Accelerat ing t he world's research.

Reported pathological childhood

experiences associated with the
development of borderline
personality disorder
Ruth Lewis

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Prevalence and severit y of childhood adversit y in adolescent s wit h BPD, psychiat rically healt …
Blaise Aguirre

Axis I Comorbidit y of Borderline Personalit y Disorder

Elyse Dubo

Clinical Feat ures and Impairment in Women wit h Borderline Personalit y Disorder (BPD) wit h Post t rau…
Donna Bender
 ZANARINI,
PATHOLOGICAL
WILLIAM
Am J Psychiatry CHILDHOOD
S, LEWIS,
154:8, August ET
EXPERIENCES
AL.
1997

Reported Pathological Childhood Experiences Associated

With the Development of Borderline Personality Disorder

Mary C. Zanarini, Ed.D., Amy A. Williams, B.S., Ruth E. Lewis, Ph.D.,

R. Bradford Reich, M.D., Soledad C. Vera, M.A., Margaret F. Marino, Ph.D.,
Alexandra Levin, B.A., Lynne Yong, B.A., and Frances R. Frankenburg, M.D.

O bjective: T he purpose of this study w as to assess a full range of pathological childhood

ex periences reported by patients w ith criteria-defined borderline personality disorder and com-
parison patients w ith other personality disorders. M ethod: T he pathological childhood ex pe-
riences reported by 467 inpatients w ith personality disorders w ere assessed by interview ers
w ho used a semistructured research interview and w ere blind to clinical diagnosis. R esults: O f
the 358 patients w ith borderline personality disorder, 91% reported having been abused, and
92% reported having been neglected, before the age of 18. T he borderline patients w ere sig-
nificantly more lik ely than the 109 patients w ith other personality disorders to report having
been emotionally and physically abused by a caretak er and sex ually abused by a noncaretaker.
T hey w ere also significantly m ore lik ely to report having a caretak er w ithdraw from them
emotionally, treat them inconsistently, deny their thoughts and feelings, place them in the role
of a parent, and fail to provide them w ith needed protection. T he borderline patients w ith a
childhood history of sexual abuse w ere significantly m ore lik ely than those w ithout such a
history to report having ex perienced all but one of the types of abuse and neglect studied.
W hen all significant risk factors w ere considered together, four w ere found to be significant
predictors of a borderline diagnosis: female gender, sex ual abuse by a m ale noncaretaker,
emotional denial by a m ale caretak er, and inconsistent treatm ent by a fem ale caretak er. Con-
clusions: T he results suggest that sex ual abuse is neither necessary nor sufficient for the devel-
opm ent of borderline personality disorder and that other childhood ex periences, particularly
neglect by caretak ers of both genders, represent significant risk factors.
(Am J Psychiatry 1997; 154:1101–1106)

I nitial studies of the childhood experiences of pa-

tients with borderline personality disorder focused
on the etiological role of early separations (1–4) and
(12–19). The studies that also investigated the preva-
lence of childhood sexual abuse by noncaretakers
found that overall rates of childhood sexual abuse re-
various forms of disturbed parental involvement (3– ported by borderline patients ranged from 16% to 71%
10). H owever, more recent studies have focused on the (11, 13–18). While most of the relevant studies found
role of childhood abuse in the etiology of borderline that a childhood history of physical abuse is commonly
personality disorder (11–19). These studies found that reported by borderline patients, only two studies found
both physical and sexual abuse are commonly reported that such a history discriminated borderline patients
by borderline patients. M ore specifically, they found from patients with other types of personality disorder
that 10% –73% of borderline patients report having (12, 14). In contrast, all of the relevant studies found
been physically abused by a parent or other adult care- that a significantly higher percentage of borderline pa-
taker (11–19) and that 0% –33% report having had an tients than near-neighbor comparison subjects reported
incestuous relationship with a full-time adult caretaker a childhood history of sexual abuse (11–15, 18, 19).
These last results have been interpreted by some to
mean that sexual abuse is the main etiological factor in
Received Sept. 25, 1996; revisions received Dec. 31, 1996, and Feb. the development of borderline personality disorder
7 and M arch 25, 1997; accepted April 4, 1997. From the Laboratory (20). They have also been interpreted to mean that pa-
for the Study of Adult Development, M cLean H ospital; and the De- tients meeting current criteria for borderline personality
partment of Psychiatry, H arvard M edical School, Boston. Address
reprint requests to Dr. Z anarini, M cLean H ospital, 115 M ill St., Bel- disorder might be better conceptualized as suffering
mont, M A 02178. from a chronic form of posttraumatic stress disorder
Supported in part by N IM H grant M H -47588. rather than borderline personality disorder (20, 21).

