Mother and Infant Home Visiting Program Evaluation-Strong Start

The study's conceptual framework has three broad aspects: (1) inputs (factors influencing service delivery), (2) outputs (services delivered) and (3) outcomes for families. Each site in MIHOPE-Strong Start will use either the NFP or HFA home visiting model.

Community resources provide a foundation from which programs operate and are considered an input that influence how services are provided to families. In particular, these determine the outside referral services available to home visiting programs and the opportunities available to families in both the program and control groups. By connecting pregnant women with services, home visiting programs can change mothers' health care use, health behaviors, and health status, which in turn can lead to improvements in birth outcomes. In communities where services are limited, however, home visiting programs may have limited capacity to improve mothers' connections to such supports. Furthermore, a program's ability to improve outcomes is influenced by the "treatment differential," or the difference between service receipt among families who receive home visiting and families who do not receive home visiting. The more that control group members receive services that are similar to those provided by HFA and NFP, the smaller the "treatment differential." Community context also includes community characteristics that could affect norms toward use of social services and health care, or other influences on control group help-seeking and program group responses to home visiting programs. These contextual factors can affect program impacts in both positive and negative directions. The service model defines the program plan. It includes information such as the intended goals of the home visiting program; the expected frequency, duration, and content of home visits; and intended linkages with other services.

The MIHOPE-Strong Start implementation study will document the extent to which the two evidence-based models and their local counterparts have defined clear, coherent, and well-specified service plans for helping parents to obtain prenatal care, other services needed to reduce the risk of poor birth outcomes, and infant health care in the first year of life - a critical prerequisite to delivering services that could affect these outcomes.

In MIHOPE-Strong Start, the "service model" can be defined at two levels: the service model is defined by the national evidence-based model and then it may be refined or adapted by the agency that is operating a local program. It is important to clearly understand how HFA and NFP define their models since these models showed efficacy in prior research. At the same time, local programs often deliberately adapt models to fit their local contexts. Given inconsistent impacts on birth outcomes in past studies, in fact, such adaptations could be one path to improvements in program impacts.

There are multiple organizational influences on how a home visiting program defines its service model and its implementation system. These organizations include the local program operator, the purveyor of the evidence-based model that has been adopted (HFA or NFP), the state MIECHV (Maternal, Infant, and Early Childhood Home Visiting Program) grantee (if the site is participating in MIECHV), and community organizations with which the local agency collaborates.

The implementation system includes the resources for carrying out the service model. It incorporates policies and procedures for staff recruitment, training, supervision and evaluation; assessment tools, protocols and curricula to guide service delivery; administrative supports; organizational climate regarding fidelity and the use of evidence-based practices; available consultation to address issues beyond the home visitor's skills and expertise; and the program's relationships with other organizations to facilitate referral and service coordination.

Other program characteristics that affect the services delivered include the attributes of staff in a given program. NFP specifies that home visitors should be registered nurses with a minimum of a baccalaureate degree in nursing. HFA gives local sites considerable discretion in this regard. Moreover, an individual staff person's own psychological well-being can influence how they approach their work with families. In addition to these global attributes, staff may vary in their degree of focus, confidence, and competence in carrying out responsibilities with respect to particular outcomes, due to variations in staff training and supervision. The implementation study will therefore be designed to understand the extent to which not only the service model, but also the implementation system and individual staff, are focused on activities that are expected to improve birth outcomes and maternal and infant health care use.

The attributes of families who enroll in a given home visiting program will also affect the program's opportunities to affect birth outcomes. HFA and NFP specify the characteristics of families that their programs can serve, yet even within each national model, local programs sometimes vary in the families they target either because of community characteristics, or because they vary in their processes for family recruitment. Baseline attributes of families who enroll can, in turn, influence services because staff are expected to tailor services to the family's strengths, needs, and concerns; because families vary in their understanding of the program and the benefits they are likely to derive from it; and because parents vary in their capacity, whether psychosocial or because of material resources, to engage with the services offered.

