Behavior, Family Function, School Performance, and Predictors of Persistent Disturbance in Sexually Abused Children

(1)

Behavior,

Family

Function,

School

Performance,

and

Predictors

of

Persistent

Disturbance

in Sexually

Abused

Children

Jan E. Paradise, MD*; Lynda Roses; Lynn A. Sleeper, ScD; and Madelaine Nathanson, PhDII

ABSTRACT. Objective. Although sexual abuse is

widely considered to have severe sequelae, most studies of children’s status after sexual abuse have had major limitations, including retrospective design, small sample

sizes, selective enrollment of subjects, no comparison

groups, and lack of information about potentially con-founding characteristics of studied children. The aim of this study was to clarify the impact of sexual abuse on children’s psychological well-being.

Methods. We prospectively studied 154 children who

were brought to urban, university-affiliated hospitals for assessment of recently disclosed sexual abuse and

com-pared them with a control group of 53 demographically

similar children not known to have been sexually abused.

Results. In comparison with the control children,

fewer of the sexually abused children had health insur-ance and more had received psychiatric care unrelated to the sexual abuse. Most initial ratings of behavior, mater-nal psychiatric status, family function, and school per-formance were less favorable among the sexually abused than among the control children (P < .05). At follow-up 6 months later, the psychiatric status of the mothers of the abused children apparently improved, but the children’s difficulties largely persisted. Baseline characteristics of the abused children significantly or suggestively associ-ated with persisting problematic behavior were older age (P = .04), lower maternal educational attainment (P = .06),

poorer maternal psychiatric status (P = .04) and lower

fam-ily integration (P < .001). These four factors accounted for 31% of the variance in the children’s behavior at 6-month follow-up (P < .001). Unexpectedly, characteristics of the children’s sexual abuse experiences did not predict their later behavioral status.

Conclusions. These findings suggest that preexisting, long-standing adverse psychosocial circumstances may contribute importantly to persistently problematic behav-ior and school performance among sexually abused chil-dren. The findings also suggest that it is children’s pre-existing psychosocial circumstances, rather than the

abuse, that determine, at least in part, the nature of their

functional outcomes. Pediatrics 199493:452-459; child

abuse, child behavior, child psychiatry, child welfare, sexual child abuse.

From the *Dp1.tfllent of Pediatrics, Boston City Hospital and Boston

Urn-versity SchoolofMedicine; Boston University SchoolofPublic Health; §the

New England Research Institute; and lithe Philadelphia Child Guidance

Clinic, Philadelphia, PA.

Lynda Rose’s current address is Goldman School of Graduate Dentistry, Boston University, Boston, MA 02118.

Presented in part at the 30th Annual Meeting of the Ambulatory Pediatric Association, May 9, 1990, Anaheim, CA, and at the 8th Biennial National

Symposium on Child Victimization, May 21, 1992, Washington, DC.

Received for publication Feb. 19, 1993; accepted Sep. 13, 1993.

Reprint requests to (J.E.P.) Dept of Pediatrics, Boston City Hospital, 818 Harrison Aye, Boston, MA 02118.

ABBREVIATIONS. CBCL, Child Behavior Checklist; CPS, child

protective services; SCL-90, Symptom Checklist-90-Revised.

Sexual abuse is common in the United States1 and

is generally considered to have severe psychosocial

and behavioral consequences,217 ranging from

run-rnng away3 and truancy4 to depression,9”#{176}

unsatis-factory adult sexual relationships,11”2 and an elevated

risk of repeated victimization. 5,117 However,

ques-tions about the validity of these observations remain.

Many studies of the sequelae of child sexual abuse

have methodologic limitations induding:

retrospec-tive design; small, narrowly selected, or incompletely

described samples; absence of information about

eligible but unstudied subjects; a lack of comparison

groups; and questionnaires without demonstrated

validity or reliability.2”8”9 In addition, although many

investigators have suggested that the association

be-tween poor psychosocial outcomes and reported

sexual abuse may be confounded by coexisting

factors-including poverty,20 social isolation and

separation of children from parents,10’ parental

un-supportiveness, psychiatric disturbance, and

sub-stance abuse5”2’2-and by physically handicapping

conditions, no contemporary reports have

at-tempted to distinguish the possible effects of sexual

abuse from the effects of such comorbid conditions.

