Microwave endometrial ablation versus thermal balloon endometrial ablation (MEATBall): 5-year follow up of a randomised controlled trial




To compare long-term outcomes following microwave endometrial ablation (MEA™) and thermal balloon ablation (TBall).


Follow up of a prospective, double-blind randomised controlled trial at 5 years.


A teaching hospital in the UK.


A total of 320 women eligible for and requesting endometrial ablation.


Eligible women were randomised in a 1:1 ratio to undergo MEA or Tball. Postal questionnaires were sent to participants at a minimum of 5 years postoperatively to determine satisfaction with outcome, menstrual status, bleeding scores and quality of life measurement. Subsequent surgery was ascertained from the women and the hospital operative database.

Main outcome measures

The primary outcome measure was overall satisfaction with treatment. Secondary outcomes included evaluation of menstrual loss, change in quality of life scores and subsequent surgery.


Of the women originally randomised 217/314 (69.1%) returned questionnaires. Nonresponders were assumed to be treatment failures for data analysis. The primary outcome of satisfaction was similar in both groups (58% for MEA™ versus 53% for TBall, difference 5%; 95% CI −6 to 16%). Amenorrhoea rates were high following both techniques (51% versus 45%, difference 6%; 95% CI −5 to 17%). There was no significant difference in the hysterectomy rates between the two arms (9% versus 7%, difference 2%; 95% CI −5 to 9%).


At 5 years post-treatment there were no significant clinical differences in patient satisfaction, menstrual status, quality of life scores or hysterectomy rates between MEA™ and Thermachoice 3, thermal balloon ablation.


Few randomised trials comparing surgical interventions measure follow up of outcomes in the long term. This can lead to difficulty in establishing the true impact and worth of an intervention. Endometrial ablation is a well-researched treatment for heavy menstrual loss and a number of second-generation ablative techniques are recommended by the National Institute for Health and Care Excellence (NICE) for the treatment of this complaint.[1] Few of the commercially available second-generation techniques available have been evaluated in independent, adequately powered randomised trials with meaningful clinical and economic outcomes. Fewer still have undergone scrutiny in the long term.

Both microwave endometrial ablation (MEA™)[2] and Thermal Balloon (TBall)[3] endometrial ablation are NICE-recommended ablative techniques[1]. Microwave ablation, although recently removed from the UK market, has been independently evaluated in randomised trials with published follow-up data at 5 years[4] and 10 years.[5] This made it the ideal comparator for Thermachoice 3™, a thermal balloon which required re-evaluation as data in the literature pertained to earlier, perhaps less effective, versions of the balloon.[3, 6] The original results from this cohort demonstrated that the microwave technique was quicker to perform and required less postoperative analgesia, but clinical outcomes at 1 year were comparable between the two techniques.[7] Microwave ablation was likely to be more cost effective at 1 year.[8]

In this paper we will present and discuss the outcomes at 5 years post treatment.


Full methodological and operative details can be found in the original paper.[7] The trial protocol is held by the funding body: Chief Scientist Office, Scottish Government Health Directorates, ref. CZH/4/117 and also by the Health Services Research Unit, Aberdeen University. Local ethics committee approval was obtained for long-term follow up at 5 years. Women complaining of heavy menstrual loss who desired and were eligible for endometrial ablation were recruited from the gynaecology department of Aberdeen Royal Infirmary between January 2003 and January 2005. Eligible women were premenopausal, had completed their families, and gave their informed consent to participate within the trial. They had a uterine cavity length of <12 cm, no histopathological abnormalities of the endometrium and fibroids if present were <3 cm and not obstructing the uterine cavity. The women did not routinely undergo hysteroscopy before recruitment unless an abnormality was identified on transvaginal ultrasound scan. Women with previous caesarean section were included if scar thickness was <10 mm.

Sample size

The original power study determined that 290 recruits were required to give an 80% power of demonstrating a 12% difference in those totally or generally satisfied with treatment. Additionally, this number gives 80% power to detect a 15% difference in amenorrhoea rates (2P < 0.05) and 90% power to detect a difference in menstrual scores (pictorial assessment blood loss chart) of 10, again with significance at the 5% level.

In all, 320 recruits were randomised in a ratio of 1:1 to the MEA™ and TBall arms of the study after obtaining informed consent. Six post randomisation exclusions occurred so 314 women were treated in the trial. The mode of treatment was not revealed to the women, or the statistician.


