Personal, reproductive, and familial characteristics associated with bilateral oophorectomy in premenopausal women: A population-based case-control study.

The overall study design and the clinical characteristics of the women included in the MOA-2 study were reported elsewhere [ 12 – 14 ]. In brief, MOA-2 included a cohort of premenopausal women who underwent bilateral oophorectomy for a nonmalignant indication, and a corresponding cohort of age-matched referent women. Both cohorts were representative of the geographically defined population of Olmsted County, Minnesota (USA) for the 20-year period 1988–2007. All data collection was through the records-linkage system of the Rochester Epidemiology Project (REP) that has been described elsewhere [ 15 – 18 ]. The women originally sampled to serve as exposed and referent women for the cohort analyses were re-labeled as cases and controls to be used in the case-control analyses reported here. A physician (LGR) and a trained nurse abstractor reviewed the medical records of all women who received a surgical code for unilateral or bilateral oophorectomy. For those women confirmed to have undergone surgery, detailed information about surgical characteristics was abstracted (e.g., indication for the surgery, pathology of the removed ovaries, and pathology of the removed uterus, if applicable). In addition, for both women with oophorectomy and their age-matched controls, the complete medical records were reviewed to collect an extensive series of demographic, social, and reproductive history data, and information about adult life characteristics and family history of cancer. Only characteristics documented before the index date were considered in the case-control analyses. Data were abstracted and recorded using an electronic data entry application. The application provided real time data checks (e.g., range of valid values), and comprehensive data checks were performed regularly during abstraction. To increase the consistency of the data collected, the two data abstractors followed a manual of instructions providing definitions and examples for the characteristics of interest. The manual was updated iteratively during the data collection phase. Information about income was derived from the 2000 United States Census (Summary File 3) [ 19 ]. Each woman was assigned the median household income for the census block group in which she lived at the index date. Cases and controls were compared using conditional logistic regression models for matched pairs, and the associations were measured using odds ratios and 95% confidence intervals. Because sociodemographic characteristics were considered possible confounding variables [ 12 ], analyses for personal, reproductive, and familial characteristics were adjusted for race (white; non-white), education (≤12; 13–16; >16 years), and household income (quartiles: <$42,000; $42,000–56,999; $57,000–71,999; ≥$72,000). We conducted a set of analyses including the complete sample, and three sets of analyses stratified by age at the index date (≤45 years and 46–49 years), by indication (benign ovarian condition and no ovarian indication), and by calendar year (1988–1997 and 1998–2007). We also conducted a set of sensitivity analyses for the overall sample after excluding 165 case-control pairs in which the control had undergone hysterectomy before the index date and 24 pairs in which the case had not undergone hysterectomy as of the index date. All analyses were conducted using SAS v.9.4 (SAS Institute), and tests of statistical significance were conducted at the two-tailed alpha level of 0.05. All study procedures and ethical aspects were approved by the institutional review boards of both Mayo Clinic and Olmsted Medical Center. Because the data collection was historical, women did not need to provide a study-specific informed consent but rather a general consent to use their medical records for research (Minnesota legal requirements) [ 16 , 17 ].

