The long‐term effectiveness of endometrial ablation for menorrhagia remains under investigation. Several large retrospective case series of laser and electrosurgical techniques for ablation have reported success rates of 79–92%, with varying lengths of follow‐up.1–6 The outcomes studied have included the incidence of subsequent hysterectomy, rate of patient‐perceived “adequate” control of menorrhagia, and surveys of patient satisfaction.
Most reports of endometrial ablation outcomes report the incidence of hysterectomy as a simple proportion at a given interval.6 This method of statistical evaluation is limited by its failure to incorporate the effects of varying lengths of follow‐up. For patients with menorrhagia who undergo endometrial ablation, applying survival analysis to describe the outcome of interest (eg, the incidence of subsequent hysterectomy) produces a more accurate description of the outcome over time. The present retrospective cohort study used survival analysis to evaluate the probability of a woman having hysterectomy or oligomenorrhea and determine the factors affecting these outcomes in patients who underwent rollerball endometrial ablation in a teaching hospital setting at a health maintenance organization.
MATERIALS AND METHODS
From a retrospective database review of the surgical procedures done from January 1991 through December 1996 at Kaiser Permanente Medical Center in Santa Clara, California, we identified 296 consecutive women who had undergone initial endometrial ablation with or without resection of polyps or myomas. Thirty‐one had a preoperative diagnosis of postmenopausal bleeding and were excluded from the study.
Preoperative endometrial suppression with hormonal therapy, prescribed in 220 of the patients (83%), consisted of either monthly 3.75‐mg injections of a GnRH agonist, Lupron Depot (TAP Pharmaceuticals, Deerfield, IL) or varying doses of oral medroxyprogesterone or injectable depot medroxyprogesterone acetate (Provera and Depo‐Provera, Pharmacia & Upjohn Co., Kalamazoo, MI). Thirty‐one of the women (11.3%) also underwent sharp or suction curettage immediately before the hysteroscopy was begun. Intraoperatively, a standard 24F operative hysteroscope (Karl Storz Endoscopy, Culver City, CA) equipped with a 2.5‐ or 5‐mm rollerball attachment connected to a Force 2 electrosurgical generator (Valleylab, Boulder, CO) was used at settings of 70–100 W of pure cutting or blend current. A solution of 3% sorbitol was used as the distension medium. Ablation of the endometrial lining alone was done in 178 of the procedures (64.7%); ablation with resection of polyps or myomas was done in 59 procedures (21.5%); and ablation in addition to diagnostic or operative laparoscopy was done in 22 procedures (8.0%). The presence of myomas was identified by the surgeon's visual inspection or by pathologic examination of the resected tissue. All ablation procedures were done by any of 20 attending surgeons with the assistance of resident physicians.
The use of prophylactic preoperative antibiotics was not routinely documented in the outpatient records and therefore was not included in this analysis. The follow‐up status was assessed by review of the last recorded gynecologic evaluation done at any Kaiser Permanente facility in Northern California before August 1998.
The statistical analysis of data used the Statistical Analysis System, version 6.11, software (SAS Institute, Cary, NC). The survival analysis used Kaplan‐Meier plots to evaluate the outcome of hysterectomy over time in the study population. Cox proportional hazards models were used to test the hypotheses that age, surgical history, preoperative hormonal endometrial preparation, addition of myomectomy or polypectomy, or presence of myoma at surgery would affect outcome of hysterectomy or subsequent oligomenorrhea. Multivariate analysis was used to identify the subset of variables that most accurately predicted the risk of needing hysterectomy. These analytic models used all the outcome information available for the study subjects despite the varying length of follow‐up time after completion of the surgical procedure. For the purposes of the medical record review, oligomenorrhea was defined to include amenorrhea and occasional or minimal vaginal bleeding. These outcomes were grouped together as representing successful treatment of preoperative menorrhagia.
From January 1991 through December 1996, 275 roller‐ball endometrial ablation procedures were done in 265 women with menorrhagia; 241 (90.9%) had been offered endometrial ablation for treatment of menorrhagia, 22 (8.3%) for menometrorrhagia, and two (0.8%) for abnormal bleeding in combination with serious medical disorders (systemic lupus erythematosus and aplastic anemia in one case and anemia associated with chronic renal failure in the other). For most patients in this study, medical therapy for control of abnormal uterine bleeding had either failed or was refused. For 25 patients, no follow‐up information was available for the period extending from time of the procedure to August 1998; the outcomes were analyzed for the remaining subset of 240 women. The mean length of follow‐up for all 240 women was 31.2 months (range 0.25–76.9; median 28.9).
The mean age of the patients at the time of surgery was 41.6 years (range 25–57), and the average obstetric history was 2.7 pregnancies and 2.2 births. In 196 of the study subjects (73.9%), endometrial biopsies were done preoperatively to evaluate for the presence of atypia; another 18 women underwent either D&C or diagnostic hysteroscopy before ablation. Thus, the medical record review revealed that a total of 214 of the women (80.8%) had undergone some type of preoperative endometrial evaluation.
