Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

DOI 10.1007/s00127-011-0393-5

ORIGINAL PAPER

Gender differences in suicide methods

Valerie J. Callanan Mark S. Davis

Received: 13 August 2010 / Accepted: 4 May 2011 / Published online: 22 May 2011
Springer-Verlag 2011

Abstract
Introduction Gender differences in suicide completion
rates have been attributed to the differences in lethality of
suicide methods chosen by men and women, but few empirical studies have investigated factors other than demographic
characteristics that might explain this differential.
Methods Data from the 621 suicides in Summit County,
Ohio during 19972006 were disaggregated by gender to
compare known correlates of suicide risk on three methods
of suicidefirearm, hanging and drug poisoning.
Results Compared to women, men who completed suicide with firearms were more likely to be married and
committed the act at home. Unmarried men were likelier to
hang themselves than married men, but unmarried women
were less likely to hang themselves than married women.
Men with a history of depression were more likely to
suicide by hanging, but women with depression were half
as likely to hang themselves compared to the women
without a history of depression. Men with a history of
substance abuse were more likely to suicide by poisoning
than men without such history, but substance abuse history
had no influence on womens use of poisoning to suicide.
For both sexes, the odds of suicide by poisoning were
significantly higher for those on psychiatric medications.
Keywords Suicide methods
Gender
Firearms

Poisoning
Hanging

V. J. Callanan
Department of Sociology, The University of Akron, Akron, USA
M. S. Davis (&)
Criminal Justice Research Center, The Ohio State University,
Columbus, USA
e-mail: [email protected]

Introduction
Research on suicide in the United States has consistently
found gender differences in suicidal behavior. Most
notably, males take their own lives at nearly four times
the rate of females and comprise approximately 80% of
all suicides, even though female suicide attempt rates
are estimated to be three to four times higher than mens
[13].
One of the primary reasons given for the large gender
gap in suicide completion rates is the difference in suicide
methods used by males and females. In general, men are
more likely to use methods that ensure lethality than are
women. The most common suicide method used by men is
firearms; current statistics indicate that 56% of males who
committed suicide in 2006 used a firearm [1]. In contrast,
women are less likely than men to commit suicide with
firearms but more likely to commit suicide by poisoning
[1]. Since women are more likely to attempt suicide by
poisoning, they stand a greater chance of being rescued or
resuscitated than men, who typically use firearms, which
are more likely to inflict mortal wounds. The gender differences in choice of suicide method contribute to the large
gender differential in suicide attempts versus completions.
Why women choose less lethal means than men has been
attributed to a number of factors, including intent to die,
gender socialization, and easy availability of methods.
Although the number of female firearm suicides eclipsed
female poisoning suicides during the 1980s and 1990s, the
percentage of women that used firearms remained much
lower than that of men [3, 4]. For example, during the
1980s and 1990s, male firearm suicide rates peaked at 13.4/
100,000 (in 1990), which was nearly six times higher than
the peak rate for female firearm suicides (2.3/100,000 in
1981). Recent statistics indicate that poisoning deaths have

123

858

increased among women and now are the most common

method used [1].
Gender differences in suicide method are further complicated by age differences that vary across gender. Over
the time period 19801996, the highest rate of female
firearm suicides was found among those 4554 years old
(3.8/100,000); among males, the highest rate of firearm
suicide (43.8/100,000) was found among those ages
7584 years [5]. There are also age differences across
gender in other methods of suicide [6].
Besides sex and age, other variables may be relevant to
the chosen method of suicide. For example, the rates of
suicide methods can vary considerably over different historical periods, which may reflect not only lethality of
available methods, but also changes in unemployment rates
[7, 8], and cultural acceptance of certain methods [9].
While changes in the rates of various suicide methods have
been noted in the literature, little research has examined
suicide method at the individual level. Of those that have,
most typical are investigations of the factors associated
with a particular method of suicide. Kosky and Dundas
[10], for example, examined all hanging deaths of individuals under the age of 25 that occurred in Queensland,
Australia, during 1995 and 1996 and found employment
status, prior suicide threats and attempts, psychiatric illness, and marital separation were correlated with hanging
suicides. This approach reveals information about individuals that chose hanging, but not if these suicides differ
from suicides that did not choose hanging as their means of
death. Moreover, few multivariate analyses have compared
method of suicide between men and women. Instead, most
studies of suicide method treat gender as a control variable,
which cannot illuminate gender differences in risk factors.
The gender differential in suicide has been one of most
perplexing and controversial issues in the study of suicidal
behavior, largely because of gender biases, which have
influenced both theory and research. We believe that one
key to rejecting such biases is challenging them with
empirical studies that examine all possible factors that
could account for these differences. The detailed examination of the type and use of suicide methods is one way to
help achieve this end.
The aim of this study, therefore, is to assess gender
differences in method and place of suicide by comparing
factors relevant to suicide risk across subsamples of men
and women. This study is unique insofar that it includes a
number of factors that have been found to be correlated
with suicide risk, but have not been included in studies that
examine suicide method. We examine these factors on risk
of various suicide methods controlling for gender, and then
across gender. In splitting the sample by gender, we are able
to assess if predictors of suicide method work equivalently
for women and men. We believe a better understanding

123

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

gender differences in suicide method can not only advance

the research literature, but also could have important
implications for suicide prevention.
Gender differences in method of suicide
A review of the extant literature finds that gender differences in suicide method is usually attributed to one of three
reasons. The most common perception is that women are
more likely to use less lethal methods because they do not
really want to kill themselves. According to this line of
reasoning, women who attempt suicide are crying for help
and are not really motivated to die [3]. Thus, women are
more likely to ingest pills or other toxic substances which
increases the probability of intervention. While it is true
that method influences lethality [2], some dispute the idea
that women use less violent means of suicide because they
are not really serious about dying. A psychological autopsy
of 141 male and female suicide victims by Denning et al.
[11], for example, found no gender differences in intent to
die using the Beck Suicide Intent Scale [12]. However,
other studies using the same scale found that the intent to
die was greater for males than for females [1315].
The second perspective argues that gender socialization
decreases the likelihood that women will own, have access
to, or be as familiar with firearms compared to men. This
suggests that cultural scripts make it more likely that
women turn to other suicide methods such as poisoning or
drowning [16, 17]. Because there is no national gun registration system in the US, we cannot accurately ascertain
gender differences in gun ownership. Data from the General Social Survey (GSS), however, suggests that approximately 12% of women own guns, especially those who
live in rural areas and among those who hunt [18].
Approximately 7.5% of women own handguns, and almost
4% own long guns. Even though women are far less likely
to own guns than men, many would have access to firearms
owned by males with whom they live. Public health studies, e.g. [19, 20] find that suicide risk is strongly elevated
with the presence of firearms in the household. Wintemute
et al. [20], for example, examined risk of suicide for those
that purchased a handgun in California and found that in
the first year after purchasing this weapon, 24.5% committed suicide. Of relevance, 75.4% of the women who
bought handguns were between the ages of 21 and 44 years
old, and of this subset, fully 51.9% committed suicide
within the year after purchasing the gun.
The third explanation for the observed gender disparity
in the use of suicide methods centers on womens concern
with physical disfigurement of their body [21, 22]. Some
have suggested that women prefer to leave a beautiful
corpse because of societal emphasis on female physical
appearance [23]. Others suggest that because women are

