This study utilized administrative data from a large urban county to determine if latent risk classes could be established from the CPS records of children who experienced maltreatment re-report or maltreatment recurrence within 12 months of index case closure.
Youth who experienced maltreatment re-report or recurrence were drawn from the population of all youth who experienced a maltreatment investigation or substantiation in a major urban county over a three-year period. Five mutable risk factors were extracted from the assessment of maltreatment that was completed by the investigating caseworker. Risk factors included child emotional/physical disability, caregiver history of domestic violence, caregiver history of drug or alcohol problems, caregiver mental health problems, or caregiver parenting deficits. Covariates included youth age, race, gender, and initial maltreatment allegation type.
Latent class analysis was used to establish distinguishable risk classifications for the sample. Output included a probability of class membership and an estimated mean for the proportion of cases containing a particular risk factor in each latent class. Goodness of fit statistics included a Likelihood Ratio Test, Bayesian Information Criteria, and Entropy statistic. To examine demographic differences across the latent classes Chi-square and Fishers’ exact tests were conducted in a post-hoc analysis.
Maltreatment re-report: A 3-class model best fit the data (G = 58.15, p<.00, BIC = 26319.76, Entropy = .68, n = 4369). Class 1 accounted for 56% of the sample and could be characterized by “Few Identified Challenges.” Class 2 accounted for 26% of the sample and could be characterized by “Mental Health and Domestic Violence.” Class 3 accounted for 18% of the sample and could be characterized primarily by “Drug and Alcohol and Domestic Violence.” A higher proportion of youth age 0 – 1, Native American / Alaska Native youth, and youth referred for neglect were observed in Class 3 “Drug and Alcohol and Domestic Violence.” Allegations of physical abuse and mental/emotional abuse were higher in Class 1, “Few Identified Challenges.”
Maltreatment recurrence: A 3-class model best fit the data (G = 61.52, p < .01, BIC = 4633.71, Entropy = .64, n = 694). Class 1 accounted for 48% of the sample and could be characterized by “Few Identified Challenges”. Class 2 accounted for 15% of the sample and could be characterized by “Mental Health” challenges. Class 3 accounted for 37% of the sample and could be characterized by “Domestic Violence, Mental Health, and Parenting” challenges. No sample covariates were associated with membership in a particular recurrence class.
Mental health, domestic violence, drug and alcohol, and parenting challenges were present in the maltreatment re-report and recurrence risk classes. This underscores the complex and multidimensional challenges families who come to the attention of child welfare experience. Just under one-half of re-reporting and recurrence cases were characterized by few identified risks, however. This indicates a need for enhanced screening and assessment practices which identify maltreatment related risks. Implications for future research will be discussed.