Uterine myomas (leiomyomas) are very common in reproductive-aged women with cumulative incidence reaching up to 70% in white women and 80% in black women by age 50 years.1 Leiomyomas impose significant economic burden with an annual medical cost for women with leiomyomas reaching $8,463 higher than those without (in 2005 dollars).2 The overall economic burden of leiomyomas in the United States ranges from $5.9 to $34.4 billion (in 2010 dollars)3,4 with such a wide range indicating difficulty in measuring leiomyoma-related costs, particularly indirect costs (eg, lost wages).
Leiomyomas can cause debilitating symptoms including heavy menstrual bleeding, bulk symptoms such as pelvic pain, urinary problems, and constipation.5–7 Hysterectomy, which eliminates most leiomyoma symptoms and the possibility of formation of new leiomyomas, is the mainstay of treatment for leiomyomas. Up to 50% of all hysterectomies performed in the United States have a discharge diagnosis of leiomyomas.8–10 However, hysterectomy is not acceptable to many women, especially to those desiring fertility. Alternatives to hysterectomy for treating leiomyoma-related bulk symptoms include the following uterine-sparing options: myomectomy, uterine artery embolization, and magnetic resonance-guided, focused ultrasound surgery. However, the real-world evidence on the comparative effectiveness between alternative procedural treatments for leiomyoma-related bulk symptoms is sparse.11 Specifically, extant evidence demonstrates substantial variation in the reintervention rates and reproductive outcomes between different leiomyoma treatment procedures.12–19
Therefore, the primary objective of this study was to compare the risk of reintervention among different uterine-sparing leiomyoma procedures. The secondary objectives were to compare long-term health care utilization and reproductive outcomes between leiomyoma procedures.
MATERIALS AND METHODS
We conducted a retrospective analysis of administrative claims data from a large U.S. commercial insurance database, Optum Labs Data Warehouse, which includes both privately insured and Medicare Advantage enrollees throughout the United States. The Optum Labs Data Warehouse contains longitudinal health information for more than 100 million enrollees over the past 20 years, from geographically diverse regions across the United States, with the greatest representation from the South and Midwest.20 Health care claims data are generated as part of usual clinical practice in which care providers submit claims to payer organizations for reimbursement in standardized format (eg, UB-04 form or Centers for Medicare & Medicaid Services 1500 form). Payer organizations conduct internal checks to validate consistency and accuracy of the claims submitted. Given that the Optum Labs Data Warehouse captures data from one of the largest commercial insurance companies in the United States, the population of women with leiomyoma drawn from this database will resemble the commercially insured leiomyoma patient population in the United States. The health plans included in the database provide claims for professional (eg, physician), facility (eg, hospital), and outpatient prescription medication services. Medical (professional, facility) claims include International Classification of Diseases, 9th Revision, Clinical Modification diagnosis codes, International Classification of Diseases, 9th Revision procedure codes, Current Procedural Terminology, Version 4 procedure codes, and Healthcare Common Procedure Coding System procedure codes. Study data were accessed using techniques compliant with the Health Insurance Portability and Accountability Act of 1996. Because this study involved analysis of pre-existing, deidentified data, it was exempt from institutional review board approval.
We identified women aged 18–54 years who had diagnoses of leiomyoma and bulk symptoms and who underwent either a uterine-sparing procedure that was expected to reduce bulk or size-related symptoms (eg, myomectomy, uterine artery embolization, and magnetic resonance-guided focused ultrasound surgery) or hysterectomy between January 1, 2000, and December 31, 2013 (Appendix 1, available online at http://links.lww.com/AOG/B21, for relevant diagnosis codes, and Appendix 2, also available online at http://links.lww.com/AOG/B21, for relevant procedure codes). Bulk symptoms include pelvic pain or pressure, urinary problems, nocturia, constipation, and dyspareunia. Women’s first procedural therapy after the leiomyoma and bulk diagnoses during the study period was defined as the “index procedure,” which also defined their study cohorts. For example, “myomectomy cohort” refers to women who received myomectomy as their index procedure. The date of the index procedure was defined as the “index date.” The 2-week period before the index date was considered as the “preoperative period.” The 12-month period before the preoperative period was defined as the “baseline.” All women included in the study were required to have continuous health insurance during the preoperative and baseline periods and also for a minimum of 6 months after the index date. Beyond the minimum required enrollment of 6 months in the follow-up, women in the study were followed until they disenrolled from the health plans or the end of the study period (June 30, 2014). We excluded women who received any procedural therapies for leiomyoma before the index date as well as those with uterine cancer (Appendix 1, http://links.lww.com/AOG/B21, for codes).
