4 Evidence and interpretation
The Appraisal Committee considered evidence from a number of sources (see Appendix B).
4.1.1 A total of 13 publications relating to seven trials were identified by literature searches. One of the trials compared MEA with TCRE/RB, and six compared TBEA with TCRE, with RB, or with both. Two of these trials had a non-randomised, controlled design, and the rest were randomised controlled trials (RCTs). In addition, one manufacturer provided the translation of a small trial, published in German, that compared TBEA with RB, and another unpublished RCT comparing TBEA with TCRE – this was submitted in confidence. Another manufacturer provided details of a study comparing MEA with RB that it conducted as part of its submission to the US Food and Drug Administration (FDA). In summary, 10 trials (two MEA and eight TBEA trials) were included in this review.
4.1.2 One trial reported that at 12 months, 87% of women who had undergone MEA and 83% of women who had undergone RB had normal bleeding levels, defined as a PBAC score of less than 75. The difference could have arisen by chance (p = 0.359). Another trial that compared MEA with TCRE/RB reported a median bleeding score of 3 in both groups at 12 months, which fell at 24 months to 1 for the MEA group and 0 for the TCRE group. This bleeding score is the sum of the daily scores reported by the women, who were asked to grade the heaviness of their period on a fivepoint scale for each day of their period. The differences could have arisen by chance.
4.1.3 Only one trial, which compared MEA with TCRE/RB, reported bleeding patterns in terms of the length and severity of bleeding. Based on intention-to-treat (ITT) populations, at 12 months 6% of the MEA group and 5% of the TCRE/RB group had more than 3 days of heavy bleeding (2% MEA, 5% TCRE/RB at 24 months), and 11% of women in the MEA group required at least double their usual sanitary protection compared with 12% in the TCRE/RB group (7% MEA, 13% TCRE/RB at 24 months). The differences between the groups could have arisen by chance.
4.1.4 Amenorrhoea was reported as a clinical outcome in two MEA trials. In one trial, amenorrhoea at 12 months was reported for a median of 40% of women undergoing MEA, compared with 40% undergoing RB. At 36 months follow-up, the median amenorrhoea rates were 47% and 41% respectively (p = 0.19). The other trial reported similar median values for amenorrhoea in ITT populations at 12 months (55% in MEA versus 46% in RB, p = 0.106).
4.1.5 Two MEA trials reported patient satisfaction. In one trial, 69% of both MEA and TCRE/RB groups were totally or generally satisfied at 12 months, and 74% of those undergoing MEA and 64% of those undergoing TCRE/RB were totally or generally satisfied at 24 months. The study was underpowered to detect whether this observed clinically important difference of 10% could have arisen by chance. Another trial, which compared MEA with RB, reported that 98% of women undergoing MEA were very satisfied or satisfied at 12 months compared with 99% of those undergoing RB.
4.1.6 One trial used the SF-36 questionnaire to examine the impact of MEA and TCRE on quality of life. Following treatment, six of the eight items improved significantly compared with baseline in the MEA group, and seven items improved significantly in the TCRE group.
4.1.7 Although the duration of procedures was defined inconsistently in the trials, MEA procedures took less time than TCRE and/or RB. In one trial, the mean operating time was 11.4 minutes for MEA and 15.0 minutes for TCRE/RB (p < 0.001). The other trial reported 'anaesthesia times' of 39.3 minutes for MEA and 47.1 minutes for RB, and 'treatment times' of 3.5 minutes for MEA and 20.3 minutes for RB. These differences were all statistically significant at the p < 0.01 level.
4.1.8 One trial reported that 8% of women in the MEA group had undergone further ablation or hysterectomy at 12 months (6% hysterectomy, 1% TCRE and 1% other ablation), and 8% of women in the TCRE plus RB group had undergone hysterectomy, but none of this group had undergone further ablation. The difference could have arisen by chance. Another trial reported that 1 out of 209 women in the MEA group and 1 out of 106 women in the RB group had undergone hysterectomy after 12 months, and none required further ablation.