A m J Psychiatry 154:8, A ugust 1997 1101

PATH O LO GICAL CH ILDH O O D EXPERIEN CES

The study we report here tried to place the role of RESULTS

childhood sexual abuse in the etiology of borderline
personality disorder in perspective by studying a wide Five hundred four patients were given diagnostic in-
range of pathological childhood experiences reported terviews, and 467 (92.7% ) also participated in inter-
by a group of 358 criteria-defined borderline patients views about their childhood. Of these 467 patients, 358
and 109 comparison subjects with DSM -III-R axis II met both the DIB-R and the DSM -III-R criteria for bor-
diagnoses. It improved upon the design of earlier stud- derline personality disorder, and 109 met the DSM -III-
ies because of the large size of the patient groups, the R criteria for at least one axis II disorder other than
rigor with which they were diagnosed, the substantially borderline personality.
more inclusive list of pathological childhood experi- There were no significant demographic differences
ences that were assessed, and the fact that these assess- between the 467 patients who had a childhood inter-
ments were made blind to diagnostic status with the use view and the 37 (7.3% ) who did not. However, a signifi-
of a semistructured interview of demonstrated reliabil- cantly higher percentage of the borderline patients
ity. It also improved upon the design of most earlier (94.4% , N =338) than of the comparison subjects
studies by using multivariate analyses to assess the rela- (87.2% , N =95) had a childhood interview (χ2=6.25,
tive contributions of these childhood factors. df=1, p= 0.01). This was probably because the borderline
patients tended to have slightly longer hospitalizations
than the comparison subjects, and most of the 37 pa-
M ETH O D tients without a childhood interview had been discharged
from the hospital before they could be interviewed.
All subjects were inpatients at M cLean H ospital (Belmont, M ass.) The borderline patients were found to be quite simi-
who were admitted between M arch 1991 and December 1995. Each
patient was initially screened to determine that he or she 1) was be-
lar to the comparison subjects in terms of race (11% of
tween the ages of 18 and 50 years, 2) had normal or better intelli- each group was nonwhite) and socioeconomic back-
gence, 3) had no history or current symptoms of a serious organic ground as measured by the 5-point H ollingshead-
condition or major psychotic disorder (i.e., schizophrenia or bipolar Redlich Scale (25) (1=highest status, 5=lowest). The
I disorder), and 4) had been given a definite or probable axis II diag- mean socioeconomic status of the borderline patients
nosis by the admitting physician.
Written informed consent was obtained from each patient. Three was 2.6 (SD=1.3), and that of the comparison subjects
semistructured diagnostic interviews were then administered to the was 2.7 (SD=1.3). H owever, the borderline patients
patient by one of five interviewers who were blind to clinical diagno- were found to be slightly, but significantly, younger
sis. These instruments were 1) the Structured Clinical Interview for than the comparison subjects (mean age=27.6 years,
DSM -III-R (22)—a semistructured interview designed to assess the
lifetime prevalence of many of the most common axis I disorders de-
SD=6.8, versus mean=29.3 years, SD=9.1; t=2.10, df=
scribed in DSM -III-R, 2) the Revised Diagnostic Interview for Border- 465, p<0.04). In addition, a significantly higher per-
lines (DIB-R) (23)—a semistructured interview that can reliably dis- centage of the borderline patients (77.1% , N =276) than
tinguish clinically diagnosed borderline patients from those with of the comparison subjects (56.0% , N =61) were female
other DSM -III and DSM -III-R axis II disorders, and 3) the Diagnostic (χ2 =17.54, df=1, p=0.00003).
Interview for DSM -III-R Personality Disorders (24)—a semistruc-
tured interview that reliably assesses the presence of the 13 axis II Because a significantly higher percentage of the bor-
disorders described in DSM -III-R. All five interviewers had been derline patients were women, we conducted two sub-
trained in the administration and scoring of these instruments by the analyses to control for gender. Since the results of these
first author (M.C.Z .), who is one of the developers of both the DIB-R analyses were basically the same as those obtained for
and the Diagnostic Interview for DSM-III-R Personality Disorders.
Adequate levels of interrater reliability had been obtained during this
the mixed-gender group, we decided to present only our
training period (e.g., kappa=0.85 or higher for the DIB-R and DSM - overall results. H owever, we controlled for gender in
III-R diagnoses of borderline personality disorder). the multivariate analysis.
Information concerning pathological childhood experiences was Table 1 shows the reported rates of abuse and neglect
assessed by one of seven clinically experienced interviewers (A.A.W., of the borderline patients and the patients with other
R.E.L., R.B.R., S.C.V., M .F.M ., A.L., L.Y.), each of whom was blind
to all other data concerning each patient, including his or her current axis II diagnoses. At the Bonferroni-corrected alpha
diagnostic status. Pathological childhood experiences that were re- level of p<0.003, a significantly higher percentage of the
ported to have occurred before the age of 18 were assessed with the borderline patients reported each of the childhood ex-
Revised Childhood Experiences Q uestionnaire—a semistructured in- periences assessed except caretaker’s verbal abuse, care-
terview whose psychometric properties have been described else-
where (19). Briefly, this instrument inquires about four forms of
taker’s sexual abuse, physical neglect by a caretaker,
abuse and seven forms of neglect by full-time caretakers, sexual abuse and lack of a real relationship with a caretaker.
by noncaretakers, and 12 types of separations from full-time caretak- Ab ou t eq ual p ercent ages of borderline patients
ers that lasted 1 month or more. For an item to be given a positive (29.1% , N =104) and comparison subjects (22.9% , N=
rating, detailed information concerning the event in question had to 25) reported having experienced a prolonged separa-
be provided. In addition, the rates of physical and sexual abuse re-
ported in this study are not the result of psychotherapeutic efforts and tion from a caretaker before the age of 6 (χ2 =1.27, df=1,
thus do not represent “ recovered” memories. n.s.). The mean numbers of early childhood separations
Between-group comparisons involving categorical data were com- reported by the borderline patients and the comparison
puted by means of the chi-square statistic corrected for continuity; subjects were also basically the same: the borderline pa-
between-group comparisons involving continuous data (age and so-
cioeconomic status) were computed by means of Student’s t test. The
tients reported a mean of 2.3 (SD=2.3) early separa-
Bonferroni correction for multiple comparisons was applied where tions, while the comparison subjects reported a mean of
appropriate. 1.9 (SD=0.63) such separations (t=1.51, df=465, n.s.).

1102 A m J Psychiatry 154:8, A ugust 1997

 Z AN ARIN I, WILLIAM S, LEWIS, ET AL.

TABLE 1 . Pathological Childhood Experiences Reported by Patients With Borderline Personality Disorder and Patients With Other Personality
Disorders

Patients With Patients With

Borderline O ther
Personality Personality
Disorder Disorders
(N =358) (N =109) Analysis
Childhood Experience N % N % χ2 (df=1) pa
Caretaker’s emotional abuse 260 72.6 56 51.4 16.29 0.00005
Caretaker’s verbal abuse 273 76.3 68 62.4 7.47 0.006
Caretaker’s physical abuse 211 58.9 37 33.9 19.97 0.00001
Caretaker’s sexual abuse 98 27.4 17 15.6 5.63 0.02
N oncaretaker’s sexual abuse 200 55.9 26 23.9 33.02 0.00001
Any sexual abuse 220 61.5 35 32.1 27.85 0.00001
Any abuse 327 91.3 80 73.4 22.46 0.00001
Caretaker’s physical neglect 94 26.3 14 12.8 7.72 0.005
Caretaker’s emotional withdrawal 196 54.7 35 32.1 16.24 0.00006
Caretaker’s inconsistent treatment 187 52.2 34 31.2 14.01 0.0002
Caretaker’s denial of patient’s feelings 252 70.4 49 45.0 22.50 0.0001
Lack of real relationship with caretaker 250 69.8 61 56.0 6.62 0.01
Caretaker’s placing patient in parental role 211 58.9 43 39.4 12.02 0.0005
Caretaker’s failure to protect patient 199 55.6 36 33.0 16.12 0.00006
Any neglect 330 92.2 82 75.2 21.50 0.00001
a Application of the Bonferroni correction for multiple comparisons indicates statistical significance at p<0.003.

