3Results

Methodological Issues

Numerous factors prevented us from estimating the overall likelihood of beneficial and adverse outcomes from myomectomy, including:

1. Variable and often nonstandardized methods for reporting pre- and postintervention symptoms: The lack of standard definitions for reporting menstrual blood loss or symptoms, as well as variability in the timing of the measurement of these symptoms, prevented any meaningful comparison between studies. Without knowing the severity of the baseline symptoms, measured by a common metric, readers cannot assess the impact on those symptoms of two different interventions or the same intervention performed in different places and settings.
2. Variations in surgical technique, expertise, and experience: Adjunctive techniques appear to influence short-term outcomes. Randomized trial data suggest that the use of gonadotropin-releasing hormone (GnRH) agonists prior to surgery (Lethaby, Vollenhoven, and Sowter, 1999) and the use of vasopressin during myomectomy (Fletcher, Frederick, Hardie, et al., 1996; Hutchins, 1996) reduce blood loss. The use of barrier technologies may reduce adhesion formation, which in turn may help subsequent fertility (Farquhar, Vandekerckhove, Watson, et al., 2000). Other details of surgical technique that are not reported could conceivably affect short-term outcomes, especially complications. Surgical volume or experience with the procedure also may influence outcomes -- a correlation between volume and reduced complications has been shown for multiple surgical procedures (Harmon, Tang, Gordon, et al., 1999; Ho, 2000; Yao and Lu-Yao, 1999). This may affect the generalizability of findings for a particular procedure -- outcomes for specialized procedures, such as laparoscopic myomectomy, may not be as good for surgeons with less experience, or less volume, compared with the group reporting their results. In this case, even with the best quality reporting, one can only judge the outcomes of a particular procedure performed in a specific setting by a specific surgical team, not the outcomes of the procedure itself.
3. Variable length of followup and variations in reporting length of followup: Again, comparison was prevented by the lack of common times for measuring postintervention outcomes or by imprecise reporting.

Summary

Excess blood loss, possibly requiring transfusion, is a risk with all methods of myomectomy. Randomized trials suggest that the use of GnRH agonists prior to surgery (Lethaby, Vollenhoven, and Sowter, 1999) or vasopressin during surgery (Fletcher, Frederick, Hardie, et al., 1996; Hutchins, 1996) result in statistically significant differences in estimated blood loss; however, the clinical significance of this difference is unclear. Hysteroscopic myomectomy carries the unique risks of uterine perforation and fluid/electrolyte disturbances. There is little evidence on long-term risks of myomectomy. Several peer reviewers pointed out case reports of serious adverse events, such as death from vasoconstriction during use of vasopressin or uterine rupture in pregnancy following laparoscopic myomectomy. One limitation of our search strategy is that such case reports were not included. However, without data that would allow estimation of the frequency of such events (i.e., the number of serious adverse events divided by the total number of procedures), quantification of risk is impossible. Adverse events, including death, are possible with all invasive procedures and with many noninvasive therapies. It is just as important to have high-quality data on the frequency of adverse events as it is to have high-quality data on the effectiveness of procedures in estimating relative benefits and risks.

Case series of laparoscopic and hysteroscopic myomectomy reported consistent improvement in symptoms related to bleeding within the first 6 months of treatment. Although published studies of abdominal myomectomy that met our criteria do not provide data on symptomatic relief, there is no evidence to suggest that results from an abdominal procedure would be different from the results of a laparoscopic procedure in terms of symptoms. Earlier case studies cited in the review of Buttram and Reiter (1981) did report consistent improvement in symptoms. Although recovery time from an abdominal procedure may be longer, there is no evidence to suggest that symptomatic relief would vary; route of surgery had no effect on long-term outcomes of hysterectomy in the Maryland Women's Health Study (Rhodes, Kjerulff, Langenberg, et al., 1999). Comparison of results between studies was prevented by lack of standardization in the reporting of measures.

Methodological Issues

Although all three studies were prospective and used validated instruments for measuring at least some outcomes, the ability to use these results to predict outcomes for women with fibroids is limited by variable levels of detail in reporting results specifically for women with fibroids. Although differences in results may not have been statistically significant across groups, it is possible that there were clinically important differences.

Bilateral oophorectomy was not a predictor of failure to have improvement in symptoms or development of new symptoms in the Maine study (Carlson, Miller, and Fowler, 1994a, 1994b), but it was in the Maryland study (Rhodes, Kjerulff, Langenberg, et al., 1999), although this effect was seen in Maryland only after 2 years. This may be due to the varying length of followup. Although Kjerulff and colleagues reported that there was no relationship between the self-reported use of hormone replacement therapy and poor outcome in oophorectomized women (Kjerulff, Langenberg, Rhodes et al., 2000), it may be that adherence to hormone replacement therapy decreased over time in some cases. This could have resulted in significant differences in outcomes at 2 years but not at 1 year. These studies all took care to differentiate between procedures involving no oophorectomy, a single oophorectomy, or a bilateral oophorectomy.

Another potential reason for the failure to detect an effect of oophorectomy on outcomes is that hysterectomy alone may lead to changes in ovarian function (such as decreased sex steroid production), which in turn could lead to decreased observed differences in outcomes attributable to changes in ovarian hormone production. The literature concerning the effects of hysterectomy on ovarian function is discussed in detail in a later section.

Summary

In prospective studies, hysterectomy results in improvements in symptoms and quality of life up to 2 years after the procedure in most women with sufficiently severe symptoms. For women with less severe symptoms who do not undergo hysterectomy, as many as 30 percent may have some improvement at 1 year; however, another 20-25 percent will undergo hysterectomy during that time. In the Maine study, women with fibroids who initially did not undergo surgery appeared less likely to have resolution of symptoms than women who had pain or bleeding unrelated to fibroids. Type of hysterectomy or short-term outcomes, such as complications, do not appear to influence the likelihood of a favorable outcome. New symptoms developed in 5-12 percent of women, with menopausal symptoms being most common. Low income and preexisting psychiatric disease appear to increase the risk of a poor outcome (defined as either development of new symptoms or failure to have improvement in preoperative symptoms); bilateral oophorectomy at the time of surgery also may increase the risk of a poor outcome. Race alone did not predict a poor outcome.

Methodological Issues

We were unable to identify any randomized trials comparing hysterectomy with myomectomy for women with symptomatic fibroids. Given strong physician and patient beliefs about the risks and benefits of hysterectomy, recruitment into such trials would likely be difficult. Patient age and desire to retain childbearing potential would also contribute to difficulty in recruitment. Setting aside the issue of the inherent biases of nonrandomized study designs in evaluating therapeutic outcomes, any attempt to synthesize the results of studies of both procedures to address this question is limited by multiple factors:

• Inconsistency in methods used to report symptom type and degree of severity; such data would at least allow comparison of results between prospective studies of hysterectomy outcomes with those for myomectomy.
• Inconsistency in reporting anatomical findings (uterine size, fibroid size, number and location of fibroids).
• Inconsistency in reporting patient characteristics that might affect outcomes, both demographic (race, income) and medical (comorbidities, prior surgical procedures).
• Inconsistency in methods for reporting outcomes, variable duration of followup, and variable accounting for patients lost to followup.
• In studies of hysterectomy, inconsistency in reporting results for patients with fibroids separately from patients with other indications for hysterectomy.
• Lack of power to detect statistically or clinically significant differences in important patient characteristics or outcomes.
• Use of inappropriate statistical techniques (for example, use of parametric tests for ordinal variables such as number of fibroids or parity).

