Characteristics of Depressed Patients Who Report Childhood Sexual

Document Sample
Characteristics of Depressed Patients Who Report Childhood Sexual Powered By Docstoc
					       Characteristics of Depressed Patients Who Report
                    Childhood Sexual Abuse

                         Gemma Gladstone, B.A.(Hons), M.A.P.S.,
 Gordon Parker, D.Sc., M.D., Ph.D., F.R.A.N.Z.C.P., Kay Wilhelm, M.D., F.R.A.N.Z.C.P.,
Philip Mitchell, M.D., M.R.C.Psych., F.R.A.N.Z.C.P., and Marie-Paule Austin, F.R.A.N.Z.C.P.

               Objective: Depressed patients who had and had not been exposed to childhood sexual
            abuse were studied to determine differences in severity of depressed mood, lifetime histo-
            ries of anxiety and depression, childhood environment, and disordered personality func-
            tion. Method: Data were obtained from 269 inpatients and outpatients with major depres-
            sion (171 women and 98 men) by means of structured clinical interviews and self-report
            questionnaires. Results: Forty-six of the 269 patients reported childhood sexual abuse; 40
            of these were women. These 40 women were compared with the 131 who did not report
            childhood sexual abuse. The patients who experienced abuse did not differ from those who
            had not on psychiatrist-rated mood severity estimates, but they did have higher self-report
            depression scores. They also evidenced more self-destructive behavior, more personality
            dysfunction, and more overall adversity in their childhood environment. Childhood sexual
            abuse status was associated with more borderline personality characteristics indepen-
            dently of other negative aspects of the patients’ earlier parenting. Childhood sexual abuse
            status was linked strongly to adult self-destructiveness, as was early exposure to maternal
            indifference. Conclusions: Multivariate analyses suggest that depression is unlikely to be
            a direct consequence of childhood sexual abuse. Childhood sexual abuse appears to be
            associated with a greater chance of having experienced a broadly dysfunctional childhood
            home environment, a greater chance of having a borderline personality style, and, in turn,
            a greater chance of experiencing depression in adulthood.
               (Am J Psychiatry 1999; 156:431–437)

T    he long-term psychological sequelae of childhood
sexual abuse have been pursued in clinical and non-
                                                                      A history of childhood sexual abuse is not uncom-
                                                                    monly reported by patients with depressive disorders
clinical samples. Although linked with a plethora of                (13, 14), but inadequate and deprivational parenting
psychological, emotional, and physical disturbances in              have also been identified as important antecedents to
adulthood, childhood sexual abuse appears most dis-                 later depression (15–17). It remains unclear, therefore,
tinctively overrepresented in subjects with depressive              whether childhood sexual abuse provides an indepen-
(1–4), anxiety (5), personality (6–8), and eating disor-            dent risk factor (i.e., alone and beyond a general ad-
ders (9); in those who display self-destructiveness (10,            verse background factor) for onset, severity, and other
11); and in those with low self-esteem and high inter-              parameters of depressive disorder. We report a study
personal sensitivity (12).                                          designed to clarify these issues.

    Received Jan. 16, 1998; revision received June 4, 1998;         METHOD
accepted July 3, 1998. From the Mood Disorders Unit, Prince of
Wales Hospital and the School of Psychiatry, University of New      Subjects and Measures
South Wales; and the Department of Psychiatry, St Vincent’s Hos-
pital, Sydney, N.S.W., Australia. Address reprint requests to Dr.     Our total study group comprised 269 inpatients or outpatients
Parker, Psychiatry Unit, Prince of Wales Hospital, Randwick 2031,   who met DSM-IV criteria for a major depressive episode lasting 24
N.S.W., Australia; (e-mail).                   months or fewer; 171 of the patients were women and 98 were men.
  Supported by National Health and Medical Research Council         All were assessed by one of our mood disorders unit consultants.
Program Grant 753205.                                               Full details of the study are provided elsewhere (18). After complete
  The authors thank Kerrie Eyers, Dusan Hadzi-Pavlovic, Ian         description of the study to the patients, written informed consent
Hickie, and Christine Taylor for study assistance.                  was obtained.

Am J Psychiatry 156:3, March 1999                                                                                                   431

