306P
Can Symptom Profiles in Sexually Abused Children be Predicted? an Exploratory Analysis

Schedule:
Saturday, January 17, 2015
Bissonet, Third Floor (New Orleans Marriott)
* noted as presenting author
Matthew J. Cuellar, MSW, PhD Student/Graduate Research Assistant, University of Tennessee, Knoxville, Knoxville, TN
Rebecca Bolen, PhD, Associate Professor, University of Tennessee, Knoxville, Knoxville, TN
Background. Research has shown that a number of symptoms are related to child sexual abuse. A more complex question concerns the predictability of the type of symptom presentation for child sexual abuse victims. Although the literature has suggested that sexually abused children do not have a unique set of symptoms, reasons exist to suggest that, within sexually abused children, different symptom groups might appear. Because of the influence of children’s internalized working model(s) of attachment, it is possible that symptom groups might emerge based on the confluence of the secure/insecure continuum and the continuum representing the underactivation/ hyperactivation of the attachment system. The purpose of this exploratory study was to determine whether child victims of sexual abuse can be categorized by similarity in symptom profiles, and whether these symptom profiles are theoretically predictable based upon developmental and relational factors.

Methods. The original sample included 92 sexually abused children and their guardians who visited an outpatient hospital clinic for the children’s forensic/medical exam. After accounting for missing data on key variables, final sample size was 73. Children and their guardians completed measures capturing the children’s symptomatology and competence; guardian attachment, stress in parenting, and postdisclosure support; the child’s perception of the quality of the child/guardian relationship; and demographics and other abuse-specific variables. Using latent class analysis (LCA), children were originally grouped across 3 to 7 clusters based on specific symptom and competence scales. Comparing across the different clusters, the 4-cluster solution had the least classification errors (0.017%) and the greatest entropy R2 (96.12%). After ANOVAs compared potential predictors of cluster placement (i.e., symptom groups), forward stepwise multinomial logistic regression analysis (FSMRA) determined those variables most strongly related to cluster assignment.

Results. Clusters differed by both magnitude and shape, with clusters representing high depression and high anxiety groups, as well as more moderate groups. In addition, clusters differed by parent reports of competence. All symptom and competency variables across clusters were significant at p < .001 except for one that was significant at p < .008. When cluster assignment/symptom group was placed in the FSMRA as the DV, the final model of significant predictors correctly classified 68.1% of cases. Significant variables were the primary caregiver’s level of secure attachment, child’s greater or lesser need for psychological proximity to the caregiver, the caregiver’s perception of the levels of distractability and parental reinforcement of the child, age of abuse onset, and gender.

Conclusions and Implications. Attachment and the child’s rating of the quality of the child/guardian relationship were important predictors of cluster assignment, as well as the child’s overall developmental stage as captured by age of abuse onset. Other important variables were the caregiver’s perception of the child as an indicator of stress of parenting. These findings suggest that attempts can be made to ascertain not only the elevation of symptoms in sexually abused children, but also the type of symptomatology with which they present.