is a multi-step process that requires women to have a procedure to place coils inside the opening of the Fallopian tubes, use another method of contraception for three months after the procedure, and then have a special X-ray test in which dye is pushed into the uterus to confirm whether the tubes are blocked.
The study in question used a computer model to predict failure rates.
The study uses data in the published literature to model what happens to women who start down a path of wanting a laparoscopic sterilization or hysteroscopic sterilization, including those who do not successfully have the procedure. The computer model, called a decision analysis, calculates what could occur in a theoretical group of 100,000 women taking into account all of the potential options that could happen in each step of the process.
The authors found that pregnancy risk after hysteroscopic sterilization is primarily accrued in the first year after initiating the process because hysteroscopic sterilization is not immediately effective. Conversely, laparoscopic sterilization is immediately effective.
The major findings by Gariepy and colleagues include that pregnancy rates in the first year for women planning hysteroscopic sterilization are 57 per 1,000 women, compared with about 3 to 7 per 1,000 women for laparoscopic sterilization. The total pregnancy rate over 10 years reached 96 per 1,000 women for hysteroscopic sterilization compared to only 24 to 30 per 1,000 women with a laparoscopic procedure. The authors accounted for other methods of contraception that would be used for women who did not have a sterilization procedure, including that some women who have a failed hysteroscopic procedure would choose a laparoscopic procedure.
Let's suppose 100 000 use Essure (just to pick an easy number). Close to 1000 will get pregnant over ten-year period.
That's not statistically insignificant.
Of course, this is based on a computer model. But I have a hunch-- and it's only a hunch-- that the actual failure rate is higher because we tend not to see everything that could go wrong.
For a method that's supposed to be fullproof, it's not so fullproof.