A randomised trial of endometrial ablation versus hysterectomy for the treatment of dysfunctional uterine bleeding: outcome at four years

Transcription

1 British Journal of Obstetrics and Gynaecology April 1999, Vol106, pp A randomised trial of endometrial ablation versus hysterectomy for the treatment of dysfunctional uterine bleeding: outcome at four years Aberdeen Endometrial Ablation Trials Group Objective To assess the long term impact of initial management by endometrial ablation for women with dysfunctional uterine bleeding who would otherwise have had a hysterectomy. Design Long term follow up of randomised cohorts of women. Setting Gynaecology department of a large teaching hospital. Sample Two hundred and four women who had joined a randomised comparison of endornetrial ablation with hysterectomy, 4 to 5.6 years previously. Methods Postal questionnaire and casenote review. Main outcome measures Women's satisfaction with treatment, gynaecological symptoms and psychological outcomes at four years; further surgical treatment and differential resource use at a minimum of four years' follow up. Results Further surgical treatment was received by 39 (38%) women randomised to endometrial ablation, including six women who each had two additional treatments. At four years, the probability of receiving further surgical treatment by any method was 36% and by hysterectomy was 24% (compared with 29% and 14% respectively at one year). Satisfaction rates were high (80% ablation group us 89% hysterectomy group), the difference reflecting re-treatment. Premenstrual symptoms improved over time but more so in the hysterectomy group, who also rated their improvements in general health higher. The estimated overall mean cost of the endometrial ablation group is 93% of that of the hysterectomy group (f 1231 us &1332). Conclusions While about two out of every five women allocated to endometrial ablation eventually received further surgical treatment, hysterectomy with its associated morbidity was still avoided by 76% of women with dysfunctional uterine bleeding who would otherwise have had a hysterectomy. At four years, the difference in the costs of endometrial ablation and hysterectomy policies had narrowed. INTRODUCTION Hysteroscopic endometrial ablation is increasingly popular for the treatment of dysfunctional uterine bleeding'. Randomised comparisons with hysterectomy have shown short term benefits in terms of less operative morbidity, more rapid recovery, almost as high satisfaction levels, and lower Unlike hysterectomy, endometrial ablation does not guarantee amenorrhoea and women may require further surgery for menstrual symptoms some years after initial treatment6. For this reason the full impact of a policy of initial endometrial ablation, and hence its place in the management of heavy menstrual periods, will only be apparent after extended follow up7. While some long Correspondence: Professor A. Grant, Health Services Research Unit, Polwarth Building, Foresterhill, Aberdeen AE325 2zD, UK. term data are available from uncontrolled studies6,' the follow up period of women within randomised trials has tended to be relatively short'. We therefore reviewed cohorts of women originally recruited to one of these randomised trials between October 1990 and April 1992, allowing comparisons of endometrial ablation with hysterectomy after a minimum of four years follow up, in terms of further surgical treatment; gynaecological symptoms; satisfaction; psychosocial outcome; and costs. METHODS Full details of the original trial design have been reported elsewhere2 (as have the short term clinical, psychological and economic ~utcomes~,~.~). In summary, women were eligible if they met the following entry criteria: under 50 years of age; weighed under 100 kg; had 0 RCOG 1999 British Journal of Obstetrics and Gynaecology 360

2 ENDOMETRIAL ABLATION VERSUS HYSTERECTOMY 361 a clinical diagnosis of dysfunctional uterine bleeding (uterine size less than 10 weeks); and would have otherwise undergone hysterectomy. Two hundred and four women were recruited from the general gynaecology clinics of all consultants at this teaching hospital and randomly allocated to endometrial ablation or hysterectomy. The trial design was pragmatic, evaluating a policy of initial endometrial ablation, which recognises that some women would later receive further surgery. The original sample size was based on the wish to detect a difference of 20% in satisfaction rates (from 86% to 66%) between the policies, with 80% power at the 5% significance level. Although randomisation was primarily between hysterectomy and conservative surgery, those allocated conservative surgery were also randomly assigned to laser ablation or endometrial resection. The gonadotrophin releasing hormone agonist analogue goserelin was given to women having hysteroscopic surgery five weeks pre-operatively to prepare the endometrium. In the event, hy steroscopic surgery had to be abandoned before completion in eight cases: four were completed as a two-stage procedure and three had a hysterectomy. Seven women were sterilised, one by mini-laparotomy. Amongst the 95 women allocated hysterectomy who had an immediate hysterectomy, abdominal hysterectomy was performed in 85 (six of whom had bilateral oophorectomy and one a sub-total hysterectomy) and vaginal hysterectomy was performed in the other 10. A postal questionnaire was sent to all women four years after their initial trial management, assessing gynaecological symptoms, satisfaction, anxiety, depression, and sexual activity. The women had completed a similar questionnaire at recruitment and at six, 12 and 24 months post-operatively. Anxiety and depression were measured using the Hospital Anxiety and Depression Scale the higher the score the more depressed or anxious the women. At four years, overall satisfaction with treatment was measured using a seven point Likert scale which ranged from totally satisfied to totally dissatisfied; sexual activity questions were adapted from the Psychological Adjustment to Illness Scale. Initial non-responders were sent one reminder; if a woman still did not respond, her general practitioner was contacted and the address checked. If it was known that the woman was no longer resident at the address recorded in the trial documents, the Primary Care Record Department at Grampian Health Board was contacted for the woman s new address, if available. A review of all casenotes was conducted during May 1996 (at least four years after initial trial management) to identify re-treatments, other surgical procedures and investigations. The time from a woman s initial surgical management to any re-treatment was recorded to the nearest month. When possible, the estimated costs of re-treatments and additional procedures were those used in the previously reported economic analysis up to one year after trial entry5. Costs that had not previously been derived (such as for ultrasound scan, urethral dilatation, and cystometry) were estimated using a combination of published data * and personal correspondence with Aberdeen Royal Infirmary. Cost estimates were then applied to the data describing re-treatments derived from the casenote review. Data on patient costs up to one year were already available. Based on these, assumptions were then made about patient costs arising between one and four years as a result of re-treatments or additional procedures. All cost data are reported using 1994 costs as in the original study and any costs arising beyond that have been discounted accordingly using the UK Treasury recommended rate of 6%. All costs were subjected to a detailed sensitivity analysis to establish whether variations in these would affect conclusions. Statistical analysis All analyses were by intention to treat, that is, women remained in the group to which they had originally been randomised irrespective of any further treatment received. Data were analysed using the Statistical Package for the Social Sciences (SPSS) with calculation of 95% confidence intervals, where appropriate. The x test was applied to categorical data and the x test for trend was used when the data were ordinal. Non-parametric tests were applied to the Hospital Anxiety and Depression Scale anxiety and depression scores, due to their skewed distributions. The randomised groups were compared using the Mann-Whitney U test, and changes over time were analysed using the Wilcoxon matched pairs test. Life table analysis techniques were used to calculate the probability of re-treatment by hysterectomy, endometrial ablation, or either of these, for women who were randomised to endometrial ablation. The re-treatment curve for hysterectomy represents the probability of a women ever requiring hysterectomy (irrespective of previous re-treatment by endometrial ablation). The other curves represent the probability of the first retreatment. RESULTS Of the 204 women originally recruited to the trial, 105 had been randomly allocated to endometrial ablation and 99 to hysterectomy. These women received their initial trial management between October 1990 and April One hundred and fifty-one (74%) women returned postal follow up questionnaires at four years 0 RCOG 1999 Br J Obstet Gynaecol 106,36&366

3 ~ ~ ~~~ 362 ABERDEEN ENDOMETRIAL ABLATION TRIALS GROUP Table 1. Characteristics of trial groups at entry. Values are given as n (%) or mean [SD]. All randomised Four year questionnaire responders Endometrial Hysterectomy Endometrial Hysterectomy ablation (n = 105) (n = 99) ablation (n = 78) (n = 73) Age Marriedsteady partner Employed Children Sterilisedvasectomy Regular periods Regular dysmenorrhea Duration of menstrual problem > 2 years Cycle < 4 weeks > 7 days of bleeding Bloating Breast discomfort Irritability Headaches Depression Bowel symptoms Flushes 40.1 [5.0] 94 (90) 79 (75) 2.5 [1*0] 90 (86) 53 (50) 79 (75) 94 (90) 67 (64) 62 (59) 97 (92) 67 (64) 86 (82) 57 (54) 56 (53) 33 (31) 50 (48) 40.3 [52] 92 (93) 79 (80) 2.3 [0.9] 80 (81) 47 (47) 68 (69) 83 (84) 70 (71) 57 (58) 92 (93) 73 (74) 75 (76) 53 (54) 56 (57) 22 (22) 46 (46) 40.5 ( (90) 60 (77) 2.4 [0.9] 67 (86) 39 (50) 56 (72) 70 (90) 50 (64) 45 (58) 73 (94) 46 (59) 63 (81) 40 (51) 39 (50) 26 (33) 36 (46) 40.4 [5.1] 68 (93) 59 (81) 2.4 [1.0] 59 (81) 37 (51) 50 (68) 61 (84) 49 (67) 43 (59) 67 (92) 53 (73) 56 (77) 38 (52) 39 (53) 16 (22) 37 (51) after initial trial treatment. Table 1 summarises the characteristics at recruitment, firstly for all randomised women, and secondly, only for women who responded at four years. There were no differences between the trial responder groups with respect to known prognostic variables. A formal non-response analysis did not reveal any differences in prognostic characteristics between responders and non-responders to the four year postal questionnaire (data not shown). Table 2 details the surgical management that women actually received. Of the 204 women randomised, two women withdrew from the study choosing not to have surgery. Six women did not receive their allocated treatment, four in the endometrial ablation group (who had hysterectomy) and two in the hysterectomy group (who had endometrial ablation). Case note review was completed for 197 women, the remaining notes not being available at the time of data abstraction. The length of follow up in the casenote review ranged from 4 to 5.6 years (mean 4-8, SD 0.5). Amongst those randomised to hysterectomy, one of the women who chose to receive endometrial ablation subsequently had a hysterectomy. Amongst the women randomised to endometrial ablation, 39 (38%) received at least one additional surgical treatment for their menorrhagia. Six women in this group had two additional treatments (four had hysterectomy after re-treatment by endometrial ablation and two had repeated endometrial ablations). The reasons for Table 2. Actual surgical management over at least four years after trial entry. Values are given as n (%). All randomised Four year questionnaire responders Endometrial Endometrial ablation Hysterectomy a b 1 at i o n Hysterectomy Initial management Endometrial ablation Hysterectomy None Further surgery for menstrual periods Hysterectomy Edometrial ablation + hysterectomy Endometrial ablation Endometrial ablation (twice) Total n = 105 n= (96) 2 (2) 4 (4) 95 (96) 2 (2) n = 102 n = (20) 1 (1) 4 (4) 13 (13) 2 (2) 39 (38) 1 (1) n=78 n=73 75 (96) 1 (1) 3 (4) 72 (99) n=76 n=71 14 (18) 2 (3) 9 (12) 1 (1) 26 (34) 0 RCOG 1999 BrJ Obstet Gynaecol 106,

4 ENDOMETRIAL ABLATION VERSUS HYSTERECTOMY 363 women requiring re-treatment were bleeding (31 women), pain (3), and a combination of bleeding and pain (5). Re-treatment rates were slightly lower in the cohort of women who responded to the postal questionnaire at four years (overall 34% versus 38%) but this difference was not statistically significant. The cumulative probability of further treatment in those allocated to endometrial ablation is shown in Fig. 1. With respect to further surgical treatment (by either ablation or hysterectomy) this was 29% at one year, 36% at four years, and 40% at five years. The equivalent probabilities of receiving a hysterectomy were 14%,24% and 30%. Case note review also identified other additional surgical procedures or investigations since women's initial treatments. Eleven women randomised to endometrial ablation had further surgical procedures and/or investigations. The procedures and investigations performed were: hysteroscopy (6), laparoscopy (l), laparoscopy and hysteroscopy (l), laparoscopy and ultrasound scan (l), colposuspension (l), and ultrasound scan (1). In the hysterectomy group, two women had laparotomies, one had urethral