Background
Although ovarian conservation at hysterectomy for benign gynecologic disease has demonstrated mortality benefit in young patients and this benefit may be sustained up to age 65 years, there is a scarcity of data regarding ovarian conservation in those with a diagnosis of endometrial hyperplasia, a premalignant uterine condition.
Objective
This study aimed to examine patient, hospital, treatment, and histology characteristics related to ovarian conservation at the time of inpatient hysterectomy for endometrial hyperplasia.
Study Design
The Healthcare Cost and Utilization Project’s National Inpatient Sample was retrospectively queried to examine patients aged ≤65 years with endometrial hyperplasia who had inpatient hysterectomy from January 2016 to December 2019. The exclusion criteria included concurrent gynecologic malignancy, adnexal pathology, and lymphadenectomy. Cases were grouped by adnexal surgery status (ovarian conservation or oophorectomy). A multivariable binary logistic regression model was used to identify independent characteristics for ovarian conservation. A classification tree was constructed with recursive partitioning analysis to examine utilization patterns of ovarian conservation.
Results
Overall, 3105 patients (31.1%) underwent ovarian conservation at hysterectomy among 9975 patients. The utilization of ovarian conservation decreased gradually until age 45 years and then markedly decreased by age 52 years (63.3%–15.3%; P <.001). In a multivariable analysis, younger age, non-White, urban nonteaching centers, and vaginal hysterectomy were associated with increased utilization of ovarian conservation, whereas endometrial hyperplasia with atypia, obesity, comorbidity, large bed capacity centers, and Midwest and South regions were associated with decreased utilization of ovarian conservation (all, P <.05). A classification tree identified 17 utilization patterns for ovarian conservation, ranging from 7.8% to 100.0% (absolute rate difference, 92.2%).
Conclusion
The utilization of ovarian conservation at the time of inpatient hysterectomy in patients undergoing surgical management for endometrial hyperplasia started decreasing in their mid-40s and seemed to occur earlier than in benign hysterectomy. There was substantial variability in ovarian conservation at the time of hysterectomy for endometrial hyperplasia based on patient, hospital, surgical, and histology factors, suggesting the possible benefit of clinical practice guidelines for ovarian conservation in this population.
Why was this study conducted?
There is a scarcity of data regarding ovarian conservation at hysterectomy for endometrial hyperplasia.
Key findings
In an analysis of the National Inpatient Sample (n=9975), utilization of ovarian conservation decreased gradually until age 45 years and then markedly decreased by age 52 years (63.3%–15.3%). Younger age, non-White, urban nonteaching centers, and vaginal hysterectomy were associated with increased utilization of ovarian conservation. Endometrial hyperplasia with atypia, obesity, comorbidity, large bed capacity centers, and Midwest and South regions were associated with decreased utilization of ovarian conservation. There were 17 utilization patterns for ovarian conservation, ranging from 7.8% to 100% (absolute rate difference, 92.2%).
What does this add to what is known?
A decrease in the utilization of ovarian conservation at inpatient hysterectomy for endometrial hyperplasia occurred at a younger age than expected; substantial variability for ovarian conservation suggested the possible benefit of developing clinical practice guidelines.
Introduction
Endometrial hyperplasia, a precursor to endometrial cancer, is characterized by the disorganized proliferation of endometrial glands that occurs in response to unopposed exposure to estrogen. Although endometrial cancer is known to be the most common gynecologic cancer in the United States, the incidence of endometrial hyperplasia is estimated to be a few times higher than that of endometrial cancer and is approximately 133 per 100,000 woman-years.
