Calculating neonatal mortality using the CDC Wonder database

I’ve explained many times how I obtained the CDC mortality statistics for homebirth and hospital birth. People still have questions, so I made a video to demonstrate how to use the database.

  • The Computer Ate My Nym

    Every time you do something on the CDC wonder data, I have to go play with it again. Today I did a search where I tried for the absolute lowest risk group that I could find. So, white women, age 20-44, married, with at least a high school education, delivery by vaginal delivery, at 39+ weeks, with birth weight of at least 2.5 kg, death from any cause other than congenital malformations. And the rate of death for babies delivered in the hospital by MDs is…14.7 per 100,000. Quite similar to the rate for maternal mortality, actually. The numbers for other midwife out of hospital weren’t high enough to trigger the calculation, but my calculation made it about 75 per 100,000. Interestingly, the rate of death for CNMs in hospital was 15.8 per 100,000, slightly higher than the MDs. Maybe a coincidence, maybe an issue not yet identified?

    • The Computer Ate My Nym

      Oh, the above had one more restriction that I forgot to mention: non-Hispanic white women only. When I took off the ethnic restriction, the rate for MD delivery in the hospital actually became lower at 14.3 per 100,000 and the mortality for other midwife out of hospital became calculable at 83.7 per 100,000. I should write these up and send them to an OB conference somewhere.

      • The Bofa on the Sofa

        Can you refine the educational level at all?

        • The Computer Ate My Nym

          Yes. Actually, I just played with the education, since the MANA data included mostly women who had at least some college eduction. If you restrict to at least some college, with or without degree, the mortality for in hospital with MD becomes 13.1 per 100,000. There were “only” 15 total deaths for other midwife, not in hospital, but that comes to a ratio of 76.3 per 100,000. (This is excluding congential anomalies but including all other causes of death and including only vaginal deliveries. Also I should mention that I’m using the years 2007-10.)

          • Young CC Prof

            Let’s do a quick hypothesis test on that. Essentially I’m calculating the probability that we would see 15 deaths if the rate really were 13.1 per 100,000. What was the denominator, and are these total neonatal or early neonatal?

          • Young CC Prof

            Just taking 20,000 as the denominator, using a 2-sided test to be generous, and using the binomial probability formula as np is too small for a normal approximation, the p-value is 4 times 10 to the negative 8, that is:

            4 in 1 million chance of coincidence.

            Yup. That’s absurdly statistically significant.

          • thepragmatist

            In context of the body of literature, too, it’s very familiar, no?

          • The Bofa on the Sofa

            76/100K is 15/20K. In 20000 births, a 13/100K rate would indicate 3 deaths or so. So what is the probability of getting 15?

            I just ran a quick simulation on Excel. Using a base rate of 13/100K, I calculated how many deaths would occur in a 20K sample. I ran 10000 20K samples.

            The average was (as expected) right around 2.6 deaths/20K. The most likely outcome was,in fact, 2, which occurred 24.6% of the time.

            The most that ever happened was 10, and that happened 3 times in 10K samples. Thus, the probability that there could be even 10 in this sample is in the 1 in 10000 range.

    • PJ

      Wouldn’t it just be that they are taking on lower risk patients? As in, white married women aged between 20-44 are not immune from complications other than growth restriction and congenital malformations. They just presumably happen at a lower rate.

      • PJ

        I should have said “not immune from complications that are forseeable or that can be detected before labour.” If that makes it any clearer.

      • The Computer Ate My Nym

        But the CNMs in hospital had a slightly higher rate of neonatal mortality than MDs in hospital, despite, presumably, taking lower risk cases. The numbers are pretty similar so it could be a coincidence or related to factors other than CNM competence in general. And you’re right about these numbers not perfectly balancing risk, of course. It’s just the best I can figure out to do.

        • Young CC Prof

          We’re talking about less than 100 total deaths in each category, right? The difference there isn’t statistically significant, that is, it could easily be coincidence.

          If you extended the years further back, you might be able to see more clearly if there is a difference between MD and CNM. Or not.

        • PJ

          Oops, sorry; obviously read too fast.

      • Medwife

        Over 40 is far from low risk. I still provide care to over 40’s, but with a very low threshold for warning signs of HTN, pre-e, I induce at 39 weeks, they get biweekly NST’s starting at 36 weeks, etc. basically they are “comanagement” in terms of midwifery/OB care.

    • auntbea

      Shouldn’t you also exclude those over 35 and macrosomic babies? Also, exclude those without prenatal care?

      • The Computer Ate My Nym

        Oops. I did exclude those with no prenatal care. I didn’t exclude macrosomic because home birth midwives don’t do GTT and so will have macrosomic infants in their sample. I’ll have to see if excluding over 35 makes the numbers too low…

        • The Computer Ate My Nym

          Ok, so including only ages 20-34 and excluding macrosomic infants (>5000 grams) you get 12.2 per 100,000 for MD in hospital and 14 out of 15,249 for other midwife, out of hospital (91.8/100,000 but small numbers so unstable.)

  • Trixie

    http://www.scienceandsensibility.org/?p=7869
    Wendy Gordon is back on the Science and Sensibility thread still claiming that there is doubt about intrapartum death rates and that Judith Rooks is wrong.
    Great job, Doula Dani for not letting up on this one.

    • DaisyGrrl

      She’s taking no prisoners. It’s great! And Wendy Gordon is *still* going on about the “missing” term intrapartum death rate for hospitals. I guess she didn’t like Judith Rooks’ numbers and is desperately looking for something more favorable?

    • Medwife

      I’m so embarrassed that this farce of a paper was published by JMWH. Official journal of the ACNM. Ugh

  • Lisa from NY

    The news can’t compete with the movie, “The Business of Being Born.”

    Just look at the comments on this article:

    http://www.voanews.com/content/home-birth-movement-us-challenges-norms/1835009.html

    “In my humble opinion, for healthy, uncomplicated pregnancies, there is no safer place to give birth than home. Hospitals are staffed with medical doctors, nurses, etc. to care for people who are sick. The course of labor in a healthy woman requires an environment in which the mother can safely surrender to her birthing body. “

    • Anj Fabian

      I know of multiple cases where the mother’s body surrendered to death while birthing at home. These things tend to make the headlines.

    • Jessica S.

      “The course of labor in a healthy woman requires an environment in which the mother can safely surrender to her birthing body.”

      To which I would say, “Yes, and what safer place to “surrender” my “birthing body” than a well-equipped facility, staffed with competent experts who can save my baby and I should something go wrong?” It goes back to that naturalistic fallacy that birthing is natural and natural = safe. No way. Nature is a cruel and often arbitrary bitch.

    • Susan

      Surrender to your birthing body… screams cult.

    • auntbea

      Surrender to your body! Let your skin naturally become dry and itchy, just like it wants to! Let your vision be lopsided and slightly blurry, as your eyeballs intended! Let your hair and nails grow until they break off! Eat whatever your body tells you it wants (mine wants Dr. Pepper!)

    • The Computer Ate My Nym

      I don’t want to surrender my body, birthing or otherwise. I want control over what happens to it. I recognize that I don’t always get that, but why should I want to “surrender” it?

    • Sally RNC-NIC

      “….a peaceful, unmedicated birth.”

      Ummmmmm…….what? Good luck!

    • Sally RNC-NIC

      Standing ovation, with a slow-clap…

      “We here on our labor and delivery unit fight for seconds when an unexpected fetal distress occurs; we do drills, so we plan an emergency Caesarean and fight for seconds,” he said. “Because literally, seconds — if someone is as much as one block away from this hospital, it’s too far,” he said.

  • Ash

    Great post, the Youtube video does a very nice job of explaining things! I am getting ready to write to my state representatives and hopefully draw attention to the lack of oversight of CPMs in my state. The committee that advises licensing in my state must consist of 2 CPMs, a CNM who attends out of hospital births,an OB/GYN, and a member of the public who has had midwife care outside of a hospital. Hm. I doubt the member of the public is someone who has had a baby die or severe injury during homebirth.

    • manabanana

      Which state is this?

      • Ash

        Wisconsin

  • DaisyGrrl

    Thank you! I was having trouble following the CDC number crunching discussions. This helps enormously.

  • Squillo

    Thanks, Amy.

    Two questions:

    1. Why would one not eliminate the “no prenatal care” group from the query? Women don’t show up at planned homebirths never having seen a midwife, while it happens several thousand times a year in the hospital group.

    2. Why include the “other” and “unknown or unstated” group in the attendant group? It would seem that eliminating those two categories would largely eliminate the unplanned homebirth stats.

    I would also argue that we should eliminate deaths from “external causes” and “special causes”, as neither has any direct bearing on obstetrics or the safety of a particular birthplace.

    • Amy Tuteur, MD

      You could do those things to further refine the data and that would make MANA’s numbers look even more hideous.

    • Lisa from NY

      Good point. Women with untreated diabetes during pregnancy are more likely to end up with dead babies.

      • Karen in SC

        Especially before the glucose tolerance tests became standard. One of them is my friend who lost her first born son about 35 years ago. Still was not properly treated for GD and her second was born almost twelve pounds with SD and a brachial plexus injury.

        Two things to note about my friend – after taking insulin during her third pregnancy, she realized that her previous doctor screwed up (in her case she had told him of previous stillbirth, mom & sister had GD) and sued him. And won money in trust for her child. And with proper care, children # 3,4,5,and 6 weighed less and born healthy.

  • EllenL

    Fascinating. Wonderful. Thank you.

  • Guest RN

    That you for this – the “missing” intrapartum numbers was one of MANA’s arguments for why they had higher death rates.

    • Amy Tuteur, MD

      They are correct that the intrapartum numbers are missing and if you added them in to the CDC numbers, the homebirth group would look even worse since there are basically no intrapartum stillbirths in term infants in the hospital.

      The intrapartum stillbirths are even worse for MANA in another important way. If we consider bad outcomes from childbirth, we can grade them from bad to worse as follows:

      Bad: Baby is born, is resusictated and dies within the first month.
      Worse: Baby is born and can’t be resuscitated.
      Worst: Baby is born dead.

      So, while MANA has been trying to convince people that their neonatal mortality statistics aren’t that bad, it’s only because the midwives were so grossly incompetent that they didn’t know that the baby had died in utero.

      • Jessica S.

        Thank you for addressing that clearly, it really is despicable.