dilation and another had cystometry. Of the 33 (45%) women allocated to endometrial ablation who reported amenorrhoea at four years, 13 had received hysterectomy within four years of their initial treatment (Table 3). There was no difference between the groups in relation to current pelvic pain (1 8% versus 15%). Women allocated to endometrial ablation reported premenstrual symptoms more frequently than those allocated to hysterectomy; the difference in symptoms achieved statistical significance at the 5% level for breast tenderness, bloatedness and irritability. The majority of women in both groups were either generally or totally satisfied with their treatment (80% versus 89%) (Table 4), the difference reflecting lower rates among women re-treated (69%). Similar levels of sexual activity and sexual satisfaction (Table 4) were reported in the two groups. Most women reported that their general health had improved but those in the hysterectomy group rated their improvement higher (Table 4). At four years the median (IQR) Hospital Anxiety and Depression Scale anxiety and depression scores were similar in the two groups: 5 (3, 8) ws 5 (2.8, 7.2), and 1 (0, 3) vs 1 (0, 3). Overall, there was a significant improvement in anxiety and depression from pre-operative levels but there was no apparent additional improvement after one year. At one year, median (IQR) Hospital Anxiety and Depression Scale anxiety and depression scores were 5 (2.5,8) and 1 (0.25,3) respectively; the corresponding scores at four years were 5 (3, 8) and 1 (0,3). w C E" 5 30 L 7?! 'c 0 Y.- u) L a,.- > Time since initial surgery (months) Fig. 1. Cumulative probability of re-treatment among women allocated endometrial ablation. A = Either hysterectomy or endometrial ablation; 0 = hysterectomy; 0 = endometrial ablation. 0 RCOG 1999 Br J Obstet Gynaecol 106,

5 ~ 364 ABERDEEN ENDOMETRIAL ABLATION TRIALS GROUP Table 3. Gynaecological symptoms at four years. Values are given as n (%). Endometrial ablation Hysterectomy (n = 78) (n = 73) % difference (95% CI) Menstrual status Amenorrhoea Brown discharge Hypomenorrhoea No change Pain Pelvic pain (current) Premenstrual symptoms Breast tenderness Bloatedness Headache Irritability Mood swings Weight gain n = (45) 9 (12) 29 (40) 2 (3) n=77 14 (18) n = (62) 35 (45) 29 (38) 35 (45) 23 (30) 24 (31) n=66 65 (98) 1 (2) n = (15) n=71 28 (39) 21 (30) 21 (30) 18 (25) 18 (25) 23 (32) 3 (-9 to +16) 23 (+7 to +39) 16 (+I to +31) 8 (-7 to +23) 20 (+5 to +35) 5 (-10 to +19) -1 (-16 to +14) The estimated overall mean cost of the endometrial ablation group is 93% of that of the hysterectomy group at four years (f1231 vs f1332). As would be expected, the initial treatment costs were much lower in the endometrial ablation group (mean: f726 vs f1247) but this difference was subsequently largely offset by the higher costs of re-treatments in the ablation group (mean: $321 vs f13). Sensitivity analyses using the bounds of the confidence intervals on initial treatment costs for endometrial ablation suggested that the fouryear costs of a policy of initial endometrial ablation are between 89% and 95% those of hysterectomy. Changing the discount rate within sensible limits (0%-12%) had little effect on the results. DISCUSSION This study has three particular strengths. First, it includes the longest follow up published to date of randomised cohorts of women treated by hysterectomy or endometrial ablation; second, randomised comparisons should be free of the distorting effects of selection bias; and third, the tendency for casemix to shift towards less severely affected women with increasing use of endometrial ablation should have been avoided because all women recruited to the study had symptoms which were sufficiently troublesome to warrant hysterectomy. Case note review was completed on almost all women and this allows confident unbiased estimates of re-treatment, and additional surgical procedures and investigations. The response rate to the postal questionnaire was 74%. Although this was comparable to that observed in a similar study at two years7, there is concern that differential follow up might have introduced bias into the comparisons. Non-response analysis, however, failed to reveal any differences at trial entry between those who Table 4. Satisfaction and general health with treatment at four years. Values are given as n (%). Endometrial ablation Hysterectomy Significance* Satisfaction Totally satisfied Generally satisfied Fairly satisfied Neutral or dissatisfied Sexual satisfaction Stopped Slight decrease No change Slight increase Marked improvement General health Much better Better Unchangedlworse n=76 33 (43) 28 (37) 8 (11) 7 (9) n=69 1 (1) 12 (17) 37 (54) 11 (16) 8 (12) n=76 29 (38) 35 (46) 12 (16) n=72 41 (57) 23 (32) 3 (4) 5 (7) n = 69 5 (7) 9 (13) 33 (48) 16 (23) 6 (9) n=70 47 (67) 19 (27) 4 (6) X = 2.6; df = 1 P =0.10 xz = 0.1; df = 1 P = 0.72 *x2 test for trend. Q RCOG 1999 Br J Obster Gynaecol 106,