The recommended treatment in those with endometrial hyperplasia who have completed childbearing and are medically fit for surgery is hysterectomy. The Society of Gynecologic Oncology (SGO) in 2012 states that oophorectomy is not required at the time of hysterectomy, and the American College of Obstetrics and Gynecology (ACOG) in 2015 suggests that ovarian conservation deserves serious consideration, especially in the pre- and perimenopausal population, given the potential risks associated with surgical menopause at a young age. ,
Several studies have demonstrated that ovarian conservation in young women undergoing benign hysterectomy has a mortality benefit, and this benefit may be possibly sustained up to the age of 65 years. Endometrial hyperplasia falls on the spectrum between benign and malignant, which presents a clinical challenge and adds nuance to the decision regarding ovarian conservation in this cohort. A 2015 study demonstrated that women aged <51 years with normal ovaries were twice as likely to have oophorectomy at hysterectomy for the treatment of endometrial hyperplasia, suggesting hesitancy regarding ovarian conservation in this population. These findings may reflect a heightened concern related to the potential for occult endometrial cancer. In case of endometrial cancer, oophorectomy at total hysterectomy is the standard surgical procedure. ,
Given the absence of more concrete guidelines regarding ovarian conservation at hysterectomy for endometrial hyperplasia per SGO and ACOG as above, , the decision is left to the provider, which poses a challenge as national-level data regarding these practices is unknown. This study aimed to examine patient, hospital, and treatment characteristics related to ovarian conservation at the time of inpatient hysterectomy for endometrial hyperplasia.
Materials and Methods
Data source
The National Inpatient Sample (NIS) was queried for this study. The NIS program is a population-based all-payer database that catalogs information regarding inpatient utilization, cost, quality, and outcomes by randomly sampling 20% of admission records in each hospital, with the weighted sample representing >90% of the US population. , This is a publicly available and deidentified program that was developed as a part of the Healthcare Cost and Utilization Project and is sponsored by the Agency for Healthcare Research and Quality.
Ethics statement
The University of Southern California Institutional Review Board deemed this study exempt because of the use of publicly available deidentified data.
Study cohort
This was a retrospective cohort study examining the NIS program from January 2016 to December 2019. This study point was chosen because of the introduction of the International Classification of Disease, 10th Revision (ICD-10), codes into the NIS program. Women aged ≤65 years with endometrial hyperplasia who had hysterectomy were eligible. This age cutoff was chosen per the previous analysis demonstrating the possible benefit for all-cause mortality with ovarian conservation at hysterectomy. The case identification for endometrial hyperplasia was based on the ICD-10 diagnosis code of N85.0, which was consistent throughout the study.
Patients were excluded if they did not have a hysterectomy or no information for surgery, were aged >65 years, or had adnexal pathology and gynecologic malignancy with uterine, cervical, or ovarian cancer. Furthermore, patients who had surgical nodal evaluation were excluded. These exclusions were selected to ensure assessment of the effect of endometrial hyperplasia on adnexal surgery.
Outcome measures
The main outcome measure was patient, surgical, and hospital characteristics related to ovarian conservation at the time of inpatient hysterectomy. The co-outcomes included temporal trends of ovarian conservation at hysterectomy for endometrial hyperplasia over time and per patient age.
Patients who have an ICD-10 code for bilateral adnexectomy at the time of surgery were defined as the oophorectomy group, whereas those without the code were defined as the ovarian conservation group. This strategy for exposure assignment followed the same definition as the previous analysis.
Study covariates
Baseline information on patient characteristics, surgical treatment, and facility parameters were abstracted from the NIS program for eligible participants. This study followed the same ICD-10 codes for the extraction of information that was unchanged during the study.
- 1.
Abstracted patient characteristics included age at surgery (<45, 45–52, or >52 years) determined per trend analysis, year of surgery (2016, 2017, 2018, and 2019), race and ethnicity (White, Black, Hispanic, Asian, or others) determined per the NIS program, primary expected payer (Medicare, Medicaid, private, including, HMO, and others), median household income (per quartile), Charlson Comorbidity Index (0, 1–2, or ≥3) calculated according to the previous study, , and obesity (yes or no).
- 2.
Facility parameters included hospital bed capacity (small, middle, and large), location and teaching status (rural, urban nonteaching, and urban teaching), and regional area (Northeast, Midwest, South, and West).
- 3.