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

more concerned with peoples feelings than are men, they

are less likely to leave a mutilated corpse for their loved
ones to find [24]. Thus, women are more likely to use
methods that do not disfigure the face, such as drug
poisoning.
Since the data used in this study are of suicide completers, we cannot directly test the impact of factors such as
intent to die and other motivations that may have influenced choice of suicide method. We can, however, examine if other factors correlated with gender differences in
suicide risk are also relevant to gender differences in suicide method, which has not been done heretofore.
Correlates of suicide risk
There are a number of established correlates of suicide risk
that vary by gender that might also be correlated with
suicide method. First, studies of those who have attempted
suicide and lived, whether it was due to an unsuccessful
attempt, interruption, or a change of heart, have found
these individuals to be at much higher risk of attempting
suicide again [2527]. Although some studies suggest
about one-half of suicide attempters were not serious about
ending their lives [28], there still exists a subset of those
who intended to die, but survived. Further, follow-up
studies of suicide attempters find that those that have tried
and failed may be inclined to use a more lethal method in
subsequent attempts. One study has found that this varies
by gender insofar that men with prior attempts are more apt
to use more lethal means in subsequent attempts than are
women [29]. Because women in the United States are
nearly four times more likely to attempt suicide than men,
research on gender differences in suicide method should
control for prior suicide attempts.
Second, research has consistently found that depression
and mood disorders are highly correlated with suicidal
behavior [30]. What remains relatively unknown, however,
is whether these conditions are correlated with method of
suicide, although the one study to date that has examined
this question found no relationship [31]. Because depressive disorders are twice as high among suicidal women as
men [32], studies of suicide methods should account for the
influence of these conditions.
Some studies have found that being on psychiatric
medication increases the odds of killing oneself with these
medications [33]. This is relevant to gender differences in
suicide method because women are twice as likely to be
diagnosed with depression and mood disorders as men
[34]. This difference appears in adolescence and continues
through adulthood, and is found in many cultures and
countries [35]. If women are more likely to be diagnosed,
they are also probably more likely to be on psychiatric
medications. Studies find that women in the US are much

859

more likely to be prescribed medication for depression than

are men [36], thus increasing their odds of suicide by drug
poisoning and decreasing their likelihood of suicide by
other means. Such findings belong under the aegis of the
opportunity theory of suicide that argues availability of
suicide method is an important predictor of suicide
behaviors [8]. Therefore, studies should examine such
differential opportunities by controlling for whether the
individual had prescribed psychiatric medications at the
time of suicide.
The same can be said of controlling for reported history
of substance abuse as this is an important covariate of
suicide attempts and completions [2830] and varies by
sex. Studies of completed suicides have found that men
have higher rates of substance abuse disorder than women
[37], although some studies suggest this gender gap is
minimal among younger age suicide decedents [26, 38].
All of these gender differences in correlates of suicide risk
present opportunities to disentangle the gender differential
in suicide methods.
While gender differences in suicide method have long
been noted in the suicide literature, relatively few studies
have conducted multivariate analyses on method of suicide
risk that included more than socio-demographic variables.
One recent exception examined the influence of sex and
place of death on suicide method using the population of
completed suicides in a 3-year period from one county in
southern California [39]. The data used in this study are
remarkably similar to those of Kposowa and McElvains
data [39] in terms of dates and sample size, thus allowing
for a fairly close replication of their test of the gender
differential in suicide methods that controlled for gender,
age, marital status and place of suicide.1
Their analyses treated gender as a control variable,
however, which does not reveal potential gender interactions with other factors that may be correlated with various
methods of suicide. The data used in this study makes it
possible to assess if other factors are correlated with suicide method, and if these factors might contribute to the
gender differential in method of suicide. This study splits
the sample by gender and regresses three methods of suicidefirearms, hanging and poisoningon predictive
factors of suicide risk found in the research literature to
ascertain if these variables work similarly for men and
women in their choice of suicide method. To our knowledge, this approach has not been done to date.

The study period of Kposowa and McElvain [39] covered

19982001, which is subsumed within our study period, and they
collected a similar number of usable cases (643) to the data in this
study (621).

123

860

Data and method

The data used in the analyses come from the population of
suicides over a 10-year period (19972006) in Summit
County, OH. The information used in our study comes
from the Summit County Medical Examiners Office case
files.2 All suspicious and unusual death scenes, including
suicides, are investigated by medico-legal board certified
death investigators who must also have a police officer
certification or a Bachelors degree in criminal justice.
These investigators examine the death scene, and conduct
interviews with people at the scene and later with individuals who might have relevant knowledge of the suicide
decedent. The bodies of those who die from suspicious or
unusual circumstances are required by Ohio law to be
autopsied. The state of Ohio requires that all coroners and
medical examiners must be licensed physicians, and must
complete 32 h of continuing education every 4 years. The
Chief Deputy Medical Examiner of Summit County is an
M.D. with years of training in forensic pathology, as are the
Deputy Medical Examiners on staff.
The breadth of information contained in the case files
allowed for an examination of a wide array of factors. Each
case file included an investigation report, medical autopsy,
and almost always, a toxicology screen. Most of the
information used in the analyses comes from the investigation reports, which are much more detailed than death
certificates, and often more accurate, as studies using death
certificates often have missing data on key variables, such
as gender, e.g. [39]. The reports contain information not
only about the method of suicide, but also about the suicide
scene, such as the presence of alcohol or other drugs,
location of the suicide, and an account of the decedents
activities, disposition, and behavior shortly before death.
Current and past mental and physical health status and
treatment, substance abuse history, as well as family,
employment and legal problems are also reported. The
information is obtained from interviews with law
enforcement and other first responders to the death scene,
family members and other intimates, but could also be
later obtained from neighbors, friends, and health care
providers.
Every investigative report, including the narrative, toxicology screen and portions of the autopsy report were
numerically coded by a team of graduate students and the
study authors. The coding schemes were developed on a
pilot sample of twenty cases in consultation with the
Medical Examiners Office. Each case was independently
coded by at least three members of the research team.
Meetings between the coders and the study authors were
2

IRB approval was obtained from the authors respective

universities.

123

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

regularly held so that all cases were reviewed and to

resolve any coding disagreements.3 Items without group
consensus were coded as missing. Approximately, 20% of
all cases were audited by the study authors to check for
accuracy of coding and data entry.
The analyses began with bivariate comparisons of men
and women for method and place of suicide. This was
followed by three multivariate logistic regression analyses
in which each outcome variable was modeled in terms of
the odds of decedents using a particular method (coded as
1) and all other methods (coded as 0), controlling for
gender, place of suicide, age, race, and marital status.4
Gender is coded with men as 0 and women as 1. Place of
suicide is a categorical variable with four conditions,
public place, residence, work place, and hotel/motel; public
place is the omitted category. Age is a five category variable (\30, 3044, 4554, 5564, and 65 and older); under
30 is the omitted category. Race is a dichotomous variable
with non-White coded as 1 and White coded as zero.
Marital status is dichotomized into married (0) and nonmarried (1).
The remaining logistic regressions split the sample by
gender to compare the effects of the above variables and
five other variables relevant to suicide risk: living alone,
history of substance abuse, prior attempts, history of
depression, and being on psychiatric medication at the time
of suicide on the three methods of suicide. This information came from family members or others close to the
decedent as told to and noted in the report by the death
investigator on the scene. Each of the five variables is
measured dichotomous, with having the condition coded as
1 and the absence of the condition coded as zero, which
served as the omitted categories. If the decedent recently
had abused alcohol, prescription drug, and illicit drugs they
were coded as having a history of drug abuse.5 History of
depression was seldom a clinical diagnosis as most of this
information came from close family members or friends
instead of physicians.

Results
Descriptive analyses
Of the 621 confirmed suicides in Summit County, Ohio
from 1997 to 2006 480 were male (77.3%) and 141 were
3

Cohens Kappa was not estimated given the coding protocol.

Analysis was limited to these three suicide methods because the
other methods used had too few cases, which precluded multivariate
analyses.
5
Specific types of substance abuse were not examined due to lack of
statistical power.
4

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

861

female (22.7%), which is very close to the gender distribution reported by Kposowa and McElvain (78.2 and
21.8%, respectively), as well as national data from 1999 to
2007 (79.9 and 20.2%, respectively). Given regional differences in the race/ethnicity of populations, comparisons
to other areas or even national data are not valid. More than
90% of suicide victims in our sample were White, 7.4%
were African American, 0.8% were Asian and 0.6% Hispanic. The study decedents ranged in age from 9 to
91 years old, with a mean age of 45. Twenty-three percent
of the suicide decedents in the study sample were under age
30. Those aged 65 or older comprised 18.8% of our cases.
Almost 31% were married, over 36% were single, 26.1%
were separated, and 6.9% were widowed.
Relevant descriptive information about method and
place of suicide is presented in Table 1. Although various
methods of suicide are shown in Table 1, subsequent
analyses were limited to three (firearms, hanging and poisoning). For both sexes, suicide by use of firearms was the
main method of choice and was used in 48.7% of suicides.
Hanging was the second most common method (21.4%).
Only 10.3% of our sample committed suicide by poisoning.
Fewer than 10% of the sample chose carbon monoxide
poisoning, jumping from heights, suffocation, or other
methods, which included cutting (n = 7), putting oneself in
front of a train (n = 11), and setting oneself on fire
(n = 8).

Table 1 Differences in method and place of suicide between men

and women
Men

Women

Total

Method of suicide
Hanging or
strangulation

23.1 (111)

15.6 (22)

21.4 (33)

Use of firearms

51.8 (249)

38.3** (54)

48.7 (303)

Poisoning

6.9 (33)

22.0*** (31)

10.3 (64)

Carbon monoxide (CO)

5.6 (27)

6.4 (9)

Suffocation

2.1 (10)

5.0 (7)

2.7 (17)

Jump from high place

5.2 (25)

5.7 (8)

5.3 (33)

Other

5.4 (26)

7.1 (10)

5.8 (36)

Total

100.0 (481)

100.0 (141)

100.0 (622)

78.3 (376)

80.9 (114)

78.9 (490)

2.3 (11)

0.7 (1)

1.9 (12)

Public area

17.1 (82)

16.3 (23)

16.9 (105)

Hotel/motel

1.9 (9)

2.1 (3)

1.9 (12)

Other non-public area

0.4 (2)

5.8 (36)

Place of suicide
House
Place of business

Total

100.0 (480)

The values are in percentage

* p \ 0.05, ** p \ 0.01, *** p \ 0.001

0 (0)
100.0 (141)

0.3 (2)
100.0 (621)

As expected, there were gender differences in the

method of suicide. Although the use of firearms was the
most popular method for both men and women, 51.8% of
men committed suicide with a firearm (n = 249), but only
38.3% (n = 54) of women used this method.6 With respect
to the use of long guns versus handguns, we found that
although men were more than twice as likely to use long
guns as women (22.4 and 11.3%, respectively), both men
and women were much more likely to use handguns (77.6
and 88.7%, respectively); these gender differences in type
of gun used were not statistically significant. The second
most common method among men was hanging (23.1%),
followed by poisoning (6.9%). Unlike men, womens second preferred method was poisoning (22.0%), followed by
hanging (15.6%).
As shown in Table 1, most suicides occurred at home
(78.9%). The second most common place of suicide was in
a public area (park or wooded area, bridge, or other public
place), which constituted 16.9% of cases; followed by
place of business and hotel/motel. There were no significant gender differences in place of suicide. Fewer women
committed suicide at a place of business (0.7%) compared
to men (2.3%), but among both sexes, this place of death
was rare.
Whereas 63.4% of men and 65.0% of women in the
southern California sample study died at home, 78.3% of
men and 80.9% of women in our Ohio sample committed
suicide at home. The differences between the samples may
be attributed to differences in coding. Kposowa and
McElvain [39] reported place of death in their descriptive
statistics, whereas our study records place of suicide. These
are not synonymous inasmuch as some of those who
commit suicide at home actually die while en route to or at
a hospital.7 Whereas place of death may be important
regarding such questions as the efficacy of medical care in
specific locations [42], the location of suicide attempt is
more properly the focus for analyses of gender, method,
and place of suicide.
Although not shown, we also examined gender differences in suicide method and place of suicide conditioned
6

The percentage of men who used firearms to suicide in Summit

County during the period 19972006 was lower than the percentage
of male firearm suicides for the entire state for that time period (51.7
and 57.8%, respectively), whereas the percentage of female suicides
in the county who used firearms was higher than female firearm
suicides in the state (38.3 and 31.4%, respectively). Of interest,
firearms remain the most common means of suicide for females in our
population, similar to earlier trends reported by Kaplan et al. [40], but
different from more recent statistics that show poisoning has once
again become the most common method among women in Ohio
(40.3%) as well as the United States (34.6%) [41].
7
If we remove the cases that died in the hospital, 76.1% of their
sample committed suicide at home (408/536), which was very close to
the percent of cases in our sample that committed suicide at home.

123

862

by age. Of relevance, men 65 and older were nearly twice

as likely to use firearms as women in this age category
(79.2 vs. 41.2%, respectively). There were also age and
gender interactions among those who committed suicide by
hanging or poisoning. More than twice as many men
between the ages of 3064 hanged themselves compared to
women in this age group (23.0 vs. 10.1%, respectively). In
contrast, women between the ages of 30 and 64 were
almost three times as likely as men in this age group to use
drug poisoning (26.3 and 9.6%, respectively).
There were significant gender and age interactions in
place of suicide. Men ages 3064 were somewhat less
likely to commit suicide at home compared to women in
this age category; conversely, men 65 and older were more
likely to commit suicide at home (95.0%) than women 65
and older (76.5%). Only 4.0% of men ages 65 and older
committed suicide in a public place compared to 23.5% of
women in this age group.
Multivariate analyses
The first set of logistic regressions examines the risk (odds)
of suicide by a given method (for example, hanging) as a
function of gender and place of suicide. In these analyses,
we controlled for age, race and marital status. Multicollinearity was assessed with variance inflation factors (VIF)
and tolerance tests, which did not indicate problems with
multicollinearity.8
Table 2 presents the logistic regression results. With
respect to firearm suicides, women were less likely to use
this method, net of other factors; (OR 0.557; CI 0.369,
0.841). Additionally, firearm suicides were more likely to
occur among older individuals. Specifically, the odds of
those 65 and older of committing firearm suicide were
three times higher than the odds of those under 30 (OR
3.097; CI 1.752, 5.474). Non-married individuals were less
likely to use firearms than those who were married (OR
0.541; CI 0.364, 0.806), and the odds of committing suicide
with a firearm in the home were almost twice as high as the
odds of committing firearm suicide in a public place.
Women were less likely to commit suicide by hanging
compared to men (OR 0.570; CI 0.334, 0.971), which
departs from Kposowa and McElvain, who did not find a
gender differential. The odds of non-Whites using hanging
to suicide were considerably lower than the odds of Whites
(OR 0.385; CI 0.168, 0.882). The odds of suicide by
hanging decrease with every increase in age category, and

Although VIF and tolerance are normally associated with tests of

multicollinearity of OLS regression models, as Menard [43] argues,
[since] the concern is with the relationship among the independent
variables, the functional form of the model for the dependent variable
is irrelevant to the estimation of collinearity (p. 76).

123

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

it is more likely to occur in the home compared to public

places.
Women were far more likely than men to use drug
poisoning to suicide. More specifically, the odds of women
using this method of suicide were almost 3.5 times higher
than the odds of men (OR 3.486; CI 1.978, 6.143), net of
other factors. Poisoning suicides also were higher for those
ages 3054 compared to those under 30. The odds of suicide by poisoning were also related to place. Those who
poisoned themselves were far more likely to have done so
at home (OR 7.027; 1.653, 29.879), or at a hotel or motel
(OR 18.571; CI 2.503, 137.759) compared to the odds of
those who poisoned themselves in a public place.
The remaining analyses go beyond most studies of
gender differences in suicide method by including other
variables that are relevant to gender differences in suicide risk. These additional variables are living alone,
history of depression, history of substance abuse, prior
suicide attempts, and being on psychiatric medication at
the time of the suicide; all are dichotomous variables
with 1 indicating presence of the condition and zero the
absence. Given wide confidence intervals for some categories of place of suicide, place of suicide was recoded
into a dichotomous variable with residence coded as 1
and all other places coded as zero. The first set of
analyses includes gender as a control variable in the
models.
As seen in Table 3, when these additional variables are
introduced, the gender difference in firearm suicides was
rendered non-significant, but the effects of age, marital
status, and place of suicide retained statistical significance.
The only added variable that impacted firearm suicides was
prior attempts. That is, the odds of using a firearm to suicide were about half the odds for those with prior attempts
than those without prior suicide attempts (OR 0.497; CI
0.298, 0.898).
When the additional variables were added to the model
for hanging suicides, all of the coefficients, including
gender, retained statistical significance. Like firearm suicides, the only additional factor that impacted suicide by
hanging was prior attempts, but unlike firearm suicides,
prior attempts increased the odds of hanging (the odds
ratios were 2.5 times higher). Finally, place of suicide is
rendered insignificant for poisoning suicides when the
additional variables are introduced, but gender and age
retain statistical significance. Being on psychiatric medication, however, significantly increased the odds of poisoning suicides (OR 4.281; CI 1.919, 9.549).
In order to examine the potential impact of gender
interactions with other factors related to suicide risk on the
three methods of suicide, the sample was split by gender.
To ascertain if the odds ratios for a particular covariate
are equivalent across gender, we computed z scores and

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

863

Table 2 Logistic regression estimates of the effect of gender on methods of suicide

Variable

Firearm

Hanging

Poison

OR

95% CI

OR

95% CI

0.557**

0.369, 0.841

0.570*

0.334, 0.971

3.486***

1.978, 6.143

3044

0.713

0.435, 1.168

0.454**

0.267, 0.772

3.153*

1.202, 8.270

4554

1.194

0.709, 2.011

0.338***

0.184, 0.620

4.908**

1.831, 13.152

5564

1.929

0.992, 3.752

0.200***

0.081, 0.493

2.259

0.662, 7.711

65 and over

3.097***

1.752, 5.474

0.066***

0.026, 0.167

1.338

0.403, 4.442

0.541**
1.709

0.364, 0.806
0.932, 3.134

1.296
0.385*

0.764, 2.197
0.168, 0.882

1.828
0.957

0.940, 3.555
0.308, 2.978

Female

OR

95% CI

Age

Not married
Non-White
Place of suicide
Residence

1.887**

1.172, 3.037

3.770***

1.839, 7.730

7.027**

1.653, 29.879

Work place

0.600

0.146, 2.455

4.993*

1.216, 20.496

0.515

0.118, 2.238

2.506

0.457, 13.750

Hotel/motel

18.571***

Intercept

0.775

0.216***

0.003***

Nagelkerke R2

0.162

0.190

0.188

2.503, 137.759

OR odds ratio
* p B 0.05, ** p B 0.01, *** p B 0.001
Table 3 Odds ratios and confidence intervals of the effects of gender, place of suicide, and other factors on method of suicide
Variable

Firearm (n = 302)

Hanging (n = 133)

Poison (n = 64)

OR

95% CI

OR

95% CI

OR

95% CI

0.652

0.409, 1.040

0.480*

0.263, 0.877

3.645***

1.873, 7.093

3044

0.925

0.520, 1.644

0.395**

0.213, 0.733

4.897**

1.506, 15.922

4554

1.531

0.830, 2.823

0.305***

0.151, 0.612

5.903**

1.757, 19.824

5564

2.855**

1.302, 6.264

0.248**

0.094, 0.658

2.521

0.558, 11.390

65 and over

3.471***

1.752, 6.878

0.088***

0.032, 0.237

2.289

0.542, 9.674

Not married

0.562**

0.349, 0.906

1.402

0.779, 2.523

2.119

0.953, 4.709

Non-White

1.790

0.887, 3.613

0.408*

0.171, 0.974

0.870

0.248, 3.060

Suicide at residence

2.367***

1.376, 4.073

2.504***

1.221, 5.133

1.985

0.657, 5.991

Lived alone

0.833

0.501, 1.385

0.629

0.330, 1.198

1.715

0.809, 3.636

Drug abuse

0.808

0.504, 1.296

1.020

0.559, 1.735

1.813

0.904, 3.635

Prior attempts

0.497**

0.298, 0.828

2.556***

1.455, 4.493

1.312

0.639, 2.694

History of depression

0.945

0.631, 1.415

0.935

0.571, 1.532

0.967

0.499, 1.876

On psychiatric meds

0.546

0.292, 1.023

0.786

0.367, 1.683

4.281***

1.919, 9.549

Intercept
Nagelkerke R2

0.689

Female
Age

0.196

0.314**
0.220

0.004***

0.214

OR odds ratio
* p \ 0.05, ** p \ 0.01, *** p \ 0.001

associated probabilities that the odds ratios for men and

women were significantly different,9 see [44].
As seen in Table 4, for both sexes, the odds of committing suicide with firearms increase among those ages 45
and older but the gender gap is largest among the oldest
9

Z = (b1 - b2)/H(SEb21 ? SEb22).

suicides. The odds for men 65 and older were almost four
times greater than the odds for men under 30 (OR 3.74),
but the odds for women 65 and older were about twice as
high as women under 30 (OR 2.27); this gender difference
is marginally significant (z score = -1.613, p \ 0.10).
There were also other gender interactions on the likelihood
of firearm suicide. Men were more likely to commit suicide

123

864

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

Table 4 Odds ratios and confidence intervals of firearms suicides by gender

Variable

Men (n = 248)

Women (n = 54)

z score

OR

95% CI

OR

95% CI

3044

0.965

0.503, 1.852

0.828

0.221, 3.110

-1.156

4554

1.160

0.571, 2.356

2.994

0.816, 10.991

-0.063

5564

2.742*

1.079, 6.967

3.921

0.732, 21.010

-0.800

65 and older

3.735***

1.736, 8.032

2.273

0.425, 12.146

-1.613

Not married

0.529*

0.302, 0.926

0.900

0.307, 2.637

-0.382

Non-White
Suicide at residence

1.614
2.490**

0.720, 3.616
1.351, 4.590

2.111
1.159

0.425, 10.484
0.303, 4.440

-0.840
-1.763*

Lived alone

1.015

0.577, 1.099

0.391

0.101, 1.509

-1.943*

History of drug abuse

0.840

0.492, 1.432

0.851

0.280, 2.593

-0.903

Prior attempts

0.586

0.312, 1.099

0.364*

0.138, 0.957

-1.380

History of depression

1.020

0.638, 1.629

0.806

0.339, 1.915

-1.060

Psychiatric medication

0.442*

0.213, 0.999

0.829

0.206, 3.328

-0.412

Intercept

0.658

0.721

Nagelkerke R2

0.206

0.208

Age

OR odds ratio

p B 0.10,* p B 0.05, ** p B 0.01, *** p B 0.001

Table 5 Odds ratios and confidence intervals of hanging suicides by gender

Variable

Men (n = 111)

Women (n = 22)

z score

OR

95% CI

OR

95% CI

0.524

0.263, 1.043

0.126**

0.028, 0.563

-2.480**
-3.473***

Age
3044
4554

0.452*

0.210, 0.971

0.045**

0.007, 0.308

5564

0.377

0.129, 1.107

0.037*

0.003, 0.487

-3.690***

65 and older

0.120***

0.041, 0.353

0.054*

0.004, 0.698

-2.152**

Not married

2.218*

1.129, 4.359

0.364

0.086, 1.533

-2.845***

Non-White
Suicide at residence

0.428

0.166, 1.101

0.509

0.041, 6.273

-1.154

Lived alone

0.502

0.249, 1.012

2.463

0.385, 15.744

-0.847

History of drug abuse

1.070

0.597, 1.918

1.234

0.324, 4.695

-0.712

Prior attempts

2.719**

1.426, 5.184

3.563*

1.024, 12.397

0.447

History of depression

1.066

0.628, 1.809

0.470

0.142, 1.551

-1.756*

Psychiatric medication

Intercept

0.366

1.909

Nagelkerke R2

0.213

0.317

OR odds ratio

p B 0.10, * p B 0.05, ** p B 0.01, *** p B 0.001

in a residence (OR 2.49) than elsewhere, and were more

likely to live alone than women. Among both sexes, the
odds of using a firearm to suicide were significantly lower
among those who had prior suicide attempts than those
who did not (OR 0.442 for men and 0.364 for women), but

123

women with prior attempts were slightly less likely than

men to use a firearm (z score = -1.380, p \ 0.10).
Table 5 presents the logistic regression coefficients and
corresponding odds ratios for suicide by hanging among
men and women. Place of suicide and being on psychiatric

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

865

Table 6 Odds ratios and confidence intervals of poisoning suicides by gender

Variables

Men (n = 33)

Women (n = 31)

z score

OR

95% CI

OR

95% CI

3044

3.628

0.700, 18.798

7.002*

1.113, 44.049

-0.676

4554

8.639**

1.684, 44.323

3.410

0.500, 23.237

-2.276**

5564

3.448

0.410, 28.982

1.424

0.150, 13.502

-2.380**

65 and older

2.834

0.439, 18.315

1.339

0.109, 16.490

-2.244*

Not married

2.068

0.727, 5.886

1.923

0.509, 7.269

-1.174

Non-White
Suicide at residence

0.388
1.621

0.038, 3.994
0.440, 5.975

1.468
3.714

0.245, 8.789
0.393, 35.144

-0.119
-0.517

Lived alone

1.682

0.655, 4.317

3.102

0.711, 13.532

-0.498

History of drug abuse

2.781*

1.129, 6.853

0.985

0.270, 3.603

-2.100*

Prior attempts

0.805

0.261, 2.480

2.221

0.702, 7.028

-0.063

History of depression

0.839

0.345, 2.041

0.922

0.298, 2.847

-0.921

On psychiatric meds

5.291***

1.925, 14.450

5.296*

1.046, 26.815

-1.158

Age

Intercept

0.004***

0.007**

Nagelkerke R2

0.191

0.234

OR odds ratio, CI confidence interval

p B 0.10, * p B 0.05, ** p B 0.01, *** p B 0.001

medication were excluded from the analyses because none

of the women in this subset hanged themselves outside of a
residence or were on psychiatric medications. To make
relevant comparisons, these variables were also excluded
from the regression conducted on the male subsample.10
The odds of committing suicide by hanging decreased
with age for both sexes, but had more of an effect on
women. As seen in Table 5, for every age category above
30, the odds of women committing suicide by hanging
were lower than the odds for men. Further, marital status is
relevant and shows significant gender interaction. That is,
men who were not married were more likely to hang
themselves than married men, whereas unmarried women
were less likely to use hanging than women who were
married (OR 2.218 and 0.364, respectively; z score =
-2.845, p \ 0.001). Prior attempts increased the odds of
hanging for both men and women; there were no statistical
gender differences in the effect of prior attempts.
Table 6 displays the logistic regression coefficients and
associated odds ratios for men and women that used poison
to suicide. One of the largest predictors is age; between
both genders, the odds of using poisoning are higher for
those ages 30 and older. Among women, however, the odds
are significantly lower than the odds for men ages 45 and
older. For example, men between the ages of 45 and 54 are
more likely to use poison themselves than men under 30
10

A model was run on the male subsample that included these

factors, but neither place of suicide or being on psychiatric
medication was related to suicide hangings among men.

(OR 8.639); this effect was twice as large as it was among

women (OR 3.410); z score = -2.276, p \ 0.01. Men with
a reported history of drug and/or alcohol abuse were more
likely to suicide by poisoning than men without such a
history (OR 2.781), but substance abuse history had no
bearing on womens use of poisoning to suicide (z score for
differences in OR -2.100, p \ 0.05). Finally, both men
and women that were on psychiatric medications at the
time of suicide were significantly more likely to suicide by
poisoning than those who were not on psychiatric medications (OR 5.291 for men and 5.296 for women).
Because the odds of suicide by poisoning are elevated
for those who are on psychiatric medications at the time of
death, we attempted to ascertain if level of psychiatric
supervision moderated this relationship. Importantly, we
found (not shown) that the odds of committing suicide by
poisoning for those who are taking psychiatric medications
while under psychiatric outpatient care are no greater than
the odds of those who are taking medications but not under
psychiatric supervision.