Postal questionnaires were sent at a minimum of 5 years after the original procedure. The primary objective of this study was to detect any difference in patient satisfaction between the two treatments. Secondary outcomes were menstrual status, changes in health-related quality of life (Short Form-12 [SF-12] and EQ-5D) and any further surgery received. The questionnaire was sent to the participant. Those who did not respond were sent a postal reminder and, finally, a telephone call to ascertain whether they wished to participate. Data were entered into an SPSS (version 16, SPSS Inc., Chicago, IL, USA) database. Subsequent operations were established from the questionnaires and from the hospital database for all recruited women.

Statistical analysis

All statistical analyses were carried out using SPSS v20 (IBM Inc, Armonk, NY, USA) unless otherwise stated. Differences in proportions and 95% confidence intervals were calculated in Excel using Newcombe's method.[9] Bleeding scores, pain scores and health-related quality of life measures (EQ-5D and SF-12 scores) were compared using a linear regression model to estimate the mean difference between groups after adjusting for baseline values.[10] Changes in quality of life from baseline were compared using paired t-tests. Women who had had a hysterectomy at follow up were included in all comparisons. Intention-to-treat analysis was used, that is non-responders were assigned a negative response (that is presumed dissatisfaction) when calculating proportions for satisfaction with the procedure, willingness to recommend to a friend and amenorrhoea rates. This method of analysis was not appropriate for other menstrual outcomes or for Health-related Quality of Life measures as these values are calculated from questionnaire data, which are not available for non-respondents. The CONSORT scheme of reporting was adhered to.[11-13]


Two hundred and seventeen of the 314 (69.1%) women originally randomised returned completed questionnaires. The baseline characteristics of the 217 women successfully followed up were very similar to those of the total trial group and not statistically significantly different between either. Mean age at follow up was 48 years (SD 5.2) in both arms. The flow of participants through the trial is outlined in Figure 1.

Figure 1.

CONSORT Flowchart.


Baseline characteristics of those returning questionnaires at 5 years are shown in Table 1 and are comparable.

Table 1. Baseline patient characteristics
CharacteristicMEA™ (n = 157)TBall (= 157)
  1. Values are given as n (%) unless stated otherwise.

  2. BMI, body mass index; IQR, interquartile range; PBLAC, pictorial assessment blood loss chart; SF-12, short form 12.

  3. a

    Bleeding score value: each day of period is given a score from 1 mild to 6 severe. No bleeding received a zero value. Each day is added together to achieve a total score.

  4. b

    Pain score: each day of period is given a score from 1 mild to 6 severe. No pain received a zero value. Each day is added together to achieve a total score.

Age (years), mean (SD)43.1 (5.5)43.2 (5.1)
BMI (kg/m2), median (IQR)27.1 (5.5)27.8 (5.8)
Uterine cavity length (cm), mean (SD)8.4 (1.1)8.4 (1.2)
Previous vaginal delivery129 (83)136 (87)
Previous caesarean delivery40 (26)23 (15)
Previous cervical surgery11 (7)9 (6)
Premenstrual dysmenorrhoea117 (75)117 (75)
Double sanitary protection134 (86)129 (84)
Bleeding score, median (IQR)a24.0 (19.0–31.0)25.0 (19.0–30.0)
Pain score, median (IQR)b16.5 (10.0–22.8)16.0 (9.0–24.0)
EQ-5d, mean (SD)0.77 (0.23)0.77 (0.28)
SF-12 Physical score, mean (SD)48.6 (7.5)49.1 (8.6)
SF-12 Mental score, mean (SD)43.3 (9.9)43.9 (10.6)
PBLAC score, median (IQR)184 (94–351)194 (78–402)

There were no statistically significant differences between the two groups with respect to menstrual symptoms (Table 2). This was also the case for the number of women totally or generally satisfied with treatment MEA™ 91/157 (58.0%) and TBall 83/157 (52.9%); difference (95% CI) 5.1 (−5.9 to 16.1) and those who would recommend the treatment to a friend, MEA™ 104/157 (66.2%), TBall 89/157 (56.7%); difference (95% CI) 9.6 (−1.2 to 20.3).

