Uptake and Predictors of Opportunistic Salpingectomy for Ovarian Cancer Risk Reduction in the United States

Ovarian cancer is the deadliest gynecologic cancer and the fifth leading cause of cancer deaths among U.S. women (1). Despite advances in treatment over the past three decades, the 5-year survival of ovarian cancer remains less than 50%. Population-based screening for ovarian cancer is currently not recommended as available screening tests have high false-positive rates and are not associated with reductions in ovarian-cancer incidence and mortality (2–4). In the absence of effective screening, risk-reducing surgeries have been adopted to reduce the mortality from ovarian cancer in high-risk women (5).

Recent studies have indicated that high-grade serous ovarian cancer, the most common and lethal ovarian cancer subtype, originates in the fallopian tubes and involves the ovaries secondarily (6, 7). These findings have led to the emergence of opportunistic salpingectomy (OS), the surgical removal of the fallopian tubes during benign gynecologic surgery, as a novel ovarian cancer risk-reducing strategy. Data from 3 observational studies suggest that OS may be associated with a 42% to 64% reduction in ovarian-cancer risk (8–10). However, these studies were conducted in women undergoing OS for indications other than ovarian-cancer risk reduction and did not assess long-term morbidity. Despite limited data on efficacy, some national societies in the United States now recommend OS for ovarian-cancer risk reduction in place of tubal ligation for sterilization or during a benign hysterectomy in average-risk women (11, 12). High-risk women, including those with BRCA mutations and/or a strong family history of ovarian cancer are still advised to undergo the removal of both the ovaries (oophorectomy) and the fallopian tubes (5). However, interval OS has been proposed as an approach to delay oophorectomy in premenopausal high-risk women. Several studies are currently assessing the safety and acceptability of this approach in BRCA mutation carriers (13–15).

The extent to which guideline recommendations for OS use have led to changes in clinical practice is unclear since the majority of information regarding OS dissemination comes from studies that are limited to inpatient settings (16–18), which is not reflective of where most benign gynecologic surgeries are performed in the United States (19, 20). Prior studies have also focused on small geographic regions (21, 22), significantly limiting their generalizability. No studies have examined the predictors of OS adoption in a national sample of inpatient and outpatient surgeries.

We conducted a large observational study of over 48 million women to comprehensively characterize recent trends and predictors of OS use in a nationwide sample of inpatient and outpatient insurance claims and evaluate the impact of national guidelines supporting OS use on the speed of OS uptake.

We analyzed deidentified data from the 2010 to 2017 Truven Health Analytics MarketScan Commercial Claims and Encounters Database, an administrative claims database that contains a large sample of Americans with commercial, employer-sponsored health insurance. MarketScan includes approximately 350 payers and captures insurance claims from privately-insured patients less than 65 years old and their dependents. The database contains inpatient and outpatient encounters at hospitals and ambulatory surgical centers, and includes demographic characteristics, and admitting and discharge diagnoses and procedure codes. Each enrollee is provided with a unique identification number, which enables linking across different encounters and enrollment years. An individual's enrollment in the database may be subject to change if they lose employment, change employment, or transition to disability due to illness. The study was deemed exempt by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board.

Our study population consisted of 48,231,235 women without a history of breast or gynecologic cancer. Figure 1 shows the criteria used to identify the analytic sample. Women with at least one claim for tubal ligation, OS, or hysterectomy between 2010 and 2017 were identified (n = 1,046,525). All women undergoing an oophorectomy or a salpingo-oophorectomy during hysterectomy (n = 303,348), those with hysterectomy codes that do not provide detail regarding removal of additional adnexal structures (n = 97,021), and those with ectopic pregnancy, pelvic inflammatory disease, or pelvic infection were excluded (n = 7,200). For tubal ligation and OS for sterilization, women more than 50 years were excluded since sterilization procedures are rare in this age group (n = 863). After exclusions, 638,093 women with benign gynecologic surgeries were included in the analytic sample. Women were classified into 4 distinct groups by type of surgery: tubal ligation (n = 308,833), OS for sterilization (n = 13,462), hysterectomy alone without OS (n = 293,000), and hysterectomy and OS (n = 22,798). For brevity, hysterectomy alone will denote hysterectomy alone without OS for the rest of the article.

Surgeries were ascertained using the International Classification of Diseases (ICD) ninth and 10^th revision procedure codes and Current Procedural Terminology (CPT) codes. A set of minimally-sufficient codes were used to classify each surgical procedure (Supplementary Table S1). There were 5 groups: hysterectomy alone (group 1), hysterectomy, vague on oophorectomy or salpingo-oophorectomy (group 2; CPT codes only), bilateral oophorectomy or salpingo-oophorectomy (group 3), bilateral salpingectomy (group 4), and tubal ligation (group 5). For inpatient procedures, patients with at least one ICD code in group 1 and no ICD codes in groups 3, 4, or 5 were classified as hysterectomy alone, those with at least one ICD code in groups 1 and 4 and no ICD codes in groups 3 or 5 were classified as hysterectomy and OS, those with at least one ICD code in group 5 and no ICD codes in groups 1, 3, or 4 were classified as tubal ligation, and those with at least one ICD code in group 4 and no ICD codes in groups 1, 3, or 5 were classified as OS for sterilization. For outpatient procedures, patients with at least one CPT code in group 1 and no CPT codes in groups 2, 3, 4, or 5 were classified as hysterectomy alone, those with at least one CPT code in group 1 and CPT code 58700 and no CPT codes in groups 2, 3, or 5 were classified as hysterectomy and OS, those with at least one CPT code in group 5 and no CPT codes in groups 1, 2, 3, or 4 were classified as tubal ligation, and those with at least one CPT code in group 4 and no CPT codes in groups 1, 2, 3, or 5 were classified as OS for sterilization.

Demographic characteristics including age (<30, 31–40, 41–50, >50 years), year of surgery (2010–2017), insurance type [health maintenance organization (HMO), preferred provider organization (PPO), other], region of residence (Northeast, Midwest, South, West), and location of residence (rural, urban) were ascertained for each patient. Region of residence was based on U.S. Census delineations (23) and location of residence was mapped from the 5-digit zip code of the primary beneficiary by Truven Health Analytics. (24) Patients were classified based on the type of hysterectomy performed [abdominal, vaginal, laparoscopic (abdominal or vaginal)] based on ICD9/ICD10 procedure codes and CPT codes. Surgical indications for hysterectomy were identified based on the primary ICD9/ICD10 diagnosis code associated with the hysterectomy claim. A comorbidity score (0, ≥1 comorbidities) was calculated using the Klabunde adaptation of the Charlson comorbidity index (25). Family history of and genetic susceptibility to breast/ovarian cancer was based on ICD9/ICD10 diagnosis codes. Women were classified as having a family history of breast/ovarian cancer if they had a family history code for breast and/or ovarian cancer, but no genetic susceptibility code for breast and/or ovarian cancer during the study period. Women were classified as having a genetic susceptibility to breast/ovarian cancer if they had a genetic susceptibility code for breast and/or ovarian cancer with or without a family history code for breast and/or ovarian cancer during the study period (Supplementary Table S1).

Descriptive analyses were conducted for patient characteristics, and annual rates of surgery (tubal ligation, OS for sterilization, hysterectomy alone, and hysterectomy and OS) were calculated overall and by specific subgroups. Rates were calculated as the number of women with surgery during a specified period divided by the total number of women enrolled during the specified period (with and without surgery) and multiplied by 100,000. An interrupted time series (ITS) analysis using segmented Poisson regression was used to evaluate changes in the quarterly rate of OS from January 2010 to December 2017, with the Society of Gynecologic Oncology clinical practice guideline release in November 2013 as the interruption (11). The year 2014 was excluded to allow time for guideline implementation, resulting in a post-guideline period from January 2015 to December 2017. A scaling parameter was added to the models to allow for overdispersion of data and all models were adjusted for age and seasonal variation by using a Fourier term (26–28).

