6P
Configural Frequency Analysis of Psychotropic Medication Use Among Youth in Child Welfare Systems: A Community Perspective
Methods: We used data on 3264 children aged between 2-17.5 years surveyed in the baseline wave (April 2008 to December 2009) of the second National Survey of Child and Adolescent Well-being (NSCAW II). We linked these data to US census data in order to obtain a multilevel data set containing child-level characteristics (sociodemographic, maltreatment history, insurance status, and child need as measured by the Child Behavior Checklist [CBCL]); caregiver variables (education, employment, marital status, and income); and county characteristics (population, percent of child in poverty, education ratio, and neighborhood safety and community involvement). An analytic technique, Configural Frequency Analysis, was used to identify groupings of factors at the child- caregiver, and community levels that were predictive of medication use. CFA is a technique to identify clusters of variables that occur more (called ‘types’) or less (called ‘antitypes’) than expected by chance.
Results:414 children (12% of the weighted sample) were taking psychotropic medications. At the child level, a type consisting of Medicaid insurance, prior history of physical abuse, and a CBCL score of 64 or above occurred more than what would be expected by chance (p<0.0001). Another child-level type consisting of children aged between 6-10 years, Black Non-Hispanic race/ethnicity, female sex, and placed out-of-home was also identified as having higher than expected use of psychotropic medications (p< 0.0001). A caregiver level type containing individuals with high school or above education, above 200% of Federal Poverty Line, and full-time employed, were significantly associated with reduced medication use than would be expected by chance (p<0.0001). A community level type containing children not using medications, living in a community with few unsupervised children, safer neighborhoods, less help from neighbors, was also identified. Another community type containing children not using medication, living in a community that has a large population (66.7 percentile), medium level of percentage of children in poverty, low level of unemployment rate, and high ratio of juvenile arrests also occurred more often than expected.
Implications: The use of CFA in this manner allows us to go beyond what variables predict risk of service use (e.g., high CBCL scores and psychotropic medication use), to understand groups of variables that represent risk (e.g., for what kinds of children is a high CBCL score the most predictive?). Expansion of CFA techniques can better help child welfare agencies predict risk clusters within their served population, and contribute to the psychological well-being of children in the child welfare system.