Microwave and thermal balloon endometrial ablation for heavy ...

Health Technology Assessment 2004; Vol. 8: No. 3The effectiveness and costeffectivenessof microwaveand thermal balloon endometrialablation for heavy menstrualbleeding: a systematic review andeconomic modellingR Garside, K Stein, K Wyatt, A Roundand A PriceFebruary 2004Health Technology AssessmentNHS R&D HTA ProgrammeHTA

HTAHow to obtain copies of this and other HTA Programme reports.An electronic version of this publication, in Adobe Acrobat format, is available for downloading free ofcharge for personal use from the HTA website (http://www.hta.ac.uk). A fully searchable CD-ROM isalso available (see below).Printed copies of HTA monographs cost £20 each (post and packing free in the UK) to both public andprivate sector purchasers from our Despatch Agents.Non-UK purchasers will have to pay a small fee for post and packing. For European countries the cost is£2 per monograph and for the rest of the world £3 per monograph.You can order HTA monographs from our Despatch Agents:– fax (with credit card or official purchase order)– post (with credit card or official purchase order or cheque)– phone during office hours (credit card only).Additionally the HTA website allows you either to pay securely by credit card or to print out yourorder and then post or fax it.Contact details are as follows:HTA DespatchEmail: orders@hta.ac.ukc/o Direct Mail Works Ltd Tel: 02392 492 0004 Oakwood Business Centre Fax: 02392 478 555Downley, HAVANT PO9 2NP, UK Fax from outside the UK: +44 2392 478 555NHS libraries can subscribe free of charge. Public libraries can subscribe at a very reduced cost of£100 for each volume (normally comprising 30–40 titles). The commercial subscription rate is £300per volume. Please see our website for details. Subscriptions can only be purchased for the current orforthcoming volume.Payment methodsPaying by chequeIf you pay by cheque, the cheque must be in pounds sterling, made payable to Direct Mail Works Ltdand drawn on a bank with a UK address.Paying by credit cardThe following cards are accepted by phone, fax, post or via the website ordering pages: Delta, Eurocard,Mastercard, Solo, Switch and Visa. We advise against sending credit card details in a plain email.Paying by official purchase orderYou can post or fax these, but they must be from public bodies (i.e. NHS or universities) within the UK.We cannot at present accept purchase orders from commercial companies or from outside the UK.How do I get a copy of HTA on CD?Please use the form on the HTA website (www.hta.ac.uk/htacd.htm). Or contact Direct Mail Works (seecontact details above) by email, post, fax or phone. HTA on CD is currently free of charge worldwide.The website also provides information about the HTA Programme and lists the membership of the variouscommittees.

The effectiveness and costeffectivenessof microwaveand thermal balloon endometrialablation for heavy menstrualbleeding: a systematic review andeconomic modellingR Garside, 1* K Stein, 1 K Wyatt, 2 A Round 1and A Price 31 Peninsular Technology Assessment Group, Peninsula Medical School,Universities of Exeter and Plymouth, UK2 Research and Development Support Group, Peninsula Medical School,Universities of Exeter and Plymouth, UK3 Southampton Health Technology Assessment Centre, University ofSouthampton, UK* Corresponding authorDeclared competing interests of authors: nonePublished February 2004This report should be referenced as follows:Garside R, Stein K, Wyatt K, Round A, Price A. The effectiveness and cost-effectiveness ofmicrowave and thermal balloon endometrial ablation for heavy menstrual bleeding: asystematic review and economic modelling. Health Technol Assess 2004;8(3).Health Technology Assessment is indexed in Index Medicus/MEDLINE and Excerpta Medica/EMBASE.

NHS R&D HTA ProgrammeThe NHS R&D Health Technology Assessment (HTA) Programme was set up in 1993 to ensurethat high-quality research information on the costs, effectiveness and broader impact of healthtechnologies is produced in the most efficient way for those who use, manage and provide carein the NHS.The research reported in this monograph was commissioned by the HTA Programme on behalf ofthe National Institute for Clinical Excellence (NICE). Technology assessment reports are completedin a limited time to inform the appraisal and guidance development processes managed by NICE.The review brings together evidence on key aspects of the use of the technology concerned. However,appraisal and guidance produced by NICE are informed by a wide range of sources.The research reported in this monograph was funded as project number 01/61/01.The views expressed in this publication are those of the authors and not necessarily those of theHTA Programme, NICE or the Department of Health. The editors wish to emphasise that fundingand publication of this research by the NHS should not be taken as implicit support for anyrecommendations made by the authors.Criteria for inclusion in the HTA monograph seriesReports are published in the HTA monograph series if (1) they have resulted from workcommissioned for the HTA Programme, and (2) they are of a sufficiently high scientific qualityas assessed by the referees and editors.Reviews in Health Technology Assessment are termed ‘systematic’ when the account of the search,appraisal and synthesis methods (to minimise biases and random errors) would, in theory, permitthe replication of the review by others.HTA Programme Director:Series Editors:Managing Editors:Professor Tom WalleyDr Ken Stein, Professor John Gabbay, Dr Ruairidh Milne,Dr Chris Hyde and Dr Rob RiemsmaSally Bailey and Caroline CiupekThe editors and publisher have tried to ensure the accuracy of this report but do not accept liabilityfor damages or losses arising from material published in this report.ISSN 1366-5278© Queen’s Printer and Controller of HMSO 2004This monograph may be freely reproduced for the purposes of private research and study and may be included in professional journalsprovided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising.Applications for commercial reproduction should be addressed to HMSO,The Copyright Unit, St Clements House,2–16 Colegate, Norwich, NR3 1BQ.Published by Gray Publishing, Tunbridge Wells, Kent, on behalf of NCCHTA.Printed on acid-free paper in the UK by St Edmundsbury Press Ltd, Bury St Edmunds, Suffolk.T

Health Technology Assessment 2004; Vol. 8: No. 3AbstractThe effectiveness and cost-effectiveness of microwave andthermal balloon endometrial ablation for heavy menstrualbleeding: a systematic review and economic modellingR Garside, 1* K Stein, 1 K Wyatt, 2 A Round 1 and A Price 31 Peninsular Technology Assessment Group, Peninsula Medical School, Universities of Exeter and Plymouth, UK2 Research and Development Support Group, Peninsula Medical School, Universities of Exeter and Plymouth, UK3 Southampton Health Technology Assessment Centre, University of Southampton, UK* Corresponding authorObjectives: To estimate the clinical effectiveness andcost-effectiveness of microwave endometrial ablation(MEA) and thermal balloon endometrial ablation(TBEA) for heavy menstrual bleeding (HMB), comparedwith the existing (first-generation) endometrial ablation(EA) techniques of transcervical resection (TCRE) androllerball (RB) ablation, and hysterectomy.Data sources: Electronic databases, bibliographies ofarticles, and also experts in the field and relevantindustry bodies were asked to provide information.Review methods: A detailed search strategy wascarried out to identify systematic reviews and controlledtrials of MEA and TBEA versus first-generationtechniques for EA. In addition to electronic databasesearching, reference lists were hand-searched andinformation sought from manufacturers of EA devicesand by experts in the field. A deterministic Markovmodel was developed to assess cost-effectiveness. Datafor the model were taken from a range of sources.Results: The systematic review of first-generation EAtechniques versus hysterectomy found that EA offered analternative to hysterectomy for HMB, with fewercomplications and a shorter recovery period. Satisfactionand effectiveness were high for both MEA and TBEA.Costs were lower with EA although the differencenarrows over time. Second-generation EA techniques arean alternative treatment to first-generation techniquesfor HMB, and first-generation techniques are known tooffer an alternative to hysterectomy. Although no trials ofsecond-generation techniques and hysterectomy havebeen undertaken, it seems reasonable to assume thatsecond-generation techniques also offer an alternativesurgical treatment. Using the model to assess costeffectiveness,costs were very slightly higher for MEAwhen compared to TBEA, and differences in qualityadjustedlife-years (QALYs) were negligible. For MEAcompared with transcervical resection of the© Queen’s Printer and Controller of HMSO 2004. All rights reserved.endometrium (TCRE) and RB ablation, costs wereslightly lower with MEA and MEA accrued very slightlymore QALYs. Compared with hysterectomy, MEA costsless and accrues slightly fewer QALYs. For TBEAcompared with TCRE and RB ablation, costs were lowerwith TBEA and TBEA accrued slightly more QALYs.Compared with hysterectomy, TBEA costs moderatelyless and accrues moderately fewer QALYs.Conclusions: Overall, there were few significantdifferences between the outcomes of first- and secondgenerationtechniques including bleeding, satisfactionand QoL measures and repeat surgery rates. Secondgenerationtechniques had significantly shorteroperating and theatre times and there appear to befewer serious perioperative adverse effects withsecond-generation techniques and postoperative effectsare similar. Compared with hysterectomy, TCRE andRB are quicker to perform and result in shorterhospitalisation and faster return to work. Hysterectomyresults in more adverse effects and is more expensive,although the need for retreatment leads this differenceto decrease over time. Satisfaction with hysterectomyis initially higher, but there is no significant differenceafter 2 years. The economic model suggests thatsecond-generation techniques are more cost-effectivethan first-generation techniques of EA for HMB. BothTBEA and MEA appear to be less costly thanhysterectomy, although the latter results in moreQALYs. Further research is suggested to make directcomparisons of the cost-effectiveness of secondgenerationEA techniques, to carry out longer termfollow-up for all methods of EA in RCTs, and todevelop more sophisticated modelling studies. Furtherresearch is also recommended into HMB to establishhealth-state utility values, its surgical treatment,convalescence, complications of treatment, symptomsand patient satisfaction.iii

Health Technology Assessment 2004; Vol. 8: No. 3ContentsGlossary and list of abbreviations .............Executive summary ....................................viixi6 Conclusions ................................................ 81Acknowledgements .................................... 831 Aim of the review ...................................... 12 Background ................................................ 3Description of underlying healthproblem ...................................................... 3Current service provision ........................... 7Description of new intervention ................ 13Summary .................................................... 153 Methods ..................................................... 17Research questions ..................................... 17Review team and advisory group ............... 17General methods ........................................ 17Assessment of microwave and thermalballoon ablation ......................................... 17Economic evaluation .................................. 184 Results ........................................................ 23Systematic review – effectiveness ................ 23Included systematic reviews ....................... 23Existing systematic reviews – findings ....... 24Systematic review of hysterectomy versusfirst-generation EA techniques .................. 24Controlled trials of second-generation EAtechniques .................................................. 24Summary .................................................... 59Economic evaluation of microwave andthermal balloon ablation ............................ 59Summary .................................................... 66Economic analyses supplied by industry ... 66Summary .................................................... 73Impact on NHS budget .............................. 735 Discussion ................................................... 75Main results ................................................ 75Assumptions, limitations anduncertainties ............................................... 76Need for further research .......................... 79References .................................................. 85Appendix 1 Pictorial blood loss assessmentchart ........................................................... 89Appendix 2 Research protocol .................. 91Appendix 3 Search strategy ...................... 95Appendix 4 Excluded studies .................... 97Appendix 5 Included systematic reviews –QUOROM checklist ................................... 99Appendix 6 Included systematicreviews ........................................................ 103Appendix 7 Included controlled studydetails ......................................................... 109Appendix 8 Graphs showing sensitivityanalyses for MEA and TBEA ..................... 135Appendix 9 Quality assessment of industrysubmittedeconomic analyses ..................... 145Appendix 10 Parameters used in the industryand PenTAG economic models .................. 153Health Technology Assessment reportspublished to date ....................................... 157Health Technology AssessmentProgramme ................................................ 165v

Health Technology Assessment 2004; Vol. 8: No. 3Glossary and list of abbreviationsTechnical terms and abbreviations are used throughout this report. The meaning is usually clear fromthe context, but a glossary is provided for the non-specialist reader. In some cases, usage differs in theliterature, but the term has a constant meaning throughout this review.GlossaryAdenomyosis The presence of endometriumin the myometrium. Can cause heavymenstrual bleeding and pain.AmenorrhoeaCervixuterus.Absence of periods.The lower, narrower end of theCornua The horn-shaped top of the uterusleading to the Fallopian tubes.Cystometry A method for measuring thepressure-volume relationship of the bladder.Diathermy Use of a high-frequency electricalcurrent to produce heat that destroys tissuesthrough cutting or electrocoagulation. Thepatient’s body forms part of the circuit.DysmenorrhoeaPainful periods.Electrocautery Cauterisation of tissue usingan electric current to generate the heat.Cauterisation destroys the tissue and causesscarring.Endometriosis A condition where tissueresembling the endometrium occurs outsidethe uterus. The tissue responds to themenstrual cycle causing internal bleeding andpain.Endometrium The inner lining of the uterusthat thickens and sloughs off during themenstrual cycle.FundusThe higher, wider end of the uterus.Haematometra A collection of blood andother menstrual fluids in the uterus, whichcauses it to distend.Haematosalpinx A collection of blood in thefallopian tubes – postendometrial ablation, thismay be caused by bleeding from untreatedislands of endometrium at the cornea.Hyperplasia The abnormal increase in thenumber of normal cells in a tissue.Hypomenorrhoea Regular periods withblood loss less than normal.Hysterectomy The surgical removal of theuterus; may include removal of the cervix.Hegar A German gynaecologist who gave hisname to a series of graduated, cylindricalinstruments used to dilate the cervix.Hysteroscope An instrument using fibreoptictechnology that allows direct visualisationof the uterine cavity. Channels in theinstrument allow instruments to be inserted toperform ablations.Iatrogenic An adverse effect inadvertentlyinduced through treatment.Laparoscope A device used in surgery thatallows visualisation through the use of fibreoptics.EumenorrhoeaNormal periods.LeiomyomasFibroids.Fibroids Benign, smooth muscle tumours ofthe uterus.continuedvii© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Glossary and list of abbreviationsGlossary continuedMenopause Cessation of menstruation,usually around age 50 years.Peri-menopausalmenopause.Around the time of theMeno-metrorrhagiamenstrual bleeding.Frequent, excessivePolyp A mass of tissue on the mucosal lining.In this case, in the uterus.Menorrhagia Heavy menstrual bleeding,clinically defined as more than 80 ml of bloodper cycle, but more usually defined subjectivelyby the woman.Menstruation The cyclic, physiologicaldischarge of blood and mucosal tissues throughthe vagina from the non-pregnant uterus. It isunder hormonal control and recurs atapproximately 4-week intervals.Metrorrhagia Irregular, sometimesprolonged, menstrual bleeding.Myometriumthe uterus.NecrosisThe outer muscular layer ofCell death.OligomenorrhoeaFew or scanty periods.Post-ablation sterilisation syndrome Inpreviously sterilised women accumulation ofthe blood in the Fallopian tubes, which maycause severe pelvic pain.Premenstrual syndrome A combination ofemotional and physical features that occurcyclically in women. May include moodchanges, bloating, breast tenderness, fatigueand other symptoms.PyrexiaFever.Salpingo-oophorectomy Surgical removal ofthe Fallopian tubes and the ovaries.Uterus The womb. A hollow, muscular, pearshapedorgan in which the embryo isnourished.Pelvic inflammatory disease Aninflammatory process that may be caused bysexually transmitted infection, ovarian cysticdisease or infections after childbirth.viii

Health Technology Assessment 2004; Vol. 8: No. 3List of abbreviationsANCOVAanalysis of covarianceMRCMedical Research CouncilANOVAanalysis of varianceMRImagnetic resonance imagingBMICID&Cbody mass indexconfidence intervaldilation and curettageNICENSAIDNational Institute for ClinicalExcellencenon-steroidal anti-inflammatorydrugDUBdysfunctional uterine bleedingORodds ratioDVTEAFDAdeep vein thrombosisendometrial ablationFood and Drug Administration(USA)PBACPIDPMSpictorial blood loss assessmentchartpelvic inflammatory diseasepremenstrual syndromeGAgeneral anaestheticQALYquality-adjusted life-yearGIgastrointestinalQoLquality of lifeGnRHHESgonadotrophin-releasinghormoneHospital Episode StatisticsRBRCOGrollerballRoyal College of Obstetriciansand GynaecologistsHMBHTAICERITTIUDheavy menstrual bleedinghydrotherm ablatorincremental cost-effectivenessratiointention-to-treatintrauterine deviceRCTRRSDSESF-36randomised controlled trialrelative riskstandard deviationstandard errorShort Form with 36 ItemsIUSLALHRHLNGLTFUintrauterine systemlocal anaestheticluteinising hormone-releasinghormonelevonorgestrellost to follow-upTBEATCRETTOTVSthermal balloon endometrialablationtranscervical resection of theendometriumtime trade-offtransvaginal ultrasound(sonography)MEAmicrowave endometrial ablationUTIurinary tract infectionAll abbreviations that have been used in this report are listed here unless the abbreviation is well known (e.g. NHS), orit has been used only once, or it is a non-standard abbreviation used only in figures/tables/appendices in which casethe abbreviation is defined in the figure legend or at the end of the table.ix© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Executive summaryObjectiveThe aim of the project was to estimate the clinicaleffectiveness and cost-effectiveness of microwaveendometrial ablation (MEA) and thermalballoon endometrial ablation (TBEA) for heavymenstrual bleeding (HMB) compared withthe existing (first-generation) endometrialablation (EA) techniques of transcervicalresection (TCRE) and rollerball (RB) ablation,and hysterectomy.Description of proposed serviceThe technologies examined in this review areMEA and TBEA for the treatment of HMB. Both ofthese, also referred to as second-generation EAtechniques, aim to destroy the endometrial lining ofthe uterus, thereby reducing or eliminatingmenstrual bleeding. To achieve endometrialdestruction, TBEA uses a balloon catheter in whichhot water is circulated for a prescribed amount oftime. MA uses microwaves of a wavelength that willbe absorbed to a defined depth of tissue. Bothtreatments may be performed under local orgeneral anaesthetic and are performed withoutdirect visualisation of the uterus.Epidemiology and backgroundHMB (or menorrhagia) is defined as the cyclicalloss of more than 80 ml of blood over severalconsecutive cycles. HMB is a common complaintfor which one in 20 women aged 30–49 yearsconsult their general practitioner each year(approximately 1.5 million women in England andWales). Quality of life may be impaired by suchbleeding.Current treatments for HMB include various drugregimens, such as tranexamic acid, mefenamicacid, the combined pill and the progestogenreleasingintrauterine system. Danazol, gestrinoneand gonadotrophin-releasing hormone (GnRH)analogues may be used as second-line medicaltreatment. Current surgical interventions includehysterectomy or minimally invasive proceduressuch as TCRE and RB ablation.Over 51,000 hysterectomies were performed inthe public sector in England in 1999–2000. Inabout half of these cases, HMB would havebeen the presenting complaint, and in half ofthese, the uterus would have been normal. In1998–9 more than 16,000 admissions for EA wererecorded.This report assesses the effectiveness and costeffectivenessof MEA and TBEA compared withspecific existing surgical techniques for HMB,that is, first-generation EA techniques [byresection (TCRE) and/or RB] and hysterectomy.Number and quality of studiesand direction of evidenceA detailed search strategy was carried out toidentify systematic reviews and controlled trials ofMEA and TBEA versus first-generation techniquesfor EA. In addition to electronic databasesearching, reference lists were hand-searched andinformation sought from manufacturers of EAdevices and by experts in the field.Two good-quality systematic reviews, of theeffectiveness of hysterectomy versus firstgenerationablation methods and endometrialdestruction techniques for HMB (2002), wereincluded.Two randomised controlled trials (RCTs) of MEAand eight trials of TBEA versus first-generationtechniques were included. These trials includea total of 1561 women, with sample sizesranging from 20 to 322 (median 143). Twoof the TBEA trials were non-RCTs and the restwere RCTs.The quality of the trials was variable. The MEAtrials included more participants than TBEA trialsand were of higher quality and applicability to theUK. Two TBEA studies were not randomised;controls in one were women who underwentfirst-generation EA at the same institution, and inthe other two consecutive cohorts were compared.Of the RCTs, seven used appropriate allocation togroups; one MEA study reported blind assessmentof outcomes; one MEA and four TBEA studiesxi© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Executive summaryxiishowed that the groups were comparable atbaseline and six studies (one MEA and five TBEA)gave the same intervention and control treatmentto all women. Both MEA studies usedsubcutaneous GnRH analogues as an endometrialpre-thinning agent in both intervention andcontrol groups. Of the TBEA trials, two gave adilation and curettage (D&C) immediately prior tothe operation in both arms of the trial, two gaveGnRH analogues to women in both arms of thetrial and one gave no pretreatment to thoseundergoing TBEA, and GnRH to those in thecontrol group. One gave D&C to womenundergoing TBEA, and GnRH to womenundergoing TCRE.Only one MEA and three TBEA studies reportedundertaking a sample size calculation. One ofthese (TBEA) did not recruit sufficient participantsto meet requirements. Loss to follow-up wasbetween 0 and 46% (median 3.5%) – the highestfigure at 5 years of follow-up (TBEA versus RB).Of the six studies that reported some loss tofollow-up, two reported using intention-to-treat(ITT) analysis, although one appears to have useddifferent denominators for some variables. Onestudy does not report loss to follow-up, but doesnot appear to have data on all recruited women.Based on the adequacy of the description ofparticipant characteristics and inclusion criteria,the generalisability of the studies was judged byreviewers as high in one MEA and three TBEAcases, medium in three TBEA studies and low inone MEA and two TBEA studies. Main outcomemeasures were measured independently ineight cases and were uncertain in two TBEAstudies.Summary of benefitsThe systematic review of first-generation EAtechniques versus hysterectomy found that EAoffered an alternative to hysterectomy for HMB,with fewer complications and a shorter recoveryperiod. Satisfaction and effectiveness were high forboth techniques. Costs were lower with EAalthough the difference narrows over time.Owing to clinical heterogeneity between trials offirst- and second-generation EA techniques, metaanalysiswas not undertaken.The included studies of MEA and TBEA did notshow a significant difference betweenamenorrhoea rates after first-generation comparedwith second-generation techniques. Only onestudy showed a first-generation technique (RB) tobe significantly superior for the outcome ofamenorrhoea measured at 2 years. The medianproportion of women with the outcome ofamenorrhoea is higher among those treated withMEA (46%) than those with TBEA (14%), althoughthe ranges overlap (MEA 36–55%; TBEA 10–40%)and the amenorrhoea rates in the MEA trials werealso higher for the control group. No comparisonbetween MEA and TBEA should be inferred onthe basis of amenorrhoea rates between secondgenerationtechniques alone as there were similardifferences between control groups across trials.No significant differences between first- andsecond-generation techniques of EA were shownfor any other measure of bleeding.No significant differences between the results offirst- and second-generation EA were found fordysmenorrhoea or premenstrual symptoms.Differences in patient satisfaction reportedbetween first- and second-generation EAtechniques were not significant. One study usedthe Short Form with 36 Items to measure qualityof life (QoL) and found that six of the measuresimproved significantly after MEA, as did seven ofthe items for women in the TCRE/RB treatmentgroup.Compared with first-generation EA techniques,second-generation techniques resulted insignificantly shorter operating and theatre times,but not in postoperative length of stay or recoverytime.Perioperative and postoperative adverse effectswere few with both first- and second-generationtechniques, but there were fewer seriousperioperative adverse effects with MEA and nonewith TBEA compared with first-generationtechniques. Postoperative adverse effect rates weresimilar.Second-generation EA techniques are analternative treatment to first-generationtechniques for HMB. First-generation techniquesare known to offer an alternative to hysterectomy.Although no trials of second-generationtechniques and hysterectomy have beenundertaken, it seems reasonable to assume thatsecond-generation techniques also offer analternative surgical treatment. No head-to-headtrials of second-generation techniques have beenundertaken and there is not enough evidence toidentify differences between the clinicaleffectiveness of TBEA and MEA.

