Randomized Evaluation of the Nurse Family Partnership in South Carolina

The Nurse-Family Partnership (NFP) was launched by pediatrician David Olds in the early 1970s in response to the observation that low-income mothers are less likely to engage in positive health and parenting behaviors that are associated with better child development outcomes (Olds et al. 1986). NFP is a prenatal and infancy home visiting program for low-income, first-time mothers and their families. Registered nurses begin visiting their clients as early in the pregnancy as possible, helping the mother-to-be make informed choices. The nurses continue visiting regularly until the child is two years old.

2016-04-01

2022-03-31

We will assess the effect of NFP on a range of short- and long-run maternal and child outcomes using administrative data that will be available for all members of both treatment and control groups. Several of the outcomes we will evaluate are required as part of the Medicaid waiver and pay-for-success financing model, and the state is committed to making these data as well as many additional long-run outcomes available to the research team.

We will begin with three outcomes of particular interest to the stakeholders: pre-term birth, birth spacing, and child injuries. These outcomes will form the basis of the pay-for-success payments. Data for pre-term birth and birth spacing will be obtained from Vital Records, collected and maintained by the South Carolina Department of Health and Environmental Control. Data on child injuries will be obtained from an all-payer encounter database overseen by the South Carolina Data Oversight Council.

Our study also aims to assess outcomes of broader interest to the academic and policy communities beyond those specified in the pay-for-success assessment of NFP. The original, smaller-scale studies of NFP suggested that it improved the mothers’ outcomes as well as the children’s, and might have long-term implications for health and well-being that would be of first-order importance in assessing the benefits of public financing of the program. We will also collect administrative data on additional outcomes including health care utilization, mortality, education, abuse and neglect, employment/earnings, use of government services, credit, and crime. South Carolina has a particularly well-developed administrative data infrastructure, and these outcomes are all measurable in administrative data collected by the state and linkable based on the identifiers that will be collected.

This broader evaluation will provide a more comprehensive assessment of the impact of the NFP program, both over a wider array of impacts and over a longer period of time. This will yield useful evidence for the State of South Carolina and for policy-makers nationally, who are interested in comprehensive analyses of the social and economic impact of visiting programs like NFP and on the health and financial consequences of including such visits as a Medicaid benefit.

This study is an individual-level randomized controlled trial, with randomization post-consent. The study will enroll low-income, first-time, Medicaid-eligible pregnant women at less than 28 weeks gestation. All eligible applicants who provide written consent will be randomly assigned either to a treatment group that is offered access to NFP or to a control group that is not. Two-thirds of those recruited will be randomized to the treatment group and one-third to the control group. Control group members will have access to the standard of care and whatever other programs and services are available in the community. Study enrollment, on-the-spot randomization, and the baseline survey will be conducted by NFP staff using encrypted tablets and Computer Assisted Personal Interview (CAPI) software.

On-the-spot randomization using a tablet (Survey CTO)

The unit of randomization is the individual (a pregnant woman who meets the eligibility criteria for program participation).

No

6,000 women (individual-level randomization without clusters)

6,000 women and their children

4,000 women in treatment group
2,000 women in control group

We have performed power calculations as part of the preparation for the Medicaid waiver and pay-for-success contract specifications. Our power analysis is based on expected effect sizes suggested by previous NFP trials and NFP data on the target population. We use the effect sizes of a 20% drop in pre-term birth (with a baseline mean rate of preterm birth of .12), 18% reduction in birth spacing of less than 2 years (baseline mean of .14), and a 13% reduction in child injury (baseline mean of .18); we assume that 95% or more of women randomized into the treatment group receive any NFP services and at most 5% of the control group that receives a home visiting program that has as much impact as NFP. NFP plans to serve 4,000 families in South Carolina between January 2016 and January 2020. Our power analysis suggests that with 4,000 families in the treatment group, a 2:1 T:C ratio (or 2,000 control group families) will achieve 80% or higher chance of detecting statistically significant effects on two out of the three primary outcomes, with the greatest power to detect child injury and the lowest to detect pre-term birth. We anticipate having substantially greater power for many of the additional outcomes that we will assess for the broader evaluation (e.g. utilization of government benefits or the mother’s subsequent arrests). We will further develop our plans for these measures during the pilot period.