TABLE 2 . Pathological Childhood Experiences Reported by Sexually Abused and Not Sexually Abused Patients With Borderline Personality
Disorder

Sexually Abused N ot Sexually

Patients Abused Patients
(N =220) (N =138) Analysis
Childhood Experience N % N % χ2 (df=1) pa
Caretaker’s emotional abuse 180 81.8 80 58.0 23.07 0.00001
Caretaker’s verbal abuse 181 82.3 92 66.7 10.56 0.001
Caretaker’s physical abuse 154 70.0 57 41.3 27.68 0.00001
Any abuse 204 92.7 107 77.5 15.85 0.00007
Caretaker’s physical neglect 76 34.5 18 13.0 19.15 0.00001
Caretaker’s emotional withdrawal 138 62.7 58 42.0 13.84 0.0002
Caretaker’s inconsistent treatment 134 60.9 53 38.4 16.32 0.00005
Caretaker’s denial of patient’s feelings 170 77.3 82 59.4 12.13 0.0005
Lack of real relationship with caretaker 165 75.0 85 61.6 6.61 0.01
Caretaker’s placing patient in parental role 151 68.6 60 43.5 21.15 0.00001
Caretaker’s failure to protect patient 156 70.9 43 31.2 52.68 0.00001
Any neglect 212 96.4 118 85.5 12.40 0.0004
a Application of the Bonferroni correction for multiple comparisons indicates statistical significance at p<0.004.

Table 2 shows the rates of abuse and neglect reported than that reported by those who were not sexually
by the 220 borderline patients with a childhood history abused (mean=0.65, SD=1.9) (t=1.08, df=356, n.s.).
of sexual abuse and the 138 without such a history. At These numerous univariate analyses reveal that re-
the Bonferroni-corrected alpha level of p<0.004, a sig- ported sexual abuse almost always occurred in conjunc-
nificantly higher percentage of the sexually abused bor- tion with at least one other type of reported abuse or
derline patients than of the borderline patients who neglect. A forward stepwise logistic regression with di-
were not sexually abused reported each of the child- agnostic status (borderline personality disorder versus
hood experiences assessed except lack of a real relation- other personality disorders) as the dependent variable
ship with a caretaker. was conducted next, to determine the relative impor-
A significantly higher percentage of the sexually tance of these numerous pathological childhood expe-
abused borderline patients (33.6% , N =74) than of riences. Seventeen independent variables that were sig-
those who were not abused (21.7% , N =30) reported nificant at a Bonferroni-corrected alpha level were
having experienced a prolonged separation from a care- studied: the patient’s gender and eight abuse and ne-
taker before the age of 6 (χ2 =5.26, df=1, p=0.02). The glect variables broken down by gender of the caretaker
mean number of early childhood separations reported (or in the case of noncaretaker sexual abuse, the gender
by the sexually abused borderline patients (mean=0.95, of the person who was not a caretaker).
SD=2.9) was also slightly but nonsignificantly higher As table 3 shows, four factors were significantly as-

A m J Psychiatry 154:8, A ugust 1997 1103

PATH O LO GICAL CH ILDH O O D EXPERIEN CES

TABLE 3 . Results of Forward Stepwise Logistic Regression to Determine Risk 33% reported by most of the studies that have
Factors Associated With the Diagnosis of Borderline Personality Disordera assessed the prevalence of this type of childhood
O dds experience (12–14, 17–19). We also found an
Variable Beta SE df p Ratio overall rate of childhood sexual abuse of 61.5% .
Female gender 0.3142 0.1245 1 0.01 1.4
This rate is quite consistent with the 40% –71%
M ale noncaretaker’s sex- reported for most other groups of borderline pa-
ual abuse 0.5579 0.1344 1 0.0001 1.7 tients (11, 13–15, 17, 18). The rates of caretakers’
Female caretaker’s incon- and overall sexual abuse that we found are much
sistent treatment 0.2833 0.1384 1 0.04 1.3 higher than the 0% –16% reported by Salzman
M ale caretaker’s denial of
patient’s feelings 0.4105 0.1243 1 0.001 1.5 et al. (16). H owever, these differences are not
a The dependent variable was a diagnosis of borderline personality disorder
surprising, as we were studying severely impaired
inpatients and Salzman et al. were studying symp-
according to the Revised Diagnostic Interview for Borderlines (23); N =358
with the diagnosis, and N =109 without the diagnosis. tomatic volunteers who had never been hospital-
ized for psychiatric reasons and who had not been
self-destructive or suicidal for 4 years before entry
sociated with an adult diagnosis of borderline person- into their study.
ality disorder: being female, a childhood history of sex- The second major finding of this study was that child-
ual abuse by a male who was not a caretaker, denial of hood experiences of both abuse and neglect were sig-
one’s thoughts and feelings by a male caretaker, and nificantly more common among the borderline patients
inconsistent treatment by a female caretaker. In terms than among the comparison subjects. M ore specifically,
of these four variables, a patient’s risk of being diag- emotional abuse, physical abuse, sexual abuse by non-
nosed as having borderline personality disorder was caretakers, overall sexual abuse, emotional withdrawal
about one and one-half times greater than the risk for by a caretaker, inconsistent treatment by a caretaker,
a patient who was male or who did not report having denial of the patient’s thoughts and feelings by a care-
had these childhood experiences. taker, being placed in the role of a parent by a care-
taker, and failure of a caretaker to provide needed
protection were reported by significantly higher per-
DISCUSSIO N centages of the borderline patients than of the compari-
son subjects (at the Bonferroni-corrected alpha level of
Four important results emerged from this study. p<0.003). The results of this study are consistent with
First, we found that childhood experiences of both those of earlier studies which found that childhood
abuse and neglect were basically ubiquitous among our experiences of both abuse and neglect are reported by a
borderline patients. M ore specifically, we found that significantly higher percentage of borderline patients
91% of the borderline patients reported some type of than comparison subjects (12, 18, 19).
childhood abuse, and 92% reported some type of child- The third major finding of this study was that the
hood neglect. In terms of specific forms of abuse, about sexually abused borderline patients seemed to come
75% of the borderline patients reported a childhood from more chaotic environments than the borderline
history of emotional abuse (i.e., frequent experiences of patients who were not sexually abused. The sexually
being shamed or humiliated, being frustrated by being abused patients were significantly more likely (at the
given mixed messages, or being put in impossible situ- Bonferroni-corrected alpha level of p<0.004) to report
ations) or verbal abuse, while about 60% reported a having been emotionally, verbally, and physically
childhood history of physical or sexual abuse. In terms abused. They were also significantly more likely (at the
of specific forms of neglect, about 70% of the border- Bonferroni-corrected alpha level of p<0.004) to report
line patients reported a caretaker’s denial of their having been physically neglected by a caretaker, having
thoughts or feelings and reported that they lacked a real a caretaker withdraw from them emotionally, being
emotional relationship with one or more caretakers. treated inconsistently by a caretaker, having a caretaker
About 60% reported often being put in the position of deny their thoughts and feelings, being placed in the
a parent, where they felt the need to take care of a care- role of a parent by a caretaker, and having a caretaker
taker or other family members. About one-half re- fail to provide needed protection. This, to the best of
ported a caretaker’s withdrawing from them emotion- our knowledge, represents a new finding. It also empha-
ally, treating them inconsistently, and failing to provide sizes that sexual abuse does not typically occur in fami-
them with needed protection; and about one-fourth re- lies where the preborderline child is otherwise well
ported being physically neglected during childhood. cared for but, rather, in a context of ongoing abuse and
The results of this study are consistent with those of neglect. This finding suggests, in addition, that the
earlier studies which found that a high percentage of childhood sexual abuse reported by borderline patients
borderline patients report having been abused and/or may represent a marker of the severity of the familial
neglected during childhood (11–19). In terms of sexual dysfunction they experienced, as well as being a trau-
abuse, we found that 27.4% of the borderline patients matic event or series of events in itself.
reported being sexually abused by a caretaker during The fourth major finding of this study is that when
childhood. This rate falls within the range of the 25% – all significantly different pathological childhood expe-

