Download Childhood physical and sexual abuse and lifetime number of suicide

ARTICLE IN PRESS
Behaviour Research and Therapy 45 (2007) 539–547
www.elsevier.com/locate/brat
Childhood physical and sexual abuse and lifetime number
of suicide attempts: A persistent and theoretically
important relationship
Thomas E. Joiner Jr., Natalie J. Sachs-Ericsson, LaRicka R. Wingate,
Jessica S. Brown, Michael D. Anestis, Edward A. Selby
Department of Psychology, Florida State University, Tallahassee, FL 32306-1270, USA
Received 16 November 2005; received in revised form 11 April 2006; accepted 26 April 2006
Abstract
Background: Research to date has indicated that childhood abuse is associated with suicide, though little research has
examined the unique contribution of specific types of abuse to suicidal behavior. We predict that childhood physical and
violent sexual abuse will have a greater effect on suicide attempts than molestation and verbal abuse.
Methods: The National Comorbidity Survey data were used to test these predictions while controlling for a number of
psychiatric and psychosocial variables.
Results: As expected, childhood physical and violent sexual abuse showed similar effects on lifetime suicide attempts,
which were stronger than the effects of molestation and verbal abuse.
Limitations: This was a cross-sectional, retrospective study, so true causality cannot be shown. Some measurement
limitations exist. Additionally, effect sizes were small but still significant.
Conclusions: While all forms of childhood abuse are troubling and create risk for future psychopathology and suicidality,
the present study indicates that childhood physical and violent sexual abuse should be seen as greater risk factors for future
suicide attempts than molestation and verbal abuse.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Physical abuse; Sexual abuse; Suicide; Family history
Introduction
Abuse during childhood can result in a number of difficulties later in life, including increased risk for death
by suicide. Previous research has demonstrated that a connection between childhood abuse and suicide exists
(Glowinski et al., 2001; Roy, 2003, etc.), but few studies have examined the effects of different forms of
childhood abuse in relation to later suicide attempts. Although, any form of abuse has the potential to have
deleterious effects on an individual, it would seem that more severe forms of childhood abuse would infer
Corresponding author.
E-mail address: [email protected] (T.E. Joiner Jr.).
0005-7967/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.brat.2006.04.007
ARTICLE IN PRESS
540
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
greater risk for suicide upon those victims due to psychological and environmental consequences surrounding
such severe abuse relative to less severe forms of abuse.
Past work has examined the link between childhood physical abuse and suicidal behavior, and most but not
all affirm the connection. For example, when examining families of 67 adolescent victims of suicide in
comparison to 67 demographically matched community controls. Brent et al. (1994) found that the suicide
victims had higher rates of childhood physical abuse than the controls. Glowinski et al. (2001) interviewed
over 3000 female adolescent twins, and found that childhood physical abuse was one of the factors most
associated with a suicide attempt history. Roy (2003) studied 280 people with alcohol use disorders, and found
that childhood physical abuse was associated with lifetime suicide attempts. McHolm, MacMillan, and
Jamieson (2003) reported similar results in depressed women, as did Anderson, Tiro, Price, Bender, and
Kaslow (2002) in low-income African-American women. Kaplan, Pelcovitz, Salzinger, Mandel, and Weiner
(1997) did not find a difference in suicide histories between 99 abused and 99 non-abused white, middle class,
suburban adolescents, though victims of intrafamilial sexual abuse were excluded from participating in the
study.
Considerably less work has evaluated the unique contribution of a particular form of abuse, controlling for
other forms, to suicidality. Among adults with alcohol use disorders, Roy (2003) reported that different forms
of abuse each related to lifetime suicide attempts, with similar effect sizes, but the forms of abuse were not
controlled for each other. Anderson et al. (2002) reported a similar pattern among low-income AfricanAmerican women. By contrast, Soloff, Lynch, and Kelly (2002) studied women with borderline personality
disorder, and found that childhood sexual abuse but not physical abuse was related to adult suicide attempts.
Brown, Cohen, Johnson, and Smailes (1999) reported that various forms of childhood abuse were linked to
later suicidal behavior among community participants, but that childhood sexual abuse showed the strongest
independent link.
Controlling for the effects of forms of abuse on each other in predicting outcomes is important for at least
two reasons. First, and more generally, it allows specificity regarding which form is most linked to outcomes.
