Expectant Management

We did not identify any studies intentionally designed to determine outcomes of no intervention, also called expectant management or watchful waiting. However, 16 small RCTs compared a treatment to no intervention, typically trials that compared a medication with placebo. Of these trials two were of good quality, six were fair, and eight were poor quality. The evidence, based on an average followup time of five months (range: 3 to 12 months), suggests the size of fibroids does not meaningfully change over these short timespans. One study reported a four percent reduction in size and those that reported volume measures documented an average increase in size of about 9 cm³, which is about one-fifth the size of a golf ball. The two studies that followed postmenopausal women for a full year did not detect an increase in total volume of fibroids.

Likewise, bleeding characteristics, such as days of bleeding and severity of bleeding as measured by hemoglobin, heaviness of periods, and severity of heavy bleeding episodes, did not change meaningfully during followup for those without active management. The proportion of the 514 women enrolled in these trials who presented specifically with problem bleeding, as opposed to other fibroid-related symptoms, is not known. However, the data suggests that women with fibroids should not expect that bleeding patterns will worsen over the near term.

Summary/Strength of Evidence (SOE): None of these studies were designed to evaluate expectant management. The number of women in the literature followed without intervention is small, and these participants may be fundamentally different from other women with fibroids since they were willing to risk randomization to no intervention. For these reasons, the evidence is insufficient to inform choice of expectant management over other options.

Medical Management

We identified 43 studies assessing effectiveness of medical treatment (including progesterone delivered by IUD) for uterine fibroids. Ten studies had placebo or no treatment comparison groups. Approximately one third were industry sponsored. The longest duration of followup after the end of treatment was 60 months in one study. Women included in the studies were predominately premenopausal (39 studies). Four studies evaluated therapies in postmenopausal women. We assessed four as good quality, 12 as fair quality, and 27 as poor quality for effectiveness outcomes.

GnRH (Gonadotropin-Releasing Hormone)Agonists

Eighteen studies (reported in twenty publications) addressed GnRH agonists, which included seven studies of “add-back” therapy (addition of a second agent to a GnRH agonist). The studies included 912 participants, and followup was typically limited to the immediate end of treatment. Only six studies followed women post-treatment, for 3 to 6 months. GnRH agonists reduce the size of fibroids, with reductions in volume of fibroids documented between 64 and 175 cm³ and reductions in the total volume of the uterus between 131 and 610 cm³.

Six studies reported complete absence of bleeding during treatment, three noting statistical significance for clinically important reduction from baseline. No study reported an increase in bleeding or worsening in hemoglobin or hematocrit. Individual women in several studies discontinued treatment because bleeding became more irregular or did not decrease.

Pain symptoms improved by GnRH treatment in four studies included pelvic pressure, pelvic and abdominal pain, and dysmenorrhea. Two studies reported similar improvements but without statistical comparisons of baseline to followup. Studies consistently reported significant improvement in measures of quality of life symptoms (days of bleeding, heavy menstrual bleeding, pelvic pressure, pelvic pain, urinary frequency, and constipation). Harms associated with GnRH included onset of menopausal symptoms, unfavorable changes in lipid profile, declines in cognitive function and memory, and bone loss, although some of these can be ameliorated with hormonal add-back therapy. Extended followup of women after they discontinue GnRH agonists is limited.

Summary/SOE: Moderate strength of evidence supports that GnRH agonists (with and without add-back therapy) reduce the size of fibroids, the overall size of the uterus, and bleeding symptoms. Low strength of evidence suggests that fibroid-related quality of life improves with and without add-back therapy.

Progesterone Receptor Agents: Anti-Progestins, Selective Receptor Modulators, and Levonorgestrel IUD

Seven studies provided data about outcomes of the anti-progestin mifepristone treatment. Average length of time for off-medication followup was 11 months with the longest untreated followup being 18 months. All studies observed a decrease in the size of fibroids at the completion of the period of active treatment. The magnitude of change in size of the largest fibroid ranged from a decrease of 37 cm³ to 95 cm³, with an average decrease of 71 cm³ among the 575 women receiving mifepristone. Total uterine volume also decreased across women receiving mifepristone.