The present prospective study was undertaken to

clarify the impact of sexual abuse on children’s

psy-chological well-being and on their ability to function

effectively. The study had three objectives: to learn

whether and, so, in what ways and to what extent

a group of sexually abused children differed in

be-havior and school performance soon after the time

of their disdosures of abuse from a demographically

similar group of nonabused children; to detail

changes, if any, that developed in the abused

chil-dren’s behavior and school performance during the

ensuing 6 months (resources were insufficient to

permit longer follow-up); and to identify initial

characteristics among the sexually abused children

that were associated with subsequent problematic

behavior.

METHODS Setting and Subjects

Subjects for this study were enrolled between June 1985 and

March 1986 at The Children’s Hospital of Philadelphia. After the

principal investigator’s relocation, subjects were enrolled between

February 1987 and January 1988 at Boston City Hospital and,

during alternate months, at The Children’s Hospital, Boston.

(2)

brought to one of these hospitals within 8weeks after a disclosure

that they had been sexually abused and if a study interview could

be arranged within 8 weeks after the hospital visit. Only one child

per family was enrolled. New cases were identified by reviewing

the daily diagnostic rosters of the hospital’s emergency and social

service departments, inpatient services, and child abuse and

pe-diatric gynecology clinics.

A control group was recruited from among 4- to 12-year-old

children who received primary care in the general medical clinic

or the Emergency Department of either The Children’s Hospital of

Philadelphia or The Children’s Hospital, Boston, who were free of

major illness, and who had no history of sexual abuse,

genitouri-nary complaints, or behavior problems that might indicate sexual

abuse. The parents were invited to participate in a study of the

emotional, social, and physical health of children with medical

problems. During the enrollment procedure, these parents were

informed that their children would be compared with a group of

sexually abused children, and were instructed not to enroll if they had any suspicion that their children had been sexually abused.

Study Definition of Sexual Abuse

A child was considered sexually abused if he or she had been

physically touched by aperson 5 or more years older in a way

considered sexual by the child, the parent, or the involved cmi-clans; had been touched sexually by a peer and objected to the contact; or had a newly diagnosed sexually transmitted disease.

Because limiting this study to cases of sexual abuse confirmed

after investigation by either state child protective service (CPS)

workers or law enforcement officials would delay enrollment and

might result in biased selection of more serious cases or more

articulate children, the children were enrolled on the basis of their

initial histories. After enrollment, however, each case of abuse was

rated as definite, probable, possible, or uncertain using the follow-ing criteria: definite-the child described the abuse in substantial

detail or the alleged abuser admitted the abuse or pled guilty in

criminal court; probable-the child provided a relatively detailed

history of the abuse or had a sexually transmitted disease, or the

abuser was found guilty at criminal trial; possthle-the child

pro-vided a suggestive but nonspecific history of abuse or had an

unexplained genital injury; and uncertain-the child provided a

vague history or failed to confirm abuse alleged by a parent.

Uncertain cases were included only if they were handled as bona

fide cases by either CPS or law enforcement officials.

Parent Interviews and Questionnaires

Parents were queried during two semistructured interviews

that included standardized questionnaires. The first interview was

conducted within 8 weeks of the initial hospital visit (for abused

children) or study enrollment (for control children); the second

was conducted 6 months after the first. Mothers were interviewed

preferentially because they were more likely than fathers to reside

with their children and to have met the researchers at the hospital.

Information about children abused by a parent was solicited from

the nonabusing parent. Trained research assistants conducted the

interviews and helped parents complete the questionnaires.

Par-ents were reimbursed for transportation expenses related to the study.

To assess children’s behavior, the parent report form of the

Child Behavior Checklist (CBCL) was used.24 fl

question-naire, a total behavior problem score is generated from a parent’s

rating of 118 problematic behaviors as not true, sometimes true, or

often true of the child. Normed T scores are available for four

sex/age categories of children. Total T scores above 63 (90th per-centile) are deemed clinically significant. To assess parent’s

psy-chiatric status, the Symptom Checklist-90-Revised (SCL-90) was

used.27’28 This 90-item inventory generates measures of the

quan-tity, intensity, and overall severity of adults’ psychiatric

symp-toms. The Family Concept Q Sort was used to elicit parents’

per-ceptions of their families.3’ In this task, subjects sort 80 cards

containing descriptions of family traits and behaviors into nine

categories ranging from least like to most like their families. From the ratings assigned to the cards, two major factors are generated: integration (a measure of family cohesion, communicativeness, and home-centeredness), and adaptive coping (a measure of fam-ily effectiveness, confidence, and sense of control in dealing with the world outside itself).