Forms of abuse often occur together as well as with other negative childhood events. For example, Felitti and
colleagues (1998) found that when respondents experienced one childhood adversity, the probability of having
experienced another was approximately 80%. Kessler (2002) argued that researchers often look at one
childhood adversity, such as sexual abuse, and assume subsequent problems are related to that adversity,
ignoring the possibility that sexual abuse often occurs within a cluster of other childhood difficulties that may
themselves account for the subsequent problems. Similarly, Mullen, Martin, Anderson, Romans, and
Herbison (1993) argued that childhood abuse may be just one of several elements that occur within a context
of family problems, and these other elements may account for the relationship between abuse and negative life
outcomes. Family characteristics associated with childhood abuse include parental psychopathology, family
conflict, low social economic status, parental loss or absence, and parental divorce (Felitti et al., 1998; Fleming
et al., 1997; Kenny & McEachern, 2000; Molnar et al., 2001; Romans et al., 1995; Sidebotham & Golding,
2001; Zuravin & Fontanella, 1999). In order to understand the implications of various forms of abuse for
suicide, it is important to try to distinguish their influence from other co-occurring adversities.
A related and particularly relevant reason to control for the effects of forms of abuse on each other in
predicting outcomes is that it allows tests of important facets of theories that make differential predictions
about forms of abuse and their associations to outcomes. According to Joiner’s (2005) theory of suicidal
behavior, few people want to die by suicide, but also, and perhaps more importantly, few people can. The
central idea of Joiner’s theory is that serious suicidal behavior requires each of three interpersonalpsychological precursors: (1) the acquired capability, through habituation to pain and fear, to enact lethal selfinjury; (2) the sense that one is a burden on loved ones; and (3) the sense that one does not belong to or is not
connected with a valued group or relationship. This theory is compatible with and adds to other prominent
models of suicidality (e.g., Baumeister, 1990; Beck, 1996; Linehan, 1993). Difficult life events such as
childhood abuse, especially when severe, may have the potential to be extremely painful and fear-inducing.
According to Joiner’s model, then, more severe and painful forms of childhood abuse should be greater risk
factors for suicide than less painful forms of abuse.
The aquired capability to enact lethal self-injury portion of Joiner’s model posits that those who, through
various painful experiences, have habituated to pain and fear are the only ones capable of completing
ARTICLE IN PRESS
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
541
suicide—anyone else is unable to do so, even if they desire death. According to this theory, repeated exposure
to pain and provocation may cause habituation to the ‘‘taboo’’ and prohibited quality of suicidal behavior,
thus diminishing the fear and pain associated with self-harm. A number of studies support the relationship
between repeated exposure to injury/pain and suicidality (e.g., Dhossche, Snell, & Larder, 2000; Menninger,
1936; Phillips, McElroy, Keck, & Pope, 1993; Rosenthal, Rinzler, Wallsh, & Klausner, 1972; Veale, Boocock,
Gournay, & Dryden, 1996). In this framework, repeated painful experiences may result in the ability to enact
future lethal self-injury; childhood physical and violent sexual abuse may constitute pathways by which this
occurs.
Childhood physical abuse and certain forms of childhood sexual abuse may be more closely linked to
acquisition of lethality than other forms of abuse (i.e., neglect, verbal abuse), because they are, on average,
more physically painful than the other forms of abuse. Regarding childhood sexual abuse, there is
evidence that more painful forms (e.g., severe forced abuse) are more associated with suicidality
than less painful forms (Mullen et al., 1993; Stepakoff, 1998). According to Joiner’s model, habituation to
pain and fear combines with desire for death to result in serious suicidal behavior, and that desire for
death stems from feeling a burden on loved ones and others, and feeling disconnected and alienated from
others.