Health Technology Assessment 2004; Vol. 8: No. 3CostsCosts of technologies were estimated for 2002.The costs of TBEA and MEA were similar at£1273 and £1295 per procedure, respectively.Methods used to calculate costs may not have beensufficiently sensitive to measure such smallapparent differences with precision. The cost ofsecond-generation ablation is slightly less thancombined TCRE and RB ablation at £1614 butslightly more than RB at £1191. Abdominalhysterectomy costs £2275.Cost-effectivenessA deterministic Markov model was developed toassess cost-effectiveness. Data for the model weretaken from a range of sources. For MEA comparedwith TBEA, costs were very slightly higher forMEA (£1448 versus £1324 per woman), anddifferences in quality-adjusted life-years (QALYs)were negligible (8360.70 versus 8360.77 for thewhole cohort). For MEA compared with TCRE andRB ablation, costs were slightly lower with MEA(£1448 versus £1732 TCRE, £1752 RB and £1785TCRE/RB combined) and MEA accrued veryslightly more QALYs (8.361 versus 8.357 TCRE,8.360 RB and 8.358 TCRE/RB). Compared withhysterectomy, MEA costs less (£1448 versus £2320)and accrues slightly fewer QALYs (8.361 versus8.774).For TBEA compared with TCRE and RB ablation,costs were lower with TBEA (£1324 versus £1732TCRE, £1752 RB and £1785 TCRE/RB combined)and TBEA accrued slightly more QALYs (8.361versus 8.357 TCRE, 8.360 RB and 8.358TCRE/RB). Compared with hysterectomy, TBEAcosts moderately less (£1324 versus £2320) andaccrues moderately less QALYs (8.361 versus 8.774).Sensitivity analysesThe economic model was found to be particularlysensitive to changes in the utility value for womenwho had recovered from having an EA, in otherwords, women who were ‘well’. To a lesser extent,recurrence of HMB and the cost of the procedureswere also important in the analysis.Limitations of the calculationsGiven the paucity of data about utility values forthe health states relating to HMB, EA and postconvalescence,accurate estimates of costs perQALY are difficult to ascertain. As absolute costsand QALYs for MEA and TBEA are very similar,small changes in inputs relating to aspects of theprocedure that affect costs can lead to largechanges in the model outputs. There must,therefore, be considerable uncertainty about theprecision of these results. In particular, we are notconfident that available data are significantlyrobust to support comparison between secondgenerationtechniques.Other important issues regardingimplicationsLonger term follow-up is required to collectfurther data on failure rates and subsequent retreatment.TBEA is not suitable for women with largeruterine cavities (>12 cm) and those with uterinepathology or abnormalities. This may account foras many as 60% of women with HMB, althoughestimates are uncertain.Notes on the generalisability ofthe findingsOf the 10 included trials, five TBEA studiesexcluded women with fibroids and one TBEAstudy included only women with fibroids. Thismay not represent those women consideredsuitable for EA in routine practice and mayinfluence effectiveness. In addition, only one study(of MEA) uses self-reported menorrhagia as aninclusion criteria, as would be usual in clinicalpractice. For the five studies (one of MEA and fourof TBEA) using stringent measurements of HMBbased on high pictorial blood loss assessmentchart scores, higher rates of satisfaction may resultas all have objectively measured menorrhagiainitially. Such women have been shown to ratetreatment as more satisfactory than women withless bleeding. Finally, one TBEA study includessome women who are post-menopausal but whodid not wish to stop taking hormone treatment.The authors believe that this group is unlikely,currently, to be treated by EA in the UK.ConclusionsBoth MEA and TBEA techniques appear to offereffective alternatives in the surgical treatment ofwomen with HMB.xiii© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Executive summarySecond-generation techniques are quicker toperform and appear to provide similar outcomesto first-generation approaches. First-generationtechniques are associated with fewer adverseeffects than hysterectomy and there is evidence infavour of greater safety for second- over firstgenerationtechniques. In trials between first- andsecond-generation techniques, there were very fewsignificant differences in the main clinicaloutcomes.In essence, there seems to be little discernibledifference between second-generation techniqueson the basis of currently available data. However,TBEA may be suitable for fewer women as it hasmore restrictions on uterine size, abnormality andpathology. Both MEA and TBEA appear to offersimilar outcomes to older ablation techniques atsimilar or lower costs. It is not possible to predictwhich patients will become amenorrhagic and thedifferences are small. If amenorrhoea is thepreferred outcome, hysterectomy is the mosteffective technology, but with higher costs. Thecost–utility ratio for hysterectomy versus EA iswithin the range considered by decision-makers torepresent acceptable value for money.Need for further research●●Head-to-head comparisons of secondgenerationEA techniques should be considered.Longer term follow-up for all methods of EA inRCTs will provide better information about●●●●●failure rates and repeat procedures, in additionto checking whether longer term complicationsare an issue.More sophisticated modelling studies mayimprove estimates of cost-effectiveness, takinginto account population heterogeneity, andwould permit exploration of issues relevant toimplementation such as waiting times anddetailed budget impact.Given the importance of the utility values indetermining the cost-effectiveness of treatmentsfor HMB, further research to establish utilitiesfor the states of HMB, its surgical treatment,convalescence and complications of treatmentwould be valuable.Future studies of HMB should use validatedQoL measures and established modes ofmeasuring patient satisfaction both with theprocedure and with the outcomes.Further research into the effect of theconstellation of symptoms associated withmenstruation (such as pain, bloating andbreast tenderness) and the part that thesesymptoms play in women’s perceptions ofbleeding and the effect of its treatmentcould help to establish which women willfind treatment of bleeding aloneacceptable.Alternative models of care for EA should befurther investigated, including differentoperators (non-consultant medical staff andspecialist nurses) and different settings (officeversus operating theatre).xiv

Health Technology Assessment 2004; Vol. 8: No. 3Chapter 1Aim of the reviewThe aim of the project was to estimate theclinical effectiveness and cost-effectiveness ofmicrowave endometrial ablation (MEA) andthermal balloon endometrial ablation (TBEA) forheavy menstrual bleeding (HMB) compared withthe existing (first-generation) endometrial ablation(EA) techniques of transcervical resection (TCRE)and rollerball (RB) ablation, and hysterectomy.1© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Chapter 2BackgroundDescription of underlying healthproblemHMB (menorrhagia) affects many women. One in20 women aged 30–49 years consults her generalpractitioner (GP) with this complaint each year,approximately 1.5 million women in England andWales. 1 Referrals for menstrual disorders accountfor about 20% of all those to specialistgynaecology services. 2 By its nature, HMB is achronic, cyclical problem that may have physical,emotional and social impacts in addition toaffecting a woman’s ability to carry out hernormal activities. A study of 348 women in generalpractice found that over half said HMB was thecause of anxiety or depression and moodinessor irritability. In addition, over one-third saidHMB interfered with relationships, spoilt theirsex life and interfered with hobbiesor holidays. For 14% of women, HMB had animpact on their ability to carry out their job. 3Regular blood loss of 50–60 ml per cyclewill lead to a negative iron balance for mostwomen. 4Defining menorrhagiaMenorrhagia is objectively defined as the loss ofmore than 80 ml of blood per cycle over severalconsecutive cycles. 5 However, objectivemeasurement is difficult and several studies haveshown that between 35 and 60% of women whopresent with the complaint of HMB haveobjectively measured blood loss in the normalrange. 6,7 Conversely, there is also a proportionof women who do not seek help although theycan be shown to have ‘abnormally heavy’ bloodloss. 8Issues in the measurement of HMB, associatedproblems and their impact are discussed further inthe section ‘Measurement of blood loss’ below.Causes of HMBPossible causes of HMB are shown in Table 1. Nonpathologicalcauses are poorly understood and areusually referred to under the name dysfunctionaluterine bleeding, which is the commonest cause. 9Studies examining the efficacy of drug treatmentsin women with HMB have suggested that womenwho fail to respond to effective drug treatmentmay have an underlying cause that may only bedetected at later hysterectomy. 11,12 It isrecommended in the Royal College ofObstetricians and Gynaecologists (RCOG)guidelines that women should be examined bytransvaginal ultrasound (sonography) (TVS) orhysteroscopy for polyps or fibroids. 13 A largeprospective study in Italy of 793 women referredfor HMB who had a uterus

Background4The current ‘gold standard’ method of measuringblood loss is the alkaline hematin technique. 16Although this method has been modified byseveral researchers (e.g. Gannon and colleagues 17 )to simplify and quicken the procedure, all versionsrequire women to collect their used sanitary wear.This is subsequently treated to extracthaemoglobin, which is then measured and relatedback to actual blood loss. This method is rarelyused outside a research setting.Another method of assessing menstrual blood lossis the pictorial blood loss assessment chart(PBAC). 18 This is a simple scoring system, whichtakes into account the number of items of sanitarywear used and the degree of staining of each item(see Appendix 1). This technique is now morewidely used than the alkaline haematin methodalthough a recent study showed that, in a group of103 women with menorrhagia, there was poorcorrelation between actual measured blood lossand PBAC score. 19 Furthermore, methods that relyon directly or indirectly estimating blood loss fromthe effect on sanitary wear do not take account ofextraneous blood loss (blood lost during changingsanitary wear).Another indirect method for estimating bloodloss is the ‘menstrual pictogram’. 20 This issimilar to the PBAC but also asks women todistinguish between the absorbency of the towelor tampon and to estimate extraneous bloodloss.Objectivity and subjectivity in HMBSubjective and objective estimates of menstrualblood loss do not correlate well. Some women withbleeding within the normal range describe theirbleeding as heavy, whereas some with objectivelymeasured HMB regard their bleeding asnormal. 19,21 A recent study validating a newtechnique of assessing blood loss investigatedwomen presenting at clinic with HMB andcontrols who considered their blood loss to be‘normal’. Only 36% of women presenting with thecomplaint of HMB had their condition objectivelyverified and 14% of the controls had blood loss inexcess of 80 ml despite considering their loss to benormal. 20Clearly, women’s expectations of normal menstrualloss are important in determining the definitionof bleeding as a ‘problem’. Such expectationsmay also have an influence on the demand forand perceived success of interventions. Forexample, over 50% of women who have surgeryfor HMB do not have objectively measured bloodloss of 80 ml or more. 7 Interpretation of blood losshas an impact on the effectiveness of treatment:one study found that women with objectivelyconfirmed menorrhagia were more likely torate the outcome following surgery as ‘successful’than those presenting for surgery without aconfirmed, objective measurement ofmenorrhagia. 17Associated menstrual symptomsThe presence of other menstrual symptoms mayhave an impact on perceptions of bleeding andaccount for some of the difference betweenobjective and subjective estimates of menorrhagia.A recent study found that women perceived theirbleeding to be heavier if they were alsoexperiencing associated pain. 22 The 39th ScientificStudy Group of the RCOG on Disorders of theMenstrual Cycle, recommended that “decisionsrelated to the treatment of menstrual cycledisorders must be based on all the relevantsymptoms”. 23 A study of 348 women presentingwith HMB in general practice found that over halfdescribed themselves as having painful periods inaddition to HMB. 3The definition of HMB, and correspondingdemand for specialist treatment, may also beaffected by the perceptions of GPs in response tothe clinical history of a woman presenting withmenstrual symptoms. In a study of 952 women inScotland, Warner and colleagues found that,among women referred to specialist gynaecologyservices, 78% were reported by their GP to haveHMB whereas only 38% of women reported thatmenstrual loss was a severe problem to the GP. 24Again, this may affect perceived treatmentoutcome if women are treated for HMB whileanother menstrual symptom was their primeconcern.Measuring the impact of HMBThe impact of any condition can be measuredusing one of three types of quality of life (QoL)scale:● Condition-specific scales These have theadvantage of incorporating attributes of QoLthat are specifically affected by the condition ofinterest. They may therefore be more sensitiveto small but important changes and may beconsidered to have greater face validity (that is,they include items that are of importance tosufferers and reflect their experience andconcerns).● Generic scales These have the advantage ofallowing comparison between conditions of

Health Technology Assessment 2004; Vol. 8: No. 3impact on QoL. However, they may be relativelyinsensitive to aspects of a particular condition.They may provide a single index or a profile ofscores across dimensions of QoL.● Preference-based scales A particular type ofgeneric measure, these elicit the respondent’spreference for a given health state and, ifappropriately scaled, provide weights that canbe used in cost–utility analyses.A recent systematic review of QoL measures usedin studies of HMB found 15 generic and twocondition-specific scales reported in 19 scaledevelopment, epidemiological and interventionstudies. 25 Quality of the scales was judged using achecklist derived from generic QoL measureappraisal tools, broadly assessing face validity andmeasurement properties. The authors, Clark andcolleagues, conclude that measurement scales inHMB perform better in relation to measurementproperties than face validity and that improvedcondition-specific measures are required to assessthe impact of HMB on QoL. 25Condition-specific scalesTwo condition-specific outcome measures havebeen developed for women with HMB: theMenorrhagia Outcomes Questionnaire 26 and theMulti-attribute Questionnaire. 27 The MenorrhagiaOutcomes Questionnaire includes items onsymptoms and satisfaction with care, physicalfunction, psychological and social well-being,global judgement of health and QoL and personalconstructs. The Multi-attribute Questionnaireincludes items on practical difficulties, socialfunction, psychological function, physical health,interruption to work and family life.Generic measuresA range of generic measures of QoL have beenused in HMB: the Short Form with 36 Items (SF-36), Nottingham Health Profile, health statusstructured history and single global item. The SF-36 was the most frequently cited in the systematicreview by Clark and colleagues, and is generally awell-validated measure used to assess healthrelatedQoL. 25 This includes items on globalhealth perception, physical function, socialfunction, role – physical and mental, pain, mentalhealth and energy/vitality. The validity of the SF-36 in assessing the QoL in women with HMB hasbeen determined in a population of womenpresenting with HMB in a study by Jenkinson andcolleagues. 28 Although the authors commentedthat it was a “feasible” means of looking at qualityof life, responding to changes over time, theysubsequently suggested that the SF-36 may havesome problems when applied to this group ofwomen. 29 In interviews with 49 women with HMBwho had completed the SF-36, Jenkinson andcolleagues found that women commented on somequestions being difficult to answer orinappropriate for women with HMB, which mayaffect the measure’s validity. 29 In addition,comparing the results given by 425 women withHMB with those from the Oxford healthy lifestylesurvey in a general population sample (n = 9219),the authors found that internal reliability, asassessed with Cronbach’s -statistic, was lower inthe HMB group, especially for general healthperception and mental health scales.Clark and colleagues 25 also report the use ofgeneric measures that address particular aspects ofQoL such as physical (Modified Townsend Score),mental (General Health Questionnaire) and sexualhealth (Revised Sabbatsberg Sexual Rating Scale)and social function (Lifestyle Index) in studies ofwomen with HMB bleeding.Preference-based measuresClark and colleagues report the use of the EQ5Din two intervention studies as a measure of QoL inHMB. The EQ5D includes a multi-attribute scale,with dimensions of mobility, self-care, usualactivities, pain/discomfort and anxiety/depression,and a global rating scale for QoL (visual analoguescale). Both studies on HMB used the visualanalogue scale for global QoL rating.Table 2 below shows the baseline ratings for QoLin women with HMB, compared to those in arange of other conditions, measured using arange of approaches to obtain a utility estimate.These values are taken from the websitehttp://www.healthpriorities.uci.edu.The value of 0.55 for menorrhagia in Table 2 maybe considered low – endometrial cancer, chest paindue to myocardial infarction and recurrence ofbreast cancer after initial surgery, for example, areall estimated to carry higher values for utility. Inthe same study, women were asked to rate theirown current health state, which had a mean of0.65 [standard error (SE) 0.04] and a median of0.75 (range 0–1.0), higher than that given for thestate of menorrhagia, which the author ascribes tomost women not menstruating at the time of theinterview. The author acknowledges that there areproblems eliciting values for chronic health statesthat may affect QoL on a daily basis but for whichthe worst effects are episodic. Even in the extremecases most HMB remains cyclical and is notusually a permanent condition. The discrepancies5© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