Table 2. Menstrual outcomes
 MEA (%) (n = 157)TBall (%) (=157)Difference (95% CI)
  1. Values are given as n (%) unless stated otherwise.

  2. BMI, body mass index; IQR, interquartile range; PBLAC, pictorial assessment blood loss chart; SF-12, short form 12.

  3. *Bleeding score value: each day of period is given a score from 1 mild to 5 severe. No bleeding received a zero value. Each day is added together to achieve a total score.

  4. **Pain score: each day of period is given a score from 1 mild to 5 severe. No pain received a zero value. Each day is added together to achieve a total score.

Amenorrhoea 80 (51.0)70 (44.6)6.4 (−4.7 to 17.4)
 (n = 114)(n = 103) 
Periods light 31 (27.2)34 (33.0)−5.8 (−18.0 to 6.4)
Unchanged or heavier 3 (2.6)6 (5.8)−3.2 (−8.6 to 2.2)
Bleeding score
Median (interquartile range)0 (0–2)0 (0–4)−0.6 (−2.0 to 0.7)
Mean (SD)2.3 (4.8)3.0 (5.3)
Pain score
Median (interquartile range)0 (0–1.5)0 (0–4)−1.3 (−2.9 to 0.3)
Mean (SD)2.7 (5.6)3.7 (6.5)

The amenorrhoea rates are high and comparable for both modalities for those returning questionnaires. Even if all non-responders are assumed to be failures and still bleeding (except for non-responders known to have had a hysterectomy), then this gives intention-to-treat amenorrhoea rates of 51% (80/157) for MEA™ and 45% (70/157) for TBall (risk difference 6.4%, 95% CI −4.7 to 17.4%).

There were no significant differences in quality of life—mean (SD) between groups for SF-12 (physical functioning): MEA™ 51.1% (10.1), TBall 52.6% (8.6), difference (95% CI) −0.9 (−3.4 to 1.6); SF-12 (mental functioning): MEA™ 49.1% (9.4), TBall 49.4% (10.1), difference (95% CI) 0.2 (−2.4 to 2.8); or EQ-5D: MEA™ 0.83 (0.26), TBall 0.83 (0.26), difference (95% CI) 0.01 (−0.06 to 0.07). SF-12 demonstrated significant improvement from the baseline quality of life scores for both categories (< 0.01 for both physical and mental functioning) and this was not altered significantly from the 12-month follow up (Figure 2).

Figure 2.

SF-12 physical and mental component scores.

Subsequent treatment received or continued at 5 years

At a minimum of 5 years after treatment, the majority of women had not required further gynaecological surgery. There was no statistically significant difference in women undergoing hysterectomy, with ten women (8.8%) in the MEA™ arm and seven (6.8%) in the TBall arm (difference of 2.0%, 95% CI −5.1 to 9.1%). Following MEA™, of the ten hysterectomies, four were for heavy bleeding, one for cyclical pain, four for pain and bleeding and one for prolapse. In the TBall arm, one hysterectomy was for pain alone, one for heavy bleeding, four for combined pain and bleeding and one for fibroid pressure symptoms. There was one repeat ablation in the TBall arm and none in the MEA™ arm.


Main findings

The two trial arms remained balanced at the 5-year follow up despite the dropout of recruits. Those women who did not respond were analysed as treatment failures for the principal outcome measure, which gives the impression of deterioration in satisfaction with treatment over time. Although this is the accepted method of analysis it does not lend itself to comparison with results from other trials that have not used this method.

It is more likely, however, given the low hysterectomy rates, that satisfaction rates are higher and are likely to lie between the rates ascertained by intention-to-treat analysis and rates reported by responders (totally or generally satisfied, 91/111, 82%, for MEA™ versus 83/99, 84%, for TBall). The same argument also applies to amenorrhoea rates, which (by intention-to-treat analysis), for MEA™, assuming all non-responders are still bleeding, is 51%, but 68% for responders. For TBall the corresponding rates are 45 and 61%.

Although no statistically significant differences could be demonstrated between the two treatments it is important to point out that with only 219 responders there was not adequate power to detect smaller perhaps meaningful differences in clinical outcomes between the two ablative techniques. Given that hysterectomy rates were low and comparable it does support the fact that there is little difference in outcome in the long term. Subsequent operations were determined not only from the patient questionnaire but also from the hospital database. As fewer than 10% of women had left the region and this is the only hospital supplying gynaecological care, it is likely that these figures are representative of the original recruited cohort. Hysterectomy rates of around 16% have been previously reported for follow up after 5 years for MEA™.[5]

Women recruited had a subjective complaint of heavy bleeding and formal menstrual blood loss measurements were not performed. The majority of woman had a preoperative ultrasound and endometrial biopsy but hysteroscopy was not performed preoperatively unless clinically indicated. All women underwent hysteroscopy immediately before insertion of the ablative device to ensure no false passage or perforation. This mirrors recommended clinical practice, thereby increasing the generalisability of the results. Also, the procedures were all undertaken by a trainee and not a consultant hysteroscopic surgeon with extensive experience of both techniques. It is important to reiterate that during the treatment phase with the thermal balloon the pressure was maintained between 160 and 180 mmHg by injecting further dextrose solution during the treatment phase. This, along with the active impellor in Thermachoice 3™, circulating the heated fluid, may account for the better results achieved by the balloon in this study when compared with previous trials.

Heavy menstrual loss is known to cause significant deterioration in quality of life, and SF-12 and EQ-5D were used although neither has been formally validated for menstrual disorders. Neither is a condition-specific quality of life tool but SF-36 has been used in the past to demonstrate reduction in generic quality of life values for menorrhagia[14] and has also been shown to return to normative values following endometrial ablation for heavy menstrual loss.[3, 15] Importantly, normative values for the healthy female population are known for SF-12.[16] EQ-5D is used for economic calculations that were not repeated at 5-year follow up. The physical component of the SF-12 score demonstrated improved scores from baseline for responders and was maintained at normative levels. The mental component followed a similar pattern, with an overall improvement for both arms at 12 months, maintained at the normative values (49.2) by 5 years. Importantly, no significant difference was noted between the groups.

Strengths and limitations

The strength of this trial is that it was an independent, government-funded, double-blind randomised controlled trial and hence bias is minimised. The procedures were performed on women with a generic complaint of heavy menstrual loss where the women had decided upon endometrial ablation as treatment. It is entirely possible that some would not have true menorrhagia, but this reflects standard practice in the UK and through randomisation, prognostic factors should be equalised. There were limited pre-selection criteria with cavities up to 12 cm treated and no exclusion of smaller fibroids, which enhances the generalisability, as does the fact that the procedures were all performed by a trainee.

It is unfortunate that the numbers of women returning questionnaires failed to meet the original power study requirements, but the number of hysterectomies performed at 5 years was comparable and low at under 10% in each arm, which is reassuring. An operating pressure of between 160 and 180 mmHg was maintained throughout the TBall treatment phase, which was not the manufacturer's recommendation. A failure to do this may lead to inferior results.

Long-term follow up of interventional randomised trials is rare but offers invaluable information for the healthcare purchaser and potential patient. Five-year follow-up data are available for Novasure,[6] Microwave ablation[4] and now Thermachoice 3 and are highly informative, particularly as health-related quality of life data are also available in addition to menstrual outcomes. Ten-year data are also available for transcervical resection of the endometrium, microwave ablation[5] and Novasure,[17] and although it is commendable that these data are available, it perhaps offers principally reassurance of endometrial ablation's long-term safety and efficacy rather than the ability to discriminate between different techniques. This is because almost 50% of the recruits from the trials are menopausal by this stage, making menstrual status a meaningless outcome,[18] whereas almost all repeat surgeries, including hysterectomies for treatment failure, occur within the first 3 years of endometrial ablation.[19]


This trial confirms that Thermachoice 3™ achieves better results than earlier models of this balloon device.[2, 6] It is simple to use requiring minimal dilatation of the cervix and has been successfully used under local anaesthetic. Although treatment times are slower than MEA™ and short-term costs are slightly higher, clinical and quality of life outcomes are comparable with low hysterectomy rates. These results should be used and quoted when critically assessing Thermachoice 3™ rather than trials involving previous models, which are now obsolete.


This trial confirms that Thermachoice 3™ endometrial ablation achieves comparable long-term results to Microwave Endometrial Ablation, which is one of the most robustly evaluated second-generation devices. Since MEA™ has been removed from the market in the UK by a competing company, who purchased the distribution rights, it is important to prove that another method of ablation achieves encouraging long-term results. These results enhance choice for the gynaecologist and reduce the likelihood of a monopoly situation arising for the supply of endometrial ablation technology.

Disclosure of interest

KC has received financial support to attend conferences from Microsulis PLC and Ethicon Women's Health and Urology.

Contribution to authorship

AS recruited and undertook follow up of all women for this study, input the data and wrote the manuscript. AE undertook the statistical analysis and edited the manuscript. KC designed the trial and methodology, supervised the original procedures, co-wrote the manuscript and is guarantor.

Details of ethics approval

Funding was received from the Grampian Research Ethics Committee No. 00/0023.


This study was originally funded by a grant awarded by the Chief Scientist's Office, Scottish Government Health Directorates, ref CZH/4/117. Costs for the 5-year follow up were funded by the Gynaecology Endoscopy Research Fund, through the Endowments Department, Aberdeen Royal Infirmary.


The original trial up to 1-year follow up was funded by a grant awarded by the Chief Scientist's Office, Scottish Executive Health Department. We would like to thank all the women who participated. The views expressed are those of the authors.

Commentary on ‘Microwave endometrial ablation versus thermal balloon endometrial ablation’

The paper by Sambrook et al., reports the 5-year results of a randomised trial comparing two endometrial ablation devices and provides evidence that the ThermaChoice III system is associated with similar clinical outcomes to the now discontinued microwave endometrial ablation (MEA) device. Patient satisfaction was relatively high and hysterectomy rates were seemingly low. The authors should be commended for the execution of a high-quality double-blinded surgical trial.

However, ‘looking forward via the rearview mirror’, one is compelled to ask, just what is the clinical condition that has been treated in this study? Understandably, since recruitment was initiated in 2004, before the advent of the new FIGO systems for terminology and classification of causes of chronic abnormal uterine bleeding in the reproductive years (Munro et al. Int J Gynaecol Obstet 2011;113:3–13), terms like ‘menorrhagia’ and ‘heavy menstrual loss’ abound, and are used as if they were diagnoses unto themselves. However, we now know that the term ‘menorrhagia’ had so many different meanings in the literature that the International Federation of Obstetrics and Gynecology (FIGO) has deemed it obsolete. ‘Heavy menstrual bleeding’ should now be considered a symptom rather than a diagnosis (Fraser et al. Fertil Steril 2007;87:466–76).

Evaluation of this patient population for the cause of their chronic abnormal uterine bleeding reveals that a proportion of women had endometrial polyps and submucous leiomyomas. Furthermore, there was no apparent evaluation of ovulatory status, the presence or absence of coagulopathies, and whether or not adenomyosis was present. Assessment for adenomyosis would have been of interest, in part because the prevailing reasons for hysterectomy involved one or a combination of pain and bleeding, possible manifestations of endometrial glands and stroma within the myometrium. Indeed, the investigators describe, in the original Methods section, the removal of identified polyps immediately before performing the procedure, a circumstance that might confound the results leaving one to ask ‘Are the patients better because of the polypectomy or the ablation?’ Understanding the disorder-specific response of pharmacotherapy and minimally invasive surgical interventions such as endometrial ablation is important to inform therapeutic decision-making and patient choices. The design of future studies, if based around FIGO's PALM–COEIN system, will help us to sort out some of these questions.

On a historical note: the authors refer to endometrial ablation devices like microwave endometrial ablation and the Thermachoice balloon as ‘second-generation’, suggesting that resectoscopic techniques represent the ‘first-generation’ of endometrial ablation systems. Indeed, this appears to be a case of ‘generational amnesia’ because the literature contains multiple series of nonhysteroscopic/resectoscopic radiofrequency and cryotherapeutic systems starting in 1937 (Bardenheuer. Zentralbl Gynakol 1937;4:209–11). Steam-based endometrial ablation techniques were also described in the late nineteenth century. These techniques long pre-dated the introduction of hysteroscopic and resectoscopic techniques. It is important to maintain an awareness of our collective history and not forget the unique contributions of those who have come before us.

Disclosure of interests

The author is a consultant for Ethicon Inc., manufacturer of ThermaChoice; Hologic Inc., manufacturer of NovaSure; Boston Scientific, manufacturer of the Hydrothermablator; and Bayer HealthCare; manufacturer of the Mirena LNG-IUS-20.

  • MG Munro

  • Department of Obstetrics & Gynecology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA