Académique Documents
Professionnel Documents
Culture Documents
model of personality provides a relatively comprehensive coverage of personality traits, it can be used to explore and generate
hypotheses about phenomena that have been relatively underinvestigated or about which there is relatively little theorizing.
We are aware of no theory that makes explicit predictions
about the contributions of specific personality traits to specific
dimensions of suicidal behavior in particular demographic and
diagnostic groups. Most personality theories of suicidal behavior lack the specificity warranted by the epidemiological data.
For example, despite long-established age and gender differences in suicidal behavior (Durkheim, 1897/1951; Monk, 1987),
clinical writings (e.g., Buie & Maltsberger, 1989; Hendin, 1991)
have typically emphasized the role of hostility, independent of
age, gender, or any other demographic or contextual variable.
Use of an omnibus personality questionnaire grounded in the
FFM increases the likelihood that traits central to late-life suicidal behavior are not overlooked, even if they are ignored in clinical and theoretical writings. Indeed, the FFM may be construed
as hypothesis-generating. Proponents of the FFM argue that it
provides a fixed reference point from which to assess a variety
of different scales (Costa & McCrae, 1992; Marshall, Wortman,
Vickers, Kusulas, & Hervig, 1994). It therefore overcomes a
perennial problem in personality psychology: Scales with different labels measure the same trait, while those with the same
label measure different traits.
Among others, Kagan (1994), McAdams (1994), and Block
(1995) offer less optimistic opinions of the FFM. Kagan (1994)
critiques its basic premises, including the scientific utility of a
natural language approach to personality, self-report measures,
and factor-analysis itself. He ultimately concedes that, even
though the five factors "omit too much information" and are
"insufficiently differentiated... [they] do tell us something of
interest" (pp. 45-46). McAdams (1994) also takes issue with
the basic premises and criticizes trait assessments in general on
the grounds that they fail to provide causal explanations for
human behavior, disregard the conditional and contextual nature of human experience, and fail to provide enough detailed
Completed suicide may be the most preventable lethal complication of depressive disorders in older adults.
Identification of risk factors for suicidal behavior has therefore become a major public health priority. Using data collected on SI depressed patients 50 years of age and older, we report analyses designed to determine the associations between the personality traits that constitute the Five Factor Model of personality and measures of suicidal behavior and
ideation. We hypothesized that low Extraversion would be associated with a lifetime history of attempted suicide, and
high Neuroticism would be associated with suicidal ideation. Results were generally consistent with the hypotheses. We
also observed a relationship between Openness to Experience and suicidal ideation. These findings suggest that longstanding patterns of behaving, thinking, and feeling contribute to suicidal behavior and thoughts in older adults and
highlight the need to consider personality traits in crafting and targeting prevention strategies.
information to predict specific behaviors in certain circumstances. Block (1995) generally accepts the premises upon
which FFM research is based, though he is somewhat critical of
the "arbitrariness" (p. 189) of factor analysis and the overreliance on self- and peer-report data. He also raises a number
of technical concerns, such as the high intercorrelations among
the ostensibly uncorrelated five factors. Still, the FFM has withstood criticism from those who share, and do not share, its basic
assumptions (Costa & McCrae, 1995; McCrae & Costa, 1997),
and it has proven useful in research on health outcomes in older
adults (Hooker, Frazier, & Monahan, 1994; Hooker, Monahan,
Bowman, Frazier, & Shifren, 1998; Hooker, Monahan, Shifren,
& Hutchinson, 1992). Those achievements may be sufficient
justification for its continued application to questions of public
health significance pertaining to older adults.
METHODS
Participants
Participants were drawn from a larger, ongoing, case-control
study of attempted suicide in major depressive disorder.
Depressed inpatients 50 years of age and older who were
admitted to the hospital following a suicide attempt were compared with similarly depressed age- (5 years) and gendermatched inpatients whose admissions were not precipitated by
a suicide attempt. Although there is significant heterogeneity in
the prevalent diagnoses of young adult suicides, after age 50,
the psychiatric diagnoses associated with completed suicide become increasingly homogeneous, and affective disorders are
present in over 70% of cases (Conwell et al., 1996). Thus, by
P19
P20
DUBERSTE1NETAL.
choosing 50 as the lower age limit for study entry, we are able
to control for affective disorder without excluding a large portion of people at risk for completed suicide.
