Homebirth Debate... The Research

"I could never take a chance like that with my baby!"

The homebirth debate. What does the research really say about the safety of homebirth?

By Jamie Mahurin Smith

Sooner or later, just about any woman planning a home birth will hear some version of this reaction. Most people here in the US assume that hospital birth must be a safer choice than birth at home. "Studies shows that homebirth is safe for low-risk women," you might respond. But what, exactly, do the studies show? When you run across anti-homebirth sentiment in the newspaper or on the web, what's it based on?

I am not a statistician or an epidemiologist: I am a mother who looked long and hard at the safety data on homebirth before deciding to stay home for the births of my two youngest sons. I'll always be glad I did. I wish the discussion weren't so heated, and so rife with misinformation.

Why is it so hard to get good information?
The randomized controlled trial (RCT) is widely accepted as the gold standard in medical research. To study homebirth in this way, we'd take 20,000 low-risk pregnant women and assign half of them to deliver at home and half to deliver in the hospital, observing outcomes in each group.

It's never going to happen, though, because large-scale randomization of birthplace is neither ethical nor feasible. And when women choose their own birthplace, a researcher can't be certain exactly how she is similar to or different from other low-risk women. Do women who opt for homebirth share some characteristics that affect birth outcomes? Are they different, somehow, from women who reject homebirth as an option? There are no definitive answers. Given that we can't do a perfect study, let's talk about what researchers can accomplish.

Who's got the power?
Statistical power tells you something about the certainty with which a researcher can report his or her conclusions. Power increases with size -- that's one of the advantages of our hypothetical RCT with 20,000 women. As a general rule, a study of 5000 women can tell you more than a study of 1000 women because statistical blips tend to even out over larger samples. Be cautious about conclusions drawn from small samples, whether they're favorable toward homebirth, as in this study of this study of homebirth in Switzerland, or suspicious about its safety, as in this Irish study.

Looking back or looking forward?

Hindsight is 20/20, we say in most contexts. But in research, you'll see more clearly when you look ahead than when you look behind. The abstract of any study will tell you whether it's retrospective (one that looks back at a set of events) or prospective (one in which the researcher planned in advance to collect a specified set of data on a group of women). One of the strengths of the study published in 2005 by Kenneth Johnson and Betty-Anne Daviss was its prospective design: more than 400 certified professional midwives agreed in advance to collect information about all births they attended in the year 2000.

It's axiomatic in epidemiology that retrospective studies are more vulnerable to bias, and there are a couple of egregious examples in the homebirth literature. The Irish study linked in the previous section, on the hazards of small sample
size, had the additional weakness of being a retrospective study; this critique points out that the study's beginning and ending dates seem calculated to make outcomes look as poor as possible. In a case like that, one has to wonder about the authors' agenda.

In the US, researcher Jenny Pang made headlines in 2002 when her study of homebirth in Washington state concluded that homebirth raised the risk of perinatal mortality by one- or two-tenths of one percent. Her study, however, was retrospective: she looked back at birth certificates that didn't indicate planned birthplace, instead of asking women ahead of time to designate where they planned to deliver their babies.

Serious perinatal problems like neonatal death or brain damage occur only rarely, and misclassifying one or two of them can alter a study's conclusions. When you read about a retrospective study, always ask how the researchers validated their findings. If they haven't corroborated their data (in the Pang study, for instance, that might have involved contacting a subset of the families to ask where they initially planned to give birth), take the conclusions with a big spoonful of salt.

Apples to...cantaloupes?

In comparing homebirth and hospital birth, it's crucial to compare apples with apples. If a study looked only at outcomes for women who delivered at home, it would give falsely positive information about homebirth because it would exclude
the women who transferred to the hospital during labor -- in other words, most of the women who experienced complications. Studying homebirth safety requires a researcher to include outcomes for all the women who planned at the beginning of labor to deliver at home, whether the birth ultimately occurred at home or in the hospital.

Similarly, studies on hospital birth outcomes must take into account iatrogenic complications -- those caused by medical intervention. A further weakness of Jenny Pang's study was the choice to consider only vaginal births among the
hospital cohort. One of the drawbacks to planning hospital birth is the likelihood, approaching 30% in this country, that you'll wind up with a C-section. In focusing exclusively on vaginal births, Pang was free to disregard all the complications resulting from unnecessary cesareans.

C-section complications affect not only mothers, whose risk of dying triples, but their babies as well. New research published in the September 2006 issue of Birth concluded that the mortality rate for babies delivered by C-section was almost triple that of babies born vaginally, even when known risk factors were accounted for.

