A systematic review comparing hysterectomy with less-invasive treatments for abnormal uterine bleeding.

Abnormal uterine bleeding (AUB) affects up to 30% of women during their reproductive years. ( 1 – 4 ) AUB has a substantial impact on women’s quality of life, associated loss of productivity, and major health care costs. ( 5 – 8 ) AUB is a symptom of several different underlying conditions, which have been newly classified by the International Federation of Gynecology and Obstetrics (FIGO) Menstrual Disorders Group. ( 9 , 10 ) Under this system, causes of AUB are classified as polyps (AUB-P), adenomyosis (AUB-A), leiomyomas (AUB-L), malignancy and premalignant conditions (AUB-M), coagulaopaty (AUB-C), ovulatory disorders (AUB-O), endometrial disorders (AUB-E), iatrogenic (AUB-I), and “not classified” (AUB-N). Because of confusion surrounding the term “dysfunctional uterine bleeding” or “DUB”, this term is not included in the naming system and women who were formerly described as having “DUB” generally fit into the ovulatory disorders and endometrial hemostatic disorders categories if coagulopathy has been excluded. ( 9 , 10 ) Although hysterectomy is considered the “definitive” treatment for AUB regardless of etiology, less invasive options are increasingly available, including medical therapy, the levonorgestrel intrauterine system (LNG-IUS), and endometrial ablation. Better characterization of the relative efficacy of these alternatives compared to hysterectomy will improve patient counseling, facilitate informed decision-making, and reduce the burden of unnecessary procedures on both the patient and the health care system. The Systematic Review Group of the Society of Gynecologic Surgeons (SGS SRG) performed a systematic review of randomized control trials (RCTs) comparing hysterectomy with other treatment modalities for women with AUB presumed secondary to ovulatory (AUB-O) or endometrial dysfunction (AUB-E) under this new FIGO classification system. The objective of this study was to determine the relative efficacy of less invasive treatments compared to hysterectomy for bleeding control, change in quality of life, pain, sexual health, patient satisfaction, additional needed treatments, and adverse events. The systematic review was conducted with the goal of producing an evidence-based guideline on decision-making around hysterectomy for AUB-O and AUB-E. (If JMIG chooses to publish the clinical practice guidelines will insert the location of the guideline in the journal here) The SGS SRG, including gynecologic surgeons and systematic review methodologists, performed a systematic search to identify RCTs comparing treatments for AUB in women with ovulatory disorders (AUB-O), endometrial hemostatic dysfunction (AUB-E), or fibroids (AUB-L). A working document defining parameters for a literature search was created. ( 11 ) We searched MEDLINE from inception to January 14, 2011 for English language human studies. We included search terms for AUB, menorrhagia, fibroids, hysterectomy, myomectomy, uterine artery embolization, ablation, intrauterine devices, medical therapies, RCTs, and related terms. Details of the full search were reported in a previous publication. ( 12 ) We included RCTs that compared hysterectomy (via any route) to endometrial ablation, LNG-IUS, or medical therapies as treatments of AUB secondary to presumed ovulatory disorders (AUB-O) or endometrial hemostatic dysfunction (AUB-E), and reported an outcome of interest. We excluded RCTs that included only participants with AUB attributed to fibroids (AUB-L), though studies that included women with AUB-L in addition to AUB-O and AUB-E were not excluded. This review was conducted as part of a broader systematic review that addressed all treatments for AUB secondary to ovulatory disorders (AUB-O), endometrial hemostatic dysfunction (AUB-E), or fibroids (AUB-L). Data from studies were extracted by members of the SRG, most of whom had experience from a prior systematic review.( 13 , 14 ) Individual extractions were confirmed by a second member and discrepancies were resolved by consensus among all members of the SRG. Complete details on the definitions of variables of interest, outcome classification, and coding have been published.( 12 ) In brief, 114 outcomes were identified and ranked as “critical”, “important”, and “low-importance” for clinical decision-making. Using a ballot system all members of the SGS SRG used a structured process (explained in detail in a previous publication of the review group) to grade the importance of each outcome with a scale suggested by the GRADE Working Group ( 12 , 15 , 16 ). “Critical” and “important” outcomes were organized into 7 overarching domains considered in this review (bleeding, quality of life, pain, sexual health, patient satisfaction, need for subsequent surgical treatment, and adverse events).