To the degree that any form of abuse facilitates either lethality (through habituation to pain and
provocation) or desire for death (through increased feelings of burdensomeness or disconnection),
Joiner’s (2005) model would view it as a risk for later suicidal behavior. However, we predict the strongest
effects of all for childhood physical and violent sexual abuse, not only because they are physically painful,
but also because they imply burdensomeness and disconnection. They therefore potentially increase risk
through every mechanism specified in the model. Although habituation, burdensomeness, and disconnection
are not directly measured within this sample, we nonetheless anticipate a direct relationship between the
exposure to a particularly physically painful experience—physical abuse and/or violent sexual abuse—and
number of suicide attempts. Regarding molestation and verbal abuse, we predict weaker effects. Fergusson
(1996, 2000) measured both violent and non-violent childhood sexual abuse and found both forms to be
related to later suicidal behavior; however, he did not distinguish between forms of abuse that contributed
directly to failed suicide attempts versus those that preceded a successful completion. Joiner’s (2005) model
contends that, while the non-violent forms of sexual abuse likely contribute to future suicidal behavior by
causing or intensifying perceived burdensomeness and/or thwarted belongingness, their failure to expose the
victim to physical pain makes them contributory, but insufficient causes of completed suicide. The victims of
non-violent sexual abuse are thus at a higher risk for a non-lethal suicide attempt rather than for completed
suicide. After this first attempt or any other exposure to physically painful or provocative experiences,
however, they will become more habituated to physical pain and their risk for death by suicide will
increase dramatically. In light of prior findings, we thus expect to find that, even when controlling for a host
of established risk factors for suicidal behavior, violent childhood physical and sexual abuse will serve
as significantly stronger predictors for the number of lifetime suicide attempts than will molestation and
verbal abuse.
Method
The current study draws on the National Comorbidity Survey (NCS), which is a nationwide epidemiological
study designed to assess the prevalence and psychosocial correlates of DSM-III-R psychiatric disorders (for
more information on the NCS see Kessler et al., 1997). The present study will examine those with complete
data on all study variables (N for this study ¼ 5838).
Sampling and weighting
The participants were selected through a multistage area probability sample based on household, within a
stratified sample of counties within the United States. Response rate was 82.6%. More detail on the sampling
and weighting can be found in Kessler (1994) and Kessler et al. (1994).
ARTICLE IN PRESS
542
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
Participants
Participants were interviewed in their home and informed consent was obtained from each participant.
There was an equal distribution of men ðN ¼ 2938Þ and women ðN ¼ 2939Þ within the sample. The average
age of the participants was 33.2 years (SD ¼ 10.70). Ethnicity was as follows: 75.6% Caucasian, 11.6%
African-American, 9.4% Hispanic, and 3.4% categorized as ‘‘Other.’’
Measures
Lifetime number of suicide attempts
NCS participants were asked whether they had ever attempted suicide, and if so, how many times. Those
who stated they had never attempted suicide were coded ‘‘0’’; all others were coded with their lifetime number
of attempts.
Childhood experiences of abuse
Participants were provided with a list of negative events which included a question about childhood physical
abuse. Participants who reported that they had been physically abused as a child were coded ‘‘1’’ and those
who reported that they had not been physically abused as a child were coded ‘‘0’’.
The list of traumatic events included questions about being raped (i.e., ‘‘someone had sexual intercourse
with you when you did not want to by threatening you or using some degree of force’’). If the participant
responded positively to the question, the participant was then asked the age at which the event first occurred.
Those participants who indicated that the rape occurred before age 15, were coded ‘‘1’’, and all other
participants were coded, ‘‘0’’.
Participants also were asked specifically about molestation (i.e., ‘‘someone touched or felt your genitals
when you did not want them to’’). If the participant responded positively the participant was then asked the
age at which the event first occurred. Those participants who indicated that they had been molested before the
age of 15 were coded ‘‘1’’ and all other participants were coded ‘‘0’’.
In a subsequent section of the NCS survey participants were asked several questions about their early family
life including questions related to experiences of childhood verbal abuse. Participants were handed a list of
specific behaviors related to verbal abuse, and asked how often any of the things on the list happened to them
when they were growing up. The list included insults, swearing, and doing or saying something spiteful.
Participants responded about the frequency of these experiences using a 4-point Likert-type scale anchored by
1 ‘‘Often’’ and 4 ‘‘Never’’. Participants who indicated that they ‘‘often’’ experienced verbal abuse by a parent
or stepparent, were coded ‘‘1’’, and all other participants were coded ‘‘0’’.
It should be noted that the assessments of physical abuse, sexual abuse, molestation, and verbal abuse vary
in terms of their depth, specificity, and quality. Physical abuse was assessed with one question, as were sexual
abuse and molestation, whereas verbal abuse was assessed by several different items. The measures on sexual
abuse and molestation were quite specific, whereas the measure on physical abuse was more general.
