Over the last decade, the rate of U.S. births happening outside of hospitals has been steadily on the rise, as more expectant mothers seek out a relaxed, private space to bring their children into the world. In six states, including Alaska, Pennsylvania and Washington, more than three percent of all births occur beyond the maternity ward.

But how safe are home births?

Perhaps the answer lies in the Netherlands, the only developed nation in the world where such arrangements are relatively common. Roughly a quarter of the country's births took place at private homes between 2000 and 2008.

The infant mortality rate is also higher in the Netherlands than in European countries with similar medical resources. But as more Dutch mothers have switched in recent years to delivering their babies in hospitals, rather than their homes, doctors have noticed a drop in newborn deaths.

Researchers wanted to understand why, between 1980 and 2009, the country's infant mortality rate fell from 4.25 deaths per 1,000 births to 2.42 deaths per thousand births. Over the same period, the share of deliveries in maternity wards swelled from about 61 percent to 73 percent. On the surface, it appeared the home method may simply be riskier for babies.

But the story isn't that simple. What happened to the women during their home births depended a lot on their income, and relatedly, their access to routine medical care before and after pregnancy, according to a study published this month in the American Economic Journal: Applied Economics. 

Researchers tracked about 356,000 Dutch women who gave birth between 2000 and 2008. None in the sample had any documented health problems, making them eligible within the country's universal health-care system to choose between a home birth with a midwife, but no doctor, and a hospital delivery.

A woman whose household earned above the country's median income didn't face a greater risk of losing her child if she opted for a home birth over the maternity ward, the study found. This group's 28-day infant mortality rate, no matter where they gave birth, was about 2 per 1,000 births.

But a poorer woman who preferred a home birth was more likely to encounter tragedy. The 28-day infant mortality rate for them more than doubled, from about 2 deaths per 1,000 births to 5 deaths per 1,000 births.

One possible reason is that low-income families have less access to medical care, in general, even in a country with universal health coverage like the Netherlands. They may live farther away from a hospital and lacking transportation or simply time between jobs, skip the routine appointments higher-income couples tend to prioritize.

Low-income families could also miss certain health issues and incorrectly identify as "low risk."

"Some women classified as 'low risk' benefit from a hospital birth and their babies would have been saved if their mothers had delivered in a hospital," said study author N. Meltem Daysal, an economics professor at the University of Southern Denmark.  "Low-income women benefit especially from a hospital birth."

The implication abroad could serve as a warning in the United States, where out-of-hospital births are becoming more popular, increasing from 1.3 percent in 2011 to 1.4 percent in 2012, according to the latest numbers from the CDC.

The more relevant message, Daysal emphasized: Proximity to medical care is measurably important to infant health, especially for poor families. "Any mortality reductions from a hospital birth are likely due to the medical care provided after delivery,” the authors wrote. “A hospital birth may reduce infant mortality through various channels, such as the availability of better facilities and equipment, potentially better hygiene or the proximity to other medical services.”

Causation, of course, is difficult to prove.

Between 1980 and 2009, the infant mortality rate in the Netherlands fell from 4.25 deaths per 1,000 births to 2.42 deaths per thousand births. Over the same period, the share of deliveries in maternity wards swelled from about 61 percent to 73 percent.

Far fewer American women deliver their babies at home, and yet America's infant mortality rate is much higher at 6 deaths per 1,000 births.

Some findings may still compel expectant mothers to think twice about where they give birth, Daysal said, and the care they seek in the following weeks. Most of the mortality decline in the Netherlands came from newborns who weigh more than 5.5 pounds -- the least likely to be "high-risk," the authors note, and are therefore eligible for home births.