( 12 ) Adverse events were categorized as “major” [need for urgent re-operation/conversion to another surgery (hysterectomy or laparotomy), cystotomy, ureteral injury, bowel injury, pelvic or vaginal vault hematoma, death, myocardial infarction/cardiorespiratory event, deep vein thrombosis/thromboembolism, pulmonary embolus/thromboembolic event, stroke, admission to the intensive care unit, and “other”] and “minor” [wound infection, pelvic infection/cuff cellulitis, febrile morbidity, “other infection” (which included urinary tract infection, pneumonia, other infections), wound hematoma/wound rupture (which also included wound hematoma/seroma and unspecified hematomas), need for transfusion, re-admission without re-operation, fluid overload, uterine perforation, ileus or obstruction, non-urgent surgery presumed secondary to complications from the initial procedure. Re-operations which were not presumed secondary to complications from the initial procedure were not reliably reported (for example, subsequent surgery for ovarian cysts or adhesions) and therefore not included. We assessed the methodological quality of each study using predefined criteria from a three-category system modified from the Agency for Healthcare Research and Quality.( 15 ) Studies were graded as good (A), fair (B), or poor (C) quality based on the likelihood of biases and the completeness of reporting. Grades for different outcomes could vary within the same study. We decided a priori that meta-analyses would be performed only when three or more trials used similar interventions and outcome definitions. To grade the overall strength of evidence , we followed the Grades for Recommendation, Assessment, Development and Evaluation (GRADE) system, with four ratings: high, moderate, low, and very low. ( 16 ) As part of a public vetting process, the review and guidelines were presented for public comment at the 36 th SGS Annual Scientific Meeting April 2010 and posted on the SGS website, where public comments were solicited for 4 weeks.

The literature search identified 5,503 citations. Screening of citations and potentially relevant full-text articles resulted in 117 eligible articles (115 which were extracted) and 81 distinct RCTs relevant to the broader review. ( Figure 1 ) This review is based upon the 18 articles from 9 trials that directly compared hysterectomy with another intervention and met our eligibility criteria. ( Table 1 ) Seven RCTs (six rated as “B” quality and one rated as “A” quality) with 1,167 women, compared hysterectomy to endometrial ablation with follow-up from 4–48 months.( 17 – 26 ) Two of the seven trials included women with submucous myomas (AUB-L-SM). ( 23 , 26 ). In the hysterectomy arm of these seven trials, 49% (n=395) had a TAH, 26% (n=206) had a TVH, and 25% (n=199) had a laparoscopic approach to hysterectomy. All trials utilized resectoscopic methods of endometrial ablation [4 used resection only, one resection or laser ( 20 ), one resection or rollerball ( 23 ), and one electrodessication or thermal balloon ( 25 )]. Overall quality of the evidence for these studies was low to moderate for each outcome domain. The heterogeneity in specific outcomes reported precluded meta-analyses for this comparison. Seven studies reported bleeding outcomes. Amenorrhea rates for the endometrial ablation groups ranged from 13–64%, versus an implied 100% following hysterectomy in most studies (one study reported one woman who had bleeding following hysterectomy). ( 25 ) Because of the heterogeneity of outcomes measures between studies comparing hysterectomy and ablation (different questionnaires), there were insufficient data to systematically compare bleeding control outcomes other than amenorrhea for the evidence profiles. The strength of evidence was moderate that bleeding is better controlled following hysterectomy than ablation. Six studies evaluated quality of life. Overall, studies showed significant improvement above baseline in quality of life scores after treatment with both ablation and hysterectomy. Several studies found no difference between treatment arms in scores on various validated quality of life assessment tools.