Endometrial ablation; less is more? Historical cohort study comparing long-term outcomes from two time periods and two treatment modalities for 854 women.

Menorrhagia is a significant health problem in premenopausal women, with an estimated annual incidence for seeking medical help of 10/1000 women years [ 1 , 2 ]. Menorrhagia can reduce quality of life, and cause iron deficiency anemia [ 3 ]. Treatment of menorrhagia and reversal of anemia gives an increased quality of life [ 4 ]. When specific causes of abnormal uterine bleeding such as endometrial polyps, endometrial neoplasia or hematologic bleeding disorders have been excluded, treatment may be initiated. First-line treatment has traditionally consisted of medical therapy such as progestin, either as an oral medication or as a hormone-releasing intrauterine device (IUD). However, medical therapy is often ineffective. A systematic 2016 Cochrane review revealed that 59% of women randomized to receive medical treatment for abnormal uterine bleeding underwent surgery within two years, and 77% after five years [ 5 ]. A study by Famuyide from 2017 concluded that initial radiofrequency endometrial ablation compared to medical therapy offered superior reduction in menstrual blood loss and improvement in quality of life without significant differences in total costs of care [ 6 ]. Hysterectomy is a definitive treatment of the bleeding disorder but is costly and associated with a considerable risk of severe complications [ 7 , 8 ]. During the 1980s, less invasive surgical procedures were developed; removal or destruction of the endometrium while preserving the uterus. Following the first hysteroscopic procedures, several non-hysteroscopic endometrial ablation methods have evolved during the last 25 years, including thermal balloon ablation, microwave ablation, free-fluid thermal ablation, bipolar radiofrequency ablation and cryotherapy. These newer techniques are generally quicker, have lower complication rates and are technically less demanding, allowing a more widespread use of endometrial ablation techniques [ 7 , 9 , 10 ]. We have earlier evaluated a large cohort of 386 patients treated with hysteroscopic resection during the time-period 1992–1998 in our department [ 11 ]. A bipolar radiofrequency endometrial ablation method was introduced as an alternative to hysteroscopic resection in our department in 2004. When altering treatment regimes, and specifically when introducing a new surgical modality, it is important to evaluate if this change is beneficial. Hence, the objectives of this study were to evaluate the outcomes of two different minimally invasive techniques for treating abnormal menstrual bleeding in our department in the period 2006–2014, and to compare the results with a previous cohort of hysteroscopic resections only.

During the study period, 2006–2014 a total of 772 women were surgically treated with endometrial resection (75%) or endometrial ablation (25%) ( Fig 1 ). A total of 468 patients were included after consent. Of these 450 (96%) answered the questionnaire. The procedures included were 312 (67%) hysteroscopic resections of the endometrium (with or without resection of fibroids), 21 (4%) hysteroscopic resections of fibroids only, and 135 (29%) endometrial ablations. Inclusion and exclusion flowchart for women treated for abnormal uterine bleeding with minimally invasive surgery at Haukeland University hospital during 2006–2014. Preoperative clinical characteristics are shown in Table 1 . Comparison of two minimal invasive surgical methods performed at Haukeland University hospital during 2006–2014. a Confidence Interval Cases missing b n = 10 c n = 263 d n = 167 e n = 182 f n = 127 g n = 367 h n = 75 i n = 163 j Out of 140 cases noted as having any myoma n = number k n = 85 The 303 women declining to participate were not different from the consenting women regarding age (median 44 years, 95% CI 43–46, p = 0.753 Mann-Whitney). However, more women treated by endometrial ablation (70%) consented to participate, compared with those treated with hysteroscopic resection (58%, Fig 1 ). Women treated with hysteroscopic fibroid resection only (n = 21) were younger, with a median age of 40 years (95% CI 34–44) and with a lower median parity of 0.5 (95% CI 0–1, both p ≤0.003, Kruskal-Wallis test) than women treated with hysteroscopic endometrial resection or endometrial ablation. In our further statistical analysis, the resections of fibroids only are grouped together with the endometrial resections. Women treated with endometrial ablation had a slightly thinner endometrium measured by preoperative ultrasound and a lower rate of previous sterilization procedure than the hysteroscopic resection group. In spite of similar median age in the two groups, women treated with endometrial resection had a significantly higher proportion of postmenopausal status than in the endometrial ablation group. Previous medical treatment to correct menorrhagia was reported in 310 (66%) of cases. Medical treatment consisted of fibrinolytic inhibitors 158/310 (34%) and/or a hormone-releasing intrauterine device (IUD) 148 (32%). Of postmenopausal women, 17/45 (38%) used any hormonal replacement therapy. Perioperative findings comparing the two minimally invasive treatments are shown in Table 2 . In total, there were significantly more complications in the hysteroscopic resection group. Overall complication rate was significantly higher if fibroids were present: 16% versus 8% (p = 0.031 Chi-square test, data not shown). a One woman scheduled for endometrial ablation converted to hysteroscopic resection due to technical problems, thus 334 completed endometrial resections and 134 endometrial ablations are used for follow-up data. b CI = confidence interval cases missing c n = 7 d n = 316 e Defined as complication if surgeon initiated measures (such as intrauterine balloon tamponade or intravenous fibrinolytic medications) to control excessive bleeding f Discontinued procedure, 1 converted to hysteroscopic resection, 1 had a successfully completed ablation later same day. n = number In total 39 different surgeons performed the procedures during the 2006–2014 cohort. Four of the surgeons were classified as experienced and performed 273 (58%) of the procedures, of which 60% were hysteroscopies. Medium experienced surgeons performed 133 (28%) procedures and the inexperienced performed 62 (13%) procedures, of which 46 (74%) were endometrial ablations. The surgeons classified as inexperienced performed only 16 (5%) of the hysteroscopic resections. There were no statistical significant differences in complication rates among the different groups of surgeons, (p = 0.073). Histopathologic examination of the hysteroscopic resectate detected malignancy in 2 (0.04%) women, one despite benign preoperative histological sample and one where preoperative endometrial sample was missing. Both women had a hysterectomy performed during the first postoperative month. The short-term and long-term postoperative variables were not significantly different when comparing the two treatments ( Table 3 ). a One woman scheduled for endometrial ablation converted to hysteroscopic resection due to technical problems, thus 334 completed endometrial resections and 134 endometrial ablations are used for follow-up data. b CI = confidence interval Cases missing c n = 17 d n = 86 e n = 34 f n = 208, n = number Only 20 women (4%) reported persistent bleeding disorder in the questionnaire, 15 after hysteroscopic resection and 5 after endometrial ablation (p = 0.647 Chi-square test). We found no significant difference in pain change amongst the endometrial resection and the endometrial ablation groups, irrespective of pre-existing dysmenorrhea. The median follow-up time for the 2006-2014-cohort was 76 months ( Fig 2A ). During this period, 54 (16%) women in the hysteroscopic resection group had a subsequent hysterectomy performed, compared with 11 (8%) women in the endometrial ablation group (p = 0.024). The time until hysterectomy was median 15 months (95% CI 10–28) in the endometrial ablation group, and 19 months in the hysteroscopy group (95% CI 6–38, p = 0.726). The proportion of patients reporting satisfaction with the received treatment was identical for both treatments. Surgeon experience did not significantly influence the rate of hysterectomy. Rate of subsequent hysterectomy after hysteroscopic resection or endometrial ablation for 843 women treated during 1992–1998 and 2006–2014. A. Grouped based on surgeon experience level. B. Grouped based on time-cohort and surgical technique. Regarding the preoperative clinical characteristics, the two time-cohorts differ only in a significantly higher proportion of postmenopausal women and more myomas described preoperatively while fewer women were subjected to sterilization in the present cohort (2006–2014) ( Table 4 ). Women treated by hysteroscopic resection (n = 334) and endometrial ablation (n = 135) at Haukeland University hospital during 2006–2014, compared to a cohort treated by hysteroscopic resection only during 1992–1998. a Confidence Interval Cases missing b n = 18 c n = 338 d n = 282 e n = 118 f n = 205 g of n = 333 women with documented myoma n = number h n = 222 i n = 190 Comparing perioperative data from the two different time-cohorts, the procedures during 2006–2014 were more often performed in general anesthesia rather than regional anesthesia, had a shorter operation time and a lower complication rate ( Table 5 ). Women treated by hysteroscopic resection (n = 334) or endometrial ablation (n = 134) during 2006–2014, compared with hysteroscopic resection during 1992–1998. a CI = confidence interval cases missing b n = 15 c Defined as complication if surgeon initiated measures (such as intrauterine balloon tamponade or intravenous fibrinolytic medications) to control excessive bleeding, n = number The questionnaire used to evaluate outcomes were answered by 96% of included patients in our present cohort. The outcomes regarding early post-operative bleeding, pain and rate of subsequent hysterectomy were similar when comparing the two time-cohorts ( Table 6 and Fig 2B ). However, a higher proportion of women in the present cohort reported to be satisfied with the procedure and more women became amenorrhoic, compared with the previous cohort. a CI = confidence interval Cases missing: a n = 365 b n = 312 c n = 299 d n = e n = 190 f n = 198 g n = 164 n = numbers When assessing the risk of subsequent hysterectomy after endometrial resection or ablation we excluded cases with only myoma resection performed. Neither endometrial height, menopausal status, presence of myomas nor former sterilization were of individual (univariate) statistical significance. When adjusting for potential confounders (age at treatment and surgeon experience), neither the new treatment modality (endometrial ablation) nor patient being treated during the present cohort had any increased risk of subsequent hysterectomy ( Table 7 ). Uni- and multivariate logistic regression for 823 patients treated at Haukeland University Hospital 1992–1998 and 2006–2014. n: number of patients, OR: odds ratio Assessing patient self-reported satisfaction only former sterilization was of statistical significance when testing in univariate analysis. Neither age, surgeon experience nor complication rates were significantly different. The rate of patients reporting to be satisfied with their surgery was borderline higher in the 2006–2014 cohort with an OR of 1.77 (95% CI 1.03–3.03, p = 0.040) adjusted for type surgery and sterilization status ( S1 Table ). Type surgery was not independently a factor influencing patient satisfaction.

