
1986 7.2 25.6 11.5 3.5 Employed women 0.04 (0.00) 5.01 <0.001
1987 7.8 29.9 15.3 3.8 Unemployed women
1988 7.0 22.7 12.0 3.2 seeking a new job 0.26 (0.08) 3.29 <0.01
1989 6.8 24.0 14.0 3.5 seeking their first job 0.51 (0.07) 7.27 <0.001
1990 7.2 33.6 19.4 4.6 Quadratic regression model
1991 8.1 36.5 21.7 4.4 Unemployed men
1992 8.6 35.2 21.0 4.1 seeking their first job
1993 8.5 35.9 35.1 4.2 year = −32.58 (11.03) 2.95 <0.01
1994 8.4 32.8 35.1 3.9 year 2 = 0.19 (0.06) 3.17 <0.01
Women Unemployed women
1982 1.9 1.7 0.9 0.9 seeking their first job
1983 1.6 2.9 0.5 1.7 year = −8.14 2.85 <0.01
1984 1.8 3.2 1.9 1.7 year 2 = 0.04 3.03 <0.01
1985 1.8 2.3 1.3 1.2
1986 2.1 3.2 1.1 1.5
1987 1.9 4.5 2.3 2.3
1988 2.0 3.3 2.0 1.6
those for the employed always has a value
1989 1.8 2.6 1.5 1.4 higher than one both for men and (to a lesser
1990 2.1 6.1 4.4 2.8 extent) for women. Women show lower suicide
1991 2.2 5.7 5.3 2.5
1992 2.2 6.3 4.7 2.7 rates than men irrespective of economic
1993 2.3 4.0 6.3 1.6 position. Suicide rates among unemployed
1994 2.2 4.5 7.0 2.0 women, however, show the same trends as for
unemployed men, being higher than among
1282 men in search of their first job; 1863 were employed women. The COA is three times
employed women, 483 unemployed women, higher for unemployed men (COA 1982 =
242 women in search of their first job. 2.33; 1994 = 3.08) than for employed men
Table 1 shows the distribution by gender (COA 1982 = 0.87; 1994 = 0.78), and this dif-
over the study period of the related suicide ference tends to become more marked with the
rates per 100 000. Suicide rates among the passing of time. As far as women are con-
unemployed are clearly and constantly higher cerned, the COA for the unemployed tends to
than those among the employed: the ratio worsen with the passing of time (1982 = 0.60;
between suicide rates for the unemployed and 1994 = 1.37), with an increment of 128 %,

9.0 13
Male unemployment percentage

8.5
12
8.0
Male suicide rates

11
7.5

7.0 10

6.5 9
6.0
8
5.5
7
5.0

4.5 6
1980 1982 1984 1986 1988 1990 1992 1994 1996 1980 1982 1984 1986 1988 1990 1992 1994 1996

16 3.2
Female unemployment percentage

15 3.0
Female suicide rates

14 2.8

13 2.6

12 2.4

11 2.2

10 2.0

9 1.8
1980 1982 1984 1986 1988 1990 1992 1994 1996 1980 1982 1984 1986 1988 1990 1992 1994 1996
Figure 1 Trends in suicide rate/100 000 among the economically active and unemployment rate percentage, 1982–1994
(n = 13 years).
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

