Learning from Null Results: A Machine Learning Approach to Exploring Heterogeneity in the Impact of Nurse Home Visiting in South Carolina

Effective interventions to address adversity in the transition to parenthood for low-income families have tremendous potential to transform adult outcomes. Home visiting programs, such as the Nurse Family Partnership, are considered “evidence based” policy by policymakers and clinicians seeking to improve outcomes for low-income pregnant people and their children in early life. We conducted a randomized controlled trial of the expansion of the Nurse Family Partnership in South Carolina. We found limited evidence of impact on the study’s primary adverse birth outcomes. We seek add-on funding to enable us to pursue two additional objectives for the evaluation. First, we will pursue a machine learning approach to understanding heterogeneity of impacts across participants’ characteristics and contextual factors, such as telehealth implementation during COVID-19. Second, we will host a dissemination event that will allow for a comprehensive presentation, discussion of trial results, and conversations about necessary next steps toward more effective interventions to improve maternal and child well-being for low-income families.