Medical Information

The hospital records of all enrolled children and eligible but

unenrolled sexually abused children were reviewed. For each

en-rolled abused child, the recorded description of the sexual abuse

episode was compared with information provided directly by the

parent to identify and reconcile discrepancies. To estimate the extent to which abused children had been selectively referred for

care, their previous care at the hospitals was rated as either

none-limited (no prior care or <3 brief hospital visits) or substantial (3

brief hospital visits or more comprehensive services).

School Information

For children attending school, data on academic performance,

classroom behavior, and attendance were extracted from school

records. For children in the first and higher grades, academic

performance in reading or English, mathematics, science, and

so-cial studies was rated. To compare children graded by differing

systems (eg, numbers, letters), academic performance was ranked

using specified criteria as either above, at, or below grade level. Assessments of children’s classroom behavior were based on

teachers’ ratings, as either satisfactory or unsatisfactory, of ten

behavioral dimensions used by the Philadelphia public schools

(eg, “demonstrates responsibility,” “shows appropriate classroom behavior.”) Ratings from schools using other dimensions were

assigned to the Philadelphia dimensions that seemed most similar. For example, the dimensions “works independently’ and “follows

directions” from other schools were assigned to the Philadelphia

dimension “demonstrates responsibility.” To compare children

rated on differing numbers of dimensions, each child’s behavior

was summarized as the percentage of available dimensions that

the teacher had rated as unsatisfactory.

Evaluation and Treatment Unrelated to the Study

Of the enrolled abused children and their families, 47% had

received psychosocial evaluation or treatment for the reported

sexual abuse before the first study interview; a total of 71%

re-ceived such care by the end of the 6-month study period.

Consent

The study was approved by the institutional review boards of the three participating hospitals. Copies of children’s school records were obtained using ordinary request forms signed by

parents or guardians. Information about criminal prosecutions in

Massachusetts was obtained with the approval of the state’s

Criminal History Systems Board.

Statistical Analysis

Data were analyzed using the Statistical Analysis System. To compare the abused and control groups, unpaired Student’s t tests were used for normally distributed, continuous variables, and Wilcoxon rank-sum tests were used for ordinal scaled data. To

analyze categorical variables, Fisher’s exact test and the Pearson test were used. Changes in continuous variables between the times

of study entry and follow-up were examined using paired Stu-dent’s ttests. Baseline predictors of 6-month outcomes were

iden-rifled in univariate analyses and subsequently assessed in a

step-wise multiple linear regression model. To enter and remain in the

regression model, criteria for predictors were, respectively, P .15

and P .10.

Levels of statistical significance were set at P .05 for simple

comparisons and at P .017 for pairwise comparisons between

values for sexually abused children, control children, and

ques-tionnaire norms (Bonferroni t tests).u fl statistical tests were

two-tailed.

RESULTS

Enrollment and Follow-Up

Of 260 sexually abused children eligible for the

study, 154 (59%) were enrolled. Seventy-six children

(29%) could not be enrolled because of parental

re-fusal, 28 (11 %) because a parent could not be

con-tacted within 8 weeks of the child’s hospital visit, in

most cases because there was no home telephone, and

at Viet Nam:AAP Sponsored on September 1, 2020

www.aappublications.org/news

(3)

in two cases (1%) because children were

psychiatri-cally hospitalized. Based on the total numbers of cases

of child sexual abuse substantiated after investigation

by the Philadelphia County CPS agency (Joseph

Spear, Director, Protective Service Programs, Office of

Children, Youth, and Families, Pennsylvania

Depart-ment of Public Weifare, personal communication,

July 10, 1991) and by the Boston/Brookline CPS

agency, and on the proportion of the age-eligible

children identified who were reported to police but

not to CPS agencies, it is estimated that the children

identified and studied comprised 42 and 25%,

respec-tively, of all new age-eligible cases of child sexual

abuse occurring in the two regions during the project

periods. Fifty-three control children were also

en-rolled. Seventy-three percent of the abused children

and 77% of the control children were from the

Phila-deiphia area. The parents of 110 abused children

(71%) and 24 control children (45%) returned for

6-month interviews (P < .001). The mean interval

be-tween initial and follow-up interviews was 6.4

months for sexually abused children and 7.2 months

for control children.