Study covariates for surgical treatment included hysterectomy modality (abdominal, laparoscopic, laparoscopy-assisted vaginal, and vaginal) and use of robotic-assisted surgery (yes or no).
Statistical consideration
The utilization rate of ovarian conservation at hysterectomy was calculated for each demographic factor. Differences in continuous, ordinal, or categorical variables were assessed with the Mann-Whitney U test, Fisher exact test, and chi-square test, as appropriate, in univariable analysis. Temporal trends of ovarian conservation over time were assessed with the Cochran-Armitage trend test.
In the multivariable analysis, a binary logistic regression model was fitted to determine independent characteristics associated with ovarian conservation at hysterectomy. Initial covariate selection was set at a P <.05 in the univariable analysis. Conditional backward selection was performed with the stopping rule of P <.05 in the final model. Effect size for ovarian conservation compared with that for oophorectomy was expressed with an adjusted odds ratio (aOR) with a corresponding 95% confidence interval (CI). Multicollinearity was assessed among the covariates in the modeling.
Utilization patterns of ovarian conservation at hysterectomy for endometrial hyperplasia were assessed by constructing a classification tree with recursive partitioning analysis. All independent factors for ovarian conservation were entered in the modeling, and the chi-square automatic interaction detector method was used with a stopping rule of a maximum of 3 layers. In the determined terminal nodes, the utilization rate of ovarian conservation was computed in each node.
In sensitivity analysis, the measured outcomes were assessed per the subtypes of endometrial hyperplasia based on the presence of atypia. Specifically, (1) endometrial hyperplasia with atypia (ICD-10 diagnosis code of N85.02) and (2) endometrial hyperplasia without atypia (ICD-10 diagnosis code of N85.01) were examined. In each subtype, independent characteristics of ovarian conservation and utilization pattern were assessed.
Linear segmented regression with log transformation was used to assess the temporal trend of ovarian conservation at hysterectomy based on patient age. The identified inflection points were determined in an automated fashion in the analysis that was used for the age clustering. In each segment, statistical significance of the slope was determined.
The weighted values for national estimates provided by the NIS program were utilized, and statistical interpretation was based on a 2-tailed hypothesis. A P value of <.05 was considered statistically significant. IBM SPSS Statistics (version 28.0; IBM, SPSS Inc, Armonk, NY) was used for all analysis. The Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines were consulted to summarize the performance of the cohort study.
Results
Cohort-level characteristics
Overall, 9975 patients aged ≤65 years underwent hysterectomy for endometrial hyperplasia during the study. The cohort-level demographics are displayed in Table 1 . The median age was 52 (interquartile range [IQR], 46–58) years. Most patients were White (57.9%), privately insured (64.9%), nonobese (56.3%), had no comorbidity (61.9%), and underwent abdominal hysterectomy (56.1%) at large (52.3%) urban teaching (70.1%) centers.
| Characteristic | n (%) a | Ovarian conservation (%) b | P value |
|---|---|---|---|
| n | 9975 (100) | 31.1 | |
| Age (y) | 52 (46–58) | 45 (40–50) | <.001 |
| <45 | 2120 (21.3) | 68.2 | |
| 45–52 | 3080 (30.9) | 34.7 | |
| >52 | 4775 (47.9) | 12.4 | |
| Year | .395 | ||
| 2016 | 3480 (34.9) | 31.6 | |
