DEPRESSION AND ANXIETY 25:742–751 (2008)

Research Article
CHILDHOOD TRAUMA AND OBSESSIVE-COMPULSIVE
SYMPTOMS
Carol A. Mathews, M.D.,1 Nirmaljit Kaur, M.D.,2 and Murray B. Stein, M.D., M.P.H.3

Childhood trauma is known to predispose to a variety of psychiatric disorders,

including mood, anxiety, eating, and personality disorders. However, the
relationship between childhood trauma and obsessive-compulsive symptoms has
not been well studied. This study examines the relationship between childhood
trauma, personality facets, and obsessive-compulsive symptoms in 938 college
students using the Childhood Trauma Questionnaire, the Leyton Obsessional
Inventory, and the NEO Personality Inventory-Revised. Between 13 and 30% of
subjects met criteria for childhood trauma, with emotional neglect the most
commonly reported experience. There was a small but significant association
between obsessive-compulsive symptoms and childhood trauma, specifically
emotional abuse and physical neglect, all of which was accounted for by co-
occurring anxiety symptoms. An independent association was also seen between
emotional abuse, physical abuse, and high levels of obsessive-compulsive
symptoms (‘‘probable obsessive-compulsive disorder’’), which remained signifi-
cant in the context of co-occurring anxiety symptoms. A similar association was
seen between obsessive-compulsive symptoms and conscientiousness, and between
emotional neglect and sexual abuse and conscientiousness, suggesting that an
indirect role for childhood trauma in the development of obsessive-compulsive
symptoms may also exist. Depression and Anxiety 25:742–751, 2008.
& 2007 Wiley-Liss, Inc.

Key words: obsessive-compulsive symptoms; childhood trauma; neglect; abuse;

personality traits; NEO-PI-R; CTQ; LOI

INTRODUCTION
Childhood
1
Department of Psychiatry, University of California, San
trauma (physical, emotional, or sexual Francisco, San Francisco, California
abuse, and emotional or physical neglect) has been 2
Department of Psychiatry, Palo Alto VAMC and Stanford
linked to specific neurobiological changes, and is University, Palo Alto, California
associated with a variety of long-term adverse effects, 3
Department of Psychiatry, University of California, San
including an increase in the risk of developing Diego, La Jolla, California
psychiatric illnesses and adult at-risk behaviors [Bern-
Contract grant sponsor: Supported in part by grant MH-64122 to
stein et al., 1998; Bierer et al., 2003; Breslau, 2002; Chu
MBS and by grant RR15533 to CAM.
and Dill, 1991; Fonagy, 2000; Ford et al., 2000; Gearon
Correspondence to: Dr. Carol A. Mathews, Department of
et al., 2003; Haller and Miles, 2004; Heim and
Nemeroff, 2001; Heim et al., 2002; Kendler et al., Psychiatry, University of California, 401 Parnassus Ave, Box
F-0984, San Francisco, CA 94143-0894.
2004; Langeland et al., 2004; Nemeroff, 2004; Ray-
E-mail: [email protected]
worth et al., 2004]. Despite evidence that childhood
trauma has broad-based and far-ranging effects, its Received for publication 22 June 2006; Revised 10 January 2007;
relationship to only a few psychiatric syndromes, Accepted 30 January 2007
particularly posttraumatic stress disorder (PTSD), DOI 10.1002/da.20316
dissociative disorders, depression, and borderline Published online 7 June 2007 in Wiley InterScience (www.
personality disorder, has been studied in detail [Bernet interscience.wiley.com).

r 2007 Wiley-Liss, Inc.

 Research Article: Childhood Trauma 743

and Stein, 1999; Gearon et al., 2003; Haller and Miles, Mathews et al., 2004; Roussos et al., 2003]. Obsessive
2004; Hubbard et al., 1995; Johnson et al., 2003; symptoms are present at low levels throughout the
Langeland et al., 2004; Levitan et al., 1998; Windle lifespan, do not necessarily cause anxiety, and at certain
et al., 1995; Yen et al., 2002; Zlotnick et al., 2001]. developmental stages are thought to be adaptive. For
Substantial research has shown that high levels of example, in early childhood, obsessive-compulsive
obsessive-compulsive symptoms (OCS), particularly in traits or behaviors are important in the emergence of
adolescence and early adulthood, are associated with an behavioral and cognitive control and as an adaptive
increased risk of Axis I disorders, including obsessive- reaction to common childhood fears and anxieties.
compulsive disorder (OCD), depression, eating dis- Similarly, in early adolescence and young adulthood,
orders, polysubstance abuse, panic and generalized low levels of OCS can act as an aid to developing a
anxiety disorder (GAD), and also with multiple other sense of control and self-efficacy during separation
impairments (increases in somatic complaints, avoi- /individuation from parents and other authority figures
dant, dependent, and obsessive personality disorders, [Evans et al., 1999; Feygin et al., 2006; Franzblau,
tendency to be more withdrawn, worse peer relation- 1997; Zohar and Felz, 2001]. However, individuals
ships, lower marriage rates, higher divorce rates, and with high OCS scores are at substantially increased risk
lower life satisfaction) [Berg et al., 1988; Brynska and for the development of OCD and other disorders
Wolanczyk, 2005; Grabe et al., 2000; King et al., 1995; compared with those with low or moderate scores
Leonard et al., 1990; Maggini et al., 2001; Mancini [Berg et al., 1989; Roussos et al., 2003]. For this reason,
et al., 1999; Roussos et al., 2003; Thomsen, 1993; we examined both the relationship between childhood
Westen et al., 2005; Zohar and Bruno, 1997]. There is trauma and total OCS score, as well as the relationship
known to be a strong overlap between non-OCD between childhood trauma and high OCS score (which
anxiety symptoms and OCD, as well as between these we refer to as ‘‘probable OCD’’) with the hypothesis
symptoms and childhood trauma [Donahue, 2005; that childhood trauma would be more likely to cause a
Heim and Nemeroff, 2001; Tukel et al., 2002]. PTSD maladaptive response, and thus would be more strongly
symptoms in particular have a high degree of overlap associated with probable OCD. We further hypothe-
with obsessive symptoms, in part due to symptoms sized that specific personality characteristics that have
such as intrusive unwanted thoughts, and rituals and/or been proposed as vulnerability traits for OCS (e.g.,
avoidance behaviors developed to avoid such thoughts neuroticism, conscientiousness, and introversion, or
[Huppert et al., 2005]. However, little is known about low extraversion) would also be associated with higher
the relationship between childhood trauma and OCS levels of childhood trauma, and that, given the known
or OCD, despite the known comorbidities between overlap between OCS and other anxiety symptoms,
OCS and neuropsychiatric disorders associated with much of the association between childhood trauma and
childhood trauma [Angst et al., 2004; Carter et al., OCS, if any, would be mediated by co-occurring
2004; Gershuny et al., 2003; Grabe et al., 2001; Heim anxiety symptoms [Bienvenu et al., 2004; Gershuny
and Nemeroff, 2001; Huppert et al., 2005; Nemeroff, et al., 2000; Rector et al., 2002; Samuels et al., 2000].
2004; Rayworth et al., 2004]. To date, only one study
examining the association between childhood trauma
and OCD has been published [Lochner et al., 2002]. METHODS
This study, which assessed physical, emotional, and
sexual abuse and physical and emotional neglect using SUBJECTS
the Childhood Trauma Questionnaire in 74 women Subjects were 938 undergraduate students enrolled
with OCD, 36 women with trichotillomania, and 31 in an introductory psychology course who agreed to
control women, found that levels of childhood trauma participate in the study for extra credit in the course.
in general, and emotional neglect in particular, were Subjects were primarily, but not exclusively, freshman.
higher in subjects with OCD and trichotillomania than The study was approved by Institutional Review
in controls. This work has yet to be replicated, nor has Boards at the University of California San Diego
there been any work published examining the role of School of Medicine and San Diego State University.
childhood trauma on OCS in non-clinical populations Study participation was voluntary and informed con-
or in male subjects. sent was obtained from all subjects. Six hundred eighty-
In this study, we examine the relationship between one participants were female (73%) and 246 were male
childhood trauma and OCS in a population of male (27%). Eleven students did not report their gender.
and female college students. We hypothesized that Sixty-three percent of the participants were 18 years or
childhood trauma, particularly emotional abuse and younger, 23% were 19 years, and 8% were 20 years,
neglect, would be associated with an increase in OCS. with the remainder falling between 22 and 25 years.
The distribution of OCS among adolescents and young Fifty-three percent of subjects were Caucasians, 12%
adults has a right skew, with the majority of subjects were Latino, 9% were Asian American, 7% were
having low OCS levels, and only a small proportion Filipinos, 4% were African American, 1% were Native
having high OCS levels at or above the suggested American, and 14% reported that they were of mixed
cutoffs for clinical significance [Berg et al., 1988; or other races.
Depression and Anxiety
744 Mathews et al.

