The New York Times provided an estimate of"How Likely Is It That Birth Control Could Let You Down?" over a 10 year period. They estimate how many women will have at least one unplanned pregnancy in a ten year period depending on the type of contraception they use. For "typical" use, 86% of women using male condoms and 61% of women who used the pill would have had at least 1 unplanned pregnancy in 10 years.  92% of women are predicted to have had at least one unplanned pregnancy using the withdrawal method.

 

But can these figures be trusted?  Are they supported by real data?

 

It is important to understand how these numbers are calculated.  These numbers are extrapolated from data on the numbers of unplanned pregnancy in a single year (the first year).  In a study conducted in the USA, data showed that 18 out of women 100 experience an unplanned pregnancy using male condoms in a single year. This in itself is a larger percentage than many of us would have thought, and certainly should encourage us to be careful that we use contraception properly. In the study, "typical" usage takes into account how people actually use condoms, and this is not equal to the rates predicted under "perfect" use.

If one woman gets unexpectedly pregnant with probability 0.18, then the probability that at least one of a group of ten women get pregnant is $1-(1-0.18)^{10}=0.86$. So we would expect at least one pregnancy to occur with 86% probability. But does this really imply, as the New York Times suggests, that 86% of women experience an unexpected pregnancy during a 10 year period?

 

Of course it doesn't! And here's why. The New York Times assumes that the way is which individuals use contraception in their first year has no bearing on how they use contraception in their second year.  In statistical terms, they assume that the years are independent. Contraception usage, however, is not independent between years.  A fact that the New York Times fails to acknowledge.

 

So how does non-independence affect the numbers?  Below, we plot the expected number of pregnancies using the 'New York Times' method for different contraceptives.  Here we extrapolate from data on the number of unplanned pregnancies experienced by women during a 3 month period (numbers from Kost et al. 2008).  The expected percentage of unplanned pregnancies (the solid lines) are given for each type of contraceptive method. However, the colour-coded solid circles show the actual percentage of women experiencing an unplanned pregnancy after 3, 6 and 12 months from the start of a study (Kost et al. 2008).  In every case the actual numbers of unplanned pregnancies are below the expected rates of unplanned pregnancies.

 

 

Using these NY Times method, we would expect about 90% of women using the fertility awareness method to have experienced an unplanned pregnancy within 5 years. This is very unlikely to be correct. Unfortunately, as the figure below illustrates, we can't tell for sure how likely you or anyone else is to get pregnant using this method. Just that it is less than 90%.

 

 

So what are the real numbers of unplanned pregnancies over longer time periods? Are they more similar to the dashed line in the above figure, and not the solid lines?  This can only be determined from longitudinal studies.  If the assumptions of mathematical models are not appropriate, then their predictions can over (or under) estimate what happens in the real world.  Whilst the actual numbers of unplanned pregnancies are still high, they are unlikely to be as high as what the New York Times would have you believe.