A Poisson regression model with robust variance was utilized to identify predictors of OS use. Covariates were selected a priori based on clinical rationale and literature review. Multivariable-adjusted models were run separately for OS for sterilization and hysterectomy and OS for benign indications. Models were stratified by year of diagnosis (2010–2013 and 2015–2017) to examine the impact of clinical-practice guideline release on the predictors of OS use. All models were adjusted for year of surgery, age, region, location, type of insurance, place of service, family history of and genetic susceptibility to breast/ovarian cancer, and underlying comorbidities. Multivariable models for hysterectomy and OS were additionally adjusted for hysterectomy route and surgical indication. We also examined the impact of location and region of residence on OS use in a subset of women with a family history of and genetic susceptibility to breast and/or ovarian cancer since this group is at the highest risk of ovarian cancer. These models were adjusted for year of surgery, age, type of insurance plan, place of service, and underlying comorbidities. A sensitivity analysis was performed by adding hysterectomy and bilateral salpingo-oophorectomies over the same period to the analytic sample to examine its impact on the proportional uptake of OS during benign hysterectomy. All analyses were conducted in Stata, version 15.1 (Stata Corporation) and all tests of statistical significance were two-sided.

Among 322,295 women undergoing sterilization, 13,462 OS and 308,833 tubal ligations were identified, and among 315,798 women undergoing a benign hysterectomy, 22,798 OS and 293,000 hysterectomies alone were identified (Table 1). The median age at surgery was similar for women undergoing tubal ligation [35 years; interquartile range (IQR): 31–39] and OS for sterilization (36 years; IQR: 32–40) and hysterectomy (43 years; IQR: 39–48) and hysterectomy and OS [44 years; IQR: 39–47). Regardless of surgery type, most women resided in the South (47.0%), in urban locations (79.7%), and had a PPO health plan (61.0%). About 80% of OS for sterilization and 23% of hysterectomy and OS were performed in outpatient settings (44% of all OS procedures). The most common hysterectomy routes for a hysterectomy and OS were laparoscopic (48.5%) followed by abdominal (30.0%). About 6.5% of women had a documented family history of breast/ovarian cancer (8.0% of OS for sterilization; 9.3% of hysterectomy and OS) and 0.3% had a documented genetic susceptibility to breast/ovarian cancer (0.5% of OS for sterilization; 0.3% of hysterectomy and OS). Most women undergoing surgeries had no diagnosed comorbidities. Demographic characteristics (age, region, location, type of health plan) of women undergoing surgeries did not differ from the overall study population.

All tubal ligations and OS alone included in the study were performed for sterilization. About 23% of tubal ligations and 16% of OS alone were performed postpartum, most of which were performed during a cesarean section (61% of tubal ligations and 60% of OS). Two thirds of hysterectomies alone were performed for uterine fibroids (31.7%) and abnormal uterine bleeding (31.7%). Other common indications for hysterectomy alone included prolapse (10.7%), pain/other symptoms of the genital tract (8.9%), and endometriosis (7.1%). Women with a diagnosis of uterine fibroids accounted for 46% of all hysterectomy and OS procedures. Other common indications for hysterectomy and OS included abnormal uterine bleeding (27.1%), pain/other symptoms of the genital tract (6.0%), prolapse (5.6%), and endometriosis (4.2%).

From 2010 to 2017, there was a 1,683% increase (3 to 47 per 100,000 women) in the annual age-adjusted incidence rate of OS for sterilization and a 659% increase (5 to 38 per 100,000 women) in the incidence rate of hysterectomy and OS. Over the same period, there was a 42% (294 to 169 per 100,000 women) decrease in the annual age-adjusted rate of tubal ligation and a 76% decrease (341 to 81 per 100,000 women) in the rate of hysterectomy alone (Fig. 2). The proportion of sterilization encounters involving OS increased from 1% in 2010 to 20% in 2017 and the proportion of benign hysterectomy encounters involving OS increased from 1% in 2010 to 32% in 2017 (Fig. 3). Incidence rates of OS increased across all age groups, in all geographic regions, in rural and urban locations, across all health plans, in women with and without underlying comorbidities, and in women with and without a family history of/genetic susceptibility to breast/ovarian cancer (Supplementary Table S2).