4.1.9 Four TBEA trials reported changes in PBAC score. One trial, which compared TBEA with RB, reported that at 12 months, 73% of the TBEA and 70% of the RB group had normal bleeding levels, defined as a PBAC score of less than 100. Another trial reported that 71% of the TBEA and 79% of the RB group had normal bleeding levels at 12 months, defined by a more stringent criterion (that is, a PBAC score of less than 76). This second study reported mean PBAC scores of 41.1 in the TBEA group and 40.2 in the RB group (mean score reductions of 343.2 and 345.5, respectively). Another trial did not report actual PBAC scores, but stated that these were significantly better for the TBEA group than for the RB group at 24 months (p = 0.01), although not at 6 or 12 months. This trial measured treatment success as a post-operative PBAC score of less than 185, and 78% of women in the TBEA group and 76% of women in the TCRE group achieved this at 24 months. Results from the fourth trial were submitted to the Institute in confidence.
4.1.10 At 24 months, between 5% and 8% of patients who had undergone TBEA, and between 9% and 15% of those who had undergone TCRE or RB, were still experiencing HMB. At 60 months, these figures were 2% for the TBEA group and 1% for the TCRE or RB group. No trial reported statistically significant differences between the groups for recurrent HMB.
4.1.11 Amenorrhoea was reported as a clinical outcome in five TBEA trials. Amenorrhoea at 12 months was reported in between 10% and 40% of women for TBEA, and between 17% and 30% for TCRE/RB. The differences were statistically significant in only one trial (14% for TBEA versus 22% for RB, p < 0.05). At 36 months, 13% of women undergoing TBEA and 21% of women undergoing RB had amenorrhoea, and at 60 months 10% of women undergoing TBEA and 14% of those undergoing RB had amenorrhoea. These results are for ITT populations.
4.1.12 Six TBEA trials reported patient satisfaction. Of these, five reported non-significant differences in patient satisfaction between TBEA and TCRE and/or RB groups. The proportion of women who were satisfied or very satisfied with the treatment ranged between 79% and 100% in TBEA groups, and between 54% and 100% in TCRE and/or RB groups at 12 months. The trial with the longest follow-up reported that 42% of women in the TBEA group and 44% of women in the RB group were satisfied at 60 months (ITT populations). Only one trial reported statistically significant differences between the TBEA and the TCRE and/or RB groups. In this trial, 43% of women undergoing TBEA evaluated the treatment outcome as 'excellent' at 12 months compared with 24% of women undergoing TCRE and RB. These figures were 35% and 4% respectively at 24 months.
4.1.13 Five trials consistently reported shorter procedure times for TBEA compared with TCRE and/or RB. Of these, two studies reported the percentages of operations that took less than 30 minutes. For TBEA these percentages were 65% and 100%, and for TCRE and RB they were 24% and 53% respectively. These differences were significant in both studies (p < 0.05). The mean operating times were between 11.5 and 24 minutes in the TBEA groups compared with between 37 and 45 minutes in the TCRE and/or RB groups. The differences were statistically significant in all trials.
4.1.14 Six trials reported the proportion of women who required further intervention. At 12 months, between 1% and 10% of women in the TBEA group required further interventions compared with between 2% and 16% in the TCRE and/or RB groups. In one trial, 5% of women undergoing TBEA and 10% of women undergoing TCRE plus RB had had an additional procedure, and these percentages rose to 6% and 15% respectively at 24 months. This difference in the repeat surgery rate was statistically significant (p < 0.01). In the trial with the longest follow-up period, repeat procedures had been done for 15 of the 76 women (19.7%) in the TBEA group (13 hysterectomies and two repeat ablations), compared with 9 of 71 women (12.7%) in the RB group (seven hysterectomies, two repeat ablations, and one dilatation and curettage) at 60 months.
4.2.1 Only one published study was identified. Three economic analyses were made available to the Institute as part of manufacturers' submissions, and the Assessment Group developed its own model.
4.2.2 The published study compared the costs of vaginal hysterectomy, TBEA and RB in 147 women in France. The total costs for each treatment group were calculated 24–36 months after the surgery, taking into account the subsequent resource use only (for example, re-interventions). The total costs were estimated to be around £3670 for vaginal hysterectomy, £870 for TBEA and £910 for RB (€5321, €1263 and €1320 respectively, converted to pounds sterling at 2003 rates).
4.2.3 The Microsulis model suggested that MEA is less costly and more effective than other EA techniques, and therefore is a dominant strategy. However, hysterectomy was more effective but more costly than MEA, at an incremental cost-effectiveness ratio (ICER) of around £4600 per quality-adjusted life-year (QALY).
4.2.4 The Cavaterm model estimated that TBEA is cost saving when compared with hysterectomy or other EA techniques. Cost per treatment success was £767 for Cavaterm, £828 for Thermachoice, £865 for TCRE or RB, and £2050 for hysterectomy, based on RCT data only.
4.2.5 The Thermachoice model, which used the cost estimates from the published French study (see Section 4.2.2), estimated that the ICERs for hysterectomy and TCRE compared with TBEA were £1197 (€1736) and £950 (€1378) respectively per additional woman with amenorrhoea, £13,648 (€19,789) and £11,552 (€16,751) per additional woman with eumenorrhoea or less, and £9748 (€14,135) and £18,379 (€26,650) per additional satisfied patient.
4.2.6 The Assessment Group's model was a Markov model, which examined the progress of six hypothetical cohorts of women with HMB treated separately by TBEA, MEA, TCRE, TCRE and RB, RB, or hysterectomy. The model took the perspective of the NHS and calculated incremental cost utility between different treatment options over 10 years. This model concluded that the second-generation techniques (MEA and TBEA) are more cost effective than the first-generation techniques (TCRE and/or RB). Although base-case analysis showed that TBEA dominated MEA (in other words TBEA was less costly and more effective than MEA), the overall differences in costs and utilities were negligible, and moreover the results were sensitive to small changes in utility values. Both TBEA and MEA were dominated by hysterectomy; however the model did not take into account either patient preference or suitability. The ICER of hysterectomy versus second-generation EA techniques was around £2000 per QALY in the base-case analysis.
4.3.1 The Committee reviewed the data available on the clinical and cost effectiveness of TBEA and MEA, having considered evidence on the nature of the condition and the value placed on the benefits of these treatments from women with HMB, those who represent them, and clinical experts. It was also mindful of the need to take account of the effective use of NHS resources.
4.3.2 Based on the available evidence on the effectiveness of TBEA and MEA, the Committee concluded that TBEA and MEA are likely to be as effective as first-generation EA techniques in terms of reducing abnormal menstrual bleeding patterns in women with HMB. However, the Committee considered that there was not sufficient evidence to differentiate between TBEA and MEA in terms of their overall effectiveness when all potential outcomes were considered jointly.
4.3.3 The Committee took into account the potentially less invasive nature of the second-generation techniques and the possibility that they could be performed under local anaesthesia as outpatient procedures. The Committee was also mindful of the potential advantages of delivering these treatments (TBEA and MEA) under local anaesthesia and in an outpatient setting. However they heard from consultees that the application of EA techniques other than under general anaesthesia was by no means universal in the NHS.
4.3.4 Having reviewed the economic models submitted to the Institute, the Committee concluded that TBEA and MEA are cost-effective treatment alternatives for HMB. However, the Committee concluded that in the absence of reliable effectiveness data (particularly from head-to-head trials), it was not possible to draw conclusions on the relative clinical and cost effectiveness of TBEA and MEA. Additionally, the Committee was persuaded that the relative merits of these techniques varied greatly for individual patients, and was highly dependent on the specific outcome that was appropriate for any particular patient. The Committee therefore considered that the issue of choice for the individual rendered differences in overall effectiveness between the techniques less relevant. It concluded that these techniques may separately be appropriate for specific subgroups of women, and the choice between them should be made by the woman and the clinician responsible for her treatment, following informed discussion.
4.3.5 Having consulted with experts, the Committee concluded that the continued availability of first-generation EA techniques is important because for some women with HMB these techniques may remain the most appropriate options. The NHS should consider locally how it will ensure that both second-generation techniques are available, in order to facilitate appropriate choices for individual patients.
4.3.6 The Committee accepted that hysterectomy is the only option that can guarantee amenorrhoea, but considered that it should not be offered to patients by default, even when the desired outcome is the complete cessation of menstruation. The potential risks and benefits of all available options should be clearly explained to the individual. The patient and the clinician responsible for treatment should make the decision jointly.