1104 A m J Psychiatry 154:8, A ugust 1997

 Z AN ARIN I, WILLIAM S, LEWIS, ET AL.

riences are considered together, sexual abuse (particu- found in previous studies (11–19). However, despite
larly, sexual abuse by a noncaretaker) seems to be an the fact that we made strenuous efforts to assess the
important factor in the etiology of borderline personal- validity of reports of childhood experiences of neglect
ity disorder but that other factors (particularly, neglect through the use of detailed case vignettes, the possibility
by caretakers of both genders) also play an important remains that what we are reporting in this area reflects
role. This too, to the best of our knowledge, represents the heightened sensitivity of borderline patients to per-
a new finding. It also underscores the idea that the de- ceived failures in parenting.
velopment of borderline personality disorder seems to In future research there is a need for studies of care-
be linked to the lack of optimal parenting by caretakers fully defined groups of borderline patients that simulta-
of both genders (26). neously investigate basic temperament, a full range of
This finding concerning neglect by both of the par- pathological and protective childhood experiences,
ents and sexual abuse by noncaretakers makes intuitive caretaker psychopathology to which patients may have
clinical sense. It may be that neglect by both of the par- been exposed, and the possible genetic influence of a
ents puts the preborderline child at risk for being sexu- family history of selected psychiatric disorders. Longi-
ally abused by making it clear to potential perpetrators tudinal studies that prospectively assess the effects of
that no one will notice or care if the child is abused. childhood sexual abuse on the long-term adult func-
Such neglect may also put the preborderline child at risk tioning of borderline patients are also needed. In addi-
for being sexually abused by leaving him or her with a tion, studies of high-risk children, as well as twin and/or
strong unmet need for attention, care, and closeness adopted-away studies, are needed. Finally, prospective
that may be misinterpreted and/or manipulated by un- studies that examine the effects of difficult childhood
scrupulous, sexually predatory individuals. experiences in a large group of “ normal” children
H owever, the pathological childhood experiences would help to clarify the role of these experiences in the
noted by our borderline patients are almost certainly development of later personality disorders. In this re-
not the “ whole story” of the etiology of borderline gard, it is important to note that over 30% of the com-
personality disorder. Rather, we believe that a multifac- parison subjects in our study reported a childhood his-
torial model of the etiology of borderline personality tory of physical or sexual abuse.
disorder best captures the complexity of borderline psy-
chopathology (27, 28). This model suggests that bor-
derline symptoms and their comorbid manifestations CO N CLUSIO N S
are the final product of a complex admixture of innate
temperament, difficult childhood experiences, and rela- Overall, the results of this study suggest that child-
tively subtle forms of neurological and biochemical dys- hood sexual abuse is neither necessary nor sufficient for
function (which may be sequelae of these childhood ex- the development of borderline personality disorder. For
periences or innate vulnerabilities). about 60% of our borderline patients, childhood sexual
This model is supported by recent research which has abuse appears to be an important etiological factor.
found that borderline personality disorder is associated H owever, this abuse usually seems to be embedded in
with a temperament characterized by a high degree of an atmosphere of general chaos and neglect by both
neuroticism (i.e., emotional pain) as well as a low de- parents. For the rest of our patients, other forms of
gree of agreeableness (i.e., strong individuality) (29– abuse in conjunction with various forms of neglect
31). Borderline personality disorder has also been probably play a more central etiological role.
found to be the only axis II disorder that is associated In time, the etiology of borderline personality disor-
with a high degree of both harm avoidance (i.e., com- der will be understood more fully. While enormous
pulsivity) and novelty seeking (i.e., impulsivity) (32). In strides have been made in the last decade, research into
addition, a series of studies have found that borderline the multifactorial basis of borderline personality disor-
patients often suffer from difficult-to-diagnose forms of der has barely begun. For now, we suggest patience and
neurological dysfunction (33–36), and biochemical common sense in conceptualizing the role of pathologi-
studies have found decreased serotonergic activity in cal childhood experiences in the etiology of borderline
patients with problematic impulsivity, including pa- personality disorder.
tients with criteria-defined borderline personality disor-
der (37, 38). REFEREN CES
The major limitation of the present study is that its
findings may not be generalizable to less severely dis- 1. Akiskal H S, Chen SE, Davis GC, Puzantian VR, Kashgarian M ,
Bolinger JM : Borderline: an adjective in search of a noun. J Clin
turbed borderline outpatients. Another limitation is Psychiatry 1985; 46:41–48
that all of the information concerning pathological 2. Bradley SJ: The relationship of early maternal separation to bor-
childhood experiences was obtained through retrospec- derline personality in children and adolescents: a pilot study. Am
tive self-report. Although many of the types of child- J Psychiatry 1979; 136:424–426
hood experiences assessed in this study were not as- 3. Soloff PH , M illward JW: Developmental histories of borderline
patients. Compr Psychiatry 1983; 24:574–588
sessed in other studies, those that were, such as physical 4. Walsh F: The family of the borderline patient, in The Borderline
abuse and various types of sexual abuse, were reported Patient. Edited by Grinker RR, Werble B. N ew York, Jason
by percentages of borderline patients similar to those Aronson, 1977, pp 158–177