   Patients completed the Beck Depression Inventory (19) as well as         years; 40 (87%) of these 46 patients were women. Be-
a detailed self-report questionnaire covering sociodemographic vari-        cause of the preponderance of childhood sexual abuse
ables, previous treatments, and psychiatric family history. Earlier
parenting experiences were assessed by using the Parental Bonding           among women, we elected to study only women; there-
Instrument (20) and the Measure of Parental Style (18).                     fore, our study group comprised the 40 women who re-
                                                                            ported childhood sexual abuse compared with the 131
Clinical Assessment Interviews                                              women who did not report childhood sexual abuse.
   Patients participated in a semistructured interview with a research        The mean ages of the women who did or did not re-
psychologist followed by a clinical interview with a consultant psy-        port childhood sexual abuse were 39.1 (SD=13.5, range=
chiatrist. The first interview focused on lifetime and current episodes     19–74) and 44.1 (SD=15.0, range=18–77), respectively
of depression and anxiety disorders (formalized by use of the com-          (t=1.9, df=169, p=0.06); the trend for a between-group
puterized Composite International Diagnostic Interview [21]). The
interview determined a wide range of current and lifetime depressive        difference in age led us to control for current age in all
symptoms, together with histories of drug and alcohol use.                  analyses. The women who did or did not report child-
   The interviewing psychiatrist assessed clinical features, generated      hood sexual abuse did not differ significantly on other
DSM-III-R and DSM-IV depressive diagnoses, completed the 21-item            sociodemographic variables examined, including mean
Hamilton depression rating scale (22) and the DSM-III-R Global As-          number of years of education, employment, marital sta-
sessment of Functioning Scale, rated severity of depression for the cur-
rent episode on a 0–3-point scale, and asked a set of structured ques-      tus, and whether they were currently involved in an inti-
tions about past and present suicidal or self-injurious behaviors.          mate relationship. Of the 40 women who experienced
                                                                            childhood sexual abuse, five (13%) had been abused by
Childhood Environment                                                       a parent only, 28 (70%) had been abused by someone
   The interviewing psychiatrist assessed specific characteristics of       other than a parent, and seven (18%) had been abused
the patient’s childhood environment in the first 16 years. Using 3-         by both a parent and someone other than a parent.
point rating scales (0=no, 1=possibly/likely, 2=definitely), the psychi-
atrist judged whether the patient had been exposed to a number of           Lifetime and Current Depression and Anxiety
early adverse parenting characteristics, including whether the patient
had experienced sexual abuse by a parent or by another person be-              Few clinical features distinguished patients with and
fore age 16. Patients were assigned to the childhood sexual abuse
group if the interviewing psychiatrist rated their exposure to sexual       without a history of childhood sexual abuse. However,
abuse as likely or definite. The inclusion of patients whose exposure       patients who reported childhood sexual abuse had
was rated as likely was based on literature suggesting a general un-        made more visits to their psychiatrist in the previous 2
derestimation bias in reports of childhood sexual abuse (1, 14).            years (mean=26.8, SD=39.0, versus mean=15.1, SD=
Assessment of Disordered Personality Functioning                            26.9) (F=4.10, df=1, p=0.04). In addition, more pa-
                                                                            tients who reported childhood sexual abuse admitted
   The interviewing psychiatrist judged whether the patient had a           to moderate or severe levels of hopelessness at inter-
substantial personality disturbance preceding the first depressive ep-      view (37 [93%] versus 97 [74%]) (χ2=4.90, df=1, p=
isode and whether the patient had a long-standing pattern of inter-
personal sensitivity resulting in substantial social or occupational
                                                                            0.02). The groups were comparable in percentages of
impairment. At the end of the interview, the psychiatrist used a 6-         self-reported helplessness, sinfulness, guilt, pessimism,
point scale (0=not at all to 5=extreme) to rate the extent to which         worthlessness, and impaired self-image. More patients
each of 15 personality vignette descriptors matched the patient’s           who were exposed to childhood sexual abuse than pa-
long-term personality style. These descriptors were derived from the        tients who were not reported significant feelings of
10 DSM-IV personality disorders plus one DSM-IV personality dis-
order listed for further study (depressive personality disorder), three     self-annoyance or self-anger (34 [85%] versus 81
DSM-III-R personality disorder classes (passive-aggressive, self-de-        [62%]) (χ2=6.04, df=1, p=0.01), and more reported
feating, and sadistic), and an anxious personality descriptor (nervy,       that they could lose control of their anger (19 [48%]
tense, and a worrier). In addition, the psychiatrist used a 4-point         versus 34 [26%]) (χ2=5.17, df=1, p=0.02).
scale (0=no, 1=possibly, 2=probably, 3=definitely) to rate the extent
to which the patient’s personality style was disordered across eight           There was a difference between groups in severity of
parameters defined by Millon (23): inflexible/defective, causing sig-       depression as measured by self-report Beck depression
nificant personal discomfort, reducing opportunities, inability to          scores: the patients who had been exposed to child-
function effectively and efficiently, inability to adjust to the environ-   hood sexual abuse had a mean score of 33.7 (SD=9.5),
ment, vicious or self-defeating cycles, tenuous stability under stress,
and personal discomfort to others. The psychiatrist also rated the de-
                                                                            compared with 26.9 (SD=11.4) for those who had not
gree of functionality (1=functional, 2=probably dysfunctional, 3=           (F=12.65, df=1, p=0.001). There were no significant
definitely dysfunctional) across five domains where disordered per-         differences on other depression measures: clinician-
sonality style might be manifested (23): intimate relationships, fam-       rated Hamilton depression scores were 23.8 and 22.0,
ily relationships, peer relationships, work, and work relationships.        respectively, for patients who had and had not been ex-
Total parameter and domain scores were obtained by summing the
eight parameter and five domain scores, respectively.                       posed to childhood sexual abuse. The consultants’ rat-
                                                                            ing of depression severity yielded similar scores for the
                                                                            two groups (mean=2.25 and mean=2.22, respectively),
RESULTS                                                                     as did current Global Assessment of Functioning Scale
                                                                            scores (mean=47.2 and mean=50.1, respectively).
Reported Incidence of Childhood Sexual Abuse                                   No differences in lifetime history measures were ob-
                                                                            served: patients who reported childhood sexual abuse
  Forty-six (17%) of the 269 patients reported experi-                      had had their first episode of major depression at 25.4
encing childhood sexual abuse before the age of 16                          years compared with 32.0 years for those with no