When you're reading research, the fine print matters. Is the study considering antepartum deaths (those that occur before the onset of labor), intrapartum deaths (those that occur during labor), neonatal deaths (those that occur in the first 28 days of life), or some combination? If a baby dies as a result of a congenital malformation, is that included in the statistics or not? If a mother makes a higher-risk choice (breech birth at home, or declining assessment after 42 weeks), are those outcomes included? In Mark Durand's 1992 study of births at The Farm, every single death is included. As you'd expect, the mortality rate in his study is quite a bit higher (17/1707) than that of more recent studies, which frequently exclude antepartum deaths, higher-risk births, and deaths resulting from congenital anomalies.

How much does it matter?

Flip a nickel a thousand times and you'll get about five hundred heads and five hundred tails. If you get 983 heads and 17 tails, you can be pretty sure something strange happened to your nickel at the mint. It's possible you could
get that breakdown of heads and tails with an ordinary coin, but those numbers should make you wonder what's going on. (Perhaps the mint was infiltrated by a team of modern architecture enthusiasts, who scraped off the images of Monticello because they believe Frank Lloyd Wright deserves representation on America's currency. Or perhaps not.)

This idea is what researchers mean by "statistical significance." Are the differences observed between two groups important, or are they just the random variation you can expect to see whenever you study real events among real people? If you see a big difference between large groups, the likelihood is higher that it will be statistically significant.

As an example, Patricia Janssen's study of homebirth in British Columbia found more infant deaths among babies born at home. The National Birthday Trust study of 7800 women in the UK found the opposite: there were more infant deaths among babies born to low-risk women in the hospital. In neither study did the differences reach statistical significance -- both authors concluded that we can't tell whether home or hospital is safer for low-risk women and their full-term
babies.

Who says?

Whenever you read research, it's worthwhile to consider who's behind it. Was it published in a peer-reviewed journal? What's in the authors' disclosure statements? Who gains or loses by publication of these results? If you google the authors' names, what affiliations do you find? What do the rapid responses say? If the authors' conclusion is at odds with established points of view, what explanation do they offer? If you google "critique" with the title of the study, do you find any responses to the publication? Henci Goer and Faith Gibson, for instance, have written instructive critiques of the Jenny Pang study. Keep in mind that homebirth is a hot topic and both supporters and opponents tend to be passionate about their viewpoints.

What can we conclude?

A final consideration about homebirth research: no single study can be considered definitive. A given study can tell you about one group of practitioners, one population of women. Things like differing practitioner protocols, varying population density and its effect on transport times, cultural factors like willingness to undergo prenatal testing for congenital anomalies, or ethnic considerations like the increased rate of fatal genetic defects among the Old Order Amish families that account for many of the homebirths in the US -- all of these variables and many more can affect outcomes. Despite the difficulty of conducting good research, epidemiologists agree: planned, attended homebirth is a safe option for low-risk women carrying singleton vertex babies.

Now what?

If you are reading this and considering homebirth, you may be uncertain about where to go from here. Perhaps you are facing family pressure to go to the hospital for your baby's birth; perhaps you and your husband disagree on the wisest choice. Perhaps the idea that homebirth could be as safe as hospital birth seems completely counterintuitive to you.

In addition to the articles linked above, you may find it useful to read birth stories, written both by families and by midwives. What happens when things don't go quite as planned? How do homebirth midwives handle complications? How do women planning homebirths keep themselves as low-risk as possible?

The World Health Organization says that a woman should give birth wherever she feels safe, whether that's at home, in the hospital, or in a freestanding birth center. The American Public Health Association encourages increased access to out-of-hospital birth services, stating, "births to healthy mothers, who are not considered at medical risk after comprehensive screening by trained professionals, can occur safely in various settings, including out-of-hospital birth centers and homes."

You may find it useful, too, to take a look at one of the largest studies of homebirth ever conducted. Peter Schlenzka, a doctoral student in sociology at Stanford, analyzed more than 800,000 California birth certificates according to planned birthplace for his dissertation. "Apparent disadvantages of the obstetric approach have such large order of magnitude," he concluded, "that in any clinical trial it would be considered unethical to continue with the obstetric treatment."

The research on homebirth can tell you a lot about populations of women. It can help you to quantify the risks to you and your baby of birth at home versus birth in the hospital. In the end, though, your plans for your baby's birth aren't about
populations but about individuals -- where do you feel safest? Which set of risks would you rather face? What does your husband have to say? Where does your preferred birth attendant practice? When you imagine delivering in the hospital,
what do you think about? What about when you imagine birth at home?

You will always remember the day your baby was born, and the available research says low-risk women can deliver safely at home or in the hospital. The choice is yours to make.

Comment on this article.