Children’s Characteristics

Selected sociodemographic characteristics of the

enrolled sexually abused and control children are

summarized in Table 1. More of the abused children

were white (P = .04), fewer had health insurance

(P = .02), fewer had received care at a study hospital

previously (P < .001), and more had received

psychi-atric care previously (P = .04). In comparison with the

families who had Medicaid insurance, the families

who had no insurance had higher incomes (P = .01)

and more employed parents (P < .001). Although the

difference did not reach statistical significance, more

of the abused than the control children had mothers

with lower educational attainment (P = .06). By

defi-mtion, none of the control children had histories,

documented or undocumented, of sexual abuse.

The sexually abused children who were eligible for

the study but not enrolled were comparable to the

enrolled abused children with respect to sex,

ethnic-ity, relationship to the adult respondent, health

in-surance status, and prior treatment at the hospital, but

fewer unenrolled children had home telephones (71

vs 87%, P = .001), and more had documented

epi-sodes of prior physical abuse or neglect (21 vs 12%, P = .06).

Table 2 shows selected characteristics of the

en-rolled children’s sexual abuse episodes. The abuse

in-volved vaginal or rectal penetration and/or

oral-genital contacts in 77% of the children. The sexual

abuse episodes of the eligible but unenrolled abused

children were similar.

Behavior Problems Reported by Parents

The enrolled children’s scores for behavior

prob-lems at entry and at 6-month follow-up are

summa-rized in Table 3. At entry, mean scores for the sexually

abused children in each of the four sex/age categories

were significantly higher (P < .001) than the respective

norms. Scores for the control children also tended to

be higher than the norms, but the difference reached

TABLE 1. Sociodemographic Characteristics of Sexually

Abused Children and Control Children*

Characteristic Sexually Control

Abused Children

Children (n = 53)

(n = 154)

Age, y (mean ± SD)+ 7.4 ± 2.6 7.3 ± 2.3

Sex (% female) 78 74

Ethnicity, %

White 28 13j

Black 68 85

Hispanic/other 5 2

Adult respondent, %

Mother 85 91

Father 5 4

Other relative 6 4

Foster parent 5 2

Health insurance, %

Medicaid 49 68

Private insurance 20 19

None 31 13

Maternal education, y 11.9 ± 2.4 12.6 ± 2.211

(mean ± SD)

Annual household

income, %

<$5000 14 17

$5000-9999 43 34

$10000.-19999 26 24

$20 000-29 999 7 20

$30000 10 5

Prior treatment at hospital, %

None/limited 55 211

Substantial 45 79

Telephone in home, % 87 94

Prior psychiatric care 23 10

unrelated to sexual abuse, %

Prior physical abuse 12 4

or neglect, (%)

* Percentages may not add to 100% due to rounding.

t At report of abuse in abused children; at enrollment in control

children. :1:P = .04.

§P= .02.

lIP =.06. IP < .001.

significance only in the largest sex/age category, ie,

girls 6 years of age (P < .001). Also, it was only in

this sex/age category that the difference in behavior

scores between abused and control children reached

significance (P = .008).

Comparison of the abused and control children’s

behavior at follow-up is problematic because attrition

was greater in the control group than in the abused

group and, within the control group, was greater

among better-scoring than among worse-scoring

chil-then. Among children evaluated at follow-up, the

only significant change in behavior score occurred in

the subgroup of younger abused boys, whose mean

score declined. At follow-up, 43% of the abused

chil-dren overall, including 39% of the younger girls, 48%

of the older girls, and 44% of the older boys, received

behavior problem scores in the clinical range (ie,

above the 90th percentile).

Mothers’ Psychiatric Symptoms and Reported

Family Function

Mothers’ reports about their own psychiatric

symp-toms and their families’ function are summarized in

(4)

TABLE 2. Characteristics of Sexual Abuse Episodes of En-rolled Abused Children*

Characteristic of Sexual

Abuse Episode

Enrolled Children

(n = 154)

25.6 ± 14.3

95 32 22 37 7 3 45 9 22 24 39 12 49 67 25 23 44 41 9 6 38 11 24 14 14 70 30 88 12 86 7 6 <2 d 2-30 d 1-6 mo >6 mo

Age of perpetrator, y (mean ± SD)

Sex of perpetrator (% male)

Relationship of perpetrator to child, %

Close relativet

Extended family member Acquaintance

Stranger Other

Duration of abuse, %

<2 days

2-30 days

1-6 mo

>6 mo

Location of abuse, %

Victim’s home Relative’s home

Other

Type of sexual contact, %

Vaginal or rectal penetration

Oral-genital contact

Other contacts only

Time from last abuse to disclosure

Circumstance precipitating disclosure, %

Volunteered by child

Abuse witnessed Child’s behavior Physical problem

Other

Child abuse report, % Filed

Not applicable

Outcome of CPS investigation, %

Report substantiated Report not substantiated

Child’s disposition after investigation, %

No change in residence

Moved to another relative’s home Removed from family

Likelihood that abuse occurred, %

Definitely 62

Probably 28

Possibly or uncertain 11

*Not all variables were known for all children. Percentages may

not add to 100% due to rounding.

tClose relatives include biological parents, step-parents, grand-parents, biological siblings, aunts and uncles.