432                                                                                               Am J Psychiatry 156:3, March 1999
                                                                                               GLADSTONE, PARKER, WILHELM, ET AL.

TABLE 1. Scores for Parenting of Fathers and Mothers of De-             TABLE 2. Self-Report Items on the Measure of Parental Style
pressed Women With and Without Childhood Sexual Abuse                   That Were Most Strongly Associated With Childhood Sexual
                                           No                           Abuse Among Depressed Womena
                        Childhood      Childhood                        Self-Report Item                                Ranked Odds Ratiob
                          Sexual        Sexual                          Father
                          Abuse         Abuse                             Physically violent to me                              12.6
Measure of Parental      (N=40)a       (N=131)a                           Critical of me                                          6.4
Behavior                Mean SD        Mean SD         Fb       df        Sought to make me feel guilty                           5.7
Father                                                                    Made me feel in danger                                  5.7
  Parental Bonding                                                      Mother
    Instrument                                                            Made me feel in danger                                  9.4
    Care                  13.3    9.5 21.6 9.5 18.1c 1, 108               Made me feel unsafe                                     7.7
    Overprotection        16.3    8.4 14.5 9.1         0.9    1, 108      Rejecting of me                                         6.4
  Measure of Paren-                                                       Would forget about me                                   6.0
    tal Style                                                           a Logistic regression analyses (Wald chi-square test); the odds ra-
    Indifference           8.7    6.0     3.9 5.0 26.7c 1, 110            tio quantifies overrepresentation in the group with childhood sex-
    Overcontrol            5.1    3.4     3.3 3.2      7.0    1, 109      ual abuse. Age was used as the categorical covariate.
    Abuse                  6.6    5.7     2.4 3.8 22.4c 1, 109          b All are significant after Bonferroni correction at p<0.002 (30 self-
Mother                                                                    report items), two-tailed (df=1).
  Parental Bonding
    Care                  17.1 11.9 22.1 9.5           7.0    1, 113    Style. There were no differences between the groups in
    Overprotection        16.6 10.5 16.3 9.1           0.1    1, 113    Parental Bonding Instrument overprotection scores
  Measure of Paren-
    tal Style
                                                                        and Measure of Parental Style overcontrol scores for
    Indifference           7.6    6.3     3.0 4.5 21.7c 1, 113          either parent.
    Overcontrol            6.0    3.8     4.2 3.6      4.8    1, 114       All Measure of Parental Style items were examined
    Abuse                  5.4    5.0     2.1 3.2 24.0c 1, 114          individually for group differences (patients who had
a Some variations in group Ns due to incomplete parental self-re-
                                                                        been exposed to childhood sexual abuse versus those
  port data.                                                            who had not). Items with the four highest odds ratios
b Multivariate analysis using a two-tailed general linear model, with
  age as the continuous covariate.                                      (i.e., those items from the Measure of Parental Style
c Exceeds Bonferroni adjusted F critical value (8.09).                  most strongly associated with childhood sexual abuse)
                                                                        were ranked in descending order for fathers and moth-
                                                                        ers separately. Items of greatest distinction for fathers
childhood sexual abuse; the reported number of life-                    identified physical violence, criticism, making the child
time episodes of depression were 4.0 and 3.2, respec-                   feel guilty, and making the child feel in danger. For
tively; and duration of the current depressive episode                  mothers, distinctive items reflected lack of protection
was 35.2 and 30.4 weeks, respectively. The two                          and maternal distance (table 2).
groups of patients had similar rates of nonmelan-
                                                                           Table 3 lists the percentages of patients with and
cholic depression according to the criteria of DSM-III-
                                                                        without exposure to childhood sexual abuse who re-
R (22 [55%] and 75 [57%]) and DSM-IV (23 [58%]
                                                                        ported certain parental environment characteristics
versus 79 [60%]).
                                                                        during childhood. Lack of protection against threat
   Lifetime prevalence of all anxiety disorders was sim-                posed by others was the most overrepresented experi-
ilar for both groups: 50% of the patients in each group                 ence affirmed by the childhood sexual abuse group.
were given a diagnosis of any lifetime anxiety disorder.                Other overrepresented characteristics were being made
Social phobia was the most prevalent anxiety disorder                   to feel unsafe, a dysfunctional father, verbal abuse, and
for both of the groups: it occurred in 13 (33%) of the                  exposure to an unstable relationship between parents.
patients who had been exposed to childhood sexual                       Hence, the general nature of the childhood environ-
abuse and 33 (25%) of those who had not.                                ment (as measured by the Parental Bonding Instru-
                                                                        ment, the Measure of Parental Style, and interview-
Parental Environment                                                    rated parental environment items) was able to distin-
                                                                        guish patients with a history of childhood sexual abuse
   Significantly more patients who had been exposed to                  distinctly from those with no childhood sexual abuse.
childhood sexual abuse reported having an alcoholic
father than did those who had not (17 [43%] versus 26                   Self-Injurious and Drug Use Behaviors
[20%]) (χ2=6.77, df=1, p=0.01). No other differences
in psychiatric family history were evident.                               There were some notable differences in self-injurious
   Table 1 presents the patients’ scores on the Parental                behaviors between the two groups (table 4). Signifi-
Bonding Instrument and the Measure of Parental Style.                   cantly more patients who had been exposed to child-
Patients who had been exposed to childhood sexual                       hood sexual abuse had made a suicide attempt during
abuse had significantly lower care scores for fathers                   their current depression, and they had also made more
only on the Parental Bonding Instrument. These pa-                      previous suicide attempts. Significantly more patients
tients had significantly higher indifference and abuse                  who had been exposed to childhood sexual abuse had
scores for both parents on the Measure of Parental                      a previous history of self-injury or self-mutilation, and

Am J Psychiatry 156:3, March 1999                                                                                                        433

TABLE 3. Characteristics of Parental Environment Among De-            TABLE 4. Self-Injurious Behaviors and Drug and Alcohol Use
pressed Women With and Without Childhood Sexual Abuse                 Among Depressed Women With and Without Childhood Sex-
                                                                      ual Abuse
                          Childhood       Childhood                                                                   No
                           Sexual           Sexual                                                Childhood       Childhood
Characteristic of           Abuse           Abuse          Ranked                                  Sexual           Sexual
Parental                   (N=40)          (N=131)          Odds                                   Abuse            Abuse
Environment               N      %        N      %         Ratioa                                  (N=40)          (N=131)         Odds
Lack of protection                                                    Variable                     N      %        N       %       Ratioa
 against threaten-                                                    Current suicidal
 ing others               27     68       22      17        10.1b       thoughts or plansb         34      85      89       68      2.5
Making child feel                                                     Current suicide
 unsafe                   21     53       22      17         6.1b       attemptb                   16      40      23       18      3.1c
Generally dysfunc-                                                    Previous suicide
 tional father            33     83       59      45         5.7b       attempts (>2)              13      33      15       12      4.1c
Verbal abuse              25     63       33      25         5.0b     Several previous sui-
Dysfunctional mari-                                                     cide attempts (>5)          6      15       1        1     26.3c
 tal relationship         32     82       64      49         4.7b     Current self-injurious
Parents separated                                                       actb                       10      25      16       12      2.1
 from each other          21     53       24      19         4.7b     History of self-injurious
Rejection                 25     63       41      33         4.0b       acts                       14      35      15       11      3.9c
Lack of support in                                                    Current alcohol
 times of distress        25     63       40      31         3.9b       dependenceb                 2       5       3        2      1.9
Physical abuse            16     40       20      15         3.8b     Previous alcohol
Generally dysfunc-                                                      dependence                  9      23       9        7      4.7c
 tional mother            27     68       55      42         2.9      Current benzodiaze-
Unpredictability          21     53       39      30         2.6        pine useb                   8      20      21       16      0.6
Physical violence                                                     Previous benzodiaze-
 between parents          14     35       26      20         2.4        pine dependence             5      13      18       14      1.0
Multiple parent                                                       Current illicit drug useb     3       8       6        5      1.3
 figures                  10     25       16      12         2.3      a Odds ratios were obtained by using logistic regression analyses
Depressed mother                                                        (Wald chi-square test) with age as the categorical covariate.
 and/or father            24     60       51      40         2.2      b Behavior evident in current depressive episode.
Criticism                 22     55       52      40         1.8      c Significant after Bonferroni adjustment at p<0.005, two-tailed
Raised by single                                                        (df=1).
 parent                    7     18       14      11         1.7
a Odds   ratios were obtained by using logistic regression analyses
  (Wald chi-square test) with age as the categorical covariate.       was significantly higher for this group was the border-
b Significant after Bonferroni adjustment at p<0.003, two-tailed
                                                                      line descriptor.
                                                                      Multivariate Analyses
more had been dependent on alcohol in the past. How-
ever, current alcohol dependence, past and present                       Separate logistic regression analyses were used to
benzodiazepine use, and illicit drug use were similar                 identify any influence of the borderline personality
across both groups. These results may have been af-                   style variable on the association between childhood
fected by the fact that we excluded from the study pa-                sexual abuse and suicidal and self-injurious behaviors.
tients with drug- or alcohol-induced depression.                      Borderline scores, childhood sexual abuse status, and
                                                                      age (less than 35, 35–50, or more than 50 years) as a
Personality                                                           categorical covariate were entered into a logistic re-
                                                                      gression model with current suicide attempt as the de-
  More patients with exposure to childhood sexual                     pendent variable. The chi-square improvement statis-
abuse than patients with no exposure (26 [65%] versus                 tics were 7.20 (df=1, p=0.01) for borderline; 1.37 (df=
42 [32%]) were rated as having evidence of substantial                1, p=0.24) for childhood sexual abuse, and 0.37 (df=1,
personality disturbance before their current depressive               p=0.83) for age. These results indicate that higher
episode (χ2=12.30, df=1, p=0.001). Similarly, patients                scores on the borderline item were significantly associ-
who had been exposed to childhood sexual abuse were                   ated with a greater likelihood of there having been a
significantly more likely to admit to a long-standing                 suicide attempt during the current episode. Being in the
pattern of interpersonal sensitivity (27 [68%] versus                 childhood sexual abuse group did not add further pre-
50 [38%]) (χ2=8.80, df=1, p=0.003).                                   dictive value to the chance of making a current at-
  Table 5 examines the differences between the groups                 tempt. This analysis was repeated with several suicide
on a variety of measures of disordered personality                    attempts (more than five) as the dependent variable.
function as well as styles underlying separate person-                Again the analysis indicated that the borderline de-
ality disorders. Nonsignificant trends were observed                  scriptor was a dominant predictor of having made
for higher domain and parameter dysfunction scores                    more than five previous attempts at suicide (χ2=7.43,
for the patients with a history of childhood sexual                   df=1, p=0.006) and that the childhood sexual abuse
abuse. However, the only personality item score that                  variable (χ2=1.04, df=1, p=0.31) and age (χ2=2.82, df=

