Understanding statistics

Those who want homebirth to be, if not outlawed, at least so marginalised and dirty that noone with any sense would go near it, are able to use statistics to coerce their listeners into compliance, and to assure authorities that they are acting 'in the public interest'. There have been a couple of outstanding examples of this phenomenon in the past year.

Firstly, remember the Australian Medical Journal's publication of Planned home and hospital births in South Australia, 1991-2006: differences in outcomes (Kennare et al 2010), using shameless distortion of facts gathered in the research. I wrote about it and set up links to the paper at the MiPP blog in January this year. Alarm bells sounded, and media picked up the story from the abstract: X7 higher risk of intrapartum death and X27-fold higher risk of death from intrapartum asphyxia in the planned home births group.


The second doozie [for readers who are unfamiliar with this word, it's Australian slang - not sure what it really means, but it seems to fit here] is the Wax et al 2010 paper on maternal and newborn outcomes for homebirths in North America.

Various reliable midwifery organisations have critiqued this paper for its methodology and conclusions. The Medscape "Attention-Grabbing No Doubt, But Uninformative" comment by Andrew Vickers, copied in full (below) is worth reading.

Midwives can also understand statistics.

"Home Birth Triples the Neonatal Death Rate": Attention-Grabbing No Doubt, But Uninformative
Andrew J. Vickers, PhD

Posted: 09/27/2010

Home birth, according to a position statement from the American College of Obstetricians and Gynecologists, is "trendy" and "fashionable." Moreover, women who choose to deliver a baby at home "place the process of giving birth over the goal of having a healthy baby."[1] Interesting thoughts, I guess, but hardly evidence-based. Has anyone actually interviewed home-birthing parents to determine, for example, that they rate having a healthy baby at, say, 5 out of 10, whereas being allowed to listen to druid chanting during the second stage of labor is rated an 8? And with respect to being fashionable, have researchers really evaluated the wardrobes of home-birthers compared with those choosing to labor in the hospital, finding in the former a higher proportion of Marc Jacobs and Manolo Blahnik?

So it is nice to finally see some data that quantify the relative benefits and harms of home birth. Joseph R. Wax and colleagues conducted a meta-analysis, combining data from 12 studies including more than a half million deliveries, in order to report on a wide variety of outcomes, including process (eg, use of epidural), maternal morbidity (eg, vaginal laceration), neonatal morbidity (eg, prematurity), and mortality (of both mother and child).[2] What isn't so nice is the spin. The study authors themselves, who are from a department of obstetrics, report a highly alarming statistic -- that home birth is associated with a triple the risk for neonatal death. The American College of Nurse-Midwives, predictably enough, finds fault with the methodology of the study and cautions against overinterpretation of the findings.[3]

I am sympathetic toward the critiques. A meta-analysis is only as good as the studies that are entered, and it is somewhat disconcerting to see a mixture of prospective and retrospective observational studies all mixed in with a single randomized trial. (On which point, it is even more disconcerting to find that the paper referenced for the randomized trial was a discussion piece, not a trial report.) But for the sake of argument, let's assume that the paper is perfect and accurately represents the true outcomes of home and hospital delivery.

First off, how should we interpret a "tripling of death rates"? This is what statisticians call a relative risk, and it is widely known to be problematic for decision-making. As a simple example,[4] would you buy a pair of slippers if I told you that they were 90% off? Well, no, you would want to know how much they cost. It is the same with risk; it is the absolute amount that matters. The classic example is the contraceptive pill and breast cancer. One estimate is that the pill raises the risk for early breast cancer by 50%. This sounds pretty scary until you realize that most women's risk is so low that this translates to about 1 woman with breast cancer for every 10,000 on the pill. Most women would feel that is a risk worth taking, given the benefits of the pill and the possible harms of the alternative: pregnancy, which after all, has dangers of its own.

In place of a "tripling in death rate," the more informative statistic is the absolute increase in neonatal death associated with home birth. On the basis of the results tables, it is possible to calculate that this turns out to be 1 neonatal death per 1000 women who choose home birth. However, the results tables show that those women would also experience some benefits, including 40 fewer premature labors, 45 fewer cesarean sections, 140 fewer vaginal lacerations, and 140 fewer epidurals. This type of cost-benefit analysis -- trading off neonatal mortality against maternal morbidity -- can seem sort of cold-blooded. But if the only thing we cared about was a healthy baby, then we'd do cesareans on all pregnant women at 38 weeks (as well as insist that all women conceive once they turned 21). We implicitly trade off risks and benefits anytime we consider a medical procedure. Let's do it explicitly rather than implicitly, on the basis of decision-analytic statistics such as absolute risk, rather than headline-grabbing statistics such as a "tripling of the death rate."

References
1. American College of Obstetrics and Gynecologists. ACOG statement on home births. Medscape OB/GYN and Women's Health. 2010.Available at: http://ww.medscape.com/viewarticle/725383 Accessed September 9, 2010.
2. Wax JR, Lucas FL, Lamont M, Pinette MG, Carlin A, Blackstone J. Maternal and newborn outcomes in planned home birth vs. planned hospital births: a metaanalysis. Am J Obstet Gynecol. 2010;203:e1-e8.
3. American College of Nurse Midwives. The American College of Nurse Midwives expresses concerns with recent ACOG statement on home births. Medscape OB/GYN and Women's Health. 2010. Available at: http://ww.medscape.com/viewarticle/725382 Accessed September 9, 2010.
4. Vickers AJ. Top scientific papers vs. furry green slippers: which should you trust? Medscape Business of Medicine, 2010. Available at: http://ww.medscape.com/viewarticle/722723 Accessed September 9, 2010.