The Comparing Options for Management: PAtient-centered REsults for Uterine Fibroids (COMPARE-UF) registry: rationale and design.

Nine Clinical Centers with expertise in UF care are collaborating institutions: Mayo Clinic Collaborative Network, University of California Fibroid Network, Henry Ford Health System, University of Mississippi Medical Center, the University of Michigan, University of North Carolina, Partners Healthcare/Harvard Medical School Collaboration, Inova Health Systems, and the Department of Defense Collaborative Sites ( Table 3 ). These centers represent a mix of academic and community practice settings and were chosen, in part, to maximize geographic and demographic diversity of enrolled patients ( Table 3 ). Both rural and urban populations are well represented, and most centers are affiliated with networks of community-based primary care. The Figure displays the location of core Clinical Centers and key collaborating affiliated subsites, along with the county-specific proportion of African American women among women aged 20–54 years by quintile (derived from 2012 US Census estimates). 27 Five of the centers are located in counties in the highest quintile, with 2 of the others in the 4th quintile. Other racial/ethnic groups are well represented ( Table 3 ). Initial COMPARE-UF recruitment began at 3 vanguard sites in November 2015 and will continue through at least March 2019. As of January 23, 2018, 2031 COMPARE-UF subjects have been enrolled. In contrast to nationwide statistics where hysterectomy predominates, 8 , 9 the COMPARE-UF sites have increased enrollment for women seeking alternatives to hysterectomy ( Table 4 ). Hysterectomy was elected by 38% of participants; myomectomies comprised 46% of all procedures; laparoscopic or robotic myomectomies were performed most frequently and accounted for 19% of all procedures ( Table 4 ). The sites have also recruited younger women where 16% of participants are <35 years old, and 40% are <40 years old ( Table 4 ). Similarly, despite comprising only 13% of the US population, African American women comprise 42% of enrollment in COMPARE-UF ( Table 5 ). Similar or greater numbers of African American women elected each type of myomectomy and uterine artery embolization compared with white women ( Table 5 ). The COMPARE-UF Protocol Committee formulated both baseline and follow-up questionnaires to meet the key objectives of the study ( Supplementary Figures 1 , 2 , and 3 ). These instruments may be useful for other CE studies of fibroids tumors.

Comparing the safety and effectiveness of UF treatment options is a major research priority that reflects the substantial burden of UF shared by patients, providers, insurers, and society. 10 , 11 , 28 It is especially important to gain CE data regarding alternatives to hysterectomy that have been understudied because alternatives are much less commonly performed than hysterectomy but are sought strongly by many women and especially African American women. 1 , 8 , 9 Early recruitment data for COMPARE-UF suggest that this registry is poised to answer some of these key CE questions. The fact that more than one-half of the enrolled women underwent an alternative to hysterectomy provides a much needed source of data for the examination of the safety and efficacy of these uterine-sparing treatment options. Similarly, the fact that a sizable proportion of women are 40% of enrollees suggests that data for this most significant subpopulation can be obtained and that potential differences in outcomes by race can be studied. In contrast to studies in which outcomes are assessed from government or insurance databases that lack detailed clinical data, COMPARE-UF can determine whether outcomes are similar in women with similar baseline symptoms or equivalent uterine anatomy. There are a number of specific factors that contribute to uncertainty about optimal treatment choices for UF, including the diverse clinical presentation of UF, the limited utility of administrative data (ie, most women with fibroid tumors are not covered by Medicaid, a common administrative data source for other conditions 29 ), lack of sufficient relevant clinical detail, 30 and treatments that frequently are either off-label, (such as surgical procedures), or involve the use of devices that typically require less rigorous preapproval research compared with drugs. 31 – 33 Race and ethnicity also play a role with black and Hispanic women more often undergo inpatient, as opposed to ambulatory, surgeries, and black women both having higher rates of myomectomies than other women and prioritizing uterine sparing options. 1 , 34 , 35 Additionally, recruitment into randomized trials for UF or abnormal menstrual bleeding, which typically is the goal standard, has been historically difficult, especially in the United States. 36 – 38 COMPARE-UF will address this knowledge gap by enrolling a large, nationally representative sample of patients who undergo procedures for UF. The registry collects detailed information on UF characteristics, symptoms, and quality of life before the procedure, which allows for robust adjustment of CE analyses. This infrastructure and methods will provide a basis for expansion into other therapeutic treatments. Diverse patient data from women across the United States are being compiled to provide patient-centered analyses for UF treatment. Given the huge burden imposed on women by UF, these results are urgently needed.

