Maternal Hospital Screening for Risk and Child Protective Service Interactions

Brief Summary of Abstract:

Families were screened for risk when admitted to hospitals for delivery and assessed for postnatal home visiting services. The rates of Child Protective Services (CPS) involvement were compared between hospitals which implemented risk screening compared to those with no such program. Using logistic regression, it was found the risk screening process accurately identified families at elevated risk of CPS involvement while controlling for variation in birth characteristics between hospitals. Substantial differences in mother and child characteristics were found given hospital screening implementation, and counterfactual analyses indicated there was no detectable screening bias.

Background and Purpose:

Identifying family child care needs early has been targeted as a solution to curbing future child protective services interactions. Currently, a risk screening process for mothers at the time of delivery to target home visiting referrals for families at heightened risk of Child Protective Service (CPS) involvement is implemented in select hospital within a geography. We use administrative records to identify which families were targeted as high risk and their subsequent involvement with CPS versus births identified as low risk or born in hospitals without risk screening.

Methods:

Families with a child born between 2011 and 2017 were selected for the study. The population of births (n=285,327) were linked to hospital screening and child protective service records using a probabilistic, machine learning linkage program called ChoiceMaker. This linkage allowed us to follow families in both hospital and CPS records, with individual trajectories of service usage from birth to age 5. The demographic and case characteristics of all families were gathered from birth records, with chi square tests identifying how subsets of populations differ based on experienced service usage. Logistic regression used the 2011 and 2012 birth cohorts to identify the relationship of the screening program and subsequent CPS interactions for children.

Results:

Hospitals with risk screening programs had mothers who were younger, lower education attainment, and a larger proportion covered with public insurance. Mothers in hospitals with the risk screening program were not at elevated risk versus hospitals without risk screening. Mothers with “high risk” scores were indeed 31% more likely to have a referral to CPS within 5 years from birth, 34% more likely to have a substantiation, but no increased risk for foster care placement. Families enrolled with home visiting programs were 84% more likely to be referred to CPS, 92% more likely to have a referral for maltreatment substantiated, and 30% more likely to experience foster care placement.

Conclusions and Implications:

Hospital risk screening programs offer opportunities for providers to intervene with at-risk families before CPS must be involved. Families screened in as high risk were also contacted by CPS at higher rates, indicating the risk screening process is correctly identifying families which would benefit from intervention. However, the significant relationship between home visiting enrollment and CPS involvement indicates bias for at-risk families. Surveillance of these families by mandated reporters could be capturing referrals which would otherwise go undetected in a hospital without risk screening processes.