Supplementary Table 1 shows the results of case-control analyses for race, education, and income overall and in strata by age at oophorectomy and by indication for the oophorectomy. Non-white race was significantly less common in cases than controls overall and in all stratified analyses. However, the numbers for non-white women were small. Cases had significantly fewer years of education than controls overall, in the age stratum ≤45 years, and in women with a benign indication (with a dose-effect trend). Finally, cases had significantly lower income in women with age ≤45 years and in women with a benign indication (with a dose-effect trend). Figure 1 provides details about the indications for the oophorectomy and the pathological findings in the ovaries removed. Of the 1,653 pairs of ovaries removed for any indication, 847 (51.2%) were found to be healthy at pathological examination (bolded numbers). Interestingly, 333 women underwent removal of their ovaries and uterus in the absence of any recognized ovarian or uterine condition (shaded boxes in right lower corner). The only indication in these women was excessive bleeding or abdominal pain. Table 1 shows the results of case-control analyses for personal characteristics in the overall sample after adjusting for race, education, and income. Infertility, higher weight (with a dose-effect trend), and higher body mass index (BMI; with a dose-effect trend) were more common in women who underwent bilateral oophorectomy. By contrast, older age at menarche (with a dose-effect trend), any use of oral contraceptives, longer use of oral contraceptives (with a dose-effect trend), history of breast feeding, and fibrocystic breast disease were less common in women who underwent bilateral oophorectomy ( Table 1 ; all ages, all indications). The results were similar in a set of sensitivity analyses in which we removed controls who had undergone hysterectomy and cases who had not undergone hysterectomy as of the index date (data not shown). Table 2 shows case-control analyses for bilateral oophorectomy stratified by age at the time of oophorectomy (or index date; all indications). The median age at oophorectomy was 41 years (IQR, 38–44) for the ≤45 year stratum and 47 years (IQR, 47–48) for the 46–49 year stratum. The associations were more pronounced for the younger stratum, and some characteristics were significantly associated with bilateral oophorectomy only in the younger stratum. In particular, any oral contraceptive use, older age at menarche (with a dose-effect trend), ≥3 pregnancies (with a dose-effect trend), ≥3 live births (with a dose-effect trend), any induced abortion, breast feeding, and fibrocystic breast disease were less common, whereas infertility and smoking (with a dose-effect trend) were more common in cases than in controls only in the younger age stratum. By contrast, higher weight and higher BMI (with a dose-effect trend) were more common and longer use of oral contraceptives were less common in both age strata. Figure 2 shows the dose-effect trend analyses in women who underwent bilateral oophorectomy before age 46 years (only significant trends are shown). Table 3 shows case-control analyses for bilateral oophorectomy stratified by indication for the oophorectomy (all ages). Of the 675 women with a benign indication, 654 (96.9%) had a concurrent or preceding hysterectomy. Of the 978 women without an ovarian indication, 975 (99.7%) had a concurrent or preceding hysterectomy. Among 1,653 referent women, only 165 (10%) had concurrent or preceding hysterectomy. The women without an ovarian indication were similarly distributed in the younger (52.4%) and older stratum (47.7%) by age of bilateral oophorectomy. By contrast, women with an ovarian indication were more commonly in the younger age stratum (76.9%) than in the older age stratum (23.1%). Some results were different or in opposite directions in women who had a benign ovarian indication compared to women who did not have an ovarian indication for bilateral oophorectomy. Having ≥3 pregnancies was less common in women who underwent oophorectomy with a benign ovarian indication but more common in women without an ovarian indication. Infertility was more common in the benign indication stratum but not in the no indication stratum. Older age at menarche, history of breast feeding, and fibrocystic breast disease were less common in the benign condition stratum but not in the no indication stratum. Smoking and longer duration of smoking were more common in the no indication stratum but not in the benign indication stratum. By contrast, the associations for any use and longer use of oral contraceptives (less common) and higher weight (more common) were similar in the two indication strata ( Table 3 ). Figure 3 shows the dose-effects trend analyses for women with a benign ovarian indication (all ages; only significant trends). Figure 4 shows the dose-effect trend analyses for women without an ovarian indication (all ages; only significant trends). Supplementary Table 2 shows case-control analyses for bilateral oophorectomy stratified by calendar year (all ages and all indications). The results were not noticeably different in the two strata. Table 4 shows the case-control analyses for family history of cancer stratified by indication for the oophorectomy (all ages). Women with a benign ovarian indication had an increased frequency of reported family history of ovarian (first-degree relatives) and colorectal cancer (any relative). Women without a benign ovarian indication had an increased frequency of reported family history of uterine cancer (first-degree relatives), and other cancers (first-degree relatives).

