We propose to examine the effectiveness of an early childhood home visiting program to reduce child maltreatment after dissemination to at-risk mothers in Connecticut, through analysis of administrative data. Child maltreatment has life-long impacts on children and is difficult to prevent, but there is evidence that home visiting can reduce maltreatment in the research setting. The effectiveness of home visiting in decreasing child maltreatment after dissemination to community settings is not well established. One challenge in evaluating maltreatment accurately is surveillance bias, which is the increased chance of a report that a family in the home visiting program has, compared to if they were not in the program. These extra reports are hypothesized to be for less severe situations. One way to assess maltreatment that avoids surveillance bias is to examine medical care for injuries, which can be considered a proxy measure for child maltreatment. Although many injuries are accidental, if one group has more injuries than another, this may be evidence that they either have a less safe environment (potential neglect) or have been physically abused. In Connecticut, approximately 40% of first-time mothers are screened for maltreatment risk each year for a voluntary home visiting program; this screening takes place in birth hospitals, through referrals from prenatal care providers, and other social services. Home visits take place for up to 5 years. We propose to examine child maltreatment by comparing families in the program to demographically-similar families who were not in the program. We will create a comparison group using birth certificate data from children born during the same time period, but not in the program for a variety of reasons (including program availability and not being screened). We utilized propensity score matching to identify similar families using demographic and health behavior data available from the birth certificates (such as neighborhood characteristics, adequate prenatal care, and maternal age). We have a sample of 4,886 program families and 23,904 comparison families. We will link these groups to both Child Protective Services (CPS) data and Medicaid/CHIP data to create 2 separate datasets. Using logistic regression and survival analysis with CPS data, we will compare investigated and substantiated reports; child removals from the home; and timing and severity of reports. Surveillance bias will be estimated using the difference-in-differences approach. Using logistic regression and the Medicaid/CHIP data, we will determine if injuries and injury-related hospitalizations are decreased in the program group. This project would lead to a better understanding of how to evaluate maltreatment in the context of home visiting.
Project Relevance One goal of home visiting programs for new mothers is to reduce child abuse and neglect, a serious public health problem with life-long impacts on children. Although home visiting research trials have produced evidence for reductions in maltreatment, this evidence was mixed and few studies have examined these outcomes in community-replication sites. This project will examine the impact of a statewide home visiting program on child maltreatment, measured both by Child Protective Services determinations and by health care utilization for injuries.