After introduction of endometrial ablation as a supplement to hysteroscopic resection, the complication rate and operation time in our department has declined, compared with a previous cohort of hysteroscopic resection. Although demanding less surgeon experience, the patient reported outcomes were slightly better and the subsequent hysterectomy rate non-inferior for patients operated during 2006–2014 when endometrial ablation was part of routine treatment for abnormal uterine bleeding. When medical treatment is not feasible, endometrial ablation may be a good alternative to hysteroscopic resection in properly selected patients.

In this historical cohort study patients were identified from the electronic patient files from Department of Obstetrics and Gynecology, Haukeland University Hospital, Bergen, Norway. We included all patients with a Nordic Classification of Surgical Procedures (NCSP) code LCB 25 (Hysteroscopic excision of lesion), LCB28 (Hysteroscopic endometrial excision) or LCA16 (Endometrial destruction) treated during the period 2006–2014 [ 12 ]. Regarding procedure LCB25, excisions performed on polyps or uterine septa were excluded. The surgical indication for the vast majority of patients was menorrhagia unresponsive to medical treatment. According to the department guidelines, the uterine cavity length should be less than 12 cm and submucosal fibroids should be less than 5 cm, evaluated by preoperative transvaginal ultrasound. Submucosal fibroid distorting the cavity was an exclusion criterion for endometrial ablation. Endometrial samples were collected from all patients prior to surgery to exclude neoplasia. In general, no specific medical pre-treatment regimen was used. All procedures were performed under general anesthesia or spinal block. The endometrial ablation procedures also included a paracervical nerve block. The endometrial resections were performed using a rigid 9 mm hysteroscope, bipolar loop diathermy and isotonic (0.9%) saline. The endometrial ablation procedures were performed using NovaSure® Endometrial Ablation System, (Hologic, Marlborough, Massachusetts, USA) a 6 mm diameter probe using bipolar radiofrequency diathermy for vaporization and coagulation. We reviewed hospital records for patient information. Preoperative data included age, parity, if sterilization had been performed, preoperative medical treatment, symptoms, and ultrasound findings. Perioperative data included type of procedure, operation time and complications. Any repeat procedure or subsequent hysterectomy was noted. Between two and nine years after surgery, patients were mailed a questionnaire ( S1 Fig and S2 Fig ) with questions regarding short-term (one month post-operative) effects of treatment. Patient self-report of overall satisfaction rate finalized the questionnaire. For peri- and postoperative analysis, the procedures were categorized into two groups: hysteroscopic endometrial resection with or without resection of fibroids and bipolar radiofrequency endometrial ablation. In the logistic regression assessing risk of subsequent hysterectomy, cases with only myoma resection performed were excluded, as this primarily is a fertility sparing treatment as opposed to full endometrial resection/ablation. Events registered as complications were: discontinued procedure, uterine perforation, fluid deficiency >1000ml and bleeding (defined as >500ml or bleeding with the need of additional hemostatic treatment such as internal uterine compression by catheter balloon or intravenous antifibrinolytic medication). We classified surgeons in three categories; as experienced (>50 procedures), medium experienced (10–50 procedures), or inexperienced (<10 procedures). For comparison, and evaluation of our present treatment algorithm, we used data from a previous study of women treated with endometrial resection during the period 1992–1998 [ 11 ]. Data collection (from patient files and patient questionnaire), inclusion criteria and definitions have been exactly the same for both cohorts. Statistical analyses were performed using the statistical package in SPSS version 25, IBM, Armonk, New York, USA. Categorical variables were compared by Chi-square test or Fisher exact test. Linear variables were compared by non-parametrical tests, Mann-Whitney U-test or Kruskal-Wallis test. Survival analysis regarding time until reoperation/hysterectomy by the Kaplan-Meier log rank method were applied to compare the two different operative procedures and to compare the two time cohorts. Logistic regression assessing risk of later hysterectomy and patient reported satisfaction was performed, evaluating preoperative clinical factors significantly different between the two surgical methods or between the two time cohorts. Factors identified as significant in univariate regression analysis were included in the multivariate model. Statistical significance is defined as p<0.05 and all tests are two-sided. The regional ethical board (REK 2015/892) has approved this study and all women have given written consent to participation. The study is being presented according to the STROBE guidelines [ 13 ].

Norwegian version (original) of questionnaire used. (DOCX) Click here for additional data file. English translation of questionnaire used. (DOCX) Click here for additional data file. Uni- and multivariate logistic regression for 508 patients treated at Haukeland University Hospital 1992–1998 and 2006–2014. (DOCX) Click here for additional data file.