Spacefem (spacefem) wrote,

Pregnancy, Due Dates, and the length of Human Gestation: My Statistics

Several of you may remember that towards the end of my pregnancy, I got really interested in statistics of due date accuracy and human gestation. I researched, made tables, got mad at data... and finally launched my own online survey to ask women who'd had babies when they had them. And then JUST when the data was getting interesting I went and had a baby and couldn't really look at the results. But the survey didn't die, and I've been meaning to post a sort of epilogue about what I found.

The short story: that "40 weeks" idea is a pretty damn good guess at the average. I mean seriously. When I looked at just the 141 spontaneous deliveries after 37 weeks ("term"), both the mean and median date fell right smack on 280 days, or 40 weeks 0 days. This means a woman is equally likely to have her baby before the date than after it. And the most common days to have a baby are all clustered around that date, which lends to the theory that we can apply a normal curve to this (I love normal curves).

But I'd like to explore all this a lot more, not just my statistics but my reasons for the survey and answers to frequent questions/criticisms. Let's go for a ride, shall we?

Why did you do the survey?
When I was pregnant I heard a lot of conflicting information about what a "due date" meant and when babies were born. Some people saw the "due date" like a deadline... if the baby hasn't arrived yet, it's late, it's time to freak out. Some people said due dates were a bad example of western intervention, used by doctors as an excuse to cut us all open. Some people said that first time moms should expect to go late, it's normal and you should chill out. I found some scattered statistics about what week babies came... and a lot of those graphs just said "X% were born at 40 weeks" but didn't explain what that meant (during the 40th week of pregnancy? 40 weeks accomplished? On day 280?). And trust me if you're pregnant, you start counting DAYS when that time comes around. There is a huge difference between 40 weeks 0 days and 40 weeks 5 days.

On top of this mess, there's this 1990 study people keep bringing up called The length of uncomplicated human gestation. No one has access to the whole study, no one knows the sample size of their study or what data they threw out. It says that half of all first time moms (primiparas) go into labor eight days past the due dates we give them... this has been labeled "Mittendorf's Rule" after the first study author. But it makes no sense to me, because if 288 days was really the median that means that more than half of all pregnancies would end past 41 weeks... don't you think we would have noticed something like that by now? It doesn't gel.

There is good weekly data from the CDC that talks about birthweight and weight gain and inductions and all kinds of good stuff. But it's not the daily breakdown I wanted, and it doesn't break down weekly stats by whether the births were spontaneous or induced. So to make this story short, if you find good daily data send it to me. I couldn't find it.

You realize that an internet survey isn't exactly scientific, right? It's going to be so biased!
Science wasn't really my goal. By that I mean I'm not terribly interested in how the human body works... I'm interested in how life is for women like me. The women who are reading this and visiting my website are mostly internet users from english-speaking countries... well, so are my survey participants. I see that as a good thing.

Where does "40 weeks" come from?
The most common way to calculate a due date is to set it at 40 weeks (280 days) past the woman's last menstrual period. This is known as Naegele's Rule, after a German doctor who published the method in 1806. Naegele did not arrive at this method very scientifically, but I personally believe that we would not be using it if it was totally off.

What'd you find?
At the time I'm writing this there are 211 responses. 152 (72%) of participants went into labor spontaneously (were not induced).

Considering all spontaneous labors, the average baby was born at 39 weeks, 5 days and the median was 40 weeks, 0 days (meaning half the babies were born before that date, half after it). When I threw out births before 37 weeks (pre-term), the mean changed to 40 weeks, 0 days. For reference, the standard deviation was about 9 days.

Here are the non-induced deliveries by date:
Probability of spontaneous delivery on a given day

So babies come on their due dates?
Babies come around their due dates. Only 12 out of 211 particpants (5.6%) spontaneously arrived on the exact day. But if you look around that date, the numbers get bigger fast... 55 babies (25%) arrived spontaneously between 39 weeks 4 days and 40 weeks 3 days. 87 (40%) arrived spontaneously within a week of their due date. Throwing out inductions makes those percentages even higher: 61% of all non-induced babies arrived within a week of their due date.