434                                                                                             Am J Psychiatry 156:3, March 1999
                                                                                   GLADSTONE, PARKER, WILHELM, ET AL.

1, p=0.24) did not add any predictive value beyond          TABLE 5. Measures of Personality Disorder and Style Among
that of the borderline descriptor for the existing rela-    Depressed Women With and Without Childhood Sexual Abuse
tionship between childhood sexual abuse and previous                                                      No
suicide attempts.                                                                       Childhood      Childhood
                                                                                         Sexual          Sexual
   A further analysis that used previous self-injury as                                   Abuse          Abuse
the dependent variable yielded the same result. Border-                                  (N=40)         (N=131)        F (df=1,
line scores (χ 2 =26.37, df=1, p<0.001) rather than         Personality Variable        Mean SD        Mean SD          168)a
childhood sexual abuse status (χ2=0.26, df=1, p=0.69)       Disordered personality
were associated with the increased likelihood of a his-        functioning
tory of self-injury.                                           Total domain score         8.3 2.6        6.9      2.2      7.3
   Thus, the borderline descriptor stood out as a signif-      Total parameter score     10.5 6.6        6.4      6.1      8.2
                                                            DSM personality classb
icant mediating influence of self-harmful behaviors in         Borderline                 2.0 1.8        0.4      0.9    43.3c
that it appeared to account for the connection between         Paranoid                   1.3 1.3        0.7      1.0      6.6
childhood sexual abuse and self-harm. Because of its           Self-defeating             1.5 1.6        0.8      1.2      6.1
obvious significance, the borderline variable was ex-          Histrionic                 1.0 1.3        0.5      1.0      4.2
amined against other rated parental characteristics.           Avoidant                   3.0 1.4        2.3      1.5      3.0
                                                               Depressive                 1.6 1.7        0.9      1.4      2.7
First, all parental environment predictors were corre-         Obsessive                  1.0 1.3        1.2      1.5      0.5
lated with borderline scores to assess whether any             Dependent                  2.0 1.4        1.7      1.5      0.5
overlap existed. Modest but significant relationships          Anxious                    2.5 1.4        2.6      1.2      0.2
were observed between borderline scores and paternal        a Multivariate analysis using a two-tailed general linear model with

care (r=–0.34, p<0.01), maternal indifference (r=0.40,        age as the continuous covariate.
                                                            b Analysis includes nine of the original 15 personality items. Six
p<0.01), maternal abuse (r=0.30, p<0.01), paternal in-
                                                              items were excluded from the analysis because of low response
difference (r=0.37, p<0.01), paternal care (r=–0.29, p<       rates in both groups.
0.01), and childhood sexual abuse status (r=0.48, p<        c F value exceeds Bonferroni-adjusted F critical value (8.27).

