February 22, 2018

On due dates (Or: why confidence intervals are often better than means)

A very pregnant woman (not my wife). Image by Nina Matthews.

A very pregnant woman (not my wife). Image by Nina Matthews.

As regular readers will know, my wife is pregnant.  In fact, today is the due date.  As of this writing (2:30 EST), labour has yet to commence.

Throughout the pregnancy, my wife and I have constantly reminded ourselves that the due date is a gross over-simplification.  It’s pretty obvious that it’s a ball-park estimate at best.  And yet, providing a precise date creates some weird psychological tension, even when you know that it shouldn’t be taken too seriously.

For example:

  • If today is the “due date”, does that mean that tomorrow the baby is “over due”?
  • Can the baby spoil, like over due milk?
  • Does it mean there is something wrong with the baby?
  • Seriously: when is this baby going to come out??!?!
  • Etc.

These are all questions that are naturally raised by pronouncing a specific due date, and by repeating it at every appointment over a 9 month period (and by being asked about the due date every time you pass someone in the street).

due dates

Image via spacefem.

This is especially annoying given than only ~10% of women give birth on their actual due date (which is actually quite impressive, given that it’s calculated 6+ months in advance). For a visual representation of what this looks like, check out the above figure compiled from a survey by spacefem (the original figure is interactive, so it’s worth clicking through). While births are most likely right around the due date, that is in no way guaranteed (another very cool interactive figure at spacefem shows that roughly 60% of women give birth within a week of their due date, yet only ~10% give birth on the due date itself)   So everyone knows that the due date is a radical oversimplification, and yet we stick with it.

There is another way

Consider if instead of providing a specific date, a pregnant woman was provided instead with a range of dates.  So rather than giving us a due date of August 19, the doctor could have instead said that there is a 60% chance that the baby will be born between August 12 and 26th.  Or that there is a 95% chance the baby will be born between July 19 and September 19.  You could easily create a figure like the one above, with the actual dates plugged in.

The nerds among us will recognize that I’m suggesting that instead of providing women with a due date in the form of a mean (e.g. one specific date), I’m suggesting they be provided with a 60% or 95% confidence interval.  I personally think it would be much more useful (and more accurate) information, while being arguably more intuitive than the due date, which always seems a bit random and arbitrary to  me.

So if there are any healthcare providers out there, I strongly urge you to try presenting the due date in a different way.  You could even do some A/B testing to see which format women prefer.

And seriously, when is this baby going to come out!??!?


(c) Obesity Panacea – Read entire story here.