Materials
.86 to .92 (Costa & McCrae, 1992). Longitudinal studies conducted over periods of up to 7 years have frequently reported
test-retest correlation coefficients greater than .6, attesting to
the stability of these five domains (Costa & McCrae, 1992).
Although the 60-item NEO-FFI has been used in gerontology
research (e.g., Hooker et al., 1994) and in research on depressed
outpatients (Bagby et al., 1998), we are unaware of any study
that has used the 240-item NEO-PI-R with older, depressed
inpatients.
History and number of suicide attempts.For decades, the
standard approach to research on personality and attempted
suicide involved a static group comparison of individuals seeking health care following a suicide attempt with individuals
seeking care for another reason ("nonattempters"). This approach is limited primarily because a portion of those described as nonattempters have attempted suicide in the past. As
a general principle, when personality traits increase risk for
certain adverse health outcomes, such as attempted suicide,
risk refers to the entire lifecourse and is not confined to the period of time during which subjects are enrolled in a study.
Thus, in the present study we examined the relationship between personality and (a) lifetime suicide attempter status, and
(b) number of suicide attempts.
Our data on the number of suicide attempts were based, in
part, on participants' responses to the questions: "How many
times all together in your life have you actually done something
with the intention of taking your life?" and "How many suicide
attempts have you made in your life?" With respect to the latter
question, past self-destructive behaviors were coded as suicide
Spectrum of Suicidal Behavior Scale (SSB).Project coordinators used this 5-point ordinal scale (Pfeffer, Stokes, &
Shindledecker, 1991) to rate the participants' most serious suicidal behavior over the past month. Thus, the SSB and SSI
cover different time frames (month prior to hospitalization vs.
week prior to interview). Participants were rated a 1 (nonsuicidal) if there was "no evidence of any self-destructive or suicidal
thoughts or actions," a 2 if there is evidence of suicidal ideation,
a 3 if they made a suicidal threat, a 4 if they made a mild suicide attempt, or a 5 if they made a serious suicide attempt. In
the present study, the SSB served primarily as a measure of suicidal ideation in the month prior to admission. For analytic purposes, we therefore dichotomized SSB scores (1 vs. other) and
estimated its reliability by means of the kappa-coefficient (K =
.54) using chart documentation of preadmission suicidal behavior as the criterion.
Structured Clinical Interview for DSM-III-R.This instrument was used to establish Axis I psychiatric diagnoses
RESULTS
Descriptive statistics and zero-order correlations (Kendall's
tau) between the NEO factors and continuous outcomes (SSI
and number of suicide attempts) are presented in Table 1. There
relationships between the personality variables and dichotomous endpoints. Table 2 shows that women, suicide ideators,
and death ideators obtained higher Neuroticism scores, and
those who had attempted suicide obtained lower scores on
Extraversion. Of the 10 intercorrelations among the 5 NEO
variables, five had absolute values less than .07; the highest
value was .46. Therefore, multicollinearity did not appear to
pose any problems for the regression analyses.
Presence and Number of Suicide Attempts
The first logistic regression sought to determine whether the
personality variables were associated with having made a suicide attempt. As shown in Table 3, those who obtained lower
Extraversion scores were more likely to have made a lifetime
P21
P22
DUBERSTEIN ET AL.
Table 1. Unadjusted Relationship Between NEO-PI and Continuous Variables: Kendall's Tau
Continuous Variable
SD
Ne
Ex
Op
Ag
Co
Age
Number of lifetime SA
Total SSI score
81
81
81
61.30
9.6
1.6
10.3
-.29***
.19*
2g***
.04
-.19*
-.15
-.17*
.04
.11
-.06
-.04
.07
-.15
-.04
-.14
0.85
7.52
Ne = Neuroticism; Ex = Extraversion; Op = Openness to Experience; Ag = Agreeableness; Co = Conscientiousness; SA= Suicide Attempts; SSI = Scale for
Suicidal Ideation.
Ne
Ex
-0.15
2.66**
-2.03*
-1.50
3.11***
3.01***
-1.06
-1.24
Op
Co
Ag
0.85
-0.44
1.48
1.80
1.76
1.91
-1.77
0.14
-1.23
-1.21
-0.86
1.31
-1.60
-1.64
-0.69
Lifetime SA = participants with a lifetime history of at least one suicide attempt were contrasted with all others. Suicidal ideation - SSB = participants who scored
a 1 (absence of suicidal ideation or suicide attempt in the month prior to hospitalization) were contrasted with all others; Suicidal ideation - SSI = participants who
reported suicidal ideation in the week prior to interview (scored a 1 or higher on the Scale for Suicidal Ideation) were contrasted with all others; Death ideation - SSI
= participants who acknowledged that their wish to die outweighed their wish to live in the week prior to interview (scored a 1 or higher on the death ideation items
of the Scale for Suicidal Ideation) were contrasted with all others; Ne = Neuroticism; Ex = Extraversion; Op = Openness to Experience; Ag = Agreeableness: Co =
Conscientiousness. *p < .05; **p < .01 ;***/}< .001. All rf/s = 79, except where otherwise noted.
"Women scored higher than men.
h
Unequal variance (p < .05), there was greater heterogeneity among those with a lifetime SA; d/'tbr t-test = 78.
Predicted Variable
Lifetime SA
Number of lifetime S A (outliers removed)
No suicidal ideas-SSB
Analysis
Model
1
2
3
Logistic
Poisson
Logistic
Predictor(s)
Coeff
SE
X:(D
p value
Extraversion
Extraversion
Openness
-.032
.016
-.026
-.054
+.038
.007
.025
4.43
14.06
.035
.0002
.02
.05
.01
.05
.02
Agreeableness
Suicidal ideation-SSI
Death ideation-SSI
4
5
Logistic
Logistic
Age
Openness
Neuroticism
-.087
.021
.038
+.038
+.037
.021
.018
5.65
3.73
6.38
3.87
4.49
Lifetime SA = participants with a lifetime history of at least one suicide attempt were contrasted with all others; Suicidal ideation - SSB = participants who scored
a 1 (absence of suicidal ideation or suicide attempt in the month prior to hospitalization) were contrasted with all others; Suicidal ideation - SSI = participants who
reported suicidal ideation in the week prior to interview (scored a 1 or higher on the Scale for Suicidal Ideation) were contrasted with all others; Death ideation - SSI
= participants who acknowledged that their wish to die outweighed their wish to live in the week prior to interview (scored a 1 or higher on the death ideation items
of the Scale for Suicidal Ideation) were contrasted with all others; CoetT = coefficient.
goodness of fit was not significant, x2 (8) = 9.90, p = .27, indicating a satisfactory fit. Next, we examined predictors of the
number of previous suicide attempts using Poisson regression.
Higher Neuroticism and lower Extraversion emerged as significant predictors; however, 4 participants were outliers. In each
case, the predicted number of suicide attempts was lower than
the actual number. All 4 had at least two Axis I diagnoses; 3 of
the 4 had psychotic features. Concerned that the nature and interpretation of our findings may have been unduly influenced
by this relatively small group reporting numerous suicide attempts, we removed the outliers and conducted another Poisson
regression. The results (Table 3, Analysis 2) partially duplicated
the previous analysis. Again, Extraversion was a strong predictor, but Neuroticism was not, despite its significance in the pre-
1 .97'*
1 .69"
DISCUSSION
These findings reinforce the notion that personality traits
ought to be seriously considered as potential risk factors for
late-life suicidal behavior and ideation. Even in this demographically and diagnostically homogenous group of psychiatric inpatients, personality traits were important predictors of
suicide attempts and suicidal ideation. Although there were
some negative findings, regression analyses supported hypothesized associations between Extraversion and attempted suicide
and between Neuroticism and suicidal ideation. These analyses
also generated a novel hypothesis linking Openness to suicidal
ideation.
Substantive Findings
Three findings are especially noteworthy. First, higher
Extraversion distinguishes people who have never made an attempt from those who have. Extraversion is positively associated with positive affect (Clark, Watson, & Mineka, 1994) and
increased social support (e.g., Von Dras & Siegler, 1997), and
negatively correlated with trait, but not state, hopelessness
(Young et ah, 1996). Extraverted individuals may be less likely
to engage in suicidal behavior even in the midst of a depressive
episode because they are more likely to recruit and affectively
benefit from friendships and family relations, perhaps as a result of better social skills (cf. Zweig & Hinrichsen, 1993).
Suicide attempts among those who are low in Extraversion may
reflect a tendency to take matters into one's own hands, rather
than attempt to recruit help from others. Strategies for treating
young adult suicide attempters (Linehan, 1993) have been informed by data linking personal concerns (Linehan et ah, 1986)
or personality dimensions (Rudd, Joiner, & Rajab, 1996) with
suicidal behavior, but similar data on older adults are rare.
Future research aimed at identifying the mediators of the relationship between Extraversion and attempter status may lead to
interventions designed to decrease the risk of nonfatal suicidal
behavior. This is important in part because suicide attempts
may exacerbate the physical morbidity (Gallo et ah, 1997; Katz,
1996 ) and mortality risks (Gallo et ah, 1997; Penninx et ah,
1999; Zubenko et ah, 1997) frequently associated with late-life
depressive disorders.
with the unadjusted analyses, which showed a relationship between Neuroticism and the SSB score (Table 2). The HosmerLemeshow for the multiple regression was nonsignificant, \2
(8) = 10.75, p = .22, indicating a reasonable fit. When we dichotomized the SSI score (0 vs. > 0) and created two groups,
suicide ideators and nonideators, the logistic regression (Analysis 4) yielded one significant predictor (age), in contrast to the
unadjusted analyses, which implicated Neuroticism in suicidal
ideation (Table 2). We also conducted a linear regression with
the total score on the SSI as the dependent variable. Again, only
age was associated with that outcome, F(l,73) - 4.56,p = .04,
but there was a trend for those higher in Openness to obtain
higher SSI scores, F( 1,73) = 3.25, p = .07. Next, we created
two groups, those who reported death ideation in response to
Items 1-3 of the SSI and those who did not. Table 3 (Analysis
5) shows that higher Neuroticism and higher Openness
P23
P24
DUBERSTE1NETAL.
Conclusion
Our findings suggest that suicidal thoughts and behavior are
rooted in longstanding patterns of behaving, thinking, and feeling, and highlight the need to consider personality traits in crafting and targeting prevention strategies. Suicidal ideas and
behavior are not an inevitable consequence of aging, disease,
disability, or even depression. The current findings thus
challenge an ageist stereotype that has probably contributed
to a lack of interest in preventing late-life suicide (AARP
Foundation/Center for Mental Health Services, 1997). On the
other hand, findings must be regarded as preliminary. Several
lines of preintervention research ought to be pursued.
The "state-trait problem" potentially confounds research on
psychiatric inpatients, many of whom may be in acute distress
while completing questionnaires or participating in interviews.
Although it is unlikely that our observation that suicidal behavior and ideas are associated with Extraversion, Neuroticism,
and Openness, follow-up data, collected when patients no
longer meet diagnostic criteria for major depressive disorder,
would be useful (e.g., Santor, Bagby, & Joffe, 1997). Future research may also benefit from informant reports and projective,
physiological, or other nonverbal sources of psychological
information.
By collecting data on a relatively homogeneous group of
older persons with major depression, we sought to decrease the
probability that potentially confounding effects of age or major
psychiatric diagnosis would obscure relationships between personality and suicide variables. Still, heterogeneity was apparent
in the analyses on previous suicide attempts. These analyses
suggested that Neuroticism may contribute to multiple suicide
attempts in those with comorbidity. Further research on larger
samples may be necessary to determine whether the personality
traits associated with suicidal behavior in depressed patients
with psychiatric comorbidity differ from those without comorbidity.
Even in the absence of consensus concerning the ideal design, sampling strategy, and statistical analysis required to determine whether the constructs of suicidal ideation, attempted
suicide, and completed suicide are categorically distinct (cf.
Flett, Vredenburg, & Krames, 1997; Meehl, 1992), tragedies
may be prevented for now simply by acknowledging that developing risk-identification and prevention strategies based on assumptions implicit in the severity continuum model could be
misguided. Variables associated with the absence of reported
suicidal ideation, such as low Openness, may not confer decreased risk for completed suicide. Paradoxically, in some
patients who are low in Openness, the absence of reported suicidal ideation may confer increased suicide risk.
This study uncovered the possibility that different personality variables are associated with attempted suicide and suicidal
ideation, with Extraversion associated with the former and
Openness more closely tied to the latter. We are not arguing for
eliminating the severity continuum model of suicide; rather, we
are suggesting that the categorical model has much to offer. It is
P25
Hosmer, D. W., & Lemeshow, S. (1989). Applied logistic regression. New York:
John Wiley & Sons, Inc.
Kagan, J. (with Snidman, N., Arcus, D., & Reznick, J. S.). (1994). Galen's
Available: http://www.cdc.gov/nchswww/data/gm250_97.pdf.
Chatterjee, S., & Hadi, A. J. (1988). Sensitivity analysis in linear regression.
New York: John Wiley & Sons, Inc.
103-116.
Cloninger, C. R., Svrakic, D. M., & Przybeck, T. R. (1993). A psychobiological
model of temperament and character. Archives of General Psychiatry, 50,
975-990.
Congressional Record. (1997). Resolution recognizing suicide as a national
D. (1996). Relationships of age and Axis I diagnosis in victims of completed suicide: A psychological autopsy study. American Journal of
Psychiatry, 153, 1001-1008.
Costa, P. T., Jr., & McCrae, R. R. (1987). Neuroticism, somatic complaints, and
disease: Is the bark worse than the bite? Journal of Personality, 55,
299-316.
Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory
and NEO Five Factor Inventory': Professional manual. Odessa, FL:
Aging, 3, 230-232.
Hooker, K., Frazier, L. D., & Monahan, D. (1994). Personality and coping
among caregivers of spouses with dementia. Gerontologist, 34, 386-392.
Hooker, K., Monahan, D., Bowman, S. R., Frazier, L. D., & Shifren, K. (1998).
Personality counts for a lot: Predictors of mental and physical health of
Psychology, 6, 880-881.
Lyness, J. M., Duberstein, P. R., King, D. A., Cox, C., & Caine, E. D. (1998).
Medical illness burden, trait neuroticism, and depression in primary care elderly. American Journal of Psychiatry, 155,969-971.
Lyness, J. M., Noel, T. K., Cox, C., King, D. A., Conwell, Y, & Caine, E. D.
(1997). Screening for depression in elderly primary care patients. Archives
of Internal Medicine, 757,449-454.
Maris, R. W. (1992). The relationship of nonfatal suicide attempts to completed
Guilford Press.
Marshall, G. N., Wortman, C. B., Vickers, R. R., Jr., Kusulas, J. W, & Hervig,
L. K. (1994). The five-factor model of personality as a framework for
personality-health research. Journal of Personality and Social Psychology,
67, 278-286.
McAdams, D. P. (1994). Can personality change? Levels of stability and growth
Meehl, P. E. (1992). Factors and taxa, traits and types, differences of degree and
differences in kind. Journal of Personality, 60, 117-124.
Monk, M. (1987). Epidemiology of suicide. Epidemiologic Reviews, 9,51-69.
Moscicki, E. K. (1989). Epidemiologic surveys as tools for studying suicidal
MD: Author.
O'Carroll, P. W., Berman, A. L., Maris, R. W., Moscicki, E. K., Tanney, B. L.,
& Silverman, M. M. (1996). Beyond the Tower of Babel: A nomenclature
for suicidology. Suicide and Life-Threatening Behavior, 26,237-252.
Penninx, B. W. J. H., Geerlings, S. W., Deeg, D. J. H., van Eijk, J. T. M.,
van Tilburg, W., & Beekman, A. T. F. (1999). Minor and major depression
and the risk of death in older persons. Archives of General Psychiatry, 56,
889-895.
Pfeffer, C. R., Stokes, P., & Shindledecker, R. (1991). Suicidal behavior and
hypothalamic-pituitary-adrenocortical axis indices in child psychiatric inpatients. Biological Psychiatry, 29,909-917.
Roy, A. (1998). Is introversion a risk factor for suicidal behaviour in depression? Psychological Medicine, 28, 1457-1461.
P26
DUBERSTEINETAL
Young, M. A., Fogg, L. F, Scheftner, W., Fawcett, J., Akiskal, H., & Maser, J.
Zubenko, G. S., Mulsant, B. H., Sweet, R. A., Pasternak, R. E., & Tu, X. M.
(1997). Mortality of elderly patients with psychiatric disorders. American
Journal of Psychiatry. 150, 1687-1692.
Zweig, R. A., & Hinrichsen, G. A. (1993). Factors associated with suicide attempts by depressed older adults: A prospective study. American Journal of
Psychiatrv, 150, 1687-1692.