Discussion
Using data gleaned from the case files of deaths ruled as
suicide by the Summit County Medical Examiners Office,
we extended prior research on suicide populations with
detailed information on prior suicide attempts, living
arrangements, reported mental health, history of substance
abuse, and being on psychiatric medication. While many of

123

866

these factors operated similarly for men and women, there

were several important gender differences.
This study found that firearms are the most common
method of suicide for both men and women. While this is
not a surprising result for male suicides, this differs from
Kposowa and McElvain [39] who found that firearm suicide was the second most common suicide method among
women (roughly 26%), whereas poisoning was the most
common method. In contrast, over 38% of female suicides
in our sample were committed by firearms, making it the
most common method employed, but still much lower than
the percentage of male firearm suicides. Since our data
come from Summit County, Ohio and theirs from southern
California, this raises the possibility of regional differences
in suicide methods for women but to our knowledge, no
study has been published that examines gender differences
in suicide methods by state.11
We also found a significant gender difference in suicide
by hanging or strangulation; net of other factors, the odds
of women using this method were half that of the odds for
men. Women were far more likely to suicide by drug
poisoning than men, although nearly twice the proportion
of women used this method in southern California as
women in Summit County, OH (40 and 22%, respectively).
Our study also confirmed the age differences in suicide
method noted by prior studies [40, 46]. The risk of suicide
by hanging decreases with age for both men and women,
perhaps reflective of the physical exertion required to
commit suicide in this manner. Conversely, the risk of
firearm suicide increases with age, and even more so for
older men. Finally, the risk of suicide by drug poisoning is
lowest among those under 30, irrespective of gender.
We also found significant gender and age interactions
across all suicide methods. The odds of firearm suicide
were high among both women and men 65 and older, but
were significantly higher for men than for women. Suicide
by hanging decreased with age for both genders, but this
decrease was even greater for women than it was for men.
Last, the odds of using poison to suicide were higher for
both men and women 45 and older, but even higher for
men than for women.
The results also indicate method of suicide differs by
place. We found that individuals are more likely to suicide
at home than in a public place, irrespective of method. This
consistent finding punctuates the importance of the home
as a focus of preventive strategies. For example, suicide
research has established that the odds of firearm suicide are
elevated with the presence of a gun in the home [47, 48].
11
Kaplan and Geling [45] found that female firearm suicide rates
were higher in the Pacific region that included California than in the
East North Central region that included Ohio. But aggregating data by
region may obscure differences in firearm ownership, possession and
use that could vary by state or smaller geographic areas.

123

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

Given the extraordinarily high rates of suicide among older

males, and also that the highest rates of firearm suicides are
found among older women, prevention efforts that target
warnings to this age group about the risk associated with
the presence of firearms in the home might reap significant
results [46]. There is research that suggests that the elimination of first-choice method may not necessarily lead
suicidal individuals to substitute a second method [49, 50;
but see 51]. Employees of agencies and organizations that
serve the at-home elderly population should be educated to
assess the risk to occupants with firearms in the home. The
restriction of access to firearms, however, must be weighed
against the personal protection concerns of this group, a
rapidly growing segment of the US population.
Not all suicide methods are as easily addressed. Hanging, one of more lethal methods, was employed by more
than one-fifth of the suicides in this study. In any given
location, there are numerous materials such as ropes, cords,
and fabric from which a ligature can be fashioned, as there
are several structures from which a ligature can be suspended. Suicide prevention in such cases is thus a greater
challenge, and it may be that it should focus on identifying
and addressing mental health symptoms or other etiological
factors.
For both women and men, being on psychiatric medications was a significant predictor of suicide by drug poisoning. Because we did not have information from the
prescribing doctor, we could not ascertain if the prescribed
drugs were those that killed the decedent. However, even if
we had these data, we would not be able to distill whether
taking psychotropic drugs elevates the risk of suicidal
ideation and subsequent action, as discussed by Pomerantz
[52], or if the prescription drug simply provided a means to
an end. Although recent research tends to find that antidepressants reduces the risk of suicidal ideation [5355], it
may also provide a greater risk for suicide by these medications. Future research should attempt to disentangle
these risks [56]. Regardless, mental health professionals
already recognize the important role they can play in
working with primary care physicians and psychiatrists
who prescribe psychotropic medications [57].
Given the relationship between being on psychiatric
medication and committing suicide with poison, at least
some preventive implications are clear. Those living with,
or caring for individuals on psychiatric medications should
consider restricting access to the available supply. This is
routinely used effectively in institutional settings and, like
firearms, could help restrict access to easy means. Likewise, other available poisons such as cleaning solvents
could be locked away until needed. Although such measures may seem extreme, suicide-proofing a household,
much like child-proofing, may save lives. This is particularly true for women, who are much more likely than

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

men to have been prescribed medications for psychiatric

problems [36], although that was not the case for the study
sample.
Prior suicide attempts emerged as a puzzling variable in
these analyses. Both men and women who used firearms to
suicide were significantly less likely to have had prior
suicide attempts. We also found the opposite to be true of
suicide by hanging: prior attempts increased the odds of
using this method of suicide. Exactly why this occurs
cannot be discerned from our data, but there are a couple of
possible explanations that future studies could explore. It
could be that attempters who are not resolute about taking
their lives avoid the use of firearms, a highly lethal method.
Hanging might be considered by those who attempt suicide
multiple times as a method from which they might survive,
as they have in the past. Conversely, those who are
extremely serious about ending their lives undertake a
single attempt that employs lethal means such as firearms.
Interviews with attempters could reveal the reasons for
their choice of method, and detailed psychological autopsies might shed light on such rationales for completers.
Being married and committing the suicide at home
increased the odds of firearm suicide among men, but these
factors had no influence on female firearm suicides. Among
women, the odds of committing firearm suicide were lower
for those who lived alone, but living situation had no
influence on firearm suicides among men. With respect to
hanging, unmarried men were much more likely to use this
method than married men were, but there was no relationship between marital status and suicide by hanging
among women. Finally, men who had a history of substance abuse were somewhat more likely to suicide by drug
poisoning than men without substance abuse histories;
there was no relationship between substance abuse history
and drug poisoning among women.
Other variables might help explain gender differentials
in suicide that were not part of this analysis. Physiological
factors, for example, could dictate in part the choice of
method. Specifically, if women are more likely to have
shorter arms, and long arms are a prerequisite for pulling
the trigger on a shotgun or rifle, then it follows that women
may be more likely to choose handguns. Much the same
argument can be made for caliber of weapons used to
commit suicide. Is it possible, for instance, that women are
more likely to use smaller caliber handguns (e.g., .22 or
.25), whereas men seem to favor larger calibers (e.g., 9 mm
or 0.357)? Data such as those available through the
National Violent Death Reporting System may facilitate
the examination of such questions.
The social learning of suicide through popular culture
may also play a role in the gender differential. In films,
men are far more likely to use firearms than are women
[58]. Alternatively, women are frequently depicted