Probability of spontaneous delivery within x days of a given date

Did you consider that augmenting labor can change the birth date?
No. Because I don't care that much. Most augmentations can only change things by one day, if that. I also feel like I would have had to consider the whole range of things... there's the doctor that starts Pitocin when a woman is at 3cm and there's the doctor who's completely hands off until reaching for the foreceps at the end. Both of them decreased labor time and might have change the day a baby was born, but to very different degrees. And anyway, my goal was to tell women who know nothing when they might have a baby, since most of these dates are totally a surprise. When you're very pregnant, there's no way to tell when labor will begin and you go looking for internet charts. When you're in active labor, there aren't as many questions in your mind.

Does birthweight increase as the weeks go along?
It tends to, yes.
Birthweight vs. duration of pregnancy

The CDC has good data on this but I included the weight question in my survey just for funzies, and the trendline on that graph shows an average weight gain of about 18 grams per day. That's 126 grams per week, or about 4 ounces. But really birthweights were all over the place. That chart shows babies who were perfectly average despite being two weeks overdue, and 10-pounders who were a week early.

The western world likes to induce babies rather than let them go too late, doesn't this skew your results by removing late babies from the "late" end of the spectrum where they should be?
Not as much as you'd think. Only 15% (9) of the inductions were after 41 weeks. By that time, 95% of all spontenous labors have already happened. At 41 weeks 0 days, 132/153 (86%) of spontaneous labors had already happened. But only nine inductions happened after that date. Maybe all those women might have safely gone into labor the day after they were induced, but that would still mean that 81% of labors would be before 41 weeks. Yes, it is hard to determine what the stats really look like if we reduce inductions and let everyone go, but again the purpose of my survey is to help women figure out when they *might* go into labor, and once you set an induction date your story, and curiousity, changes.

This graph shows your odds of eventually being induced if you're pregnant on a certain date. By the time those odds start increasing dramatically, 90% of babies are already born. This tells me that letting those women go on forever would not change the spontaneous distribution much.
Probability of induction after a given day

So what does it mean to "go late"?
A due date isn't some magic number, it's just the top of the curve. So on day 281 of your pregnancy, you're still riding the top of that curve... you just happen to be on the other side of it. The most popular days to have a baby on my survey:
40 weeks 0 days - 9% of spontaneous births after 37 weeks
40 weeks 5 days - 8%
39 weeks 4 days - 7%

Notice that the second most popular date is an "overdue" one? That's probably because 50% of my participants were still pregnant on their due dates. But take heart: of all those late women, 67% had their babies by the end of the week (by two weeks, it was 90%)! So don't think of yourself as overdue. You're just... due.

Not all due dates are based on LMP.
I assumed that due dates were at 280 days LMP no matter how they were determined. Oddly enough though, when due date was determined by LMP labor tended to start closer to the due date. The average LMP-determined due date was one day off, the average ultrasound one was six days off. The average ovulation-determined date was 2 days off. So if you want a good due date ladies, track those periods.

How do dates change with subsequent pregnancies?
Second time moms had their babies two days earlier than first time moms, on average. My data's not really complete there though, only 21% of my participants were second time moms. I only had four participants who had two or more previous births and I feel like that's really not enough to make any assumptions. But really people, two days difference is nothing to write home about, I think we can safely say that you're not destined to go a week overdue just because you've never had a baby before.

So, did you meet your goals in doing this survey?
I'm happy with it. Scientific or not, it was a good way to channel my nervous energy during my last week of pregnancy. It made me feel very good about the idea of being overdue... people would bug me with the endless "had that baby yet?" and I'd respond to say that I felt more than ready to go late, I'd totally be at work the week after my due date, that sort of thing.

Of course, I was wrong, I didn't go late. In the midst of my big project I had a baby at 39 weeks, 5 days. But I knew I was at the top of that curve, so it's not like I could really be surprised.

What are your goals for the project now?
The survey continues! You're welcome to tell your friends or take it yourself if you've had a baby. I'd really like to get at least 1000 responses, then re-post some analysis like this.

So check it out here:

Also if you'd like to do your own analysis, here's the data in a constantly updated format:
Table of raw results:
Graphs and calculations:

And finally, if you are pregnant, my pregnancy calendar presents you with a table of due date probabilities that you might just find handy:
Tags: math, pregnancy
  • Post a new comment


    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded