Wednesday, January 3, 2007

Crunchin' Some EC Numbers [Serge]

JivinJ asks in the comments:

I'm trying to crunch the numbers to figure out what the effectiveness of EC would have been estimated at if the WHO used the pregnancy risk number from the Canadian study.

Correct me if I'm wrong but my estimations from trying to decipher how they figured out the effectiveness (is it percentage of pregnancies divided by pregnancy risk?) show that if the WHO study used the pregnancy risk number (4.12) from the Canadian study instead of the 7.7 garnered from the traditional method of estimating pregnancy risk and then took the percentage of women who became pregnant from the WHO study then the effectiveness of EC would be somewhere around 75% and not the 85% cited for the WHO.
I actually believe its worse than that. James Trussell, who I believe was an author in the WHO study, uses a meta analysis of all of the studies about plan B when discussing its efficacy. The 1.1% crude pregnancy rate found in that study has not been reproduced. Trussel states:

Therefore, the chance that pregnancy would occur in the absence of emergency contraception is estimated indirectly using published data on the probability of pregnancy on each day of the menstrual cycle.14,15 This estimate is compared to the actual number of pregnancies observed after treatment in observational treatment trials. Effectiveness is calculated as 1-O/E, where O and E are the observed and expected number of pregnancies, respectively...

Eight studies of the levonorgestrel regimen that included a total of more than 9,500 women reported estimates of effectiveness (a reduction in a woman’s chance of pregnancy) between 59% and 94%.7,8,9,17,18,19,20,21 A meta-analysis of eight studies of the combined regimen including more than 3,800 women concluded that the regimen prevents about 74% of expected pregnancies; the proportion ranged from 56% to 89% in the different studies.
Using the 7.7% rate for the traditional method of estimating pregnancies, Trussel's numbers are about a 2% rate of pregnancy when a woman is using Plan B EC (from the 74% number). This translates to an effectiveness of 51%.

One more thing. As the authors of the Canadian study state, even the revised method of estimating pregnancy may not be very accurate. This study of using ultrasound techniques to determine ovulation time was far more accurate than the calender method.

Bottom line: if the Canadian study is accurate, and the data from the meta-analysis of Plan B is used, then its maximum effectiveness is 51%. I believe further data will show that it is actually less than that.


  1. Thanks Serge.

    So the high effectiveness number used by EC proponents uses both a bad pregnancy rate number (around 1%) and a bad pregnancy risk number.


  2. I believe you are correct, and that is a great way to say it. As I review the literature, I don't believe anyone would claim a >90% effectiveness of EC - even though it remains on the label. There is even more evidence that I haven't revealed here.


All comments are moderated. We reject all comments containing obscenity. We reserve the right to reject any and all comments that are considered inappropriate or off-topic without explanation.