These inputs - the service model, implementation system, and characteristics of home visitors and families - all affect the outputs, or the services that families receive. Because home visiting programs rely heavily on referrals to other community organizations to meet families' needs, these outputs can include services provided directly by home visiting staff and referrals to other services.

NFP and HFA programs are designed to affect mothers' prenatal outcomes, including use of recommended levels of prenatal care, prenatal health behaviors related to birth outcomes such as smoking and use of alcohol or other substances, and mothers' prenatal health. These prenatal outcomes may influence birth outcomes, and birth outcomes may directly affect infant health outcomes, health care use and costs. By improving birth outcomes, parenting behaviors may also improve. Furthermore, home visiting services may indirectly improve infant health and health care use regardless of impacts on birth outcomes by improving parenting behaviors.

Based on this framework, MIHOPE-Strong Start will address the following broad research questions:

• What is the impact of home visiting programs that use one of these two evidence-based models on birth outcomes, maternal and infant health, and health care use up to the first year postpartum? How do impacts vary for key subgroups, such as smokers and young mothers?
• What is the impact of programs using each evidence-based model on the outcomes of MIHOPE Strong Start? The design is also intended to provide information that would allow actuaries at the Centers for Medicare and Medicaid Services (CMS) to estimate the effects of the programs on Medicaid costs.

The implementation study for MIHOPE-Strong Start will document the key features of HFA's and NFP's service models and implementation systems (at the national and local levels) that are expected to affect birth and health outcomes. The implementation research will answer these specific questions:

• How is each evidence-based service model - HFA and NFP - defined?
• How do local home visiting programs specify or adapt their service models relative to the national models with which they are affiliated?
• To what extent are local service models and implementation systems focused on preterm birth and related outcomes?
• What dosage of services do families actually receive in local programs and how much does it differ from the intended dosage?
• What kinds of referrals are provided to community services that could affect birth outcomes and the child's and mother's health?
• How do programs' inputs (such as the two evidence-based models, the extent of focus on birth outcomes, family characteristics, staff attributes, and community characteristics) relate to achieved outputs (in particular, the dosage of services received and referrals provided)?

Finally, the study will examine the intersection of impacts and implementation to answer the research question:

• How do home visiting programs using these two evidence-based models achieve their results?

To provide unbiased estimates of the effects of home visiting programs, families who are recruited into the study will be randomly assigned either to a program group that can receive home visiting services or to a control group that can use other services available in the community. Although the feasibility of carrying out random assignment must be assessed community-by-community, discussions with states and local programs thus far indicate that the need for home visiting services far exceeds the capacity of local programs in most places, allowing for the ethical creation of a control group. Other than home visiting services from the programs participating in MIHOPE-Strong Start, control group members can receive services available in the community for which they would normally be eligible. Control group members will receive referrals to such services. After their child reaches one year of age, those assigned to the control group will be able to receive home visiting services, if they are eligible for the local program. The evaluation will adhere to all ethical standards for program evaluation and has undergone human subjects review by the MDRC Institutional Review Board.

An impact analysis will estimate the effects of home visiting on prenatal health care use, birth outcomes, infant health, and maternal and infant health care use until the infant is one year old. The analysis will start with an analysis for the full sample, by evidence-based model (HFA and NFP), and for key subgroups. In all three cases, results will be presented for an "intent-to-treat" analysis that compares all program group members-regardless of whether they actually received home visiting services-with all control group members, some of whom may have received home visiting outside the MIECHV program. State Medicaid and vital records data will be collected from each of the 18 states and will provide all follow-up data, as well as some baseline data about sample members.

An implementation study, designed to complement the impact study, will collect information on community context, influential organizations, the service model, the implementation system, home visitors, families, and actual service delivery. The proposed evaluation will rely on multiple sources of data to understand how home visiting programs are implemented and what factors affect the quality of implementation. These data include information from each model's management information systems, interviews with home visitors and program managers at local sites, interviews with state administrators, and community characteristic information from the U.S. Census. Collecting basic implementation data across such a large number of sites will enable MIHOPE-Strong Start to provide evidence about which program variations are most effective at improving birth outcomes and maternal and infant health care use. It will also provide information about how programs can be designed to best improve these outcomes in the future.