Please visit the new home of Majikthise at

« Abu Ghraib photos show rape, U.S. general says (updated) | Main | Song from the Uproar »

May 28, 2009

Withdrawal study: Ur doin' it wrong

The conventional wisdom is that withdrawal is to contraception what bulimia is to weight-loss. Both methods have a certain mechanical plausibility, but no responsible physician would recommend either one.

Sex educators tend to regard withdrawal less as a form of contraception and more as an excuse not to use birth control. A researcher from the Guttmacher Institute and her co-authors are urging sex educators to reconsider, leading off a recent paper in the journal Contraception with the following bold hypothesis: "[Withdrawal] might more aptly be referred to as a method that is almost as effective as the male condom." (.pdf)

In the large print, the paper says the typical use failure rate for withdrawal is 18%. That sounds pretty good juxtaposed with the 17% typical use failure they cite for condoms. A footnote adds some additional context: "Notably, the typical-use failure rate for withdrawal is more variable, ranging from 14%-24%, compared to a confidence interval of 15%-21% for condoms." 

According to the Guttmacher Institute's website, withdrawal's perfect use and typical use failure rates are 4% and 17.4%, respectively. The typical use figure on the website comes from the same study that the authors of the Contraception paper cited. (To put these numbers in perspective, 85% of couples who don't use any contraception will get pregnant within a year.)

The authors have gotten a lot of play out of their claim that, when used perfectly, withdrawal has an estimated annual failure rate of 4%, compared to 2% for perfect condom use. Where did that impressive sounding 4% perfect use figure come from?

A footnote in Guttmacher's online table explains: "Most perfect-use rates have been clinically evaluated, but some are based on clinical expertise or “best guesses” (such as some forms of periodic abstinence, withdrawal and no method use)." So, that dramatic 4% number comes down to the best guess of someone at the Guttmacher Institute? If so, the authors should have made that clear in the journal article.

The authors define perfect use as the man pulling out before ejaculation every single time, a definition that risks begging the question. If you only count the couples where the man always manages to pull out in time, every single time for a year, withdrawal might look pretty good. The "oops factor" is, after all, the main reasons sex educators discourage the practice. If you're going to define perfect use as perfect execution, you also have to compare the percentage of people who try to use withdrawal and fail vs. those who manage to screw up other methods.

The authors also admit that withdrawal is basically scientifically untested as a form of contraception. What little research exists comes from interviews and surveys, often from small studies of very specific populations, like Asian Canadians seeking abortions and Turkish factory workers. There has apparently never been prospective clinical trial of withdrawal.

The paper ends with a recommendation that, "[h]ealth care providers and health educators should discuss withdrawal as a legitimate, if slightly less effective, contraceptive method in the same way they do condoms and diaphragms." That's a totally irresponsible conclusion based on their evidence.

It's one thing to say that withdrawal seems to be better than nothing, it's quite another to suggest the practice as a long-term birth control method on par with condoms.

Predictably, the ABC website is running an uncritical story about the study under the headline: "'Pulling Out' Method Gets New Respect: Study says withdrawal is "better than nothing. Women react with 'sheer disbelief.'" Oh, those silly women with their obsession with logic and evidence. 


TrackBack URL for this entry:

Listed below are links to weblogs that reference Withdrawal study: Ur doin' it wrong:


Maybe the sheer disbelief is about the idea that the man will in fact pull out.

It's no accident that "I won't come in your mouth" is counted as one of the Great All-Time Lies.

I imagine that sort of dishonesty applies to other orifices as well.

What percentage of teenage boys, do you suppose, have enough experience and control to reliably pull out?

I'm glad you are objecting to this article -- everywhere else on the net people seem to just be linking to it uncritically.
Here's another big problem with their thesis: its badly innumerate. They say that there is only a very small difference between the success rates of condom use and the withdrawal method, which are 2% (best use) and 17% (typical) for condoms and 4% (best use) and 18% (typical) for withdrawal. Now the difference in success rates for typical use is indeed pretty small. But the difference for best use is enormous. Indeed, the data state that if you are correctly using one of these forms of contraception, you are TWICE as likely to avoid pregnancy if you use condoms than if you use the withdrawal method. That is not remotely a trivial difference. Not to mention the question-begging you point out about what "best use" means for the withdrawal method.

Any way, glad to see someone critiquing this stupid study instead of just prominently linking to it.

The thing that surprises me most is the appallingly bad numbers for typical condom use. Given that fertilization is only possible for a few days each month the number of instances where some sperm makes it into the vagina has to be quite large in order to get a 17% pregnancy rate in a single year. Unless typical condom use includes quite a lot of breakage or simply not using anything I can't make sense of that number.

My own experience suggests breakage is a minimal problem, but perhaps there are practices that make it more common.

Perfect and typical use in condoms is incredibly confusing, for sure. The reason is that the stats are measured on a per year basis. So, if you put that you use condoms as your main form of birth control on the survey, then you go into that column. You are considered a perfect user if you use them every time, with lube, checking the expiration date, and putting them on correctly. But a lot of people in the typical use column only use them some of the time.

It sounds crazy to put "some of the time" into a "using protection" column, but as a public health measurement, it's actually pretty good. There's a lot of evidence to show that even spotty condom use has public health benefits, since as HIV transmission studies of prostitutes in HIV-heavy areas that show markedly lower transmission rates even if they use condoms somewhat sporadically. Or the 17% number that shows that people who at least try most of the time (but not all of the time) have much lower pregnancy rates.

My gut is that the numbers they have can't be right, and more research is needed. I know people who got pregnant using withdrawal correctly, even though that's supposedly hard to do. But then again, I know people who got pregnant on the pill and condoms, though those were typical use situations.

With condoms, there's also the issue of proper disposal. This requires the man to pull out before the erection ends, dispose of the condom, and wash his hands thoroughly before touching the woman again. If he doesn't, they run the risk of penis-to-hand-to-hand-to-vagina impregnation.

I've also been entirely skeptical of condom use pregnancy rates. I've used condoms inside a (contented, active) marriage for 25 years with zero failure.

And how the foo do you break one of them anyway? Stop buying decade old condoms out of truck stop vending machines, or use some lubrication.

Amanda's comment on post-coitus cross-contamination is well taken, however: it may not be completely romantic, but the male needs to hop up, dispose, wash hands and genitals and only then run back for post-coital bonding rituals.

Shows you care.

>...the male needs to hop up, dispose, wash hands and genitals and only then run back...

Some cleanup, sure. But if human semen were this robust, uber-fertile teenage girls wouldn't even be allowed to stand in movie lineups with men without wearing tivak suits.

Sperm remains viable outside the body for a few minutes. It's enough to get a girl pregnant after sex if you're not careful, but not to have any contamination in public in close quarters.

Alon is correct, Bruce: sperm are hardier than you suppose, and smearing a few on the vestibule of the vagina can lead to wigglers making it all the way.

(Sorry Alon, I attributed you previous comment to Amanda)

I find the success rate of vasectomy satisfactory. I've tested it more than a thousand times. :-)

I don't understand why more (hetero) sexually active women don't opt for continuous contraception. No period, no PMS, no pregnancy, no ill health affect. Every day is the same as every other, hormonally. The female OB/GYNs have been doing it for decades.

When I have tried to discuss this with young women, I get an earful of superstition. Women don't have to have a monthly period to be healthy. You can free yourself from unwanted pregnancy and the hassle of menstruation. The 7 blanks were put in the pill package by the marketing department and the catholic church. There is no scientific or medical reason for them to be there.

The failure rates are given per year, not per intercourse. With perfect use, condoms are supposed to result in pregnancy once per 50 woman-years.

Thanks for this: this is a piece of what I was saying to every press person who asked me about it...and yet.

When you really look at the data, there are so many obvious potholes (like how small the group in the census was using withdrawal as opposed to those using condoms, like the fact that this figure was based, at best, on self-reporting just like the self-reporting the writer criticized elsewhere, etc.), I walked away from it feeling like that opening paragraph was more of a stylistic attempt at grabbing the reader's attention than an earnest statement.

And like that's all of the paper most people read.

I don't understand why more (hetero) sexually active women don't opt for continuous contraception.

There are a lot of female-bodied people who react very badly to Hormonal Birth Control -- not just physical problems such as nausea, weight gain, or vaginal dryness, but also mood issues, and total loss of libido.

Also, not all uterus-owning people can manage continuous contraception. Some can, but other people's uteruses are more stroppy and will insist on bleeding after a month, or two, or three, until you cut the pills and remind them what a real hormone-drop looks like so they'll shut up for a while.

The day that HBC isn't a "try all the pills and see which ones work with your individual body chemistry" die-roll, but is instead guaranteed no-side-effects? I'll start HBC. Of course, I'll probably be 90 years old by then and not need it anymore... :p (Advance faster, medical technology! Advance faster!)

Generic avodart [Dutasteride] is responsible for the smooth progress of getting better urinary flow. It may also lessen the likelihood of a prostatic surgery later on. This drug really shrinks the prostate causing improvement not only in symptoms but also the cause.

I can't beileve that someone would actually publish such misleading results as this. I was shocked an appauled when a friend of mine actually tried to justify her unprotected sex with this guttmacher article!!

The comments to this entry are closed.