Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner
The FIBROID Registry

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

One Year Outcome

Among patients initially eligible for long-term followup (N = 2,112), 6-month followup data collection concluded on August 4, 2003, with 1,797 (85.1 percent) completed cases and 12-month data collection concluded on March 1, 2004, with 1,701 (80.5 percent) completed cases.

Symptom and HRQOL Outcomes

The primary measures of outcome are the symptom and health-related quality of life (HRQOL) scores from the UFS-QOL. As noted earlier, the two scales range from 0 to 100, but in an inverse manner: higher scores on the HRQOL scale indicate a better score, while a lower score is better on the symptom scale, indicating fewer symptoms. By way of reference, in the original validation, the mean symptom score for normals was 22.5 (std. dev. 21.1) and the HRQOL total score for normals was 86.4 (std. dev. 17.7).

The summary from the current study is presented in Table 14. Although sample sizes differed between the two time points, several steps were taken to prevent bias that could be present when excluding cases.

We compared scores in three populations. The first was the cleanest population consisting of only those who were available at both time points, had not had a hysterectomy in the interim and for whom scores were available. The second population included patients who had had a hysterectomy in the interim along with the first. The third population consisted of all patients such that patients who were missing a score (for any reason) were imputed the mean score of patients who had had hysterectomy (to be conservative, assuming that patients needing a hysterectomy would have had the worst scores). The mean differences etc. reported here are from this last population consisting of all 2112 patients.

At baseline, the mean symptom score was 58.93 (std. dev. 20.81). The mean symptom score at 6 months was 19.86 (std. dev. 18.60). At 12 months, very similar scores were noted, with the mean at 19.23 (std. dev. 17.94). The mean HRQOL score prior to therapy was 47.32 (std. dev. 22.92). The mean HRQOL score at 6 months was 85.05 (std. dev. 20.05) with a median of 93.95. At 12 months, HRQOL mean score was 86.68 (std. dev. 18.15). When compared to baseline, paired (dependent) t tests of pre- and post-procedure scores indicate that the symptom score improved by a mean of 37.03 points (s.e. 0.607, p < 0.0001) at 6 months and by a mean of 37.98 points (s.e. 0.5816, p < 0.0001) at 12 months.

Less than 7 percent of patients failed to improve their symptom score at 6 months and less than 6 percent failed to do so at 12 months. Ninety percent of patients had at least a 10 percent improvement at 6 months and at 12 months. Similarly, for HRQOL scores, at 6 months, 7.23 percent of patients failed to improve their score and 85.38 percent had greater than a 10 percent increase.

Table 15 presents the change in symptom scores based on the percentage change over baseline. More than 60 percent of patients had a greater than 50 percent improvement in their score at both 6 and 12 months. These same findings are presented graphically in Figure 2 (6 KB) and Figure 3 (6 KB).

At this time, there are no agreed-upon definitions of success and failure based on these scores. Patients display a spectrum of responses to the procedure as measured by this questionnaire. We are currently examining the effect of different definitions on outcome and hope to identify an appropriate threshold using a variety of techniques (e.g., by constructing receiver operating curves [ROC] of percent change in QOL score versus satisfaction/dissatisfaction).

Need for Additional Fibroid-Related Care

Another way to view outcome is in terms of the frequency with which additional medical or surgical therapy is necessary. In Table 16 it can be seen that about 7 percent of patients, at both 6 months and 12 months, needed some hormonal or other therapy for fibroid-related symptoms; most common (3 percent) were birth control pills.

Subsequent gynecologic interventions over the course of the first year are summarized in Table 17. Within the first 6 months following therapy, 3.56 percent of patients required a gynecologic intervention and an additional 5.88 percent needed interventions between 6 and 12 months. Of these interventions, the most common major intervention was hysterectomy, with a total of 49 performed in the interval from baseline to 12 months. Other major interventions included 25 myomectomies and a repeat UAE in 21. The indications for these interventions were not recorded.

Among the interventions, the second most common after hysterectomy was D&C. Again the indications are not available, but it is presumed that in most cases the D&C was done to treat fibroid expulsion and/or vaginal discharge.

Table 18 presents the need for emergency room visits and unexpected hospitalization in the first year. Emergency room visits were needed for 3.78 percent of patients within 6 months post-procedure and for an additional 1.18 percent of patients between 6 and 12 months. Unexpected hospitalization for fibroid-related problems had occurred in 2.89 percent of patients at 6 months and in an additional 2.18 percent at 12 months. The total with unplanned ER or hospital care is less than the total of those with any visit to the ER plus those with any hospitalizations, as some patients had experienced both.

Menstrual Cycles after Embolization

Changes in menstrual cyclicity are the main measure of potential effects of UAE on ovarian function. Although hot flashes were recorded within the first 30 days, new onset of hot flashes was not included in subsequent questionnaires, primarily because previous data suggested that these symptoms are usually transient and do not correlate with ovarian function. Changes in menstrual cyclicity may therefore be a more sensitive marker of ovarian function. At baseline, 2.6 percent of patients were post-menopausal by self-report. At six months after the procedure, 14 percent of patients were not having cycles and 16 percent of patients were not having cycles at one year (including those post-menopausal at baseline) (Table 19).

The causes of the amenorrhea were only asked in a revised followup form and responses are available from 171 of the 256 patients with amenorrhea at 6 months and 245 of 273 at 12 months. Of these groups, 57 percent (98/171) at 6 months and 60.8 percent (149/245) at 12 months had spontaneous amenorrhea. The balance with no periods had other causes, including hysterectomy, pregnancy and medications as the reason for amenorrhea. Of the 149 patients with spontaneous amenorrhea 12 months after embolization, 24 were self-described as post-menopausal before the procedure. Of the remaining 125 that had onset of spontaneous amenorrhea after UAE, 3 (2.01 percent) were aged 35 to 40, 15 (12.0 percent) were 40 to 45, 53 (42.40 percent) were aged 45 to 50, and 54 (43.20 percent) were over age 50.

Based on 125 patients known to have cycles before the procedure and amenorrhea after, the rate of procedure-related amenorrhea at 12 months is 7.3 percent. However, whether these patients had cycles for some time after the procedure then stopped or whether the procedure precipitated amenorrhea is not known.

Additional Outcome Measures

There were additional measures of outcome included at each interval based on questions raised in prior studies. These included assessment of sexual functioning after embolization and the presence of post-procedure vaginal discharge (Table 20). Approximately 9 percent of patients at 6 months and 12 percent of patients at one year either agreed or strongly agreed that they were experiencing unwanted changes in sexual function. Whether these changes correlate with the onset of amenorrhea (i.e., the onset of menopause) will require additional study.

Vaginal discharge had occurred at some time after the procedure in 17 percent of patients during the first 6 months after treatment and 16 percent between 6 and 12 months. From the wording of the original questions, it is unclear whether some of these were reported twice, whether there was continuing discharge in some patients or whether some patients had recurrent discharge. This subject also will require additional study.

Regardless of timing of onset, the duration of vaginal discharge was brief, with a mean duration of 11.37 days (std. dev. 14.28, median 6.5, interquartile range 3-14 days) at 6 months and 8.44 days (std. dev. 8.22, median 6, interquartile range 3-14 days) at 12 months. Roughly a third of these patients sought therapy from a physician; approximately half of those received medication for the discharge and in most patients the treatment was successful.

Finally, when patients were asked whether they would recommend the UAE procedure to family members or friends, 84.26 percent agreed or strongly agreed at 6 months and 82.01 percent agreed or strongly agreed at 12 months (Table 21), which aligns with the degree of symptom control demonstrated by the UFS-QOL scores after therapy. Additional analysis will be directed toward determining whether there are factors that predict satisfaction with the procedure outcome.

Return to Contents

Analysis of Failure of Symptom Control at 12 months

A central question with this or any therapy is how effective the symptom control is after therapy; in other words, how frequently does a patient fail to improve? Just as important is the question of whether failure of a procedure can be predicted based on demographic, medical, or procedure-related factors. At this time during the followup of the registry, these questions can begin to be addressed.

Prior to the analysis, a failed procedure must be defined. In the context of the FIBROID Registry and this analysis, a failure was defined as when a patient did not have at least 10 percent improvement in the symptom score at 12 months. Further, since the intent of the UAE procedure is to provide an alternative to hysterectomy, any hysterectomy within 12 months after embolization is also considered a failure.

Only 49 hysterectomies occurred, however, and this low number of occurrences limits the analysis. For this reason, a broader definition is used here: the procedure is considered a failure if any gynecologic intervention is needed in the first year. This is a very broad definition, but does allow the inclusion of repeat UAE and myomectomy, as well as more minor procedures, such as D&C. Using this definition, the demographic, medical, and operator- and procedure-related variables were examined using univariable and multivariable models.

In Table 22, the results of the univariable analysis are presented. The only demographic or medical history item with a statistically significant finding was race, suggesting that African-Americans may be less likely to fail to improve (p = 0.04). Fibroid morphology may play a role, as those patients with dominant serosal and intramural fibroids more likely to fail than those with submucosal dominant fibroids (p = 0.03).

Those with fewer fibroids (p = 0.0006) and larger fibroids also appear to be predisposed to failure (p = 0.0040 Chi square analysis, p = 0.0013 Wilcoxon two-sample test). Those treated with unilateral embolization are more likely to fail (p = 0.014), which is not surprising. Failure to embolize both sides will commonly result in untreated fibroids that are likely to cause continued symptoms.

As illustrated in Table 23, the multivariable analysis eliminates several of the above factors when other factors are held constant. The strongest predictor of failure is unilateral (versus bilateral) embolization, which as noted above is not surprising. There is a slightly lower likelihood of failure with increasing age, although the impact is very minor for a single year (hazard ratio 0.997 per year increase in age). However, this difference becomes more apparent when considering a 5 year increase in age (hazard ratio 0.89), a 10 year increase (hazard ratio 0.792), or a 15 year increase (hazard ratio 0.705).

Smoking history had an increased risk of failure. The reason is unclear, given that smokers tend to have lower levels of estrogen. One would therefore expect that, if anything, smoking would, like advancing age, reduce the likelihood of recurrence. Perhaps additional analysis and review of data at 24 and 36 months will provide an answer.

The findings regarding preoperative anatomy are also interesting—essentially, patients with fewer, larger fibroids are more likely to fail than patients with multiple, smaller fibroids. Again this effect becomes more apparent when considering a 5 cm increase in fibroid size (hazard ratio 1.51) or a 10 cm increase in size (hazard ratio 2.28). Similarly, with fibroid number, a patient with five fibroids compared to one has a substantially lower risk of failure (hazard ratio 0.436).

Whether these effects represent differences in the relative decrease in uterine and fibroid size and subsequent residual symptoms or abdominal wall distortion, or perhaps that subsequent interventions may be more likely in larger fibroids, is unknown. These questions may be answered as additional data is gathered.

Return to Contents

Pregnancy After Uterine Embolization

One goal of the registry was to determine the frequency of pregnancy after uterine embolization and the outcomes from those pregnancies. After only 12 months of followup, it is clearly too early to expect many pregnancies.

During the first 6 months after UAE, there were 4 pregnancies, 2 ending in miscarriage and 2 ongoing. By 12 months, there had been 12 pregnancies, 4 ongoing. Among the 8 that were not ongoing, there were the 2 miscarriages already mentioned, one termination, four reports of no loss, and one case missing. Of the four reports of no loss, only two reported information on delivery, one by planned cesarean section and one by unplanned cesarean. No other pregnancy-related adverse events have been reported thus far. Data collection continues on other pregnancies.

While data continue to be collected on pregnancies occurring among patients in the registry, the numbers are too small for analysis at this time. Because each patient was asked about their interest in future pregnancy at the time of registration in the study, those interested in pregnancy can be characterized to some extent.

In Table 24 and Table 25, univariable and multivariable analysis was used to determine the odds ratios for patients interested in pregnancy. In the first table, those definitely planning to be pregnant are compared to the rest. In the second table, those definitely planning to be pregnant and those who want to keep the option open are compared to those not interested.

As expected, the older the patient, the less likely she is to plan on pregnancy. Those without children are much more likely to want to become pregnant (odds ratio 5.28), as are those that had never been pregnant (odds ratio 3.78). Patients with suspected infertility were also much more likely to be interested in pregnancy (adjusted odds ratio 2.61).

Similar findings are evident from Table 25. In terms of demographic factors, African-American women are 74 percent more likely to want to get pregnant than non-African-American women and are 27 percent more likely than non-African-American women to want to keep their options open. Among all patients, every year of increase in age, women are 26 percent less likely to want to get pregnant but only 20 percent less likely to want to keep the option open.

Although one of the primary interests of the FIBROID Registry was to examine the impact of UAE on fertility and pregnancy-related outcomes, these data suggest the following:

  1. Given the small number of women who stated definite plans for pregnancy (191 at enrollment), their relatively advanced age (mean 37 years), their underlying fibroids, and the high proportion with previous diagnoses of infertility, the number of pregnancies during followup is likely to be quite small; even a 50 percent fecundity rate (a very optimistic estimate in this population) would mean fewer than 100 pregnancies, generating very little power to draw any conclusions about safety regarding rare pregnancy outcomes, especially if none are observed.
  2. Women definitely planning pregnancy are, not surprisingly, younger, but "young" in this case, is relative—most are over 35. Interestingly, they are more likely to be African-American, even after adjusting for age (thus accounting for potential confounding secondary to the fact that African-American women with fibroids are younger than white women, in this registry and in every other reported study of fibroids). They are more likely to be nulliparous, and to have a previous diagnosis of infertility. Maternal age over 35, African-American race, nulliparity, and a history of infertility are all associated with an increased risk of adverse pregnancy outcomes, including preterm delivery, intrauterine growth restriction, stillbirth, and cesarean section. The important implication of this is that women undergoing UAE who plan on attempting pregnancy are already in a group at high risk for adverse pregnancy outcomes, even before accounting for their underlying condition (which is clearly associated with adverse pregnancy outcomes), and the planned procedure. Any attempt to compare the results of pregnancies after UAE to published data regarding other treatments for fibroids that does not attempt to adjust for these baseline characteristics will be inadequate.
  3. We believe that, given these considerations, it is unlikely that the FIBROID Registry is going to be able to provide definitive data on the effect of UAE on fecundity or pregnancy outcomes, especially if reported adverse event rates are low. Achieving sufficient numbers will likely require population-based studies, perhaps outside of the United States. Patients and practitioners should be informed of this continuing uncertainty.

Return to Contents

Chapter 4. Discussion

For many fibroid therapies there has been relatively little study of outcomes and a study with the breadth of the Registry has never been completed. A recent systematic review of the literature conducted as part of AHRQ's Evidence-based Practice Center program concluded that the available evidence on management of uterine fibroids was of poor quality.1,22 Even the best studies of hysterectomy, which reported both short- and long-term outcomes, and used standardized, validated instruments for measuring physical and psychological outcomes4,5,63 are limited by inclusion of women undergoing hysterectomy for a variety of indications.

The number of women undergoing hysterectomy for fibroids was relatively small in these studies: 3163,1454, and 6255, and results were not reported separately by indication. Although the authors of all these studies stated that results did not differ among indications, the potential impact of factors such as preoperative size of the uterus, number and location of fibroids, and specific type of symptoms on outcomes is not reported. It was in the context of this history of limited clinical research of fibroid therapies that the Registry was conceived.

The FIBROID Registry was able to enroll over 3,000 women undergoing a procedure for symptomatic uterine fibroids, and thus represents the largest available cohort of women undergoing a single therapy for this condition. Despite the large number of data elements collected, data were complete for over 95 percent of patients, with the exception of the most complex data, the UFS-QOL and uterine anatomy parameters. This is even more impressive given the voluntary nature of the registry, with no direct registry resources being provided to sites. Given that new procedures can be introduced without the need for approval from the FDA or other regulatory agencies, voluntary registries organized by specialty medical societies early after the introduction of a new procedure may be one of the few ways to collect prospective data on outcomes in ways that avoid the limitations of single-center studies or studies based on administrative data.

For example, registries of laparoscopic cholecystectomy organized by State or regional medical societies were instrumental in providing data about the relationship between safety and physician experience.64 One of the particular strengths of this registry is the availability of data on patient characteristics, such as obesity and uterine anatomy, which may affect the difficulty of the procedure, and therefore allow more meaningful risk adjustment than is offered by administrative data.65 It is also unusual in the breadth of its participating sites and thus provides a window into the outcomes experienced in a broad spectrum of practices of varying size and among operators of varying experience. The experience of SIR with the FIBROID Registry can be a model for other specialty societies as new procedures are introduced into practice.

Another key aspect of this registry is the use of a disease-specific validated questionnaire to measure symptom status and health-related quality of life.19 Previous studies have used generic quality-of-life instruments, which may not be as sensitive to small changes in QOL. Because most of the clinical impact of fibroids is based on subjective symptoms, it was important to try to provide the most reliable and repeatable measure of that status for evaluation of patient outcome over time.

Although missing data elements were more common with patients entered later in the course of the registry, the proportion of data elements which were complete remained above 90 percent throughout, a proportion that is as good as or better than other voluntary registries.66 Retention in the cohort agreeing to long-term followup was 80 percent at 12 months, again comparable to other registries.

The baseline data in the registry suggests that women undergoing UAE are similar in many measurable respects to women undergoing surgical management of fibroids, particularly hysterectomy. These differences are also similar to those reported in other large series of women undergoing UAE, such as the Ontario Uterine Fibroid Embolization Trial.67 In particular, differences between African-American and white women in age, body mass index, uterine anatomy, and symptom severity are similar to reported differences in large series of women undergoing a variety of invasive procedures for leiomyomas, including hysterectomy5 and myomectomy34 as well as UAE. Elucidation of the multiple potential genetic, environmental, cultural and socioeconomic causes of this phenomenon should be a high research priority.

Return to Contents

Thirty-day Outcome

Despite the diversity of sites and operators, the technical details and short-term outcomes were remarkably similar across sites and operators. We found no difference in the length of procedure, length of stay, incidence of adverse events in hospital or in the first 30 days after therapy based on site experience. Similarly, there were few predictors of adverse events related to baseline demographic factors. There were relatively small increased odds of an adverse event associated with African-American race, smoking, prior medical therapy, existing comorbidities, and prior therapies (OR range 1.14 to 1.32).

The incidence of all adverse events was quite low and serious complications were rare (0.66 percent in hospital, 4.8 percent post-discharge). This is particularly remarkable when the definition of major complication is considered. In this study, any patient with an initial hospitalization longer than 2 days, or post-discharge reevaluation in the emergency room or readmission were considered to have a major complication, even if the reevaluation did not result in admission or if the admission was for only 1 night.

Only three patients underwent a hysterectomy in the first month after the procedure (0.09 percent) and only 32 patients had additional surgical interventions of any type (1.17 percent). One concern early on with UAE was that it might be dangerous and in particular that uterine injury leading to major infection and hysterectomy might be a substantial risk. The current study reinforces the results of earlier single-center studies.9,10,13,68

This study's findings regarding adverse events differ from those from studies that have assessed the risk of leiomyoma surgery. For hysterectomy, increasing uterine size is associated with increasing risk of complications after hysterectomy, particularly blood loss.33,69 Febrile morbidity after hysterectomy is associated with abdominal approach and blood loss greater than 750 cc.70 Similar findings have been noted with myomectomy, with increased odds ratios for uterine size (OR 1.86), number of leiomyomas (OR 1.83) and comorbidities (OR 2.77).34

The lack of an impact of these same factors for uterine embolization suggests that embolization may be an attractive alternative to surgical therapies for fibroids for patients who may be at increased risk for operative complications. However, because the incidence of adverse events was rare, we ultimately had limited power to determine reliable predictors of complications.

In a recent review of severe complications after hysterectomy71, operative approach also impacted the frequency of complications, with laparoscopic and vaginal procedures associated with a higher incidence of severe complications. That study gathered data on 37,512 hysterectomies performed in the United Kingdom in a 12-month period. The investigators defined severe complications as death, deep venous thrombosis, pulmonary embolism, myocardial infarction, renal failure, cerebrovascular accident, septicemia, necrotizing fasciitis, secondary hemorrhage, fistula, ureteric obstruction, and visceral damage. All of these would be classified as a Class D, E, or F complication in the SIR system in use in the current study.

The incidence of severe complications from hysterectomy for all indications was 3.5 percent and the risk of a complication was higher when the indication was fibroids (4.4 percent, adjusted odds ratio 1.3). This complication rate is higher than those with similar severity associated with UAE in the current study. There were 14 deaths in the hysterectomy study for a crude mortality rate of 3.8/1000. By contrast, there were no deaths in the current UAE study and D, E, and F complications occurred in 1.4 percent of patients.

Return to Contents

Six- and 12-month Outcomes

As is evident from both the mean symptom and quality-of-life score changes, there is a dramatic positive impact from UAE for most patients. More than 60 percent of patients had a 50 percent or greater improvement in both symptom and HRQOL scores at 6 and 12 months. Moreover, nearly all patients had a positive impact, with 90 percent showing a 10 percent or greater increase in symptom score and 87 percent showing a similar improvement in quality-of-life scores. The proportion displaying a marked improvement in scores also is evident in the graphical representation of the results (Figure 2 [6 KB] and Figure 3 [6 KB]).

Thus, over the course of the first year after treatment, the symptoms associated with fibroids respond well to embolization and this change has a dramatic impact on the broad range of elements of health-related quality of life. Despite the wide range of patients treated in this study and the various sites where they were treated, there are remarkably uniform results. The degree of improvement is reflected in the high degree of satisfaction among treated patients.

To view the outcome from another perspective, the frequency of additional fibroid therapy after UAE was measured. Only 5 percent of patients needed continued or adjunctive use of hormone-based therapy and 5 percent needed additional gynecologic interventions. The low intervention rate speaks to the efficacy of the procedure in controlling symptoms, but also is a surrogate measure for detecting late complications that might require intervention. Another measure of late complications is the frequency of unplanned hospital admissions or emergency room visits; in this study only 4 percent needed such care by 6 months and less than 3 percent needed it between 6 and 12 months.

As noted earlier, there is little information in the gynecologic literature regarding this type of re-intervention or additional care beyond the immediate peri-operative recovery of 30 days. Thus it is difficult to gauge whether the UAE procedure represents an improvement in terms of late interventions. It is clear from the earlier discussion that any uterine-sparing therapy can result in the need for additional care and intervention. Further analysis is needed on the current data set to determine how commonly additional interventions are associated with a failure of improvement in symptom or quality-of life-score.

In the registry, 7.3 percent of patients had cessation of menses 12 months after embolization, although the structure of the questionnaires do not allow for an exact measure of the rate at which permanent amenorrhea occurred as a result of the treatment. Nearly all these patients were aged 45 or greater. Prior studies have noted the onset of menopause as a result of the procedure, but the range has been from 2 to 5 percent in most9,10 and in nearly all cases this has occurred in women nearing the typical age of menopause.

Other negative outcomes included the occurrence of short-term vaginal discharge in 17 percent of patients. No further information regarding etiology is available. In one study of outcome from hysterectomy and UAE at one year17,yeast infection over the first year was the most common adverse event after 30 days in both groups (5 percent of UAE patients, 10 percent of hysterectomy patients). Chronic vaginal discharge related to endometrial injury can occur after UAE9, but that type of discharge typically lasts for months rather than the mean duration of 11 days noted in this study.

Finally, unwanted changes in sexual function did occur in 10 to 12 percent of patients. At this time, we have not performed an analysis of the relationship of those with this symptom with any baseline factors or other adverse events such as amenorrhea.

There were relatively few predictors of outcome among either the measured baseline characteristics or the procedure itself. The strongest indicator is a technical failure (completion of only a unilateral embolization when most patients require bilateral treatment). The size and number of fibroids also influence outcome, but the impact is only moderate, with a hazard ratio of 1.086 per cm for fibroid size and 0.847 for each additional fibroid. This finding will need additional analysis in separate studies that have more detailed imaging outcome to determine whether it is a failure to infarct, a failure to shrink, or other factors that impact this observation.

Similarly, the reason for a greater likelihood of failure in younger patients is not self-evident and will require additional study. Further, as additional data are collected in the study, it is likely that the predictors of late failure (recurrence) may change. At this time it is too early to assume that reliable predictors of failure or recurrence are known.

Return to Contents

Potential Limitations

The FIBROID Registry offers an opportunity to collect prospective data on symptom recurrence/persistence, development of new symptoms, and subsequent need for therapy, adverse events, and disease-specific QOL (using a validated instrument) on by far the largest group of women ever to undergo a single procedure for uterine fibroids, at a relatively low cost. However, we recognize that there are potential limitations to both this cohort, and the proposed approach.

One limitation is loss-to-followup. At 6 months, data were available for 85 percent of the original cohort, and at 12 months, for 80.5 percent. These rates are comparable to other registries, especially in younger, more mobile populations, but they still raise the possibility that the effectiveness of UAE may be somewhat lower than the current data suggests, especially if women lost to followup are more likely to have had poorer responses to their procedure.

Although there is anatomic information from baseline evaluation of the patients in the registry, these data were not validated by a core imaging lab. Further, there is no imaging followup included in the registry. The investigators believe that clinical outcomes, as measured by a validated, reliable disease-specific quality-of-life instrument, and need for subsequent interventions, are most important. The lack of such data is one of the major limitations of the available literature on fibroids.

Moreover, numerous studies have shown an acute decrease in size after UAE.9,10,72 Although the degree to which this decrease persists is certainly of scientific interest, regrowth in the absence of clinical symptoms is of debatable importance. Also, at least one study of outcome has shown that the degree of uterine or fibroid volume decrease does not predict symptom control.73 Because funds for followup imaging were not available in the registry, the questions regarding the relationship of uterine anatomy and outcome will require new studies.

The registry has no comparison group, which limits its utility. Clearly, randomized trials comparing UAE to other treatments for fibroids are the ideal and preferred method for determining relative efficacy. However, previous attempts at randomized trials of hysterectomy versus conservative therapies for menorrhagia have been extremely difficult to perform in the United States, primarily because of extremely low recruitment, which investigators have attributed to strong patient and physician preferences and beliefs for and against hysterectomy.

In one recently published small "randomized" trial from Spain, patients were randomized to two groups: one group had hysterectomy without being told of the alternative of UAE, and one was given the choice of UAE versus hysterectomy; only one patient in the second group elected hysterectomy.74 It seems unlikely that a randomized trial of hysterectomy versus UAE performed under U.S. standards of informed consent would be feasible in the near future.

Parallel registries of myomectomy and hysterectomy would be an alternative approach, but given that the sites participating in this registry were selected primarily for their experience and expertise in UAE, the volume of gynecological surgeries performed at those centers may or may not be comparable. In addition, while many of the same demographic, anatomic, and QOL measures can be used, data on events of the procedure itself would require design of additional case report forms, and recruitment and training of new gynecological investigators and coordinators.

Because the demographic characteristics of the FIBROID cohort are closer to those typical of women undergoing hysterectomy than to myomectomy, we will at least be able to qualitatively compare changes from baseline QOL measures to those reported in published large cohort studies of women undergoing hysterectomy for benign indications, including fibroids.4,5 Furthermore, the data obtained from the registry will serve as a resource for those planning additional studies.

The validity of physician reports of short-term outcomes, which have already been recorded, is another potential source of misreporting. Similarly, the validity of patient self-report of QOL and subsequent interventions is also unknown. It is important to note that physician underreporting of adverse events within 30 days of the procedure will not affect estimation of event rates in the long-term followup study proposed here. Inaccuracies of site reporting would affect only the planned multivariable analyses of predictors of outcomes.

Thus, for the purposes of the proposed long-term followup study, the use of a validated, reliable QOL instrument and patient self-report of additional interventions should have more than adequate validity. Use of adverse event data from the initial procedure and 30-day followup, which has not been validated, might affect interpretation of predictors of long-term outcomes, but the likely direction of any bias introduced by this would result in underestimation of the beneficial outcomes of uncomplicated UAE.

We recognize that the results obtained from the FIBROID Registry, whether for short- or long-term outcomes, may not be universally generalizable. Physicians participating in a voluntary registry may be quite different from those who decline to participate. As discussed above, patients undergoing a procedure early in its adaptation by practitioners may be different from patients who undergo a procedure after it is well established. Lack of initial insurance coverage, especially by Medicaid, may make this group different in important socioeconomic characteristics. Outcomes may be different early in the "learning curve" of a given physician or hospital. Patients who are lost to followup may differ from those for whom data is available.

Many of these concerns apply to randomized studies or any other clinical studies. Given the extent to which the baseline characteristics of these patients have been recorded, there is sufficient detail to allow assessment of this population with those treated in other studies.

Return to Contents

UAE FIBROID Registry Project Dissemination Plan

Upon completion of scientific analysis and interpretation of data, the UAE FIBROID Registry Steering Committee is planning for the submission of three manuscripts originating from baseline, short-term and mid-term followup. Publication plans spanning the next year are outlined below.

  1. The Fibroid Registry for Outcomes Data (FIBROID) for Uterine Artery Embolization—Baseline Patient Characteristics: The purpose of this manuscript will be to describe the baseline patient characteristics from the national sample of women contained in the UAE FIBROID Registry undergoing uterine artery embolization for uterine leiomyomas.
  2. Short-term Outcomes—UAE at 30-days Postprocedure: The purpose of this manuscript will be to examine UAE FIBROID Registry data at 30-days post-procedure.
  3. One-year Outcomes: The purpose of this manuscript will be to examine UAE FIBROID Registry data at 12-months postprocedure, including the analyses included here.
  4. Future Publication Topic Areas: It is important to note that the longest prospective followup for any treatment of fibroids is 2 years. Accumulation of at least 3 years of followup data would allow dissemination of a wealth of new knowledge about fibroid treatment in the medical/scientific literature. The UAE FIBROID Registry Steering Committee strongly believes that to achieve maximum impact of the data contained in the registry all efforts should be made to publish registry data in the gynecological literature and other high-impact medical journals.

Opportunities to publish long-term patient outcomes from cases enrolled in the UAE FIBROID Registry will involve further exploration of how long the procedure lasts and the factors that may predict which patients will receive long-term relief and those who will likely experience recurrence of symptoms. Other topics under current consideration include:

  • Characteristics of women seeking pregnancy.
  • 24-month outcomes—need for additional therapy.
  • 36-month outcomes—need for additional therapy.
  • Further analysis of baseline and outcome factors and their relation to quality-of-life subscales, symptom and overall QOL scores.

Return to Contents


The registry provides a unique window into the outcomes from uterine embolization as practiced in a broad variety of clinical settings. While the results of this study will not provide women with information as to whether UAE is better or worse than other available treatments, it will provide valuable information on the likelihood of important health outcomes. In addition, researchers planning comparative studies, especially randomized trials, will be able to use the data from this study to ensure that future studies collect the right data, are adequately powered and are conducted efficiently.

Based on the data presented here, it can be concluded that embolization is safe, with complication rates that compare favorably with those reported for surgical interventions for leiomyomas, and that, for the majority of patients, the procedure requires only limited hospitalization and a short recovery period. The large majority of patients have marked improvement in symptoms and health-related quality of life at 6 and 12 months after the procedure. We await additional longer-term data from the ongoing followup post-treatment to further assess the durability of symptom control.

Return to Contents
Proceed to Next Section


The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care