The diagnostic utility of sexual behavior problems in diagnosing sexual abuse in a forensic child abuse evaluation clinic
Introduction
The desire for a reliable indicator of sexual abuse has led clinicians to look at sexual behavior problems as both a screening and as a diagnostic tool. Although sexually abused children as a group exhibit more sexual behavior problems than nonabused children Adams et al 1995, Cosentino et al 1995, Dubowitz et al 1993, Friedrich 1993b, Friedrich et al 1992, Friedrich et al 1986, Gale et al 1988, Goldston et al 1989, Hibbard and Hartman 1992, McClellan et al 1997, Mian et al 1996, Wells et al 1997, Wells et al 1995, White et al 1988, the utility of behavioral measures in a diagnostic assessment remains in question. In a research review, Kendall-Tackett, Williams, and Finkelhor (1993) found that sexual behavior problems and post-traumatic stress disorder could differentiate sexually abused children from nonabused children. However, they also found that a third of sexually abused children exhibit no such symptoms. This asymptomatic group of sexually abused children thus limits the utility of such behavioral tools because of the risk of increasing the false negative rate of diagnosis.
Another potential problem for these tools is the false positive rate. How good a discriminator for sexual abuse is the presence of sexual behavior problems? Are there other determinants of sexual behavior problems apart from sexual abuse? The various studies that have indicated an association between sexual behavior problems and sexual abuse raise four methodological questions. What methods were used to measure sexual behavior problems and how reliable are these methods? How was the diagnosis of sexual abuse made and, importantly, were sexual behavior problems used in the diagnostic assessment?
Early studies of the relationship between sexual behavior problems and sexual abuse relied primarily on the Child Behavior Checklist (CBCL) (Achenbach, 1991) to measure sexual behavior problems. The CBCL is a questionnaire that asks caretakers to rate the frequency of a variety of behaviors over the previous 6 months. The CBCL divides behavior problems into internalizing and externalizing problems. Internalizing problems include withdrawal, somatic concerns, depression, and anxiety. Externalizing problems include aggression, delinquency, and conduct problems. Six items of the CBCL comprise the sex problem scale (Friedrich et al., 1986). These items are: behaves like the opposite sex, plays with own sex parts too much, has sexual problems, thinks about sex too much, wishes to be the opposite sex, and plays with own sex parts in public. The sex problem scale is scored for only a limited number of age groups.
Studies using the CBCL’s sex problem scale found that sexually abused children have higher scores than nonabused children Dubowitz et al 1993, Friedrich et al 1987, Hibbard and Hartman 1992. Using three items of the sex problem scale, Dykman et al. (1997) found that sexually abused boys exhibited significantly more sexual behavior problems than sexually abused girls. However, a limitation of the scale is that the handful of items on the CBCL sex problem scale do not adequately sample the range of sexual behaviors generally exhibited by children Achenbach 1991, Friedrich et al 1992.
The Structured Interview for Symptoms Associated with Sexual Abuse (SASA) was developed as a screening tool for sexual abuse in children (Wells, McCann, Adams, Voris, & Dahl, 1997). The interview inquires about 26 symptoms, five of which pertain to sexual behavior. Sexually abused boys were found to be more likely to demonstrate curiosity about sex and to be sexually aggressive. As with the CBCL, sexual behaviors were only a small subset of a range of behavior problems identified by the instrument.
Faced with the limitations of the CBCL sex problem scale, Friedrich (1997) developed a comprehensive measure of sexual behavior problems, the Child Sexual Behavior Inventory (CSBI). After the development of three preliminary unpublished versions of the instrument—the CSBI (Friedrich et al., 1992), the CSBI-R (Friedrich, 1993b), and the CSBI-3 (Friedrich, 1995)—the CSBI was formally published in 1997 by Psychological Assessment Resources (Friedrich, 1997). All versions of the instrument measure sexual behavior problems based on the numerical frequency of sexual behaviors over the 6 month period prior to assessment. All three use the child’s primary female caretaker as the informant. The published CSBI includes a total of 38 sexual behavior items.
Versions of the CSBI have been administered to samples of nonabused children in an attempt to identify the overall frequency of sexual behaviors in children Friedrich et al 1998, Friedrich et al 1991. These studies indicate that nonabused children ages 2 to 12 years exhibit a variety of sexual behaviors, the most frequent being self-stimulating behaviors, exhibitionism, and problems with personal boundaries. Intrusive sexual behaviors were reported infrequently in these samples. An inverse relationship between sexual behaviors and age was also identified, with older children demonstrating less sexual behavior. These findings are consistent with other reports that some sexual behaviors, particularly self-stimulatory behaviors, curiosity about sexuality, and exposure of genitals to other children, are normal childhood behaviors, at least in certain age groups Cavanagh Johnson 1993, Lindblad et al 1995.
The utility of the CSBI in distinguishing sexually abused from nonabused children was described in two studies that used two of the earlier unpublished versions of the CSBI Friedrich 1993b, Friedrich et al 1992. Two recent studies also found differences between sexually abused children and nonabused children when utilizing versions of the CSBI Cosentino et al 1995, Mian et al 1996.
Studies of nonabused children also identified a number of additional variables related to sexual behavior problems, such as maternal education, family nudity, life stress, hours spent in daycare, and the presence of domestic violence Friedrich et al 1991, Friedrich et al 1998. Gray, Busconi, Houchens and Pithers (1997) also found a direct relationship between domestic violence and sexual behavior problems in a population of sexually aggressive children. Several studies found an association between scores on the CBCL and the CSBI Cosentino et al 1995, Friedrich et al 1991, Friedrich et al 1998, indicating that sexual behavior problems exist concurrently with, and are not necessarily independent of, other behavior problems. It thus appears that sexual behavior problems have multiple etiologies.
How the diagnosis of sexual abuse is arrived at in the studies that look at behavioral indicators is central to the utility of these indicators. If behavioral indicators are used in the diagnostic assessment, then the conclusion that they are useful becomes a self-fulfilling prediction.
In many studies, the method used to classify children as sexually abused was not clearly described. A few studies applied reasonably stringent, although infrequently found, criteria such as perpetrator confession (e.g., Wells et al., 1995). Several studies relied primarily on substantiation of abuse as determined by child protective services or law enforcement Adams et al 1995, Friedrich et al 1992, Gray et al 1997, McClellan et al 1997, Mian et al 1996, Wells et al 1995. Others used children identified as abused in specialized sex abuse diagnostic clinics Cosentino et al 1995, Dubowitz et al 1993. However, the diagnostic system utilized in these clinics was not clearly described. Some studies provided no clear description of the criteria used in classifying children as sexually abused Dykman et al 1997, White et al 1988.
Professionals assigned the task of investigating sexual abuse may be influenced by the presence of sexual behavior problems when making a diagnosis of abuse. For example, Kendall-Tackett and Watson (1991) found that professionals were more likely to believe that sexual abuse had occurred when children under the age of 6 were exhibiting advanced knowledge about sex. Thus, it is possible that, in the studies cited above, some children are included in abuse samples whose diagnosis of abuse was influenced by the presence of sexual behavior problems. Such confounding of variables would constitute a methodological weakness in any conclusion that sexual abuse is associated with sexual behavior problems.
This methodological weakness led us to analyze the diagnostic utility of sexual behavior problems in a sample of children assessed for sexual abuse in a forensic child abuse program where behavioral indicators are not used in the diagnostic process. Thus, this sample of children offers an opportunity to examine the relationship between sexual abuse diagnosis and sexual behavior problems in a setting where diagnosis and behavior are independent measures.
Section snippets
Methods
The sample of 247 children constitute all children evaluated at the Spurwink Child Abuse Program (CAP) for sexual abuse allegations between September 1, 1997, and March 31, 1999, for whom results from the Child Sexual Behavior Inventory (CSBI) are available. Because the CSBI is normed for children age 2 to 12 years in age, children evaluated who were younger than 2 years or older than 12 years were excluded from the sample. The two largest sources of referrals for these children were medical
Results
The diagnostic results are summarized in Table 3. The team found evidence of sexual abuse (Strong or Moderate) in 24% of the cases, while no evidence of abuse was found in 61%. The remaining 15% were classified as Don’t Know.
Descriptive statistics for the five behavioral variables are presented in Table 4. The children on average obtained high scores on the five behavioral variables. Because all five variables are expressed as T-scores, the expected sample mean would be 50 and the standard
Discussion
This study found no significant relationship between sexual abuse diagnosis and sexual behavior problems as measured by the Child Sexual Behavior Inventory among children assessed for sexual abuse in a multidisciplinary forensic evaluation setting. Similarly, the study found no significant relationship between sexual abuse diagnosis and behavior problems as measured by the Child Behavior Checklist. There was a high level of correlation between the sexual behavior problem scores and the other
Conclusions
The study found no significant relationship between a diagnosis of sexual abuse and the presence or absence of sexual behavior problems in children referred for a forensic sexual abuse evaluation. Although some sexually abused children exhibit serious sexual behavior problems, this study indicates that community professionals should use caution when using sexual behavior problems as a diagnostic indicator of abuse. Indeed, such use runs the risk of increasing the false positive rate if a child
Acknowledgements
The authors would like to thank William Friedrich and David Finkelhor, for their review of preliminary manuscripts, and Robert I. Kabacoff for his statistical consultation.
References (31)
- et al.
Sexually inappropriate behaviors in seriously mentally ill children and adolescents
Child Abuse & Neglect
(1995) - et al.
Sexual behavior problems and psychopathology symptoms in sexually abused girls
Journal of the American Academy of Child and Adolescent Psychiatry
(1995) - et al.
A follow up study of behavior problems associated with child sexual abuse
Child Abuse & Neglect
(1993) Sexual victimization and sexual behavior in childrena review of recent literature
Child Abuse & Neglect
(1993)- et al.
Sexual abuse in young childrenits clinical presentation and characteristic patterns
Child Abuse & Neglect
(1988) - et al.
Behavioral problems in alleged sexual abuse victims
Child Abuse & Neglect
(1992) - et al.
Preschoolers’ sexual behavior at daycare centersan epidemiological study
Child Abuse & Neglect
(1995) - et al.
Relationship between sexual abuse, gender, and sexually inappropriate behaviors in seriously mentally ill youths
Journal of the American Academy of Child and Adolescent Psychiatry
(1997) - et al.
The effects of sexual abuse on 3–5 year old girls
Child Abuse & Neglect
(1996) - et al.
A validational study of the structured interview of symptoms associated with sexual abuse (SASA) using three samples of sexually abused, allegedly abused, and nonabused boys
Child Abuse & Neglect
(1997)