Each year, more than 300,000 women request permanent surgical sterilization in the United States [1], and laparoscopic bilateral tubal ligation remains the most widely used operative approach to achieve this. However, in 2002 a non-incisional technique became available for bilateral tubal occlusion via hysteroscopic placement of metal inserts at the utero-tubal junction. Hysteroscopic sterilization (HS) brings several advantages over standard laparoscopy by eliminating the need for abdominal access [23], reducing overall cost [4], and minimizing anesthesia requirements [5]. The only HS method currently available is Essure (Bayer HealthCare AG, Whippany, NJ, USA) which was the focus of the current study. While latex free, the HS implants do contain acrylonitrile butadiene styrene, polytetrafluoroethylene polyvinyl pyrrolidone, polyethylene terephthalate and nickel, among other components [6].

Worldwide, at least 750,000 women have undergone HS and this contraceptive method is reported to be safe and effective [78]. In France, government health authorities recommend HS as the "first line alternative" for women over age 40 who request permanent contraception [9]. Even its off-label use in the management of hydrosalpinges (before in vitro fertilization and embryo transfer) has been mostly favorably reported [10]. Efficacy data provided by the device manufacturer claims that HS is 99.83% effective at preventing pregnancy over a five year interval, with the proviso that the implant is only used for approved indications and according to "perfect use" guidelines [11]. However, a recent investigation [12] questioned several aspects of the prior reports and, crucially, presented a very different efficacy assessment of HS. Using an evidence-based Markov model to estimate pregnancy rates among HS users over a 10 year interval, pregnancy probability at year 1 and over 10 years was found to be substantially higher with HS compared to laparoscopic sterilization [12]. Disagreement on contraceptive efficacy notwithstanding, the lay press has called attention to serious safety concerns associated with HS [13], thus underscoring the urgent need for more data on this contraceptive method. Moreover, clinical experience with HS remains somewhat limited as the total worldwide output of Essure publications is currently less than 200 despite more than a decade of clinical use. Against this background, the current analysis aimed to improve the understanding of HS by offering descriptive information collected directly from Essure patients who experienced an unintended pregnancy after undergoing this contraceptive procedure.

A multi-item research questionnaire was developed specifically for women who had a medically confirmed pregnancy following HS with Essure. Queries were structured to permit numerical or text responses (Table 1) and then configured electronically for an internet interface. The sample consisted of women who had already joined a closed, invitation-only support group for patients who underwent HS and had at least one subsequent unplanned pregnancy. Although participants were required to register via computer to access the questionnaire site, there was no cost to do so and respondents received nothing of value in exchange for their responses.

Participation in this investigation was open to all members of a support group (n=115), itself a subsidiary of the closed, web-based "Essure Problems" patient education and advocacy group (n=16,772) which is currently the world's largest comprehensive internet resource for Essure patients (https://www.facebook.com/groups/Essureproblems/). The questionnaire was posted in English and maintained on a dedicated internet site for a five week period, beginning February 6, 2015. Incomplete questionnaires were excluded and responses were electronically tracked to block duplicate submissions from the same individual. Height and weight information was received from participants either in Imperial (US) or SI (Système International, metric) units, with automated metric conversion for analysis. Only investigators had access to running totals of the questionnaire during the study.

This study proposal was submitted to an independent ethics committee prior to questionnaire implementation. Because the investigation tabulated data in an anonymous, nonidentifiable manner and involved no direct contact with study participants, the research protocol was judged as "no risk to human subjects" and exempted from review.

HS has been available to patients in the United States since 2002, although this is the first study to gather data directly from women with an unplanned pregnancy while using this method of birth control. Our analysis contributes to a larger understanding of HS by extending the observations of previous studies, while also challenging some others. For example, this analysis supports the findings of prior investigators who have noted reduced contraceptive efficacy when a method requires multiple steps (such as Essure) compared to simpler, single-step methods like intrauterine devices and contraceptive implants [1215]. The current data align with this observation as patient compliance with the use of a "back up" birth control method (for three months immediately following HS) was only 73.8%, and the rate of follow-up with the required HSG after HS was even worse at 66%. Because this study group was educated (>40% had at least two years college) and most had access to insurance (>85% had private health coverage), it is difficult to identify specific characteristics that might explain this low self-reported adherence.