| 2017 | 2615 (26.2) | 30.0 | |
| 2018 | 2055 (20.6) | 32.1 | |
| 2019 | 1825 (18.3) | 30.7 | |
| Race and ethnicity | <.001 | ||
| White | 5780 (57.9) | 25.3 | |
| Black | 1450 (14.5) | 31.7 | |
| Hispanic | 1530 (15.3) | 45.8 | |
| Asian | 485 (4.9) | 36.1 | |
| Others | 730 (7.3) | 42.5 | |
| Primary expected payer | <.001 | ||
| Medicare | 1005 (10.1) | 17.9 | |
| Medicaid | 1955 (19.6) | 41.4 | |
| Private including HMO | 6475 (64.9) | 29.7 | |
| Others | 540 (5.4) | 35.2 | |
| Median household income | .004 | ||
| QT1 (lowest) | 2800 (28.1) | 33.2 | |
| QT2 | 2640 (26.5) | 30.9 | |
| QT3 | 2395 (24.0) | 31.3 | |
| QT4 (highest) | 1980 (19.8) | 28.0 | |
| Unknown | 160 (1.6) | 34.4 | |
| Charlson Comorbidity Index | <.001 | ||
| 0 | 6175 (61.9) | 34.9 | |
| 1 | 2040 (20.5) | 25.7 | |
| 2 | 945 (9.5) | 26.5 | |
| ≥3 | 815 (8.2) | 21.5 | |
| Obesity | <.001 | ||
| No | 5615 (56.3) | 34.3 | |
| Yes | 4360 (43.7) | 27.1 | |
| Hp bed capacity | .019 | ||
| Small | 1685 (16.9) | 32.6 | |
| Middle | 3070 (30.8) | 32.4 | |
| Large | 5220 (52.3) | 29.9 | |
| Hp setting and location | <.001 | ||
| Rural | 1050 (10.5) | 33.3 | |
| Urban nonteaching | 1930 (19.3) | 34.7 | |
| Urban teaching | 6995 (70.1) | 29.8 | |
| Hp region | <.001 | ||
| Northeast | 2005 (20.1) | 32.4 | |
| Midwest | 2110 (21.2) | 26.1 | |
| South | 3630 (36.4) | 28.8 | |
| West | 2230 (22.4) | 38.6 | |
| Hysterectomy type | <.001 | ||
| Abdominal | 5600 (56.1) | 29.3 | |
| Laparoscopic c | 2040 (20.5) | 26.5 | |
| Laparoscopy-assisted vaginal c | 1860 (18.6) | 30.9 | |
| Vaginal | 475 (4.8) | 73.7 | |
| Robotic assisted | <.001 | ||
| No | 8030 (80.5) | 33.1 | |
| Yes | 1945 (19.5) | 22.9 | |
| Endometrial hyperplasia | <.001 | ||
| Nonatypia | 3480 (34.9) | 31.8 | |
| Atypia | 3720 (37.3) | 23.8 | |
| NOS | 2775 (27.8) | 40.2 |
a Data are presented as number (percentage) or median (interquartile range)
b Utilization of ovarian conservation at hysterectomy is shown as percentage per row level
Trends of ovarian conservation
Overall, 3105 patients (31.1%) had ovarian conservation at the time of surgery. The median age for those who had ovarian conservation at surgery was 45 years (IQR, 33–42).
A time trend of ovarian conservation examined in a yearly fashion is shown in Figure 1 . The rate of ovarian conservation at hysterectomy was unchanged during the study from 31.6% to 30.7% ( P trend=.807). In addition, the utilization of ovarian conservation at hysterectomy was unchanged for endometrial hyperplasia with atypia (n=3720) from 23.2% to 23.9% ( P trend=.772) and endometrial hyperplasia without atypia (n=3480) from 30.8% to 33.8% ( P trend=.130) ( Figure 1 ).
An age-specific trend of ovarian conservation at hysterectomy is shown in Figure 2 . Ovarian conservation rates decreased gradually until the age of 45 years, ranging from 81.0% to 63.3% ( P trend=.007), after which time the rate sharply and significantly decreased by 15.6% (95% CI, 20.2–10.8; P trend<.001) in 1-year age increments from 63.3% at age 45 years to 15.3% at age 52 years. This was followed by a continued decrease from 15.3% to 5.5% between ages 52 and 65 years ( P trend=.004).
When the cohort was stratified on the basis of the presence of atypia ( Supplemental Figure 1 ), the inflection point of decreasing ovarian conservation occurred earlier in the endometrial hyperplasia with the atypia group than in the endometrial hyperplasia without atypia group (43 and 45 years, respectively). Patients in the atypia group were more likely to be older and obese and have comorbidities than those in the nonatypia group ( Supplemental Table 1 ).