All studies of mifepristone that assessed bleeding reported reduced heaviness of bleeding. Two comparisons found mifepristone superior to placebo. Women were more or equally likely to have decreased bleeding or absent menses on the lower doses compared with the higher doses. Each of six studies that evaluated pelvic pain before and at conclusion of treatment noted substantial improvements (present in 68% to100% at baseline compared with 9% to 28% after 3 months of treatment); findings were similar at the conclusion of 6 to 9 months of treatment and maintained post-treatment in roughly 60 percent to 90 percent of women in three RCTs with longer term followup. Four studies reporting the outcome consistently noted significant improvements in quality of life measures. Some studies suggested fibroids do resume growth after treatment. Few participants in these trials pursued other treatment during or after active treatment. Harms included spotting, elevations in liver function enzymes, and endometrial hyperplasia.

Six studies investigated treatment with ulipristal, a selective progesterone receptor modulator. All studies found ulipristal effective for reducing the size of individual fibroids and the overall fibroid burden as measured by total fibroid or uterine volume. Ulipristal, as intended, resulted in absent menses for the majority of women during treatment (range: 62% to 100%) and improved or stabilized hematocrit or hemoglobin. The exception was among women who had submucous fibroids who, in one study, had less improvement in bleeding. Compared with placebo, all ulipristal doses improved fibroid-related quality of life, and two of the six trials also documented improvement in pain. Durability of effect at 6 months was assessed in two studies: one found minimal resumption of fibroid growth (8.1%) after completion of treatment regardless of ulipristal dose. A second found that the 10 mg dose sustained fibroid size reduction while there was increase in size on the 5 mg dose that was similar to placebo. Ulipristal was associated with hot flashes, hyperplasia, and increases in liver function enzymes, although it is not clear if these effects persist after treatment ends.

Levonorgestrel (LNG) IUD improved bleeding (the only outcome of interest reported); however, the single available trial was of poor quality including lack of participant masking.

Summary/SOE: There is moderate strength of evidence that both mifepristone and ulipristal effectively reduce the size of fibroids and bleeding symptoms, while improving quality of life. Duration of effects is uncertain. Evidence is insufficient to choose between higher and lower doses. Evidence was insufficient to assess the effectiveness of the LNG-IUD on any outcomes

Estrogen Receptor Agents

Six studies included agents that act at the estrogen receptor. Three studies, two of fair quality and one of poor quality, investigated raloxifene (which acts as an anti-estrogen in breast and endometrial tissue) in comparison with placebo. Fibroid size decreased by 4.4 cm³ to 34.2 cm³ in two studies of raloxifene and did not change size in another raloxifene study (To put this in perspective, 40 cm³ is the volume of a golf ball). In raloxifene studies with premenopausal women, neither bleeding pattern (in 3 studies) nor hemoglobin levels (in one study) were improved compared with placebo. Among postmenopausal women, the percent of treatment cycles without bleeding was similar and the number of episodes of spotting and severity of bleeding were similar among women in the treated and control group.

A single poor quality study evaluated tamoxifen, which acts as an anti-estrogen within breast tissue and as an estrogen ligand in the endometrium. Tamoxifen use in premenopausal women did not influence length or severity of bleeding compared with placebo. Change in fibroid characteristics was not reported. Women receiving tamoxifen had less pain after four months of treatment compared with placebo.

Two poor quality RCTs had a total of 42 women receiving hormone replacement therapy (HRT) (transdermal estrogen replacement plus cyclic oral medroxyprogesterone acetate) after menopause. They compared hormone therapy to tibolone (not available in United States) for menopausal symptom management with attention to whether treatment increased size of fibroids. Growth was approximately 10 cm³, which is a quarter the size of a golf ball. In the longer study there was no further growth between 6 and 12 months.

No studies reported any serious adverse effects.

Summary/SOE: Studies provide low strength of evidence that, if prescribed to women with fibroids for other conditions such as breast cancer prophylaxis, raloxifene will not cause significant growth of existing fibroids or exacerbate bleeding. Evidence is insufficient to assess if tamoxifen or HRT does or does not promote fibroid growth.

Procedures for Uterine Fibroids

We identified 28 studies assessing procedures for uterine fibroids. Most studies compared similar procedures, two compared different procedures, and nine compared procedures against surgery or medications. The longest duration of followup after the end of treatment was 5 years in two studies. Participants were predominately premenopausal. We assessed 6 studies as good quality, 8 as fair quality, and 14 as poor quality for effectiveness outcomes.

Uterine Artery Embolization (UAE) and Occlusion

We identified 21 studies that randomized women to UAE or uterine artery occlusion. We assessed five studies as good quality, eight as fair quality, and eight as poor quality for effectiveness outcomes. Fibroid and uterine volume decreased significantly and consistently following UAE (up to 12 months postprocedure) regardless of the embolization agent or size of particles used to occlude the fibroid arteries. Longer-term followup reports from the Embolization for the Treatment of Symptomatic Uterine Fibroid Tumors Study, called the EMMY Trial, confirmed that fibroid and uterine volume reductions persist up to 5 years after UAE; however, 28 percent (23/81) of women underwent subsequent hysterectomy. Subsequent treatment was reported in 11 trials with length of followup ranging from 6 to 60 months (Table 23 in the full report). Hysterectomy was the most frequent intervention (8.9%) followed by repeat embolization (4.2%), myomectomy (3.6%), medication or IUD (1.1%) and endometrial ablation (0.1%).^13-23

For UAE, bleeding effects were consistent, with declines in bleeding or bleeding-related measures reported in seven studies. Pain improved in up to 84 percent of women in two studies reporting this outcome. Only eight studies of UAE measured quality of life, which consistently improved postprocedure with durability in the two studies with longer-term followup. Treatment satisfaction was high in seven studies reporting this outcome.

No women receiving UAE required transfusion. Risk of major complications during and following UAE ranged from 1.2 to 6.9 percent around the time of the procedure up to about 5 percent by 2 years. The probability of “major complications” as defined by authors was high in two studies that reported long-term followup (21% at 5 years in the Randomised comparison of uterine artery embolisation with surgical treatment in patients with symptomatic uterine fibroids Study, termed the REST Trial, and 16.8% at 32 months in a second study) in large part because they considered any subsequent procedure a complication, while other authors did not. Fertility outcomes were not evaluated in the UAE-only trials. When available, pregnancy outcomes after UAE are presented in the sections that compare UAE to other procedures or surgeries.

Trials of uterine artery occlusion were small, and a variety of intervention methods were described, which prohibits conclusions.

Summary/SOE: There was high strength of evidence that UAE is effective for reducing the size of fibroids and total uterine volume. Moderate strength of evidence finds that bleeding and quality of life is improved following embolization. The effect of UAE on reproductive outcomes is not well studied and evidence is insufficient to guide care or determine safety.

High Intensity Focused Ultrasound for Fibroid Ablation

Six studies (reported in 7 publications) assessed high intensity focused ultrasound (HIFU) for fibroid ablation, but only one fair quality pilot study (n=20) used magnetic resonance imaging (MRI) guidance, which is used in the United States. The other studies were rated as poor quality primarily due to lack of masking participants and outcome assessors to the intervention received. In four studies reporting effects on fibroid size, the magnitude of fibroid volume reduction was greater at 12 months after ultrasound destruction than at 1 month post-treatment. One year after treatment fibroid volume decreased by averages of 90 and 170 cm³ in two studies. Studies did not report on bleeding, pain, or pregnancy outcomes. One study addressed quality of life but did not report baseline data, and one reported improvements in sexual function. One study reported no transfusions among 48 participants. No study reported major complications.

Summary/SOE: HIFU reduced fibroid and uterine size, but strength of evidence is low because of short followup and poor quality of overall study design. Evidence related to patient reported outcomes is insufficient.

Radiofrequency Fibroid Ablation

Two RCTs, both assessed as poor quality, addressed radiofrequency ablation of fibroids. Both studies reported the technique was successful in delivering treatment to 5 percent or more of the targeted fibroid volume. Studies provided limited data on effects of ablation on bleeding, quality of life, and subsequent pregnancies, but did not report pain outcomes and noted no major complications.

Summary/SOE: The strength of evidence for radiofrequency ablation is insufficient to inform care.

Surgical Management

We identified 37 randomized trials with at least one arm that assessed surgical intervention (endometrial ablation, myomectomy, or hysterectomy) for uterine fibroids. One compared myomectomy to hysterectomy, the remainder evaluated outcomes in women treated by UAE, uterine artery occlusion, HIFU, or medication. The longest duration of followup after the end of treatment was 60 months in two studies. Women included in the studies were predominately premenopausal. We assessed nine as good quality, 11 as fair quality, and 17 as poor quality for effectiveness outcomes.

Endometrial Ablation

One fair quality study addressed endometrial ablation and reported significant decreases in bleeding after both rollerball and thermal balloon ablation, with similar rates of reintervention (9%) in both groups. More women in the rollerball group had complications. More than a third of women receiving each intervention noted dissatisfaction with ablation results on a single item with three levels of satisfaction. Dissatisfaction was associated with failure to achieve amenorrhea.

Myomectomy

Fifteen RCTs reported health outcomes after myomectomy and five additional studies provided information about harms. We assessed four studies as good quality, six as fair quality, and 10 as poor quality. Studies did not report changes in fibroid characteristics (e.g., proportion of fibroids removed at myomectomy, decrease in uterine volume). One small study reported improved bleeding patterns one year after surgery in 12 of 15 women. Another described change in undefined symptoms with relief reported by 51 of 88 women. Fibroid recurrence, reported in five studies, ranged from 2.5 percent to 25 percent with duration of followup ranging from 6 to 60 months. Two additional studies reported no recurrences at 6 month followup. Laparoscopic myomectomy was associated with faster return to usual activity than comparator surgeries in three studies and with improved quality of life compared with hysterectomy in one study. Laparoscopic myomectomy was also associated with better fertility and pregnancy outcomes over other myomectomy techniques in one study (n=136), but outcomes were comparable in another RCT (n=131). Transfusion rates were most often zero (1,040 participants in 10 studies). Six studies reported 25 transfusions among 502 participants treated by myomectomy, and seven did not report transfusion. Intraoperative conversion from myomectomy to another procedure ranged from 0 to 17 percent in eight studies (n=658). Harms associated with myomectomy included transfusion and pelvic organ injury. Harms did not differ among techniques.

Summary/SOE: Myomectomy is associated with improved quality of life (low strength of evidence). Since myomectomy removes fibroids, the effect on fibroid size is not measured, and strength of evidence is not graded. Evidence is insufficient to determine if myomectomy meaningfully improves bleeding patterns or anemia. Myomectomy is an option for women desiring future fertility, though evidence is insufficient to define potential benefit.

Hysterectomy

We identified 14 RCTs assessing hysterectomy (via abdominal, vaginal or laparoscopic-assisted approaches) in women with uterine fibroids. Seven reported harms only (i.e., did not report final health outcomes for effectiveness). Assessment duration (where clearly reported) in comparative studies ranged from 15 days to 24 months. We assessed five as good quality, three as fair quality, and six as poor quality for effectiveness outcomes. Among seven studies reporting health outcomes, one noted a decrease in hemoglobin postoperatively, and two reported increases in hemoglobin levels 24 months after surgery. Pain symptoms improved in three RCTs reporting outcomes of fibroid-related pain. Time to return to usual activity after hysterectomy averaged 30 to 40 days in three studies. One study reported faster recovery (mean 22 days) after laparoscopic hysterectomy. Women reported good or very good satisfaction postoperatively, though one study reported worsened physical health compared with baseline measures at 5-year followup.

The percent of women needing transfusion following hysterectomy ranged from 0 to 20 percent in 890 women from 11 studies. An event of organ perforation occurred in one study, but overall risk across studies cannot be calculated since bowel and bladder injury were not uniformly reported across studies.

Summary/SOE: Hysterectomy de facto reduces uterine size and bleeding; the strength of this evidence cannot be graded. There is low strength of evidence that hysterectomy improves quality of life.

Direct Comparisons of Interventions

In total, 17 studies compared the effectiveness of two or more interventions. We identified four studies designed to compare outcomes across two or more of drugs (e.g., GnRH vs. ulipristal acetate). Overall these studies were small and inadequately powered for providing definitive evidence of greater effectiveness.

Two studies compared HIFU to other interventions, with greater tumor destruction after radiofrequency ablation compared with HIFU, and comparable sexual function after HIFU or myomectomy, but faster recovery in the HIFU groups.

Other direct comparisons of procedures included comparisons of UAE to myomectomy or hysterectomy. Technical success and quality of life were similar between UAE and myomectomy but reintervention rates were higher with UAE. Because of low power to detect differences, the evidence is insufficient to determine if pregnancy outcomes are better after myomectomy compared to UAE. Likewise ovarian reserve was assessed in only one study comparing UAE to hysterectomy. Symptom relief and quality of life outcomes were generally similar between UAE and hysterectomy (regardless of surgical approach), with faster recovery associated with UAE. Subsequent treatment rates were higher in the UAE group than in the hysterectomy group at each time point in followup; however, the majority of women randomly assigned to have UAE avoided hysterectomy for the duration of followup, which included 5 years of surveillance in the largest study. Fewer than one in three women with UAE required additional treatment.

Summary/SOE: Because of low power to detect differences, the evidence is insufficient to determine if outcomes differ in direct comparisons of procedure or medications

Analysis of Subsequent Treatment Following Initial Treatment for Uterine Fibroids

From data reported in 38 studies, we estimated the probabilities of receiving additional treatment for fibroids after randomization to initial treatment with medical management, UAE, or myomectomy. For women in their 30s, the model predicted that the probability of subsequent intervention for fibroids over 2 years varied from 6 to7 percent after medical treatment or myomectomy, to 44 percent after UAE. For women in their 40s and 50s, modelled 2-year reintervention rates were 9 to19 percent following medical treatment or UAE, and 0 percent after myomectomy.

Overall, fewer than half of women had another intervention within 24 months. UAE was most often followed by myomectomy among those in their 30s, and by hysterectomy among those in their 50s. Younger women who initially had myomectomy were most likely to have repeat myomectomies over the 2 years of followup. After medical treatment, very few women in any age group had subsequent treatment within 2 years.

Comparative Effectiveness

Studies comparing different categories of intervention were rare. Most were single studies of the specific comparison investigated. Two studies compared UAE to myomectomy; 3 studies compared UAE to hysterectomy; and four studies compared different drugs, but because of quality and size of these single comparison studies, evidence is insufficient to guide choices between medications and procedures.

Subsequent Intervention

For each of these interventions (uterine artery embolization, myomectomy, and medical management) and the subsequent treatment possibilities, the meta-analysis estimates 44 percent of women in their 30s received subsequent intervention after UAE. The findings are intriguing but inadequate to guide care about the sequence of treatments that may have the best outcomes because the overall quality of trials is limited and few women were followed long enough to identify treatment patterns. It is likely that much fewer than half of women will choose subsequent treatment in the near-term after an initial intervention. However, we can also speculate that the priorities which led women to participate in the initial trials reflected the intensity of treatments they were most interested in pursuing so it is not surprising surgeries were most followed by other procedures promptly (within 6 months) by those were not satisfied with initial results while those who enrolled in medication trials were less likely to pursue more aggressive options. Because of the limited roster of studies that followed women for 6, 12, or 24 months, this analysis does not substitute for study of treatment trajectories in which all initial treatments can be followed by all possible combinations of next treatments.

Applicability

Overall, our findings are widely applicable to the general population of women seeking treatment for uterine fibroids. For KQs 1 and 2 we set inclusion criteria for this review to women of any age with uterine fibroids with patient outcome data beyond intermediate outcomes only. We excluded studies in pregnant women, and restricted our synthesis to treatments currently available in the United States. Over 40 percent of the studies were conducted in European countries and another 27 percent were conducted in the United States or Canada. Although the outcomes collected may differ by country and by healthcare setting, the interventions were selected to be comparable so that the results reported in this review are expected to apply to women with fibroids in the United States.

Evaluation of expectant management was not an explicit aim of any trial. Sixteen studies with placebo arms or no treatment arms that included 514 women served as a surrogate. This population is not an ideal substitute as participants in the trials presumably hoped to receive active treatment and may report their status differently than women willing to be randomized to watchful waiting. This could restrict applicability but since the majority of studies included a plausible level of participant masking, they would be unlikely to know if they were on an active agent.

Medical management of fibroids was assessed in over 2,800 predominately premenopausal women from 43 studies (15 industry-sponsored and 11 conducted in the United States). Procedures, including UAE, HIFU, and radiofrequency fibroid ablation were evaluated in 28 studies including almost 2,000 women. Surgical studies evaluated hysterectomy, myomectomy, and endometrial ablation in over 3,000 women. Although none of the surgical studies were conducted in the United States, the surgical procedures are comparable to those widely available to women in the United States.

Data in these studies were inadequate to assess applicability based on patient characteristics (age, race/ethnicity, pregnancy intention, or menopausal status) or fibroid characteristics (size, position, and number) that could influence effectiveness outcomes.

While there are limitations in the literature as discussed below, the information that is available from these trials is relevant to contemporary practice. In summary, this review is generally applicable to women in the United States seeking one of the many treatment choices currently available for fibroids.

For KQ3 and 4: Data can only be systematically obtained from published research or publically available research. Those represented in the literature may differ from the universe of women having surgery for fibroids in the US. The literature about risk and outcomes of leiomyosarcoma does not separate cases well by type or surgery or menopausal status. Prospective studies which include a greater representation of myomectomy patients may be more applicable for discussing risk among younger women.

Limitations of the Systematic Review Process

Methodologic choices constrain the findings of this report. We chose to focus on publications in the English-language literature, to restrict to randomized clinical trials for the comparative effectiveness synthesis (KQs 1 and 2), and to review only those studies that included at least one intervention that is available in the United States. Similar reviews have documented in the past that language restrictions do not increase risk of omitting high quality trials. This is especially true for the topic of fibroids because the fibroid research community is small. Our technical expert panel and authors are familiar with prior and ongoing work and helped assure relevant studies have not been overlooked. Restricting to trials allowed us to sharply focus on proof of effectiveness. Because all individuals whose outcomes were assessed in these studies were randomly assigned to the intervention received, provider and patient biases in intervention choice are reduced and risk of confounding, difficult to fully assess or adjust for in cohort studies, is minimized. Random assignment to intervention arm in trials reduces bias and allows aggregation and summary of the findings by study arm, as presented in this report.

Restricting the review to randomized controlled trials limited ability to detect the full range of harms since studies were generally not designed or powered to evaluate harms, and many had a short duration of follow up.

Our analysis of subsequent intervention after a first intervention could be biased by the types of studies that reported this data and by differences in the willingness of women to be randomized to different types of intervention.

For meta-estimates related to leiomyosarcoma risk, available evidence based on pathology specimens for estimating presence of leiomyosarcoma in a mass believed to be a fibroid is accruing rapidly and will likely continue after the production of this report. Our estimates and that of Pritts and colleagues¹⁰ find that the estimates are lower in data from more contemporary prospective cohorts of women having surgery compared to retrospectively collected data. This may be explained by inaccuracies in retrospectively collected data even when pathology specimen banks are used to index a full population of surgical patients. On the other hand, it is important to note that prospectively collected data may enroll younger patients and a larger proportion undergoing myomectomies. As a rule, data for estimating rare events is volatile, which is demonstrated by the changes in prevalence estimates when we exclude studies that included hysteroscopy.

The risk of inaccuracy is especially true in understanding and estimating the degree to which morcellation method influences survival when a woman is found to have a leiomyosarcoma that was believed to be a fibroid. More nuanced data is needed to be able to include exact ages and uniform staging data at an individual level and to account for secular trends in aggressiveness of treatment and range of modalities used in treating leiomyosarcomas.

Focusing on interventions available in the United States, and excluding those that cannot be obtained here could neglect a promising intervention but does restrict the report to data that is of immediate value to women and their care providers who must make decisions among available options. We have included some interventions that are not widely available in the United States such as HIFU without MRI guidance. For any woman, her geographic location, local provider training or insurance coverage may restrict the availability of some options.

Limitations of the Evidence Base

While the literature about the effectiveness of uterine fibroids treatment has grown since the last evidence report in 2007, significant gaps in knowledge persist. The 97 studies unique intervention arms enrolled 9,179 women, an average of 98 women per study with a range of 16 to 451. Individual studies were often small and powered to address a single continuous outcome such as hematocrit or score on a quality of life scale.

Our analytic framework was created by expert consensus to reflect the outcomes that matter to women when making decisions. The available literature has substantial gaps in collecting this information as indicated by the number of studies that addressed each of our eight primary outcomes. Fibroid characteristics and symptom status were the most frequently reported outcomes, addressed in 65 percent and 59 percent of the studies respectively, though assessment techniques and measures varied. Other key outcomes including quality of life and satisfaction with outcomes (39%), sexual function (12%), and future reproductive outcomes (8%) were addressed in only a handful of studies. Detailed descriptions of subsequent treatment were reported in 29 percent of the included trials.

Little continuity exists in approaches to measuring outcomes and use of unvalidated measures is common. When data are combined across studies for a particular intervention, risk of serious rare harms cannot be fully assessed. In many instances ability to synthesize evidence across studies is absent, weak because of biased collection methods (e.g., assessors not blind to intervention), or difficult to aggregate across studies because of use of different metrics.

Paucity of “similar” articles (populations, settings, patient characteristics, and outcomes measured) also precludes efforts to pool data about characteristics of the study populations as they contribute to predicting outcomes. No studies were appropriately powered to understand whether specific groups of patients, such as those closer to menopause or with a specific symptom pattern have outcomes that are modified by those characteristics. Lastly, the lack of direct comparisons limits the information this can provide to help a woman or her care provider make an evidence-driven selection among choices in the context of the patient’s priorities.

Limitations about leiomyosarcoma data: Because cases are rare and detailed age data for non-cases is lacking, only rough models of risk by age can be produced. ²⁵ We need more prospective studies that include nuanced data such as patient and fibroid characteristics, patient age, menopausal status and surgical approach, hormonal exposures and genetic factors so that we can give more accurate estimates to patients.

Future Research

Key components of study design, analysis, and reporting remain the leading weaknesses of the literature for each topic addressed in this review. Overall, the literature identified is limited by the following gaps and problems discussed in detail in the full report. Future research should aim to remediate these concerns:

• Ability to assess internal and external validity
• Study populations of adequate size for assessing key outcomes
• Use of standard nomenclature and validated measures
• Analysis methods matched to the outcomes of interest
• Direct comparisons of treatment options
• Formal development of patient centered outcome measures for fibroid care

A range of content priorities also need to be addressed. These include the burden of disease and societal costs from loss of ability to function well in the usual family or occupational roles. Transitions associated with appearance of uterine fibroids, growth patterns, and influences on growth (e.g., concurrent medical conditions like diabetes, use of medications like hormonal contraception, influence of lactation and duration) are high-priority topics, as are predictors of symptom development and resolution. Variation in care-seeking behaviors, differences in severity at presentation, and health and quality of life outcomes are other matters that investigators should attempt to address. Likewise reproductive outcomes such as fertility and risk of future pregnancy complications are very important to women as they make decisions about fibroid treatment and the current literature is insufficient to guide choices. In addition, the current literature cannot address from trials whether disparities between white and black women in the age at appearance of fibroids and in the number and size of fibroids also foreshadows different treatment outcomes and durability of results.

In current practice, women without symptoms may forego intervention because of the general belief that care should be aimed at improving symptoms or addressing a specific clinical concern such as difficulty conceiving or recurrent pregnancy loss. Although foregoing intervention can be wise in the absence of data that the intervention will prevent future difficulties, research on the natural history of fibroids and likelihood of developing symptoms or other health effects is limited. No data is available to indicate whether use of therapeutics short of surgery might forestall or prevent future changes in fibroids or appearance of symptoms. The concept of preventive strategies is appealing. However, as long as the etiology of fibroids remains unclear, preliminary trials are not assessing lifestyle interventions, and the prospect for dietary management, nutritional supplementation, exercise, hormonal management, or other prevention trials is slim.

The clinical research agenda will likely depend on new translational research and large-scale epidemiology studies. Much remains to be learned that will require large-scale prospective observational studies of sufficient size and rigor to support time-to-event analysis of outcomes, such as that being conducted in the COMPARE Uterine Fibroids 10,000 woman registry supported by AHRQ and PCORI. These studies may afford greater power to examine effect modification and to determine trajectories of care over a reproductive lifespan for women with fibroids. Additionally, such studies will be better able to estimate both common and rare harms, including the risk of occult leiomyosarcoma. Research effort must be focused on documenting first the course and consequences of uterine fibroids using optimal imaging strategies. We must then deploy more robust statistical techniques across aggregated data in order to understand the modifiers of that course and of the effectiveness of treatment, so that we can offer women an accurate account of the likely outcome of intervention choices based on their individual status.