It should also be noted that each of the four abuse variables was skewed, as would be expected. A total of
1.6% of the sample reported sexual abuse; 4.2% reported physical abuse; 5.9% reported molestation; and
8.4% reported verbal abuse. Here again, the indices of physical abuse and sexual abuse were the least
advantaged in terms of distributional properties; therefore, skew would not be a convincing explanation for
any supportive findings for physical abuse and sexual abuse. Intercorrelations between the four abuse
measures ranged from 0.10 to 0.26 (p’so0.01); multicollinearity is thus unlikely to affect analyses.
Participants’ psychiatric diagnoses
The Composite International Diagnostic Interview (CIDI, World Health Organization, 1990), a semistructured interview, was used to assess the participant’s lifetime DSM-III-R psychiatric diagnoses. The
reliability and validity of the CIDI have been established (Wittchen, 1994).
ARTICLE IN PRESS
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
543
Family of origin variables
Several variables were identified as potential correlates of childhood abuse. Participants were asked to
indicate if, before the age of 15, they had experienced the divorce of their parents (1 ¼ yes; 0 ¼ no) or the death
of either parent (1 ¼ either parent died; 0 ¼ neither parent died). Moreover, family conflict was assessed on a
scale from 1 (a lot) to 4 (none). Participants were also asked to compare their family’s income to the average
family in their community at the time they were growing up, on a scale from 5 (better off) to 1 (a lot worse off).
Family history of psychiatric symptoms
Participants were asked about psychiatric symptoms of each of their parents, within each of the following
domains: suicide attempts, depression, mania, anxiety, substance use, and antisocial personality disorder. The
items for each domain were based on Family History Research Diagnostic Criteria (Andreasen et al., 1977).
Rather than exclude from the analyses those participants who had no knowledge of their mother’s
psychiatric symptoms ðn ¼ 364Þ or father’s psychiatric symptoms ðn ¼ 894Þ, we created two additional
dichotomous variables: ‘‘Knowledge of Mother’s symptoms’’ and ‘‘Knowledge of Father’s symptoms’’.
Participants who could not recall a parent’s symptoms were coded ‘‘0’’, those who could recall were coded ‘‘1’’.
Data-analytic strategy
We were interested in the association between forms of childhood abuse and lifetime number of suicide
attempts, when respondents’ age and gender, psychiatric history, family psychiatric history, and family of
origin variables were controlled. The four forms of childhood abuse (i.e., physical, sexual, molestation, verbal)
and all of the covariates were simultaneously entered into a regression equation predicting lifetime number of
suicide attempts. This approach allows evaluation of the specific association between a given form of abuse
and lifetime suicide attempts, controlling for a host of key covariates as well as for other forms of abuse.
Results
Table 1 shows the results of the regression equation predicting lifetime number of suicide attempts, with
respondents’ age and gender, psychiatric history, family psychiatric history, family of origin variables, and the
four forms of childhood abuse (i.e., physical, sexual, molestation, verbal) entered as predictors. Before
addressing results relevant to our main predictions, we first note other findings of interest. Both age and
gender were associated with lifetime suicide attempts, such that younger people and women reported more
attempts. Most disorders were uniquely associated with lifetime suicide attempts. There were relatively few
effects for the family variables; exceptions were father’s anxiety (more anxiety related to fewer attempts),
father’s antisocial personality disorder (more antisocial behavior related to more attempts), and father’s
suicide attempts (more paternal attempts related to more attempts for participants).
In total, 272 individuals responded that they had made at least one suicide attempt. The range in number of
suicide attempts for those who indicated that they had, in fact, made at least one attempts was 1 to 10
(mean ¼ 1.57; SD ¼ 1.102). A total of 190 individuals reported a total of one suicide attempts, while 38
reported two attempts, 24 reported three attempts, 13 reported four attempts, five reported five attempts, and
one reported ten attempts.
The effects for childhood physical abuse (partial correlation ¼ 0.074, t ¼ 5:98, po0:05) and sexual abuse
(partial correlation ¼ 0.065, t ¼ 5:40, po0:05) were relatively pronounced, similar to one another, and
exceeded effects for molestation (partial correlation ¼ 0.033, t ¼ 2:88, po0:05) and for verbal abuse (partial
correlation ¼ 0.017, t ¼ 1:68, p ¼ ns). Treating these correlations as independent (though they are not),1 and
applying the z-test for the difference between independent correlations (see Cohen & Cohen, 1983, pp. 53–54),
these correlations appear to significantly differ from one another (z’s48.64, po0:05).2
1
This test was developed for use with independent, zero-order correlations; because we used partial correlations, these comparisons
should be viewed as reasonable estimates (see Cohen & Cohen, 1983, p. 53).