Our study identified a number of personal, reproductive, and familial characteristics that may have influenced the decision to undergo a surgery resulting in bilateral oophorectomy. These characteristics have not been previously investigated in large epidemiologic studies. Infertility, higher weight, and higher body mass index were more common, whereas older age at menarche, use of oral contraceptives, history of breast feeding, and fibrocystic breast disease were less common in cases than controls. The associations were more extreme for bilateral oophorectomies performed at ages ≤45 years compared to 46–49 years, and were different in women with or without a benign ovarian indication. Family histories of ovarian and colorectal cancers were more common in women who underwent bilateral oophorectomy with a benign ovarian indication compared with referent women, whereas family histories of uterine cancer or other cancers were more common in women who underwent bilateral oophorectomy without a benign ovarian indication compared with referent women. Our findings are consistent with the findings from some previous studies. A study that compared women who underwent hysterectomy with bilateral oophorectomy to women who underwent hysterectomy alone in the state of New York reported an association with family history of breast or ovarian cancer, and with a personal history of breast cancer, ovarian cyst, or endometriosis. In addition, both race (lower rate in African American and Hispanic women) and insurance status were associated with the performance of bilateral oophorectomy [ 20 ]. Another study conducted in Michigan showed that family history of cancer and personal history of endometrial hyperplasia, endometriosis, and cervical dysplasia were associated with bilateral oophorectomy [ 21 ]. A study in California showed higher risk of bilateral oophorectomy without an ovarian indication in Hispanic or African American women. The study also reported higher risk in urban hospitals and in hospitals with low California Medicaid utilization rates [ 2 ]. Some of the associations observed in our study for bilateral oophorectomy were consistent with the associations observed in a study of ovarian cancer. A European study showed an association between oral contraceptive use and greater number of full-term pregnancies with lower risk of ovarian cancer [ 22 ]. In our study, oral contraceptive use was associated with reduced risk of bilateral oophorectomy, and a greater number of pregnancies was associated with reduced risk of bilateral oophorectomy performed at age ≤45 years, or performed for a benign ovarian indication. These findings provide additional evidence that oral contraceptive use may have positive long-term effects. A total of 675 women (40.8%) who underwent bilateral oophorectomy had a benign ovarian condition listed as the indication for the surgery ( Figure 1 ) [ 12 ]. For these women, the associations that we observed might in part be interpreted as risk or protective factors for the specific ovarian conditions that prompted the surgery (benign tumor, cyst, endometriosis, or other benign ovarian condition). For many of these women, the removal of both ovaries was not needed to control the benign ovarian conditions, and 25% of these women had normal ovaries at pathology. More conservative practices may be considered for these women in the future. A total of 978 women (59.2%) who underwent bilateral oophorectomy did not have any specified ovarian indication ( Figure 1 ) [ 12 ]. Women without a benign ovarian condition were historically considered to have “prophylactic”, “elective”, or “incidental” bilateral oophorectomy. In most of these women, the presumed healthy ovaries were removed at the time of a hysterectomy that was performed for another gynecological indication. Therefore, the risk and protective factors that we observed might in part relate to the uterine conditions or symptoms that prompted the hysterectomy (most commonly, excessive bleeding, pelvic pain, fibroids, or prolapse), or to intraoperative-events, surgeons’ preferences, and women’s preferences and past experiences. For example, higher body mass index is associated with an increased risk of excessive bleeding that may be an indication for hysterectomy, in turn possibly leading to a decision to also remove the ovaries at the same time. Only 46 women (4.7%) without an ovarian indication were recorded to have a positive family history for ovarian cancer in first-degree relatives (mothers, sisters, or daughters), and family history was not recorded by the surgeon as an indication for the oophorectomy in these women. Similarly, a total of 36 controls (3.7%) were also recorded to have a positive family history. Therefore, family history of ovarian cancer was not significantly different in cases and controls. The associations with family history of uterine cancer and other cancers suggest that a concern of women about the risk of cancer in general may have played a role in electing to remove their presumed healthy ovaries, even in the absence of a documented increased risk of ovarian cancer in their families. Based on current knowledge and guidelines, these 978 women with no ovarian condition and no documented increased risk of malignancy had no clear indication for removing their presumed healthy ovaries [ 13 , 14 ]. The historical practice of bilateral oophorectomy for the prophylaxis of ovarian or breast cancer even in women at average risk of ovarian cancer, and the lack of awareness of the multiple long-term sequelae of bilateral oophorectomy, might have led the gynecologists to offer the oophorectomy as an option. However, women’s preferences and previous life experiences related to sexuality and reproduction may also have played a role in the decision. For example, women who had a higher number of pregnancies or live births were more likely to undergo the removal of presumed healthy ovaries, even though contraceptive methods were widely available during the study period. As shown in our previous study [ 10 , 11 ], some women had undergone prior abdominal