O4347: Child Welfare Outcomes Among Children with and without Nurse Family Partnership: A Propensity Score Matched Assessment in Florida from 2010-2023

Background: Nurse-Family Partnership (NFP) trials provide the strongest evidence for the prevention of child maltreatment through home visiting programs. Home visiting programs have been widely implemented and serve diverse populations at risk for child welfare involvement; yet, program effectiveness on child welfare outcomes distal to a report or a verified report are limited. We examine child welfare outcomes subsequent to a maltreatment report among NFP-served and comparison children and hypothesized that NFP participation would be associated with improved child welfare outcomes.

Methods: We linked Florida NFP and birth certificate datasets to identify first born children of primiparous birthing individuals in Florida between January 1, 2010 and September 7, 2023. Using multivariable logistic regression, we created propensity scores for NFP participation by dyad sociodemographic characteristics: age; race; ethnicity; birthplace; education level; marital status; prenatal care; insurance and child’s year and county of birth. Our sample was 38,257 children (6436 NFP and 31,821 comparison). A Florida child welfare administrative dataset was used to link NFP-served and comparison children with a maltreatment report during the study period. We generated odds ratios where the sample at risk for each subsequent outcome was the previous outcome- for example, the odds of an investigated report among those with any report, the odds of a verified report among those with an investigated report, and the odds of being removed from the home among those with a verified report. The final three outcomes - kinship placement, family permanency and termination of maternal rights - were all calculated among those ever removed from the home. We estimated program effects on maltreatment prevention and family preservation child welfare outcomes for the two cohorts of children reported to child welfare using generalized linear mixed models.

Results: We identified 1,597 NFP and 6,392 comparison children with a child welfare report. NFP children had significantly greater odds of reports (Odds Ratio [OR] 1.33, 95% confidence interval [CI] 1.24 -1.41) but not investigations (OR 0.91, CI 0.79-1.05), verification (OR 1.06, CI 0.92-1.21), removal (OR 0.99, CI 0.78-1.24), kinship placement (OR 1.21, CI 0.78-1.88), familial permanency (OR 0.71, CI 0.48-1.04) or termination of parental rights (OR 1.24, CI 0.62-2.48).

Conclusion: Among children with a child welfare report, post-report child welfare outcomes did not differ between NFP children and comparison children. This finding ran counter to our hypothesis perhaps because (1) in this context, NFP does not provide the same child welfare benefits, (2) despite our use of propensity scores, our intervention and comparison populations are not comparable, or (3) we did not have the sample size needed to detect program effects. Future research should increase sample sizes to detect what maybe small but significant program effects and develop approaches to explicitly account for child welfare policy and practice heterogeneity.