2
In an additional analysis, rape after age 15 was added as a covariate in an effort to explore whether the differential effects of various
forms of childhood abuse remained significant when controlling for the effects of abuse later in life. The independent variables retained
ARTICLE IN PRESS
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
544
Table 1
Respondent variables, psychiatric history, family psychiatric history, family of origin variables, childhood abuse, and parental history of
suicide attempts predicting respondent suicide attempts
Order of
entry of
set
Predictors in set
F for set
1
Respondent’s age and gender
Age
Gender
19.20*
Respondent’s psychiatric history
Alcohol dependence
Substance dependence
Major depression
Dysthymia
Bipolar disorder
Generalized anxiety disorder
Simple phobia
Panic disorder
Agoraphobia without panic disorder
Post-traumatic stress disorder
Adult antisocial behavior
Number of depressive episodes
57.54*
Family psychiatric history
Knowledge of father’s symptoms
Knowledge of mother’s symptoms
Maternal depression
Maternal anxiety
Maternal substance use
Maternal adult antisocial behavior
Paternal depression
Paternal anxiety
Paternal substance use
Paternal adult antisocial behavior
35.99*
Family of origin variables
Parents divorced before age 15
Death of a parent
Parental conflict
Income
31.03*
Parental history of suicide attempts
Paternal attempt
Maternal attempt
29.92*
Childhood experiences of abuse
Verbal abuse
Rape before age 15
Molestation before age 15
Physical abuse
29.61*
2
3
4
5
6
df
Partial
correlation
Model R2
2,5835
5835
5835
0.08
0.05
0.07
0.01
3.55*
5.15*
14,5823
5823
5823
5823
5823
5823
5823
5823
5823
5823
5823
5823
5823
0.34
0.04
0.07
0.08
0.02
0.15
0.03
0.02
0.04
0.02
0.15
0.03
0.04
0.12
3.27*
5.11*
6.21*
1.32
11.83*
1.91
1.43
3.33*
1.49
11.39*
2.07*
3.11*
24,5813
5813
5813
5813
5813
5813
5813
5813
5813
5813
5813
0.35
0.03
0.01
0.00
0.01
0.01
0.05
0.00
0.03
0.02
0.04
0.13
1.96
0.64
0.31
0.85
0.45
3.91*
0.23
2.07*
1.28
3.32*
28,5809
5809
5809
5809
5809
0.35
0.01
0.01
0.03
0.01
0.13
0.10
0.93
2.07*
0.51
30,5807
5807
5807
0.37
0.06
0.02
0.13
4.73*
1.64
34,5803
5803
5803
5803
5803
0.37
0.02
0.07
0.04
0.08
0.14
1.68
5.40*
2.88*
5.98*
t for within set
predictors
*po0.05.
Discussion
Few studies have evaluated the unique contribution of a particular form of abuse, controlling for other
forms, to suicidality. In this study, we did so through the differential analysis of violent versus non-violent
(footnote continued)
their significance and remained in the same order of magnitude, however, we believe that abuse after age 15 does not fit into the model we
sought to examine in this study and have therefore chosen to report the results without considering this additional covariate.
ARTICLE IN PRESS
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
545
childhood abuse to later instances of suicidal behavior. Results of the study showed that the association
between childhood physical and sexual abuse to lifetime suicide attempts was persistent in the face of
numerous covariates, each of which are considered strong suicide- and abuse-related variables. Furthermore,
as predicted, childhood physical and violent sexual abuse related more closely to lifetime suicide attempts than
did verbal abuse and molestation. We view these results as supportive of a key aspect of Joiner’s (2005)
theory—that painful and provocative experiences confer increased risk for later suicidal behavior.
The covariates in this study, both demographic variables like age, gender, and family of origin, along with
clinical variables like individual and family psychiatric histories as well as childhood abuse all are among the
most well documented correlates of serious suicidal behavior. Covariance of these variables from the
association between childhood abuse and the lifetime number of suicide attempts represents a rigorous test.
When controlling for these covariates we found that the occurrence of serious suicidality, as well as a
significant relationship between childhood abuse and number of suicide attempts remained.
Before discussing further implications of our results, it is important to consider several cautions and
considerations regarding the work. First, our statistical procedures yielded relatively small effect sizes.
