A woman holds her baby inside a mobile home near Essex, England.Kirsty Wigglesworth / AP

Nurse-Family Partnership is a decades-old program in which registered nurses visit poor, first-time single mothers, starting early in their pregnancies and continuing until the child turns two. The nurses nudge the mothers to give up smoking and drinking, give them tips on child development, and answer questions about their own health or that of their children.

In the U.S., Nurse-Family partnership is considered “proven” to work. It’s gone through three randomized trials with big sample sizes—400 to 1,189 women—and it has consistently shown to boost the health of the mother, to improve the child’s well-being and safety, and to curb preventable deaths. One recent analysis found that though the program costs only $1.6 billion, it will save the federal government $3 billion in spending on things like Medicaid and food stamps by 2031. It’s already up and running in more than 600 U.S. counties, and a provision of the Affordable Care Act offers the states $1.5 billion to expand home-visiting programs like NFP.

That’s why a new study published in the Lancet is so surprising. The U.K. has a similar program, called, somewhat confusingly, Family-Nurse Partnership. The British government thought it was working so well that in 2013 they expanded it. But the results of a randomized trial involving 1,645 mothers, published last week, found that the program did basically nothing. Just as many moms who were visited by nurses had a second pregnancy within two years. More of the kids who were receiving the nurse visits wound up in the hospital than those in the control condition.

“Programme continuation is not justified on the basis of available evidence,” the study authors wrote.

This isn’t a sign that home-visiting for new moms isn’t effective after all. Similar models have shown modest, but good, results in the Netherlands and Germany.

One reason the U.K. results were so lackluster might be that the Lancet study compared home-visiting to “usual care,” or the health care new moms would receive anyway. But since the U.K. has universal health care and the U.S. does not, it could be that there was less of a difference between the two British groups than the two American ones. (The study also focused on young moms, using their youth as a proxy for low income, so its subjects might have been less disadvantaged than their American counterparts to begin with.)

Furthermore, most of the American studies on NFP were conducted by the program’s own developer, David Olds, a professor at the University of Colorado Denver. Scientifically, it’s not ideal to have the same person who came up with a program trying to determine whether the program is any good.

Overall, the Lancet study shows how difficult it is to design the perfect program to cure society’s ills, especially entrenched, complicated ones like poverty and poor health.

Everyone wants to help kids and moms, but these programs are hard to evaluate. How do you prove that a child would have gotten abused or sick had the nurse not come? How do you control for cultural and socioeconomic factors, or for the fact that some moms really click with their nurse visitors, and some don’t? Even in the U.S., NFP yields different results when the moms were visited by lay people rather than nurses.

As this study shows, even the most proven, ironclad programs struggle with these problems. It makes improving child and maternal health that much harder for policymakers.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.