Hysteroscopic device implantation, a minimally invasive alternative to traditional laparoscopic sterilization for women, was demonstrated to have high effectiveness and low peri-procedural complications in early clinical studies.1,2 After wide commercialization, safety concerns related to hysteroscopic sterilization were raised,3,4 with 23,773 adverse events reported to the U.S. Food and Drug Administration (FDA) by the end of 2017.5 In September 2015, an FDA panel discussion led to the label change in 2016 that included a black box warning and a Patient Decision Checklist.6 Recently, Bayer announced their intention to withdraw their commercial hysteroscopic sterilization product (Essure) after 2019 due to declining sales.
Our short-term study found that hysteroscopic sterilization was associated with a significantly higher risk of reoperation when compared with laparoscopic sterilization.7 Following our study, investigators from France and the United Kingdom8,9 also reported elevated risks of tubal re-intervention after hysteroscopic sterilization. A U.S. study using commercial claims data10 reported that hysteroscopic sterilization was associated with increased hysteroscopic interventions, but fewer abdominal surgeries, when compared with laparoscopic sterilization.
Previous studies assessing outcomes after hysteroscopic sterilization had median follow-ups ranging from 1–3 years. We sought to assess longer-term outcomes after interval sterilizations in a cohort of women with median follow-up of 7 years. In doing so, we provide evidence to the many women who had hysteroscopic sterilization performed prior to market removal. In addition, there are consumer concerns about the systemic effect of the nickel component in the hysteroscopic implant.11 Our secondary aim was to assess cancer incidences after interval sterilizations.
We used data from the New York State Department of Health Statewide Planning and Research Cooperative System. This is an all-age group, all-payer database that collects patient and treatment information for every hospital discharge, outpatient and ambulatory surgery, and emergency department visit in New York State. Procedures performed in operating rooms in extended outpatient surgical settings of hospitals were captured as outpatient surgical procedures. A unique personal identifier is assigned to every patient and encrypted to allow longitudinal analyses, without compromising the confidentiality of the records.
This observational cohort study included women undergoing hysteroscopic and laparoscopic sterilization procedures in outpatient and ambulatory surgical settings between 2005 and 2016 in New York State. We included procedures performed in outpatient or ambulatory surgical settings to maintain a clean comparison with outpatient laparoscopic procedures, because some of the procedures performed in inpatient settings may be more complex. We selected our cohort using Current Procedural Terminology, Fourth Edition (CPT-4) and International Classification of Diseases, Ninth Revision, Clinical Modification codes (Appendix 1, available online at http://links.lww.com/AOG/B250).7 We used the same exclusion criteria from our previous study (Fig. 1). Laparoscopic sterilization was defined as laparoscopic ligation or occlusion of fallopian tubes. We excluded patients who had previous sterilization procedures and patients who underwent sterilization via laparotomy or concurrent with other abdominal procedures. We defined the first sterilization procedure recorded for each patient in the Statewide Planning and Research Cooperative System as the index procedure and followed up until the end of the study period (December 31, 2016). The presence of a CPT modifier code 52 (reduced service modifier) likely represented index sterilization procedures that were aborted or not completed due to patient conditions. We defined these procedures as reduced index procedures, which were included in the cohort based on the intention-to-treat principle.
We examined the following patient characteristics: age, race and ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, and other), insurance payer (Medicare, Medicaid, commercial, and other), comorbidities (uterine leiomyoma, endometriosis, obesity, anemia, cardiovascular disease, diabetes, and depression), previous pelvic inflammatory disease (PID), previous menstrual disorders, history of major abdominal surgery, history of cesarean delivery, and history of breast cancer, gynecologic cancer, and other solid malignant tumors. We identified common comorbidities using International Classification of Diseases, Ninth Revision, Clinical Modification algorithms validated by Elixhauser.12 Coronary artery disease, hypertension, congestive heart failure, and peripheral vascular disease were combined to create one variable for cardiovascular comorbidity.