However, we statistically controlled for other variables that could possibly account for the relationship
between childhood abuse and number of suicide attempts. The reported effect sizes not only indicate the size
of the relationship between childhood abuse and suicide attempts, but also speak to the size of the relationship
between childhood abuse and suicide attempts above and beyond (or controlling for) the relationship between
suicide attempts and the other variables. This is a stringent test, and for this reason, large effect sizes are not
expected. Note, however, that the relationships found were in fact statistically significant.
Another consideration to note is that the study was cross-sectional and retrospective. Since so few people
attempt, and even fewer die by suicide, it is a difficult topic to study prospectively. Additionally, childhood
sexual abuse is difficult to study prospectively, as the study of children is usually consented for and
reported by parents; if parents are abusing their children, they are less likely to participate in studies and
less likely to report abuse. Because of the cross-sectional design of the study, we are unable to make clear
causal inferences, as we cannot definitively establish a timeline with independent variables occurring
before the measurement of suicidality. Given that data suggest that only a small percentage of children
attempt suicide, and the large majority of suicides occur in adults (McIntosh, 2003), it may be logical to infer
that, in a majority of the cases, the child abuse occurred before the suicide attempt. Another common
concern with the use of retrospective data is the possibility that participants are unable to recall, or incorrectly
recall, information. Additionally, Robins et al. (1985) found that victims of childhood trauma who
managed to experience more positive overall outcomes tended to underreport childhood difficulties. If such
were the case in this sample, it could potentially artificially inflate the differences between childhood abuse
victims who did and did not attempt suicide later in their lives. In this study, however, events that were
assessed were likely to be salient and severe (e.g., sexual abuse, physical abuse, and suicide attempt), which
should minimize reporting biases. Added steps were taken, including preliminary studies, to insure the
accurate collection of data but it is still possible that the data are limited because information was gathered
retrospectively.
Finally, Joiner’s (2005) model emphasizes mechanisms like habituation and opponent processes. It is
important to highlight that these mechanisms were not directly examined in our study, and we eagerly await
future research that directly addresses these issues. In such research, it may be of interest to include measures
relevant to habituation and opponent processes (e.g., fear of death) and to evaluate whether these measures
mediate the association between past abuse and future suicidality. Additionally, analyses that include data on
frequency of physical and violent sexual abuse would provide substantial additional insight into the matter.
A significant strength of our study is that the NCS questionnaire allowed the possibility to assess not only
sexual abuse, but more specifically, to separately evaluate violent as well as less physically painful forms of
abuse (i.e., molestation). Our results showed a distinction such that the more physically painful abuse was
more closely related to number of later suicide attempts than was less painful abuse. Assessment approaches
that combine types of sexual abuse may find weaker effects than the approaches that separately examine
physically painful versus less painful sexual abuse. A suggestion for further study is the continued separate
evaluation of different forms of sexual abuse. In line with the results of this study and Joiner’s (2005) theory of
suicide, we would expect that results would consistently find a stronger relationship between the more
ARTICLE IN PRESS
546
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
physically severe sexual abuse and number of lifetime suicide attempts, and a weaker relationship between less
physically severe sexual abuse and number of lifetime suicide attempts.
An additional strength of our study is that we examined suicidality in terms of the number of actual
occurrences of suicide attempts. Because relatively few people attempt suicide, and even fewer die by suicide,
research on predictive variables is often difficult to conduct. In many instances, the construct of suicidality is
measured by self-report forms assessing suicidal ideation. As has been demonstrated, those who have made
multiple suicide attempts are at higher risk for eventual suicide than others. Our work specifically studied
those people who have attempted suicide and are at higher risk. Future work may address those people who
actually die by suicide. In line with results reported here, we would expect that individuals who suffered
physical or violent sexual abuse as children would be amongst those most likely to fit within that category.
These results may also help to inform clinical work. For clinicians, it is essential to be able to evaluate risk
for suicide in clients in order to select the appropriate course of action. Suicide risk assessment strategies have
already pointed to the importance of placing individuals who have previously made multiple suicide attempts
and have a history of abuse into a category of at least moderate risk (Joiner, Walker, Rudd, & Jobes, 1999).
However, these results allow for a more fine-grained distinction. Multiple attempters who have suffered
physical or violent sexual abuse should be placed into a higher risk category than those who have suffered
verbal abuse or molestation. Therefore, it is important for clinicians to determine not only whether or not a
client has suffered abuse, but also what type of abuse it was in order to evaluate risk for suicide.