0.01). Although some overlap did exist, interpretations
concerning the independent role or effect of each pre-
dictor variable could not be made because correlation       several previous suicide attempts, high maternal indif-
coefficients were not high.                                 ference scores and childhood sexual abuse were
   A linear regression analysis examined components         equally predictive characteristics.
of the parental environment (care scores on the Paren-
tal Bonding Instrument, indifference and abuse scores
on the Measure of Parental Style, childhood sexual          DISCUSSION
abuse status) associated with higher borderline scores.
The regression model indicated that having a history          Childhood sexual abuse has long been held to have a
of childhood sexual abuse was the only significant          number of short-term and long-term consequences, in-
variable (t=4.66, df=7, 94, p<0.001) and that the           cluding depression in adulthood. Having been sexually
other measures of parental environment were nonpre-         abused in childhood may increase the likelihood of de-
dictive. This result indicates that childhood sexual        pression or may influence the expression and severity
abuse status contributed to higher borderline scores        of depression, either directly or indirectly, as a conse-
irrespective of other aberrant characteristics of the       quence of a more general aversive early environment.
parental environment.                                       We studied a group of female patients with primary
   Similar analyses were conducted for current suicide      major depression, focusing on identifying nuances of
attempt, previous attempts, and history of self-injury,     the parental environment for those reporting child-
with the aim of isolating any independent effects of        hood sexual abuse and then considering any role that
childhood sexual abuse above and beyond other aber-         childhood sexual abuse might bring to adult depres-
rant parental characteristics on these adult self-harm      sion. Our study relied to a large degree on self-report
behaviors. For current suicide attempt, the only char-      data, a method that risks a range of retrospective bi-
acteristic of the early environment that was a signifi-     ases, particularly in view of the fact that patients with
cant predictor was childhood sexual abuse status (χ2=       a borderline personality style are recognized as tending
4.12, df=1, p=0.04). For several previous suicide at-       to blame others. Thus, while we explore causal links,
tempts, significant predictors were higher maternal         such noncausal possibilities must be conceded.
indifference scores on the Measure of Parental Style          We found no sociodemographic differences between
(χ2=4.41, df=1, p=0.03) and having a history of child-      the patients who did or did not report childhood sex-
hood sexual abuse (χ2=4.44, df=1, p=0.03). For his-         ual abuse. Depressive features did not distinguish these
tory of self-injury, childhood sexual abuse alone and       two groups of patients, but patient-rated Beck depres-
independent of other characteristics had significant        sion scores were significantly higher for those who re-
predictive value (χ2=4.39, df=1, p=0.03). Thus, for         ported childhood sexual abuse. However, psychiatrist-
current suicide attempt and history of self-injury,         rated estimates of the severity of depression did not
childhood sexual abuse was the dominant predictive          differentiate the two groups. Nor was there any differ-
characteristic of the early environment, whereas for        ence in the incidence of lifetime anxiety disorders. We

Am J Psychiatry 156:3, March 1999                                                                                          435