BackgroundTABLE 2 Examples of utility values for HMB and other health statesHealth state Utility Source How value obtainedMenorrhagia 0.55 Sculpher 30 Women with menorrhagia,time trade-offMenopause, symptoms of 0.99 Weinstein 31 Author judgementBreast cancer, reversible complication 0.99 Carter et al. 32 Standard gamble, clinicalexpertsBreast cancer chemotherapy after surgery, major toxicity 0.8 Hillner and Smith 33 Clinician judgementBreast cancer chemotherapy after surgery, minor toxicity 0.9 Hillner and Smith 33 Clinician judgementBreast cancer after surgery, first recurrence 0.7 Hillner and Smith 33 Clinician judgementBreast cancer after surgery, after first recurrence 0.85 Hillner and Smith 33 Clinician judgementBreast cancer after surgery, second recurrence 0.5 Hillner and Smith 33 Clinician judgementBreast cancer after surgery, after second recurrence 0.7 Hillner and Smith 33 Clinician judgementEndometrial cancer 0.9 Hillner et al. 34 Clinical judgement0.95 Carter et al. 32 Standard gamble – clinicalexpertsMyocardial infarction, chest pain 0.67 Tsevat et al. 35 Patient rating scaleLower third molar extraction, mild postoperative pain 0.7011 Brickley et al. 36 Patient rating scaleLower third molar extraction, moderate postoperative pain 0.4262 Brickley et al. 36 Patient rating scaleLower third molar extraction, severe postoperative pain 0.1583 Brickley et al. 36 Patient rating scaleLower third molar, no extraction, occasional low-grade pain 0.6571 Brickley et al. 36 Patient rating scaleGallstones, symptoms or chronic pain 0.95 Weinstein et al. 37 Author judgementGallstones, acute surgical complication 0.92 Bass et al. 38 Clinical expert rating scaleGallstones, endoscopic sphincterotomy 0.9 Bass et al. 38 Clinical expert rating scaleGallstones, surgical scar 0.993 Bass et al. 38 Clinical expert rating scale6may also be due to different techniques foreliciting utility values and their use in differentgroups (clinicians or sufferers). Research by Dolanand Kind 39 has also suggested that inconsistencyrates in respondents’ own ratings are higher forinterview than postal survey studies and are alsoaffected by age and educational attainment.Although utility provides a metric that can be usedto compare the value of technologies acrossdifferent conditions, the variation in valuesdemonstrated here should be borne in mind bythose interpreting such analyses.Patient satisfaction measurementPatient satisfaction is widely used as a primaryoutcome measure in studies of treatments forHMB. It is not a measure of the impact of HMB,but is discussed here alongside other outcomemeasures in HMB.“Satisfactory” means “adequate … leaving no roomfor complaint … meeting expectations or needs”. 40Satisfaction is necessarily a subjective and relativeconcept. In this context, it is the extent to which aservice meets users’ expectations. It is not clearwhether satisfaction can be measured on acontinuum, from dissatisfied through to satisfied,or whether factors resulting in satisfaction aredifferent from those leading to dissatisfaction.Satisfaction with services is related to patientcharacteristics, 40 notably age and health status.Older people are more likely to report highersatisfaction with healthcare, for reasons that arepoorly understood. The relationship betweenhealth status and satisfaction is notstraightforward. Among hospitalised patients,worse health is generally associated with lowerreported satisfaction with healthcare. One studyreviewed by Crow and colleagues, 40 showedimprovements in health resulted in highersatisfaction, although another study showed thatsatisfaction was related more to health status ondischarge than on improvement in health statusduring the hospital stay.The relationship between health status andsatisfaction is important in the current context assatisfaction is a key outcome in trials of EA. Thedebate on this point is balanced. On the onehand, satisfaction can be determined by the

Health Technology Assessment 2004; Vol. 8: No. 3experience of the care setting, which may have aminimal relationship with change in health status– such as whether staff were polite or the wardsurroundings aesthetically pleasing. Therefore,satisfaction may be regarded as a poor outcomemeasure by which to judge the effectiveness of ahealth technology. On the other hand, satisfactionis a global measure that incorporates process andoutcome aspects of the health technology andtherefore may be considered as a legitimatemeasure. The authors of this assessment regardpatient satisfaction as an important measure ofoutcome, but as a complement to appropriatemeasures of QoL.Patient satisfaction measures come in a wide rangeof formats. 40 In common with other types ofmeasure, they are prone to several importantbiases arising from design and delivery. Singleitemsatisfaction measures, such as have been usedin trials of EA, may be less valid than wellconstructedmulti-item scales. 40The range of methods for eliciting satisfactionratings is large, and details are frequently notreported. It is therefore difficult to considerwhether satisfaction in one study is similar to thatmeasured in another, rendering comparisonbetween technologies difficult on this measure.Satisfaction can be interpreted according togeneral or personal referents. In other words,people may report on their satisfaction with theirpersonal care, or whether they felt the care was, ingeneral, satisfactory. Adopting these differentperspectives produces systematically differentratings of satisfaction, with the general referentmore likely to produce a higher rating.Finally, several important response biases occur insatisfaction measurement:accounts of the development and validation of themeasures used are not available. While the use ofsimilar methods to measure subjective satisfactionfor women in both arms of an RCT may provide acomparative measure between these groups, it mayremain unclear exactly what is being measured forthe reasons outlined above. In addition, the rangeof techniques and scales used to elicit a measure ofsatisfaction across studies precludes pooling ofresults through meta-analysis. Finally, somewomen who are recorded as being satisfied withablation treatment will have had a subsequenthysterectomy, which is known to confer highsatisfaction rates in clinical trials.Current service provisionTreatment for HMB aims to improve QoL throughreducing menstrual loss. Two evidence-basedguidelines for the management of menorrhagia,one for medical management 5 (1998) and one formanagement in secondary care 13 (1999), havebeen produced by the RCOG. It is recommendedby the RCOG that women with HMB shouldreceive hysteroscopy and/or TVS to examine foruterine pathology. In addition, endometrial biopsymay be required to diagnose carcinoma orhyperplasia. Dilation and curettage (D&C) is nolonger considered the best way to assess abnormalbleeding. 13Drug therapyFor women presenting with HMB, a number ofdrug treatment options are available. These areaddressed by the RCOG guidelines. Some women,whose bleeding is relatively manageable, and forwhom investigation has shown no underlyingpathology, may benefit from counselling andreassurance that the experience is normal. Forthese women, watchful waiting is appropriate.●●●Social desirability bias – where the respondentgives what they believe to be the questioner’spreferred response, this may be a particularissue in face-to-face interview where theinterviewer is a member of the team providingcare.Cognitive consistency pressure – whereresponses are given congruent with theircontinued use of the service.Acquiescent response sets – the tendency torespond positively to all questions.The extent to which these potential biases areaddressed in the patient satisfaction measuresused in studies of EA cannot be judged as detailedAccording to RCOG guidelines, if treatment isrequired, HMB should initially be treatedmedically for at least three cycles. 5 However, one1991 study of 205 women in an English healthauthority found that only about half of patientsreferred to a gynaecologist had previously beenprescribed drug therapy by their GP. 2 The RCOGguidelines for medical management state thattranexamic acid (an anti-fibrinolytic drug) andmefenamic acid [a non-steroidal anti-inflammatorydrug (NSAID)] are considered effective treatmentsin the initial management of HMB. 5 A metaanalysisof seven studies found that tranexamicacid reduced menstrual blood loss by 47%. 21 Ameta-analysis of 10 trials found that mefenamic7© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Background8acid reduced blood loss by 29%. 41 Treatments haveside-effects such as headache, diarrhoea, nausea,vomiting, dizziness, fatigue and skin irritation.Although these are usually mild, they may affectup to 50–80% of women taking thesemedications. 5Women requiring contraception in addition totreatments for HMB may benefit from combinedoral contraceptives (COCs) or the progesterone[levonorgestrel (LNG)] releasing intrauterinedevice (IUD) [LNG intrauterine system (IUS),marketed as Mirena]. This was originallydesigned as a contraceptive device but has beenlicensed for use in HMB since 2001. Both areconsidered effective although hormone treatmentshave well-known side-effects. 5Although evidence suggests that tranexamic acid isthe most effective drug treatment for HMB, arecent UK survey of primary care prescribingshowed that 35% of treatment prescriptions forHMB were for this. 42 Women for whom one typeof medical treatment has been unsuccessful maybe reluctant to try alternative medication, eventhough this may be more effective. Prescribingpractice in primary care may therefore affectreferral and surgery rates in secondary care. Widevariations have been described in all aspects ofmanagement for HMB: general practicemanagement, referral patterns and rates ofhysterectomy. 13Surgical treatmentIf drug therapy is not effective, surgicalinterventions, including EA techniques andhysterectomy, may be considered. For womenreferred to a gynaecologist following the failure ofmedical management in primary care, surgicalintervention is likely. In an RCT of medicalmanagement versus transcervical resection of theendometrium (TCRE) in secondary care, of 94women randomised to receive medical treatment,only 10% remained in this arm after 5 years. Atotal of 77% of women had undergone subsequentsurgery, 18% having had a hysterectomy (in twocases in addition to TCRE treatment.) 43Furthermore, this study found that women whoreceived EA initially were significantly more likelyto be totally satisfied with their treatment thanthose women initially given medical treatment insecondary care (39% versus 61%; p = 0.01).Incidence of surgical operations for HMBThere were 51,858 hysterectomies in the publicsector in England in 1999–2000, includingoperations coded as secondary procedures in theOPCS Hospital Episode Statistics (HES). About80% of these are likely to have been abdominalhysterectomies. 44 About half of all hysterectomiesare likely to be for HMB. In 1998–9, there were16,219 admissions for EA. Hysterectomy andablation have a large place in private practice,although no numbers are available for operationsperformed. In addition, it is possible that changesin practice in the private sector may influencepatient behaviour in the NHS. For example, aquicker uptake of new minimal interventionablation techniques into private practice couldremove some patients wishing to avoidhysterectomy from the NHS, while some womenwishing immediate hysterectomy may prefer to payprivately rather than wait for an NHS operation.Early enthusiasts felt that EA might replacehysterectomy for HMB. In reality, diffusion hasnot been straightforward. A study of Englishhospital admission data between 1989–90 and1995–6 concluded that EA was not replacinghysterectomy. 45 However, since then numbers ofEAs have increased whilst hysterectomies havefallen. The rise in the numbers of EA proceduresfor 1997 coincides with the introduction ofsecond-generation devices into clinical practice(Amso NN, University of Wales, Cardiff: personalcommunication, 2002).Figure 1 plots HES codes Q08 and Q09 combined forhysterectomy and Q16 and Q17 combined for EA.Although there appears to be a trend towardincreased ablation and decreased hysterectomy,these figures may mask more complex localvariations. A recent study 46 in the USA examinedthe diffusion of EA using State Inpatient andAmbulatory Surgery Databases of the HealthcareCost and Utilization Project in six states for1990–7. Whereas the rate of EA increased in allstates, the rate of hysterectomy decreased in three,remained static in two and increased in one. Theratio of hysterectomy rate to EA rate decreased inall states. The combined rate of EA andhysterectomy increased in all but one state. Theauthors suggest that EA is being used as anadjunct to rather than a replacement therapy forHMB. It is possible that the availability of EA maydecrease the threshold for surgical treatment.HysterectomyHysterectomy is the only treatment for HMB thatcan guarantee complete removal of symptoms(amenorrhoea) in all women. In the UK, 20% ofwomen will have a hysterectomy by the age of55 years. 47 In about half of all hysterectomies,

Health Technology Assessment 2004; Vol. 8: No. 3800007000060000Number5000040000HysterectomyEndometrial ablation300002000010000090–91 91–92 92–93 93–94 94–95 95–96Year96–97 97–98 98–99 99–00FIGURE 1 Number of hysterectomies and EA operations in EnglandHMB is the presenting complaint and in half ofhysterectomies performed for HMB, a normaluterus is removed. 48Different approaches to hysterectomy are possible.In abdominal hysterectomy the uterus isapproached through the anterior abdominal wall,via a vertical or horizontal incision. In vaginalhysterectomy, the uterus is removed through thevagina and may be carried out with the assistanceof a laparoscope. Different degrees ofhysterectomy are also possible: removing thecomplete uterus (total hysterectomy), leaving thecervix (sub-total hysterectomy) and removing theovaries and Fallopian tubes in addition to theuterus (total hysterectomy with bilateral salpingooophorectomy).The VALUE study of over 37,000hysterectomies performed in the UK in 1994–5found that two-thirds were abdominal (of which4% were sub-total) and that ovaries were removedin 57% of hysterectomies. 48Hysterectomy is an inpatient procedure and fullrecovery may take 4–6 weeks. One in 30 womensuffer perioperative adverse events (Table 3).Postoperative complications affect at least one in10 women and include incontinence and otherurinary problems, fatigue, infection, pelvic pain,hot flushes, dry vagina and sexual problems. Inaddition, women undergoing bilateral salpingooophorectomyat the time of hysterectomy willexperience the menopause. 48A systematic review of studies examining the effectof hysterectomy on sexuality found little evidencethat hysterectomy had a detrimental affect. Inmost women, sexuality was unchanged orenhanced following the operation. However, thequality of the trials included in the review wasconsidered generally poor. 49 There is evidencethat in the long term, women who haveundergone hysterectomy may suffer increased riskof some symptoms such as urinary incontinence, 509© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

BackgroundTABLE 3 Adverse events following hysterectomyVery common (>1/10) Common (>1/100, 1/1000,

Health Technology Assessment 2004; Vol. 8: No. 3Lumen of uterusEpitheliumCapillariesUterine glandsVenous sinusoidsLamina propria ofconnective tissueSpiral (coiled)arteryStraight arteryMyometriumEndometriumResectoscopeTreatedendometriumEndometrial veinSmooth musclefibresRadial arteryFIGURE 4 Transcervical resection. Fromwww.gynalternatives.com/ablation.htm, by permission of P IndmanArcuate arteryUterine arteryFIGURE 3 Section through the endometrium. Modified fromHuman anatomy, Marieb EN, Mallatt J, © 2001 by PearsonEducation Inc. Reprinted by permission of Pearson Education, Inc.lining. TCRE provides good samples ofendometrium for biopsy. TCRE may also be usedfor the removal of fibroids, usually those notlarger than 2 cm. The operation takes13–45 minutes 52 and may be done as a day-caseprocedure.The RB technique also requires visualisation andirrigation using a resectoscope. A RB electrode isused rather than a cutting loop. A current ispassed through the ball and this is moved acrossthe surface of the endometrium, therebydestroying the tissue. 56 Because the RB fits betterin the thin-walled uterine horns and lessens thechance of perforation, some surgeons use acombination of cutting loop and RB equipment inthe same ablation procedure. As no ‘chips’ ofremoved endometrium are generated with RBcoagulation, there may be better visibility throughthe hysteroscope than with TCRE. RB also resultsin fewer operative adverse effects. 54 In theUK, it is usual for TCRE to be supplemented byRB at the fundus and in the thin parts of theuterus around the openings of the Fallopiantubes. 10Possible perioperative adverse effects with TCREand RB include electrosurgical burns, uterineperforation, haemorrhage, gas embolism,infection and fluid overload (which may causecongestive cardiac failure, hypertension,haemolysis, coma and death). Strategies foravoiding fluid absorption include maintaining theminimum intrauterine pressure for safe surgery,having an efficient system to retrieve circulatedfluid and maintaining an account of fluidvolumes. 57 Fluid overload may be of particularconcern when fibroids are being removed, asopen blood vessels are capable of rapid fluidabsorption.The MISTLETOE study examined complicationswith first-generation EA techniques. Possibleadverse effects, both operative and postoperative,are shown in Table 4.The Endometrial Ablation Group (a specialinterest group) consensus paper (2002) 58concluded that EA is contraindicated whenthere is:1. uterine malignancy or its precursors2. acute pelvic infection3. desire for future pregnancy4. excessive cavity length (>12 cm).In addition, the Group recommends that womenundergoing EA are counselled that:1. Amenorrhoea cannot be guaranteed, and itsoccurrence depends on technique, operator11© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

BackgroundTABLE 4 Adverse effects with first-generation EA techniquesVery common (>1/10) Common (>1/100, 1/1000,

Health Technology Assessment 2004; Vol. 8: No. 3Description of new interventionSecond-generation EA techniquesSince the 1990s, several new methods of EA havebeen developed. These are often referred to assecond-generation techniques. They do notrequire direct visualisation of the uterine cavityand employ a variety of means to destroy theendometrium – circulation of heated saline withinthe uterine cavity, use of a diode laser (ELITT),punctual vaporising methods, photodynamicmethods, radiofrequency, microwaves, a ballooncatheter filled with heated fluid and cryotherapy.Apart from the direct circulation of hot liquidwithin the uterus, none of the second-generationmethods require direct visualisation of the uterus.The treatments are much less dependent on theskill of the surgeon than first-generationtechniques, and much more dependent on thereliability of the machines used to ensure safetyand efficacy. For this review we have been asked toconsider thermal balloon and microwaveendometrial techniques, both of which areperformed without direct visualisation of theuterine cavity and require no distension fluid.MEAThe MEA technique was developed in Bath,England, in 1993. The microwave frequency(9.2 GHz) was chosen to ensure that tissuepenetration was no more than 6 mm. An 8-mmapplicator inserted through the cervix delivers themicrowaves using a dielectrically loadedwaveguide. 64 Power is controlled by the surgeonusing a footswitch and the temperature inside theuterus is monitored by thermocouples on thesurface of the waveguide. Prior to microwaveablation treatment, oral and vaginal thinningagents may be given. Immediately prior to MEA,hysteroscopy is performed to exclude falsepassages, wall damage and perforation.The uterus is measured and the measurement ischecked with a metal rule. Under general or localanaesthetic, the cervix is dilated to Hegar 8 or 9and the length of the uterine cavity is measured.The microwave probe is inserted until the tipreaches the fundus. Graduated centimetremarkings on the applicator shaft confirm thelength and if these three measurements of uterinelength are the same, the device is activated. 65When, after a few seconds, the temperaturereaches 80°C, the probe is moved laterally so thatthe tip is placed in one of the uterine cornu. Thetemperature briefly falls and rises again and when80°C is reached again the probe is moved to theother cornual region and the procedure repeated.5 mm catheterCervixBalloonEndometrial liningFIGURE 5 Thermal balloon ablation. Modified fromGynaecology, 3rd ed, Shaw RW, Soutter P, Stanton SL, © 2002,with permission from Elsevier.Maintaining a temperature of 70–90°C, the probeis withdrawn with side-to-side movements. Thetemperature measured by the thermocouple isactually the heat transmitted back from the tissuethrough the plastic sheath to the applicator shaft.Tissue temperature is higher than these measuredlevels during active treatment. As a marker on theprobe appears at the external os, the applicator isswitched off to avoid treating the endocervix. Theprocedure takes 2–3 minutes. 64 Following theprocedure, analgesia is provided as required. Awatery discharge for about 3 weeks is usual. 65MEA is contraindicated where there has beenprevious uterine surgery and where previousclassical Caesarean section has left a uterine scarthinner than 8 mm.TBEAThe TBEA relies on transfer of heat from heatedliquid within a balloon that is inserted into theuterine cavity (see Figure 5). Several devices areavailable including Thermachoice andCavaterm. All systems involve an electroniccontroller, a single-use latex or silicone ballooncatheter (5 mm) that houses a heating elementand two thermocouples, and an umbilical cable.The thermal balloon cannot be used on womenwith large or irregular uterine cavities as theballoon must be in direct contact with the uterinewall to cause ablation. Cavaterm is contraindicated13© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

