After decades of continuous decline, many developed countries are now experiencing sharp increases in home birth rates. For example, home births in the US increased by almost 30% between 2004 and 2009, while the fraction of home births in the UK almost tripled between 1990 and 2006. Proponents of home births generally argue that labor is a natural process that is inherently safe in a majority of cases: medical interference in the delivery process might increase risk, e.g. because cesarean sections are more likely to be performed and this increases the risk of complications in subsequent pregnancies. On the other hand, a hospital birth may provide better facilities and equipment, better hygiene (sterility) and proximity to other medical services in case of emergency.
There is little debate on the benefits of using more intensive medical treatments for high-risk babies. As such, research has mostly focused on low-risk newborns. The ideal way to study the impact of delivery location (home versus hospital) on low-risk newborn health would be a randomized controlled trial. In this case, one would randomly assign expecting mothers to give birth at home or in a hospital and then compare the differences in the health of their newborns. However, previous efforts to conduct such a trial proved to be impossible because pregnant women strongly object to having their birth location randomized. Hence, our knowledge of how home births impact infant health comes exclusively from studies relying on observational data.
The majority of the research on this topic is conducted by medical scientists in studies that compare average outcomes between low-risk women planning to give birth at home versus the hospital after taking into account observable differences (e.g. age, gestational age) between the two groups. However, such correlation studies can be misleading: even among low-risk women, those who give birth at home or in a hospital may have different risk factors, with riskier deliveries usually taking place in the hospital. Therefore, babies born in a hospital often seem to have more health problems than babies born at home.
To address this issue, Reyn van Ewijk, Mircea Trandafir, and I compared two groups of low-risk women who are identical, except that the women in one group have a higher probability of delivering in a hospital only because they live closer to a hospital. We conducted our study using data from the Netherlands, the only developed country where home births are widespread. We found that giving birth in a hospital leads to large reductions in infant mortality. Our back-of-the-envelope calculations suggested that the rise in hospital births explained roughly 46–49% of the reduction in infant mortality in the Netherlands between 1980 and 2009. However, we also found that hospital births mainly helped newborns of low-risk lower-income women with no effects on the health of the babies of low-risk higher-income mothers.
These results caution against policies that encourage home births for all low-risk women. The Netherlands is a country where maternity care is provided using a rigorous process of risk selection based on both past medical history and the current health status and development of the mother and the fetus. The Dutch maternity system is explicitly geared toward midwife-supervised home births. Yet, even with a relatively sophisticated model of risk selection, we find that the babies of some women classified as low-risk benefit from hospital deliveries. That these women are lower-income suggests that risk selection is more difficult and less precise for women with lower socio-economic status. Having a good understanding of risk selection must go hand-in-hand with crafting policies about childbirth technologies.
© N. Meltem Daysal
Early-life medical care and human capital accumulation, by N. Meltem Daysal
We recognize that IZA World of Labor articles may prompt discussion and possibly controversy. Opinion pieces, such as the one above, capture ideas and debates concisely, and anchor them with real-world examples. Opinions stated here do not necessarily reflect those of the IZA.