Summary

The choice of hysterectomy versus myomectomy appears to reflect patient and provider preferences; for example, for women desiring to retain fertility, myomectomy is clearly the only option. The relative risks and benefits of the two procedures for other women remain unclear. For myomectomy, there appears to be a rough correlation between the number of fibroids removed and the risk of immediate complications and recurrence of symptoms (see next section). For women with symptoms and no desire to retain fertility, hysterectomy appears to result in significant improvements in symptomatic severity and quality of life; myomectomy may result in a lower risk of intraoperative complications. Data are not sufficient to allow comparison of myomectomy with hysterectomy in terms of long-term outcomes. Evidence on the possible effects of hysterectomy on longer term outcomes, such as ovarian function, is discussed later in this chapter under Question 8.

Methodological Issues

We were unable to identify any randomized trials comparing hysterectomy with myomectomy for women with asymptomatic fibroids. Given strong physician and patient beliefs about the risks and benefits of hysterectomy, recruitment into such trials would likely be difficult. Given the lack of a compelling argument for such surgery, justifying the trial would be difficult both scientifically and ethically. This question may be one for which modeling is appropriate; however, better data on the natural history of asymptomatic fibroids are required before such modeling could be performed.

Summary

There is no evidence to support performing either myomectomy or hysterectomy in asymptomatic women. Clearly, there are risks involved with both surgical procedures. Although fibroids may be associated with some adverse pregnancy outcomes in fertile women (see Question 8, discussion), there are no data that myomectomy in asymptomatic women with uterine fibroids reduces the risk of such complications.

Methodological Issues

Many included studies did not stratify results by the number of fibroids removed. Most of those that reported the number of fibroids removed used means as the summary statistic rather than the more appropriate median. The effect of the number of fibroids removed on outcomes was seldom examined in a multivariate analysis. In our analysis of the primary data, we were able to examine only in-hospital and immediate postoperative events.

Summary

In those studies that allowed comparison of results between women undergoing myomectomy for single versus multiple fibroids, there is a consistent pattern of better long-term results for women with single fibroids in terms of pregnancy rates, risk of recurrence, and need for subsequent surgery. Results from our primary data analysis also suggest that women undergoing removal of a single fibroid have fewer complications than women undergoing multiple myomectomies.

The effect of the number of fibroids on immediate short-term surgical outcomes is likely to be related to technical difficulty -- removal of a single fibroid is likely to take less time and result in less blood loss than removal of multiple fibroids. Whether there is a threshold number of fibroids, considering other factors such as size and location, where the perioperative risk of myomectomy becomes unacceptable is not at all clear from the available data. The effect of the number of fibroids on longer term outcomes may be related to technical factors (e.g., failure to remove very small fibroids) or biological factors (e.g., women with multiple fibroids may be more susceptible to fibroid development, or removal of large fibroids may allow smaller, undetected fibroids to grow more rapidly).

Both perioperative outcomes and longer term outcomes appear to be better for women with a single clinically detectable fibroid who undergo myomectomy than for those with multiple fibroids. However, whether there is a threshold number of fibroids above which myomectomy has either an unacceptably high complication risk or an unacceptably low likelihood of long-term benefit cannot be determined from the available data.

No Treatment

For women with asymptomatic fibroids, there is no evidence that failure to treat with medical or surgical therapy will result in harm, no evidence that medical or surgical treatment will result in benefits, and for surgical management at least, clear evidence of risks from the treatment itself.

For women with symptomatic fibroids, the likelihood of success with no treatment is likely to be a function of the nature of the patient's symptoms and her proximity to menopause, although data do not allow us to quantify this probability. In the Maine Women's Health Study, 68 percent of women with fibroids were managed with observation alone. At the end of 1 year, there were no significant changes in the amount of bleeding or pain or in the degree to which patients were bothered by these symptoms (Carlson, Miller, and Fowler, 1994b). Twenty-five percent of all women in this cohort eventually underwent hysterectomy during the followup period, although the data do not describe the risk for women with fibroids alone. For women with bleeding as the primary symptom, the probability of worsening of symptoms may well be inversely related to the likelihood of undergoing menopause; the effect of hormone replacement therapy on this is uncertain. For women with noncyclic pain, especially symptoms associated with an enlarged uterus, the degree to which menopause will resolve these symptoms is also uncertain.

Oral Contraceptives, NSAIDs, and Progestins (see under Question 6)

Given the scarcity of data on the effectiveness of these medical treatments in relieving symptoms related to fibroids, it is difficult to identify appropriate candidates for the treatments. In symptomatic women for whom there is no contraindication to their use, there is no evidence to suggest that a trial will result in harm and limited evidence that it may result in an improvement in symptoms.

GnRH Agonists (see under Question 6)

Although there is consistent evidence based on randomized trials that preoperative use of GnRH agonists reduces some morbidity associated with both hysterectomy and myomectomy, especially blood loss (Lethaby, Vollenhoven, and Sowter, 1999), the clinical relevance of the reduction is unclear. Statistically significant reductions in length of surgery (a difference of 6.6 minutes), blood loss (7.8 cc), reduction in vertical incisions, and length of stay after hysterectomy were observed; there was not a significant difference in transfusion rates. Women who are candidates for either hysterectomy or myomectomy may benefit from preoperative use of these agents (e.g., if anemia secondary to heavy menstrual blood loss would increase the likelihood of perioperative morbidity). However, additional study of the cost-effectiveness of routine use of these agents prior to surgery and of the impact of side effects on short-term quality of life, is needed. For perimenopausal women with symptoms, the use of these agents may avoid the need for additional therapy; however, there are no data to allow estimation of which perimenopausal women are most likely to benefit from their use.

Uterine Artery Embolization (see under Question 4)

Uterine artery embolization appears to hold promise as an alternative to other invasive procedures. There are no randomized trial data and only a few series with more than 20 cases (Goodwin, McLucas, Lee, et al., 1999; Hutchins, Worthington-Kirsch, and Berkowitz, 1999; Pelage, Le Dref, Soyer, et al., 2000; Siskin, Stainken, Dowling, et al., 2000; Spies, Scialli, Jha, et al., 1999; Spies, Warren, Mathias, et al., 1999; Worthington-Kirsch, Popky, and Hutchins, 1998). Resolution of bleeding and bulk symptoms range from 80-90 percent in these series. Although postprocedure pain requiring overnight hospitalization and narcotics is common, reported serious complications are rare. In all these studies, followup has been relatively short, with loss to followup in the largest series of close to 50 percent for 1-year outcomes (Hutchins, Worthington-Kirsch, and Berkowitz, 1999). Data on outcomes related to pregnancy are limited; almost every series reports at least one pregnancy, and there has been one series of 12 cases reported (see under Question 8). Notably, there appears to be a serious effort among those performing this procedure to collect prospective data using validated instruments (Spies, Warren, Mathias, et al., 1999) within a registry organized by the Society for Cardiovascular and Interventional Radiologists (J Spies, personal communication). Given the relatively short reported experience, the most appropriate candidates for this procedure would be women willing to participate in research protocols or registries performed at centers with significant experience and technical expertise. This is especially true for women considering future pregnancy.

Myomectomy (see under Questions 1 and 8)

The risks and benefits of myomectomy appear to be partly related to the location and number of fibroids, as well as the presenting symptoms. Women with a single clinically detectable fibroid appear to have fewer perioperative complications and a lower chance of recurrence following myomectomy than women with multiple fibroids. For women with submucous fibroids, the probability of success with hysteroscopic myomectomy appears to be related to the depth of invasion into the myometrium (Nezhat, Roemisch, Nezhat, et al., 1998). The likelihood of benefit from myomectomy appears to be related to presenting symptoms -- women with bleeding and submucous fibroids appear to have favorable short-term responses; pain symptoms may be less likely to resolve with any type of myomectomy (see under Questions 1 and 8). There are no data to support performing myomectomy in asymptomatic women.

Hysterectomy (see under Question 1)

Hysterectomy is clearly inappropriate for women who wish to retain the ability to carry a pregnancy. For women who do not desire to retain this ability, data from two large prospective cohort studies suggest that physical and psychological outcomes at 1 year after surgery in women with symptomatic fibroids are favorable (Carlson, Miller, and Fowler, 1994a; Kjerulff, Langenberg, Rhodes, et al., 2000). Women with preexisting depression or anxiety appear to be more likely to have persistent symptoms or develop new symptoms after hysterectomy for all benign indications (Kjerulff, Langenberg, Seidman, et al., 1996; Rhodes, Kjerulff, Langenberg, et al., 1999), suggesting that in such women, depending on severity and types of symptoms, hysterectomy may be less likely to result in an overall improvement in quality of life compared with women without depression and anxiety. Interestingly, in-hospital complications did not significantly affect the likelihood of having favorable outcomes in the longer term (Kjerulff, Langenberg, Rhodes, et al., 2000). Data supporting suggested rationales for performing hysterectomy in asymptomatic women are lacking (Friedman and Haas, 1993) (see also under Question 1).

No Treatment, Hormonal Therapy, or Treatment with NSAIDs

GnRH Agonist Therapy

Myomectomy

Embolization

Additional Medical Treatment after Initial Medical or Invasive Therapy

We were unable to identify any studies that addressed the specific question of increased morbidity associated with additional medical treatment for patients with persistent or recurrent symptoms after conservative surgical or nonsurgical treatment related to new side effects, drug interactions, or changes in anatomy or physiology related to invasive procedures. There is some evidence that uterine artery embolization may result in decreased ovarian hormone production (see under Question 9). Theoretically, this could result in decreases in bone density that would be further exacerbated by additional treatment with GnRH agonists. However, we did not identify any data to support this hypothesis. Also, there are no data to suggest that embolization, myomectomy, or other invasive therapies result in changes in pelvic organ anatomy or physiology that would increase the likelihood of side effects of other pharmacological agents such as steroid hormones or NSAIDs.

Additional Invasive Therapy after Initial Medical Treatment

We did not identify any studies that documented an increased risk of perioperative morbidity related to use of estrogenic compounds, NSAIDs, increasing uterine size, or development of other comorbid conditions.

Additional Invasive Therapy after Initial Surgery or Radiological Procedure

None of the articles that addressed the need for additional surgery or other invasive therapy after an initial procedure (see under Question 4) reported on morbidity related to these additional procedures.

No Treatment

The results of conservative management (no treatment) were described in the placebo arms of four randomized controlled trials of hormone treatment that followed patients for long periods (more than 6 months) (Friedman, Harrison-Atlas, Barbieri, et al., 1989; Friedman, Hoffman, Comite, et al., 1991; Gregoriou, Vitoratos, Papadias, et al., 1997; Schlaff, Zerhouni, Huth, et al., 1989) and in one cohort study in which the majority of nonsurgically managed patients received observation alone (Carlson, Miller, and Fowler, 1994b).

One of the most useful studies comparing nonsurgical with surgical management of fibroids was a large cohort study (Carlson, Miller, and Fowler, 1994b). A cohort of women with fibroids (uterine size 8 weeks or greater) was assembled from multiple sites in Maine and interviewed prospectively. One hundred and twenty-three women were managed with hysterectomy, and 106 received one of a variety of nonsurgical treatments (observation alone, 68 percent; NSAIDs or iron, 18 percent; hormone therapy, 14 percent). There were significant differences between the two groups at baseline. The patients who were managed without surgery had fewer days of bleeding and dysmenorrhea than those who underwent hysterectomy. In addition, the patients managed without surgery had more positive feelings about their symptoms and better quality of life at baseline than those who underwent surgery. By 1 year, the women who underwent hysterectomy had increased quality-of-life scores and, compared with nonsurgical patients, felt better about their symptoms. Approximately 23 percent of nonsurgical patients underwent hysterectomy in the 1 year followup period.

Baseline differences in the symptoms, attitudes, and quality of life between women managed nonsurgically and those managed surgically suggest that women who were managed nonsurgically were less severely affected by their fibroids. Because of this imbalance, we are limited in our ability to draw inferences about the relative effectiveness of nonsurgical and surgical management from this study. Furthermore, because nonsurgical management included several different treatment approaches and results were not reported separately for the various interventions, comparisons among the various nonsurgical treatments used are impossible.

Long-term observation (at least 1 year) or no treatment was described in the control arm of a trial of the synthetic steroid tibolone, a drug with mixed estrogenic, progestogenic, and androgenic properties (Gregoriou, Vitoratos, Papadias, et al., 1997). In addition, several placebo-controlled studies of the GnRH agonist leuprolide described 6-month outcomes of placebo-treated patients (Friedman, Harrison-Atlas, Barbieri, et al., 1989; Friedman, Hoffman, Comite, et al., 1991; Schlaff, Zerhouni, Huth, et al., 1989) (see Table 19). All of these studies were conducted among women who did not want to have surgery (Friedman, Harrison-Atlas, Barbieri, et al., 1989; Friedman, Hoffman, Comite, et al., 1991; Gregoriou, Vitoratos, Papadias, et al., 1997; Schlaff, Zerhouni, Huth, et al., 1989). Two described hemoglobin or hematocrit decreases over time, in addition to changes in fibroid or uterine size (Friedman, Harrison-Atlas, Barbieri, et al., 1989; Friedman, Hoffman, Comite, et al., 1991). None of the studies reported the number of women who eventually underwent surgery.

Consistent with the Maine Women's Health Study, in which the majority of nonsurgically treated patients received no treatment, these studies showed that "no treatment" resulted in no reduction in uterine size, no alleviation of symptoms, and no resolution of anemia.

Medical Therapy

Costs for a 3-month course of medical therapy were estimated by:

• Identifying minimum and maximum average wholesale prices for medications (including generics if available) using the 2000 "Red Book" of pharmaceutical prices.
• Estimating monthly pill usage based on potential dosing regimens (e.g., up to 7 days of NSAID therapy, or 10 days of oral medroxyprogesterone acetate).
• Estimating monthly costs based on wholesale prices for a given number of dispensed units (e.g., if 100 pills were dispensed at a given price and 21 would be used during a month, the monthly cost was estimated as 0.21 x cost/100 pills).

Costs are given for 3 months to facilitate comparison with 3 months of GnRH agonist therapy, the maximum time GnRH would be given prior to surgical management.

Results are shown in Table 23. NSAIDs are the least expensive treatment. The figures in Table 23 represent prolonged monthly courses of 7 days; costs for women who respond to shorter courses would be lower. Long-acting medroxyprogesterone acetate is next; unfortunately, there are no data on its effectiveness in women with fibroids. Oral progestins and oral contraceptives are next, at two to five times the cost of NSAIDs. GnRH agonists are the most expensive.

Estimation of cost-effectiveness would require better data on effectiveness, as well as on costs. One analysis of the use of adjunctive GnRH agonists prior to surgery suggested that their use might be cost-effective because of decreased complication rates and increased ability to perform less expensive vaginal hysterectomy (Bradham, Stovall, and Thompson, 1995). Replication of this analysis using more recently published effectiveness and cost data would be worthwhile.

Myomectomy

Estimated mean charges for women 18-44 years of age undergoing myomectomy in the NIS for 1997 were $8,884. Mean total costs for women undergoing the procedure at Duke University Medical Center from 1993 through 1998 (normalized to 1999 dollars) by age are shown in Table 24. The total mean Duke cost was $5,171, consistent with a 50-60 percent cost-to-charge ratio compared with the NIS data. Of these costs, 30.4 percent were attributable to inpatient unit costs, and 54.0 percent were attributable to surgical services.

Hysterectomy

Mean Duke costs by age group are shown in Table 25 (all hysterectomies combined). Overall mean cost was $5,961, with 32.6 percent attributable to inpatient unit costs and 49.0 percent attributable to surgical services.

Epidemiology

Using data from the Nurse's Health Survey, Marshall and colleagues determined that the incidence of uterine fibroids among black women is approximately three times that among whites (Marshall, Spiegelman, Barbieri, et al., 1997). In addition, they found that uterine fibroids are diagnosed earlier in black women than in white women, with the highest incidence of diagnosis being from 35-40 years versus 40-44 years for whites. A study by Kjerulff and colleagues corroborated this finding, reporting a mean age of diagnosis for blacks of 37.5 ± 7.9 versus 41.6 ± 6.6 for whites (Kjerulff, Guzinski, Langenberg, et al., 1993).

Management

Although the time from diagnosis of uterine fibroids to hysterectomy has been shown to be longer for blacks than for whites (3.9 ± 5.0 versus 2.8 ± 3.6 years) (Marshall, Spiegelman, Barbieri, et al., 1997), black women overall undergo hysterectomy at a younger age than white women due to their earlier average age at diagnosis. In one prospective cohort, the average age at hysterectomy for blacks was 41.7 ± 6.1 versus 44.6 ± 5.8 for whites (Kjerulff, Langenberg, Seidman, et al., 1996). In one large epidemiological study of more than 53,000 women, uterine fibroids accounted for 65.4 percent of the hysterectomies in black women versus only 28.5 percent in white women (Kjerulff, Guzinski, Langenberg, et al., 1993). Moreover, at the time of hysterectomy, black women have been shown on average to have both more myomas and a larger uterine size (Hillis, Marchbanks, and Peterson, 1996; Kjerulff, Langenberg, Seidman, et al., 1996). Kjerulff and colleagues found that 30 percent of black women who underwent hysterectomy had uteri greater than 500 grams versus only 15 percent of white women (Kjerulff, Langenberg, Seidman, et al., 1996). This difference has potentially important implications because Hillis and colleagues demonstrated that women undergoing hysterectomy with a uterus of > 500 grams were 1.6 (1.0-4.0) times more likely to develop operative or postoperative complications, particularly cuff cellulitis (relative risk [RR], 2.8 [95% confidence interval (CI), 1.3-6.2]) and transfusion (RR, 2.4 [95% CI, 1.3-4.3]).

At the time just prior to hysterectomy, Kjerulff and colleagues found that black women were more likely to be anemic than white women (56 percent vs. 38 percent) and more likely to complain of severe pelvic pain (59 percent vs. 41 percent ) (Kjerulff, Langenberg, Seidman, et al., 1996). There was no evaluation of postoperative symptomatology in this study, and none of the other studies evaluated symptomatology along racial lines.

We identified no studies that evaluated racial differences in response to, or provision of, medical therapies for uterine fibroids, such as NSAIDs, GnRH agonists, or other hormonal preparations. Similarly, there are no published data evaluating racial distinctions concerning myomectomy, uterine artery embolization, or surgical procedures other than hysterectomy.

Preliminary analysis of data from the 1997 Nationwide Inpatient Sample (NIS, 1997) shows the following: 45,740 women were identified as undergoing either myomectomy (n = 7,749) or hysterectomy (n = 37,991) for fibroids. Figures 2 through 5 show the estimated age- and race-specific incidences of myomectomy and hysterectomy based on 1997 NIS and U.S. Census Bureau estimates, as well as the projected race-specific cumulative incidences of each procedure based on these data. Bivariate analysis revealed that, at the time of their procedure, 45 percent of black women were younger than 40 versus 35 percent of Hispanic women and only 27 percent of white women (p < 0.0001). Data on number of prior children, contraceptive use, or history of sterilization are not available within the NIS data. Hispanics were more likely to have Medicaid/Medicare than their black and white counterparts: the figures were 23 percent, 17 percent, and 17 percent respectively (p < 0.0001). Blacks were more likely than either whites or Hispanics to earn $25,000 or less (41 percent vs. 29 percent vs. 26 percent) and less likely to earn more than $35,000 (24 percent vs. 31 percent vs. 36 percent) (p < 0.0001). Blacks were more likely than either whites or Hispanics to earn $25,000 or less (41 percent vs. 29 percent vs. 26 percent) and less likely to earn more than $35,000 (24 percent vs. 31 percent vs. 36 percent (p < 0.0001).

Thirteen percent of white women were treated with myomectomy versus 22 percent of black women and 17 percent of Hispanic women. Results for both the unadjusted and adjusted models were similar and revealed that, when undergoing a procedure for fibroids, black women are 1.6 (95% CI, 1.5-1.7) times more likely to receive a myomectomy than white women. Similarly, Hispanic women are 1.3 (95% CI, 1.2-1.5) times more likely to receive a myomectomy than white women. These findings were consistent for every age group of women. This suggests that, at the least, minority women are not less likely to be offered conservative surgical therapy than white women. There also may be cultural differences that lead some minority women to be more likely to request conservative therapy.

Preliminary analysis of the primary data abstraction of 239 patients undergoing myomectomy at Duke University Medical Center also shows some interesting findings. Black women were more than twice as likely as white women to have an in-hospital complication (odds ratio [OR], 2.48) or transfusion (OR, 2.2), but this increased risk was eliminated after adjusting for uterine size and number of fibroids (adjusted OR for complications 1.36 [95% CI, 0.56-3.15]; adjusted OR for transfusion 0.9 [95% CI, 0.27-2.76]). Other factors, such as body mass index (BMI) or insurance status, were not independent predictors of complications. This suggests that, for myomectomy, an increased complication rate in black women is largely attributable to differences in the underlying disease pathology. Currently, we are analyzing data on hysterectomy from the same time period to see if a similar pattern is observed.

Summary

Though it seems clear that black women are more likely to develop uterine fibroids than are white women and to have them diagnosed earlier, little else is known concerning racial differences in the natural history of fibroids.

Hysterectomy is the best-studied issue with regard to race and fibroids, yet many questions remain unanswered. Previous work has shown that at the time of hysterectomy, black women are more likely than white women to have severe pain, to have a uterus greater than 500 grams, and to be anemic. They also are more likely to have a longer delay from diagnosis to hysterectomy. The reasons for these observed differences are unclear. Studies examining the role of physician and patient beliefs in explaining this difference are needed, as are studies evaluating patient symptomatology pre- and postsurgery. Even more fundamental is determining whether the higher incidence of fibroids in black women explains their overall higher hysterectomy rates compared with white women (49.5 versus 41.2 per 10,000 women) (Kjerulff, Guzinski, Langenberg, et al., 1993). Our analysis of NIS data suggests that black women also are more likely to undergo myomectomy than white women and, in fact, are more likely to undergo myomectomy than hysterectomy for fibroids at every age group, despite having larger and more numerous fibroids. Further examination of these differences is needed. Finally, though no studies looking at hysterectomy rates for fibroids alone have been performed, black women have been found to be at 1.4 (95% CI, 1.3-1.5) times higher risk for surgical complications than white women. Given that black women appear to have larger uteri at the time of hysterectomy and uterine size is associated with an increased rate of complications, studies examining the relationship between uterine size, other comorbidities, race, and complications are needed. Preliminary data from our primary chart abstraction seem to support this relationship for myomectomies.

As stated above, we were unable to identify studies that examined racial differences and medical or invasive therapies other than hysterectomy for fibroids. Therefore, any studies evaluating epidemiological issues, response rates, side effects, and patient symptomatology in regard to race with these therapies would be a contribution.

Epidemiology

The rate of uterine fibroid diagnosis in the Nurses Health Study is highest between ages 40-44 (Marshall, Spiegelman, Barbieri, et al., 1997). The highest percentage of hysterectomies for fibroids is performed in the age group 40-49 (Kjerulff, Langenberg, Seidman, et al., 1996). From 1988-1990, the rate of admissions nationally for fibroids as the primary indication was 49.3 (95% CI, 43.6-55.0) per 10,000 women, with the peak age group for those admitted being 25-34 years (Velebil, Wingo, Xia, et al., 1995). Differences in these ages may represent differences in diagnostic methods, differences in patient populations, or differences in data sources used. One study of women complaining of abnormal uterine bleeding found submucous fibroids (by hysteroscopy) in 26 percent of premenopausal women, 24 percent of postmenopausal women on hormone replacement therapy, and 15 percent of postmenopausal women not taking hormones (Akkad, Habiba, Ismail, et al., 1995). In a case control study of more than 1,500 women, no relationship between oral contraceptive use and the development of fibroids was found (Parazzini, Negri, La Vecchia, et al., 1992).

Management

Summary

There is little evidence to allow clear conclusions about the effect of age or menopausal status on the risks and benefits of therapies for symptomatic fibroids. GnRH agonists, myomectomy (especially hysteroscopic myomectomy), and uterine artery embolization all appear more likely to induce prolonged amenorrhea in perimenopausal women compared with premenopausal women, but there are no data on long-term side effects with or without the use of hormone replacement therapy. In addition, there are no data that allow prediction of which perimenopausal women would be most likely to respond to conservative therapy.

Epidemiology

Estimates of the prevalence of fibroids during pregnancy range from 1 to 4 percent (Exacoustos and Rosati, 1993; Katz, Dotters, and Droegemeuller, 1989; Kuhlmann, Gartner, Schindler, et al., 1997; Piazze Garnica, Gallo, Marzano, et al., 1995).

Fibroids in Pregnancy

Most fibroids, especially those less than 5 centimeters, appear to remain stable in size during pregnancy. One small prospective cohort of 29 women found that 78 percent of fibroids either remained the same or decreased in size during pregnancy. Further, in the 22 percent of women whose fibroids increased in size, the increase in volume was less than 25 percent (Aharoni, Reiter, Golan, et al., 1988). Another prospective study of 134 women reported that 85 percent of fibroids either remained stable or decreased in size during the second and third trimesters, with 62 percent of fibroids 5 centimeters or less becoming undetectable (Strobelt, Ghidini, Cavallone, et al., 1994). Fibroids greater than 5 centimeters in this study were more likely to increase in size (26.2 percent vs. 9.7 percent, p = 0.03).

There is convincing evidence that the presence of uterine fibroids is associated with pregnancy complications. In one large retrospective cohort of more than 12,000 women, it was estimated that 1/500 pregnancies involved complications due to fibroids, and that 10 percent of women with fibroids experience pregnancy complications (Katz, Dotters, and Droegemeuller, 1989). Unfortunately, no comparisons in complication rates between women with and without fibroids were reported. A population-based series of more than 6,000 singleton live births in Washington state by Coronado and colleagues found that pregnant women with fibroids were 1.9 (95% CI, 1.6-2.2) times more likely to experience antepartum pregnancy complications than women without fibroids (Coronado, Marshall, and Schwartz, 2000). The pregnancy complications that have been found to be associated with the presence of fibroids include pain, first trimester bleeding, polyhydramnios, premature rupture of membranes, and abruption (Coronado, Marshall, and Schwartz, 2000; Exacoustos and Rosati, 1993; Rice, Kay, and Mahony, 1989). In particular, abruption has been associated with submucosal fibroids, fibroids greater than 200 cc, and a placental location that is superimposed over a fibroid (Exacoustos and Rosati, 1993; Rice, Kay, and Mahony, 1989). Pain has been associated with fundal or isthmic location of the fibroids.

Pregnant women with uterine fibroids also are more likely to undergo laparotomy than are pregnant women without fibroids. One large retrospective cohort of more than 120,000 women found that 1.1 pregnant women in 10,000 underwent laparotomy due to fibroids. Although these numbers are small, 28 percent of the women experienced pregnancy loss, with 21 percent undergoing hysterectomy at the time of the procedure (Burton, Grimes, and March, 1989).

Studies have found that fibroids are associated not only with pregnancy complications, but also with complications during labor and delivery. The Coronado study determined that women with fibroids were 1.9 (95% CI, 1.7-2.2) times more likely to have labor complications and also more likely to experience delivery complications than women without fibroids. In particular, they found that these women were more likely to experience dysfunctional labor (OR, 1.9 [95% CI, 1.3-2.7]), breech presentation (OR, 4.0 [95% CI, 3.0-5.2]), and cesarean delivery (OR, 6.4 [95% CI, 5.5-7.5]). In another series, 13 of 492 women with fibroids (2.6 percent) underwent hysterectomy at the time of delivery (no comparison group data were provided) and had higher rates of postpartum sepsis (4 percent vs. 0.4 percent in women without fibroids, p<0.001) (Exacoustos and Rosati, 1993). In an Italian series, the cesarean section (c-section) rate was 76 percent (Piazze Garnica, Gallo, Marzano, et al., 1995), and it was 73 percent in a Malaysian series (Hasan, Arumugam, and Sivanesaratnam, 1991). Consistent associations of fibroids with preterm labor have not been identified.

Only one study has looked at neonatal outcomes in relationship to the presence of fibroids during pregnancy. Coronado and colleagues found that the newborns of women with fibroids were 2.5 (95% CI, 1.5-4.2) times more likely to have a 5-minute Apgar less than 7; 1.9 (95% CI, 1.3-2.3) times more likely to have any malformation; and 2.0 (95% CI, 1.5-2.6) times more likely to have a birthweight below 2500 grams.

These observed associations between fibroids and adverse pregnancy outcomes do not necessarily mean that fibroids are causative of all of these complications. Confounding may play a role. For example, black women are more likely to have fibroids than women from other ethnic groups, and they also are more likely to have pregnancy complications such as preterm labor or low birthweight. Age is another potential confounder; the incidence of fibroids increases with age, as does the incidence of many pregnancy complications. Detection bias also may play a role -- women with pregnancy complications are more likely to undergo multiple ultrasound examinations, and they are more likely to undergo c-section. Therefore, they have more opportunities for detection of fibroids than women with uncomplicated pregnancies. In particular, this may explain some of the observed associations with low Apgar scores, low birthweight, and malformations, since both increased ultrasound surveillance and increased c-section rates would be expected in these circumstances. Physician decisionmaking about operative delivery based on beliefs about risks of fibroids may play a role in higher operative delivery rates in patients with fibroids.