ASSESSMENTS personality domains assessed by the NEO-PI-R

(neuroticism, extraversion, openness to experience,
All participants completed a battery of questionnaires agreeableness, and conscientiousness) was collected,
within a 2-hour time period. OCS were assessed with a neuroticism, extraversion and conscientiousness were
short form of the Leyton Obsessional Inventory (LOI- the primary facets of interest. As practical constraints
SF) [Cooper, 1970; Mathews et al., 2004]. The LOI-SF prevented direct interviews of the participants, clinical
is a 30-item self-report instrument that was designed as diagnoses for Axis I psychiatric disorders were not
a screening instrument for clinically significant OCS. obtained.
The original LOI, from which the short form was
derived, has excellent internal consistency, as well as STATISTICAL ANALYSES
good convergent, divergent, and discriminative validity,
differentiating OCD and clinically significant OCS All analyses were generated using Stata 8.0. w2
analyses were used to examine differences between
from other anxiety symptoms [Cooper, 1970; Richter
the proportion of students of different ethnicities and
et al., 1994; Stanley et al., 1993]. Although the
between males and females with clinically significant
psychometric properties of the LOI-SF have not been
childhood trauma, anxiety, PTSD, and OCS scores.
reported, it was chosen for this study because it is
The proportion of students with potentially clinically
directly derived from the LOI, and because it was not
feasible to administer the entire LOI, which consists of significant scores on the CTQ-SF, LOI-SF, PCL-C,
and STAIT was determined using commonly accepted
70 two-part questions, to the subjects. Exposure to
cutoffs for more than minimal (i.e., low, moderate, or
childhood trauma was assessed using the 28-item short
severe) abuse or neglect (CTQ-SF subscores of Z9 for
form of the Childhood Trauma Questionnaire (CTQ-
emotional abuse, Z10 for emotional neglect, Z8 for
SF) [Bernstein et al., 2003; Scher et al., 2001]. The
physical neglect, Z8 for physical abuse, and Z6 for
CTQ-SF is a self-report instrument that consists of five
sexual abuse), moderate PTSD and anxiety symptoms,
subscales assessing emotional, physical, and sexual
and high levels of OCS (Z50 for the PCL-C, Z50 for
abuse, and emotional and physical neglect. Typical
the STAIT, and Z20 for the LOI-SF) [Mathews et al.,
items on the CTQ-SF include ‘‘When I was growing
2004; Paivio and Cramer, 2004; Spielberger, 1970;
up, people in my family called me things like ‘stupid’,
‘lazy’, or ‘ugly’’’ (emotional abuse), ‘‘When I was Weathers, 1991]. The prevalence of subjects with any
childhood trauma experience (defined as meeting the
growing up, someone tried to make me do sexual
cutoff score for one or more of the CTQ subscales) was
things or watch sexual things’’ (sexual abuse), and
‘‘When I was growing up, I didn’t have enough to eat’’ also assessed. Regression analyses were used to examine
the relationships between childhood trauma scores and
(physical neglect). Duration and intensity of traumatic
OCS, and to examine the relationships between
experiences are not assessed. Instead, extent of the
personality style (as measured by the NEO-PI-R) and
traumatic experience is measured using a score that is
OCS and childhood trauma. CTQ-SF, LOI-SF, PCL-
generated for each subscale and reflects total number of
C, and STAIT scores were included as continuous
items endorsed. Individual items are scored on a scale
variables in the linear regressions. Regressions were
of 0 ‘‘never true’’ to 4 ‘‘very often true’’. Each subscale
conducted in a backward stepwise manner, with
score is categorized into four groups: none or minimal,
variables with Po0.1 sequentially excluded from the
low to moderate, moderate to severe, and severe to
model until the best-fit model was reached. Gender,
extreme [Bernstein and Fink, 1998]. The CTQ-SF has
been well validated in both clinical and non-clinical age, and ethnicity were controlled for in all analyses. As
a comparison, the relationships between childhood
populations, and has excellent reliability (.70 – .93) for
trauma and PTSD and anxiety were also assessed.
all of the subscales, with the lowest reliability for
physical neglect and the highest for sexual abuse
[Bernstein and Fink, 1998; Paivio and Cramer, 2004]. RESULTS
Anxiety was assessed using the State Trait Anxiety
Inventory, trait version (STAIT), a well-accepted CHARACTERISTICS OF THE STUDY
measure of trait anxiety that has excellent reliability SAMPLE
(.65–.86), and is highly correlated with other measures Between 13 and 30% of the total sample met
of trait anxiety [Spielberger et al., 1970]. Symptoms of criteria for significant childhood trauma of a particular
PTSD were assessed with the PTSD Checklist-Civilian subtype, with approximately 50% meeting criteria for
Version (PCL-C), a 17-item self-report scale that is any childhood trauma (Table 1). Emotional neglect was
based on DSM-IV criteria [Weathers et al., 1991]. the most commonly reported experience, and sexual
Reliability is high for the PCL-C (.92), and it has abuse was the least common. Males reported signifi-
excellent convergent validity and discriminant validity cantly more trauma overall, as well as more physical
[Ruggiero et al., 2003]. Personality traits were assessed abuse, physical neglect, and emotional neglect than
in a subset of students (N 5 128) using the Revised females; there were no significant gender differences in
NEO Personality Inventory (NEO-PI-R) [Costa and the rates of emotional or sexual abuse. In general,
McCrae, 1992]. Although information on all five Caucasian students reported the lowest rates of
Depression and Anxiety
 Research Article: Childhood Trauma 745