Characteristics related to ovarian conservation
The results of univariable analysis examining the differences in the baseline characteristics for ovarian conservation at hysterectomy are shown in Table 1 . All measured study covariates except for the year factor were significantly associated with ovarian conservation (all, P <.05).
In multivariable analysis ( Table 2 ), (1) patient characteristic with younger age at surgery and Black or Hispanic race and ethnicity, (2) hospital characteristic with urban nonteaching centers, and (3) surgical characteristic with vaginal hysterectomy were independently associated with increased utilization of ovarian conservation (all, P <.05). On the contrary, (1) patient characteristic with obesity and increased number of comorbidities and (2) hospital characteristic with large bed capacity located in the Midwest or South were associated with decreased utilization of ovarian conservation (all, P <.05) ( Table 2 ).
| Characteristic | aOR (95% CI) | P value |
|---|---|---|
| Age (y) | <.001 a | |
| <45 | 17.9 (15.6–20.6) | <.001 |
| 45–52 | 3.94 (3.47–4.47) | <.001 |
| >52 | 1 | |
| Race and ethnicity | <.001 a | |
| White | 1 | |
| Black | 1.90 (1.63–2.22) | <.001 |
| Hispanic | 1.61 (1.39–1.86) | <.001 |
| Asian | 1.23 (0.98–1.55) | .078 |
| Others | 1.98 (1.64–2.40) | <.001 |
| Charlson Comorbidity Index | .001 a | |
| 0 | 1 | |
| 1 | 0.90 (0.78–1.03) | .125 |
| 2 | 0.88 (0.73–1.06) | .172 |
| ≥3 | 0.66 (0.53–0.81) | <.001 |
| Obesity | ||
| No | 1 | |
| Yes | 0.88 (0.78–0.99) | .029 |
| Hp bed capacity | .004 a | |
| Small | 1 | |
| Middle | 0.95 (0.82–1.10) | .491 |
| Large | 0.81 (0.70–0.94) | .005 |
| Hp setting and location | .014 a | |
| Rural | 1.09 (0.92–1.30) | .224 |
| Urban nonteaching | 1.22 (1.07–1.39) | .004 |
| Urban teaching | 1 | |
| Hp region | <.001 a | |
| Northeast | 1 | |
| Midwest | 0.70 (0.60–0.82) | <.001 |
| South | 0.55 (0.47–0.63) | <.001 |
| West | 0.96 (0.82–1.13) | .612 |
| Hysterectomy type | <.001 a | |
| Abdominal | 0.89 (0.77–1.02) | .082 |
| Laparoscopic b | 1 | |
| Laparoscopy-assisted vaginal c | 1.10 (0.93–1.30) | .250 |
| Vaginal | 10.9 (8.41–14.1) | <.001 |
| Endometrial hyperplasia | <.001 a | |
| No atypia | 1 | |
| Atypia | 0.75 (0.66–0.85) | <.001 |
| NOS | 1.40 (1.23–1.58) | <.001 |
In addition, patients with endometrial hyperplasia with atypia were 25% less likely to undergo ovarian conservation at hysterectomy than those without atypia in multivariable analysis (23.8% vs 31.8%, aOR, 0.75; 95% CI, 0.66–0.85, P <.001) ( Table 2 ). Sensitivity analyses demonstrated largely similar trends among those with endometrial hyperplasia with atypia and without atypia ( Supplemental Tables 2 and 3 ).
Utilization patterns of ovarian conservation
The utilization patterns of ovarian conservation at hysterectomy were assessed ( Figure 3 and Table 3 ). A classification tree model identified 17 unique patterns of ovarian conservation based on patient factors (age, comorbidity, and obesity), treatment facility factors (hospital region, capacity, and teaching status), surgical factors (hysterectomy mode), and endometrial hyperplasia type. Patient age at hysterectomy was the first indicator for allocation in the initial layer, followed by hysterectomy type and comorbidities in the second layer.