To date, studies on patient compliance with the post-HS HSG have been limited and have reported variable results [16]. Specifically, during clinical trials for HS patient compliance with post-placement HSG was very high, exceeding 90% [17]. Insured patients have been reported to have good adherence rates for keeping their HSG appointment [18] although lower income women may have a post-HS HSG compliance rate below 20% [19]. Not surprisingly, clinics that place a nurse in charge of scheduling HSG appointments for HS patients, calling them with appointment reminders, and tracking HSG compliance will have a significantly better HSG follow-up rate, compared with no outreach effort [20]. A retrospective cohort study of 638 patients who had HS over a six-year period [21] reported that 57% of HS procedures were performed in an operating room (not in a doctor's office), but observed no association between success in bilateral device placement or occlusion and any patient characteristic, irrespective of the clinical setting (hospital vs. office). However, their investigation acknowledged that private insurance, patient age, and performance of HS in the office setting were positively associated with likelihood of compliance with post-HS hysterosalpingography [21]. Other investigators have found that not speaking English or having to travel large distances are not associated with "no show" status at the post-HS HSG appointment to confirm proper placement [16]. While all of our study participants were English-speaking, we were unable to assess distance traveled for HSG or the possible role of economic status in our sample, as these parameters were not collected in our study. No clinical data on post-HSG counseling following HS have yet been published, and indeed for those women in our study group who did comply with the HSG three months after HS, we do not know if the radiograph confirmed bilateral tubal occlusion or not.

Our information on prior smoking status (and change in smoking status subsequent to HS), prior abdominal/pelvic instrumentation, and total number of pregnancies (and deliveries) before HS, joins observations from other reports in describing features of HS patients. This study also supplies original data regarding what types of birth control were being contemplated by patients when their choice for HS was made, as well as the role of a husband/partner on a woman's decision to undergo HS. It is hoped that future research will include these parameters going forward.

This evaluation has some limitations which should be acknowledged. For example, the internet-based sampling methodology used to collect study data warrants comment because error could be introduced if any incorrect patient information was included. However, validation of actual procedure notes and lab test records supplied by a random audit of >10% of the total respondent group assuaged these concerns. Moreover, patient questionnaire-based data analyses may be affected by low response and recall bias. In this investigation the response rate approached 90%, an unusual feature we attribute to the participants being generally well educated and highly motivated, obtained from a closed-membership internet support group. Additionally, these data were collected from patients who self-selected into a voluntary support group, so it cannot be known if these respondents are representative of all unplanned pregnancies after HS. Importantly, estimating contraceptive failure rate based on these data is inappropriate (and outside the scope of our study) because it was impossible to determine with accuracy the geographical origins of study participants. Recall bias also cannot be excluded, but a planned follow-up study involving detailed, independent review of patient medical records will be helpful in this regard. Finally, we were unable to assess characteristics of the practitioners who performed HS in this dataset, so there could be specific physician factors associated with contraceptive performance with HS that escaped detection.

Considering the patient data presented here, previous risk models [12] based on a relatively high first-year failure rate (i.e., pregnancy occurring within the first year after HS) appear valid but may still require some recalibration. Our analysis of contraceptive failure with HS did identify a peak unplanned pregnancy incidence within the first 12 months of device placement, but mean interval to pregnancy after HS was closer to 20 months. We may thus speculate that even if these patients had availed of an alternate method of contraception in the first three months following HS, many unplanned pregnancies would have still occurred. Since contraceptive efficacy for HS is often reported based on total number of HS devices sold during a reference period rather than the actual number of patients who undergo HS (Fig. 2), the published failure rate for HS may not be reliable [9]. Calculating contraceptive efficacy using absolute patient count instead of product inventory data may give a more accurate impression of actual or "realworld" HS experience, as described previously [12]. Additional research is needed to complete our understanding of outcomes following HS.