TABLE 1. Proportion of college students meeting criteria for probable childhood trauma, clinically significant OCS,
PTSD, and anxiety symptoms

Probable
Emotional Physical Sexual Emotional Physical Any Probable Probable anxiety
abuse abuse abuse neglect neglect childhood OCD PTSD disorder
(C 5 8) (C 5 7) (C 5 5) (C 5 9) (C 5 7) trauma (C 5 20) (C 5 50) (C 5 50)

Total sample (N 5 927) 28.7 19.3 13.8 30.3 20.9 50.5 1.9 7.6 20.5
Males (N 5 246) 27.2 29.3 13.8 37.0 34.6 57.7 1.2 8.5 19.1
Females (N 5 681) 29.1 15.9 13.5 27.6 15.6 42.3 1.9 7.3 21.0
w2 0.30 20.8 0.01 7.6 39.8 7.4 0.5 0.4 0.4
P value 0.58 o0.0001 0.90 0.006 o0.0001 0.006 0.48 0.55 0.53
African Americans (N 5 38) 39.5 44.7 29.0 34.2 29.0 71.1 0 2.6 18.4
Asians (N 5 83) 42.2 30.1 15.7 47.0 38.6 67.5 4.8 4.8 25.3
Latinos (N 5 110) 30.0 22.7 21.8 34.6 22.7 59.1 1.8 5.5 23.6
Native Americans (N 5 7) 28.6 28.6 28.6 42.9 0 42.9 0 0 14.3
Filipinos (N 5 65) 32.3 35.4 15.4 41.5 26.2 60.0 6.2 10.8 24.6
Caucasians (N 5 493) 22.7 12.2 10.1 25.2 15.2 41.2 1.0 6.1 18.7
Mixed/other (N 5 126) 35.7 20.6 12.7 29.4 25.4 56.4 2.4 15.9 21.4
w2 21.7 50.26 20.77 22.72 31.6 40.6 12.85 18.81 3.85
P value 0.001 o0.0001 0.002 0.001 o0.0001 o0.0001 0.045 0.004 0.697

C 5 cutoff score. P values are uncorrected. Owing to multiple testing, a P value of r0.008 is considered statistically significant.
OCS, obsessive-compulsive symptoms; PTSD, post-traumatic stress disorder.

childhood trauma, whereas Asians and African Americans (F 5 21.46, Po0.00001). Because ethnicity was
reported the highest. Approximately 2% of students strongly associated with total CTQ-SF score, the
met the cutoff for significant OCS, whereas 7.6% met analyses were repeated for Caucasians only, with
the criteria for probable PTSD, and 20.5% met criteria similar results (data not shown).
for clinically significant anxiety. In light of a previous report that childhood trauma
The rates of reported childhood trauma in our was associated with OCD, and in order to test the
sample, while high, are consistent with what has been hypothesis that childhood trauma would be preferen-
previously reported in non-clinical populations [Briere tially associated with high, potentially clinically sig-
and Elliott, 2003; Duran et al., 2004; Paivio and Cramer, nificant, levels of OCS, we also examined the rela-
2004; Spertus et al., 2003; Thombs et al., 2006]. tionship of childhood trauma and ‘‘probable OCD’’
However, contrary to previous findings, we found (defined as LOI-SF score Z20; [Mathews et al., 2004]).
higher reported rates of childhood trauma among males In this analysis, emotional abuse was positively
when compared with females [Paivio and Cramer, 2004; associated and physical abuse was negatively associated
Thombs et al., 2006]. This is most likely an artifact due with probable OCD. The total best-fit model for
to skewing of the sample, which consisted of more non- probable OCD accounted for approximately 11% of
Caucasian male subjects than female subjects (30 versus the variance (Table 3). To determine whether child-
24%, w2 5 4.8, P 5 0.028). We found substantially more hood trauma was independently associated with prob-
childhood trauma in non-Caucasian subjects than in able OCD, we compared the scatterplot of total CTQ-
Caucasian subjects (Table 1), a finding that has been SF score by total LOI-SF score for all individuals with
reported previously, with Asian, Hispanic, African that of individuals with probable OCD only and
American, and Native American subjects reporting up examined the regression fit. The slope of the regression
to two time the rates of childhood trauma reported by line for all individuals was positive, and showed that
Caucasian subjects [Hussey et al., 2006]. LOI-SF score increased in a linear fashion with
increasing CTQ-SF score. In contrast, the slope of
the regression line for subjects with probable OCD was
OBSESSIVE-COMPULSIVE SYMPTOMS
flat, suggesting that there was no additional relation-
AND CHILDHOOD TRAUMA
ship between trauma and LOI-SF score in this subset
We examined the relationship between overall levels of subjects.
of OCS and childhood trauma using a backward
stepwise regression approach. Emotional abuse and
CHILDHOOD TRAUMA, ANXIETY, AND
physical neglect were significantly associated with total
LOI-SF score, as was ethnicity, whereas emotional OBSESSIVE-COMPULSIVE SYMPTOMS
neglect, physical abuse, and sexual abuse were excluded As a comparison, we examined the relationship
from the best-fit model (Table 2). The total model between childhood trauma and PTSD and anxiety
accounted for approximately 8.4% of the variance symptoms in our sample. The relationship between
Depression and Anxiety
746 Mathews et al.

TABLE 2. Best-fit regression model for obsessive-compulsive symptoms as measured by total LOI-SF score

Coefficient (95% CI) Standard error T score P value

Emotional abuse 0.18 (0.08–0.27) 0.04 3.67 o0.0001

Physical neglect 0.21 (0.06–0.36) 0.08 2.80 0.005
African Americana 2.37 (0.74–4.00) 0.83 2.86 0.04
Asiana 2.19 (1.02–3.36) 0.60 3.67 o0.0001
Filipinoa 2.98 (1.69–4.26) 0.65 4.56 o0.0001
Latinoa 1.37 (0.72–2.75) 0.52 3.35 0.001
Mixed/othera 1.27 (0.31–2.23) 0.49 2.60 0.009
Constant 6.56 (5.36–7.66) 0.56 11.75 o0.00001

Excluded from the model were emotional neglect, sexual abuse, physical abuse, gender, age, and Native American ethnicity. F 5 21.46,
Po0.00001, adjusted R2 5 0.084.
P values are uncorrected. Owing to multiple testing, a P value of r0.0125 for the total model is considered statistically significant.
LOI-SF, short form of the Leyton Obsessional Inventory.
a
Comparison is Caucasian ethnicity.

TABLE 3. Best-fit regression model for probable OCD, defined as LOI-SF score Z20

Odds ratio (95% CI) Standard error z score P value

Emotional abuse 1.22 (1.08–1.37) 0.07 3.34 0.001

Physical abuse 0.77 (0.58–1.01) 0.11 1.92 0.055
Asiana 4.77 (1.40–16.24) 2.98 2.50 0.012
Filipinoa 4.64 (1.13–19.08) 3.34 2.13 0.033

Excluded from the model were emotional neglect, physical neglect, sexual abuse, African American, Latino, Native American, and mixed/other
ethnicities, gender, and age. LR w2 5 16.87, P 5 0.0020, pseudo R2 5 0.11.
P values are uncorrected. Due to multiple testing, a P value of r0.0125 for the total model is considered statistically significant.
OCS, obsessive-compulsive symptoms; LOI-SF, short form of the Leyton Obsessional Inventory.
a
Comparison is Caucasian ethnicity.

childhood abuse and PTSD has been well substantiated the final best-fit model included emotional abuse,
in the literature [Cloitre et al., 1997; Ford and Kidd, physical abuse, and PCL-C score, but not STAIT score
1998; Ford et al., 2000; Gearon et al., 2003]. For this (Table 4).
reason, we expected to find a strong positive relation- In a further attempt to determine whether
ship between PTSD symptoms and childhood trauma, the association between OCS and childhood trauma
particularly abuse, as well as a positive, but possibly was accounted for by overlap with anxiety symptoms,
weaker, relationship between other anxiety symptoms we next examined the association between the indivi-
and childhood trauma. As expected, emotional abuse, dual LOI-SF items and emotional abuse, the subtype
emotional neglect, and sexual abuse were all strongly of childhood trauma that was associated both with
positively correlated with total PCL-C score. The total probable OCD and with total LOI-SF score. To assess
model, which also included ethnicity but not age or the independent contribution of the individual LOI-SF
gender, accounted for 14.6% of the variance items, total STAIT score and total PCL-C score were
(F 5 19.89, Po0.00001). Similarly, emotional abuse included as covariates. The LOI-SF items retained in
and neglect were strongly positively correlated with the best-fit model for emotional abuse were primarily
total STAIT score. The total model, which also obsessive-compulsive specific symptoms (as compared
included ethnicity, accounted for 15% of the variance with items that might have some overlap with other
(F 5 26.6, Po0.00001). anxiety symptoms), and included: ‘‘I am excessively
We then examined the relationship between OCS concerned about cleanliness’’ (P 5 0.011), ‘‘I take a
and childhood trauma in the context of comorbid rather long time to complete my washing in the
anxiety and PTSD symptoms by conducting a back- morning’’ (P 5 0.012), ‘‘I frequently get nasty thoughts
wards stepwise regression analysis that included total and have trouble getting rid of them’’ (P 5 0.018), ‘‘I
STAIT and PCL-C scores as covariates. When STAIT usually have serious doubts about the simple everyday
and PCL-C total scores were included in the analysis things I do’’ (P 5 0.034), ‘‘My parents were very strict
using total LOI-SF score as the outcome, the final during my childhood’’ (P 5 0.055), and ‘‘Some numbers
best-fit model included STAIT total score, PCL-C are extremely unlucky’’ (P 5 0.088). These items, in
total score, ethnicity, and gender, but none of the addition to total STAIT score and total PCL-C score,
childhood trauma subscales. In contrast, when the same accounted for 15.3% of the total variance (F 5 21.32,
analysis was done using probable OCD as the outcome, Po0.00001).
Depression and Anxiety
 Research Article: Childhood Trauma 747

TABLE 4. Best-fit regression model for probable OCD, defined as LOI-SF score Z20, including STAIT total score and
PCLC total score as covariates

Odds ratio (95% CI) Standard error z score P value

Emotional abuse 1.16 (1.01–1.32) 0.08 2.13 0.033

Physical abuse 0.72 (0.52–1.00) 0.12 1.97 0.049
Asiana 3.40 (0.88–13.16) 2.35 1.77 0.077
PCL-C total 1.08 (1.04–1.13) 0.02 3.77 o0.0001

Excluded from the model were emotional neglect, physical neglect, sexual abuse, STAIT total score, African American, Filipino, Latino, Native
American, and mixed/other ethnicities, gender, and age. LR w2 5 26.05, Po0.0001, pseudo R2 5 0.17.
P values are uncorrected. Owing to multiple testing, a P value of r0.0125 for the total model is considered statistically significant.
OCS, obsessive-compulsive symptoms; LOI-SF, short form of the Leyton Obsessional Inventory; STAIT, State Trait Anxiety Inventory, trait
Version; PCL-C, PTSD Checklist-Civilian Version.
a
Comparison is Caucasian ethnicity.