BackgroundTABLE 5 Contraindications for the three second-generation methods of EAContraindication Microwave Cavaterm ThermachoiceUterine cavity size (cm) >14 >10 >12Pervious surgery or trauma leading to uterine wall thickness of atleast 8 mm ✓ – –Previous classical Caesarean section as scar would be positioned inthe operative field ✓ – –Previous ablation/resection as this thins the uterine wall ✓ – –Fibroids distorting the uterine cavity ✓ – –Repeat ablations should never be performed in conjunction withmechanical preparation ✓ – –D&C should not be performed as preparation ✓ – –Women who are pregnant or who wish to become so should notundergo EA ✓ ✓ ✓Active pelvic inflammatory infection ✓ ✓ –Undiagnosed vaginal bleeding ✓ ✓ –Known or suspected endometrial carcinoma ✓ ✓ ✓Gross abnormalities such as myomas that prevent the balloon lyinguniformly on the endometrium – ✓ ✓Separate uterus (septum dividing the uterus in two) or otherabnormalities/lesions that would result in inadequate balloon contact – ✓ ✓Uterine wall weakness – ✓ –Cervical canal 10 cm from theinternal os to the fundus, and Thermachoice whenthe cavity is >12 cm in length.With the Thermachoice device, the cervix isdilated to about 5 mm. After insertion into theuterine cavity, the balloon is filled with sterile fluid(5% dextrose in water) and expands to fit thecavity. Intrauterine pressure is stabilised to160–180 mmHg. The fluid is then heated at 87°Cfor 8 minutes. Newer versions of the balloon use aconvection circulation approach to distribute heatmore evenly and a silicone balloon. Pressure,temperature and time are continuously monitoredand controlled by computer. Automatic shut-off isevoked if parameters are exceeded. Passive heattransfer causes cauterisation of the endometrium.NSAIDs are given postoperatively. The treatedlining sloughs off over the following week to 10days.heated at a target temperature of 78°C for10 minutes, during which time the fluid iscirculated vigorously.Endometrial thinning agents are notrecommended. The endometrium may be prethinnedby curettage immediately prior to theprocedure. NSAIDs are given to reduceperioperative cramping.Prognostic factors for the failure of TBEA, basedon a study of 130 women who underwent TBEAwith Thermachoice in The Netherlands, areyounger age, retroverted uterus, pretreatmentendometrial thickness of at least 4 mm andduration of menstruation. 66Adverse effects with secondgenerationEA devicesAdverse effects include:14The process is similar for the Cavaterm device,with some differences in detail. The cervix isdilated to about 6 mm. After insertion, a siliconeballoon is filled with sterile 5% glucose solution toa pressure of 230–240 mmHg. The liquid is●●●●●uterine infectionperforationvisceral burnbleedinghaematometra

Health Technology Assessment 2004; Vol. 8: No. 3● laceration● intra-abdominal injurySummary● cyclical pain.Chapter 2: BackgroundThe differences between the second-generation● HMB is a common complaint among womentechniques considered in this assessment reportaged 30–49 years.are summarised in Table 5, which shows the● Blood loss measurement may be direct ormanufacturers’ descriptions of contraindicationsindirect, objective or subjective. Objectivefor the Microsulis microwave device and the twoand subjective measures do not correlate welltypes of thermal balloon, Cavaterm andyet the clinical definition of HMB (>80 mlThermachoice.blood loss) is not often used outside aresearch setting. Perceptions of HMB may beUse of local anaestheticfurther influenced by other associatedUse of local anaesthetic (LA) is a stated advantagemenstrual symptoms.of second-generation EA techniques, although this● The impact of menorrhagia is largely onwill not be suitable for all women. Ninety-eightQoL, although anaemia may also occur.women in the UK undergoing microwave ablationMeasuring the impact of HMB has beentook part in a partially randomised trial of generalanaesthetic (GA) and LA. 67 attempted using a range of generic andSixty-two women (63%)disease-specific measures. In addition,expressed a preference and were about equallysatisfaction with treatment has been regardeddivided between preferring GA and preferring LA.as an important outcome, although there areThe remainder were randomised. The proceduredifficulties in interpreting its meaning.was considered acceptable under GA in both● A number of medical and surgical treatmentpreferred (100%) and randomised (97%) groups.options are currently available. SurgicalHowever, under LA, 97% of those who chose thistreatments include hysterectomy, which offersmethod and 85% of those allocated to LA founda permanent solution, but is major surgerythe procedure acceptable. The trial authorsand has associated morbidity and mortality,suggest that LA should therefore be an option,and more minimally invasive hysteroscopicrather than standard procedure. In addition, fivesurgical techniques such as resection and RB(16%) of the 32 women choosing LA actuallyablation, which rely on considerable surgeonrequired GA owing to dilation difficulties (n = 3),skill and also have associated morbidity andequipment failure (n = 1) and in one case due toreported mortalities. This report assesses twoidentifying a submucosal fibroid that required GAnewer ablation techniques that destroy thefor removal. The trial found that the operationendometrial lining through microwave ortime was not reduced in the randomised arms, butthermal energy.was in the preference groups (19 versus 25minutes). 67If LA is chosen, it has been suggested that danazolmay be a preferable pre-operative endometrialthinning agent, as goserelin may increase cervicalresistance. 6715© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Chapter 3MethodsMethods for reviewing the effectiveness ofMEA and TBEA were specified a priori andare outlined in the research protocol (Appendix 2).Research questions●●What is the effectiveness of MEA and TBEA inthe treatment of HMB?What is the cost-effectiveness of MEA andTBEA in the treatment of HMB?Review team and advisory groupThe review was carried out by a review teamcomprising Dr Ken Stein, Ruth Garside,Dr Katrina Wyatt, Dr Ali Round and Alison Price.In addition, an external advisory group of clinicalexperts provided advice during the assessmentand comments on an early draft. Details of thisgroup appear in the Acknowledgements(p. 83).General methodsThe methods of the review generally adhered toguidance laid out in the York Centre for Reviewsand Dissemination guidelines.Interventions considered were thermal balloonand microwave methods of EA for HMB.First-generation methods of EA, TCRE, RB andcombined methods are considered as comparators.In order to provide a more complete picture ofsurgical management of HMB, information abouthysterectomy compared with first-generationmethods was examined using an existingsystematic review of these treatments, updatedwith further literature searches.Assessment of microwave andthermal balloon ablationSearch strategyElectronic databases were searched for publishedstudies, recently completed and ongoing research.Appendix 3 shows the databases searched and thestrategy in full. Bibliographies of articles were alsosearched for further relevant papers. Experts inthe field and relevant industry bodies were alsoasked to provide information.Inclusion and exclusion criteriaSystematic reviews, RCTs and controlled trials ofMEA and TBEA versus TCRE, RB or TCRE andRB combined were included.Systematic reviews and RCTs of first-generationEA techniques versus hysterectomy published after1999 were included.Studies were excluded if they were:●●●●●●animal modelspreclinical and biological studiesnarrative reviews, editorials, opinionsnon-controlled studiesnon-English language papersreports published as meeting abstracts only.Identification of studies was made in two stages:abstracts were examined independently forinclusion by two researchers (RG and KS).Disagreements were resolved by discussion. Theninclusion and exclusion of full-text articles wasmade independently by two researchers (RG, KW)and disagreements were resolved by discussionwith a third (KS).Data extraction strategyData were extracted by one researcher (RG) andchecked by another (KW). Actual numbers wereextracted where possible and, when necessary,analyses were repeated on an intention-to-treat(ITT) basis from original data.Quality assessment strategyRelevant systematic reviews were assessed usingthe QUOROM checklist, 68 which uses thefollowing criteria:1. The clinical question is made explicit.2. The database and other information sourcesin detail and any restrictions.3. Inclusion and exclusion criteria arespecified.17© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Methods184. The selection criteria, methods for validityassessment, data abstraction, studycharacteristics and quantitative data synthesisin sufficient detail to permit replication.5. Characteristics of the included and excludedRCTs, details of study design, interventionsand outcomes are reported. How clinicalheterogeneity was assessed is reported.6. Principal measures of effects, method ofcombining results, handling of missing data,how statistical heterogeneity is assessed.Rationale for (and a priori) subgroup analysis,and any assessment of publication bias areprovided.7. A profile summarising trial flow through thesystematic review is shown.8. Descriptive data for each included trial aregiven.9. Agreement on the selection and validityassessment is reported.10. Simple summary statistics and data needed tocalculate effect sizes and confidence intervalsin ITT analyses are given.Assessments of the quality of RCTs were performedusing quality indicators as given below. Owing tothe nature of the intervention, the presence ofblinding to treatment received was not consideredan appropriate measure of quality, althoughconcealment of allocation and blind assessment ofoutcomes remain valid as quality markers.Internal validityTrial characteristics:1. appropriate method of randomisation2. blind assessment of outcomes3. number of women randomised, excluded andlost to follow-up (LTFU)4. whether an ITT analysis is performed5. whether a power calculation is done6. timing, duration and location of study.External validityStudy participants:1. age and any other recorded characteristics ofwomen in studies2. inclusion criteria3. exclusion criteria4. length of follow-up.Generalisability was categorised as high (detaileddescription of the exclusion criteria and patientgroup), medium (description of exclusion criteriaand patient group), or low (no description ofexclusion criteria or patient group).Interventions used:1. type of EA technique and route of hysterectomysurgery2. endometrial thinning agents used.Methods of analysisThere was considerable clinical and methodologicalheterogeneity among the studies included in thereview. Quantitative synthesis through meta-analysiswas therefore not undertaken. Study results aretabulated and, for outcomes where there aremultiple data points at the same follow-up pointand with similar methods of outcome measurement,these are illustrated using forest plots.Economic evaluationCost-effectiveness modelA state transition (Markov) model was developedby the authors using Microsoft Excel. Thestructure was informed by clinical input. Themodel examines the progress of five hypotheticalcohorts of women with HMB who are treatedseparately by either thermal balloon, microwave,TCRE or RB EA, or hysterectomy. The modeltakes the perspective of the NHS and calculatesincremental cost–utility between options.Main assumptionsStructure of the economic modelThe clinical pathway modelled is shown in thedecision tree in Figure 6.The structure of the model is shown in more detailin Figure 7 (pathway for patients undergoing anytype of EA) and Figure 8 (pathway for patientsundergoing hysterectomy). Health states areshown in boxes and arrows show the transitionsthat can occur. For example, from hysterectomy,patients can either move to a state ofconvalescence (recovery from the operation in theabsence of complications), have complications ordie through direct or other causes.The health states and pathways are the same forall types of EA. The health states in the EA modelare as follows:●●●Menorrhagia – all patients in the cohort havepreoperative HMB.EA – the women undergo EA by MEA, TBEA orresection.Complication – following EA, some women willexperience complications in the perioperativeor immediately postoperative period.

Health Technology Assessment 2004; Vol. 8: No. 3MenorrhagiaEndometrialablationRepeat ablationHysterectomyHysterectomyHysterectomyWellWellFIGURE 6 Clinical pathway modelledMenorrhagiaEndometrialablationComplicationWellRecurrentmenorrhagia2nd endometrialablationHysterectomyto hysterectomy pathDeadFIGURE 7 Influence diagram for EA path●●●●Well – following EA or complication, women arewell.Recurrent menorrhagia – following EA, HMBmay reoccur (treatment failure) at any time,including immediately postoperatively. Womenmay stay in this state, or be retreated, or have ahysterectomy.Repeat EA – if HMB recurs postoperatively,women may choose to have a second ablation.Only one repeat EA is permitted. Repeatablations are by the same technique as theinitial ablation.Hysterectomy – if HMB recurs after the firstablation, women may choose to have●hysterectomy. All those failing a second ablationwill be treated by hysterectomy. These womenthen follow the pathway outlined in thehysterectomy diagram in Figure 8.Death – it is possible to die from causes otherthan EA during any health state. Athysterectomy and EA, women may also die as adirect result of the surgical procedure.The health states in the hysterectomy model(shown in Figure 8) are as follows:●Menorrhagia – all women in the cohort havepreoperative HMB.19© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

MethodsMenorrhagia Hysterectomy ComplicationWellConvalescenceDeadFIGURE 8 Influence diagram for hysterectomy path●●●●●Hysterectomy – all women undergohysterectomy.Complication – following hysterectomy, somewomen will experience complications in theperioperative or immediately postoperativeperiod. The effects of these may last for amedian of 1–2 months.Convalescence – following hysterectomy bothwith and without complications, a period ofconvalescence is experienced.Well – following convalescence, women are well.Death – it is possible to die from causes otherthan hysterectomy from any health state. Athysterectomy, women may also die as a directresult of the surgical procedure.Only perioperative and complications immediatelyfollowing the procedure are modelled; subjectscannot enter the health state ‘complications’ fromany state except that of the operation.After an unsuccessful EA treatment, HMB canreturn at any time (treatment failure), includingimmediately after the procedure. Recurrentmenorrhagia has been assumed to be mostly evidentin the first 3 years. This was based on evidence inthis assessment (see Tables 9 and 20). It was assumedthat the total number of women with recurrentmenorrhagia counted at each point of follow-upwould include both those reporting HMB and thosewho had undergone a previous repeat procedure.20A cohort of 1000 women eligible for eachprocedure are modelled for each operation. Thestarting age of women in the model is 42, basedon the median age of women in the trials of EAincluded in this review (see Table 6). The modelruns for a total of 10 years. The model assumesthat all women become menopausal after 10 years,at age 52 years, which is the average age ofmenopause in the UK.Each cycle is 1 month long. In reality,complications following a second-generationablation may be experienced for less than 1 month.The death rate from causes other than procedureis based on values for women in the Life Tables ofEngland and Wales for the years 1992–2000starting at age 42 years and correspondinglyincreasing each year. 69Clinical processesHysterectomy is assumed to be abdominalhysterectomy in the economic model as two-thirdsof UK hysterectomies are by this route. 48If EA of any type fails, repeat ablation orhysterectomy is offered. The model assumes that90% of those with recurrent menorrhagia will havea repeat procedure, with 60% having repeat EAand 30% having a hysterectomy. This furtherprocedure takes place within 6 months ofmenorrhagia returning. Only one repeat ablationis offered; if the treatment fails a second time,only hysterectomy is available. About 90% ofwomen with recurrence following repeat EA have ahysterectomy within 6 months.There is no convalescence state after ablation as allwomen are assumed to have fully recovered within1 month and this is the cycle length.Convalescence following ablation is thereforecaptured in the utility value for the EA health state.Parameters includedThe following parameters were included in themodel:●The proportion of women who have recurrentmenorrhagia following EA.

Health Technology Assessment 2004; Vol. 8: No. 3●●●●●Death rates directly associated with each type ofoperation.Complication rates associated with eachprocedure, and with repeat procedures.The proportion of women with recurrentmenorrhagia who receive repeat ablation orhysterectomy.Utility values associated with each health stateshown in Figures 7 and 8.Costs of each procedure (including cost ofequipment, preoperative endometrial thinning,time in theatre, proportion of womenundergoing ablation who have GA and LA, timespent in hospital post-procedure).Sources of estimatesThe initial search for this assessment was broad inscope. In populating the model, a hierarchy ofevidence was used. First, data from good-qualitysystematic reviews of RCTs were sought (includingdata obtained as part this report’s effectivenessassessment). If these were not available, then datafrom good-quality individual RCTs were sought.Where these were not available, large prospective,observational studies conducted in the UK wereused. Finally, if no published evidence could befound, the opinion of clinical experts was sought.The exceptions to this hierarchy were data forperioperative complications and death. Theinfrequency of these events means that the smallRCTs provide imprecise estimates. Large nationalaudits of hysterectomy and first-generation EAexist – the VALUE and the MISTLETOE studies(see the section ‘Adverse effect data from othersources’, p. 57). These were therefore used as theyare likely to provide more accurate informationabout rare events. For complications followingrepeated ablation, data were taken from aprospective cohort study of 800 primary and 75repeat ablations. 70 For second-generationtechniques, large cohort studies investigatingcomplication rates were used. 71,72treatment have been assumed to have the sameutility value.The utility value of convalescence afterhysterectomy is assumed to be one-third less thanthe state of ‘well’ following recovery followinghysterectomy.Resource use and costsAspects of care in the modelIn order to calculate the costs of each of theprocedures, a range of health service costs wereobtained. A cost per procedure for each type of EAtechnique and for hysterectomy was calculatedbased on the details described below. Data forcosts were taken from the literature and fromSouthampton University Hospital costings unit.The cost of procedure includes costs ofendometrial thinning agents, anaesthetics,dedicated equipment, operating time andinpatient stay.Preoperative treatmentIt is assumed that once referred to secondary care,all women with HMB will have the causeinvestigated. The RCOG recommends that womenreceive a TVS initially, in order to identify thosewho have an abnormal uterine cavity. 13 Thisshould be followed up by hysteroscopy as required.Hysteroscopic examination may be carried outunder either LA or GA. The majority of womenhave the latter. A biopsy is also undertaken toexclude endometrial carcinoma or hyperplasia andshould be undertaken even where hysteroscopy orultrasound suggests a normal uterus. 13 This mayalso be done as an outpatient, blind procedure, forexample using the Pipelle sampler.The economic model assumes that all women withHMB receive these investigations as routine careprior to being offered any treatment. These costshave not, therefore, been included in the model asthey are not relevant to the marginal analysis.Utility values for different health states fallbetween one (perfect health) and zero (dead). Inthis model, the state of being well is less than oneas it encompasses general health values for womenof this age. Health state utility values were takenfrom the literature and are shown in Table 25. Onepublished cost–utility analysis of surgery formenorrhagia 30 describes utility values that wereobtained from 60 women with menorrhagia usinga set of scenarios describing health states relatingto menorrhagia and its treatment, using the timetrade-off (TTO) technique. Menorrhagia andrecurrent menorrhagia following a failedAll patients undergoing first-generation ablationsand MEA are assumed to receive 4–5 weeks of pretreatmentwith thinning agents: oral danazol(200 mg daily) if undergoing LA treatment or theluteinising hormone-releasing hormone (LHRH)analogue Zoladex if undergoing GA.Surgical proceduresDetails for average length of stay in hospital andwaiting time for hysterectomy are taken from HES2000–1 (Code Q07 – abdominal hysterectomy)for the UK. These data were used because theygive average national figures and the surgery21© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Methodscoding for hysterectomy contains only abdominalhysterectomies. Duration of surgery forhysterectomy is the mean time of surgery inminutes taken from a systematic review carried outin 1999. 52Details of resource use for first-generation EA weretaken from a systematic review rather than routineNHS statistics which give costs at HealthcareResource Group level. 9 The HES code for firstgenerationEA may also include a number of otherprocedures (at Southampton Hospital theseinclude a variety of procedures such aspolypectomy, diagnostic examination of the uterusand occlusion of Fallopian tubes) which maydistort the actual costs of EA. Instead, the meansfrom the systematic review were used. 52 HES datafor 2000–1 were used to obtain waiting times forsurgery. 73It is assumed that all hysterectomies areundertaken with GA. Data on the proportion offirst-generation EA procedures using LA weretaken from a systematic review 9 and those figuresfor second-generation techniques were taken froma patient preference RCT of GA and LA for MEA.In this study of 98 women in Scotland, 63% had apreference about which type of anaesthetic theypreferred, of whom 52% chose LA. 67 This hasbeen assumed to be the proportion of women whowould choose LA in the clinical setting.Equipment costThere are two main types of TBEA equipmentused in the UK, Cavaterm and Thermachoice, andone type of microwave equipment, made byMicrosulis Medical Ltd. Equipment costs werebased on details provided by the manufacturers ofthese devices. The cost of thermal balloon is themean cost of the two devices.Staff costsIt is assumed that all hysterectomy and all first EAtechniques are undertaken by a consultant. Staffneeded in the operating theatre for a GAprocedure are assumed to include a junioranaesthetist, a trolley nurse, instrument nurse andcirculating nurse. Given the relative simplicity ofsecond-generation ablation techniques, the costswere also calculated assuming that a more juniorsurgeon (registrar) undertook the operation.DiscountingCosts were discounted at 6% and benefits at 1.5%.AnalysesAn incremental analysis of costs and benefits wasperformed for each of the following comparisons:●●●●●●●●●MEA versus TBEAMEA versus TCREMEA versus TCRE and RBMEA versus RBMEA versus hysterectomyTBEA versus TCRETBEA versus TCRE and RBTBEA versus RBTBEA versus hysterectomy.Dealing with uncertaintyTo examine uncertainty within the model, one-waysensitivity analyses were undertaken to establishwhich estimates have the greatest effect on themarginal cost–utility for TBEA and MEA. Thesensitivity analysis focused on:●●●●●●●●complication ratesdeath rates due to the procedurepercentage of women with recurrentmenorrhagiapercentage of women with recurrentmenorrhagia who have repeat procedure andhave hysterectomypercentage of women failing the ablation afterrepeat procedureutility values for EA state, well and menorrhagiaaspects of procedure costs including proportionof procedures done under anaesthetic andlength of hospital stayduration of the model.Industry submissionsThree submissions from industry were provided tothe National Institute for Clinical Excellence(NICE) by manufacturers of thermal balloon andmicrowave ablation equipment. The submissionswere used in a number of ways. First, they wereexamined for additional information that met theinclusion criteria for the systematic review ofeffectiveness or the economic model. Second, theeconomic evaluations they provided wereappraised using the frameworks proposed bySculpher and colleagues 74 for decision analyticmodels and Drummond and colleagues 75 forgeneral cost-effectiveness analyses.Finally, a brief comparison of the modelconstructed by the review team and those suppliedby industry was undertaken.22