A clear understanding of the role of physically painful childhood trauma in later suicide attempts would be
invaluable to both researchers and clinicians, allowing them to gain further insight into the mechanisms
behind what remains a tragic and relatively mysterious phenomenon. We found that the association between
childhood physical and violent sexual abuse to number of lifetime suicide attempts was persistent even when
numerous significant covariates were controlled for. Additionally, we found that the relation of childhood
physical and violent sexual abuse to number of lifetime suicide attempts exceeded the relation between verbal
abuse and molestation to number of lifetime suicide attempts. We look forward to future work that addresses
this general framework relating childhood abuse to suicidal behavior, as well as the specific idea proposed
in Joiner’s (2005) theory—that painful and provocative experiences confer increased risk for later suicidal
behavior.
Acknowledgment
Preparation of this article was supported, in part, by a grant from the John Simon Guggenheim Memorial
Foundation.
References
Andreasen, N. C., Endicott, J., Spitzer, R. L., & Winokur, G. (1977). The family history method using diagnostic criteria: Reliability and
validity. Archives of General Psychiatry, 34(10), 1229–1235.
Anderson, P. L., Tiro, J. A., Price, A. W., Bender, M. A., & Kaslow, N. J. (2002). Additive impact of childhood emotional, physical, and
sexual abuse on suicide attempts among low-income African-American women. Suicide & Life-Threatening Behavior, 32, 131–138.
Baumeister, R. F. (1990). Suicide as escape from self. Psychological Review, 97, 90–110.
Beck, A. T. (1996). Beyond belief: A theory of modes, personality, and psychopathology. In P. M. Salkovskis (Ed.), Frontiers of cognitive
therapy (pp. 1–25). New York: Guilford.
Brent, D. A., Perper, J. A., Moritz, G., Liotus, L., Schweers, J., Balach, L., et al. (1994). Familial risk factors for adolescent suicide:
A case–control study. Acta Psychiatrica Scandinavica, 89, 52–58.
Brown, J., Cohen, P., Johnson, J. G., & Smailes, E. M. (1999). Childhood abuse and neglect: Specificity and effects on adolescent and
young adult depression and suicidality. Journal of the American Academy of Child & Adolescent Psychiatry, 38, 1490–1496.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Erlbaum.
Dhossche, D., Snell, K. S., & Larder, S. (2000). A case–control study of tattoos in young suicide victims as a possible marker of risk.
Journal of Affective Disorders, 59, 165–168.
Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., et al. (1998). Relationship of childhood abuse
and household dysfunction to many of the leading causes of death in adults: The adverse childhood experiences (ACE) study. American
Journal of Preventative Medicine, 14, 245–258.
ARTICLE IN PRESS
T.E. Joiner Jr. et al. / Behaviour Research and Therapy 45 (2007) 539–547
547
Fergusson, D. M., Lynskey, M. T., & Horwood, L. J. (1996). Childhood sexual abuse and psychiatric disorder in young adulthood, II:
Psychiatric outcomes of childhood sexual abuse. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 1365–1374.
Fergusson, D. M., Horwood, L. J., & Woodward, L. J. (2000). The stability of child abuse reports: A longitudinal study of the reporting
behaviour of young adults. Psychological Medicine, 30, 529–544.
Fleming, J., Mullen, P., & Bammer, G. (1997). A study of potential risk factors for sexual abuse in childhood. Child Abuse and Neglect, 21,
49–58.
Glowinski, A. L., Bucholz, K. K., Nelson, E. C., Fu, Q., Madden, P., Reich, W., et al. (2001). Suicide attempts in an adolescent female
twin sample. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 1300–1307.
Joiner, T. (2005). Why people die by suicide. Cambridge, MA: Harvard University Press.
Joiner, T., Walker, R. L., Rudd, M. D., & Jobes, D. A. (1999). Scientizing and routinizing the assessment of suicidality in outpatient
practice. Professional Psychology: Research and Practice, 30, 447–453.
Kaplan, S. J., Pelcovitz, D., Salzinger, S., Mandel, F., & Weiner, M. (1997). Adolescent physical abuse and suicide attempts. Journal of the
American Academy of Child & Adolescent Psychiatry, 36, 799–808.
Kenny, M. C., & McEachern, A. G. (2000). Racial, ethnic, and cultural factors of childhood sexual abuse: A selected review of the
literature. Clinical Psychology Review, 20, 905–922.