Summary

Although the prevalence of fibroids in pregnancy is well established, their behavior during pregnancy and impact on pregnancy outcomes remain to be clarified. Moreover, although pregnant women with fibroids appear to be at a greater risk for complications, the nature and incidence of these complications needs to be more clearly defined through prospective cohorts. For example, the tumor characteristics (such as size, number, and location) associated with each particular complication need to be assessed in a manner similar to that done for abruption. More studies looking at neonatal outcomes in relation to the presence of fibroids in pregnancy are also needed.

The benefit of myomectomy in improving pregnancy complications remains unclear. No randomized studies, or even large cohort or case control studies, have examined the possible benefits and harm resulting from previous myomectomy. We also did not identify any studies comparing pregnancy outcomes pre- and postmyomectomy in the same woman. If fibroids are truly associated with an increased risk of second and third trimester pregnancy complications, then this would justify myomectomy in some asymptomatic women planning pregnancy. However, given the uncertainties surrounding the nature of any risk, its likelihood in a given patient, and the potential adverse effects of myomectomy itself on both the ability to conceive and the ability to undergo labor, additional data are needed before such a strategy can be widely recommended. As a corollary, no studies have provided convincing evidence as to the benefit of c-section after myomectomy. Thus, more studies examining the pregnancy outcomes of women postmyomectomy would be useful. However, since there is a documented risk of uterine rupture during labor in women with prior classical c-sections (an analogous anatomical change), elective cesarean in women with prior myomectomies appears to be a reasonable option.

There are extremely limited data available on pregnancy outcomes after uterine artery embolization. Well-designed prospective studies (possibly including myomectomy patients as a comparison group) would be extremely useful.

Epidemiology

Given the increasing incidence of fibroids with age and the fact that many infertility patients present for care at a time in life when fibroid incidence is particularly high, it is not surprising that fibroids are a relatively common finding in women with infertility. Prevalences as high as 13 percent in women with primary infertility have been reported (Valle, 1980). Direct evidence that fibroids may adversely affect fertility comes from studies of women undergoing assisted reproduction, where the presence of uterine fibroids was associated with significantly decreased implantation and pregnancy rates (Eldar-Geva, Meagher, Healy, et al., 1998; Stovall, Parrish, Van Voorhis, et al., 1998). Some of this difference may be related to distortion of the uterine cavity. One small case series (n = 26) noted an implantation rate of 2.7 percent for women with abnormal cavities versus 8.9 percent for those with normal appearing cavities (Farhi, Ashkenazi, Feldberg, et al., 1995). In a larger case series (n = 406), no significant difference in pregnancy rates was noted in women without fibroids (33.5 percent in 367 patients) and those with fibroids that were not intramural or submucosal (38.5 percent in 39 patients) (Ramzy, Sattar, Amin et al., 1998).

Infertility and Treatment of Fibroids

The literature addressing the efficacy of myomectomy for improving fertility comprises many small case series; there are no controlled trials or cohort studies involving comparison groups of infertile women with fibroids who do not undergo myomectomy. Comparing the results of these many case series is difficult because in most cases, "infertility" is not explicitly defined, other causes for infertility are not adjusted for, and in many the followup period is not stated. Of the 11 studies we identified that had at least a 1-year followup period, 26-75 percent of patients previously "infertile" became pregnant, with 75-94 percent of these pregnancies being successfully carried to term (Babaknia, Rock, and Jones, 1978; Berkeley, DeCherney, and Polan, 1983; Chong, Thong, Tan, et al., 1988; Cravello, D'Ercole, Boubli, et al., 1995; Dequesne and Schmidt, 1996; Dubuisson, Chapron, Chavet, et al., 1996; Gehlbach, Sousa, Carpenter, et al., 1993; Phillips, Nathanson, Meltzer, et al., 1995; Starks, 1988; Vercellini, Maddalena, De Giorgi, et al., 1999; Vercellini, Zaina, Yaylayan, et al., 1999). There have been no studies comparing the efficacy of abdominal versus laparoscopic versus hysteroscopic versus laser myomectomy. The reported pregnancy and delivery rates in previously "infertile" patients for all of these procedures, however, are similar (Babaknia, Rock, and Jones, 1978; Berkeley, DeCherney, and Polan, 1983; Chong, Thong, Tan, et al., 1988; Cravello, D'Ercole, Boubli, et al., 1995; Dequesne and Schmidt, 1996; Dubuisson, Chapron, Chavet, et al., 1996; Gehlbach, Sousa, Carpenter, et al., 1993; Phillips, Nathanson, Meltzer, et al., 1995; Starks, 1988; Vercellini, Maddalena, De Giorgi, et al., 1999; Vercellini, Zaina, Yaylayan, et al., 1999). This similarity may be due in part to the wide range of reported rates for each type of procedure, resulting in the failure to detect true differences. In addition, older studies rarely use life-table methods to estimate rates.

Although only case series have been performed, the addition of GnRH agonists prior to myomectomy does not appear to improve pregnancy rates over myomectomy alone (Kuhlmann, Gartner, Schindler, et al., 1997; Narayan, Rajat, and Goswamy, 1994; Sudik, Husch, Steller, et al., 1996). There have been no studies looking at GnRH agonists alone as pretreatment prior to attempting pregnancy in women with infertility.

Two small case series examined the effect of primary versus secondary infertility in relation to myomectomy. In both of these studies, patients with secondary infertility had higher pregnancy rates than those with primary, 37-41 percent versus 45-66 percent (Babaknia, Rock, and Jones, 1978; Darai, Dechaud, Benifla, et al., 1997).

Summary

Infertility is a multifactorial problem. Although fibroids may contribute to fertility problems, it remains unclear how large a role they play. Therefore, research examining the relationship between fibroids and infertility must be meticulous about both noting and accounting for other causes of infertility. Moreover, more investigations examining the effect of fibroid size and location on infertility are needed. A positive effect of myomectomy on fertility is consistently shown in uncontrolled studies, but methodological issues preclude definite conclusions on the overall effectiveness of myomectomy in infertility patients with fibroids. It is difficult to make meaningful comparisons across existing studies because inclusion criteria are unclear, as are the denominators of the pregnancy rates. Thus, future research should explicitly state the criteria being used to define "infertility, pregnancy, conception, and live birth," as well as length of followup.

No studies have been performed comparing the risks and benefits for treatment of infertility of the different types of myomectomies. Research is also needed on the potential role of medical treatment or uterine artery embolization in the treatment of infertility associated with fibroids.

After Hysterectomy

We identified 16 studies that addressed the question of subsequent ovarian function in premenopausal women who undergo hysterectomy without removal of the ovaries. This literature is conflicting but suggests that women who have a hysterectomy (with ovarian conservation) may undergo menopause earlier than women who do not. A proposed mechanism for this phenomenon is reduced blood flow to the ovaries following hysterectomy.

Siddle and colleagues demonstrated a significantly lower mean age of menopause in women after hysterectomy compared with controls who did not undergo hysterectomy (45.4 ± 4.0 vs. 49.4 ± 4.0, p < 0.001) in a retrospective study of 316 subjects. Forty-four percent of subjects with a previous hysterectomy developed ovarian failure by age 45 (defined by the presence of vasomotor symptoms, vaginal dryness, and elevated follicle-stimulating hormone [FSH] levels) compared with 13 percent in the control group. One population-based cross-sectional study found that twice as many women with prior hysterectomy visited their physician with menopausal symptoms as age-matched controls (Roos, 1984). Another cross-sectional study of women aged 39-60 from the Netherlands found that women with a previous hysterectomy were significantly more likely to have moderate to severe climacteric complaints, especially vaginal dryness and hot flashes, than age-matched controls who had not undergone hysterectomy (Oldenhave, Jaszmann, Everaerd, et al., 1993). The largest difference was noted in the youngest patient population (age 39-41). In contrast, two prospective studies failed to demonstrate any alteration in ovarian function in women who undergo hysterectomy (Bhattacharya, Mollison, Pinion, et al., 1996; Coppen, Bishop, Beard, et al., 1981). The only randomized trial to examine this issue found no significant difference in the proportion of subjects with hot flashes or elevated FSH levels 2 years after total vaginal hysterectomy compared with endometrial ablation for menorrhagia (Bhattacharya, Mollison, Pinion, et al., 1996). A study of a prospective cohort of women who had hormone level determinations before and 3 years after hysterectomy found no significant change in estrogen, FSH, or luteinizing hormone (LH) levels (Coppen, Bishop, Beard, et al., 1981). These two prospective studies are limited by their relatively short followup, however. Differences between the findings of the prospective and retrospective studies may reflect differences in time between hysterectomy and presentation with symptoms, differences in the indication for hysterectomy, and possible confounding by perimenopause. Women with abnormal bleeding because of perimenopausal changes may be more likely to undergo hysterectomy, and they also may be more likely to experience decreased ovarian steroid production within a short time after undergoing hysterectomy.

The prospect of earlier menopause in women with previous hysterectomy raises concern about the development of osteoporosis and cardiovascular disease in this population. Two cross-sectional studies have demonstrated lower bone density compared with controls in women with prior hysterectomy, despite ovarian conservation (Hreshchyshyn, Hopkins, Zylstra, et al., 1988; Watson, Studd, Garnett, et al., 1995). However, a third and larger cross-sectional study found no difference in bone mineral density and significantly lower FSH levels (p < 0.05) in women who had undergone premenopausal hysterectomy compared with women who had intact uteri (Ravn, Lind, and Nilas, 1995). The effect of premenopausal hysterectomy with ovarian conservation on the risk of subsequent cardiovascular disease is also conflicting. In the Framingham Study, women ages 45 to 54 with premenopausal hysterectomy had a relative odds ratio of new-onset coronary heart disease 2.7 times greater than women of the same age without previous hysterectomy (p < 0.01) (Gordon, Kannel, Hjortland, et al., 1978). In contrast, the Nurse's Health Study demonstrated no increased risk of cardiovascular events in women undergoing hysterectomy alone (RR 0.7 [95% CI, 0.2-1.2]), while women who underwent hysterectomy with bilateral salpingo-oophorectomy (BSO) who did not receive estrogen replacement therapy had an increased risk (RR 2.2 [95% CI, 1.2-4.2]) (Colditz, Willett, Stampfer, et al., 1987).

After Other Invasive Therapies

We identified no studies that evaluated ovarian function after myomectomy. However, studies of uterine artery embolization have shown transient or permanent amenorrhea in some women, suggesting the possibility of altered ovarian function with this technique (Goodwin, McLucas, Lee, et al., 1999; Spies, Scialli, Jha, et al., 1999).

Summary

Some, but not all, retrospective and cross-sectional studies suggest that women who undergo hysterectomy but retain their ovaries may experience menopause earlier than women who do not undergo hysterectomy. Prospective studies with longer followup are necessary to confirm this relationship. There may be some confounding due to perimenopause: perimenopausal women are more likely to have anovulatory bleeding, which may lead to an increased likelihood of hysterectomy. Since, by definition, these women would already have impending ovarian failure, there would be an apparent association between the hysterectomy and more rapid onset of ovarian failure. Preliminary data suggest that uterine artery embolization may alter ovarian function in some women. There currently are no data regarding the effect of myomectomy on ovarian function.

Epidemiology

There is no information on the prevalence of sexual dysfunction in women with uterine fibroids. It is unknown if women with uterine fibroids, symptomatic or asymptomatic, have altered sexual function compared with the general population. Furthermore, there is no information available regarding the sexual function of women with symptomatic uterine fibroids compared with women who have similar symptoms but do not have fibroids.

Theoretically, uterine fibroids could affect sexual function through several mechanisms. First, any mass effect caused by uterine fibroids could result in discomfort to a woman or her partner during intercourse. Second, symptoms that result from the presence of uterine fibroids, such as menorrhagia, dysmenorrhea, pelvic pain, or lower urinary tract dysfunction, may affect sexual function through physiological and/or psychological mechanisms. Third, physiological alterations in the uterus that result from the presence of uterine fibroids -- including altered blood flow, alterations in growth factors, and increased prostaglandin production -- may have some as yet undefined impact on sexual response.

Sexual Function after Hysterectomy for Uterine Fibroids

There is very little information regarding the effect on sexual function of hysterectomy performed specifically for symptomatic uterine fibroids. The vast majority of prospective studies that examined sexual function after hysterectomy did not stratify their results by preoperative symptoms or diagnoses. The three randomized trials that examined sexual outcomes after hysterectomy specifically excluded subjects with fibroids (Alexander, Naji, Pinion, et al., 1996; Crosignani, Vercellini, Apolone, et al., 1997; Dwyer, Hutton, and Stirrat, 1993). From 23 to 71 percent of the women in the prospective cohort studies that examined sexual function after hysterectomy had a preoperative diagnosis of uterine fibroids, yet only one study (Carlson, Miller, and Fowler, 1994a) provided results that related specifically to sexual function after hysterectomy for uterine fibroids (Bernhard, 1992; Candiani, Fedele, Parazzini, et al., 1991; Carlson, Miller, and Fowler, 1994a; Clarke, Black, Rowe, et al., 1995; Helstrom, Lundberg, Sorbom, et al., 1993; Lambden, Bellamy, Ogburn-Russell, et al., 1997; Weber, Walters, Schover, et al., 1999). The Maine Women's Health Study, a prospective cohort of 418 women who underwent hysterectomy, demonstrated a significant reduction in dyspareunia and an improvement in sexual interest and sexual enjoyment 1 year after hysterectomy (Carlson, Miller, and Fowler, 1994a). Thirty-five percent of women in this cohort had a diagnosis of uterine fibroids, and subgroup analysis revealed similar improvements in dyspareunia, sexual interest, and sexual enjoyment for this group (p < 0.005 for each). Given the wide variety of symptoms and diagnoses that can serve as indications for hysterectomy, it is unlikely that postoperative sexual function is completely independent of preoperative diagnosis. Extrapolating the results of studies that examine the effect of hysterectomy on sexual function independent of preoperative diagnosis may be hazardous. More studies that look specifically at the effect of hysterectomy for symptomatic uterine fibroids on sexual function will have to be done to confirm the findings of the Maine Women's Health Study.

Hysterectomy and Sexual Function

We identified 17 retrospective and 14 prospective studies that provided information on the effect of hysterectomy on postoperative sexual functioning. From this literature we were able to evaluate the short-term effect of hysterectomy on the following aspects of sexual function: dyspareunia, frequency of intercourse, orgasm, libido/sexual interest, vaginal dryness, and overall sexual function. The long-term effect of hysterectomy on sexual functioning could not be adequately evaluated based on the current literature because there are no prospective studies with followup greater than 2 years after hysterectomy, and none of the retrospective studies had the appropriate design to be informative.