CPS, child protective services.

§See “Methods” for definitions of categories.

Table 4. At entry, the number, intensity, and overall

severity of psychiatric symptoms of the sexually

abused children’s mothers were higher (P < .001) than

both the respective norms and the corresponding

scores of the control children’s mothers. At entry also,

the abused children’s families had poorer integration

than the norm (P < .001) and poorer adaptive coping

skills than both the norm and the families of the

con-trol children (P < .001). Family function scores of the

children sexually abused by relatives did not differ

significantly from those of the children abused by

nonrelatives.

At follow-up, the mothers of sexually abused

chil-then who completed evaluation reported fewer psy-chiatric symptoms and a lower intensity of symptoms than they had reported at entry (P < .001).

Neverthe-less, they continued to report more psychiatric

symp-tomatology (P < .001) and poorer overall family

func-tion (P < .01) than the norm. The mothers of control

children who completed follow-up also reported

more psychiatric symptomatology than the norm at

follow-up (P < .001). The absence of change between

entry and follow-up in the control mothers’ level of

symptoms indicates that attrition was greater among

control mothers who initially had reported less

symp-tomatology.

School Placement, Performance, and Behavior

At entry, 119 (77%) of the sexually abused children

and 40 (75%) of the control children were enrolled in

school; of these, 84 and 29 children, respectively,

at-tended first and higher grades. The school data

col-lected at entry reflected periods entirely preceding the

abuse in 45% and before first disclosure of the abuse

in 92% of the sexually abused children.

Data on the children’s school placement,

perfor-mance, and behavior are shown in Table 5. At study

entry, 31 % of the abused children and 11 % of the

con-trol children were assigned to special educational

classrooms or to grades 1 or 2 years below those

ap-propnate for their chronological ages (P = .02). The

abused children’s academic performance was worse

than the control children’s in reading, science, and

social studies (P < .01). Nevertheless, 52% of the

abused children performed at or above grade level in

all of the tallied subjects. School absence related to

disclosure or treatment of the sexual abuse did not

appear to contribute to poor academic performance,

since the absence rates of the abused and control

chil-dren were comparable (12 vs 9%, P = NS). At study

entry, the abused children received a higher

propor-tion of unsatisfactory classroom behavior ratings than

did the control children (P = .04).

None of these differences between the sexually

abused and control children remained evident at

follow-up, due mainly to disproportionately higher

attrition among control children whose school

place-ments, performance, and behavior had been

satisfac-tory at study entry.

Relationships Between Parents’ and Teachers’

Evaluations of Abused Children

Correlations between parents’ and teachers’ ratings

of the abused children’s behavior were 0.36 (P = .001)

at study entry and 0.42 (P < .001) at follow-up. Of 76

sexually abused children whose parents rated their

behavior at home and whose school performance,

be-havior, or both could be assessed at follow-up, only

22 (29%) were functioning in the normal range in all

rated domains.

Potential Predictors of Behavior Problems Among

Sexually Abused Children

To evaluate the extent to which baseline

character-istics of the sexually abused children predicted the

children’s behavioral status at the 6-month follow-up

(5)

TABLE 3. Behavior Problem T Scores of Sexually Abused

and at 6-Month Followup*

and Control Children at Study Entry

Sex/Age Category At Entry

(Norm for

cat’’o SD) Sexually Control

0 7 - Abused Children

At Follow-up .