surgeries (e.g., appendectomy, tubal resection, or Cesarean section), and requested the oophorectomy in the belief that it might definitively eliminate pain or other distress. These women may have been unaware or in denial of the possible psychological and emotional origins of their pain and distress, and the gynecologists may have underestimated the possible long-term harmful consequences of bilateral oophorectomy [ 10 , 11 ]. Our case-control study has a number of strengths. First, details about the surgical procedure, prior risk factors, and conditions present at the index date were abstracted from the medical records included in a records-linkage system without direct involvement of the women included in the study (recall bias was minimized). Second, the non-participation was minimized because the data collection was historical and women did not need to provide a study-specific informed consent, but only a general research authorization (as per Minnesota legal requirements) [ 16 , 17 ]. Third, control women comprised an unrestricted sample representative of the general population rather than women who underwent hysterectomy with ovarian conservation. We elected not to use hysterectomy with ovarian conservation as a control group because hysterectomy itself may be associated with similar risk or protective factors [ 23 , 24 ]. Finally, the population studied included all race and ethnicity groups regardless of socioeconomic status, insurance status, and health care delivery setting [ 17 ]. First, because the characteristics considered were abstracted from medical records in a records-linkage system, absence of information for some characteristics was considered evidence that the characteristic was not present. On the other hand, we could not impute the value for some other missing characteristics, such as age at menarche, length of oral contraceptive use, and history of breast feeding. These missing values may have introduced a bias because, in general, controls had more missing values than cases ( Table 1 , footnote a). Second, the oophorectomies took place over 20 years, from 1988 through 2007, and surgical practices and estrogen use have changed over time. However, we conducted a set of secondary analyses stratified by decade of the surgery (1988–1997 vs. 1998–2007) and found similar associations for most of the characteristics. Third, because income was derived from census data for only one point in time and at the census block group level, some misclassification of income may have occurred. However, the year 2000 was approximately the central year of the study period, and the methods used were identical for women with and without bilateral oophorectomy (non-differential misclassification). Finally, our study focused on a single geographically defined US population, and the observed associations may differ in other populations. However, the demographic and socioeconomic characteristics of our population are similar to those of the upper Midwest and of a large segment of the entire United States population [ 16 , 18 ]. Replication of this study in other populations in the USA and worldwide will allow for useful comparisons.

We identified a number of personal, reproductive, and familial characteristics that were associated with bilateral oophorectomy over a 20-year period. Some of these characteristics (e.g., family history of cancer) may have influenced the decision of the women to undergo and of the gynecologist to perform the bilateral oophorectomy. Some other characteristics (e.g., age at menarche or breast feeding) may not have been considered in the decision making. Understanding the characteristics that were associated with the practice in the past is important for decision-making about bilateral oophorectomy in the future. Mounting research evidence suggests that bilateral oophorectomy should be limited to the treatment of ovarian malignancy or to the prevention of cancer when women carry a genetic variant known to increase their risk of cancer significantly (e.g., variants of the BRCA-1 or BRCA-2 genes) [ 9 , 13 , 14 ]. For other women, the risks of endocrine disruption appear to exceed the benefits, as discussed in detail in our previous publications from MOA-2 [ 13 , 14 ].

Bilateral oophorectomy continues to be performed in isolation or more commonly with hysterectomy in women before the age of natural menopause [ 1 , 2 ]. In the majority of cases, these surgeries are performed to treat nonmalignant gynecological symptoms or conditions. In addition, a large number of bilateral oophorectomies are performed at the time of a hysterectomy without a specific ovarian indication. For example, recent data from California suggest that approximately 38% of women undergo bilateral oophorectomy at the time of a hysterectomy in the absence of a documented ovarian condition [ 2 ]. This practice reflects the unresolved controversy about the advantages and disadvantages of removing healthy ovaries in premenopausal women for the prevention of ovarian and breast cancer [ 3 – 9 ]. In addition to a family history of ovarian cancer, intraoperative events, surgeons’ preferences, women’s preferences and past experiences, and social, reproductive, and familial factors may be involved in the decision to remove healthy ovaries [ 10 , 11 ]. We recently reported on the association between adverse childhood or adult experiences and the risk of bilateral oophorectomy [ 10 , 11 ]. However, other characteristics associated with bilateral oophorectomy have not been investigated extensively. We conducted a case-control study to investigate the personal, reproductive, and familial characteristics associated with bilateral oophorectomy in the M ayo Clinic Cohort Study of O ophorectomy and A ging 2 (MOA-2). We report a series of case-control analyses contrasting premenopausal women who underwent bilateral oophorectomy to their respective age-matched controls in a geographically defined US population with a special focus on women who underwent oophorectomy at younger ages and on women who did not have a specified ovarian indication.