Results of the ITS analysis of OS for sterilization revealed that there was a 250% [incidence rate ratio (IRR): 3.51; 95% confidence interval (CI), 2.53–4.86] increase in the age-adjusted quarterly rate of surgery immediately after guideline release followed by a 14% (IRR, 1.14; 95% CI, 1.10–1.19) sustained increase in the rate of surgery after compared with before guideline release. For hysterectomy and OS, there was a 109% (IRR, 2.09; 95% CI, 1.65–2.63) increase in the age-adjusted quarterly rate of surgery immediately after guideline release followed by a 3% (IRR, 0.97; 95% CI, 0.95–0.99) sustained decrease in the rate of surgery after compared with before guideline release. Significant declines in tubal ligation and hysterectomy rates were observed, and these declines were temporally associated with the guideline release (Supplementary Table S3). No significant differences were observed in age-adjusted surgery rates in ITS analyses stratified by geographic region, rural/urban location, type of health plan, underlying comorbidities, and family history of/genetic susceptibility to breast/ovarian cancer.

Results of the multivariable analyses of predictors of OS use are summarized in Table 2. The year of surgery had the strongest association with OS use, with women more likely to undergo OS between 2014 to 2017 than 2010 to 2013. Compared with women ≤30 years, women 31 to 40 years [risk ratio (RR), 1.23; 95% CI, 1.18–1.28] and 41 to 50 years (RR, 1.35; 95% CI, 1.29–1.43) were more likely to undergo an OS for sterilization. Among women ≤50 years, the likelihood of hysterectomy and OS increased with increasing age; however, women more than 50 years (RR, 0.83; 95% CI, 0.76–0.90) were less likely to undergo a hysterectomy and OS than women ≤30 years.

Women with a family history of breast/ovarian cancer (RR, 1.38; 95% CI, 1.31–1.46) and genetic susceptibility to breast/ovarian cancer (RR, 2.84; 95% CI, 2.31–3.49) were more likely to undergo an OS for sterilization than women with no family history of and no genetic susceptibility to breast/ovarian cancer. There was no association between family history of breast/ovarian cancer and the likelihood of undergoing a hysterectomy and OS (RR, 1.02; 95% CI, 0.98–1.06), whereas women with a genetic susceptibility to breast/ovarian cancer were less likely to undergo a hysterectomy and OS (RR, 0.70; 95% CI, 0.55–0.90). Underlying comorbidities were not associated with the likelihood of undergoing an OS for sterilization; however, women with 1 or more comorbidities were more likely to undergo a hysterectomy and OS than women with no comorbidities.

OS was more likely to be performed during an abdominal than a laparoscopic hysterectomy (RR, 1.06; 95% CI, 1.02–1.10) and less likely to be performed during a vaginal than a laparoscopic hysterectomy (RR, 0.35; 95% CI, 0.33–0.37). Compared with other surgical indications, women with uterine fibroids were more likely to undergo a hysterectomy and OS (RR, 1.22; 95% CI, 1.19–1.25). OS for sterilization was more likely to be performed as outpatient than inpatient surgery (RR, 3.42; 95% CI, 3.28–3.56), whereas hysterectomy and OS was less likely to be performed as outpatient than inpatient surgery (RR, 0.34; 95% CI, 0.32–0.35). Type of insurance plan was not associated with the likelihood of undergoing an OS for sterilization; however, women enrolled in HMO and other health plans were more likely to undergo a hysterectomy and OS than women with PPO health plans.

OS uptake differed significantly by region and location of residence. Women residing in the Midwest and South were less likely to undergo an OS for sterilization and hysterectomy and OS than women in the Northeast. In contrast, women residing in the West were more likely to undergo a hysterectomy and OS than women in the Northeast. Additionally, women residing in rural locations were less likely to undergo OS procedures than women in urban locations.

Overall results did not differ significantly in multivariable-adjusted models stratified by year of surgery (Supplementary Tables S4 and S5). Of note, differences in OS uptake by region of residence decreased after compared with before guideline release, whereas differences in OS uptake by location of residence increased after compared with before guideline release.

To further examine differences in current practice patterns, we evaluated predictors of OS use in a subset of women with a family history of and genetic susceptibility to breast/ovarian cancer who may be more likely to opt for OS if offered. The median age of OS in women with a family history of and genetic susceptibility to breast/ovarian cancer (OS for sterilization: 38 years, IQR: 34–41; hysterectomy and OS: 46 years, IQR: 40–50) was higher than women with no family history of and no genetic susceptibility to breast/ovarian cancer (OS for sterilization: 34 years, IQR: 30–38; hysterectomy and OS: 43 years, IRR: 39–48). Predictors of OS uptake in this high-risk subset did not differ substantially from the overall analysis. Differences in OS uptake were strongly related to the year of surgery and region and location of residence (Supplementary Table S6). After adjusting for year of surgery, age, type of health insurance plan, place of service, and underlying comorbidities, women with a family history of and genetic susceptibility to breast/ovarian cancer residing in rural and urban areas of the Midwest and South were less likely to undergo an OS for sterilization and hysterectomy and OS than those residing in urban areas of the Northeast (P_{heterogeneity} < 0.001; Table 3).

In sensitivity analysis, including hysterectomy and bilateral salpingo-oophorectomy to the analytic sample did not significantly alter the overall results. With bilateral salpingo-oophorectomies included, the proportion of benign hysterectomy encounters involving OS increased from 1% in 2010 to 24% in 2017. Additionally, in ITS analysis, there was no change in the quarterly age-adjusted rate of hysterectomy and bilateral salpingo-oophorectomy immediately after guideline release (IRR, 1.07; 95% CI, 0.90–1.12) or after compared with before guideline release (IRR, 1.01; 95% CI, 0.95–1.07).

To our knowledge, this is one of the first studies to characterize OS uptake in women undergoing sterilization and benign hysterectomies in inpatient and outpatient settings. Thus, our study offers the first complete look at the dissemination of OS in the United States. We observed that between 2010 and 2017, the age-adjusted rate of OS for sterilization and hysterectomy and OS increased 17.8-fold (95% CI, 16.2–19.5) and 7.6-fold (95% CI, 5.5–10.4), respectively. We also noted substantial variation in OS use, with 94% of OS performed in women ≤50 years, higher uptake of OS for sterilization in women with a family history of and genetic susceptibility to breast/ovarian cancer, and higher uptake of OS in the Northeast and West and urban locations. Additionally, OS use during hysterectomy was higher among women undergoing an abdominal hysterectomy and women with uterine fibroids. Significant variation in OS uptake, notably by region and location of residence, was also observed in analyses limited to high-risk women. These findings could reflect provider and patient knowledge and preferences regarding OS use, insurance coverage, and access to the fallopian tubes during surgery.

The reasons for the rapid increase in OS are likely multifactorial. In addition to the multiple clinical-practice guidelines advocating for the use of OS for ovarian-cancer risk reduction (11, 12, 29), other factors that may have contributed to the rapid adoption of OS include the lack of effective screening tests for ovarian cancer, the high morbidity and mortality associated with ovarian cancer, the high prevalence of gynecologic surgeries currently eligible for OS, and provider enthusiasm. The rapid increase in OS could also be driven by insurance reimbursements, which can significantly influence hospital policies and clinical practice. According to national payment amounts from Medicare, a standard that many other payers use to determine reimbursement, the average payment in 2017 for performing a tubal ligation (CPT 58670) in the office was $371.45, and the payment for performing a bilateral salpingectomy (CPT 58700) was $797.45 (30). The finding that women more than 50 years of age were less likely to undergo a hysterectomy and OS may be attributed to the fact that these women are closer to the age of natural menopause and may therefore be considered for a bilateral oophorectomy instead of OS for ovarian-cancer risk reduction. In addition to the rapid adoption of OS during benign hysterectomy, the decline in hysterectomy rates observed in this study may also be due to the increased use of uterine-sparing procedures for uterine fibroids, abnormal bleeding, and pelvic pain, the most common indications for hysterectomy in reproductive-aged women (31, 32).

To date, only 3 studies have assessed predictors of OS use during a benign hysterectomy in the United States. (17, 18, 33). Prior studies assessing OS use during hysterectomy demonstrated increased uptake of OS in young women, large-volume hospitals, the Northeast, and women with underlying comorbidities. Our study confirmed these findings and identified additional factors driving OS uptake. It is important to note that all these prior studies relied on datasets limited to inpatient procedures only, providing a biased and incomplete view of trends in OS uptake. Approximately 44% of OS procedures in our database were performed in outpatient settings (80% of OS for sterilization and 23% of hysterectomy and OS). Additionally, these studies included narrow age-ranges, lacked data from more recent time periods, had small sample sizes, and lacked information on family history of and genetic susceptibility to breast/ovarian cancer.

In discussing OS uptake, it is also important to highlight the limited evidence to support OS use for ovarian-cancer risk reduction. Only 3 studies have assessed ovarian cancer risk after OS. These studies suggest that OS is associated with a 42% to 64% lower risk of ovarian cancer (8–10). However, none of these studies evaluated OS specifically for ovarian-cancer risk reduction but rather focused on OS for specific indications such as pelvic infection and endometriosis. Additionally, there is limited data on the potential adverse effects associated with OS. A registry-based study found that women undergoing hysterectomy and OS had an increased risk of menopausal symptoms 1 year after surgery compared with hysterectomy alone (RR, 1.33; 95% CI, 1.04–1.69; ref. 34). Studies evaluating the impact of OS on short-term ovarian function have found no detrimental effects (35–37).

Although our analysis benefits from the inclusion of a very large number of privately-insured women, our results may not be generalizable to women who have public insurance plans such as Medicaid and Medicare. However, it is estimated that about 60% of women less than 65 years of age received their health coverage through employment-sponsored health insurance plans in 2017, representing the largest segment of United States healthcare users (38). Additionally, we could not capture individual patient and physician perceptions and attitudes, which likely influenced OS uptake. Furthermore, our dataset lacked several important factors, such as race/ethnicity, education, and body mass index, that likely influenced OS uptake. Finally, our analysis is limited by the nature of claims data, which may involve errors in coding or incomplete documentation for some conditions, such as a family history of and genetic susceptibility to cancer, that could lead to misclassification of covariates.

In conclusion, we found that clinicians in the United States have rapidly adopted OS for ovarian-cancer risk reduction with significant variation in use by demographic and clinical factors, even in women at high risk of ovarian cancer. Given the rapid uptake of OS in average-risk women, prospective studies are needed to evaluate the impact of OS on ovarian cancer incidence and mortality and potential adverse effects, such as premature menopause. This information could be used to inform future clinical practice. Current studies are focused on young BRCA1 carriers. From a public health standpoint, identifying subsets of women who would benefit the most from OS for ovarian-cancer risk reduction and developing initiatives to address barriers to equitable OS uptake is needed.

C.E. Joshu reports grants from American Cancer Society during the conduct of the study. K. Visvanathan reports grants from Breast Cancer Research Foundation; and grants from P30 CA006973 during the conduct of the study. No disclosures were reported by the other authors.

P.S. Karia: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. C.E. Joshu: Resources, supervision, validation, writing–review and editing. K. Visvanathan: Conceptualization, resources, supervision, funding acquisition, validation, writing–review and editing.

This work was supported by the NIH under award numbers: T32 CA009314 (P.S. Karia), T32 CA094061 (P.S. Karia), and P30 CA006973 (C.E. Joshu and K. Visvanathan), and the Breast Cancer Research Foundation (K. Visvanathan).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.