PERSONALITY TRAITS, OBSESSIVE- DISCUSSION

COMPULSIVE SYMPTOMS, AND The results of this study suggest that childhood
CHILDHOOD TRAUMA trauma, particularly emotional trauma, plays a small
To explore whether childhood trauma played a role but significant role in the development of clinically
in the development of OCS via the development of significant, maladaptive levels of OCS, independent of
specific personality traits, we examined the relation- comorbid anxiety symptoms, both directly, and possi-
ships between the five personality facets of the NEO- bly also indirectly, through the personality facet of
PI-R and OCS, and between the facets and childhood conscientiousness. We initially hypothesized that (1)
trauma using backward stepwise regression analyses. childhood trauma, particularly emotional abuse and
Ethnicity, gender, age, PCL-C total score, and STAIT neglect, would be associated with OCS, (2) the
total score were included as potential confounding association would be stronger for high levels of OCS
variables. Of the five facets, only conscientiousness was (probable OCD) than for total OCS score, suggesting
significantly associated with OCS (P 5 0.0008). Other that childhood trauma is associated with the develop-
variables remaining in the best-fit model, which ment of maladaptive levels of OCS, (3) much of the
accounted for 27% of the variance, included ethnicity association between childhood trauma and OCS would
and PCL-C total score (F 5 9.10, Po0.00001, adjusted be mediated through comorbid anxiety symptoms, and
R2 5 0.2655). None of the NEO personality facets was (4) specific personality characteristics that have been
significantly associated with probable OCD. proposed as vulnerability traits for OCS would also be
In contrast, childhood trauma was significantly associated with higher levels of childhood trauma.
associated with all of the personality traits measured We found that childhood trauma, most consistently
by the NEO. Childhood sexual, physical, and emo- emotional abuse, but also physical abuse and physical
tional abuse were significantly correlated with neuroti- neglect, had a small but significant association with
cism, accounting for 11% of the total variance OCS (accounting for approximately 5–8% of the total
(F 5 6.36, P 5 0.0005). Emotional abuse showed a variance). Contrary to our hypothesis, however, we
strong positive association with neuroticism, whereas found no evidence that development of high levels
physical and sexual abuse showed a somewhat weaker of OCS is more related to the experience of child-
negative association. Childhood emotional neglect and hood trauma than to low to moderate OCS;
sexual abuse were significantly correlated with con- the scatterplots and fitted regression lines actually
scientiousness, accounting for 7.8% of the total suggested the reverse, that childhood trauma is more
variance. Sexual abuse was positively correlated with highly associated with total OCS score than with high
conscientiousness, whereas emotional neglect was OCS score.
negatively correlated with conscientiousness (Table 5). Surprisingly, physical abuse was associated with a
Childhood neglect (both physical and emotional) but reduced risk of probable OCD, rather than an
not abuse was significantly associated with extraver- increased risk, in contrast to physical neglect, which
sion, along with Asian and ethnicity, accounting for was associated with an increased total OCS score, and
10.7% of the total variance (F 5 5.45, P 5 0.0016). was not associated with probable OCD. These find-
Physical neglect was positively associated with extra- ings, in particular, the physical abuse finding, are
version, whereas emotional neglect, the most strongly unexpected, and replication in additional samples is
correlated variable, was negatively associated. needed to confirm the observed associations.
Emotional neglect and physical abuse were both As expected, we found that comorbid anxiety
negatively associated with agreeableness (F 5 4.80, symptoms accounted for the majority of the relation-
0.01), whereas no form of childhood trauma was ship between OCS and childhood trauma. Both non-
significantly associated with openness. specific anxiety and PTSD symptoms were more highly

Depression and Anxiety

748 Mathews et al.

TABLE 5. Best-fit regression model for conscientiousness as measured by the NEO-PI-R

Coefficient (95% CI) Standard error T score P value

Sexual abuse 0.31 (0.03–0.66) 0.17 1.81 0.073

Emotional neglect 0.46 ( 0.73– 0.18) 0.14 3.26 0.001
Constant 32.17 (29.29–35.05) 1.45 22.13 o0.000

Excluded from the model were physical neglect, physical abuse, emotional abuse, age, gender, and ethnicity. F 5 5.75, P 5 0.0042, adjusted
R2 5 0.078. P values are uncorrected. Owing to multiple testing, a P value of r0.0125 for the total model is considered statistically significant.
NEO-PI-R, Revised NEO Personality Inventory.