found that patients with a history of childhood sexual          Key Measure of Parental Style items for fathers por-
abuse reported significantly more adverse childhood          trayed a paternal relationship characterized by more
environment experiences and that multiple types of pa-       general abuse and threat; for mothers, dominant items
rental dysfunction were indicated. Patients who were         suggested a relationship characterized by lack of pro-
exposed to childhood sexual abuse also differed in re-       tection, feeling in danger, and maternal disconnected-
porting higher self-injury and suicide attempt rates and     ness. Therefore, these patients, in addition to reporting
in having more disordered personality functioning.           sexual abuse, also recalled more severe degrees of
   The latter differences, together with the fact that pa-   other parental dysfunction and provided evidence of
tients with a history of childhood sexual abuse re-          greater parental marital problems.
ported more psychiatric visits and had higher self-re-          Our multivariate analyses allowed us to speculate on
port scores on hopelessness, self-disapproval, and the       the contribution and sequential associations of aber-
Beck inventory, may be consequences of childhood             rant parental environment, adult self-injury, and per-
sexual abuse or may be explained by differences ac-          sonality dysfunction for the patients with a history of
counted for by overrepresentation of an identified per-      childhood sexual abuse. The question as to whether
sonality disturbance (i.e., borderline style). Therefore,    childhood sexual abuse may be pathogenic in any
in this study group, childhood sexual abuse could not        unique form or is pathogenic as part of a childhood en-
be clearly linked to greater severity of depressive symp-    vironment with ubiquitous adversity has received at-
toms per se but was significantly linked to antecedent       tention in recent years.
environmental factors and long-standing patterns of             For our study group, childhood sexual abuse ap-
dysfunctional behaviors.                                     peared an equally predictive factor, along with high
                                                             maternal indifference, of making multiple suicide at-
   It has been argued extensively that anomalous child-
                                                             tempts in adulthood, a finding highlighted in previous
hood environments can play a causal antecedent role
                                                             studies (14, 27–29). However, childhood sexual abuse
in the development of depressive disorders. Even
                                                             was identified as a single predictor of self-destructive-
though most of our patients with exposure to child-
                                                             ness in the form of past self-injury and a current sui-
hood sexual abuse had been abused by someone other
                                                             cide attempt. In relation to personality, we found that
than a parent, they also reported a more detrimental         the relationship between childhood sexual abuse and
parental environment than did depressed patients with        adult behaviors of self-harm (history of self-injury, sev-
no history of childhood sexual abuse. Although no            eral past suicide attempts, and current suicide attempt)
more likely to have a family history of mental illness,      was determined more by the influence of higher bor-
patients who had been exposed to childhood sexual            derline descriptor scores. Thus, borderline disorder
abuse were more likely to have grown up with an alco-        might act as a mediating factor enabling a pattern of
holic father, a finding that has been reported by others     long-standing self-destructiveness.
(24). Alcohol dependence, being raised by an alcoholic          Figueroa et al. (30) argued that interpersonal sensi-
parent, early life stressors, and chronic psychosocial       tivity may act as a temperamental substrate with which
turbulence, including dissatisfaction and self-destruc-      sexual abuse experiences interact to effect a borderline
tiveness, have also been linked to characterological de-     diagnosis. For our study group, a pattern of long-
pressions (25, 26).                                          standing interpersonal sensitivity was affirmed for sig-
   Our patients with a history of childhood sexual           nificantly more patients with a history of childhood
abuse recalled a family environment characterized by         sexual abuse than patients with no childhood sexual
low levels of parental care (particularly from fathers)      abuse (68% versus 38%), a finding that provides fur-
as well as high levels of parental indifference and          ther support for this view.
abuse. Similar themes, particularly that of low levels of       After identifying the apparently significant role of
care, have been reported in previous studies (2, 5),         borderline personality, we aimed to track its possible
some of which also used the Parental Bonding Instru-         evolution for our depressed group. When childhood
ment (14, 27). Both of the latter studies, however,          sexual abuse was considered in conjunction with other
found an association between childhood sexual abuse          negative aspects of the childhood environment, we
and high levels of parental overprotection, whereas the      found that childhood sexual abuse status was a better
present study found no evidence for higher Parental          predictor of higher borderline personality style scores
Bonding Instrument overprotection scores for either          than other parental environment characteristics.
parent. In contrast, lack of protection against threat       Therefore, we speculate that childhood sexual abuse
posed by others was the most commonly reported pa-           may well act to effect a borderline personality style as
rental environment characteristic associated with            a dominant antecedent, not just as an equally negative
childhood sexual abuse (table 3), which is consistent        component within a dysfunctional family style. Thus,
with our finding that 70% of our patients with a his-        although borderline characteristics appeared to engulf
tory of childhood sexual abuse were abused by some-          the association between childhood sexual abuse and
one other than a parent. The items in table 3 identified     self-harm, they could, in fact, be traced to a history of
the childhood sexual abuse environment typically as          childhood sexual abuse. Even though such specula-
nonprotective and nonsupportive, with conflict, vio-         tions risk circularity, we were able to disentangle some
lence, and marital turbulence.                               close associations for our study group, namely, that

436                                                                                Am J Psychiatry 156:3, March 1999
                                                                                          GLADSTONE, PARKER, WILHELM, ET AL.

childhood sexual abuse contributed most strongly (and                13. Palmer RL, Oppenheimer R, Dignon A, Chaloner DA, Howells
possibly independently) to higher borderline scores but                  K: Childhood sexual experience with adults reported by
                                                                         women with eating disorders: an extended series. Br J Psy-
not exclusively to some expressions of self-harm.                        chiatry 1990; 156:699–703
   Even though childhood sexual abuse status had                     14. Mullen PE, Martin JL, Anderson JC, Romans SE, Herbison
strong links to self-destructive behavior (alone or with                 GP: Childhood sexual abuse and mental health in adult life. Br
a pattern of maternal indifference), and was a likely                    J Psychiatry 1993; 163:721–732
                                                                     15. Parker G: Parental affectionless control as an antecedent to
causative and dominant factor in the expression of                       adult depression. Arch Gen Psychiatry 1983; 40:956–960
borderline features, it was also closely associated with             16. Parker G, Hadzi-Pavlovic D, Greenwald S, Weissman M: Low
an otherwise adverse early environment. Thus, a his-                     parental care as a risk factor to lifetime depression in a com-
tory of childhood sexual abuse appears associated with                   munity sample. J Affect Disord 1995; 33:173–180
                                                                     17. Parker G, Gladstone G: Parental characteristics as influences
a greater chance of exposure to earlier dysfunctional                    on adjustment in adulthood, in Handbook of Social Support
family factors, which, in turn, are associated with a                    and the Family. Edited by Pierce GB, Sarason BR, Sarason
greater risk of depression, disordered personality func-                 EG. New York, Plenum, 1986, pp 195–218
tion, and other psychopathology.                                     18. Parker G, Roussos J, Hadzi-Pavlovic D, Mitchell P, Wilhelm K,
                                                                         Austin M-P: The development of a refined measure of dys-
                        REFERENCES                                       functional parenting and assessment of its relevance in pa-
                                                                         tients with affective disorders. Psychol Med 1997; 27:1193–
 1. Bifulco A, Brown GW, Adler Z: Early sexual abuse and clinical
                                                                     19. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J: An in-
    depression in adult life. Br J Psychiatry 1991; 159:115–122
                                                                         ventory for measuring depression. Arch Gen Psychiatry 1961;
 2. Boudewyn AC, Liem JH: Childhood sexual abuse as a precur-            4:561–571
    sor to depression and self-destructive behavior in adulthood.    20. Parker G, Tupling H, Brown LB: A parental bonding instru-
    J Trauma Stress 1995; 8:445–459                                      ment. Br J Med Psychol 1979; 52:1–10
 3. McCauley J, Kern DE, Kolodner K, Dill L, Schroeder AF,           21. World Health Organisation: Composite International Inter-
    Dechant HK, Ryden J, Derogatis LR, Bass EB: Clinical char-           view, Version 1.2 (CIDI-A). Sydney, WHO Research and Train-
    acteristics of women with a history of childhood abuse—un-           ing Centre, 1983
    healed wounds. JAMA 1997; 277:1362–1368                          22. Hamilton M: Development of a rating scale for primary depres-
 4. Mennen FE, Meadow D: A preliminary study of the factors re-          sive illness. Br J Soc Clin Psychol 1967; 6:278–296
    lated to trauma in childhood sexual abuse. J Family Violence     23. Millon TA: A theoretical derivation of pathological personali-
    1994; 9:125–142                                                      ties, in Contemporary Directions in Psychopathology: Toward
 5. Pribor EF, Dinwiddie SH: Psychiatric correlates of incest in         the DSM-IV. Edited by Millon T, Klerman G. New York, Guilford
    childhood. Am J Psychiatry 1992; 149:52–56                           Press, 1986, pp 639–669
 6. Bryer JB, Nelson BA, Miller JB, Krol PA: Childhood sexual and    24. Brown GR, Anderson B: Psychiatric morbidity in adult inpa-
    physical abuse as factors in adult psychiatric illness. Am J         tients with childhood histories of sexual and physical abuse.
    Psychiatry 1987; 144:1426–1430                                       Am J Psychiatry 1991; 148:55–61
 7. Herman JL, Perry JC, van der Kolk BA: Childhood trauma in        25. Akiskal HS, Rosenthal TL, Haykal RF, Lemmi H, Rosenthal
    borderline personality disorder. Am J Psychiatry 1989; 146:          RH, Scott-Strauss A: Characterological depressions: clinical
    490–495                                                              and sleep EEG findings separating “subaffective dysthymias”
                                                                         from “character spectrum disorders.” Arch Gen Psychiatry
 8. Nigg JT, Silk KR, Westen D, Lohr NE, Gold LJ, Goodrich S,
                                                                         1980; 37:777–783
    Ogata S: Object representations in the early memories of sex-
                                                                     26. VanValkenburg C, Akiskal HS, Puzantian V: Depression spec-
    ually abused borderline patients. Am J Psychiatry 1991; 148:
                                                                         trum disease or character spectrum disorders? a clinical
                                                                         study of major depressives with familial alcoholism or sociop-
 9. Oppenheimer R, Howells K, Palmer RL, Chaloner DA: Ad-                athy. Compr Psychiatry 1983; 24:589–595
    verse sexual experiences in childhood and clinical eating dis-   27. Romans SE, Martin J, Mullen P: Women’s self-esteem—a
    orders: a preliminary description. J Psychiatr Res 1985; 19:         community study of women who report and do not report
    357–361                                                              childhood sexual abuse. Br J Psychiatry 1996; 169:696–704
10. van der Kolk BA, Perry JC, Herman JL: Childhood origins of       28. Spaccarelli S, Fuchs C: Variability in symptom expression
    self-destructive behavior. Am J Psychiatry 1991; 148:1665–           among sexually abused girls—developing multivariate mod-
    1671                                                                 els. J Clin Child Psychol 1997; 26:24–35
11. Brodsky BS, Cloitre M, Dulit RA: Relationship of dissociation    29. Stern AE, Lynch DL, Oates RK, O’Toole BI, Cooney G: Self-
    to self-mutilation and childhood abuse in borderline personal-       esteem, depression, behavior and family functioning in sexu-
    ity disorder. Am J Psychiatry 1995; 152:1788–1792                    ally abused girls. J Child Psychol Psychiatry 1995; 36:1077–
12. Fox KM, Gilbert BO: The interpersonal and psychological              1089
    functioning of women who experienced childhood physical          30. Figueroa EF, Silk KR, Huth A, Lohr NE: History of childhood
    abuse, incest, and parental alcoholism. Child Abuse Negl             sexual abuse and general psychopathology. Compr Psychia-
    1994; 18:849–858                                                     try 1997; 38:23–30

Am J Psychiatry 156:3, March 1999                                                                                                  437