867

poisoning themselves. Inasmuch as social behavior mimics

cinematic behavior, we perhaps should not be surprised at
such differentials.
There are a number of limitations of this study that
should be noted. The data cover just 10 years of suicide in
an urban county located in the North Central region of the
United States, which do not permit us to make inferences
regarding suicides in other geographic regions or to time
periods outside 19972006. If we think that changing
gender roles over time could account for greater use of
firearms by women committing suicide, then it would be
instructive to examine data covering several decades. Such
a longitudinal look at gender and suicide method might
help identify certain sociological variables, such as labor
force participation, related to changes in female firearm
suicides.
As rich as the data from the medical examiner were,
they presented limitations. For one, these did not include
certain types of information that could have theoretical or
practical importance. For example, the documents we used
did not consistently record who owned the firearm or
exactly where and how it was stored, precluding analyses
with this information. Future studies should attempt to
ascertain the specifics of ownership, possession, and storage because limiting the accessibility of firearms is
important [39, 59]. This is particularly true in the case for
older men. It has been argued that removing firearms from
the households of older adults would prevent suicides [30].
There is research, which suggests that the elimination of
first-choice method may not necessarily lead suicidal
individuals to substitute a second method [49, 50]. If a
firearm is not easily accessible when the individual decides
to commit suicide, the lack of preferred method could
result in saving a life.
It must be noted that the analysis of suicide by drug
poisoning is particularly problematic. Classification of a
drug poisoning death as suicide, accidental, or undetermined is controversial inasmuch as it could be a social
construction, which perpetuates gendered explanations. For
example, recent research suggests potential misclassification of suicides may account for apparent racial and
ethnic gaps in suicide [60]. Gender differences, therefore,
might attenuate or disappear if the true cause of death is
known.
Finally, the data regarding psychological problems such
as substance abuse and depression must be interpreted with
caution since they are not clinical diagnoses but information gleaned from family and friends. While this may
introduce error into our measures, we argue that much
gained from analyses that use such data. These data allow
us to examine those labeled as substance abusers or suffering from depression by lay people, which can add to the
literature. Studies of mental health problems and suicide

123

868

completers have typically used samples of clinically

diagnosed individuals, e.g. [61, 62]. Thus, studies such as
ours may shed light on individuals who appear to have
mental health problems that did not seek treatment.
Moreover, this approach is also justified given the gender,
race, class, and age differentials in seeking treatment for
mental health problems [63, 64].
A richer picture of the interrelationship among gender,
suicide method and other variables could be painted with
data such as those collected through psychological
autopsies. Despite the challenges inherent in an approach
that relies on contacting survivors and other informants
[65, 66], it is possible that these kinds of data would
permit the analyst to tease out subtleties not discernable
with death certificate or medical examiner data alone. For
example, it may be that mental health diagnosis and
treatment history figure significantly in the interplay
between gender and selection of suicide method. Alternatively, it is possible that romantic relationship status,
again not part of typical archival data sources on suicide,
plays a role in the selection of suicide method. All these
questions argue for the concatenation of multiple, individual-level datasets in order to expand the analytical
possibilities and develop a fuller understanding of
suicide.
Finally, this study also illustrates the importance of
examining gender differences, not merely controlling for
gender. We found several variables that operated differently for men and women; treating gender as a control
variable masked these differences and often rendered other
factors insignificant. This is of particular importance for
understanding womens suicide behaviors. Since women
are far less likely to commit suicide than men are, data sets
of completed suicides will have far more men than women.
Thus, analyses that merely control for gender will more
likely reflect predictors that are relevant for men, but may
not be for women, and could lead to erroneous prevention
strategies and policies. Clearly, suicide research would
greatly benefit from unpacking the relationships between
suicide behavior and gender.
This study reinforces not only the importance of replication in science, but also the role of social science in the
service of suicide prevention and education. It is through
this approach that we will get closer to research findings
that help reduce the incidence and prevalence of suicide.

References
1. Centers for Disease Control and Prevention (2009) Suicide
fact sheet http://www.cdc.gov/violenceprevention/pdf/SuicideFactSheet-a.pdf. Accessed 8 Dec 2009

123

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

2. Krug EG, Dahlberg LL, Mercy JA, Zwi AB, Lozano R (2002)
World report on violence and health. World Health Organization,
Geneva
3. Moscicki E (1994) Gender differences in completed and
attempted suicides. Ann Epidemiol 4:152158
4. Cannetto SS, Lester D (1995) The epidemiology of womens
suicidal behavior. In: Canetto SS, Lester D (eds) Women and
suicidal behavior. Springer, New York, pp 3557
5. Romero MP, Wintemute GJ (2002) The epidemiology of firearm
suicide in the United States. J Urban Health 79:3948
6. Tadros G, Salib E (2000) Age and methods in fatal self harm
(FSH). Is there a link? Int J Geriatr Psychiatry 15:848852
7. Cantor C, Neulinger K (2000) The epidemiology of suicide and
attempted suicide among young Australians. Aust N Z J Psychiatry 34:370387
8. Stack S (2000) Suicide: a 15-year review of the sociological
literature. Part I: cultural and economic factors. Suicide Life
Threat Behav 30:145162
9. Marks MS, Geling O (1999) Sociodemographic and geographic
patterns of firearm suicide in the United States, 19891993.
Health Place 5:179185
10. Kosky RJ, Dundas P (2000) Death by hanging: implications for
prevention of an important method of youth suicide. Aust N Z J
Psychiatry 34:836841
11. Denning DG, Conwell Y, King D, Cox C (2000) Method choice,
intent and gender in completed suicide. Suicide Life Threat
Behav 30:282288
12. Beck AT, Schuyler D, Herman I (1974) Development of suicidal
intent scales. In: Beck AT, Resnick HLP, Letteri DJ (eds) The
prediction of suicide. Charles Press, Bowie, pp 4546
13. Haw C, Hawton K, Houston K, Townsend E (2003) Correlates of
relative lethality and suicidal intent among deliberate self-harm
patients. Suicide Life Threat Behav 33:353364
14. Nock MK, Kessler RC (2006) Prevalence of and risk factors for
suicide attempts versus suicide gestures: analysis of the National
Comorbidity Survey. J Abnorm Psychol 115:616623
15. Townsend E, Hawton K, Harriss L, Bale E, Bond A (2001)
Substances used in deliberate self-poisoning 19851997: trends
and associations with age, gender, repetition and suicide intent.
Soc Psychiatry Psychiatr Epidemiol 36:228234
16. Cannetto SS, Sakinofsky I (1998) The gender paradox in suicide.
Suicide Life Threat Behav 28:123
17. McIntosh JL, Santos JF (1982) Changing patterns in methods of
suicide by race and sex. Suicide Life Threat Behav 12:221233
18. Smith TW, Smith RJ (1995) Changes in firearm ownership
among women, 19801994. J Crim Law Criminol 86:133149
19. Kellermann AL, Rivara FP, Somes G, Reay DT, Francisco J,
Banton JG, Prodzinski J, Fligner C, Hackman BB (1992) Suicide
in the home in relation to gun ownership. N Engl J Med
327:467472
20. Wintemute GJ, Parham CA, Beaumont JJ, Wright M, Drake C
(1999) Mortality among recent purchasers of handguns. N Engl J
Med 341:15831589
21. Lester D (2000) Why people kill themselves. Charles Thomas,
Springfield
22. Stack S, Wasserman I (2009) Gender and suicide risk: the role of
wound site. Suicide Life Threat Behav 39:1320
23. Schmeling A, Strauch H, Rothschild MA (2001) Female suicides
in Berlin with the use of firearms. Forensic Sci Int 124:178181
24. Kaplan AG, Klein RB (1989) Women and suicide. In: Jacobs D,
Brown HN (eds) Suicide: understanding and responding: Harvard
Medical School perspectives. International Universities Press,
Madison, pp 257282
25. Barber ME, Marzuk PM, Leon AC, Portera L (1998) Aborted
suicide attempts: a new classification of suicidal behavior. Am J
Psychiatry 155:385389