A m J Psychiatry 154:8, A ugust 1997 1105

PATH O LO GICAL CH ILDH O O D EXPERIEN CES

5. Frank H , H offman N : Borderline empathy: an empirical investi- 23. Z anarini M C, Gunderson JG, Frankenburg FR, Chauncey DL:
gation. Compr Psychiatry 1986; 27:387–395 The Revised Diagnostic Interview for Borderlines: discriminating
6. Frank H , Paris J: Recollections of family experience in borderline BPD from other axis II disorders. J Personality Disorders 1989;
patients. Arch Gen Psychiatry 1981; 38:1031–1034 3:10–18
7. Goldberg RL, M ann LS, Wise TN , Segall EA: Parental qualities 24. Z anarini M C, Frankenburg FR, Chauncey DL, Gunderson JG:
as perceived by borderline personality disorders. H illside J Clin The Diagnostic Interview for Personality Disorders: interrater
Psychiatry 1985; 7:134–140 and test-retest reliability. Compr Psychiatry 1987; 28:467–480
8. Grinker RR, Werble B, Drye RC: The Borderline Syndrome: A 25. H ollingshead AB: Two-Factor Index of Social Position. N ew H a-
Behavioral Study of Ego-Functions. N ew York, Basic Books, ven, Conn, Yale University, 1965
1968 26. Links PS, Boiago I, H uxley G, Steiner M , M itton JE: Sexual abuse
9. Gunderson J, Kerr J, Englund D: The families of borderlines: a and biparental failure as etiologic models in borderline person-
comparative study. Arch Gen Psychiatry 1980; 37:27–33 ality disorder, in Family Environment and Borderline Personality
10. Paris J, Frank H : Perceptions of parental bonding in borderline Disorder. Edited by Links PS. Washington, DC, American Psy-
patients. Am J Psychiatry 1989; 146:1498–1499 chiatric Press, 1990, pp 105–120
11. H erman JL, Perry JC, van der Kolk BA: Childhood trauma in 27. Paris J: Borderline Personality Disorder: A M ultidimensional Ap-
borderline personality disorder. Am J Psychiatry 1989; 146: proach. Washington, DC, American Psychiatric Press, 1994
490–495 28. Z anarini M C, Weinberg E: Borderline personality disorder: im-
12. Links PS, Steiner M, O fford DR, Eppel A: Characteristics of bor- pulsive and compulsive features, in Impulsivity and Compulsiv-
derline personality disorder: a Canadian study. Can J Psychiatry ity. Edited by O ldham JM , H ollander E, Skodol AE. Washing-
1988; 33:336–340 ton, DC, American Psychiatric Press, 1996, pp 37–58
13. O gata SN , Silk KR, Goodrich S, Lohr N E, Westen D, H ill EM : 29. Clarkin JF, H ull JW, Cantor J, Sanderson C: Borderline person-
Childhood sexual and physical abuse in adult patients with bor- ality disorder and personality traits: a comparison of SCID-II
derline personality disorder. Am J Psychiatry 1990; 147:1008– BPD and N EO -PI. Psychol Assessment 1993; 5:472–476
1013 30. Soldz S, Budman S, Demby A, M erry J: Representation of per-
14. Paris J, Z weig-Frank H , Guzder J: Psychological risk factors for sonality disorders in circumplex and five-factor space: explora-
borderline personality disorder in female patients. Compr Psy- tions with a clinical sample. Psychol Assessment 1993; 5:41–52