Suicide and unemployment 697

Employed Employed
40 8
Unemployed Unemployed
35 Never employed 7 Never employed

30 6

25 5

Women
Men
20 4

15 3

10 2

5 1

0 0
80 82 84 86 88 90 92 94 96 80 82 84 86 88 90 92 94 96
Figure 2 Trends in suicide rates/100 000 in Italy (1982–1994) by gender and economic position: employed, unemployed
and seeking their first job (never employed).
whereas among employed women the change in figure 1: women are subject to less evident
of COA over time is considerably less (1982 = influences and variations. Over time, each per-
0.65; 1994 = 0.68). centage increase in the unemployment rate was
Over time, suicide rates increase for both associated with an increase in total male suicide
men and women (fig 1). The increase is higher mortality among the economically active equal
among the unemployed than among the to 1.39 deaths per 100 000 (95% confidence
employed in both genders (table 2). Each year, intervals 0.67 to 2.11, SE = 0.32, t = 4.26, p =
total suicide mortality increases by 1.62 deaths 0.001). The link between unemployment rate
per 100 000 (95% confidence intervals 1.04 to and suicide rates is greater among unemployed
2.20) among male unemployed and by 2.40 men (OLS, people seeking a new job: â = 4.28,
deaths per 100 000 (95% confidence intervals 95% confidence intervals 0.43 to 8.13, SE =
1.78 to 3.02) among men in search of their first 1.75, t = 2.44, p = 0.03; people seeking their
job. The increase is lower among unemployed first job: â = 8.65, 95% confidence intervals
women, but still higher than among employed 5.22 to 12.08, SE = 1.55, t = 5.55, p = 0.0002)
women (table 2). When related to suicides in than among employed men (OLS: â = 0.62,
both genders among those seeking their first 95% confidence intervals 0.20 to 1.04, SE =
job the results of the quadratic model have a 0.19, t = 3.29, p = 0.007). The association
significantly better fit than does the simple lin- between female unemployment rate and female
ear regression model: for men, adjusted r2 = suicide rates among economically active
0.92, p < 0.0001 in the quadratic model, women is less significant (OLS: â = 0.10, 95%
against adjusted r2 = 0.85, p < 0.001 in the confidence intervals 0.06 to 0.27, SE = 0.09, t
simple linear regression model; for women, = 1.39, p = NS). Suicide rates among women
adjusted r2 = 0.89, p < 0.0001 in the quadratic show a greater link to male than female unem-
model, compared with adjusted r2 = 0.81, p < ployment rate percentage. This is especially