Health Technology Assessment 2004; Vol. 8: No. 3Chapter 4ResultsSystematic review – effectivenessThis section describes the studies identifiedthrough the search strategy and those included inthis assessment. The quality and main findings ofsystematic reviews and controlled trials are thendescribed.Studies identifiedThe search for controlled studies including MEAor TBEA identified 216 abstracts. A total of 68full-text articles were acquired (see Appendix 4 forfurther details of excluded papers). Fifteen ofthese were possible controlled studies. A total of13 trial reports relating to 10 studies wereidentified as suitable for inclusion.The search to update the Cochrane review of firstgenerationtechniques and hysterectomy identified80 additional abstracts, of which 13 full-textarticles were obtained, none of which wereultimately included (see Appendix 4 for details ofinclusion and exclusion)Included systematic reviewsEight Cochrane reviews have examined treatmentsfor HMB. Five review the evidence for variousmedical methods of controlling HMB: oralcontraceptives, 76 cyclical progestogens, 77danazol, 78 NSAIDs 79 and antifibrinolytics. 80 Onereviews the evidence for the progesteronereleasingIUD. 81 One examines the use ofpreoperative thinning agents before hysteroscopicsurgery. 55Two reviews were included in the currentevaluation, on endometrial destruction techniquesfor HMB 9 and TCRE and RB versushysterectomy for HMB. 52Quality of included systematic reviewsSee Appendix 5 for a summary of the QUOROMchecklist used to assess quality. 68 Both reviews useda structured format. The clinical problems, andrationale for the interventions examined wereoutlined in the background sections and reviewobjectives were described. Sources of data andadditional sources of data were described, anddetails of study selection criteria (population,intervention, and study design) given. Norestrictions on publication status, language or yearof publication were listed.In both reviews, methodological quality ofincluded RCTs was assessed in relation toadequate concealment prior to randomisation, thepresence of power calculation for sample size, ITTanalysis and attrition rates.In both reviews, data were extractedindependently by two reviewers. Heterogeneitywas examined by inspecting the scatter in datapoints on graphs and the overlap of theconfidence intervals (CIs) and by checking theresults of statistical tests for heterogeneity.Dichotomous data were pooled as Peto (fixedeffect) odds ratios (ORs) with 95% CI, apart fromone outcome (use of LA) in the review ofendometrial destruction techniques, 9 which used arandom effects model. Continuous data werepooled using weighted mean difference with 95%CI. For a number of outcomes comparing pooledfirst-generation and all second-generation EAtechniques in one review 9 the data presented inthe graphs and those reported in the text weredifferent. For one outcome (postoperativeamenorrhoea) the text data suggested thatthe difference between the techniques wassignificant whereas the data presented graphicallydid not.Sensitivity analyses were planned a priori andperformed in the review of EA versushysterectomy. 52 It is stated that this did not changethe direction of results although point estimatesare not given. Sensitivity analyses were notplanned a priori in the other review. 9Diagrammatic descriptions of the flow of trialsthrough the inclusion and exclusion processeswere not included in either review. Details of thestudy characteristics were tabulated in both reviewsalthough no references to individual studies weregiven in the tables in one. 9 The level of agreementon selection and validity assessment was notreported in either review. Neither review discussedpotential biases in the review.23© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Results24Existing systematic reviews –findingsDetails of the data extracted from the existingsystematic reviews are given in Appendix 6.Systematic review ofhysterectomy versus firstgenerationEA techniquesFive RCTs were included in the review, includinga total of 752 participants. Follow-up was between1 and 4 years (median 2 years).The Cochrane review of hysterectomy versus firstgenerationEA techniques 52 found that there was asignificant advantage in improved HMB andsatisfaction rates up to 2 years, but not beyond(OR = 0.31, 95% CI = 0.16 to 0.59) for womenundergoing hysterectomy. However, duration ofsurgery, hospital stay and time to return to workwere all shorter following EA [weighted meandifference (WMD) = 23.1 minutes, 95% CI 23.8 to22.3; WMD 4.0 days, 95% CI 4.9 to 4.8; WMD4.6 weeks, 95% CI 4.8 to 4.4 respectively]. Mostadverse effects, both major and minor, were morelikely with hysterectomy – sepsis, bloodtransfusion, urinary retention, anaemia, pyrexia,haematoma and hypergranulation tissue. Onlyfluid overload was more likely with first-generationEA. Other adverse effects showed no differencebetween the groups.The reviewers concluded that first-generation EAtechniques offer an alternative to hysterectomy forHMB and that effectiveness and satisfaction ratesfor both procedures were high. The higher rate ofcomplications and longer recovery period forhysterectomy were offset by permanent relief fromsymptoms. Costs were lower for EA but, owing tore-treatment in the EA group, the differencenarrows over time with EA costing between 5 and11% less than hysterectomy at 4 years.Systematic review of endometrialdestruction techniquesThe Cochrane review of endometrial destructiontechniques 9 identified two RCTs 82,83 of TBEAversus RB. Three papers were published on one ofthese studies at 12, 82 24, 84 and 36 months 85 followup.One paper, relating to a study comparing MEAwith combined TCRE and RB, was also included. 86In addition, six further RCTs were included. Threetrials compared first-generation methods, and twocompared other second-generation techniques[vesta system, heated saline hydrotherm ablator(HTA)] with first-generation techniques.The studies contained a total of 1595 participantsand follow-up was between 6 and 15 months(median 12 months).For TBEA, some anomalies were found;amenorrhoea was more likely in the RB group at12 and 36 months (OR = 0.55, 95% CI 0.31 to0.99 and OR = 0.5, 95% CI 0.25 to 0.97,respectively) but not at 24 months. Likewise, whileadditional surgery was significantly more likely inthe RB group at 24 months (OR 0.35, 95% CI0.12 to 0.99), this was not seen at 12 or36 months. Other outcomes were not found to besignificantly different.For MEA, most outcomes were not significantlydifferent from the TCRE group. Odds ofhaemorrhage were lower in the MEA group(OR = 0.14, 95% CI 0.02 to 0.8), whereasequipment failure was more likely (OR = 4.07,95% CI 1.1 to 15).The review concluded that, overall, secondgenerationtechniques had similar success ratesand were significantly quicker to perform (WMD= 11 minutes, 95% CI –18.6 to –2.6) than firstgenerationtechniques and were significantly morelikely to be performed under LA (OR = 7.6, 95%CI 1.1 to 52.7) However, equipment failure wasmore likely in second-generation techniques(OR = 4.1, 95% CI 1.1 to 15.0).However, as noted above, there are differences inthe text and graph figures for some of thefindings. Attempts to contact the authors to clarifythese data were unsuccessful.The study concluded that second-generationtechniques compare favourably with firstgenerationtechniques but that equipmentproblems needed to be resolved.As the systematic review included only RCTs, didnot include an economic assessment and hadundertaken the primary search in 2001, weperformed a new search for this assessment asoutlined in Appendix 3. The results are describedin the next section.Controlled trials of secondgenerationEA techniquesA total of 14 publications were found using thesearch strategy shown in Appendix 3. Three wereof MEA 86–88 and 11 were of TBEA. 82–85,89–94However, two of the MEA papers report on the

Health Technology Assessment 2004; Vol. 8: No. 3same trial at 12 (Cooper and colleagues) 86 and24 months (Bain and colleagues) 87 of follow-up. Inthis report, these papers will be referred to by thefirst and main trial publication, Cooper andcolleagues (1999). 86 Four of the TBEA papersreport the same trial at 12 months (Meyer andcolleagues), 82 24 months (Grainger andcolleagues), 84 36 months (Loffer) 85 and 60 months(Loffer and Graigner) 94 of follow-up. In addition,an erratum page appeared for the paper by Lofferwhich corrected the labelling of figures in theoriginal and added a chart that had been omittedfrom the original publication. 95 These will bereferred to in this report by the first, mainpublication, Meyer and colleagues (1998). 82Of the included trials, three were provided byindustry. Wallsten Medical, the makers ofCavaterm, provided a translation of a small RCTof TBEA versus RB ablation that had beenpublished in German 83 and confidential,unpublished trial details of an RCT of TBEAversus TCRE. Details of this second study havebeen removed from the public version of thisassessment. Microsulis Medical, the manufacturersof MEA equipment, provided details of an RCTthey conducted as part of their submission to theUS Food and Drug Administration (FDA) approvalprocess. 88 Our assessment of this study is based onthe information that they supplied.In summary, two MEA and eight TBEA trials wereincluded in the review although data were takenfrom all published accounts of these trials. Detailsof these studies are described below andsummarised in Table 6, with summary details inAppendix 7.Most of the included studies are RCTs. Two arenon-RCTs. One study, by Gervaise and colleagues 90(TBEA versus TCRE), obtained patient, surgicaland outcome details from the hospital notes ofthose undergoing TCRE at their institution duringthe same time period as women were undergoingTBEA. Another by Bongers and colleagues 89(TBEA versus TCRE) is a prospective cohortcomparison of women undergoing EA at one Dutchhospital where all ablations between 1992 and1994 were TCRE and all after 1995 were TBEA.Details of trialsPublication date/country and sample sizeThe studies were published between 1996 and2002 with recruitment between 1992 and 2001.The TBEA versus RB studies by Romer 83 and Zon-Rabelink 93 did not state the dates of recruitment.The number of women randomised in each trialranged from 20 to 322 (median 143). A total of1561 women were included in all trials of secondgenerationEA techniques.The MEA versus TCRE/RB study by Cooper andcolleagues 86 was based at a single centre in the UKwhereas the Microsulis study 88 (MEA versus RB)recruited women from eight sites in the UK andthe USA. Bongers and colleagues 89 (TBEA versusTCRE) recruited women from a single centre inThe Netherlands. The TBEA versus TCRE studyby Gervaise and colleagues 90 recruited womenfrom a single centre in France. Pellicano andcolleagues 92 (TBEA versus TCRE/RB) used a singlecentre in Italy. Soysal and colleagues 91 (TBEAversus RB) recruited women from a single centre inTurkey. The study by Meyer and colleagues 82(TBEA versus RB) recruited women from multiplecentres in the USA and Canada and that by Brunand colleagues 96 (TBEA versus TCRE) frommultiple centres in France. The Zon-Rabelinkstudy 93 (TBEA versus RB) is from The Netherlandsbut the number of centres involved is not stated.Indications for surgeryThe indication for surgery was variously describedas dysfunctional menstrual bleeding, 86menorrhagia, 82,85,93 menorrhagia ormetrorrhagia, 96 excessive menstrual bleeding, 84recurrent therapy refractory menorrhagia, 83menorrhagia unresponsive to medicaltreatment, 89,92 abnormal uterine bleeding 88 andabnormal menstrual bleeding. 90 Methods ofmeasuring bleeding also varied. The MEA versusTCRE/RB study by Cooper and colleagues 86included women who self-defined their menstrualloss as heavy. Brun and colleagues 96 (TBEA versusTCRE) used a PBAC score of >80 as an inclusioncriterion whereas Meyer and colleagues 82 (TBEAversus RB) and Soysal and colleagues 91 (TBEAversus RB) used a PBAC score of at least 150. TheMicrosulis study 88 (MEA versus RB) and the Zon-Rabelink study 93 (TBEA versus RB) defined HMBas a score of 185 or more. Gervaise andcolleagues 90 (TBEA versus TCRE) quantified HMBthrough the number of pads used per cycle. Nodescription of how HMB was measured is given byBongers and colleagues 89 (TBEA versus TCRE),Romer 83 (TBEA versus RB) or Pellicano andcolleagues 92 (TBEA versus TCRE/RB).Participant characteristicsThe median average age of the women includedin the studies was 42.6 years (range 40.2–46.3) forthe intervention arms and 43.2 years (range40–47.4 years) in the control arms. TheMicrosulis study 88 (MEA versus RB) and that by25© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Results26Zon-Rabelink 93 (TBEA versus RB) did not reportthe ages of participants.Fibroids >2 cm in diameter were reported in 12%of the women in the MEA trial by Cooper andcolleagues 86 (MEA versus TCRE/RB). Fibroids

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 6 Characteristics of trials reported in all included papers and treatmentAuthor/ No. of Average age Women with Intervention Control Pretreatment Surgeon experience Anaesthetic Length ofdate/design patients (years) fibroids treat- follow-upexcluded? ment (months)Cooper et al.,1999 86RCTBain et al.,2002 87RCTMicrosulis2002 88RCTBongers et al.,2000 89Non-RCT263 MEA 41.1 (SD 6.7)TCRE/RB 42.0(SD 8.4)263 MEA 41.4 (SD 5.4)TCRE/RB 42.2(SD 5.8)No Microwave TCRE/RB 3.6 mg goserelin5 weeks priorNo Microwave TCRE/RB 3.6 mg goserelin5 weeks prior322 Not stated No Microwave RB Leuprolide acetatedepot 3–5 weeksprior152 TBEA 42.5 (SD 6.3)TCRE 43.2 (SD 6.4)YesThermachoiceTBEAAt least 50 prior TCREs,at least 5 prior MEAsAt least 50 prior TCREs,at least 5 prior MEAsNot stated100% GA 12GA 24GA:MEA 37%RB 76%TCRE D&C Not stated GA and spinal 2412Brun et al.,2002 96RCTMeyer et al.,1998 82RCTGrainger et al.,2000 84RCTLoffer, 2001 85RCT51 TBEA 45.5 (±6.0,35–59)TCRE 46.7 (SD 6.0,33–46)275 TBEA 40.2 (SD 4.9)30–51RB 40.9 (SD 5.2)29–50Not clearYesCavatermthermal balloonThermachoicethermal balloon255 Not stated Yes Thermachoicethermal balloon255 Not stated Yes Thermachoicethermal balloonTCRED&C immediatelyprior to procedureRB None stated All had extensiveexperience of RBRBRB3-minute curettageusing 5-mm curetteprior to ablationTimed 3-minutesuction curettagegiven to all prior toablationExperienced surgeons Not stated 3All experienced in RBand trained in TBEAAll experienced in RBand trained in TBEAGA:TBEA 53%RB 84%12Not stated 24LA, LA withsedation andGA. More GAwith RB36continuedHealth Technology Assessment 2004; Vol. 8: No. 327

28TABLE 6 Characteristics of trials reported in all included papers and treatment (cont’d)Author/ No. of Average age Women with Intervention Control Pretreatment Surgeon experience Anaesthetic Length ofdate/design patients (years) fibroids treat- follow-upexcluded? ment (months)ResultsLoffer andGrainger, 2002 94255 TBEA 40.4RB 40.9YesThermachoicethermal balloonRB3-minute suctioncurettageAll experienced in RBand trained in TBEANot stated 60RCTGervaise et al.,1999 90Non-RCT147 TBEA 46.3 (±1.434–66TCRE 47.4 (±0.2)34–65YesThermachoiceballoonTCRE None Not stated GA TCRE GAand LA (38%)for TBEA18Pellicano et al.,2002 92RCT96 TBEA 42.6 (±4.4)TCRE/RB: 43.2(±3.5)SubmucousCavatermballoonTCRE/RBTreatment groupnone. Control groupGnRH 6 and2 weeks prior tosurgerySurgeons ‘proficient’ inTCRESpinal 24Romer, 1998 83RCT20 TBEA 42 (37–52)RB 40 (37–50)YesCavatermballoonRB2× monthlyinjections of GnRH(leuprolide 3.75 mg)operation 2 weeksafter injectionNot stated All GA 9–15Soysal et al.,2001 91RCT96 TBEA 43.6 (±2.5,40–49)RB 44.3 (±2.6,40–49)No, all patientshad fibroidsThermachoiceballoonRB2× monthlyinjections of GnRHanalogue (3.6 mggoserelin acetate)One experiencedsurgeon performed allRB, TBEA by staffsurgeons supervised byresidentsAll RB GA, allTBEA LA12Zon-Rabelink,2001 93RCT139 Not stated Not stated ThermachoiceballoonRBPretreatment withZoladex 6 and2 weeks prior tosurgeryNot stated Not stated 24SD, standard deviation.