chiatry 1994; 35:301–305 31. Trull TJ: DSM -III-R personality disorders and the five-factor
15. Paris J, Z weig-Frank H , Guzder J: Risk factors for borderline model of personality: an empirical comparison. J Abnorm Psy-
personality in male outpatients. J N erv M ent Dis 1994; 182: chol 1992; 101:553–560
375–380 32. Svrakic DM , Whitehead C, Przybeck TR, Cloninger CR: Differ-
16. Salzman JP, Salzman C, Wolfson AN , Albanese M , Looper J, ential diagnosis of personality disorders by the seven-factor
O stacher M , Schwartz J, Chinman G, Land W, M iyawaki E: As- model of temperament and character. Arch Gen Psychiatry
sociation between borderline personality structure and history of 1993; 50:991–999
childhood abuse in adult volunteers. Compr Psychiatry 1993; 33. Andrulonis PA, Vogel N G: Comparison of borderline personal-
34:254–257 ity subcategories to schizophrenic and affective disorders. Br J
17. Shearer SL, Peters CP, Q uaytman M S, O gden RL: Frequency and Psychiatry 1984; 144:358–363
correlates of childhood sexual and physical abuse histories in 34. Cowdry RW, Pickar D, Davies R: Symptoms and EEG findings
adult female borderline patients. Am J Psychiatry 1990; 147: in the borderline syndrome. Int J Psychiatry M ed 1985–1986;
214–216 15:201–211
18. Westen D, Ludolph P, M isle B, Ruffins S, Block J: Physical and 35. Gardner D, Lucas PB, Cowdry RW: Soft sign neurological ab-
sexual abuse in adolescent girls with borderline personality dis- normalities in borderline personality disorder and normal con-
order. Am J O rthopsychiatry 1990; 60:55–66 trol subjects. J N erv M ent Dis 1987; 175:177–180
19. Z anarini M C, Gunderson JG, M arino M F, Schwartz EO , Frank- 36. Snyder S, Pitts WM : Electroencephalography of DSM-III border-
enburg FR: Childhood experiences of bor der line patients. line personality disorder. Acta Psychiatr Scand 1984; 69:129–
Compr Psychiatry 1989; 30:18–25 134
20. H erman JL, van der Kolk BA: Traumatic antecedents of border- 37. Coccaro EF, Siever LJ, Klar H M , M aurer G, Cochrane K, Coo-
line personality disorder, in Psychological Trauma. Edited by per TB: Serotonergic studies in patients with affective and per-
van der Kolk BA. Washington, DC, American Psychiatric Press, sonality disorders: correlates with suicidal and impulsive aggres-
1987, pp 111–126 sive behavior. Arch Gen Psychiatry 1989; 46:587–599
21. H erman JL: Trauma and Recovery. New York, Basic Books, 1992 38. H ollander E, Stein DJ, DeCaria CM , Cohen L, Saoud JB, Skodol
22. Spitzer RL, Williams JBW, Gibbon M , First M B: User’s Guide AE, Kellman D, Rosnick L, O ldham JM : Serotonergic sensitivity
for the Structured Clinical Interview for DSM -III-R (SCID). in borderline personality disorder: preliminary findings. Am J
Washington, DC, American Psychiatric Press, 1990 Psychiatry 1994; 151:277–280

1106 A m J Psychiatry 154:8, A ugust 1997