( 18 , 19 , 21 , 24 , 25 ) However, these studies were not powered to detect differences in quality of life scores, thus lack of statistical significance does not confirm no difference in effect. Three studies found statistically significant differences in various SF-36 dimensions favoring hysterectomy: pain( 19 ), general health ( 23 , 24 ), vitality ( 23 , 24 ), and social function ( 24 ). It should be noted that two of these three studies compared ablation specifically to minimally invasive methods of hysterectomy (the Zupi et al hysterectomies were LASH and the Croisignani et hysterectomies were TVH).( 23 , 24 ) The strength of evidence was moderate revealing no difference between hysterectomy and ablation in post-operative quality of life. Five trials assessed pain beyond the immediate post-operative period. ( 18 , 19 , 22 , 24 , 25 ). Improvement in “dysmenorrhea” was more prevalent in the hysterectomy group than the ablation group at 4 months post-procedure. ( 18 ) Three studies found pelvic pain was less prevalent in the hysterectomy group (5–19%) than the ablation group (24–64%) at 2–3 years follow-up, though only one study tested this statistically and found a significant difference.( 19 , 24 , 25 ) One study found no difference in the proportion experiencing pain at 4 years. ( 22 ) The strength of evidence on pain beyond the post-operative period was low and favored hysterectomy over ablation. Five trials reported comparisons of sexual health outcomes.( 18 , 22 – 24 , 26 ) These studies showed no difference in overall effect on sexual satisfaction and sexual health rating, although only one study used a validated questionnaire. ( 23 ) The quality of evidence for sexual health was low and revealed no difference in the effect of hysterectomy compared to ablation. Five trials compared general satisfaction between ablation and hysterectomy and four of which defined satisfaction rate as the primary outcome. ( 18 – 20 , 22 , 23 , 25 , 26 ) Outcome measures varied among studies. The majority of women in both treatment arms reported being satisfied with their treatment option (50–95% for ablation, 52–96% for hysterectomy) at 1–4 years follow-up. Three trials found no differences in satisfaction between the hysterectomy and ablation group. ( 23 , 25 , 26 ) Two trials found significantly greater satisfaction after hysterectomy than ablation. One trial reported more women “very satisfied” 1 year after hysterectomy (88% vs. 78%, p<0.05), but no significant difference at 4.8 years, possibly due to reduced statistical power due to dropouts. Another study found that at 4 months, 95% of the hysterectomy group was “quite satisfied” or “very satisfied” compared to 79% of ablation group (p=0.002). ( 19 ) Overall, the quality of evidence was very low for satisfaction outcomes which were not different between hysterectomy and ablation. The need for additional treatment after endometrial ablation, with either repeat ablation or hysterectomy, was reported as an outcome in seven trials.( 17 – 20 , 22 – 26 ) At 1–4 years follow-up, 16–42% of participants assigned to ablation had undergone an additional surgical treatment for bleeding with 10–29% treated with hysterectomy. Overall, the quality of evidence regarding additional treatments was moderate and favored hysterectomy over ablation. One “A” quality RCT (with six publications) compared hysterectomy to LNG-IUS. ( 27 – 32 ) A total of 236 women with heavy menstrual bleeding were followed for 6 months to 10 years after treatment. In the hysterectomy group 20% (n=21) had a TAH, 28% (n=30) had a TVH, and 52% (n=56) had a laparoscopic approach to hysterectomy. Women with submucous fibroids and women with “irregular bleeding” as a main complaint were excluded; 49% of participants had uterine fibroids. Overall quality of the evidence for these studies was moderate for each outcome domain. The proportion of participants who reported amenorrhea or oligomenorrhea with the LNG-IUS still “in-situ” was 51% (41/81) at 12 months and 75% (43/57) at 5 years. Women who discontinued use of the LNG-IUS (n=50, 42%) did so because of inter-menstrual bleeding (70%) and heavy bleeding (30%). Overall, the evidence favored hysterectomy over LNG-IUS for bleeding control. Both the EuroQOL-5D and the SF-36 were used to measure quality of life.( 27 – 29 ) Overall, the evidence revealed no differences in quality of life outcomes between hysterectomy and LNG-IUS. At 6 and 12 month follow-up, a greater proportion of participants in the LNG-IUS group reported lower abdominal pain. At 12 months, this difference was statistically significant (30% LNG-IUS vs. 14% hysterectomy, p=0.02). No difference was found at 5 years. ( 31 ) Overall, the evidence on pain revealed no differences between hysterectomy and LNG-IUS. Sexual function was compared between LNG-IUS and hysterectomy at 6 months, 1 year, and 5 years using McCoy sex scale scores ( 30 ) Compared to hysterectomy, the LNG-IUS group reported more sexual problems at 6 months (p=0.03), however, this effect was not seen at 12 months or 5 years. No differences were found in sexual problems at other time points or in sexual satisfaction scores at any time point. Overall, evidence for sexual health revealed no differences between hysterectomy and LNG-IUS. Satisfaction rates at 5 years were very high for both the LNG-IUS group (93%) and the hysterectomy group (94%). ( 29 ) Overall, the evidence regarding satisfaction showed no difference in satisfaction between LNG-IUS and hysterectomy. Among women randomized to LNG-IUS, 20% had undergone hysterectomy by 12-months, 42% by 5 years, and 46% by 10 years. Overall, the evidence on additional treatments favored hysterectomy over LNG-IUS. One RCT (rated as “A” or “B” quality for different outcomes) compared hysterectomy to medical therapy in 63 women with AUB presumably from DUB. The participants were followed for 2 years. ( 33 , 34 ) In the hysterectomy group 36% (n=10) had a TAH and 64% (n=18) had a TVH. The trial did not mandate a specific medication regimen and therapies used in this study included combined oral contraceptive pills (38%), cyclic progestin (16%), continuous progestin (6%), conjugated estrogen with progestin (25%), and conjugated estrogen alone (6%); 53% received a prostaglandin synthetase inhibitor, usually with a hormonal therapy. Overall quality of the evidence for these studies was low to very low for each outcome domain. Measures of bleeding control were not reported for this trial. No differences were found in the change of any SF-36 scores between medication and hysterectomy groups.( 33 ) Overall, the quality of evidence for quality of life measures was low and revealed no differences between hysterectomy and medications. No differences were found in pain between groups. ( 34 ) Overall, the quality of evidence for pain outside the post-operative period was very low and revealed no differences between hysterectomy and medications. No differences were found in SF-36 sexual satisfaction scores between groups. ( 33 ) Overall, the quality of evidence on sexual health outcomes was low and revealed no differences between hysterectomy and medications. No differences were found in change in satisfaction with symptom level between groups. ( 33 ) Overall, the quality of evidence for satisfaction was low and revealed no differences between hysterectomy and medications. At 2 year follow-up, 53% of the group randomized to medication had “crossed-over” to the hysterectomy group. ( 33 ) One patient in the hysterectomy arm required a trachelectomy 15 months after hysterectomy for persistent bleeding. Although “additional treatment” can be inferred from “crossed-over” to hysterectomy, it was not specifically assessed nor reported as an explicit “outcome” for this study. Therefore an evidence profile was not generated. Adverse events are listed in Table 5 , however reporting of adverse events was not standardized across studies and not all studies presented data on each adverse event listed. We combined the hysterectomy arms of all nine trials (ablation vs. hysterectomy, LNG-IUS vs. hysterectomy, and medications vs. hysterectomy) to obtain the complication totals for hysterectomy. In total, there were 57 “major” complications and 225 “minor” complications in the hysterectomy group which consisted of 724 patients. Rates/proportions are reported for each individual complication and not the composite of “major” and “minor” complications because one patient could have multiple complications and artificially inflate the overall complication rate. The most common “major” complication was pelvic or vaginal vault hematoma (n=28, 3.9%), with urinary tract injuries (n=7, 1% for cystotomy and n=1, 0.1% for ureteral injuries), and bowel injuries (n=2, 0.3%) being present but uncommon. For the studies comparing ablation to hysterectomy, adverse events were reported in seven studies (n=1167). In patients randomized to ablation, there were a total of seven major complications and 69 minor complications. The most common “major” complication was conversion to a different surgery or urgent re-operation, but this occurred in only 1% of patients (n=6). The most common “minor” complication was fluid overload, which occurred in 4% of patients (n=23). Two studies statistically compared adverse events between hysterectomy and ablation groups. One found no significant differences in complications between ablation and hysterectomy. ( 24 ) The second found no different in intra-operative complications, but a higher rate of complications in the hysterectomy group pre-discharge (p<0.001) and post-discharge (p=0.004). ( 26 ) There appears to be a trend to a higher rate of major and minor complications for hysterectomy than for ablation. Overall, the quality of evidence regarding adverse events was moderate and favored ablation over hysterectomy for minimizing adverse events. One study compared the LNG-IUS to hysterectomy. In this study, one major complication (a recurrent thromboembolic event) occurred in the LNG-IUS group (1%) and no minor complications. There were 12 major complications and 32 minor complications among patients receiving hysterectomy in this study and the authors reported that 30% of participants in the hysterectomy group had “post-operative complications”. Overall, the evidence for adverse events was moderate quality and favored LNG-IUS over hysterectomy. In the one trial that compared medications to hysterectomy, there were no reported complications in the medication group. In the hysterectomy group, there were 3 major complications and 1 minor complication. Overall, the quality of evidence for adverse events was low and favored medication over hysterectomy.

Designing and interpreting research on AUB treatment is challenging. Many different treatment modalities are available and AUB is associated with a wide spectrum of symptoms. Our review provides a concise assessment of the efficacy of less invasive options for managing AUB, predominantly from ovulatory dysfunction or endometrial hemostatic dysfunction, as compared to hysterectomy within seven important clinically-relevant outcome domains. Despite the high prevalence of AUB and a variety of available treatments, we found only 9 RCTs reported in 18 articles which compared hysterectomy directly to other “less invasive” treatment options. We identified important trade-offs between the relative benefits and harms of hysterectomy and endometrial ablation, LNG-IUS, and medications. Regardless of the comparator, hysterectomy was favored over other options for elimination of bleeding symptoms and need for additional treatment. Hysterectomy was also favored over ablation for pelvic pain resolution outside of the immediate postoperative period. However, these superior outcomes were achieved with the tradeoff of higher risks of adverse events. While hysterectomy is more efficacious for treatment of many AUB symptoms, we found less invasive options including endometrial ablation and LNG-IUS to be comparable to hysterectomy in achieving outcomes related to patient satisfaction and quality of life improvement. Consistent with our finding, a previous meta-analysis of individual patient data from 17 trials (six which compared hysterectomy to ablation, one which compared hysterectomy to LNG-IUS, 3 which compared ablation to LNG-IUS, and seven which compared ablation techniques) concluded that patients undergoing hysterectomy were more satisfied than patients undergoing endometrial ablation. However, satisfaction rates were very high for both groups and absolute differences were small.( 35 ) A limitation to our comparison of hysterectomy versus endometrial ablation is that all of the RCTs identified involved resectoscopic endometrial ablation techniques, which, while still used, have largely been replaced by a number of non-resectoscopic ablation techniques. Thus the applicability of the available trials to current practice is unclear. However, two recent reviews comparing non-resectoscopic to resectoscopic endometrial ablation techniques found no difference in amenorrhea rates, satisfaction rates, and rates of additional surgical interventions.( 35 , 36 ) Because of the limited body of evidence comparing medical therapies to hysterectomy, we are unable to make definitive conclusions on the relative effectiveness of individual medical therapy compared to hysterectomy. Many different medication types were used in the “medication arm” of the one RCT which directly compared medications to hysterectomy, which may help with the broad comparison of “medication vs. hysterectomy” but did not allow direct comparison of individual medical therapies to hysterectomy in terms of effectiveness. Most of the data concerning effectiveness of medical therapies are derived from RCTs comparing specific medications to either placebo or each other, which were outside of the scope of our review. Given that medications arguably represent the “least invasive” alternative to hysterectomy, the lack of evidence in this area highlights the importance of future research. For this review, we chose to include only evidence from RCTs which compared hysterectomy to uterine-preserving treatments. Although this reduced the body of evidence, using the hysterectomy comparison arm as a prerequisite for study inclusion provided a means to standardize populations across trials and to reduce the clinical heterogeneity and selection bias that stems from recruitment for treatments with varying interventional risks. However, we were challenged by the inconsistency across studies in the definitions of the study populations. There are multiple etiologies of AUB and although we reviewed trials of treatment of women with AUB presumed secondary to ovulatory dysfunction or endometrial hemostatic dysfunction (AUB-O or AUB-E), it was not always clear how the studies identified the cause of bleeding, especially whether or not they investigated possible coagulopathies as a cause (AUB-C). Additionally, two of the studies comparing ablation to hysterectomy included patients with submucous myomas, (AUB-L-SM) which could affect bleeding and treatment effectiveness. Another obstacle is that the term “DUB”, which was used in most of these trials, is used inconsistently by clinicians, researchers, and experts on AUB. ( 9 ) This is why an expert group from FIGO has recommended discontinued use of the terms “dysfunctional uterine bleeding” and “DUB” and why we have not used these terms throughout this manuscript. In addition to the small number of trials, this review was limited lack of consistency in the eligibility criteria of individual trials, and the extensive heterogeneity of studied outcomes. As discussed in greater detail in a previous publication ( 12 ), the extreme heterogeneity in specific outcomes across trials limited the interpretability of their conclusions. Likewise, the sparseness of data made it difficult to come to definitive conclusions about the efficacy of hysterectomy versus alternative treatments or to rank treatments. A clear example of this challenge was studying the efficacy in bleeding control. For most women seeking treatment for AUB, a reduction in their amount of bleeding to an acceptable level or a “normalization of menses” by their personal standards would likely constitute a satisfactory outcome. Unfortunately, amenorrhea was the only outcome that was reported in a consistent fashion across studies, allowing direct comparisons between hysterectomy and other treatments. We could not compare treatments in terms of “reduction in bleeding” or “normalization of menses”. Studies tended to report overall patient satisfaction with their interventions, but it is unclear how this may correlate with “satisfaction” with their menstrual bleeding. There is a need to standardize and validate outcome measures that are responsive to important symptoms from AUB ( 12 , 37 ) and to use them consistently going forward. We generated clinical practice guidelines for the treatment of AUB-O and AUB-E based on our systematic review (see accompanying article). Because limitations in the current body of literature comparing hysterectomy to other options allowed us to make only “weak” recommendations, we offer suggestions to address these shortcomings with additional higher quality research. Treatments not yet compared against hysterectomy, like newer endometrial ablation techniques, need to be studied in appropriate trials before we can confidently appraise their relative value. Additionally, although some studies used minimally invasive approaches to hysterectomy, overall the majority of hysterectomies for the ablation comparisons were performed abdominally (49%). It is possible that outcomes may vary by method of hysterectomy and the studies involved in this review did not stratify results by hysterectomy type. Given emphasis on increasing the proportion of hysterectomies performed by minimally invasive approaches (total vaginal hysterectomy and laparoscopic and robotic approaches to hysterectomy), the outcomes of these methods need to be more clearly studied relative to non-hysterectomy options for AUB. ( 38 ) Also, studies looking at outcomes of treatment for women with AUB should be powered to detect a meaningful difference in symptoms related to bleeding and disease-specific quality of life. ( 39 ) In summary, in comparing hysterectomy to the “less invasive” options of endometrial ablation, LNG-IUS, and medications, we identified tradeoffs between treatment effectiveness and risk of serious adverse events. Given the many options available for the treatment of AUB, it is important that clinicians are aware of the relative advantages and disadvantages of each option and able to discuss this with patients seeking treatment for AUB. Each patient should receive counseling about the efficacy as well as burden and risk of each management option in the context of her disease manifestations to facilitate an informed choice based on the available evidence.