Kessler, R. C. (1994). The national comorbidity survey of the United States. International Review of Psychiatry, 6(4), 365–376.
Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., et al. (1994). Lifetime and 12-month prevalence of the DSM-III-R psychiatric
disorders in the United States: Results from the national comorbidity survey. Archives of General Psychiatry, 51(1), 8–19.
Kessler, R. C., Krum, R. M., Warner, L. A., & Nelson, C. B. (1997). Lifetime co-occurrence of DSM-III-R alcohol abuse and dependence
with other psychiatric disorders in the national comorbidity survey. Archives of General Psychiatry, 54(4), 313–321.
Kessler, R. C. (2002). Epidemiology of depression. In I. H. Gotlib, & C. L. Hammen (Eds.), Handbook of depression (pp. 23–42).
New York: Guilford.
Linehan, M. M. (1993). Cognitive-behavioral treatment of borderline personality disorder. New York: Guilford Press.
McHolm, A. E., MacMillan, H. L., & Jamieson, E. (2003). The relationship between childhood physical abuse and sucidality among
depressed women: Results from a community sample. American Journal of Psychiatry, 160, 933–938.
McIntosh, J. L. (2003). American Association of Suicidology. Suicide data page: 2001. Modified September 26, 2003. Available from
www.suicidology.com.
Menninger, K. A. (1936). Purposive accidents as an expression of self-destructive tendencies. International Journal of Psycho-Analysis, 17,
6–16.
Molnar, B. E., Berkman, L. F., & Buka, S. L. (2001). Psychopathology, childhood sexual abuse and other childhood adversities: Relative
links to subsequent suicidal behaviour in the US. Psychological Medicine, 31, 965–977.
Mullen, P. E., Martin, J. L., Anderson, J. C., Romans, S. E., & Herbison, G. P. (1993). Childhood sexual abuse and mental health in adult
life. British Journal of Psychiatry, 163, 721–732.
Phillips, K. A., McElroy, S. L., Keck, P. E., & Pope, H. G. (1993). Body dysmorphic disorder: 30 cases of imagined ugliness. American
Journal of Psychiatry, 150, 302–308.
Robins, L. N., Schoenberg, S. P., Holmes, S. J., Ratcliff, K. S., Benham, A., & Works, J. (1985). Early home environment and
retrospective recall: A Test for concordance between siblings with and without psychiatric disorders. American Journal of
Orthopsychiatry, 55, 27–41.
Romans, S. E., Martin, J. L., Anderson, J. C., & Herbison, G. P. (1995). Sexual abuse in childhood and deliberate self-harm. American
Journal of Psychiatry, 152, 1336–1342.
Rosenthal, R. J., Rinzler, C., Wallsh, R., & Klausner, E. (1972). Wrist-cutting syndrome: The meaning of a gesture. American Journal of
Psychiatry, 128, 1363–1368.
Roy, A. (2003). Distal risk factors for suicidal behavior in alcoholics: Replications and new findings. Journal of Affective Disorders, 77,
267–271.
Sidebotham, P., Golding, J. & The ALSPAC Study Team. (2001). Child maltreatment in the ‘‘Children of the Nineties’’: A longitudinal
study of parental risk factors. Child Abuse and Neglect, 25, 1177–1200.
Soloff, P. H., Lynch, K. G., & Kelly, T. M. (2002). Childhood abuse as a risk factor for suicidal behavior in borderline personality
disorder. Journal of Personality Disorders, 16, 201–214.
Stepakoff, S. (1998). Effects of sexual victimization on suicidal ideation and behavior in US women. Suicide & Life-Threatening Behavior,
28, 107–126.
Veale, D., Boocock, A., Gournay, K., & Dryden, W. (1996). Body dysmorphic disorder: A survey of fifty cases. British Journal of
Psychiatry, 169, 196–201.
Wittchen, H. U. (1994). Reliability and validity studies of the WHO composite international diagnostic interview (CIDI): A critical review.
Journal of Psychiatry Research, 28, 57–84.
World Health Organization. (1990). Composite international diagnostic Interview. Geneva, Switzerland: WHO.
Zuravin, S. J., & Fontanella, C. (1999). The relationship between child sexual abuse and major depression among low-income women: A
function of growing up experiences? Child Maltreatment: Journal of the American Professional Society on the Abuse of Children, 4, 3–12.