On birth control failure rates (they’re all asymptotic at 100%)…

April 18, 2010

So for reasons that are nobody’s business* I’ve been thinking about birth control success/failure rates over the last couple days.  Specifically, I’ve been thinking about how they’re usually expressed–in a percentage (e.g., 95% effective).  Sometimes it’s a little clearer; for example, Planned Parenthood reports their figures as number out of 100 women per year (e.g., 95 out of 100 women per year).  What does that even mean? I think a lot of people have the idea that “95% effective” means that they only have a 5% chance of getting pregnant, ever, using it.  This isn’t the case at all.   It means that if you use that method for a year, you have a 5% chance of getting pregnant that year.  And then you have a 5% chance of getting pregnant the following year if you continue using it.  And so on.

So what if you choose a method and stick with it for a number of years?  5%, which is the annual failure rate for birth control pills, may sound really acceptable, but what about 40%?  That’s the chances that you’ll have a kid after using pills for 10 years.

This graph and table are visualizations of the same info; they show your chances of having a kid after x years of using a particular birth-control method.

Birth control failure rates over time

Chances (% probability) you will have at least one kid after x years of using birth control, by method. Click to embiggen.

Chances (% probability) you will have at least one kid after x years of using birth control, by method


x = 1

x = 5

x = 10

x = 15

No method










Diaphragms, female condoms





Male condom, fertility awareness





Hormonal pills





IUD, patch, ring, sterilization





Depo Provera; multiple methods





(Data from multiple sources; actual rather than “ideal” use reported, because everyone thinks they use perfectly and no one does.)

Of course, I feel compelled to note that your chances of getting pregnant in the next year on any particular method never change.  It’s not like if you have using condoms for 14 years without pregnancy, you suddenly have a near certainty of getting pregnant.  It’s just like Vegas–what has happened up until now doesn’t matter.

What this does mean is that if you’re thinking about what sort of birth control to use in the long term, you should decide what % risk is acceptable to you (if the answer is zero, I hope you’re celibate or pro-choice), and find a method that meets that criterion over the time period that you will be using it.

* No, it’s not what you probably think.


7 Responses to “On birth control failure rates (they’re all asymptotic at 100%)…”

  1. Do they account for how many times you do it in a year? I mean, if you have sex 365 times a year, you’re more likely to have a failure than someone who has sex once a year. Shouldn’t the percentage apply to 95% of the times you do it and not how many women per year? ‘Cuz then you’d just have to avoid that particular 5% of women each year, right?

    I don’t get these percentages.

    • amberwb Says:

      I was thinking about that when I was playing with this data… but nope, they don’t account for how often you actually have sex in failure rates–I think because of how hard that would be to accurately measure. But that would definitely make a difference.

      My graph and table are assuming continuous use (i.e., long-term partner or, I guess, always having at least one partner). It also assumes a lot of other things that aren’t true, like fertility is stable over the span of years (it declines over time… more quickly for women than men), and independence of risk between years (probably not true for things like sterilization and IUD, which tend to fail early if they fail at all).

    • amberwb Says:

      Also, while it’s not that you could avoid the 5% of women who will get pregnant, you COULD avoid your woman 5% of the time! (Google “fertility awareness method.”)

  2. Desiree Says:

    Hmm. In general I’m with you. I have trouble, though, believing that 40% of women who use the birth control pill consistently for 10 years will get pregnant. My guess would be that the 5% per year estimate there is wrong, and that the wrongness is more noticeable in the long term estimate.

    • amberwb Says:

      Actually, I think it has more to do with the fact that this analysis makes incorrect assumptions, most specifically the ecological fallacy. Hopefully my next post will be about that.

  3. mfwesq Says:

    I recall reading as well that these statistics include failures due to human error–forgetting to take the pill, skipping the condom or using it incorrectly–not merely unavoidable accidents.

    • amberwb Says:

      Sure, that’s what I mean by “actual use” rates. But those are (theoretically) equally likely to occur every year.

      This is actually exactly what I mean by “everyone thinks they use perfectly, and no one does.” It’s really tempting to ignore the actual use rate and go for the perfect use rate, thinking “Oh that doesn’t apply to me; I will never forget to take a pill.” And the actual use is, of course, on average–there may be some people who truly NEVER forget to take a pill (I really, really doubt it), and some people who forget at least once per month.

      Anyway, nothing in my post says anything about unavoidable accidents. But it doesn’t matter whether it’s avoidable–only whether it happens.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: