Due date statistics: A study on the length of pregnancy

Do babies come on their due dates? This was the question on my mind during the later stages on my pregnancy in the spring of 2010. 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. I was unable to find great data that satisfied me. So, armed, with a website, I started running a simple survey about due dates. After about 10,000 participants I decided that was enough, and I put together this site to show interesting things I learned from the results. I am not a mathematician, so there's no in-depth statistical analysis. I'm an engineer, web programmer, and (as of 2010) mom. When I started the survey I was just a very very pregnant woman wondering when my baby was going to show up. The survey ran from 2010 until 2021 when I decided to close up shop at my hosted website and archive everything at this blog. I hope it helps you. Here's the very basic plot of who took the survey:

Other data

I was unable to find daily breakdowns of when births occur, but there have been other studies about due date accuracy. I think the data is hard to find because the medical community has accepted the fact that it's impossible to exactly predict when a baby will show up, so they're okay with just knowing about when it happens, and us pregnant women are also urged to be okay with just knowing about when it will happen. But if you're a geek like me who loves statistics, that's tough.

All these studies describe the length of pregnancy in days. For reference, 280 days = 40 weeks = 9 months 6 days.

The length of human pregnancy as calculated by ultrasonographic measurement of the fetal biparietal diameter (H. Kieler, O. Axelsson, S. Nilsson, U. Waldenströ) has a very nice table of 10 published studies, which found various means between 272-283 days. The authors contribute their own results as well, recording over 800 mothers who went into labor spontaneously. When pregnancy length was calculated using ultrasound in the second trimester the mean was 280.6 days, standard deviation 9.7 days. When it was calculated using LMP, the mean was 283.6 days with a standard deviation of 10.5 days.

A study from the Swedish birth registry involving 427,582 singleton pregnancies the mean, median and modal durations of pregnancies were 281, 282 and 283 days respectively. The standard deviation was 13 days. This study did not consider whether labor was medically induced or not. (Duration of human singleton pregnancy—a population-based study, Bergsjφ P, Denman DW, Hoffman HJ, Meirik O.)

A 1999 study of over 17,000 spontaneous deliveries stated that due dates are more accurate if we add 282 days to LMP instead of 280 (Evaluation of ultrasound-estimated date of delivery in 17,450 spontaneous singleton births: do we need to modify Naegele's rule? Nguyen TH, Larsen T, Engholm G, Møller H.)

A 2003 London-based study of over 122,000 pregnancies with spontaneous onset of labor (except they call it "labour" because, well, you know) found that average gestational age at delivery was 40 weeks for white Europeans but only 39 weeks for members of the black and asian ethnic groups. I chose not to ask for ethnicity or race in my survey because it's so hard to define, so many people are a combination of races, but the study is interesting. Another note: it was made up entirely of first time ("nulliparous") mothers. (International Journal of Epidemiology 2003;33:107–113 Roshni R Patel,Philip Steer,Pat Doyle,Mark P Littleand Paul Elliott)

And finally there's this 1990 Harvard study called The length of uncomplicated human gestation (Mittendorf et al). The study looked at 31 first-time moms at one private practice to conclude that their average due date was eight days early. That's not a very big sample size, which I think is why it's the only study that found 288 days as the median. I really wish people would stop bringing it up, but it seems to be popular so I'll just say if you want to read it and come to your own conclusions, it's here.

Come on people, if 288 days was really the median that means that more than half of all pregnancies would end past 41 weeks... seems like we would have noticed something like that by now.

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.

Does your survey consider age, race, cycle length, shoe sizes, ferret ownership, etc?

It became apparent to me by reading studies that no matter what factor people added into consideration, it was still impossible to predict the start of labor. Even IVF patients, who know the HOUR their baby was created, don't get accurate due dates. So I see no point in taking all these factors into consideration, if it's going to be a range of dates lets just all throw whatever randomness we have into the pot and see what the range is. Several studies and my survey have found that there's a standard deviation of over a week... this makes the bell curve pretty flat on top. Moving a due date a day up or back based on ovulation really makes no difference because all the dates around 40W are almost equal in terms of your probability of going into labor.

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.

That said, even if it's not scientific the fact that the results are matching real scientific studies mean that I must be doing something right. The studies mentioned a few paragraphs above here seem to support the fact that the 280 day mean, 9-10 day standard deviation is the norm... well that's what my results show too. I think if the big numbers about the curves are correct than the little details are likely to be correct also.

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.

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.

Who cares what's normal? Shouldn't we just tell ourselves to accept whatever happens or be patient?

Yes and no.

On one hand, I think it's a good idea to spread the word that pregnancy does indeed usually last 40 weeks, and if a woman is 37 weeks pregnant it's not only scientifically inaccurate to tell her she's "about to pop", it's also really annoying.

On the other hand, I've seen this survey discussed with comments like, "Well obviously a bunch of women had babies past 42 weeks, this shows that we should just be happy being pregnant forever, fight those inductions, trust nature." That's a risky assessment. There is significant evidence that health risks to babies increase as pregnancy approaches 42 weeks (Fetal and perinatal mortality, United States, 2005 by MacDorman et al. is one from the CDC). I think nature is sort of a good thing to trust. Yes, it got the human race this far. Yes, I had a natural birth and recommend it. But for thousands of years 1 in 10 births didn't end well... nature was happy with the 90% batting average. But we can do better. So from that standpoint I'd prefer that you avoid using this study to tell your doctor that you intend to be 43 weeks pregnant. Deal?

What results are included in the charts?

I decided to only include births between & including days 245 (35W,0D) and 301 (43W,0D). This means that out of the 16413 records:

• 599 results before day 245 are exclude
• 52 results after day 301 are excluded

Twin births make things slightly out of the norm, so they are not included in the chart. This excludes 124 results. I didn't really ask good survey questions for twins anyway. If you are pregnant with twins and looking for data, there's a good survey going over here.

In addition, around 200 results I started noticing a big spike in the number of babies born on their due date. I believe this is because women who have their babies on their due dates are more likely to remember it, and be enthusiastic about taking a survey. To account for this I've capped the number of results shown at exactly 40 W, and 359 results are excluded.