The COMPARE-UF registry ( NCT02260752 , clinicaltrials.gov ) enrolls and follows women who elect procedural therapy for symptomatic fibroids. Neither the women nor their healthcare providers have limitations on their choice of therapy within the study. Baseline, postprocedural, and annual follow-up data are self-reported. Centralized data abstraction takes place for the baseline imaging report, any operative or procedural note, and pathology reports, where relevant. The registry also has a dedicated statistician, who will monitor ongoing recruitment and data analysis issues. COMPARE-UF has 3 initial CE objectives for procedural interventions: (1) to compare safety and efficacy in terms of durability of symptom relief and the need for additional treatment for UF, (2) to compare the impact of UF procedures on ovarian function (including ovarian reserve and time to menopause), ability to become pregnant, and maternal and neonatal outcomes of pregnancy, and (3) to compare the effectiveness of UF treatment options within key demographic subpopulations. The registry is funded by AHRQ, in collaboration with PCORI, with administrative and scientific oversight provided by AHRQ. The Duke Clinical Research Institute serves as the Research Data and Coordinating Center. Initially, a Registry Steering Committee oversaw all governance and scientific aspects of COMPARE-UF and advised the Executive Committee on issues including protocols, recruitment and retention, and strategies for sustainability. The Registry Steering Committee is comprised of the Principal Investigators of the Research Data and Coordinating Center, the participating Clinical Centers, and the Center for Medical Technology Policy, which coordinates all stakeholder activities, project officers from AHRQ and PCORI, a patient stakeholder, and 2 external advisors. At the start of funding year 3, the Executive Committee was given final governance authority, and the Registry Steering Committee shifted to an advisory role. The Executive Committee is comprised of the Principal Investigator and a coinvestigator of the Registry and Data Coordinating Center plus a representative clinical site Principal Investigator. Finally, much work of the registry is done within key committees, which includes protocol, publications, bio-specimens, and conflict of interest committees. The Stakeholder Advisory Group is an independent body comprised of approximately 30 members, which includes patient and consumer representatives, clinicians representing relevant professional societies, representatives of Federal agencies that are involved in women’s health, healthcare insurers/payers, health system administrators, and manufacturers of devices and pharmaceuticals for the treatment of UF. The diverse perspectives of these advisors are important to ensure that COMPARE-UF is relevant to the broader fibroid and healthcare community. In addition to providing input on study design and protocol development (thereby, helping to ensure that the outcomes selected, population being studied, and comparators selected are patient-centered), this group advises on strategies and mechanisms to reach vulnerable and underrepresented racial and ethnic groups and effective dissemination and translation of study findings to both clinicians and the lay community. Stakeholders are and will continue to be engaged through annual face-to-face meetings, quarterly web conferences, online surveys at critical junctures for decision-making, and targeted key informant interviews with a subset of relevant experts both before and after meeting. Patient engagement is essential to the design and conduct of COMPARE-UF. Patients have contributed to the prioritization of research questions and selection of outcomes. Continuing this philosophy, the COMPARE-UF website ( http://www.compare-uf.org/ ) keeps participants updated about the progress and results of COMPARE-UF. An additional online portal provides a platform through which patients can enter follow-up information directly. The patient reporting of outcomes is not only efficient, in terms of minimizing resources, but also is a natural approach to collecting outcomes that are patient centered. Investigators recruit consecutive eligible patients who are scheduled to undergo procedural interventions for UF at a COMPARE-UF Clinical Center or affiliated site. Initial Institutional Review Board approval was received at Duke University on October 30, 2014, and the protocol was then approved by each participating institutions’ Institutional Review Board. Recruitment for COMPARE-UF began in November 2015. Eligible women are at least 18 years old, premenopausal, defined as having had at least 1 menstrual period in the preceding year, and have documented uterine fibroids. Data are collected on the following interventions: (1) hysterectomy, (2) myomectomy (abdominal, hysteroscopic, vaginal and laparoscopic/robotic), (3) endometrial ablation, (4) radiofrequency fibroid ablation, (5) uterine artery embolization, and (6) magnetic resonance guided focused ultrasound ( Table 1 ). In early 2017, some centers observed that progestin-releasing intrauterine devices were being used increasingly at their institutions as a UF treatment instead of endometrial ablation. Therefore, in September of 2017, therapeutic use intrauterine devices were added to the list of COMPARE-UF interventions ( Table 1 ). Previous fibroid treatment is not an exclusion criterion. Full enrollment criteria are detailed in Table 1 . Depending on local regulatory approvals, eligible women may be consented and complete the baseline survey in person, via telephone, or through the electronic COMPARE-UF online portal. To monitor data quality and enrollment trends, ensure generalizability of registry findings, and facilitate statistical procedures to address potential biases, site coordinators also collect deidentified aggregate data on all eligible subjects, which included age, race, insurance status, 3-digit zip code, procedure type, and procedure date within local institutional review board approval and guidelines. Table 2 provides an overview of the data collection for enrolled participants. We will collect patient-reported outcomes annually for a minimum of 24 months after the index treatment with a goal of extending follow up indefinitely, if additional funds are obtained. All data are entered into a secure, password-protected web-based data portal. Enrolled patients receive electronic reminders to complete the follow-up survey via the portal, unless they expressed preference for a telephone interview or paper forms. Trained interviewers from the Duke Clinical Research Institute can complete surveys if requested by patients. If a patient appears to be lost to follow up, the local Clinical Center is asked to try to follow up with the patients and to review local medical records to identify additional treatments or adverse events since last contact. Because patient-centered outcomes are a key mandate for COMPARE-UF, validated measures of HR-QOL are collected at each time point ( Table 2 ). Specific instruments include (1) the uterine fibroid symptom quality of life questionnaire, the validated disease-specific HR-QOL measure; 7 , 12 – 15 (2) the EuroQOL 5D 5L, an instrument for general HR-QOL; 16 (3) the Patient Health Questionnaire–2, a screen for clinical depression; 17 , 18 and (4) the Menopause Rating Scale, a measure of climacteric symptoms. 19 , 20 Short-term patient-reported outcomes include the time from procedure to resumption of usual activities, work days missed after the procedure, and the 4 validated measures of HR-QOL. Assessed clinical outcomes include procedural complications (pain, bleeding, infection, wound complications, neuropathy) and postprocedural hospitalization (reoperation, bleeding, infection, and thromboembolic events). Additionally, the incidental diagnosis of cancer is assessed. At the annual posttreatment follow-up evaluation, we will assess the appearance of new fibroid symptoms, additional medical or procedural fibroid treatments, and new medical illnesses. Finally, key reproductive outcomes that include menopausal status, infertility, recurrent pregnancy loss, and pregnancy outcomes (conception, live birth, gestational age) will be assessed. Impairment of ovarian reserve is both a possible complication of fibroid therapies and an additional mechanism of symptom relief. Therefore, a subset of COMPARE-UF subjects is enrolled in a substudy to assess ovarian reserve. Serum samples for assay of anti-Müllerian hormone, which is a proxy measure of ovarian reserve, are collected at baseline and 1 year after treatment. This substudy is open to all women who are <45 years old. However, because fertility impairment and reproductive outcomes are of primary importance to women who elect uterine sparing therapies, the study statistician tracks enrollment by age and treatment type to optimize the power to address reproductive outcomes. Initial analyses will examine clinical characteristics and demographics of patients who receive specific UF terventions. Subsequently, comparative outcomes will be assessed. Although hysterectomy will be the primary comparator for fibroid symptom relief, myomectomy will also be used because many women with UF desire uterine conservation and/or future pregnancy. The goal of the registry is to address multiple hypotheses that have differing power and that depend on the characteristics of the women who are enrolled in the registry. Therefore, sample sizes are not based on the power for a single prespecified comparison. The projected sample sizes by March 2019 are approximately 1200 myomectomies, 1000 hysterectomies, 200 endometrial ablations, and 200 uterine artery embolizations. Although COMPARE-UF will continue to enroll patients beyond this point, the study will have good power for many important comparisons by this time. For example, we estimate >90% power for comparisons that use the uterine fibroid symptom quality of life questionnaire at 12 months for the 4 in-procedures listed earlier. 15 Because COMPARE-UF is an observational registry, we use propensity score methods to adjust for confounding, particularly confounding by indication. Propensity score methods have been favored for separating the design phase, where adjustment occurs, from analysis of outcomes. 21 The propensity score is a summary measure of the differences in baseline characteristics for patients who undergo different UF treatments. It can be used to create balance or comparability through matching, stratification, or inverse weighting. 22 In general, COMPARE-UF will use inverse propensity weighting, 23 although large differences in the treatment groups can create extreme weights that may indicate alternatives such as matching, trimming, or overlap weighting. 24 , 25 These specific options for data analysis will be determined by the context of the specific comparison. Deidentified data collected on all eligible subjects will evaluate and adjust for differences between participants and those lost to follow up and will be used to adjust for any differences between participants and nonparticipants (nonresponse bias) by inverse weighting. 26 Because COMPARE-UF’s mission patient-centered, it is essential that results be accessible, generalizable, and useful to diverse stakeholders. Thus, in addition to traditional peer-reviewed publications and presentations, we plan to collaborate with our Stakeholder Advisory Group, to identify nonacademic routes of dissemination that include newspapers, magazines, radio, television, websites of advocacy groups and social media, to partner with specialty societies to formulate evidence-based guidelines and quality metrics, or to reassess research priorities given the results of COMPARE-UF. The vision is that COMPARE-UF will provide a framework for many studies of uterine fibroids beyond the initial funding period. Given expertise on fibroids in high-volume Clinical Centers, this network provides the infrastructure for other projects that include postmarketing surveillance for new drugs and devices and genetic studies of UF.