Health Technology Assessment 2004; Vol. 8: No. 3Quality assessment of RCTsThe quality of the reports of RCTs is summarisedin Table 7.Internal validitySample sizeThe 10 studies included 20, 83 51, 96 96, 91 96, 92139, 93 147, 90 152, 89 263, 86 275 82 and 322 88women. Sample size calculations were performedin three of the RCTs 82,86,96 and one non-RCT. 89Sample size calculations were not reported byMicrosulis 88 (MEA versus RB), Pellicano andcolleagues 92 (TBEA versus TCRE/RB) Soysal andcolleagues 91 (TBEA versus RB), Gervaise andcolleagues 90 (TBEA versus TCRE), Romer 83(TBEA versus RB) or Zon-Rabelink 93 (TBEAversus RB).The trial by Cooper and colleagues 86 (MEA versusTCRE/RB) is based on an 80% power to detect a15% difference in satisfaction (p = 0.05) based on78% women satisfied with TCRE. Actual levels oftotal or general satisfaction were 77% in the MEAgroup and 75% in the TCRE and RB group at12 months (significant difference not found). Apatient questionnaire was used to measure thisoutcome.In the trial by Meyer and colleagues 82 (TBEAversus RB), sample size was calculated based on90% power to detect 20% less effectiveness in thetreatment group (p = 0.05) based on an 85%response rate for RB. “Effectiveness” is notdefined. However, 86% of women undergoingTBEA and 87% of women undergoing RB ablationwere reported as ‘very satisfied’ with treatment andthere was no significant difference in the twogroups in the percentage of women who had a90% reduction in PBAC scores (62% with TBEAversus 68% with RB).The trial by Bongers and colleagues 89 (TBEAversus TCRE) reports that assuming a 9% reinterventionrate after TCRE, a series of 150patients would be needed to show that balloonablation is equally effective; 152 women wereincluded in this trial and the re-interventionrate was 20% in the TCRE arm. It should be notedthat the sample size calculation appears only inthe abstract, and not in the body of the trialreport.The trial by Brun and colleagues 96 (TBEA versusTCRE) based the sample size calculation on 160patients giving a 90% power (p = 0.05) to detect a15% difference in efficacy. However, only 51women were actually recruited to the trial.Selection biasAllocation to intervention or control arm in theMEA trial by Cooper and colleagues 86 (MEAversus TCRE/RB) was random and treatmentallocation was concealed. Women wererandomised through a telephone call to asecretary who opened a series of sealed, opaque,sequentially numbered envelopes showing atreatment code. The sequence was predeterminedby computer-generated random number blocks of20. Allocation to study arm in the Meyer andcolleagues’ trial 82 (TBEA versus RB) was random,but there was no account of steps taken to concealallocation. Soysal and colleagues 91 (TBEA versusRB) used computer-generated randomisation andopaque, sealed envelopes for allocationconcealment. Pellicano and colleagues 92 (TBEAversus TCRE/RB) also used a computer-generatedrandom number sequence but do not report onallocation concealment. Patient characteristics inthe two arms of each of these studies appearsimilar. Zon-Rabelink 93 (TBEA versus RB) statesthat women were first stratified by age (over orunder 45 years) and parity (nulliparous or parouswomen) and then randomised with allocation viablind envelopes.Brun and colleagues 96 (TBEA versus TCRE) usedcentral allocation of patients on a 1:1 basis but themethod of randomisation is not described, nor isthe allocation concealment method reported. Assome data are missing from 24% of patients atbaseline, it is not possible to say whether thegroups were similar – mean age was similar butthe range was greater for women in the TBEAgroup (TBEA 35–59 years, TCRE 33–46 years).The Gervaise and colleagues’ study 90 (TBEAversus TCRE) was not randomised. Women in theintervention arm were consecutive patientsreceiving TBEA during the study period. Controls,who received TCRE, were matched retrospectivelyfrom the records of women receiving TCREduring the same time period. Inclusion andexclusion criteria were applied. There weresignificant differences at baseline between the twogroups, with the TCRE groups having lower parity(1.9 versus 2.4) and containing more women whowere post-menopausal (27% versus 7%) than theTBEA group, although the number of pads usedper cycle was similar. Higher parity is associatedwith increased HMB (see the section ‘Cause ofHMB’, p. 3).The Bongers and colleagues’ study 89 (TBEA versusTCRE) was not randomised. The authors reportthat consecutive women were recruited29© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Results30prospectively to the trial, with all women attendingfor EA from 1992 to 1994 undergoing TCRE andall those from 1995 to 1997 undergoing TBEA.Inclusion and exclusion criteria were applied. Thebaseline characteristics of the two groups arecomparable.The Microsulis trial 88 (MEA versus RB) and thestudy by Romer 83 (TBEA versus RB) do not reporton randomisation, allocation or blinding methods.The patient groups reported by Romer 83 seem tohave similar characteristics.Performance biasTCRE and RB ablation are skilled operations,which, like most surgical procedures, are difficultto standardise. The RCTs vary in the extent towhich standardisation of procedures are reported.All TCRE/RB ablations were undertaken by twoexperienced, senior specialist registrars in theMEA trial by Cooper and colleagues 86 (MEAversus TCRE/RB), who used a combined TCREand an electrocoagulation technique, ablating thefundus and cornual regions with an RB. Glycine(1.5%) was used as the distension medium. A 90°loop 7 mm in diameter and 3 mm deep was usedfor TCRE.No details of surgeon experience are given in theMicrosulis trial 88 (MEA versus RB) and, as this isan eight-centre trial, differences in technique andexperience are possible. Indeed, one study centreperformed all operations under GA. Analysis bycentre showed that at one centre only, patientstreated with MEA were significantly more likely tohave amenorrhoea at 12 months than thosetreated by RB (p = 0.007). It is possible that this isrelated to inexperience with the RB technique. Nosignificant differences in amenorrhoea were shownat the other seven centres.No details are given about surgeon experience byBongers and colleagues. 89 Initially, women treatedwith TBEA were treated for 8 minutes, and thiswas increased to 16 minutes for the second half ofthe study. However, looking at the extent of totalablation of the endometrium, the authorsexamined the possibility that there would be alearning curve with both TCRE and TBEAprocedures, leading to improved results over time.This was not seen.The study by Brun and colleagues 96 (TBEA versusTCRE) recruited from seven centres with eachcontributing details on 2–12 women. All thesurgeons are described as ‘experienced’ at TCREalthough variation is possible, particularly as thesewere performed to local guidelines at each centre.In the trial by Meyer and colleagues 82 (TBEAversus RB), it is stated that all surgeons hadextensive experience of RB ablation. However, thiswas a 14-centre trial so variation in technique andexperience is possible, although all are describedas ‘skilled’. Either 1.5% glycine or 3% sorbitol wasused as distension fluid and the specifics ofsurgery and equipment depended on surgeons’preference.Neither the number of surgeons performingTCRE nor surgeon experience is mentioned in thenon-randomised study by Gervaise andcolleagues 90 (TBEA versus TCRE).Pellicano and colleagues 92 (TBEA versusTCRE/RB) report that all surgeons were‘proficient’ at combined TCRE and RB ablation;2.7% sorbitol or 0.54% mannitol was used asdistension solution.Romer 83 (TBEA versus RB) and Zon-Rabelink 93(TBEA versus RB) do not report on the extent ofsurgeon experience or the operating procedure.One experienced surgeon performed all the RBablations in the study by Soysal and colleagues 91(TBEA versus RB). Glycine was the distensionmedium and a 3-mm RB electrode was used forcoagulation. Staff surgeons performed the TBEAsunder the supervision of residents.Detection biasIt is not possible to blind patients or surgeons towhich procedure was being undertaken. Althoughit would be possible to blind those who areassessing outcomes or carrying out analyses, noneof the studies report this.Attrition biasAt 24 months’ follow-up, the trial by Cooper andcolleagues 87 (MEA versus TCRE/RB) reported that14 patients (5%) were LTFU, nine (7%) in theMEA arm and five (4%) in the TCRE and RB arm.This is fewer than were reported as LTFU at12 months: 23 (9%) overall, 13 (10%) in the MEAarm and 10 (7%) in the TCRE and RB arm.Follow-up was by postal questionnaire, but at24 months those who had not returned theirquestionnaire were contacted by telephone torequest its return or to be interviewed by telephonewhere necessary. It is stated that ITT analysis isundertaken but some analyses are carried out onlyon treatment completers followed up.

Health Technology Assessment 2004; Vol. 8: No. 3The Microsulis study 88 (MEA versus RB) reportsthat 7% of women were LTFU at 12 months, 13women (6%) in the MEA arm and nine (8%) in theRB arm. All analyses are reported on an ITTbasis. In the study by Brun and colleagues 96(TBEA versus TCRE), one woman (2%) was LTFUat the time of randomisation. No further LTFUwas reported at 3 months.The study by Bongers and colleagues 89 (TBEAversus TCRE) does not report LTFU and statesthat ITT analysis is undertaken. However thepaper provides numbers and percentages for the‘satisfaction’ outcomes at 24 months that are notcalculated on an ITT basis. This suggests that 37%of TBEA and 38% of TCRE patients may havebeen LTFU by 24 months. Some other outcomes,such as amenorrhoea rates, only providepercentages so it is not possible to ascertainwhether these are calculated on an ITT basis.In the trial by Meyer and colleagues 82 (TBEAversus RB), 275 patients were randomised but15 electively withdrew before the procedure wasperformed. A further four were discovered to beineligible and one had a uterine perforation andwas not treated under protocol. These numbersare not reported consistently and are given as 11withdrew, eight ineligible and one perforation inthe paper of 3-year results; 85 255 women weretherefore treated under protocol and these arereferred to most often in all the papers as theoriginal sample. However, only the details of245 women that were available at 6 months arereported on in the 12-month paper, 82 although itis stated that there were no significant differencesbetween these and the original sample. ITTanalysis is not performed. Numbers of the original275 women allocated to treatment and controlarms are only reported in the 3-year follow-uppaper; 46% of the recruited participants wereLTFU by 60 months. This includes women fromtwo centres that did not provide 5-year follow-updata. Furthermore, in the 5-year paper, 94 patientswho had undergone repeat surgery are excludedfrom calculations of bleeding and pain outcomes.Gervaise and colleagues 90 (TBEA versus TCRE)reported no LTFU at 18 months. However, detailsof the women in the TCRE group were obtainedfrom records retrospectively, which introduces thepotential for bias (direction unknown), as theywere selected on the basis that follow-upinformation was available.Pellicano and colleagues 92 (TBEA versus RB)report 29% LTFU at 2 years.Romer 83 (TBEA versus RB) had no LTFU of theoriginal 20 women at 12 months.The study by Soysal and colleagues 91 (TBEAversus RB) lost three patients from the TBEAgroups prior to the procedure being performed,but reported no other LTFU at 12 months. Thesethree patients were excluded from analysis.Two patients were excluded after randomisation byZon-Rabelink 93 (TBEA versus RB). Both had beenallocated to the RB group. One was discovered tohave polyps at the time of operative hysteroscopy,and one was discovered to have a PBAC score of

32TABLE 7 Methodological characteristics of included controlled trialsAuthor/ No. of Adequate Blinding? Comparability Same LTFU (%) Sample ITT? General- Main Inter- Conflictsdate patients allocation of groups? interven- size calc.? isability outcome centre of interest?to groups? tion to all measured variability?patients?independentlyResultsCooper et al., 263 Yes Yes Yes Yes 12 months Yes Stated that it is, High Yes Not Yes1999 86 9% but some data applicable24 months points appear to5% use differentdenominators –missing data?Microsulis, 322 Uncertain Uncertain Uncertain Uncertain 7% Uncertain Yes Low Yes Yes Yes2002 88Bongers 152 N/A N/A Yes No Uncertain – Yes Stated yes, but Medium Yes Not None statedet al., 2000 89 maybe 38% some data appears applicableat 2 yearsto be on evaluablepatientsBrun et al., 51 Yes Uncertain Uncertain – No 2% at Yes based No Medium Yes Not None stated2002 96 baseline data random- on examinedmissing from isation 160 patients24%Meyer et al., 275 Yes Uncertain Yes Yes 12 months Yes No. Patients High Yes for None found Yes1998 82 11% lost between bleeding,24 months randomisation uncertain17% and treatment for36 months are excluded, in satisfaction22% addition some60 months data points46% appear to usedifferentdenominators– missing data?continued

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 7 Methodological characteristics of included controlled trials (cont’d)Author/ No. of Adequate Blinding? Comparability Same LTFU (%) Sample ITT? General- Main Inter- Conflictsdate patients allocation of groups? interven- size calc.? isability outcome centre of interest?to groups? tion to all measured variability?patients?independentlyGervaise 147 No No No – 27% Yes None No N/A Medium Uncertain Not Noneet al., 1999 90 of women given applicableTCRE and 7%given TBEAwere postmenopausalPellicano 96 Yes Uncertain Yes Yes 29% at No No High Yeset al., 2002 922 yearsRomer, 20 Uncertain Uncertain Yes Yes None Uncertain N/A Low Uncertain Not None1998 83 applicableSoysal et al., 96 Yes Uncertain Yes Yes None No 3 patients High Yes No None2001 91 allocated toTBEA did notreceive treatmentand wereexcluded, noother LTFUZon-Rabelink, 139 Yes Yes Uncertain Yes 2% at No No Low Yes Not stated None2001 93 2 yearsHealth Technology Assessment 2004; Vol. 8: No. 333

Results342-year follow-up, 86,89,92,93 one had an 18-monthfollow-up, 90 three had a 12-month follow-up 83,88,91and one had a 3-month follow-up. 96 Romer 83 (TBEAversus RB) followed-up patients for 9–15 months.Assessment of effectivenessReporting of outcomesOutcome percentages recorded in the followingtables are given as reported in the trials and alsohave been recalculated on an ITT basis wherenecessary. ITT figures are given in parentheses.A wide range of outcomes of surgery werereported across the studies and these are shown inTables 8–20. Broadly, the outcomes can be groupedinto the following categories:●●●●●●●●●bleeding outcomespremenstrual syndrome (PMS)-related outcomesdysmenorrhoeaanaemia/haemoglobin outcomessatisfactionQoLoperation detailsfurther surgeryadverse effects (perioperative andpostoperative).The way in which outcomes were reported differsbetween studies. For example, some bleedingoutcomes use mean PBAC scores, or changes inthese, whereas others report the numbers of womenwith various bleeding patterns. This means that it isnot always possible to compare results across studiesor to combine them for meta-analysis.Bleeding patternsAmenorrhoea, the absence of menses, is reportedby seven of the included studies, and has aconsistent definition. Table 8 shows the rates ofpostoperative amenorrhoea. Amenorrhoea at12 months was reported for a median of 45% ofwomen undergoing MEA (range 36–40%) and amedian of 15% (range 10–40%) for TBEA. At12 months, a median of 30% of womenundergoing TCRE or RB had amenorrhoea (range17–46%). The lowest percentage (10%) is found inthe TBEA arm of the trial containing women whoall had fibroids. 91Amenorrhoea at 24 months was experienced by44% of women undergoing MEA, a median of 17%(range 13–22%) of those undergoing TBEA and bya median of 28% (range 17–43%) of womenundergoing TCRE or RB. Only Meyer andcolleagues 82 report on longer term follow-up. At36 months, 12% of women undergoing TBEA and19% of women undergoing RB had amenorrhoeaand at 60 months 14% of women undergoingTBEA and 10% of those undergoing RB wereamenorrhagic.It is not clear from the data supplied by Bongersand colleagues 89 if the stated amenorrhoea ratesare based on ITT calculations. Brun andcolleagues 96 (TBEA versus TCRE), Pellicano andcolleagues 92 (TBEA versus TCRE/RB) and Zon-Rabelink 93 (TBEA versus RB) do not reportamenorrhoea.At 12 months, Meyer and colleagues 82 (TBEAversus RB) reported a statistically significantdifference between the TBEA (13%) and RB (22%)groups (p < 0.05).Figure 9 illustrates the findings for amenorrhoea at12 months for first-generation versus secondgenerationEA techniques and Figure 10 showsthose at 24 months. The size of the data pointsindicates the relative size of each study. In mostcases the CIs cross the central line, indicating thatdifferences were not statistically significant. Thesignificant difference detected by Meyer andcolleagues 82 at 12 months (TBEA versus RB) is notseen in the forest plot because the data have beenrecalculated here on an ITT basis whereas theoriginal study analysis excluded women LTFU.At 24 months, only the Meyer 82 and colleagues’study (TBEA versus TCRE) indicates a morefavourable outcome for RB. However, there was aloss to follow-up in this trial (17% at 2 years). Thestudy results have not been statistically combinedowing to clinical heterogeneity between the trials.Other recorded outcomes for bleeding are shownin Tables 9 and 10. Note that data for 24 months(Meyer and colleagues, 82 TBEA versus RB) wereestimated from data presented in graph form inthe original study report. Three trials reportedpostoperative bleeding in terms of spotting,hypomenorrhoea, eumenorrhoea, menorrhagia ormetrorrhagia; Romer 83 (TBEA versus RB) at12 months’, Meyer and colleagues 82 (TBEA versusRB) at 24 and 36 months’ follow-up, and Gervaiseand colleagues 90 (TBEA versus TCRE)immediately and at 24 months. At 24 months,5–8% of patients who had undergone TBEA and9–15% of those who had undergone TCRE or RBwere still experiencing menorrhagia. At60 months, this figure was 2% and 1%,respectively. For further details, see Table 9. Notrial reported statistically significant differencesbetween the groups for recurrent menorrhagia.

Health Technology Assessment 2004; Vol. 8: No. 3StudyOdds ratio(95% CI)CooperMicrosulisMeyerRomerSoysal0.96 (0.58 to 1.59)1.47 (0.92 to 2.34)0.56 (0.30 to 1.04)1.56 (0.24 to 9.91)0.58 (0.18 to 1.93)Favours controlFavours treatment0.1 1 10Odds ratioFIGURE 9 Forest plot of amenorrhoea at 12 months – first-generation versus second-generation EA methods (random effects model,results not pooled)StudyOdds ratio(95% CI)Cooper MEAMeyer TBEAGervaise TBEA1.21 (0.74 to 1.98)0.36 (0.20 to 0.65)0.90 (0.42 to 1.95)Overall (95% CI) 0.74 (0.34 to 1.59)Favours controlFavours treatment0.1 1 10Odds ratioFIGURE 10 Forest plot of amenorrhoea at 24 months – first-generation versus second-generation EA methods (random effects model,results not pooled)Bongers and colleagues 89 (TBEA versus TCRE)report on the number of women undergoing reinterventionsurgery who cited menorrhagia ormetrorrhagia as a reason, but it is not clear ifother women may have also suffered thesesymptoms but not undergone repeat surgery.Six trials reported changes in PBAC score. At12 months Meyer and colleagues 82 (TBEA versusRB) report that 73% of the TBEA and 70% of theRB group had a score of

36TABLE 8 Postoperative amenorrhoea – % (ITT %)Author/date Treatment Immediate post-op./ 12 months 24 months 36 months 60 months3 monthsResultsIntervention Control Intervention Control Intervention Control Intervention Control Intervention ControlCooper et al., 1999 86 MEA – – 40 (36) 40 (36) 47 (44) 41 (40) – – – –Microsulis, 2002 88 MEA – – 55 (55) 46 (46) – – – – – –Bongers et al., 2000 89 TBEA 17 (–) 36 (–) 15 (–) 22 (–) 13 (–) 17 (–) – – – –Brun et al., 2002 96 TBEA – – – – – – – – – –Meyer et al., 1998 82 TBEA – 15 (14) 15 (13) 27 (22) 17 (15) 45 (33) 15 (12) 26 (19) 33 (14) 23 (10)Gervaise et al., 1999 90 TBEA 25 (25) 38 (38) – – 36 (22) 38 (24) – – – –Pellicano, et al., 2002 92 TBEA – – – – – – – – – –Romer, 1998 83 TBEA – – 40 (40) 31 (30) – – – – – –Soysal et al., 2001 91 TBEA – – 11 (10) 17 (17) – – – – – –Zon-Rabelink, 2001 93 TBEA – – – – – – – – – –

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 9 Results: type of postoperative bleeding – % (% ITT)Author/date Length of follow-up (months) Intervention Spotting Hypomenorrhoea Eumenorrhoea Menorrhagia MetrorrhagiaCooper et al., 1999 86 12 MEA – – – – –TCRE/RB – – – – –24 MEA – – – –TCRE/RB – – – – –Microsulis, 2002 88 12 MEA – – – – –RB – – – – –Bongers et al., 2000 89 24 TBEA – – – 9 (9) a 0 3 (3) aTCRE – – – 12 (12) a 9 (9) aBrun et al., 2002 96 3 TBEA – – – – –TCRE – – – – –Meyer et al., 1998 82 12 TBEA – – – – –RB – – – – –24 TBEA 11 (9)0 45 (40) 21 (19) 9 (8)0 –RB 13 (10) 30 (22) 21 (16) 12 (9) –36 TBEA 10 (8)0 39 (33) 29 (24) 7 (6)0 –RB 16 (12) 26 (19) 25 (18) 6 (4)060 TBEA 10 (4)0 38 (17) 25 (11) 05 (2)0 –TCRE 11 (5)0 25 (11) 28 (12) 3 (1)0 –Gervaise et al., 1999 90 1 TBEA 25 (25) 22 (22) 38 (38) 11 (11) 4 (4)TCRE 38 (38) 31 (31) 13 (13) 12 (12) 5 (5)24 TBEA 36 (22) 16 (10) 34 (20) 9 (5)0 4 (3)TCRE 38 (38) 28 (28) 17 (17) 15 (15) 2 (2)Romer, 1998 83 12 TBEA – 50 (50) 10 (10) – –RB – 60 (60) 10 (10) – –Pellicano et al., 2002 92 12 TBEA – – – – –TCRE/RB – – – – –24 TBEA – – – – –TCRE/RB – – – – –Soysal et al., 2001 91 12 TBEA – – – – –RB – – – – –Zon-Rabelink, 2001 93 12 TBEA – – – – –RB – – – – –24 TBEA – – – – –RB – – – – –a These are minima – women who required repeat surgery for this type of bleeding.Health Technology Assessment 2004; Vol. 8: No. 337