Sexually Control

Abused Children

Children Children

Girls <6 y 60.7 ± 10.4 (42) 54.7 ± 8.9 (14) 57.9 ± 12.5 (234 61.6 ± 12.7(5)1

(48.2 ± 10.2)

Girls 6 y 64.5 ± 10.4 (774, 58.9 ± 10.1 (24) 64.6 ± 11.0 (654 61.6 ± 8.4 (15) (50.6 ± 9.5)

Boys <6 y 62.1 ± 13.1 (11)j 61.2 ± 11.9 (5) 51.8 ± 8.1 (6)11 48.0(1) (50.9 ± 9.7)

Boys 6 y 66.2 ± 12.1 (224 56.7 ± 11.3 (9) 64.4 ± 14.1 (164 55.0 ± 10.2(3)

(50.5 ± 9.6)

* Mean T score ± SD (No. ofchildren). Higher scores indicate more behavior problems. Scores were

not obtained at entry for two abused children and one control child. 1:P< .001 vs norm for sex/age category.

§P = .008 vs control children.

II

P< .001 for change from entry to follow-up among children who completed follow-up. 1P < .01 vs norm for sex/age category.

TABLE 4. Psychiatric Symptoms and Family Function Reported by Mothers of Sexually Abused

Children and Control Children

Mean Scores at Entry Mean Scores at Follow-up

Sexually Abused Control Sexually Abused Control

Children’s Children’s Children’s Children’s

Mothers Mothers Mothers Mothers

Psychiatric symptoms* Symptom total (22.0)

(n = 126) 37.7

(n 43)

20.5

(n = 86) 29.6,II

(n = 20)

35.6

Symptom distress index (1.37)

Global severity index (.36)

1.92

.86

1.45

.37

1.75

.61,II

1.68 .73

Family functiont (n = 131) (n = 47) (n = 101) (n = 24)

integration (5.34) 4.84 4.861 5.031 5.05

Adaptive coping (5.17) 4.M 4.98 4.83 4.91

*Higher scores indicate more symptomatology. Values in parenthesis denote norms for women.

t Higher scores indicate better integration and adaptive coping. Values in parentheses denote norms

for mothers.

j: P < .001 vs control mothers. §P < .001 vs norm.

II

P < .001 for change from study entry to follow-up among mothers who completed follow-up.

I P < .01 vs norm.

point, the children were first separated into a series of

paired subgroups determined by the presence or

ab-sence of selected baseline characteristics. The mean

behavior problem scores at follow-up of the

sub-groups in each pair were then compared (Table 6).

None of the differences between subgroups was

sta-tistically significant.

The following additional baseline variables were

considered as potential predictors: child’s age;

moth-er’s educational attainment; variables reflecting the

family’s competency and social supports (time of

hos-pital presentation, whether the mother could identify

people

who provided her with tangible assistance);

mother’s psychiatric status; and the family’s

integra-tion and adaptive coping. Seventy-four cases with

baseline information on all these variables and on all of the characteristics listed in Table 6 except

likeli-hood of abuse were included in a stepwise multiple

regression model.

In the regression model, more problematic

behav-ior in the children at 6-month follow-up was

signifi-cantly (P :S .10) associated with four variables: older

age of the child (R2 = .04, P = .04), lower educational

attainment of the child’s mother (R2 = .04, P = .06),

higher overall severity of the mother’s psychiatric

symptoms (R2 =

.03,

P = .04), and poorer family

in-tegration (R2 = .21, P < .001). Together, these variables

accounted for 31 % of the variance in children’s

be-havior problems at follow-up (P < .001). Neither

char-acteristics of the sexual abuse, nor circumstances of its

disclosure, nor clinical indicators of the competency

of the abused child’s family were significantly

asso-ciated with the children’s behavior problems at

follow-up. The power of the regression model to

de-tect an association between the children’s behavior

problem scores and any predictor variable or

combi-nation of variables was .92?

DISCUSSION

Comparisons of Sexually Abused Children, Control

Children, and Norms

At the outset of this study, a number of differences

were found between the sexually abused children and

the control children that antedated the disclosures of

abuse. Higher proportions of the abused children had

received psychotherapy unrelated to the abuse and

had relatively poor school placements, performance,

(6)

TABLE 5. School Placement, Academic Performance, and Teachers’ Behavior Ratings of Sexually

Abused and Control Children*

At Entry At Follow-up

Sexually

Abused

Control Children

Sexually Abused

Control

Children

Children Children

Special or below-grade classroom placement 31 (1164 11 (35) 32 (85) 21 (19)

Performance below grade levelt

Reading 36 (75) 13 (24) 33 (64) 36(14)

Mathematics 29 (75) 13 (24) 28 (64) 21 (14)

Science 21 (67)11 4 (23) 19 (57) 14(14)

Social studies 26 (66)1 5 (21) 17 (59) 43(14)

Percentage of behavior ratings classified as

unsatisfactoryt

0-20% 62 (79)** 80 (25) 64 (70) 57(14)

21-40% 9 (79) 4 (25) 9 (70) 21 (14)

41-60% 9 (79) 8 (25) 7 (70) 14(14)

61-80% 4 (79) 0 (25) 10 (70) 0(14)

81-100% 16(79) 8(25) 10(70) 7(14)

* Percent of group (no. of children with information available). t See “Methods” for description of variable.

:1:P = .02 vs control children. §P = .03 vs control children.

II

P = .005 vs control children. 1P = .004 vs control children.

** P = .04 vs control children.

TABLE 6. Relationship of Selected Clinical Characteristics of Sexually Abused Children to Their

Behavior Problem Scores at FollowUp*

Characteristic Mean Behavior Problem Score (No. of Children)

Characteristic Present Characteristic Absent

Female sex 62.6 (88) 59.7 (22)

Black race 63.1 (69) 60.2 (41)

Household income <$10 000 62.6 (57) 59.1 (51)

Prior abuse or neglect 67.6 (10) 61.6 (95)

Abused by dose relative 61.4 (32) 62.6 (78)

Abuse duration 1 mo 61.7 (58) 62.5 (46)

Penetrating abuse 63.4 (65) 61.1 (39)

Likelihood of abuse judged definite 62.3 (70) 61.9 (40)

Disclosure volunteered 63.9 (40) 61.0 (68)

Disclosure within 2 d 62.9 (48) 61.7 (56)

Supportive mother 61.9 (89) 62.9 (20)

Mother without dose friend 65.3 (29) 60.9 (80)

* P> .05 for all differences between mean behavior problem scores.

and behavior. They also tended to have higher rates

of prior abuse or neglect and to have mothers with

lower educational attainment, although the

differ-ences did not reach significance. Overall, these

observations are consistent with findings by

others’4’6-9 that problematic family function

accom-panies disturbances among sexually abused children,

even when the abuser is not a family member. These

data also support hypotheses18”#{176}’41 to the effect that

children with unsupportive or distressed families

may be particularly vulnerable to sexual

victimiza-tion. The epidemiologic importance of such clustered

risk states and problems in children has been

empha-sized by Statheld.

At study entry, ie, after the abuse disdosure, the behavior of the older girls, the psychiatric status of the

mothers, and the adaptive coping skills of the

chil-then’s families were all significantly less favorable

than both population-based norms and the

corre-sponding characteristics of the control children. The

control children’s ratings on these factors at entry

tended to be less favorable than the norms, but in

most instances not significantly so. This latter

obser-vation is not unexpected, inasmuch as the control

children were drawn from a largely poor, urban

population.

At follow-up, no noteworthy changes in most

as-pects of the abused children’s status were observed.

The reduction in psychiatric symptoms of the abused

children’s mothers may have reflected

recovery-whether accompanied or not by psychotherapeutic

intervention-from initial distress occasioned by

their learning of their children’s sexual abuse.

Com-parsons between the abused and control children at

follow-up must be viewed with caution because

at-trition was disproportionately greater among the

better-functioning control children and families. This

may have occurred because better-functioning

con-trol families were less motivated to maintain contact

with the research staff or because poorer-functioning families were more likely to experience the research

activities as supportive. Conceivably, providing a

(7)

nancial or other tangible incentive might have

im-proved the retention of control families.

Prevalence, Persistence, and Prediction of Disturbed Function Among Sexually Abused Children

Behavioral and academic dysfunction were

wide-spread but not universal among the sexually abused

children studied. At follow-up, of the children whose

parents rated their behavior and whose school

per-formance, behavior, or both, could be assessed, only

29% were functioning normally in all rated domains.

On the other hand, 57% of the children had normal

behavior problem scores, 52% of the students were

meeting expectations for academic achievement, and

the classroom behavior of 64% was satisfactory on

80% or more of the rated dimensions.

Gomes-Schwartz et aP found similarly that at the end of an

18-month observation period, 75% of the sexually

abused children in their sample had no more

psy-chopathology than the general population.

Several factors more prominent at entry among the

sexually abused than among the control children

were also predictive, to varying extents, of

problem-atic behavior within the group of abused children at

follow-up. These factors were a history of prior child

abuse or neglect, lower maternal educational

attain-ment, greater severity of mothers’ psychiatric

symp-toms, and poorer family integration.