The two primary outcomes of the study were 1) whether women undergoing any of the uterine-sparing leiomyoma procedures (myomectomy, uterine artery embolization, and magnetic resonance-guided, focused ultrasound surgery) as the index procedure underwent any subsequent leiomyoma procedure during follow-up (reintervention), and 2) whether they underwent other surgical procedures for treating potential complications of the index procedure. In defining the first primary outcome, if a woman had a subsequent claim for the same procedure as the index procedure within 30 days, we considered the latter as part of the initial treatment rather than a separate procedure or reintervention. For example, if a woman underwent myomectomy as the index procedure, and a second myomectomy was observed within a month, we considered the latter procedure as being related to the index procedure and did not count it as reintervention. However, if the woman underwent a uterine artery embolization or hysterectomy within 30 days after the index myomectomy, we considered it a reintervention. We also examined whether women eventually received a hysterectomy at any point during follow-up. The second outcome of interest was subsequent surgical procedures or complications related to the primary procedure after the index leiomyoma procedure (Appendix 2, http://links.lww.com/AOG/B21).
We assessed three secondary outcomes. First, we compared long-term health care utilization, including leiomyoma-related outpatient, inpatient, and emergency department (ED) visits and all-cause outpatient, inpatient, and ED visits among the leiomyoma procedures. Unique visits were defined by health care claims with dates of service on different days for outpatient and ED and number of different episodes of inpatient stay. Long-term refers to a minimum of 5 years of follow-up from the index procedure with continuous insurance. Leiomyoma-related health care utilization was defined as a claim with a leiomyoma diagnosis. Although this definition of leiomyoma-related utilization may include visits for unrelated concerns (eg, pelvic inflammatory disease) but with a leiomyoma diagnosis attached, this possibility is rather minimal given that these outcome measures are assessed after the index leiomyoma procedure. Multiple claims on the same day were counted as one event. Second, we assessed health care utilization between women who received hysterectomy as the index procedure with those who received any of the three uterine-sparing procedures as the index procedure but underwent hysterectomy during follow-up. Addressing this subaim will help quantify the difference in health care utilization between those who undergo hysterectomy as the first procedure compared with those who first try uterine-sparing procedures but eventually undergo hysterectomy. Third, we compared reproductive outcomes between women undergoing myomectomy and uterine artery embolization as the index procedure. Specifically, we assessed the rates of pregnancies in these two cohorts after the index procedure and, among those who became pregnant, the rates of other reproductive outcomes including term delivery, preterm delivery, and spontaneous abortion.
For each study woman, the database provided information on demographic and socioeconomic characteristics such as age, race, household income, and residence region, her comorbidities, Charlson-Deyo comorbidity index,21 and health care utilization measures at baseline. Categorical variables were reported in terms of counts and percents and compared between the cohorts using χ2 tests. For continuous variables, means and SDs were reported and were compared between the study cohorts using t test. For nonnormal variables such as follow-up time and duration, median and interquartile range were presented, and Kruskal-Wallis test was used to compare these outcomes between the cohorts.
To compare outcomes between the study cohorts, we use propensity score matching method. The propensity score is the probability that a patient receives a specific treatment under the study, which in our case was the probability of undergoing a specific leiomyoma treatment procedure. The propensity score is estimated using a logistic regression, accounting for the baseline covariates that influence the receipt of a procedure. Two participants from two different cohorts (procedures) but with similar propensity scores have the similar probability of receiving any of the two procedures, and therefore they can be statistically interchanged as controls for one another. In essence, the propensity score method enables creation of study cohorts that are balanced in terms of observed patient characteristics in a very simple and straightforward manner.22 The other advantages of the propensity score method compared with traditional multivariate methods have been described elsewhere.22 One-to-one propensity score matching was used to balance the differences in baseline characteristics between the study cohorts.23,24 Patient characteristics reported in Table 1 (barring patient counts and follow-up duration) were adjusted in the propensity score matching.
We created four different propensity score-matched cohorts to assess different outcomes. First, to compare rates of reintervention, subsequent other surgical procedures (which could be a consequence or complication of the index procedure) and pregnancy occurrences among women undergoing myomectomy and uterine artery embolization, we matched women undergoing myomectomy and uterine artery embolization. Second, to compare 5-year health care utilization, we again matched women undergoing myomectomy and uterine artery embolization but required at least 5 years of follow-up with continuous health insurance. Third, to compare health care utilization between women undergoing hysterectomy as the initial procedure and those undergoing hysterectomy as a subsequent procedure after an index uterine-sparing leiomyoma procedure, we matched these two cohorts of women through propensity score matching that had at least 5 years of follow-up. Fourth, to compare reproductive outcomes after pregnancy after myomectomy or uterine artery embolization, we matched women who underwent myomectomy with those who underwent uterine artery embolization and became pregnant during follow-up. As explained in the “Results” section, as a result of the small sample size, only descriptive statistics were provided for the outcomes associated with the magnetic resonance-guided, focused ultrasound surgery cohort.
To account for variable follow-up, we further analyzed the primary outcomes—the risks of subsequent leiomyoma and other surgical procedures—through Cox proportional hazard models on the corresponding matched cohorts. All analyses were conducted using SAS 9.3 and Stata 13.1.
We identified 135,522 women with leiomyoma and bulk symptoms who received one of the four procedural therapies between January 1, 2000, and December 31, 2013. The flow diagram in Figure 1 indicates how the final study sample was obtained from the Optum Labs Data Warehouse database after applying the inclusion and exclusion criteria. The overwhelming majority of women received hysterectomy as the index procedure (111,324 [82.2%]); 19,965 (14.7%) and 4,186 (3.1%) received myomectomy and uterine artery embolization, respectively (Table 1). Although not shown in the table as a result of space constraints, the percentage of women with leiomyoma receiving hysterectomy as the index procedure decreased from 87% in 2000 to 82% in 2002, which stabilized around the latter percentage for the remainder of the study period. Because only a very small number of women (n=47 [0.0003%]) underwent magnetic resonance-guided, focused ultrasound surgery, we restricted their analysis to a limited number of descriptive outcomes, and relevant data pertaining to the magnetic resonance-guided, focused ultrasound surgery cohort are presented in Appendices 3–5, available online at http://links.lww.com/AOG/B21. The mean age for the overall study sample was 43.2 years (SD 6.0). The myomectomy cohort was younger (mean age 38.4 years) and a higher proportion of them had a pregnancy during the baseline period (12.9% for the myomectomy cohort compared with 4.3% of the whole cohort). On average, women in the study were followed for approximately 3.4 years. Other notable findings are that black women opted for uterine artery embolization at a substantially higher rate than other treatment options and that women seeking magnetic resonance-guided, focused ultrasound surgery and uterine artery embolization appear to have the highest leiomyoma-related outpatient utilization (Table 1).