6 ENDOMETRIAL ABLATION VERSUS HYSTERECTOMY 365 responded in the two groups. Overall, responders may have had fewer problems than those who did not respond, as judged by less re-treatment than the overall cohort of women recruited and this observation should be borne in mind when considering the generalisability of these results. A re-treatment rate of 38% after endometrial ablation is higher than that reported from some large but uncontrolled studies'.', but more comparable to the medium term outcome of one randomised trial7. Discrepancies between the results of a few specialised centres and large scale surveys have been noted", and these have been ascribed to differences in patient selection, operator expertise and length of follow up. We recently reported a trial comparing two endometrial ablation policies (endometrial laser ablation and endometrial resection). This included women allocated hysteroscopic surgery as part of the trial reported here. We found that there were differences in case mix and expectations between those women who had been recruited when hysterectomy was an alternative and those women recruited in the later period when endometrial ablation was the only available treatmenti4. We shall be reporting long term follow up in this other trial. Initial analysis within the study reported here has not, however, detected differences in re-treatment rates or in other outcomes. It is possible that the relatively high re-treatment rate observed in this study could be partly due to a learning effect. To assess this further we divided the women in the endometrial ablation group into three equal sized cohorts according to whether they were in the first, middle, or last third operated on. This showed that re-treatments (41%, 37% and 37%) and rates of amenorrhoea (20%, 20% and 29%) were approximately equally distributed across the three cohorts of women. This does not suggest a learning effect, at least at an institutional level. Nevertheless, 76% of women allocated to endometrial ablation still avoided a hysterectomy. Also, high levels of satisfaction were reported in both groups at four years (80% vs 89% generally or totally satisfied), the difference reflecting somewhat lower rates among those who had further treatment (69%). Only two (2%) women in the hysterectomy group had further surgery (both laparotomies), fewer than the 5% reported elsewhere The amenorrhoea rate of 45% (or 33% after accounting for subsequent hysterectomy) after endometrial ablation is consistent with other reports"'. This is higher than the 22% we reported at twelve months'. The presence of pelvic pain at four years in 18% of women after endometrial ablation is consistent with a rate of 20% reported from another long term follow up of women after endometrial resection6. The finding that, even after hysterectomy, 15% of women continue to suffer from pelvic pain carries an important message for those tempted to perform hysterectomies for pain alone. Relatively few of these women feel the need to report these symptoms, judging by the small number of investigations and surgical procedures performed in this group. Premenstrual symptoms were significantly less commonly reported at four years than at entry. However, the difference was greater in the hysterectomy group, significantly so for breast tenderness, bloatedness, and imtability. Psychosocial functioning seemed to be little affected by the original mode of treatment. At four years, there was no detectable difference between the two randomised groups in terms of anxiety, depression or psychosocial adjustment, an observation which reinforces the findings at one year9 that hysterectomy per se was unlikely to cause psycho-pathology in women. The alleged detrimental effects of hysterectomy on sexual life16*17 and the possible contribution of an intact cervix towards full orgasm'' have also not been substantiated in this study. The majority of women in both groups reported either no change or some improvement in their sexual interest and satisfaction with sex (Table 4). Economic analysis at four years suggests that savings associated with endometrial ablation when compared with hysterectomy are smaller than they appeared to be at one year, because of the high re-treatment rate in this group. This gradual convergence of costs supports the work of Sculpher and colleagues'. Under varying assumptions about the initial treatment costs, the endometrial ablation policy now appears to be 5% to 11 % less costly*. However, this does not take account of possible further savings following ablation performed as a day case procedure. CONCLUSIONS Hysterectomy is associated with better general health and fewer pre-menstrual symptoms, while as many as 40% of women allocated to endometrial ablation eventually receive further surgical treatment. On the other hand, with the introduction of an endometrial ablation policy, hysterectomy and its associated morbidity can be avoided by 76% of women with dysfunctional uterine bleeding who would otherwise have had a hysterectomy. At four years, the difference in the costs of endometrial ablation and hysterectomy policies has narrowed. Acknowledgements Aberdeen Endometrial Ablation Trials Group Writing 'Further details of the full sensitivity analysis can be obtained from the authors. 0 RCOG 1999 Br J Obstet Gynaecol 106,

7 366 ABERDEEN ENDOMETRIAL ABLATION TRIALS GROUP Committee: Professor A. M. Grant, Dr S. Bhattacharya, Ms J. Mollison, Ms E. McIntosh. Other members of the Group: Dr D. R. Abramovich, Professor D. A. Alexander, Dr K. Buckingham, Mr S. Cunningham, Professor H. C. Kitchener, Dr D. E. Parkin, Dr S. B. Pinion, Mrs A. Atherton-Naji, Professor I. T. Russell, Ms I. Cameron. Grant support was provided by the Chief Scientist Office of the Scottish Office Department of Health, which also funds the Health Services Research Unit. However, the views expressed are those of the authors. References 1 Magos AL. Management of menorrhagia. BMJ 1990; 300: Pinion SB, Parkin DE, Abramovich DR et al. Randomised trial of hysterectomy, endometrial laser ablation and transcervical resection of the endometrium for dysfunctional uterine bleeding. BMJ 1994; 309: Dwyer N, Hutton J, Stirrat GM. Randomised controlled trial comparing endometrial resection with abdominal hysterectomy for the surgical treatment of menorrhagia. Br J Obstet Gynaecol1993; 100: Sculpher MJ, Bryan S, Dwyer N et al. An economic evaluation of transcervical endometrial resection versus abdominal hysterectomy for the treatment of menorrhagia. Br J Obstet Gynaecol 1993; 100: Cameron IM, Mollison J, Pinion SB, Atherton-Naji A, Buckingham K, Torgerson D. A cost comparison of hysterectomy and hysteroscopic surgery for the treatment of menorrhagia. Eur J Obstet Gynecol Reprod Biol 1996; 70: O Connor H, Magos A. Endometrial resection for the treatment of menorrhagia. NEngl JMed 1996; 333: Sculpher MJ, Dwyer N, Byford S, Stirrat GM. Randomised trial com- paring hysterectomy and transcervical resection: effect on health related quality of life and costs two years after surgery. Br J Obstet Gynaecol1996; 103: Gany R. Shelley-Jones D, Mooney P, Phillips J. Six hundred endometrial laser ablations. Obstet Gynecol 1995; 85: Alexander DA, Naji AA, Pinion SB et al. A randomised trial of hysterectomy versus endometrial ablation for dysfunctional uterine bleeding: psychiatric and psychosocial aspects. BMJ 1996; Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scad 1983; 67: Derogatis LR. The Psychsocial Adjustment to Illness Scale (PAIS). J Psychosom Res 1986; 30: Scottish Health Service costs 1994/95. National Health Service in Scotland Information and Statistics Division. Common Services AgencyICrown Copyright, O Connor H, Broadbent JAM, Magos AL, McPherson K. Medical Research Council randomised trial of endometrial ablation versus hysterectomy in management of menorrhagia. Lancet 1996; Bhattacharya S, Cameron IM, Parkin DE et al. A pragmatic randomised comparison of transcervical resection of the endometrium with endometrial laser ablation for the treatment of menorrhagia. Br J Obstet Gynuecoll997; 104: Grant JM, Hussain IY. An audit of abdominal hysterectomy over a decade in a district general hospital. Br J Obstet Gynaecol 1984; 91: Utian WH. Effect of hysterectomy, oophorectomy and oestrogen therapy on libido. Int J Gynaecol Obsrer 1975; 13: Carlson W, Nichols DH, Schiff I. Indications for hysterectomy. N Engl JMed 1993; Zussman L, Zussman S. Sunley R et al. Sexual responses after hysterectomy-oophorectomy: recent studies and reconsideration of psychogenesis. Am J Obsrer Gynecoll981; 140: Received 29 June 1998 Returned for revision 25 September 1998 Accepted 4 December RCOG 1999 Br J Obstet Gynaecol 106,