Redefining the Science and Policy of Early Childhood Intervention Programs

David Olds and his team are to be applauded for creating the Nurse-Family Partnership (NFP) home-visiting program and for steadfastly following up with the participants in the Memphis randomized controlled trial. In 2 new “child”¹  and “mother”²  articles, they report the impacts of NFP on children and mothers, respectively, when the children reach age 18.

All too often in the field of home-visiting, observers want to decide that a program either “works” or does not work. If 1 trial, 1 outcome variable, or 1 subgroup shows a null or adverse effect, there may be a rush to declare that the program failed and should be defunded. If 1 evaluation shows a positive impact for a targeted subgroup, there may be support for at least modest funding, but it is not clear that the population benefits or public health is served. This field is not like drug trials or the Food and Drug Administration,³  for which the stark goal is to “separate the relative handful of discoveries which prove to be true advances from a legion of false leads and unverifiable clinical impressions”; instead, home-visiting programs have multiple goals for the mother and child, are implemented with different subgroups in different contexts at different times, assess outcomes in diverse domains across different ages in child development, and use diverse methods of data analysis (eg, covariance, moderation, and mediation) to understand mechanisms. To have a population impact, they must be embedded in a broader system of care for families. To contextualize my comments with full disclosure, I note that I lead and study a public health approach to home-visiting that connects families with community resources, including targeted home-visiting programs similar to NFP.

The scientific question in these articles is not whether the NFP program works but how it might be effective, for whom, for which outcomes, and under what conditions. The policy question that these articles raise is how to craft a comprehensive system of care for communities in which public health and population well-being are the goals, and home-visiting is 1 tool.

As in past articles, the authors report few overall main effects on the group receiving NFP. Of the 31 outcomes reported in Tables 4 and 5 of the child article, 1 (the percentage graduating with honors) yielded a statistically significant main effect of NFP; of the 20 outcomes reported in Table 4 of the mother article, 3 (public-benefit expenditures, percentage married, and number of months current spouse is employed) yielded a significant main effect of NFP.

Consistent with the scientific goals of understanding for whom and how a program might work, the authors examine effects on subgroups. No community can afford to deliver NFP to all low-income first-time mothers, so strategic targeting is essential. Olds et al⁴  have asserted previously that NFP is best targeted to a high-risk subgroup: “one of the clearest messages that has emerged from this program of research is that the functional and economic benefits of the nurse home-visitation program are greatest for families at greater risk.” Risk is defined in the child article as being born to mothers with limited psychological resources to cope with adversity in the lower half of the distribution on an index composed of maternal intellectual functioning, mental health, and sense of mastery and/or self-efficacy (mothers’ beliefs about the importance of and confidence in accomplishing key NFP behavioral objectives) measured at baseline.¹  They hypothesize the positive impact of NFP for children of limited-resource mothers and find, among this group, that those receiving NFP fare better than controls in 8 of 31 tests.

A quandary arises from the mother article, however, which addresses the impact of NFP on government costs for the Supplemental Nutrition Assistance Program, Temporary Assistance for Needy Families, and Medicaid. The authors report that implementing NFP yields a net savings of $16 269 (computed as the cost of NFP [$12 578] minus savings [$28 847]) for the subgroup with a high level of psychological resources. In contrast, there were no overall NFP-control differences in government expenditures for women with limited psychological resources. In fact, for women with limited psychological resources, assignment to NFP is associated with a net loss of $1041 (computed as the cost of NFP [$12 578] minus savings [$11 537]; not tested for significance).

So who should receive NFP, families with limited psychological resources or those with a high level of psychological resources? Surely, the answer is not as simple as the conclusion about NFP from Social Programs That Work,⁵  which reports “sustained effects on important outcomes for mothers and their children” without qualification by subgroups. How do we reconcile findings of no positive impact of NFP on children but savings in government expenditures for families with a high level of psychological resources, and how do we reconcile a positive impact of NFP on children but no positive impact on public-benefit expenditures for families with limited psychological resources? The positive effects on children in families with limited psychological resources might suggest that NFP should be delivered to these families, but the null effect on public-benefit expenditures suggests that a financial return on investment might not be the justification.

Might it be that the positive impact of NFP on children in limited-resource families comes about directly because these families use a lot of government resources to help their children? Perhaps these mothers were empowered by NFP to navigate public resources effectively. We need to reconsider the valence attached to the outcome variable of public-benefit expenditures as a positive rather than adverse effect, in which case, the benefits of NFP for limited-resource families accrue in outcomes for children but not in financial savings from government supports.

And what about the group of low-income, first-time mothers with high levels of psychological resources? The findings suggest that without NFP, these families will cost taxpayers large amounts in Supplemental Nutrition Assistance Program, Temporary Assistance for Needy Families, and Medicaid payments without any benefits to the children. What program should these families receive?

As important as identifying who should receive NFP is the task of figuring out how to find these families to deliver this program to them. The findings clearly indicate that selecting families solely on the basis of demographic characteristics (ie, low-income, first-time mothers) is inadequate. As Olds et al⁴  previously argued, “This pattern of results challenges the position that intensive programs of this type for targeted at-risk groups ought to be made available on a universal basis. Not only is it likely to be wasteful from an economic standpoint, but such an approach may lead to a dilution of services for those families who need them most because of insufficient resources to serve everyone well.” The findings suggest we need to bring NFP to families with limited psychological resources, but given how complicated the formula is to identify this subgroup (ie, they must have low intelligence, high mental health needs, and a low sense of mastery), how do we find them at a population level to have a public health impact?

The answer is that we need to create a universal system of care in which all birthing families are reached early in pregnancy, screened for psychosocial and financial needs, and matched with the community resources they need to succeed. One family might need and benefit from NFP, whereas another family might need substance-abuse intervention, job training, or a housing loan. This is a public health approach that is accountable for population outcomes. It is the infrastructure that does not yet exist but will be necessary for targeted programs like NFP and other intensive interventions to be sustained in the long run. A system of care does not compete with targeted programs like NFP; rather, it facilitates their reach and effectiveness.

The findings in these articles call into question reliance on analyses of savings in public-benefit expenditures as the primary rationale for supporting the NFP early childhood intervention. Return on investment is the calling card that perhaps should be discarded. Young children are worth supporting even if government dollars are not saved.

The current mixed findings should not lead us to throw the baby out with the bath water, so to speak. The NFP program is a brilliant invention that has paved the way for many contemporary early childhood interventions. However, let us reconsider the return on investment, identify who benefits and in what ways, and figure out how to create a universal system of care so that families can receive the community resources they need.

NFP

Nurse-Family Partnership