As presented in Table 2, among the 24,151 women receiving a uterine-sparing index procedure other than magnetic resonance-guided, focused ultrasound surgery, 3,358 (13.9%) received a subsequent leiomyoma procedure (Table 2). Hysterectomy was the most common subsequent leiomyoma procedure with 2,150 (64.0%) undergoing hysterectomy followed by myomectomy (29.2%) and uterine artery embolization (6.7%). The median time between the index procedure and the subsequent leiomyoma procedure was 1.5 (interquartile range: 0.6–3.1) years. Overall, 2,316 (9.6%) women ended up having a hysterectomy during follow-up, which includes not only the 2,150 women described who had hysterectomy as their second procedure subsequent to the index procedure, but also those who had a leiomyoma procedure other than hysterectomy as the second procedure but ended up having hysterectomy beyond the second procedure. The median time between the index procedure and the subsequent hysterectomy was 1.8 (interquartile range: 0.7–3.4) years.
Table 3 presents the reintervention rates between propensity score-matched women in the myomectomy and uterine artery embolization cohorts. A small but statistically significant difference in overall reintervention rates between myomectomy and uterine artery embolization cohorts was found (15.0% compared with 17.1%, P=.02). Although the rates of subsequent leiomyoma procedure were similar in year 1 and 2 after the index procedure at approximately 7% and 12–13% for both myomectomy and uterine artery embolization cohorts, the differences became statistically significant from the third year onward (15.0% compared with 19.4%, P=.03 for third year). Similar patterns were observed when examining subsequent hysterectomies. Overall, 11.1% of the women undergoing myomectomy had hysterectomies compared with 13.2% of women who underwent uterine artery embolization (P<.01). The rates of hysterectomies were similar between myomectomy and uterine artery embolization cohorts in the first 2 years at approximately 5% and 9%, respectively. However, by the third year, significant difference emerged with the rates of hysterectomy in women undergoing myomectomy being 10.5% compared with 14.6% for women undergoing uterine artery embolization (P=.02). When analyzed by a Cox proportional hazard model, the uterine artery embolization cohort was 22% more likely to receive a subsequent leiomyoma procedure (hazard ratio [HR] 1.22, 95% CI 1.09–1.37; P<.001) and 32% more likely to receive a subsequent hysterectomy (HR 1.32, 95% CI 1.16–1.50; P<.001) than the myomectomy cohort (data not shown in the tables).
Significantly more women in myomectomy cohort underwent at least one subsequent surgical procedure than in the uterine artery embolization cohort: 24.6% compared with 18.1% (P<.001) with a corresponding HR of 1.38 (95% CI 1.25–1.52). In addition, 18.6% women who underwent myomectomy and 13.2% who underwent uterine artery embolization (P<.001) had subsequent surgical procedures other than leiomyoma procedures with a corresponding HR of 1.39 (95% CI 1.24–1.56) (Table 4).
Within our cohort, 31,294 (23%) women had at least 5 years of follow-up. We compared their median (interquartile range) health care utilization in terms of number of visits, including both leiomyoma-related and all-cause health care utilization during the 5 years after the index procedure (Table 5). As expected, the hysterectomy cohort had little leiomyoma-related health care utilization during follow-up, and they also had lower utilization of any leiomyoma-related outpatient services than uterine artery embolization and myomectomy cohorts. The uterine artery embolization cohort had more leiomyoma-related outpatient services and higher leiomyoma-related ED and inpatient visits but lower utilization of all-cause outpatient services than the myomectomy cohort.
In the propensity score-matched cohorts of myomectomy and uterine artery embolization who had at least 5 years of follow-up, the uterine artery embolization cohort had higher leiomyoma-related outpatient services (4.6 compared with 2.7, P<.001) but lower utilization of all-cause outpatient services (62.5 compared with 68.6, P=.009) (Appendix 6, available online at http://links.lww.com/AOG/B21).
In the propensity score-matched cohorts of women who underwent hysterectomy as their first (index) leiomyoma procedure compared with those who underwent a uterine-sparing procedure as the first procedure but subsequently received hysterectomy, the former had significantly lower leiomyoma-related or all-cause health care utilizations during the 5-year follow-up period. For example, women who underwent hysterectomy as their first procedure had 0.2 leiomyoma-related outpatient visits compared with 4.2 visits (P<.001) for those who first had a uterine-sparing procedure followed by hysterectomy; the former group of women had 59.2 all-cause outpatient visits compared with 72.0 all-cause outpatient visits for the latter group (P<.001) (Appendix 7, available online at http://links.lww.com/AOG/B21).
The percentage of women in myomectomy cohort who became pregnant was 17.8% compared with 2.0% in the corresponding uterine artery embolization cohort (7.5% compared with 2.2% in the matched cohorts, P<.001; data not shown in the tables). Among women who became pregnant after the index myomectomy or uterine artery embolization, we did propensity score matching between these two cohorts (Table 6). However, we did not find any significant differences between the two groups for any reproductive outcomes. In fact, the percentages of women who experienced any adverse reproductive outcome were identical in the two cohorts (69.4%; Table 6).
This study provides new information regarding myomectomy and uterine artery embolization, two widely used alternatives to hysterectomy for leiomyomas. Consistent with previous studies,16–18 we found that a higher percentage of women undergoing uterine artery embolization needed reintervention. In contrast to the overall reintervention rates for uterine artery embolization (17.1%) and myomectomy (15.0%) found in our study, previously reported reintervention rates for uterine artery embolization (14.0–36.7%) and myomectomy (2.7–6.1%) varied considerably; our study found a substantially higher reintervention rate for uterine artery embolization than previously reported.16–18 The current evidence on the complication rates between uterine artery embolization and myomectomy is mixed.14,18,19 Our finding of higher complication rates for the myomectomy cohort aligns with some prior studies that found higher complication rates in myomectomy than uterine artery embolization.18,19 However, note that such mixed evidence may stem from differences in complication definitions and the time periods during which complications were captured.
Reproductive outcomes including pregnancy are key criteria for many women and health care providers while deciding on leiomyoma treatment options. In this study, a much smaller percentage of women undergoing uterine artery embolization became pregnant compared with those undergoing myomectomy. There have been concerns about ovarian failure after uterine artery embolization,25 although the loss of ovarian function after uterine artery embolization was found primarily in women aged 45 years or older.26 It is noteworthy that women who underwent index uterine artery embolization were more likely to undergo hysterectomy subsequently than those undergoing index myomectomy (13.1% compared with 11.0%), potentially suggesting that patients undergoing myomectomy likely had a greater interest in subsequent pregnancy. A limitation of this study is that health care claims data do not capture women’s pregnancy intent or other relevant outcomes such as ovarian reserve reduction, and therefore the study could not make any definitive inference on pregnancy rates between uterine artery embolization and myomectomy. The lower pregnancy rate for uterine artery embolization was likely attributable in part to both patient and health care provider preference toward myomectomy for pregnancy optimization, which is also supported by the American College of Obstetricians and Gynecologists.27 This is also reflected in our data that women in the myomectomy cohort were more likely to have a pregnancy in the baseline period. A previous study found women who underwent uterine artery embolization were less likely to attempt to get pregnant, but those who did had a higher success rate than women undergoing myomectomy.19 Our study could not confirm this finding, but it was clear that between propensity-matched women from uterine artery embolization and myomectomy cohorts who became pregnant, the likelihoods of experiencing an array of adverse reproductive outcomes and delivery rates were similar albeit the small sample size after propensity matching.
Although alternatives to hysterectomy have become increasingly available in recent years, this study confirms that an overwhelming majority of the commercially insured women with leiomyoma-related bulk symptoms underwent hysterectomy. Given that hysterectomy is the definitive treatment for leiomyoma, the rate of hysterectomy with an underlying leiomyoma diagnosis continues to be high at approximately 80%, with a small decline in the early 2000s.8,28 This finding has been confirmed in our study as well with the rate stabilizing at approximately 82% since 2002. Although hysterectomy leads to lower health care utilization over 5 years compared with women who undergo uterine-sparing procedures, there may be long-term adverse consequences of hysterectomy including fracture risk, pelvic organ prolapse, cardiovascular disease risk, and risk of dementia, which may often take 20–30 years to manifest.29–32
Our study has the usual limitations of claims-based analyses including the general limitation that claims are generated for reimbursement purposes but not for research. Nevertheless, we adopted appropriate safeguards including a minimum of 1-year washout period to ensure that the study participants are newcomers to leiomyoma procedures. The very large database used in the study provides confidence in the generalizability of the results, particularly to commercially insured women in the United States. Given that black women were only approximately 10% in our study, and the fact that cumulative incidence of leiomyoma can be up to 80% in black women,1 additional studies including cohorts with higher proportion of black women (eg, Medicaid data) are warranted. Unobserved confounding factors not captured in the data may violate the propensity matching assumption and potentially bias the results.
In conclusion, more than 80% of the women with leiomyoma-related bulk symptoms underwent hysterectomy as the first-line procedural therapy. Among women who underwent uterine-sparing procedures, approximately one in seven women had a second uterine-sparing leiomyoma procedure, and 1 in 10 underwent hysterectomy over an average of 3.4 years. Compared with the myomectomy cohort, women who underwent uterine artery embolization had a higher risk of reintervention, lower risk of other surgical procedure, but a similar rate of adverse reproductive outcomes.
1. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM. High cumulative incidence of uterine leiomyoma in black and white women: ultrasound evidence. Am J Obstet Gynecol 2003;188:100–7.
2. Lee DW, Ozminkowski RJ, Carls GS, Wang S, Gibson TB, Stewart EA. The direct and indirect cost burden of clinically significant and symptomatic uterine fibroids. J Occup Environ Med 2007;49:493–506.
3. Flynn M, Jamison M, Datta S, Myers E. Health care resource use for uterine fibroid tumors in the United States. Am J Obstet Gynecol 2006;195:955–64.
4. Cardozo ER, Clark AD, Banks NK, Henne MB, Stegmann BJ, Segars JH. The estimated annual cost of uterine leiomyomata in the United States. Am J Obstet Gynecol 2012;206:211.e1–9.
5. Laughlin SK, Stewart EA. Uterine leiomyomas: individualizing the approach to a heterogeneous condition. Obstet Gynecol 2011;117:396–403.
6. Stewart EA. Uterine fibroids. Lancet 2001;357:293–8.
7. Stewart EA. Uterine fibroids: the complete guide. Baltimore (MD): Johns Hopkins University Press; 2007.
8. Merrill RM. Hysterectomy surveillance in the United States, 1997 through 2005. Med Sci Monit 2008;14:CR24–31.
9. Moore BJ, Steiner CA, Davis PH, Stocks C, Barrett ML. Trends in hysterectomies and Oophorectomies in hospital inpatient and ambulatory settings, 2005–2013: statistical brief #214. In: Healthcare Cost and Utilization Project (HCUP) statistical briefs. Rockville (MD): Agency for Healthcare Research and Quality; 2006.
10. Wright JD, Herzog TJ, Tsui J, Ananth CV, Lewin SN, Lu YS, et al. Nationwide trends in the performance of inpatient hysterectomy in the United States. Obstet Gynecol 2013;122:233–41.
11. Gliklich RE, Leavy MB, Velentgas P, Campion DM, Mohr P, Sabharwal R, et al. Identification of future research needs in the comparative management of uterine fibroid disease. A report on the priority-setting process, preliminary data analysis, and research plan. Effective Healthcare Research Report No 31. Rockville (MD): Agency for Healthcare Research and Quality; 2011.
12. McLucas B, Goodwin S, Adler L, Rappaport A, Reed R, Perrella R. Pregnancy following uterine fibroid embolization. Int J Gynaecol Obstet 2001;74:1–7.
13. Pron G, Mocarski E, Bennett J, Vilos G, Common A, Vanderburgh L, et al. Pregnancy after uterine artery embolization for leiomyomata: the Ontario multicenter trial. Obstet Gynecol 2005;105:67–76.
14. Mara M, Maskova J, Fucikova Z, Kuzel D, Belsan T, Sosna O. Midterm clinical and first reproductive results of a randomized controlled trial comparing uterine fibroid embolization and myomectomy. Cardiovasc Interv Radiol 2008;31:73–85.
15. Goldberg J, Pereira L. Pregnancy outcomes following treatment for fibroids: uterine fibroid embolization versus laparoscopic myomectomy. Curr Opin Obstet Gynecol 2006;18:402–6.
16. Broder MS, Goodwin S, Chen G, Tang LJ, Costantino MM, Nguyen MH, et al. Comparison of long-term outcomes of myomectomy and uterine artery embolization. Obstet Gynecol 2002;100:864–8.
17. Mara M, Fucikova Z, Maskova J, Kuzel D, Haakova L. Uterine fibroid embolization versus myomectomy in women wishing to preserve fertility: preliminary results of a randomized controlled trial. Eur J Obstet Gynecol Reprod Biol 2006;126:226–33.
18. Manyonda IT, Bratby M, Horst JS, Banu N, Gorti M, Belli AM. Uterine artery embolization versus myomectomy: impact on quality of life—results of the FUME (Fibroids of the Uterus: Myomectomy versus Embolization) Trial. Cardiovasc Intervent Radiol 2012;35:530–6.
19. Narayan A, Lee AS, Kuo GP, Powe N, Kim HS. Uterine artery embolization versus abdominal myomectomy: a long-term clinical outcome comparison. J Vasc Interv Radiol 2010;21:1011–7.
20. Wallace PJ, Shah ND, Dennen T, Bleicher PA, Crown WH. Optum Labs: building a novel node in the learning health care system. Health Aff (Millwood) 2014;33:1187–94.
21. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–9.
22. Borah BJ, Moriarty JP, Crown WH, Doshi JA. Applications of propensity score methods in observational comparative effectiveness and safety research: where have we come and where should we go? J Comp Eff Res 2014;3:63–78.
23. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265–81.
24. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399–424.
25. Hehenkamp WJ, Volkers NA, Broekmans FJ, de Jong FH, Themmen AP, Birnie E, et al. Loss of ovarian reserve after uterine artery embolization: a randomized comparison with hysterectomy. Hum Reprod 2007;22:1996–2005.
26. Kaump GR, Spies JB. The impact of uterine artery embolization on ovarian function. J Vasc Interv Radiol 2013;24:459–67.
27. Alternatives to hysterectomy in the management of leiomyomas. ACOG Practice Bulletin No. 96. American College of Obstetricians and Gynecologists. Obstet Gynecol 2008;112:387–400.
28. Whiteman MK, Hillis SD, Jamieson DJ, Morrow B, Podgornik MN, Brett KM, et al. Inpatient hysterectomy surveillance in the United States, 2000-2004. Am J Obstet Gynecol 2008;198:34.e1–7.
29. Melton LJ III, Achenbach SJ, Gebhart JB, Babalola EO, Atkinson EJ, Bharucha AE. Influence of hysterectomy on long-term fracture risk. Fertil Steril 2007;88:156–62.
30. Blandon RE, Bharucha AE, Melton LJ 3rd, Schleck CD, Babalola EO, Zinsmeister AR, et al. Incidence of pelvic floor repair after hysterectomy: a population-based cohort study. Am J Obstet Gynecol 2007;197:664.e1–7.
31. Howard BV, Kuller L, Langer R, Manson JE, Allen C, Assaf A, et al. Risk of cardiovascular disease by hysterectomy status, with and without oophorectomy: the Women's Health Initiative observational study. Circulation 2005;111:1462–70.
32. Phung TK, Waltoft BL, Laursen TM, Settnes A, Kessing LV, Mortensen PB, et al. Hysterectomy, oophorectomy and risk of dementia: a nationwide historical cohort study. Dement Geriatr Cogn Disord 2009;30:43–50.