Major complications were rare in our series: seven women had an excess intake of distension fluid (greater than 1500 mL) resulting in early termination of the procedure, and three patients had an excess intake of distension fluid documented at the end of the procedure. None of these women had any clinically significant sequelae. Uterine perforation was identified in four women, none of whom had excessive blood loss, required transfusion, or needed additional medical or surgical intervention. A cervical laceration was identified and repaired in one instance, and another patient had an episode of postextubation laryngospasm. Three women had incomplete ablation; two procedures were restricted by abnormal uterine anatomy, and one procedure was limited by synechiae that were caused by a prior endometrial ablation.
Postoperatively, five women were treated for endometritis; two of whom were admitted to the hospital for parenteral antibiotic therapy. One woman was diagnosed with a persistent tubo‐ovarian abscess after known perforation, and she eventually underwent hysterectomy. Three women were treated with D&C for cervical stenosis that resulted in hematometra. A woman undergoing hysterectomy for menorrhagia was found to have a bicornuate uterus, which had not been recognized at her ablation. The only known diagnosis of malignancy was a case in which leiomyosarcoma was identified by histologic examination of “polyps” removed at the time of ablation. No pregnancy was documented in any of our study subjects after they had undergone endometrial ablation.
The indications for subsequent hysterectomy in the 46 women who required additional surgical treatment included persistent menorrhagia (n = 34), dysmenorrhea or chronic pelvic pain (n = 8), persistent hematometra (n = 2), tubo‐ovarian abscess (n = 1), and leiomyosarcoma (n = 1). Of the eight women who underwent hysterectomy for dysmenorrhea as the chief complaint, seven had a history of previous tubal ligation; among all the other women who underwent hysterectomy, only 61.5% had a history of previous tubal ligation. For the women who underwent hysterectomy after endometrial ablation, a statistically significant association was found between a tubal ligation history and a preoperative diagnosis of dysmenorrhea (odds ratio 4.56). Only nine women had histologic reports available in their medical record. Four of the nine had adenomyosis described in the hysterectomy specimen; in each of these four, menorrhagia was given as the reason for their subsequent hysterectomy.
Survival analysis was performed to determine the relationship between the probability of not undergoing subsequent hysterectomy and the time since the ablation procedure. The results of this analysis are presented in Table 1, in which the number of women analyzed at each interval is also listed. Ten of the women (3.8%) in this series consented to a second endometrial ablation procedure. For six of these women, hysterectomy was eventually necessary to treat persistent menorrhagia; they were included in the survival analysis calculations using the interval between the initial procedure and the time of hysterectomy.
For the women still being monitored who did not require hysterectomy, successful control of menorrhagia was evident at long‐term follow‐up. Minimal or no vaginal bleeding was documented for 85 (91.4%) of 93 women at 3 years; for 54 (96.4%) of 56 women at 4 years; and for 28 (96.6%) of 29 women at 5 years.
Table 2 summarizes the individual effects of each hypothesized predictor variable on the probability of needing hysterectomy. A history of tubal ligation was significantly associated with an increased probability of undergoing hysterectomy after endometrial ablation to treat menorrhagia. All the methods of preoperative hormonal suppression for inhibiting growth of the endometrial lining were associated with statistically significant decreases in the probability of requiring subsequent hysterectomy. In addition, patients with oligomenorrhea at the most recent recorded visit were more likely to have a history of cesarean delivery. No statistically significant association was seen between the outcomes listed and the use of a GnRH agonist for suppression, the addition of polypectomy or myomectomy to the ablation procedure, or the presence of myoma at the time of surgery.
Multivariate analysis revealed that administration of hormonal therapy for preoperative endometrial suppression and a history of tubal ligation remained statistically significant predictors for hysterectomy (Table 3). However, a separate comparison among pretreatment with GnRH agonist, pretreatment with injectable depot medroxyprogesterone acetate or oral medroxyprogesterone acetate, and an absence of pretreatment showed no independent statistically significant effect on outcome. Hazard ratios calculated to compare the probability of hysterectomy among the three age groups revealed a statistically significant reduction in the risk of a subsequent hysterectomy when comparing women at least 45 years old with those younger than 35. Multivariate analysis revealed no combination of predictors strongly correlated with the outcome of oligomenorrhea. Table 3 reports the analysis of 231 patients for whom information was available on all the listed variables; this completeness of information was required for the patient's data to be included in the multivariate analysis.
Compared with its most commonly used alternatives, loop resection and laser ablation, rollerball coagulation to ablate the endometrial lining is easier to learn and use and associated with a lower incidence of uterine perforation.7 In addition, the lower cost of rollerball coagulation (ie, compared with that of laser ablation) makes the technique more accessible to general gynecologists. Currently, an estimated 20,000 endometrial ablation procedures are done each year in the United States.8
Ke7 summarized previously reported experiences using the rollerball electrode for endometrial ablation. Without reference to the length of follow‐up in 410 women, 31.5% achieved amenorrhea, 51.0% had a satisfactory reduction in menses, and treatment failed in 7.5%. These results were comparable with results for Nd:YAG laser ablation and for endometrial resection, summarized in the same review.7 More recently, a 60% incidence of amenorrhea and a 29% incidence of hypomenorrhea were reported after 1 year for 800 endometrial ablations, 80% of which used the rollerball technique.1
In another large series on rollerball endometrial ablation by Chullapram et al,3 75% of 142 women reported an overall satisfactory result at 36 months postoperatively. Twelve women underwent hysterectomy before 30 months of follow‐up, and no women underwent hysterectomy during the subsequent 18 months. This trend suggested a plateau effect at 30 months after the procedure. However, only 36 women were available for follow‐up at 36 months after the original procedure. The success rate in that series was not influenced by age, subjective assessment of endometrial “thickness,” type of hormonal preparation, presence of adenomyosis, or concurrent resection of myomas.3 In a smaller series, 14 of 42 patients studied required hysterectomy 5 years after endometrial ablation; these authors postulated that by extrapolation of their data, all the women would require hysterectomy within 13 years after the ablation procedure.9
Our results indicated that at 5 years of follow‐up, the probability is 71% that women who underwent rollerball endometrial ablation with or without polyp and myoma resection will have not undergone hysterectomy. Over time, the effectiveness of endometrial ablation decreased and then plateaued, with the probability of no hysterectomy declining from 93% to 71% during a 5‐year interval. The lower “success rate” in our series compared with previous reports is probably related to the statistical methods used and to technical factors. If we had reported outcomes on the basis of the incidence of hysterectomy, the analysis would have shown hysterectomy occurred in 19% and successful control of menorrhagia in 81% of the women without regard to the length of follow‐up. Thus, the survival analysis presented is a more realistic way to estimate success after rollerball ablation. Most previously described series of at least 100 patients1,2,6 reported only the incidence of hysterectomy after ablation or had shorter periods of follow‐up.
Our data also differ in that multiple surgeons and residents performed these rollerball endometrial ablations, in contrast to most other series that reported results obtained by only one or two primary operating surgeons. In addition, operative hysteroscopy with polypectomy or myomectomy was done in 21.5% of the ablation procedures in our patients. We believe that our results more accurately reflect the current practice of rollerball endometrial ablation in a community and teaching hospital setting, in which varied skill levels and techniques contribute to the overall outcomes.
Several groups have reported factors associated with successful outcomes after endometrial ablation, although not necessarily using the rollerball technique. Most of these reports4,6 found no relation between the outcome and the technical factors at the time of surgery. Lower rates of subsequent hysterectomy in women older than 45 were documented in only one study4; most studies showed no relation between age and outcome after endometrial ablation. Similar to our data, previous case series5,6 found a higher incidence of hysterectomy after repeat ablation procedures.
The effect of preoperative hormonal preparation on the outcome of ablation has been examined with varying results. One randomized controlled trial comparing administration of a GnRH agonist with no preoperative hormonal treatment found no difference in outcome, patient satisfaction, rate of amenorrhea, or amount of blood loss with subsequent menses; however, these authors noted that preoperative treatment subjectively improved the technical factors that contribute to a faster, easier procedure.10 Our data are inconsistent in that they showed a decreased hysterectomy rate among women who underwent pre‐ablation treatment with any agent but no effect when the hormonal preparations were analyzed independently. In our study, women who underwent preoperative treatment may have been given medication when the surgeon anticipated technical difficulty or a thick endometrial lining. This type of selection bias is common in observational case series and could account for our conflicting results.
The reason for the higher observed rate of hysterectomy after previous tubal ligation in this study population is not immediately clear. A post‐ablation‐tubal sterilization syndrome has been described, in which hematosalpinx develops from persistent endometrium at the cornua, causing pain and intermittent vaginal bleeding.11 The association between hysterectomy performed for dysmenorrhea and a history of tubal ligation observed in our case series may be directly related to the development of this syndrome. However, we do not believe this possibility should prevent women with menorrhagia and a previous tubal ligation from undergoing endometrial ablation.
The rarity of major complications in our series was consistent with a previous summary7 on the safety of the rollerball technique. An interesting note is that the addition of polypectomy or myomectomy to the rollerball ablation procedure did not appear to influence the outcome. Compared with the more recently introduced “blind” techniques (ie, thermal balloon probe), the ability to remove a visualized polyp or myoma would intuitively seem an advantage of using operative hysteroscopy to evaluate and treat menorrhagia. A study directly comparing the use of a blind technique and traditional hysteroscopy for ablation may elucidate any difference in effectiveness attributable to the ability to perform operative resection of polyps or myomas or any advantage in visualizing the endometrial cavity.
The results of this study disclose important outcome data to be used in counseling patients regarding their options and further supports the long‐term effectiveness of rollerball endometrial ablation. Offering a repeat ablation should be carefully considered, given the 60% rate of subsequent hysterectomy in our patients who underwent a repeat ablation procedure. Younger patients, according to our study results, may not benefit as much from conservative surgical treatment for menorrhagia with rollerball endometrial ablation. As new techniques for endometrial ablation are investigated and marketed, this reflection on our recent experience provides relevant baseline data for comparison.