Our observation that no clinical feature of the

sexual abuse itself predicted the children’s behavior

at follow-up was somewhat unexpected. Although

some investigators have found no consistent

associa-tion between children’s disturbance and the

fre-quency or duration of sexual abuse,#{176}’ the types

of sexual acts performed,20’ or the child’s

relation-ship to the abuser,#{176}’’43 others have found

associa-tions between certain of these factors and disturbance

in children or adults after sexual abuse.’1’

Eligible but Unenrolled Sexually Abused Children

The sexually abused children who were eligible for

this study but not enrolled differed from the enrolled

children in only two respects: they were somewhat

more likely to have prior histories of physical abuse

or neglect and less likely to have telephones.

Twenty-nine percent of their households lacked telephone

ser-vice, compared with only 7% of United States

house-holds generally. These findings suggest that

sexually abused children who are relatively

inacces-sible for clinical research may have more severe

dys-function and may suffer greater social isolation than

those who can more readily be studied.

Limitations of Study

Factors that may have limited this study’s capacity

to demonstrate a relationship between sexual abuse

and children’s later behavior problems include the

control group’s uneven attrition and the short

follow-up period. Because parents rated the

chil-dren’s behavior, mild or circumscribed emotional

dif-ficulties in children may not have been detected.

However, in studies of normal, psychiatrically

symp-tomatic, and abused children, parents’ CBCL ratings

of children have correlated well with reports by

chil-then themselves47’ and by teachers.49 There is also

considerable evidence that psychiatrically

symptom-atic mothers accurately report poorer adjustment in

their children.#{176}’1

Confounding by low socioeconomic status could

limit the validity or generalizability of the study’s

re-sults, especially in relation to more economically

ad-vantaged children. The CBCL was selected to

meas-ure behavior problems because it has demonstrated

stability across the range of socioeconomic strata.24

Because the household incomes of our abused and

control groups were similar, it is difficult to attribute

the differences observed between the two groups at

study entry to poverty alone. Within the abused

group, neither household income nor race (proxies for

socioeconomic status) predicted the children’s later

behavior problems, although the range of incomes

was narrow. Factors similar to those found predictive

of children’s behavior at follow-up in this study have

also been noted to correlate with psychological status

after childhood sexual abuse in middle dass

women.5”1’2

Management Implications

As suggested by Fromuth5 and Tong et al,37 the

ex-pectation that sexually abused children wifi have

poor outcomes may inadvertently produce a

self-fulfilling

prophecy. Undue emphasis on the

antici-pated impact of sexual abuse also may divert the

at-tention of parents and clinicians from other important

sources of abused children’s distress. The results of

this study suggest that preexisting, long-standing

ad-verse psychosocial circumstances other than low

so-cioeconomic status per se contribute importantly to

the problematic behavior and academic performance

seen later among some sexually abused children. The study findings further suggest that it is children’s

pre-existing psychosocial circumstances, rather than the

abuse, that determine, at least in part, the nature of

their functional outcomes. Eimer has observed

simi-larly in a prospective comparison of abused and

con-trol children that, after 8 years, the effects of

socio-economic disadvantage overshadowed those of

physical abuse and neglect. On the other hand, these

observations offer cause for guarded optimism that

some sexually abused children, especially younger

children and those with psychologically healthy

mothers or well-functioning families, may escape

se-rious untoward consequences of sexual abuse, at least

in the near term.

ACKNOWLEDGMENTS

This work was supported in part by grants from the Robert

Wood Johnson Foundation Research and Development Program

to Improve Patient Functional Status, the Children’s Hospital of

Philadelphia MediCal Associates Research and Education Fund,

and the Biomedical Research Support Fund of the Trustees of Health and Hospitals of the City of Boston, Inc. The views and

opinions expressed in this paper are those of the authors and may

not reflect those of the Robert Wood Johnson Foundation.

We are indebted to Leslie Campis, PhD, Lisa Gazdick, Shawn

Heister, Karim Khanbai, Karen O’Jennos, Joan Shirley, Victoria

Szabo, Nina Tomassini, and Ann Marie Weldon for research

as-sistance; to Matt Wehmus for computer programming to Timothy

Heeren, PhD and Janine E. Janosky, PhD for statistical analyses; to

John P. Delaney, Esq. and the Norfolk and Suffolk County,

Mas-sachusetts Victim-Witness Advocates for legal follow-up; to JoelJ.

Alpert, MD, Alvan R. Feinstein, MD, Stephen Ludwig, MD,

Carolyn Newberger, EdD, Eli Newberger, MD, and Toni Seidl,