Results38normal bleeding levels and this is reported by 87%of women in the MEA group and 83% of womenin the RB group. Soysal and colleagues 91 (TBEAversus RB) report a mean PBAC score of 41.1 inthe TBEA group (a mean reduction of 343) and amean PBAC score of 40 in the RB group (a meanreduction of 345). At 3 months, Brun andcolleagues 96 (TBEA versus TCRE) report a score of44 (±48) in the TBEA group (a decrease of 413)and a postoperative score of 75 (±78; a decreaseof 199) in the TCRE group. In addition, 71% ofthe TBEA and 79% of the RB group had ableeding score of

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 10 Postoperative PBAC scores – % (% ITT)Author/date Length of Intervention Mean PBAC PBAC PBAC PBAC PBAC Mean Bleedingfollow-up PBAC

40TABLE 10 Postoperative PBAC scores – % (% ITT) (cont’d)Author/date Length of Intervention Mean PBAC PBAC PBAC PBAC PBAC Mean Bleedingfollowed-up PBAC

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 11 Postoperative bleeding patterns – % (% ITT)Author/date Length of Intervention 3–7 days >7 days >3 days 2× sanitary Menstruation Reduction infollow-up bleeding bleeding heavy protection unchanged or number of women(months) bleeding needed worse with anaemiaCooper et al., 1999 86 12 MEA 42 (38) 5 (5) 7 (6) 12 (11) 8 (7) –TCRE /RB 41 (38) 7 (7) 6 (5) 13 (12) 9 (8) –24 MEA – – 2 (2) 14 (7)0 7 (6) –TCRE/RB – – 5 (5) 22 (13) 11 (10) –Microsulis, 2002 88 12 MEA – – – – – –RB – – – – – –Bongers et al., 2000 89 24 TBEA – – – – – –TCRE – – – – – –Brun et al., 2002 96 3 TBEA – – – – – –TCRE – – – – – –Meyer et al., 1998 82 12 TBEA – – – – – ~ 60 (55)RB – – – – – ~ 60 (49)24 TBEA – – – – – –RB – – – – – –36 TBEA – – – – – –RB – – – – – –Gervaise et al., 1999 90 1 TBEA – – – – – –TCRE – – – – – –24 TBEA – – – – – –TCRE – – – – – –Pellicano et al., 2002 92 12 TBEA – – – – 2 (5) a 0 –TCRE/RB – – – – 6 (14) a –24 TBEA – – – – 3 (8) a 0 –TCRE/RB – – – – 8 (19) a –Romer, 1998 83 12 TBEA – – – – – –RB – – – – – –Soysal et al., 2001 91 12 TBEA – – – – – –RB – – – – – –Zon–Rabelink, 2001 93 24 TBEA – – – – – –RB – – – – – –a Values for “bleeding recurs”.Health Technology Assessment 2004; Vol. 8: No. 341

42TABLE 12 Postoperative menstrual pain – % (% ITT)Author/date Length of Intervention Dysmenor- Dysmenor- Dysmenor- Pain recurs Mean pain Mild Moderate Severefollow-up rhoea rhoea same rhoea score: dysmenor- dysmenor- dysmenor-(months) decreased or worse mean, rhoea rhoea rhoearangeResultsCooper et al., 1999 86 12 MEA – 21 (19) – – 1 (0–9) – – –TCRE/RB – 18 (16) – – 1 (0–7) – – –24 MEA – 18 (17) – – 0 (0, 6) c – – –TCRE/RB – 22 (22) – – 1 (0, 8) c – – –Microsulis, 2002 88 12 MEA – – 31 (31) – – – – –RB – – 31 (31) – – – – –Bongers et al., 2000 89 24 TBEA – – 0 – – – – –TCRE – – 4 (4) b – – – – –Brun et al., 2002 96 3 TBEA – – – – – – – –TCRE – – – – – – – –Meyer et al., 1998 82 12 TBEA 70 (64) 30 (27) a – – – – – –RB 75 (62) 25 (20) a – – – – – –24 TBEA – – – – – – – –RB – – – – – – – –36 TBEA – – – – – – – –RB – – – – – – – –60 TBEA – – – – – 21 (9)0 21 (9) 5 (4)RB – – – – – 26 (12) 13 (5) 8 (4)Gervaise et al., 1999 90 1 TBEA – – – – – – – –TCRE – – – – – – – –24 TBEA – – – – – – – –TCRE – – – – – – – –Pellicano et al., 2002 92 12 TBEA – – – 1 (2)0 – – – –TCRE/RB – – – 7 (14) – – – –24 TBEA – – – 2 (4)0 – – – –TCRE/RB – – – 9 (18) – – – –continued

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 12 Postoperative menstrual pain – % (% ITT) (cont’d)Author/date Length of Intervention Dysmenor Dysmenor- Dysmenor- Pain recurs Mean pain Mild Moderate Severefollow-up -rhoea rhoea same rhoea score: dysmenor- dysmenor- dysmenor-(months) decreased or worse mean, rhoea rhoea rhoearangeRomer, 1998 83 12 TBEA – – – – – – – –RB – – – – – – – –Soysal et al., 2001 91 12 TBEA – – – – – – – –RB – – – – – – – –Zon-Rabelink, 2001 93 24 TBEA – – – – – – – –RB – – – – – – – –a Calculated from given categories “Dysmenorrhoea unchanged” and “Dysmenorrhoea increased”.b Minimum – women undergoing repeat surgery for this.c Median (25th, 75th percentile).Health Technology Assessment 2004; Vol. 8: No. 343

ResultsStudyOdds ratio(95% CI)Cooper MEAMicrosolis MEAMeyer TBEA1.16 (0.62 to 2.20)0.99 (0.60 to 1.64)1.45 (0.83 to 2.55)Favours treatmentFavours control0.1 1 10Odds ratioFIGURE 11 Forest plot of dysmenorrhoea 12 months postoperation second-generation versus first-generation EA (random effectsmodel, results not pooled)44dysmenorrhoea; however it is not stated whetherof not other women suffered dysmenorrhoea whodid not opt for further surgery.Figure 11 shows the dysmenorrhoea rates at12 months for first-generation versus secondgenerationEA techniques. The data points havebeen produced from the numbers describingdysmenorrhoea as the same or worse as preoperatively.In all cases the CIs cross the centralline indicating no statistically significantdifferences between the groups. Study results havenot been combined owing to clinical heterogeneitybetween the trials.PMS symptomsTwo studies 82,86 report postoperative PMSsymptoms although in different ways. The studyby Cooper and colleagues 86 (MEA versusTCRE/RB) reports prevalence of individualsymptoms of PMS at 12 months postoperation:bloating, breast discomfort, irritability, headachesand depression (see Table 13). There was nostatistically significant difference between thegroups for any PMS measures. Meyer andcolleagues 82 (TBEA versus RB) report the numberof women who do not have PMS symptoms at 12,24 and 36 months postoperation, and the numberof women who have moderate or severe PMS at 12(TBEA 30%, RB 24%) and 24 months (TBEA 25%,RB 22%). There were no statistically significantdifferences in PMS symptoms between the studyarms (Table 13).Satisfaction with treatmentTwo studies, Brun and colleagues 96 (TBEA versusTCRE) and Soysal and colleagues 91 (TBEA versusRB), did not report patient satisfaction. Theothers use slightly different measures ofsatisfaction.The study by Cooper and colleagues 86 (MEAversus TCRE/RB) reports whether women were“totally or generally satisfied” with treatment at 12and 24 months after their operations. At12 months, 69% of both groups were totally orgenerally satisfied and at 24 months, 74% of thoseundergoing MEA and 64% of those undergoingTCRE and RB were totally or generally satisfied.Differences between groups were not statisticallysignificant (Table 14). However, this study wasdesigned to be able to detect 20% less satisfactionin the MEA arm assuming that 85% of the TCREpatients were satisfied (90% power, 95% precision)and so is underpowered to detect if the observeddifference is significant.Cooper and colleagues 86 (MEA versus TCRE/RB)also report that 70% of women in both groupsregarded their treatment as effecting a cure or

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 13 Postoperative PMS symptoms – % (% ITT)Author/date Length of Intervention Bloating Breast Irritability Headaches Depression No PMS PMSfollow-up discomfort moderate(months)severeCooper et al., 1999 86 12 MEA 65 (58) 55 (50) 58 (52) 48 (43) 36 (33) – –TCRE/RB 51 (47) 49 (45) 52 (48) 44 (40) 40 (37) – –24 MEA – – – – – – –TCRE/RB – – – – – – –Microsulis, 2002 88 12 MEA – – – – – – –RB – – – – – – –Bongers et al., 2000 89 24 TBEA – – – – – – –TCRE – – – – – – –Brun et al., 2002 96 3 TBEA – – – – – – –TCRE – – – – – – –Meyer et al.,1998 82 12 TBEA – – – – – 27 (25) 33 (30)RB – – – – – 28 (23) 29 (24)24 TBEA – – – – – 29 (26) 29 (25)RB – – – – – 35 (27) 29 (22)36 TBEA – – – – – 32 (26) –RB – – – – – 37 (27) –60 TBEA – – – – – – –RB – – – – – – –Gervaise et al., 1999 90 1 TBEA – – – – – – –TCRE – – – – – – –24 TBEA – – – – – – –TCRE – – – – – – –Pellicano et al., 2002 92 12 TBEA – – – – – – –TCRE/RB – – – – – – –24 TBEA – – – – – – –TCRE/RB – – – – – – –Romer, 1998 93 12 TBEA – – – – – – –RB – – – – – – –Soysal et al., 2001 91 12 TBEA – – – – – – –RB – – – – – – –Zon-Rabelink, 2001 93 24 TBEA – – – – – – –RB – – – – – – –Health Technology Assessment 2004; Vol. 8: No. 345

46TABLE 14 Satisfaction with treatment and its acceptability – % (% ITT)Author/date Length Interven- Very/ Satisfied Not Not Excel- Good Moder- Totally Cure or Treat- Menstrual Reoffollow- tion perfectly very satisfied/ lent ate or acceptable ment loss commendup satisfied satisfied complaints generally improve- accept- acceptable treat-(months) no it is satisfied ment able ment?effect worseResultsCooper et al., 1999 86 12 MEA – – – – – – – 77 (69) 78 (70) 94 (84) – 91 (81)TCRE/RB – – – – – – – 75 (69) 76 (70) 90 (84) – 89 (82)24 MEA – – – – – – – 79 (74) – – 96 (89) 90 (84)TCRE/RB – – – – – – – 67 (64) – – 88 (84) 90 (87)Microsulis, 2002 88 12 MEA – 98 (98) a – 2 (2) – – – – – 99 (99) b – –RB – 99 (99) a – 1 (1) – – – – – 100 (100) b – –Bongers et al., 2000 89 3 TBEA 66 (66) 20 (20) 10 (10) 4 (4) – – – – – – – –TCRE 80 (80) 11 (11) 8 (8) 1 (1) – – – – – – – –6 TBEA 63 (51) 10 (8) 16 (13) 11 (9) – – – – – – – –TCRE 57 (52) 7 (7) 35 (32) 1 (1) – – – – – – – –12 TBEA 63 (52) 13 (10) 10 (8) 14 (12) – – – – – – – –TCRE 57 (52) 2 (1) 37 (28) 9 (7) – – – – – – – –24 TBEA 60 (36) 4 (3) 11 (6) 25 (16) – – – – – – – –TCRE 43 (27) 6 (4) 17 (11) 34 (21) – – – – – – – –Brun et al., 2002 96 3 TBEA – – – – – – – – – – – –TCRE – – – – – – – – – – – –Meyer et al., 1998 82 12 TBEA 86 (78) 10 (9) – 4 (4) – – – – – – – –RB 87 (72) 12 (10) – 1 (1) – – – – – – – –24 TBEA 86 (77) 10 (9) – 4 (3) – – – – – – – –RB 87 (66) 11 (9) – 2 (1) – – – – – – – –36 TBEA 88 (72) 9 (6) – 3 (2) – – – – – – – –RB 92 (67) 6 (4) – 2 (1) – – – – – – – –60 TBEA – 93 (42) – – – – – – – – – –RB – 100 (44) – – – – – – – – – –continued

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.TABLE 14 Satisfaction with treatment and its acceptability – % (% ITT) (cont’d)Author/date Length Interven- Very/ Satisfied Not Not Excel- Good Moder- Totally Cure or Treat- Menstrual Reoffollow- tion perfectly very satisfied/ lent ate or acceptable ment loss commendup satisfied satisfied complaints generally improve- accept- acceptable treat-(months) no it is satisfied ment able ment?effect worseGervaise et al., 1 TBEA – – – – – –– – – – – – –1999 90 TCRE – – – – – –– – – – – – –24 TBEA – – – – – – – – – – – –TCRE – – – – – – – – – – – –Pellicano et al., 2002 92 3 TBEA – – – – 27 (59) 13 (28) 0 – – – – –TCRE/RB – – – – 21 (42) 12 (24) 9 (18) – – – – –12 TBEA – – – – 20 (43) 10 (22) 5 (11) – – – – –TCRE/RB – – – – 12 (24) 12 (24) 10 (20) – – – – –24 TBEA – – – – 16 (35) 12 (26) 5 (11) – – – – –TCRE + RB – – – – 2 (4) 18 (36) 3 (6) – – – – –Romer, 1998 83 12 TBEA – 100 (100) – – – – – – – – – –RB – 100 (100) – – – – – – – – – –Soysal et al., 2001 91 12 TBEA – – 33 (31) – – – – – – – – –RB – – 39 (39) – – – – – – – – –Zon-Rabelink, 2001 93 24 TBEA – 80 (80) – – – – – – – – – –RB – 75 (73) – – – – – – – – – –a “Very satisfied” and “satisfied” combined.b “Acceptance of operation positive”.Health Technology Assessment 2004; Vol. 8: No. 347

ResultsStudyOdds ratio(95% CI)Cooper MEAMicrosulis TBEAMeyer TBEABongers TBEARomer TBEA0.98 (0.58 to 1.66)1.01 (0.47 to 2.16)1.56 (0.80 to 3.04)2.35 (1.23 to 4.50)(Excluded)Favours controlFavours treatment0.1 1 10Odds ratioFIGURE 12 Forest plot of satisfaction at 12 months’ follow-up – first-generation versus second-generation EA techniques (randomeffects model, results not pooled)48acceptable improvement in symptoms, and that84% of both groups found their treatmentacceptable. Over 80% of participants in both armswould recommend their treatment to a friend(Table 14).The Microsulis study 88 (MEA versus RB) reportsthat 98% of women undergoing MEA and 99% ofthose undergoing RB ablation were very satisfiedor satisfied and that 99% of those undergoingMEA and all those undergoing RB reported“acceptance of the operation was positive”.Bongers and colleagues 89 (TBEA versus TCRE)rated satisfaction on a four-point scale: “perfectlysatisfactory”, satisfactory”, “no treatment effect”and “complaints worsen”. At 12 months theyreport that 62% of women undergoing TBEA and53% of those undergoing TCRE were perfectlysatisfied or satisfied. At 24 months the figures were39 and 31%, respectively.Women in the trial by Meyer and colleagues 82(TBEA versus RB) were rated as “very satisfied”,“satisfied” or “not satisfied” with their treatment atall four follow-up points; 87% of women who hadundergone TBEA were reported as “very satisfied”or “satisfied” at 12 months as were 82% of thosewho had undergone RB. At 24 months, the resultswere 86 and 75%, respectively. These differenceswere not statistically significant. It should be notedthat these figures were estimated from a graphand therefore may be subject to slight inaccuracies(Table 14). At 60 months, the majority of womenfollowed up in both groups were reported to besatisfied with treatment. In addition, 22/25 womenwho had received a repeat procedure orhysterectomy by 60 months were also reported tobe satisfied with their treatment.Pellicano and colleagues 92 (TBEA versusTCRE/RB) found that satisfaction was “excellent”at 12 months for 43% of women undergoingTBEA and 24% of women undergoing TCRE andRB. These figures were 35 and 4%, respectively, at24 months. Differences between the groups werestatistically significant.Romer 83 (TBEA versus RB) states that all patientsin their trial were satisfied.Soysal and colleagues 91 (TBEA versus RB) reportthat 31% of those undergoing TBEA and 39% ofthose undergoing RB ablation were not verysatisfied. As the study reports that women wereasked if they were “very satisfied”, “satisfied” or“dissatisfied”, it is assumed that this figureincludes those in the “satisfied” and “dissatisfied”

Health Technology Assessment 2004; Vol. 8: No. 3StudyOdds ratio(95% CI)Cooper MEAMeyer TBEABongers TBEAZon-Rabelink TBEA1.40 (0.82 to 2.39)1.93 (1.05 to 3.56)1.44 (0.74 to 2.82)1.44 (0.66 to 3.15)Favours controlFavours treatment0.1 1 10Odds ratioFIGURE 13 Forest plot of satisfaction at 24 months’ follow-up – first-generation versus second-generation EA techniques (randomeffects model, results not pooled)categories, although this is not made clear.Differences between the two techniques were notstatistically significant (Table 14).Zon-Rabelink 93 (TBEA versus RB) reports that80% of women who had undergone TBEA and75% of those who had undergone RB ablationwere satisfied after 2 years. It is not stated how thiswas measured. This difference was not significant.Figures 12 and 13 illustrate satisfaction rates at 12and 24 months for first-generation versus secondgenerationEA techniques. The data points wereproduced by combining the categories for“perfectly satisfactory” and “satisfactory” in thestudy by Bongers and colleagues, 89 “satisfied” and“very satisfied” in the study by Meyer andcolleagues, 82 and the “totally and generallysatisfied” category in the study by Cooper andcolleagues. 86 The size of the data points indicatesthe relative size of each study. At 12 months, onthis dichotomous measure, Bongers andcolleagues 89 show a statistically significant effecton satisfaction in favour of TBEA, but this is notseen at 24 months. At 24 months, the Meyerstudy 82 shows satisfaction to be just significantly infavour of second-generation techniques, but this isnot seen at 12 months. Other studies do not detectsuch a difference. The study results have not beenstatistically combined owing to clinicalheterogeneity between the trials.QoLTable 15 shows various aspects of QoL reported inthe included studies, of which only two usedmeasures relating to QoL. 82,86 The trial by Cooperand colleagues 86 (MEA versus TCRE/RB) reportsthat work absence of 2 or more days wassignificantly reduced in both groups of womenwith 3% of those who had undergone MEA and6% of those undergoing TCRE and RB stillexperiencing such work absences at 12 monthspostoperation.Meyer and colleagues 82 (TBEA versus RB) report asignificant postoperative decrease in theproportion of women unable to work outside thehome at all times of follow-up, with 4% of womenin both arms unable to work outside home at36 months (see Table 15). In addition, whereastwo-thirds of women reported that HMB had asevere impact on life prior to the operation, thiswas reduced to 1% in both arms at 36 months.Differences between the groups were notsignificant. See Table 15 for more details.Only the MEA study by Cooper and colleagues 86used a QoL instrument validated in HMB, the49© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

