This study evaluates the effectiveness of two delivery models of the ProKind intervention—a German adaptation of the Nurse-Family Partnership (NFP)—in supporting high-risk families. The continuous model (CM) involves home visits exclusively conducted by NFP-trained midwives, while the tandem model (TM) combines visits by a midwife during pregnancy with follow-up visits by a social worker postnatally. A total of 755 disadvantaged pregnant women were randomly assigned to intervention (n = 394) or control groups (n = 361). Outcomes include child protection service (CPS) surveillance, foster care placements, and pediatric hospital visits. The study is based on a longitudinal, multicenter randomized controlled trial (RCT) with two follow-up phases: Phase I (birth to 36 months) and Phase II (around the child’s 7th birthday). Average treatment effects were estimated using augmented inverse probability weighting (AIPW) combined with lasso variable selection. The CM significantly reduced pediatric hospital visits in both follow-up phases. In Phase II, there were also suggestive trends toward reduced CPS surveillance and fewer foster care placements. No significant effects were found for the TM on any outcome. Full-sample analyses confirmed significant reductions in hospital visits during Phase II. These results underscore the relevance of delivery models in home-visitation programs and suggest that sustained, midwife-led interventions can yield long-term benefits for vulnerable families.