correlated with childhood trauma than was OCS, and sexual abuse, which were identified as potentially being
accounted for much more of the variance. However, we indirectly associated with OCS, mediated through
found that comorbid anxiety symptoms did not account conscientiousness. In our sample, the relationship
for all of the relationship between childhood trauma between emotional trauma and OCS was the most
and OCS, as the association between childhood trauma consistent, and in conjunction with the previous
and probable OCD remained significant when anxiety findings, suggests that although physical or sexual
and PTSD symptoms were included as covariates, trauma may contribute, emotional trauma (abuse or
suggesting an independent relationship between child- neglect) is likely to be the most relevant factor in the
hood trauma and OCS. Similarly, independent of other development of clinically significant OCS. Emotional
comorbid anxiety symptoms, the LOI questions that trauma in particular has been previously associated
are most specific to OCD, and relate to worries about with high rates of other types of anxiety symptoms,
contamination and cleaning, pathological doubt, super- including generalized anxiety, social phobia, and
stitions, and intrusive unwanted thoughts, were the PTSD, as well as with lower levels of psychosocial
most strongly associated with emotional abuse, the functioning in adulthood [Bernet and Stein, 1999;
form of childhood trauma most consistently associated Gibb et al., 2006; Krause et al., 2003; Spertus et al.,
with OCS. 2003; Young et al., 1997]. Studies on parenting style
In addition to a direct association with OCS, we also also provide indirect support for the suggestion that
identified a potential indirect relationship between emotional trauma may be a relevant contributing factor
childhood trauma and OCS, via the personality facet of to the development of clinically significant OCS.
conscientiousness. Conscientiousness, which has been Although there is little research in this area, increased
shown to be associated with OCS in other studies, was levels of OCS have been associated with a variety of
the only personality facet that was significantly maladaptive or dysfunctional parenting styles that are
associated with OCS in our sample, accounting for related to emotional abuse and neglect, including
about 10% of the total variance [Gershuny et al., 2000; parental over-control, over-protection, high expressed
Rector et al., 2002]. In addition, conscientiousness was emotion, and emotional rejection [Barrett et al., 2002;
associated with emotional neglect and with sexual Cavedo and Parker, 1994; Waters and Barrett, 2000].
abuse in our sample, accounting for about 8% of the This research does not support a direct causal role for
total variance. The fact that conscientiousness was only parenting style in the development of clinically
associated with total OCS score and not with probable significant OCS, however, suggesting instead that a
OCD suggests that any role of childhood trauma reciprocal relationship may exist [Barrett et al., 2002;
mediated through conscientiousness is most important Cavedo and Parker, 1994; Parker et al., 1997; Waters
in the development of low to moderate rather than high and Barrett, 2000]. Although outside of the scope of
OCS, and is therefore likely to be associated with this study, the relationships between parenting style,
adaptive rather than maladaptive behavior patterns. It childhood trauma, and the development of OCS and
should be noted, however, that evidence for an indirect later psychiatric psychopathology would be important
relationship between childhood trauma and OCS to investigate.
mediated through personality facets is only implied
by our results; unfortunately, we are unable to assess
direct causal links with our current data. LIMITATIONS
Our results are consistent with the only other Our findings, while intriguing, are exploratory, and
published study of childhood trauma and OCS, which clearly not conclusive. Although our results are
found a positive association between childhood trauma consistent with earlier findings, and the study has a
(most specifically, emotional neglect) and OCD in number of strengths, including the inclusion of male
female subjects [Lochner et al., 2002]. Our study subjects, the large sample size, and the use of a
identified emotional abuse as being associated with Bonferroni correction for multiple testing, it is subject
probable OCD and total OCS score, along with to a number of limitations, the most important of
physical abuse, which was associated with a reduced which is that all assessments were made using self-
risk of probable OCD, and emotional neglect and report questionnaires. We do not have objective
Depression and Anxiety
 Research Article: Childhood Trauma 749

evidence of childhood trauma in our sample, nor Breslau N. 2002. Epidemiologic studies of trauma, posttraumatic
do we have clinical assessments, and therefore do not stress disorder, and other psychiatric disorders. Can J Psychiatry
have information about OCD or other psychiatric 47:923–929.
diagnoses. We also have very small numbers of Briere J, Elliott DM. 2003. Prevalence and psychological sequelae of
individuals with high OCS scores, limiting the inter- self-reported childhood physical and sexual abuse in a general
population sample of men and women. Child Abuse Negl 27:
pretation of data regarding the role of childhood
1205–1222.
trauma and probable OCD. In addition, the study Brynska A, Wolanczyk T. 2005. Epidemiology and phenomenology of
sample, although large, is not representative of the obsessive-compulsive disorder in non-referred young adolescents: A
general population, and may not be representative of Polish perspective. Eur Child Adolesc Psychiatry 14:319–327.
young adults of similar ages who are not in college. Carter AS, Pollock RA, Suvak MK, Pauls DL. 2004. Anxiety and
The fact that we found higher rates of PTSD in males major depression comorbidity in a family study of obsessive-
in our sample, and that there were no significant compulsive disorder. Depress Anxiety 20:165–174.
differences between males and females in the rates of Cavedo LC, Parker G. 1994. Parental bonding instrument. Exploring
sexual abuse is contrary to previous findings, and for links between scores and obsessionality. Soc Psychiatry
further suggests that our sample may not be represen- Psychiatr Epidemiol 29:78–82.
Chu JA, Dill DL. 1991. Dissociation, borderline personality disorder,
tative. The results from this study should be confirmed
and childhood trauma. Am J Psychiatry 148:812–813.
in additional samples, using formal clinical assessments Cloitre M, Scarvalone P, Difede JA. 1997. Posttraumatic stress
for OCD and using other techniques to document disorder, self- and interpersonal dysfunction among sexually
presence and significance of exposure to childhood retraumatized women. J Trauma Stress 10:437–452.
trauma. Cooper J. 1970. The Leyton obsessional inventory. Psychol Med
1:48–64.
Costa PT, McCrae RR. 1992. Revised NEO Personality Inventory
(NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) profes-
REFERENCES sional manual. Odessa, Florida: Psychological Assessment
Angst J, Gamma A, Endrass J, Goodwin R, Ajdacic V, Eich D, Rossler Resources.
W. 2004. Obsessive-compulsive severity spectrum in the commu- Donahue RD. 2005. The structural relationships among generalized
nity: Prevalence, comorbidity, and course. Eur Arch Psychiatry anxiety, obsessions-compulsions, and depression at the syndrome
Clin Neurosci 254:156–164. and sub-syndrome level. Behav Res Ther 43:1587–1609.
Barrett P, Shortt A, Healy L. 2002. Do parent and child behaviours Duran B, Malcoe LH, Sanders M, Waitzkin H, Skipper B, Yager J.
differentiate families whose children have obsessive-compulsive 2004. Child maltreatment prevalence and mental disorders out-
disorder from other clinic and non-clinic families? J Child Psychol comes among American Indian women in primary care. Child
Psychiatry 43:597–607. Abuse Negl 28:131–145.
Berg CZ, Whitaker A, Davies M, Flament MF, Rapoport JL. 1988. Evans DW, Gray FL, Leckman JF. 1999. The rituals, fears and
The survey form of the Leyton Obsessional Inventory-Child phobias of young children: insights from development, psycho-
Version: Norms from an epidemiological study. J Am Acad Child pathology and neurobiology. Child Psychiatry Hum Dev 29:
Adolesc Psychiatry 27:759–763. 261–276.
Feygin DL, Swain JE, Leckman JF. 2006. The normalcy of neurosis:
Berg CZ, Rapoport JL, Whitaker A, Davies M, Leonard H, Swedo
Evolutionary origins of obsessive-compulsive disorder and related
SE, Braiman S, Lenane M. 1989. Childhood obsessive compulsive
behaviors. Prog Neuropsychopharmacol Biol Psychiatry.
disorder: A two-year prospective follow-up of a community
Fonagy P. 2000. Attachment and borderline personality disorder.
sample. J Am Acad Child Adolesc Psychiatry 28:528–533.
J Am Psychoanal Assoc 48:1129–1146; discussion 1175–1187.
Bernet CZ, Stein MB. 1999. Relationship of childhood maltreatment
Ford JD, Kidd P. 1998. Early childhood trauma and disorders of
to the onset and course of major depression in adulthood. Depress
extreme stress as predictors of treatment outcome with chronic
Anxiety 9:169–174.
posttraumatic stress disorder. J Trauma Stress 11:743–761.
Bernstein DP, Fink L. 1998. Childhood Trauma Questionnaire: A
Ford JD, Racusin R, Ellis CG, Daviss WB, Reiser J, Fleischer A,
retrospective self-report manual. San Antonio, TX: The Psycho-
Thomas J. 2000. Child maltreatment, other trauma exposure, and
logical Corporation. posttraumatic symptomatology among children with oppositional
Bernstein DP, Stein JA, Handelsman L. 1998. Predicting personality defiant and attention deficit hyperactivity disorders. Child Maltreat
pathology among adult patients with substance use disorders: 5:205–217.
Effects of childhood maltreatment. Addict Behav 23:855–868. Franzblau SH. 1997. The phenomenology of ritualized and repeated
Bernstein DP, Stein JA, Newcomb MD, Walker E, Pogge D, behaviors in nonclinical populations in the United States. Cult
Ahluvalia T, Stokes J, Handelsman L, Medrano M, Desmond D, Divers Ment Health 3:259–272.
et al. 2003. Development and validation of a brief screening version Gearon JS, Kaltman SI, Brown C, Bellack AS. 2003. Traumatic life
of the Childhood Trauma Questionnaire. Child Abuse Negl events and PTSD among women with substance use disorders and
27:169–190. schizophrenia. Psychiatr Serv 54:523–528.
Bienvenu OJ, Samuels JF, Costa PT, Reti IM, Eaton WW, Nestadt Gershuny BS, Sher KJ, Rossy L, Bishop AK. 2000. Distinguishing
G. 2004. Anxiety and depressive disorders and the five-factor manifestations of anxiety: How do personality traits of compulsive
model of personality: A higher- and lower-order personality trait checkers differ from other anxious individuals? Behav Res Ther
investigation in a community sample. Depress Anxiety 20:92–97. 38:229–241.
Bierer LM, Yehuda R, Schmeidler J, Mitropoulou V, New AS, Gershuny BS, Baer L, Radomsky AS, Wilson KA, Jenike MA. 2003.
Silverman JM, Siever LJ. 2003. Abuse and neglect in childhood: Connections among symptoms of obsessive-compulsive disorder
Relationship to personality disorder diagnoses. CNS Spectr and posttraumatic stress disorder: A case series. Behav Res Ther
8:737–754. 41:1029–1041.