Other Invasive Therapies and Sexual Function

There currently are no studies that examine the effect of abdominal, laparoscopic or hysteroscopic myomectomy on sexual function. Additionally, there are no studies that examine sexual outcomes after uterine artery embolization.

Summary

In general, hysterectomy does not appear to adversely affect sexual function in the majority of women in the first 1-2 years after surgery. Women with presurgical dysfunction may experience significant improvement. Only a small proportion of women will have a decline in sexual function after hysterectomy during this time period. The long-term effects of hysterectomy on sexual function are unknown. There are limited data about the effect on sexual function of hysterectomy performed specifically for uterine fibroids. The data that do exist, however, suggest short-term effects similar to hysterectomy in general. There is no substantial evidence that cervical preservation at the time of hysterectomy results in either improved or worsened sexual function compared with total hysterectomy. The effect of prophylactic oophorectomy on sexual function is unknown, as is the effect of postoperative estrogen replacement therapy. There are no studies that examine the effect on sexual function of other invasive therapies for uterine fibroids, such as myomectomy or uterine artery embolization.

Lower Urinary Tract Dysfunction

Dysfunction of the Colon, Rectum, and Anus

Pelvic Organ Prolapse

Epidemiology

Elective bilateral oophorectomy is performed in 40-66 percent of hysterectomies in women over the age of 40, primarily as prophylaxis for ovarian cancer (Dicker, Scally, Greenspan, et al., 1982; Lepine, Hillis, Marchbanks, et al., 1997; Pokras and Hufnagel, 1988). Uterine fibroids represent one of the most common indications for hysterectomy in the United States, and women between the ages of 40 and 49 account for the highest proportion of hysterectomies performed for fibroids (Kjerulff, Langenberg, Seidman, et al., 1996). As a result, many hysterectomies for uterine fibroids are accompanied by prophylactic oophorectomy. Factors that may affect the decision to perform a prophylactic oophorectomy at the time of hysterectomy include age, cancer risk, menopausal status, risk of osteoporosis or cardiovascular disease, and the patient's willingness to take estrogen replacement therapy. However, marked variations in physician practice styles exist (Gross, Nicholson, and Powe, 1999). One study demonstrated that women who received abdominal hysterectomy (adjusted OR, 11.42 [95% CI, 9.65-13.51]) or laparoscopically assisted hysterectomy (adjusted OR, 11.34 [95% CI, 8.13-15.81]) were significantly more likely to receive concomitant oophorectomy than women undergoing vaginal hysterectomy (Gross, Nicholson, and Powe, 1999). Significant geographic variations also were noted in this study. It is unlikely that women undergoing procedures other than hysterectomy for uterine fibroids would undergo prophylactic oophorectomy.

Benefits

Prophylactic oophorectomy is performed primarily in women who are peri- or postmenopausal to prevent subsequent ovarian cancer. Although prophylactic oophorectomy greatly reduces cancer risk, primary peritoneal cancer, which behaves much like ovarian cancer, developed in 6 of 324 women from families at high risk for ovarian cancer after ovarian removal (Piver, Jishi, Tsukada et al., 1993). The incidence of this cancer in other women is unknown.

Ovarian cancer is the fourth leading cause of cancer death in American women, with 14,000 deaths annually (Qazi and McGuire, 1995). As the majority of patients with ovarian cancer present in advanced stages of the disease, and there currently is no reliable screening tool, ovarian cancer prevention has been a point of emphasis. One study estimated that approximately 1,000 cases of ovarian cancer could be prevented annually if all women over the age of 40 who underwent hysterectomy (approximately 300,000) also received a prophylactic oophorectomy (Averette and Nguyen, 1994). Another study estimated that 700 prophylactic oophorectomies would have to be performed to prevent one case of ovarian cancer (Schweppe and Beller, 1979). However, the number of ovarian cancers prevented by the current rate of prophylactic oophorectomy is not known. A secondary benefit of elective oophorectomy at the time of hysterectomy is the potential avoidance of future surgery for benign ovarian disease. One retrospective study of 1,200 women who had undergone hysterectomy but retained at least one ovary demonstrated a 4 percent reoperation rate for benign ovarian disease (Plockinger and Kolbl, 1994).

Risks

The major risks associated with prophylactic oophorectomy include increased operative risk and the risks associated with premature estrogen deprivation. One cross-sectional study of 6,227 women found that after adjusting for type of hysterectomy, diagnosis, comorbidities, and age, elective oophorectomy was not associated with increased surgical morbidity or hospital stay (Gross, Nicholson, and Powe, 1999). Two retrospective studies of total vaginal hysterectomy also demonstrated no increased morbidity when prophylactic oophorectomy was performed (Ballard and Walters, 1996; Davies, O'Connor, and Magos, 1996). Women who undergo surgical menopause are more likely to suffer climacteric symptoms, osteoporosis (Ettinger, Genant, and Cann, 1987; Lindsay, Hart, and Clark, 1984), and increased cardiovascular disease (Colditz, Willett, Stampfer, et al., 1987). Several other risks have been suggested, including increased urinary incontinence (Rekers, Drogendijk, Valkenburg, et al., 1992) and alterations in memory function (Phillips and Sherwin, 1992), but these are less well established. Additionally, the Maryland Women's Health Study found that performing a bilateral oophorectomy at the time of hysterectomy was associated with significantly poorer outcomes 2 years after surgery (Kjerulff, Langenberg, Rhodes, et al., 2000). Hormone replacement therapy is effective in treating vasomotor symptoms (Meldrum, Erlik, Lu, et al., 1981) and preventing osteoporosis (Lindsay, Hart, and Clark, 1984; Nachtigall, Nachtigall, Nachtigall, et al., 1979; Paganini-Hill, Ross, Gerkins, et al., 1981). Many studies also suggest a cardiovascular protective effect of estrogen replacement (Anonymous. 1995; Bush, Barrett-Connor, Cowan, et al., 1987; Criqui, Suarez, Barrett-Connor, et al., 1988), but recent clinical trials have questioned this finding in women with preexisting cardiac disease (Hulley, Grady, Bush, et al., 1998). Despite the potential benefits of hormone replacement therapy, patient adherence to this intervention is poor (Ravnikar, 1987). Speroff and colleagues performed a decision analysis using Markov cohort modeling to evaluate prophylactic oophorectomy (Speroff, Dawson, Speroff, et al., 1991). They considered the influence of estrogen on coronary disease, breast cancer, and osteoporotic fracture. When adherence to estrogen replacement therapy was perfect, oophorectomy resulted in increased survival. However, when compliance with hormone replacement therapy was altered to a more realistic rate, retention of the ovaries rather than elective oophorectomy was favored. Preventive measures other than oophorectomy that have been shown to reduce the risk of ovarian cancer include oral contraceptive pills (La Vecchia and Franceschi, 1999), bilateral tubal ligation (Cornelison, Natarajan, Piver, et al., 1997), and hysterectomy without oophorectomy (Whittemore, Harris, and Itnyre, 1992). When assessing the risk and benefits of prophylactic oophorectomy, these alternative options also should be considered.

Summary

Prophylactic oophorectomy is effective in preventing ovarian cancer and appears to add minimal short-term surgical morbidity when performed with a hysterectomy. In women who are postmenopausal, the benefits of prophylactic oophorectomy may outweigh the risks; however, additional data on the potential benefits of ovarian preservation in postmenopausal women are needed. In premenopausal patients, the trade-off between benefit and harm is less clear, especially given the relatively poor adherence to estrogen replacement therapy.