0.001 in the simple linear regression model evident among women in employment (OLS: â
(table 2, fig 2). = 0.12, 95% confidence intervals 0.01 to 0.24,
All age groups are aVected by the rise in sui- SE = 0.05, t = 2.47, p = 0.03) or in search of a
cide rates (table 3). This rise is associated with first job (OLS: â = 1.64, 95% confidence inter-
a concurrent rise in unemployment rate vals 0.65 to 2.63, SE = 0.45, t = 3.64, p =
percentage. During the study period unem- 0.003).
ployment rate percentage rose from 4.7% of If the longitudinal (time series) analyses
economically active people in 1982 to 8.7% in suggests a link between risk of suicide and
1994 among men, and from 9.9% among unemployment, the ecological (geographical)
women in 1982 to 15.6% in 1994. Men seem evaluation introduces an element of uncer-
to be aVected most by this change in tainty. By considering suicide rate distribution
unemployment rate percentage, as can be seen across the 20 Italian regions along with the
Table 3 Absolute number of suicides and age specific suicide rates per 100 000 in Italy, respective unemployment rate percentages, a
1982–1993 slight negative relation between the two factors
(both for men and women) can be observed (p
18–24 25–44 45–65 65+
<0.05). This negative relation is also observed
Year Number Rate Number Rate Number Rate Number Rate when a comparison is made between the
1982–1994 changes in regional suicide and the
1982 202 1.45 701 3.54 987 5.01 1010 11.34
1983 166 1.19 665 3.35 935 4.74 1041 11.69
correspondent unemployment rates. However,
1984 182 1.30 786 3.96 1102 5.59 1057 11.87 when unemployment rates are regressed on
1985 194 1.39 811 4.09 1325 6.72 1310 14.71 suicide rates after adjustment for the regional
1986 211 1.51 856 4.32 1301 6.60 1335 15.00
1987 235 1.69 985 4.97 1378 6.99 1438 16.15 Gross Domestic Product per capita or the
1988 248 1.78 903 4.56 1182 6.00 1427 16.03 regional consumer expenditures per person
1989 229 1.64 872 4.40 1164 5.90 1404 15.77 (two indicators of an area’s wealth, the source
1990 279 2.00 1048 5.29 1183 6.00 1282 14.40
1991 281 2.02 1116 5.63 1286 6.52 1336 15.01 of which is the ISTAT Yearbook) the associ-
1992 284 2.04 1100 5.55 1264 6.41 1333 14.97 ation is no longer significant, though the
1993 307 2.20 1102 5.56 1225 6.21 1420 15.95
1994 282 2.02 1093 5.51 1138 5.76 1343 15.08
percentage of variance accounted for does not
change considerably.
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