Soc Psychiatry Psychiatr Epidemiol (2012) 47:857869

26. Rich CL, Ricketts JE, Fowler RC, Young D (1988) Some differences between men and women who commit suicide. Am J
Psychiatry 145:718722
27. Steer RA, Beck AT, Garrison B, Lester D (1988) Eventual suicide in interrupted and uninterrupted attempters: a challenge to
the cry-for-help hypothesis. Suicide Life Threat Behav
18:119128
28. Kessler RC, Borges G, Walters EE (1999) Prevalence of and risk
factors for lifetime suicide attempts in the National Comorbidity
Survey. Arch Gen Psychiatry 56:617626
29. Skogman K, Alsen M, Ojehagan A (2004) Sex differences in risk
factors for suicide after attempted suicide: a follow-up study of
1052 suicide attempters. Soc Psychiatry Psychiatr Epidemiol
39:113120
30. Conwell Y, Brent D (1995) Suicide and aging: I. Patterns of
psychiatric diagnosis. Int Psychogeriatr 7:149181
31. Grava G, Ceroni GB, Rucci P, Scudellari P (2006) Suicidal
behavior and personality structure. Suicide Life Threat Behav
36:569577
32. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M,
Eschleman S et al (1994) Lifetime and 12-month prevalence of
DSM-III-R psychiatric disorders in the United States: results
from a national comorbidity survey. Arch Gen Psychiatry
51:819
33. Jick SS, Dean AD, Jick H (1995) Antidepressants and suicide. Br
Med J 310:315318
34. Cyranowski JM, Frank E, Young E, Shear K (2000) Adolescent
onset of the of the gender difference in lifetime rates of major
depression. Arch Gen Psychiatry 57:2127
35. Hankin BL, Abramson LY (2001) Development of gender differences in depression: An elaborated cognitive vulnerabilitytransactional stress theory. Psychol Bull 127:773796
36. Norman J (2004) Gender bias in the diagnosis and treatment of
depression. Int J Ment Health 33:3243
37. Henrikksson MM, Hellivi MA, Marttunen MJ, Heikkinnen ME,
Isometsa ET, Kuoppasalmi KI, Lonnqvist JK (1993) Mental
disorders and comorbidity in suicide. Am J Psychiatry
150:935940
38. Martunnen MJ, Henriksson MM, Hillevi MA, Heikkinen ME,
Isometsa ET, Lonnqvist JK (1995) Suicide among female youth:
Characteristics and comparisons with males in age groups 13 to
22 years. J Am Acad Child Adolesc Psychiatry 34:12971307
39. Kposowa AJ, McElvain JP (2006) Gender, place, and method of
suicide. Soc Psychiatry Psychiatr Epidemiol 41:435443
40. Kaplan MS, Adamek ME, Geling O, Calderon A (1997) Firearm
suicide among older women in the U.S. Soc Sci Med
44:14271430
41. Centers for Disease Control and Prevention (2009) WISQARS
injury mortality reports. http://www.cdc.gov/injury/wisqars/.
Accessed 3 Oct 2010
42. Rawlinson C, Crews P (2003) Access to quality health services in
rural areas. Emergency medical services. Rural Healthy People
2010: a companion document to healthy people 2010. The Texas
A&M University System Health Science Center, School of Rural
Public Health, Southwest Rural Health Research Center, College
Station
43. Menard SW (2002) Applied logistic regression analysis, 2nd edn.
Sage, Thousand Oaks
44. Altman DG, Bland JM (2003) Interaction revisited: the difference
between two estimates. Br Med J 326:219
45. Kaplan MS, Geling O (1998) Firearm suicides and homicides in
the United States: regional variations and patterns of gun ownership. Soc Sci Med 46:12271233

869
46. Conwell Y, Duberstein PR, Connor K, Eberly S, Cox C, Caine
ED (2002) Access to firearms and risk for suicide in middle-aged
and older adults. Am J Geriatr Psychiatry 10:407416
47. Dahlberg LL, Ikeda RM, Kresnow M (2004) Guns in the home
and risk of a violent death in the home: Findings from a national
survey. Am J Epidemiol 160:929936
48. Kellermann AL, Somes G, Rivara FP, Lee RK, Vanton JG (1998)
Injuries and deaths due to firearms in the home. J Trauma
45:263267
49. Kreitman N, Platt S (1984) Suicide, unemployment, and domestic
gas detoxification in Britain. J Epidemiol Community Health
38:16
50. Lester D (1990) The effects of detoxification of domestic gas on
suicide in the United States. Am J Public Health 80:12
51. Daigle MS (2005) Suicide prevention through means restriction:
assessing the risk of substitution. A critical review and synthesis.
Accid Anal Prev 37:625632
52. Pomerantz JM (2008) FDA warning about suicidality: Balancing
risk and benefit. Drug Benefit Trends 20:149150
53. Brent D (2007) Antidepressant and suicidal behavior: cause or
cure? Am J Psychiatry 164:989991
54. Isacsson G, Bergman U, Rich CL (1996) Epidemiological data
suggest antidepressant reduce suicide risk among depressives.
J Affect Disord 41:18
55. Spijker J, de Graaf R, ten Have M, Nolen WA, Speckens A
(2010) Predictors of suicidality in depressive spectrum disorders
in the general population: Results of the Netherlands Mental
Health Survey and Incidence Study. Soc Psychiatry Psychiatr
Epidemiol 45:513521
56. Donovan S, Kelleher MJ, Lambourn J, Foster T (1999) The
occurrence of suicide following the prescription of antidepressant
drugs. Arch Suicide Res 5:181192
57. Mikula L, Bronson A, Reitman D (2007) Do antidepressant
medications really increase suicide risk for adolescents? Reflections on the FDAs black box warning. Behav Ther 30:99103
58. Stack S, Bowman B (2009) Suicide in movies: gender and choice
of suicide method. In: Stack S, Lester D (eds) Suicide and the
creative arts. Nova Science Publishers, New York, pp 5761
59. Oslin DW, Zubritsky C, Brown G, Mullahy M, Puliafico A, Ten
Have T (2004) Managing suicide risk late in life: access to firearms as a public health risk. Am J Geriatr Psychiatry 12:3036
60. Rockett IRH, Wang S, Stack S, De Leo D, Frost JL, Ducatman
AM, Walker RL, Kapusta ND (2010) Race/ethnicity and potential
suicide misclassification: window on a minority suicide paradox?
BMC Psychiatry 10:18
61. DeJong T, Overholser JC, Stockmeier CA (2010) Apples to
oranges? A direct comparison between suicide attempters and
suicide completers. J Affect Disord 124:9097
62. Bradvick L (2007) Violent and nonviolent methods of suicide:
different patterns may be found in men and women with sever
depression. Arch Suicide Res 11:255264
63. Sen B (2004) Adolescent propensity for depressed mood and help
seeking: race and gender differences. J Ment Health Policy Econ
7:133145
64. Sussman L, Robins L, Earls F (1987) Treatment-seeking for
depression by black and white Americans. Soc Sci Med
24:187196
65. Hawton K, Appleby L, Platt S, Foster T, Cooper J, Malmberg A,
Simkin S (1998) The psychological autopsy approach to studying
suicide: a review of methodological issues. J Affect Disord
50:269276
66. Pouliot L, De Leo D (2006) Critical issues in psychological
autopsy studies. Suicide Life Threat Behav 36:491510

123

Copyright of Social Psychiatry & Psychiatric Epidemiology is the property of Springer Science & Business
Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email articles for
individual use.