The median follow-up for our cohort was 7 years, with a range of 0–12 years. To obtain reliable estimates and comparisons, we evaluated outcomes within 7 years after the index sterilization procedures. We examined the following surgical outcomes after the index sterilization procedure: additional hysteroscopic sterilizations, additional tubal ligations or resections, and hysterectomies that were not related to gynecologic malignancies, uterine leiomyomas, or benign uterine tumors. In the secondary analysis, we examined incidence of major cancers (solid tumors) within a subcohort of patients who did not have cancer at baseline. For cancer outcomes, we included the most common solid tumors (with top 20 incidence) in the United States,13 with modification to further include all gynecologic cancers (Appendix 2, available online at http://links.lww.com/AOG/B250).
We compared baseline characteristics between patients undergoing hysteroscopic and laparoscopic sterilizations, using Student t-tests for continuous variables and χ2 tests for categorical variables. We examined subsequent gynecologic procedures after index sterilization procedures and reported results for up to 7 years. We also examined the occurrence of a reduced index sterilization procedure. The type of tubal interventions for the first subsequent surgery after complete and reduced index procedures were then summarized respectively.
We conducted time-to-event analyses to compare the risks of subsequent procedures after index hysteroscopic and laparoscopic sterilizations. Propensity score matching14 was used to adjust for differences in baseline characteristics between patients undergoing hysteroscopic and laparoscopic sterilizations. We used a multivariable logistic regression based on patient characteristics, comorbidities, previous conditions and procedures, and procedure year to obtain propensity scores. We created a missing category for race and ethnicity and performed 1:1 fixed-ratio nearest neighbor matching. A caliper width of 0.2 of standard deviation of the logit of the propensity score was used initially and then tightened to 0.1 based on matching performance. Because previous PID did not get satisfactory matching between groups with initial matching, we matched separately for patients with and without previous PID and combined the datasets for analyses. We assessed the balance achieved by matching by examining differences in baseline variables before and after matching.
We performed Kaplan-Meier analyses and obtained estimated risks of reinterventions in the propensity score matched cohort. We used Cox proportional hazard models to assess the differences in risks of reinterventions between the two groups, with a robust sandwich variance to account for matched data. We tested the proportional hazards assumptions to confirm the adequacy of the model. For the outcome of additional tubal ligation and resection, the proportional hazards assumption was not satisfied. Based on the Kaplan-Meier analysis, we included a time-dependent covariate in the model and assessed the differences in risks of additional tubal ligation and resection between the two procedure groups within 1 year, between year 1 and year 2, and after year 2. We performed a sensitivity analysis among patients whose procedure did not have a reduced index procedure modifier (without CPT modifier 52). We performed propensity score matching within this subcohort and time-to-event analyses similar to those used in the main analyses.
Within a subcohort of patients who did not have cancer at baseline, we examined the incidence of major cancers after hysteroscopic and laparoscopic sterilization. We performed propensity score matching within the subcohort and time-to-event analyses to assess the difference in cancer incidences between the two procedure groups. Due to the low occurrence, individual cancer incidence, such as ovarian cancer, was not able to yield valid statistical estimates. Therefore, cancer outcomes were aggregated to be reported as gynecologic and nongynecologic cancer incidences. All analyses were performed using SAS v9.3.
The study was approved by the Weill Cornell Medical College institutional review board (protocol No. 1209013064).
This study included 10,143 and 53,206 women who underwent interval hysteroscopic and laparoscopic sterilizations in New York State from 2005–2016. The mean age of the cohort was 34.2 years (range 18–80, Table 1). When compared with those undergoing laparoscopic sterilization, there were larger proportions of women undergoing hysteroscopic sterilization who were black (17.2% vs 13.4%) and insured by Medicaid (47.2% vs 40.4%). Women undergoing hysteroscopic sterilization were more likely to be obese (6.1% vs 3.2%) and have cardiovascular disease (6.1% vs 4.0%), but less likely to have leiomyomata (0.5% vs 1.6%) and endometriosis (0.1% vs 1.1%) than those undergoing laparoscopic sterilization (all P<.001). Women undergoing hysteroscopic sterilization were also more likely to have previous PID (10.5% vs 9.4%), menstrual disorder (7.9% vs 6.2%), abdominal surgeries (4.0% vs 3.1%), and cesarean delivery (20.6% vs 12.6%) than those undergoing laparoscopic sterilization (all P<.001). There was no significant difference in history of breast or gynecologic cancer, or other malignancies between the two groups.
In the entire cohort, 438 (4.3%) women undergoing hysteroscopic sterilization and 1,241 (2.3%) patients undergoing laparoscopic sterilization had at least one subsequent procedure within 7 years after the index procedure. Among the 438 patients who had at least one subsequent procedure after their index hysteroscopic sterilization (Fig. 2), 82 (18.7%) women had a second hysteroscopic sterilization procedure, 293 (66.9%) women had a laparoscopic tubal ligation or resection and 63 (14.4%) women had a hysterectomy for their first reoperation, respectively.
The propensity score-matched cohort consisted of 10,109 pairs of women undergoing sterilization procedures. The estimated risk of undergoing an additional hysteroscopic sterilization within 7 years after an initial hysteroscopic procedure was 0.8% (Table 2). The estimated risk of undergoing an additional tubal ligation or resection within 7 years was higher after hysteroscopic sterilization than that after laparoscopic sterilization (3.9% vs 1.6%; HR 2.89, 95% CI 2.33–3.57). The difference was most pronounced within the initial year (1.5% vs 0.2%; HR 6.39, 95% CI 4.16–9.80; Table 3, Fig. 3). After year 2, the estimated cumulative risk of additional tubal ligation or resection increased at a rate of around 0.3% for women undergoing hysteroscopic sterilization and 0.2% for women undergoing laparoscopic sterilization.
There was no significant difference in the risk of receiving a hysterectomy within 7 years after initial procedures between women undergoing hysteroscopic and laparoscopic sterilizations (0.9% vs 1.2%; HR 0.73, 95% CI 0.53–1.00; Table 2). Over time, the estimated risk of undergoing hysterectomy was 0.1–0.2% every year for women undergoing hysteroscopic and laparoscopic sterilizations (Table 3; Fig. 3). Results from the sensitivity analysis among women whose index procedures did not have a reduced procedure modifier were consistent with the main analysis (Appendix 3, available online at http://links.lww.com/AOG/B250).
In the secondary analyses of cancer incidences among women who did not have cancer at baseline, 10,053 pairs of women undergoing sterilization procedures were matched. There was no difference in the incidences of gynecologic cancer (0.1% vs 0.1%; HR 2.63, 95% CI 0.70–9.91) or other cancers (1.2% vs 1.3%; HR 1.03, 95% CI 0.78–1.36) after initial hysteroscopic and laparoscopic sterilizations (Appendix 4, available online at http://links.lww.com/AOG/B250).
We found that hysteroscopic sterilization was associated with an increased risk of additional tubal intervention within 7 years when compared with laparoscopic sterilization, with the most pronounced difference occurring during the first year after the procedures. Although limited by follow-up time, there was no increased risk of hysterectomy or any cancer after hysteroscopic sterilization during the study period.
Our pioneering study found that hysteroscopic sterilization was associated with an increased risk of repeated tubal sterilization at 1 year,7 which has been confirmed by studies from the United Kingdom and France.8,9 Another study from the United States using commercial insurance data did not find a difference in the risk of undergoing additional surgeries after hysteroscopic and laparoscopic sterilizations.10 The study by Perkins et al had two major limitations, which we addressed in our investigation. First, it did not include women insured by Medicaid, which accounted for almost half of women undergoing sterilization procedures in our study. Second, the authors grouped tubal reinterventions with other procedures including hysterectomies and exploratory surgeries such as laparotomy. We examined tubal procedures, hysterectomies, and cancer incidences separately to conduct a thorough assessment of outcomes after the index sterilization procedures.
Compared with prior studies (median follow-up 3 years, range 1–5),8–10 our study offered insights with longer follow-up. This is particularly important because we showed that the risk of subsequent tubal ligation and resection after initial sterilizations differed by follow-up time-period. The most significant increase in the risk of subsequent tubal ligation and resection after hysteroscopic sterilization occurred during the first and second years. Some of the differences in additional tubal ligation and resection within the first year may be attributable to reduced index procedures secondary to tubal spasm or patient intolerance. In addition, previous clinical trials showed that 3–4% of women undergoing hysteroscopic sterilization had unsatisfactory device location at the 3-month hysterosalpingogram follow-up,1,2 which often required a subsequent procedure to achieve sterilization.
After the 2015 FDA panel discussion, the manufacturer was requested to conduct a postmarket study of hysteroscopic sterilization, with the focus being efficacy.15 However, patient recruitment for the study has been slow and the length of the study is long. More recently, the manufacturer has decided to halt the sale of the device in the United States starting in 2019,16 after stopping marketing the device in several other countries. Under the current circumstances, studies with existing data represent important opportunities to investigate the long-term safety of the device and provide tentative answers to remaining questions. There have been reports of device removal in longer-terms (up to 10 years) after the initial hysteroscopic sterilization.17 The manufacturer estimated that more than 750,000 women received the device.18 For the hundreds of thousands of women with the device already implanted, continuous monitoring of the long-term outcomes after hysteroscopic sterilization is crucial19 given the limited evidence of the risks of the device beyond 7 years.
There are limitations to our studies. Administrative data lack clinical details and the identification of comorbidities and past conditions relied on diagnosis and procedure codes, which were subject to miscoding. Patient conditions underlying the decision for a hysteroscopic procedure may not have been fully captured. For these reasons, there might be residual confounding. We attempted to create two balanced groups with propensity score matching to account for baseline conditions. Given the significance of the differences in undergoing subsequent tubal ligation and resection in the current analysis, this association is unlikely to disappear with all confounding accounted for. In addition, we were unable to investigate the exact reasons that led to the reinterventions. Studies using other data sources are warranted to address this question. Our study did not include salpingectomy because it is a more aggressive surgical method compared with hysteroscopic and laparoscopic sterilization. With longer-term follow-up, it was possible that some patients relocated to other states and were lost to follow-up. However, by using propensity score matching, loss to follow-up is likely nondifferential between the two groups, and, thus, bias would be toward the null. Finally, our interpretations of cancer incidences must be considered within the context. To reach a more definitive conclusion for the incidences of cancers that affect older women, additional years of follow-up may be needed.
Our study demonstrated the time-varying risk of tubal reintervention associated with hysteroscopic device sterilization, in comparison with traditional laparoscopic sterilization with limited evidence of outcomes after hysteroscopic sterilization and existing reports of removals beyond 7 years, ongoing monitoring of long-term outcomes for women who received the device is warranted.
1. Kerin JF, Cooper JM, Price T, Herendael BJ, Cayuela-Font E, Cher D, et al. Hysteroscopic sterilization using a micro-insert device: results of a multicentre phase II study. Hum Reprod 2003;18:1223–30.
2. Chudnoff SG, Nichols JE Jr, Levie M. Hysteroscopic essure inserts for permanent contraception: extended follow-up results of a phase III multicenter international study. J Minim Invasive Gynecol 2015;22:951–60.
7. Mao J, Pfeifer S, Schlegel P, Sedrakyan A. Safety and efficacy of hysteroscopic sterilization compared with laparoscopic sterilization: an observational cohort study. BMJ 2015;351:h5162.
8. Antoun L, Smith P, Gupta JK, Clark TJ. The feasibility, safety, and effectiveness of hysteroscopic sterilization compared with laparoscopic sterilization. Am J Obstet Gynecol 2017;217:570 e1–6.
9. Bouillon K, Bertrand M, Bader G, Lucot JP, Dray-Spira R, Zureik M. Association of hysteroscopic vs laparoscopic sterilization with procedural, gynecological, and medical outcomes. JAMA 2018;319:375–87.
10. Perkins RB, Morgan JR, Awosogba TP, Ramanadhan S, Paasche-Orlow MK. Gynecologic outcomes after hysteroscopic and laparoscopic sterilization procedures. Obstet Gynecol 2016;128:843–52.
12. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27.
14. Rassen JA, Shelat AA, Myers J, Glynn RJ, Rothman KJ, Schneeweiss S. One-to-many propensity score matching in cohort studies. Pharmacoepidemiol Drug Saf 2012;21(suppl 2):69–80.
17. Maassen LW, van Gastel DM, Haveman I, Bongers MY, Veersema S. Removal of Essure sterilization devices: a retrospective cohort study in the Netherlands. J Minim Invasive Gynecol 2018 Oct 16 [Epub ahead of print].