698 Preti, Miotto

Discussion
KEY POINTS
The period 1982 to 1994 was characterised in
Italy by an economic crisis with growing x Despite limitation, oYcial statistic are still
unemployment rate percentage. There was also a reasonable approximation to study
an increase in suicide rates for both sexes. This suicide trends among those of working
increase was higher among the unemployed age.
than among the employed; a particularly x Periods of economic crisis with growing
significant increase was seen among those unemployment associate to a rise in
looking for a first job, with a greater increase in suicide rates, higher among unemployed
the most recent years, when unemployment than among employed.
was higher. Becoming and remaining jobless x Even if this study could not adjust for
during a period of high unemployment (such as potential confounders, the results are
the latter years of this study period) can be consistent with the view that unemploy-
assumed to be less dependent than at other ment per se increases the risk of suicide.
times on the person’s characteristics, among
which are characteristics that might also confer
a higher risk of ill health. It is noteworthy, then, authorities may be more prone to mask as an
that the unemployed in Italy have been subject accident the suicide of an employed person
to higher suicide rates in recent years than at than that of an unemployed person. This could
the beginning of the study period, when unem- help to explain the uneven distribution of
ployment was lower. suicide rates by economic position and the
The economic position of “unemployed” is changes in administrative suicide rates among
linked to a higher risk of suicide than that employed compared with unemployed people.
observed among the employed: the risk of sui- Thus the increase in administrative suicide
cide is three times higher among unemployed rates reported herein could be determined by a
than employed men, and about two times better rating of data, the tendency to hide sui-
higher among unemployed than employed cide exerting a diVerent influence in employed
women. People searching for a first job are at and unemployed cases.
higher risk: the probable age diVerence be- Another limitation is the use of data from
tween groups (people seeking a first job cross sectional surveys to construct rates
conceivably being younger on average than among social groups according to their occupa-
employed people) does not seem to act as a tional status. Respondents, for example, may
misleading factor, as suicide rates for the same be unwilling to reveal their unemployed status,
period of time in Italy increase with the age of so inflating the numbers of the employed group
the person. However, because data are rates per and diluting suicide rates among employed
100 000 rather than standardised mortality people. Seasonal fluctuations in employment,
ratio, these figures are vulnerable to changes in particularly in areas with a predominantly agri-
the age composition of the sample. Suicide cultural economy, may not be adjusted by
rates, indeed, are highly sensitive to age: this national statistics deriving from a yearly mean,
means that, if the “never employed” group yielding to incorrect measurement. On the
should “get older” during the period under other hand, the attribution of occupational sta-
study, a substantial rise in its suicide rate could tus to a suicide victim may be based on inaccu-
be expected for this reason alone. It would fur- rate rating: it could be that on occasion suicide
thermore not be surprising if the “never victims are registered in oYcial statistics on the
employed” group had increased in age given basis of most recent employment, even when
Italy’s economic circumstances since the early this employment ceased some time before-
1980s. hand. So it is possible that unemployed status is
attributed to suicide victims less often than is
CONFOUNDING FACTORS AND LIMITATIONS strictly accurate. Nevertheless, as the study of a
The lack of data on the age composition of the low base rate phenomenon like suicide requires
samples is not the only limitation of this study. a large population to detect significant diVer-
Working with oYcial suicide statistics implies a ences and specifically collected data are
risk of bias both in terms of data collection and diYcult to obtain, the use of oYcial statistics is
in the reconstruction of rates among social a reasonable compromise in the study of social
groups. Under-reporting of suicides in oYcial influences on suicide.
statistics is a worldwide problem.27 It is known The temporal relation between the rise in
that there are geographical diVerences in the unemployment rate percentage and suicide
consideration given to a self destructive act rate is, indeed, very suggestive of a causal link,
because of social and cultural factors.28 How- although the influence of concurrent factors
ever, careful studies performed in other coun- cannot be excluded: among these is the eVect
tries, as well as in Italy, indicate that underesti- of economic crisis on social integration.14 17 All
mation can only mask the dimension of the three methods used in this study, however, have
phenomenon and the weight of the influence well known limitations: comparison of suicide
exerted by social and environmental factors, rates across social groups at individual level
but not hide the importance of these exposes data to bias because of under-
influences.29 30 Whatever the reliability of oY- reporting and incompleteness. Time series
cial suicide statistics, the problem remains of analyses do not allow researchers to disentan-
the possible existence of biases in the rating of gle eVects attributable to social change as a
data according to the economic position of the whole from eVects attributable to changing
victim. For various reasons, relatives and the rates of unemployment. Aggregate level cross
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

Suicide and unemployment 699

sectional analyses are exposed to risk from the increased social stress and diYculties on a rela-
so called “ecological fallacy”. tional level that often culminate in divorce.18
Aside from the severing of familial ties,
THE “ECOLOGICAL FALLACY” however, a husband’s “undesiderable job
The term “ecological fallacy” refers to the events” are likely to associate to emotional
problems that arise from making inferences symptoms for his wife and relatives: in general
about people from studies of groups. William stressful events aVect the mental health not
Robinson first demonstrated in 1950 that the only of the people who directly experience the
correlations between two individual level vari- events but also of their close associates who are
ables frequently diVer markedly from the indirectly aVected by the events.41 42
correlations between the same two variables at Mental suVering of a depressive type is the
the aggregate level: sometime the correlations most often reported antecedent of suicide.43–45
may even be in the opposite direction.31 Studies Adverse circumstances often favour or precipi-
on social correlates of suicide are often rich in tate a depressive crisis among predisposed
examples of this kind of problem32: in this people: the quantity and quality of access to
study, as in others,11 13 33 the aggregate level social support also influences the course of a
cross sectional analysis indicates that areas with depressive episode.46 47 Job loss usually com-
higher unemployment rate percentages have prises a whole sequence of stressful events from
lower suicide rates, whereas at an individual anticipation of job loss, to job search and train-
level, the unemployed were found to have ing for re-employment, when possible.48 49 Psy-
higher suicide rates than the employed. To chological wellbeing can therefore be dimin-
explain this discrepancy many factors have ished by the experience of unemployment, and
been invoked: general opinion is that, in eVectively, losing a job seems a likely trigger for
regions with lower unemployment, becoming mental disorders, in particular those of a
unemployed implies entry into a high risk depressive type.50–52 Exclusion from ordinary
bracket (in terms of access to supportive living patterns, customs and activities arising
networks), whereas in regions where many from a lack of resources adds independently to
people are long term unemployed the develop- the stress caused by job loss, and further
ment of alternative supportive relations to increases the risk of depression and subsequent
those related to working roles is suicide.7 53–55
facilitated.11 34–36 In addition, in regions with
higher unemployment, the stigma associated RECENT STUDIES AND FUTURE DIRECTIONS
with being unemployed will be lower, and more An oft disregarded aspect in the study of the
tolerable.11 33 link between unemployment and suicide is the
The higher suicide risk among unemployed role of social class. People belonging to the less
men, although being a reflection of the trend in advantaged social classes have a significantly
the general population, suggests that the lack of higher probability of becoming unem-
a job has a diVerent influence on the two sexes. ployed.21 36 During an economic crisis men in
Men more than women associate their social professional and managerial occupations will
identity with a specific working role, the loss of tend to maintain their employment, whereas
which can seriously undermine a man’s those in semiskilled and unskilled occupations
position in society, and not only in terms of are at greater risk of job loss.36 In addition, both
access to resources. Women, on the other hand, social class at birth and adverse financial
are able to assume a well defined social position circumstances in childhood are predictive of
even in the absence of a specific working role. unemployment in adulthood.21 The risk of
The position of “housewife”, indeed, confers becoming unemployed, therefore, increases
on women precise commitments and occupa- inversely as a function of social class: the more
tion of time in defined activities, which imply disadvantaged are at higher risk.
responsibility and personal engagement. It is Breed56 and Maris,57 in USA, and later
less common for a woman than for a man to Kreitman and coworkers58 in Great Britain,
feel deprived of her social identity, as long as showed that the lowest social class also has
she can mantain a specific role in her familial higher suicide rates compared with other
network. This may justify the lesser impact of social classes. Though higher suicide rates
being unemployed on administrative suicide among the less advantaged may be a reflection
rates for women. of social drift associated with chronic mental
Taking into account the limitations of illness, the impact of poverty and material
ecological analyses, you can speculate on the deprivation, and of unemployment as an
link between the rise in male unemployment aggravating factor, could equally contribute to
percentage and the changes in suicide rates this finding. Poverty is an important risk factor
among women. The relation between the rise for both the development and the outcome of
in male unemployment rate percentage and the a mental illness.59–62 Therefore, it is not
increase in the suicide rates among unem- surprising that a person of a low socioeco-
ployed women may be a reflection of the nomic status is at risk of both a mental disor-
destructive eVect exerted by economic crisis on der and of unemployment. Recently, Neele-
the degree of family cohesion. Studies per- man and coworkers reported that low parental
formed in other contexts show that the social class is a risk factor for suicide and pre-
dissolution of the familial network is particu- mature death in the general population, as is
larly closely linked to suicide risk in the female poor academic performance, which may be
population.37–40 Unemployment also favours related to adverse circumstances in childhood
the dissolution of familial ties, as it implies and adolescence.63 In a study concerning the
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