Depression and Anxiety

750 Mathews et al.

Gibb BE, Chelminski I, Zimmerman M. 2006. Childhood emotional, Maggini C, Ampollini P, Gariboldi S, Cella PL, Peqlizza L, Marchesi
physical, and sexual abuse, and diagnoses of depressive and anxiety C. 2001. The Parma High School Epidemiological Survey:
disorders in adult psychiatric outpatients. Depress Anxiety. obsessive-compulsive symptoms. Acta Psychiatr Scand
Grabe HJ, Meyer C, Hapke U, Rumpf HJ, Freyberger HJ, Dilling H, 103:441–446.
John U. 2000. Prevalence, quality of life and psychosocial function Mancini F, Gragnani A, Orazi F, Pietrangeli MG. 1999. Obsessions
in obsessive-compulsive disorder and subclinical obsessive-com- and compulsions: Normative data on the Padua Inventory from an
pulsive disorder in northern Germany. Eur Arch Psychiatry Clin Italian non-clinical adolescent sample. Behav Res Ther 37:919–925.
Neurosci 250:262–268. Mathews CA, Jang KL, Hami S, Stein MB. 2004. The structure of
Grabe HJ, Meyer C, Hapke U, Rumpf HJ, Freyberger HJ, Dilling H, obsessionality among young adults. Depress Anxiety 20:77–85.
John U. 2001. Lifetime-comorbidity of obsessive-compulsive Nemeroff CB. 2004. Neurobiological consequences of childhood
disorder and subclinical obsessive-compulsive disorder in North- trauma. J Clin Psychiatry 65(Suppl 1):18–28.
ern Germany. Eur Arch Psychiatry Clin Neurosci 251:130–135. Paivio SC, Cramer KM. 2004. Factor structure and reliability of the
Haller DL, Miles DR. 2004. Personality disturbances in drug- Childhood Trauma Questionnaire in a Canadian undergraduate
dependent women: Relationship to childhood abuse. Am J Drug student sample. Child Abuse Negl 28:889–904.
Alcohol Abuse 30:269–286. Parker G, Roussos J, Hadzi-Pavlovic D, Mitchell P, Wilhelm K,
Heim C, Nemeroff CB. 2001. The role of childhood trauma in the Austin MP. 1997. The development of a refined measure of
neurobiology of mood and anxiety disorders: Preclinical and dysfunctional parenting and assessment of its relevance in patients
clinical studies. Biol Psychiatry 49:1023–1039. with affective disorders. Psychol Med 27:1193–1203.
Heim C, Newport DJ, Wagner D, Wilcox MM, Miller AH, Rayworth BB, Wise LA, Harlow BL. 2004. Childhood abuse and risk
Nemeroff CB. 2002. The role of early adverse experience and of eating disorders in women. Epidemiology 15:271–278.
adulthood stress in the prediction of neuroendocrine stress Rector NA, Hood K, Richter MA, Bagby RM. 2002. Obsessive-
reactivity in women: A multiple regression analysis. Depress compulsive disorder and the five-factor model of personality:
Anxiety 15:117–125. Distinction and overlap with major depressive disorder. Behav Res
Hubbard J, Realmuto GM, Northwood AK, Masten AS. 1995. Ther 40:1205–1219.
Comorbidity of psychiatric diagnoses with posttraumatic stress Richter MA, Cox BJ, Direnfeld DM. 1994. A comparison of three
disorder in survivors of childhood trauma. J Am Acad Child assessment instruments for obsessive-compulsive symptoms.
Adolesc Psychiatry 34:1167–1173. J Behav Ther Exp Psychiatry 25:143–147.
Huppert JD, Moser JS, Gershuny BS, Riggs DS, Spokas M, Filip J, Roussos A, Francis K, Koumoula A, Richardson C, Kabakos C,
Hajcak G, Parker HA, Baer L, Foa EB. 2005. The relationship Kiriakidou T, Karagianni S, Karamolegou K. 2003. The Leyton
between obsessive-compulsive and posttraumatic stress symptoms obsessional inventory—child version in Greek adolescents: Stan-
in clinical and non-clinical samples. J Anxiety Disord 19:127–136. dardization in a national school-based survey and two-year follow-
Hussey JM, Chang JJ, Kotch JB. 2006. Child maltreatment in the up. Eur Child Adolesc Psychiatry 12:58–66.
United States: Prevalence, risk factors, and adolescent health Ruggiero KJ, Del Ben K, Scotti JR, Rabalais AE. 2003. Psychometric
consequences. Pediatrics 118:933–942. properties of the PTSD Checklist-Civilian Version. J Trauma
Johnson DM, Sheahan TC, Chard KM. 2003. Personality disorders, Stress 16:495–502.
coping strategies, and posttraumatic stress disorder in women with Samuels J, Nestadt G, Bienvenu OJ, Costa PT Jr, Riddle MA, Liang
histories of childhood sexual abuse. J Child Sex Abus 12:19–39. KY, Hoehn-Saric R, Grados MA, Cullen BA. 2000. Personality
Kendler KS, Kuhn JW, Prescott CA. 2004. Childhood sexual abuse, disorders and normal personality dimensions in obsessive-compul-
stressful life events and risk for major depression in women. sive disorder. Br J Psychiatry 177:457–462.
Psychol Med 34:1475–1482. Scher CD, Stein MB, Asmundson GJ, McCreary DR, Forde DR.
King NJ, Myerson NN, Inglis S, Jenkins M, Ollendick TH. 1995. 2001. The childhood trauma questionnaire in a community
Obsessive-compulsive behaviour in children and adolescents: A sample: Psychometric properties and normative data. J Trauma
cross-sectional Australian study. J Paediatr Child Health 31: Stress 14:843–857.
527–531. Spertus IL, Yehuda R, Wong CM, Halligan S, Seremetis SV. 2003.
Krause ED, Mendelson T, Lynch TR. 2003. Childhood emotional Childhood emotional abuse and neglect as predictors of psycho-
invalidation and adult psychological distress: The mediating role of logical and physical symptoms in women presenting to a primary
emotional inhibition. Child Abuse Negl 27:199–213. care practice. Child Abuse Negl 27:1247–1258.
Langeland W, Draijer N, van den Brink W. 2004. Psychiatric Spielberger CD, Gorsuch RL, Lushene RE, Vagg PR, Jacobs GA.
comorbidity in treatment-seeking alcoholics: The role of child- 1970. The State-Trait Anxiety Inventory: STAI. Palo Alto, CA:
hood trauma and perceived parental dysfunction. Alcohol Clin Exp Consulting Psychologist Press.
Res 28:441–447. Stanley MA, Prather RC, Beck JG, Brown TC, Wagner AL, Davis
Leonard HL, Goldberger EL, Rapoport JL, Cheslow DL, Swedo SE. ML. 1993. Psychometric analysis of the Leyton Obsessional
1990. Childhood rituals: Normal development or obsessive- Inventory in patients with obsessive-compulsive and other anxiety
compulsive symptoms? J Am Acad Child Adolesc Psychiatry disorders. Psychol Assessment 5:187–192.
29:17–23. Thombs BD, Bernstein DP, Ziegelstein RC, Scher CD, Forde DR,
Levitan RD, Parikh SV, Lesage AD, Hegadoren KM, Adams M, Walker EA, Stein MB. 2006. An evaluation of screening questions
Kennedy SH, Goering PN. 1998. Major depression in individuals for childhood abuse in 2 community samples: Implications for
with a history of childhood physical or sexual abuse: Relationship clinical practice. Arch Intern Med 166:2020–2026.
to neurovegetative features, mania, and gender. Am J Psychiatry Thomsen PH. 1993. Obsessive-compulsive disorder in children and
155:1746–1752. adolescents. Self-reported obsessive-compulsive behaviour in
Lochner C, du Toit PL, Zungu-Dirwayi N, Marais A, van pupils in Denmark. Acta Psychiatr Scand 88:212–217.
Kradenburg J, Seedat S, Niehaus DJ, Stein DJ. 2002. Childhood Tukel R, Polat A, Ozdemir O, Aksut D, Turksoy N. 2002. Comorbid
trauma in obsessive-compulsive disorder, trichotillomania, and conditions in obsessive-compulsive disorder. Compr Psychiatry
controls. Depress Anxiety 15:66–68. 43:204–209.

Depression and Anxiety

 Research Article: Childhood Trauma 751

Waters TL, Barrett PM. 2000. The role of the family in childhood disorder in borderline, schizotypal, avoidant, and obsessive-
obsessive-compulsive disorder. Clin Child Fam Psychol Rev compulsive personality disorders: Findings from the collaborative
3:173–184. longitudinal personality disorders study. J Nerv Ment Dis
Weathers FW, Litz BT, Herman DS, Huska JA, Keane TM. 1991. 190:510–518.
The PTSD Checklist-Civilian Version (PCLC). National Center Young EA, Abelson JL, Curtis GC, Nesse RM. 1997. Childhood
for PTSD, Boston Veteran’s Affairs Medical Center, 150 S. adversity and vulnerability to mood and anxiety disorders. Depress
Huntington Ave, Boston, MA: F.W. Weathers. Anxiety 5:66–72.
Westen D, Dutra L, Shedler J. 2005. Assessing adolescent personality Zlotnick C, Mattia J, Zimmerman M. 2001. Clinical features of
pathology. Br J Psychiatry 186:227–238. survivors of sexual abuse with major depression. Child Abuse Negl
Windle M, Windle RC, Scheidt DM, Miller GB. 1995. Physical and 25:357–367.
sexual abuse and associated mental disorders among alcoholic Zohar AH, Bruno R. 1997. Normative and pathological obsessive-
inpatients. Am J Psychiatry 152:1322–1328. compulsive behavior and ideation in childhood: A question of
Yen S, Shea MT, Battle CL, Johnson DM, Zlotnick C, timing. J Child Psychol Psychiatry 38:993–999.
Dolan-Sewell R, Skodol AE, Grilo CM, Gunderson JG, Sanislow Zohar AH, Felz L. 2001. Ritualistic behavior in young children.
CA, et al. 2002. Traumatic exposure and posttraumatic stress J Abnorm Child Psychol 29:121–128.

Depression and Anxiety