700 Preti, Miotto

role of unemployment in violent suicide 3 La Vecchia C, Lucchini F, Levi F: Worldwide trends in sui-
cide mortality, 1955–1989. Acta Psychiatr Scand 1994;90:
attempt survivors, Beautrais and coworkers 53–64
also reported that adverse social and family 4 Diekstra RFW, Garnefski N. On the nature, magnitude, and
causality of suicidal behaviors: and International perspec-
factors antecedent to the experience of unem- tive. Suicide Life Threat Behav 1995;25:36–57.
ployment were significantly related to suicide 5 Barraclough BM, Bunch J, Nelson B, et al. A hundred cases
of suicide. Clinical aspects. Br J Psychiatry 1974;125:355–
attempt.20 In their study they found that those 78.
making serious suicide attempts reported 6 Tanney BL. Mental disorders, psychiatric patients and
suicide. In: Maris RW, Berman AL, Maltsberger JT, et al,
higher rates of unemployment than control eds. Assessment and prediction of suicide. New York:
subjects. Much of the association between Guildford Press, 1992:277–320.
7 Mòscicki EK. Identification of suicide risk factors using epi-
unemployment and suicidal behaviour in their demiologic studies. Psychiatr Clin North Am 1997;20:499–
sample was attributed to factors antecedent to 517.
8 De Leo D, Ormskerk SCR. Suicide in the elderly: general
both the onset of unemployment and the characteristics. Crisis 1991;12:3–17.
suicide attempt such as lack of formal 9 Kienhorst IWM, De Wilde EJ, Diekstra RFW. Suicidal
behaviour in adolescents. Archives of Suicide Research 1995;
educational qualifications, poor parental rela- 1:185–209.
tionship and low levels of parental care. All 10 Shepherd DM, Barraclough BM : Work and suicide: an
empirical investigation. Br J Psychiatry 1980;136:469–78.
these factors, however, are more likely to occur 11 Platt S. Parasuicide and unemployment. Br J Psychiatry
among people of low socioeconomic status64: 1986;149:401–5.
12 Pritchard C. Is there a link between suicide in young men
as the comparison group was stratified by age and unemployment? A comparison of the UK with other
and gender but not social class, this aspect had European community countries. Br J Psychiatry 1992;160:
750–6.
been lost in their analyses. A closer exploration 13 Platt S. Unemployment and suicide behaviour: a review. Soc
of the role of socioeconomic status would Sci Med 1984;19:93–115.
14 Jin R, Chandrakant PS, Svoboda TJ. The impact of unem-
doubtless oVer useful additional information ployment on health: a review of the evidence. Can Med
on suicidal behaviour and the factors aVecting Assoc J 1995;153:529–40.
15 Pritchard C. Suicide, unemployment and gender in the
its lethal outcome. British Isles and European Economic Community (1974–
Even taking into account the contribution of 1985). Soc Psychiatry Psychiatr Epidemiol 1988;23:85–9.
16 Morrel S, Taylor R, Quine S, et al. Suicide and
social class in mediating the link between job unemployment in Australia 1907–1990. Soc Sci Med,
loss and suicide, the role of unemployment in 1993;36:749–56.
17 Catalano R. The health eVects of economic insecurity. Am J
the chain leading to suicidal crisis is far from Public Health 1991;81:1148–52.
clear. Many recent studies emphasise the 18 Wasserman IM. Economy, work, occupation and suicide. In:
Maris RW, Berman AL, Maltsberger JT, et al, eds.
importance of a mental disorder to explain the Assessment and prediction of suicide New York: Guildford
higher risk of suicide among the Press, 1992:520–39.
unemployed.19 20 Unemployment in these stud- 19 Jones SC, Forster DP, Hassanyeh F. The role of unemploy-
ment in parasuicide. Psychol Med 1991;21:169–76.
ies is thought to account for a very small frac- 20 Beautrais AL, Joyce PR, Mulder RT. Unemployment and
tion of the risk, contributing merely to worsen serious suicide attempts. Psychol Med 1998;28:209–18.
21 Montgomery SM, Bartley MJ, Cook DG, et al. Health and
the decourse of a mental disorder or being a social precursors of unemployment in young men in Great
consequence of the disorder itself. Britain. J Epidemiol Community Health 1996;50:415–22.
22 Goldman N. Social factors and health: the causation-
Job loss, however, by implying a contraction selection issue revisited. Proc Natl Acad Sci USA 1994;91:
of the personal social network and a relevant 1251–5.
23 Iversen L, Andersen O, Andersen PK, et al. Unemployment
alteration of the time structure of daily life, may and mortality in Denmark, 1970–80. BMJ 1987;295:879–
lead to a reduction in surveillance, which, 84.
24 Moser KA, Goldblatt PO, Fox AJ, et al. Unemployment and
together with availability to lethal means, is mortality: comparison of the 1971 and 1981 longitudinal
another key element in suicide, particularly study census samples. BMJ 1987;294:86–90.
25 Martikainen PT, Valkonen T. Excess mortality of unem-
among mentally troubled people.55 Therefore ployed men and women during a period of rapidly increas-
unemployment, even if symptomatic of a men- ing unemployment. Lancet 1996;348:909–12.
26 Stack S. Gender and suicide among laborers. Archives of
tal disorder, should always be taken into Suicide Research 1995;1:19–26.
consideration as a risk factor for suicide. In 27 O’Carroll PW. A consideration of the validity and reliability
of suicide mortality data. Suicide Life Threat Behav
fact, unemployment has a negative influence on 1989;19:1–16.
both self esteem and the ability to use support- 28 Maris RW. Pathways to suicide. Baltimore: John Hopkins
University Press, 1981.
ive networks in a eYcient way.11–13 Unemploy- 29 Pescolido BA, Mendelsohn R. Social causation or social
ment is also a relevant source of social stress construction of suicide ? An investigation into the social
organization of oYcial rates. American Sociological Review
leading to increased family tensions, increased 1986;51:80–101.
isolation from others, and the loss of self 30 Williams P, De Salvia D, Tansella M. Suicide and the Italian
psychiatric reform: an appraisal of two data collection sys-
esteem and confidence.11 Moreover, lack of tems. Eur Arch Psychiatry Neurol Sci 1987;236:237–40.
economical independence as a result of unem- 31 Robinson WS. Ecological correlations and the behavior of
individuals. American Sociological Review 1950;15:351–7.
ployment reduces the possibility of using social 32 Selvin H. Durkheim’s suicide and problems of empirical
and health services: this may prejudice compli- research. American Journal of Sociology 1958;63:607–19.
33 Platt S, Micciolo R, Tansella M. Suicide and unemployem-
ance with therapeutically prescribed treatment, ent in Italy. Soc Sci Med 1992;34:1191–201.
contributing to worsen the course of a mental 34 Jackson PR, Warr P. Mental health of unemployed men in
diVerent part of England and Wales. BMJ 1987;295:525.
disorder. 35 Kennedy BP, Kawachi I, Prothrow-Stith D. Income
distribution and mortality: cross sectional ecological study
of the Robin Hood index in the United States. BMJ 1996;
The authors would like to thank the ISTAT staV in Cagliari for 312:1004–7.
their help and kindness; the staV of the School of Economics, of
“Medicina del Lavoro” and the “Brotzu” Hospital Libraries in 36 Bartley M, Owen C. Relation between socioeconomic
Cagliari for their assistance; Mr Thomas Eagle for help in the status, employment, and health during economic change,
revision of the English translation. 1973–1993. BMJ 1996;313:445–9.
Conflicts of interest: none. 37 Stack S. Marriage, family, religion and suicide. In: Maris
RW, Berman AL, Maltsberger JT, et al, eds. Assessment and
prediction of suicide. New York: Guildford Press, 1992:540–
1 La Vecchia C, Bollini P, Imazio C, et al. Age, period of death 52.
and birth cohort eVects on suicide mortality in Italy, 1955– 38 Breault KD. Suicide in America: a test of Durkheim’s theory
1979. Acta Psychiatr Scand 1986;74:137–43. of religious and family integration, 1933–1980. American
2 Crepet P, Caracciolo S, Casoli D, et al. Suicidal behavior in Journal of Sociology 1986;92:628–56.
Italy: data, trends and guidelines for a suicide intervention/ 39 Trovato F. The relationship between marital dissolution and
prevention policy. Suicide Life Threat Behav 1991;21:263– suicide: the Canadian case. Journal of Marriage and the
78. Family 1986;48:341–8.
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

Suicide and unemployment 701

40 Trovato F. A longitudinal analysis of divorce and suicide in 52 Viinamaki H, Koskela K, Niskanen L, et al. Unemployment
Canada. Journal of Marriage and the Family 1987;49:193– and mental wellbeing: a factory closure study in Finland.
203. Acta Psychiatr Scand 1993;88:429–33.
41 Stack S, Wasserman I. Marital status, alcohol consumption, 53 Umberson D. Family status and health behaviors: social
and suicide: an analysis of national data. Journal of control as a dimension of social integration. J Health Soc
Marriage and the Family 1993;55:1018–24. Behav 1987;28:306–19.
42 Rook K, Dooley D, Catalano R. Stress transmission: the 54 Heikkinen M, Aro H, Lonnqvist J. Rcent life events, social
eVects of husbands’ job stressors on the emotional health of support and suicide. Acta Psychiatr Scand 1994;377
their wives. Journal of Marriage and the Family1991;53:
165–77. (suppl):65–72.
43 Roy A. Risk factors for suicide in psychiatric patients. Arch 55 Chew KSY, McCleary R. A life course theory of suicide risk.
Gen Psychiatry 1982;39:1089–95. Suicide Life Threat Behav 1994;24:234–44.
44 Blumenthal S, Bell V, Neumann NV, et al. Mortality and 56 Breed W. Occupation, status, and suicide: toward a redefini-
rate of suicide of firts admission psychiatric patients. tion of anomie. American Sociological Review 1958;23:131–9.
Psychopathology 1989;22:50–6. 57 Maris R. Suicide, status, and mobility in Chicago. Social
45 Fawcett J, Busch KA, Jacobs D, et al. Suicide: A Forces 1967;46:246–56.
four-pathway clinical-biochemical model. Ann NY Acad Sci 58 Kreitman N, Carstairs V, DuVy J. Association of age and
1997;836:288–301. social class with suicide among men in Great Britain. J
46 Brown GW, Harris TO. Social origins of depression. London: Epidemiol Community Health 1991;45:195–202.
Tavistock, 1978. 59 Saraceno B, Barbui C. Poverty and mental illness. Can J
47 Brugha TS. Social support and psychiatric disorder: research Psychiatry 1997;42:285–90.
findings and guideline for clinical practice. Cambridge: Cam- 60 Holzer CE, Shea BM, Swanson JW, et al. The increased risk
bridge University Press, 1995. for specific psychiatric disorders among persons of low
48 Whelan CT. The role of income, life-style deprivation and socioeconomic status: evidence from the epidemiologic
financial strain in mediating the impact of unemployment catchment area surveys. American Journal of Social Psychia-
on psychological distress: evidence from the Republic of try 1986;4:259–71.
Ireland. Journal of Occupational and Organizational Psychol- 61 Dohrenwend BP. Socioeconomic status (SES) and psychiat-
ogy 1992;65:331–44.
49 Dooley D, Fielding J, Levi L. Health and unemployment. ric disorders. Soc Psychiatry Psychiatr Epidemiol 1990;25:
Annu Rev Public Health 1996;17:449–65. 41–7.
50 Dooley D, Catalano R, Wilson G. Depression and 62 Dohrenwend BP, Levav I, Shrout PE, et al. Socioeconomic
unemployment: panel findings from the Epidemiological status and psychiatric disorders: the causation-selection
Catchment Area study. Am J Community Psychol 1994;22: issue. Science 1992;255:946–51.
745–65. 63 Neeleman J, Wessely S, Wadsworth M. Predictors of suicide,
51 Dooley D, Catalano R, Hough R. Unemployment and alco- accidental death, and premature natural death in a general-
hol disorder in 1910 and 1990: drift versus social population birth cohort. Lancet 1998;351:93–7.
causation. Journal of Occupational and Organizational 64 Brook-Gunn J, Duncan GJ. The eVects of poverty on
Psychology 1992;65:277–90. children. The Future of Children 1997;7:55–71.
 Downloaded from [Link] on October 26, 2017 - Published by [Link]

Suicide and unemployment in Italy, 1982-1994.

A Preti and P Miotto

J Epidemiol Community Health 1999 53: 694-701

doi: 10.1136/jech.53.11.694

Updated information and services can be found at:

[Link]

These include:

Email alerting Receive free email alerts when new articles cite this article. Sign up in the
service box at the top right corner of the online article.

Notes

To request permissions go to:

[Link]

To order reprints go to:

[Link]

To subscribe to BMJ go to:

[Link]