50TABLE 15 Pre- and postoperative impact of symptoms on life – % (% ITT)Author/date Length of Intervention Unable to work 2 or more days Severe impact Moderate impact Minor impactfollow-up outside the home work absence on life on life on life(months)Preop. Postop. Preop. Postop. Preop. Postop. Preop. Postop. Preop. Postop.ResultsCooper et al., 1999 86 12 MEA – – 36 (36) 3 (3) – – – – – –TCRE/RB – – 37 (37) 7 (6) – – – – – –24 MEA – – – – – – – – – –TCRE/RB – – – – – – – – – –Microsulis, 2002 88 12 MEA – – – – – – – – – –RB – – – – – – – – – –Bongers et al., 2000 89 24 TBEA – – – – – – – – – –TCRE – – – – – – – – – –Brun et al., 2002 96 3 TBEA – – – – – – – – – –TCRE – – – – – – – – – –Meyer et al., 1998 82 12 TBEA 40 (37) 4 (4) – – 70 (66) 3 (3) – – – –RB 38 (33) 3 (2) – – 79 (67) 2 (2) – – – –24 TBEA 40 (37) 1 (1) – – – – – – – –RB 38 (36) 3 (2) – – – – – – – –36 TBEA 40 (33) 4 (4) – – 70 (66) 2 (1) 28 (28) 8 (7) 2 (1) 90 (75)RB 38 (36) 5 (4) – – 79 (67) 2 (1) 20 (20) 8 (6) 1 (1) 90 (64)60 TBEA – – – – – – – – – –RB – – – – – – – – – –Gervaise et al., 1999 90 1 TBEA – – – – – – – – – –TCRE – – – – – – – – – –24 TBEA – – – – – – – – – –TCRE – – – – – – – – – –Pellicano et al., 2002 92 3 TBEA – – – – – – – – – –TCRE/RB – – – – – – – – – –12 TBEA – – – – – – – – – –TCRE/RB – – – – – – – – – –24 TBEA – – – – – – – – – –RB – – – – – – – – – –Romer, 1998 83 12 TBEA – – – – – – – – – –RB – – – – – – – – – –Soysal et al., 2001 91 12 TBEA – – – – – – – – – –RB – – – – – – – – – –Zon-Rabelink, 2001 93 24 TBEA – – – – – – – – – –RB – – – – – – – – – –

Health Technology Assessment 2004; Vol. 8: No. 390807060Mean score50403020Preoperative MEAPreoperative TCREPostoperative MEAPostoperative TCRE100PhysicalfunctioningSocialfunctioningRole –physicalRole – Mentalemotional healthSF-36 domainEnergy/fatigue Pain General healthFIGURE 14 Mean SF-36 scores pre- and postoperation from Cooper et al., 1999 86TABLE 16 Data used in Figure 14: SF-36 scoresItem Preop. Change Postop. Preop. Change Postop.MEA postop. MEA MEA TCRE Postop. TCRE TCREPhysical functioning 84.6 0.7 85.3 82.2 2.4 84.6Social functioning 60.1 20.6 80.7 60.1 16.2 76.3Role – physical 56.5 23.9 80.4 62.9 11.3 74.2Role – emotional 61.8 17 78.8 62.6 13.7 76.3Mental health 63.6 6.3 69.9 63.8 6 69.8Energy/fatigue 44.3 12.8 57.1 43.4 12.1 55.5Pain 55.4 14.8 70.2 63.7 7.2 70.9General health 69.7 2.4 72.1 73 –2.9 70.1SF-36 (see Figure 14 and Table 16). Prior totreatment, mean scores were lower across six of theeight items than a general population of the sameage prior to treatment, and the SF-36 pain scorewas significantly lower in the MEA group than theTCRE group. Following treatment, six of the eightitems improved significantly in the MEA group, asdid seven items in the TCRE group. Analysis ofcovariance showed that the only difference betweenthe groups was on physical role, in which there wasgreater improvement with MEA than TCRE.Operation detailsTable 17 reports the results for duration ofoperations. Two studies 82,90 report on thepercentage of operations that took 50 minutes (difference significant,p < 0.05).51© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

52TABLE 17 Operation detailsAuthor/date Intervention 50 minutes Mean (SD) Mean (SD) Mean (SD) Fully Return to Return to Resumption% % operating theatre post- recovered normal work of sexual(% ITT) (% ITT) time time operative in 4 weeks: domestic (days) activity(minutes) (minutes) stay % activities (days)(% ITT) (days)ResultsCooper et al., 1999 86 MEA – – 11.4 (10.5) 20.9 (11.3) 13.4 (17.6) hours 72 (67) – – –TCRE/RB – – 15.0 (7.2)0 26.2 (8.7)0 16.7 (21.2) hours 66 (61) – – –Microsulis, 2002 88 MEA – – 3.45 (1.02) a 41.7 (25.4) b – – – – –RB – – 20.26 (15.6) a 0 50.0 (23.0) b – – – – –Bongers et al., 2000 89 TBEA – – – – – – – – –TCRE – – – – – – – – –Brun et al., 2002 96 TBEA – – – – – – – – –TCRE – – – – – – – – –Meyer et al., 1998 82 TBEA 71 (65)0 2 (2)0 – – – – – – –RB 27 (24)0 18 (14) – – – – – – –Gervaise et al., 1999 90 TBEA 100 (100) – 20.3 – – – – – –TCRE 53 (53)0 – 44.8 – – – – – –Pellicano et al., 2002 92 TBEA – – 24 (4.0) – 1.0 (0.4) days – 4.1 (±1.8) 0.7 (±0.1) 9.6 (±0.6)TCRE/RB – – 37 (6.0) – 1.3 (0.6) days – 6.2 (±3.3) 0.9 (±0.3) 9.8 (±0.7)Romer, 1998 83 TBEA – – – – – – – – –RB – – – – – – – – –Soysal et al., 2001 91 TBEA – – 11.5 (±0.8) – – – – – –RB – – 37.3 (±7.5) – – – – – –Zon-Rabelink, 2001 93 TBEA – – – – – – – – –RB – – – – – – – – –a Given as “anaesthetic time” and excluding one centre whose patients all had GA.b Given as “treatment time”.

Health Technology Assessment 2004; Vol. 8: No. 3Mean operating time is reported in four studies,although the approaches to measurementvaried. 86,90–92 A mean theatre time is also given byCooper and colleagues 86 (MEA versus TCRE/RB)In this study the mean operating time for MEAwas 11.4 minutes and for TCRE/RB it was11.5–24 minutes. However, it may be that the timereported by Gervaise (TBEA versus TCRE) asoperating time is what Cooper and colleaguesrefer to as theatre time (see Table 17). For TCREand RB, mean operating time ranges from 15.0 to44.8 minutes (median 37.3 minutes) TheMicrosulis study 88 (MEA versus RB) reports an“anaesthetic time” of 41.7 minutes for MEA and50 minutes for RB and a “treatment time” of3.45 minutes for MEA and 20.26 minutesfor RB.Differences between procedure times weresignificant in all studies at the p = 0.0001 level forCooper and colleagues 86 (MEA versus TCRE/RB)and Soysal and colleagues 91 (TBEA versus RB), at0.009 for the Microsulis study (MEA versus RB)and at the 0.05 level for the study by Gervaise andcolleagues 90 (TBEA versus TCRE). Zon-Rabelink 93(TBEA versus RB) reports that the mean operatingtime for TBEA was significantly shorter thanthat for RB (p < 0.001) but does not providethe data.Cooper and colleagues 86 (MEA versus TCRE/RB)also report on the mean postoperative stay andthe percentage of women who were fully recoveredin 4 weeks. Differences between the groups werenot statistically significant.Bongers and colleagues 89 (TBEA versus TCRE),Brun and colleagues 96 (TBEA versus RB) andRomer 83 (TBEA versus RB) do not give operatingtimes.Adverse effectsThe Microsulis study, 88 Brun and colleagues 96 andRomer 83 did not report adverse effects oftreatment. Tables 18 and 19 show intraoperativeand postoperative adverse effects reported in theother trials.Only the trial by Cooper and colleagues 86 (MEAversus TCRE/RB) reported equipment failure,which occurred in 9% of MEA operations and 2%of TCRE operations. This difference between thegroups was significant (p = 0.02). The procedurewas abandoned in 4% of both MEA and TCREprocedures. It is reported that the equipmentfailures for MEA all occurred early in the studywith a prototype microwave generator.Among all trials, only the MEA trial by Cooperand colleagues 86 and the TBEA trials by Brun andcolleagues 96 reported any intraoperative adverseeffects with second-generation techniques; in eachtrial one women was affected. Adverse effects werereported in 1% of MEAs (one blunt uterineperforation) by Cooper and colleagues 86 and in3% of TBEA procedures (one cervical burn) byBrun and colleagues. 96 TCRE and RB operationsresulted in between 0 and 27% (median 5%)intraoperative adverse effects; these included fluidoverload, cervical laceration, uterine perforationand haemorrhage (Table 18).Zon-Rabelink 93 (TBEA versus TCRE) does notgive the numbers of adverse effects occurring, butlists those experienced by women in the RB group.He also states that significantly morepostoperative pain relief was required by womenwho had undergone TBEA than those who hadundergone RB (p = 0.01), but does not give data.The recording of postoperative adverse effects maybe affected both by length of follow-up and LTFU.In both trials resulting in multiple papers 82,86 atdifferent follow-up times, an additional recordedadverse effect beyond 12 months is pregnancy. TwoTBEA studies 82,90 and the MEA study 86 reported onone pregnancy each, in all cases at 12–24 months offollow-up. No pregnancies were reported in thecontrol groups of the included trials.In the trial by Pellicano and colleagues, 92 onewoman each in the TBEA and TCRE/RB group(3%) was reported to have CIN grade one at2 years postablation procedure. This is the onlytrial reporting the outcomes of postoperativecervical smears.Zon-Rabelink 93 (TBEA versus RB) reports thatthere were no complaints in 95% of women whohad undergone TBEA and 97% of women whohad undergone RB ablation at 6 weeks of follow-up.Haemorrhage and pain are not reported in alltrials. Haemorrhage was reported after 2% of TBEAprocedures as reported by Brun and colleagues 96and after 0–12% of TCRE/RB procedures.Three studies of TBEA report postoperativeendometritis, occurring in 0–4% (median 2%)after TBEA and 1–4% (median 2%) after TCREand RB. Two studies 82,91 report postoperativehaematometra, after 0–2% of TBA procedures and1–4% of RB procedures. In addition, one study 82reports a single case of UTI in the TBEA groupand postablation sterilisation syndrome in the RBgroup (Table 18).53© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

54TABLE 18 Intraoperative adverse effects – number (%)Author/date Intervention ITT (no. Procedure Equipment Total Fluid Cervical Uterine Haemorrhage Electrolytereported abandoned failure intraoperative overload laceration/ perforation/ imbalanceon) burn lacerationResultsCooper et al., 1999 86 MEA 129 (129) 5 (4) 11 (9) 1 (1) 0 0 1 (1) 0 –TCRE/RB 134 (134) 5 (4) 3 (2) 6 (5) 0 0 1 (1) 5 (4) –Microsulis, 2002 88 MEA 215 (215) – – – – – – – –RB 107 (107) – – – – – – – –Bongers et al., 2000 89 TBEA 77 8 (10) – 0 0 – – – –TCRE 75 13 (17) – 21 (27) 20 (26) b – 1 (1) – –Brun et al., 2002 96 TBEA 21 (21) a – – 1 (3) – 1 (3) – – –TCRE 29 (29) a – – 0 – – – – –Meyer et al., 1998 82 TBEA 134 (125) – – 0 0 0 0 0 –RB 126 (114) – – 4 (3) 2 (1) 1 (1) 1 (1) 0 –Gervaise et al., 1999 90 TBEA 73 (73) – – 0 0 0 0 0 –TCRE 74 (74) – – 0 0 0 0 0 –Pellicano et al., 2002 92 TBEA 40 (46) – – 0 0 0 0 0 –TCRE/RB 42 (50) – – 8 (19) 5 (12) 1 (2) 2 (5) c 0 –Romer, 1998 83 TBEA 10 (10) – – – – – – – –RB 10 (10) – – – – – – – –Soysal et al., 2001 91 TBEA 48 (45) – – 0 0 0 0 0 –RB 48 (48) – – 5 (10) 2 (4) 1 (2) 0 0 –Zon-Rabelink, 2001 93 TBEA 77 (77) – – 0 0 0 0 0 0RB 62 (60) – – – No Yes Yes No Yesa The patient from the Brun et al. study was in fact LTFU, but it is not stated to which group this woman was allocated so ITT is not possible.b 3 (4%) >2000 ml intravasation, 20 (26%) >1000 ml intravasation.c Both of these patients had an emergency conversion to hysterectomy at the time of the procedure owing to uterine perforation.

TABLE 19 Postoperative adverse effects – number (%)© Queen’s Printer and Controller of HMSO 2004. All rights reserved.Author/date Length of Intervention Total post- Endo- UTI Haema- Urinary Fever Haem- Pain Sympto- Preg- CINfollow-up operative metritis tometra inconti- orrhage matic nancy grade 1(months) (cumulative) nence hydrosalpinxCooper et al., 1999 86 12 MEA, n = 129 4 (3) – – – – – 3 (2) 0 – – –TCRE/RB, n = 124 4 (3) – – – – – 0 3 (pelvic) (2) – – –1 (chest) (1)24 MEA, n = 120 5 (4) – – – – – – – – 1 (1) –TCRE/RB, n = 129 4 (3) – – – – – – – – – –Microsulis, 2002 88 12 MEA, n = 215 – – – – – – – – – – –RB, n = 107 – – – – – – – – – – –Bongers et al., 2000 89 24 TBEA, n = 77 – – – – – – – – – – –TCRE, n = 75 – – – – – – – – – – –Brun et al., 2002 96 3 TBEA, n = 29 – – – – – – – – – – –TCRE, n = 21 – – – – – – – – – – –Meyer et al., 1998 82 12 TBEA, n = 126 4 (3) 3 (2) 1 (1) 0 – – – – 0 –RB, n = 114 3 (3) 1 (1) 0 1 (1) – – – – 1 (1) –24 TBEA, n = 122 5 (4) – – – – – – – – 1 (1) –RB, n = 105 3 (3) – – – – – – – – 0 –36 TBEA, n = 114 5 (4) – – – – – – – – – –RB, n = 99 3 (3) – – – – – – – – – –60 TBEA, n = 61 – – – – – – – – – –RB, n = 61 – – – – – – – – – –Gervaise et al., 1999 90 24 TBEA, n = 44 1 (2) 0 – – – – – – – 1 (2)TCRE, n = 47 2 (4) 2 (4) – – – – – – – 0 –Pellicano et al., 2002 92 3 TBEA, n = 40 – – 0 – – 1 (2) 5 (12) – – – –TCRE/RB, n = 42 – – 1 (2) – – 2 (5) 4 (10) – – – –12 TBEA, n = 40 – – – – – – – – – – –TCRE/RB, n = 42 – – – – – – – – – – –24 TBEA, n = 40 – – – – – – – – – – 1 (3)TCRE/RB, n = 42 – – – – – – – – – – 1 (3)Romer, 1998 83 12 TBEA, n = 10 – – – – – – – – – –RB, n = 10 – – – – – – – – – –Soysal et al., 2001 91 12 TBEA, n = 45 3 (7) 2 (4) – 1 (2) – – – – – –RB, n = 48 3 (6) 1 (2) – 2 (4) – – – – – –Zon-Rabelink, 2001 93 12 TBEA, n = 77 – – – – – – – – – – –RB, n = 60 – – – – – – – – – – –In addition, Pellicano et al. 92 report postoperational vaginal bleeding for a mean of 7.8 days (±1) in the TCRE groups and 5.2 days (±1.8) in the TVBEA group. Visual analogue score(VAS) pain score: at discharge: TCRE 1.5 (±0.6), TBEA 1.9 (±0.3); at 3 days, TCRE 0.5 (±0.2), TBEA 0.4 (±0.1); at 7 days, TCRE 0, TBEA 0Health Technology Assessment 2004; Vol. 8: No. 355

ResultsTABLE 20 Repeat surgery – number (%) [ITT %]Author/date Length of Intervention Total Hysterectomy TCRE Otherfollow-up repeat ablation(months)surgeryCooper et al., 1999 86 12 MEA, n = 116 9 (9) [8] 8 (7) [6] a 1 (1) [1] b 1 (1) [1]TCRE/RB, n = 124 11 (9) [8] 11 (9) [8] a – 0 c24 MEA, n = 120 – – – –TCRE/RB, n = 129 – – – –Microsulis, 2002 88 12 MEA, n = 215 1 (

Health Technology Assessment 2004; Vol. 8: No. 3TABLE 21 Immediate complications reported in the MISTLETOE study – number (%)Complication Loop + ball Loop alone Ball alone TotalN = 4291 N = 3776 N = 650 N = 8717Haemorrhage 99 (2.57) 129 (3.53) 6 (0.97) 234 (2.68)Perforation 52 (1.29) 88 (2.47) 4 (0.64) 144 (1.65)Cardiovascular/respiratory 22 (0.54) 20 (0.5) 3 (0.48) 45 (0.52)Visceral burn 3 (0.07) 3 (0.08) 0 6 (0.07)Total 171 (3.98) 229 (6.06) 13 (2.00) 413 (4.74)available for follow-up are reported in the text,and shown in the table in parentheses. Data wereon an ITT basis where necessary and are shown insquare brackets in the table. For those studies withmultiple follow-up periods, the figures arecumulative for repeat procedures. Only onewoman was recorded as having a repeated secondgenerationprocedure. 86Repeat EAIn all studies, there were fewer repeat ablationsthan hysterectomy following treatment failure. At12 months, Cooper and colleagues 86 (MEA versusTCRE/RB) reported one (1%) TCRE procedure inthe intervention arm. Meyer and colleagues 82(TBEA versus RB) reported one RB procedure inthe intervention arm at 36 months. Neither ofthese studies had repeat TCRE/RB in the controlarm. By contrast, Bongers and colleagues 89 (TBEAversus TCRE) reports four (5%) repeat TCREs inthe control arm and none in the TBEA arm by24 months. At 60 months’ follow-up, Meyer andcolleagues 82 reported that two women in each ofthe intervention (TBEA 3%) and control arm (RB3%) had undergone repeat ablation.At 24 months, Gervaise and colleagues 90 (TBEAversus TCRE) report that five (7%) patients in thecontrol arm had a repeat TCRE.HysterectomyBy 12 months,

ResultsTABLE 22 Adverse effects in primary and repeat first-generation ablations 70Primary EA Second EA OR (95% CI)Adverse effect N = 800 N = 75Perforation 7 5 8.09 (2.49 to 25.88)Fluid absorption >800 dl 8 2 2.71 (0.565 to 13.00)Haemorrhage 5 0 Not calculatedTotal 20 (2.05%) 7 (9.30%) 4.01 (1.63 to 9.87)TABLE 23 Adverse effects of MEA in 1433 cases 72Type Adverse effect No. Rate/1000Major complications Visceral burn 1 0.7Minor complications Blunt perforation 4 2.6Perforation with dilator 2 1.3Endometritis 14 9.8Total 21 14.658Women having ablation by RB alone hadconsistently fewer immediate operativecomplications and fewer occasions whereemergency surgery was needed. The combinedloop and RB approach had significantly fewertotal immediate operative complications than loopalone (p < 0.00005)The overall intraoperative complication rate of4.74% in the MISTLETOE study compares wellwith the median reported adverse effects of 5% forTCRE and RB in the trials included in thisassessment.The MISTLETOE paper reports 10 deaths, ofwhich two were considered to be directly related tothe ablation procedure: one case of brain stemconing in association with malignant gliomaduring a combined procedure, and one case ofstreptococcal septicaemia 3 weeks after loopresection. Direct mortality rates were therefore2/10,000 (0.0002%) for combined procedure and3/10,000 (0.0003%) for loop alone. As these ratesare so small, it is perhaps not surprising that therelatively small trials included in this assessmentdid not record any deaths.A prospective cohort study of Canadian womenreported the rate of perioperative complications inwomen undergoing repeat ablation. 70 Data forcomplication rates following repeat ablations werenot available from the trials included in thisassessment. Eight hundred women undergoingprimary ablation and 75 women undergoingrepeat ablation by the same surgeon between 1990and 2000 were assessed. Serious complications(uterine perforation, haemorrhage and fluidabsorption) occurred in 9.3% of repeat ablationscompared with 2.05% of primary ablation(p = 0.006). Actual figures are shown in Table 22.No national audit of second-generation techniqueshas been undertaken. However, a prospectiveseries of 1433 MEA procedures (460 from one UKcentre) in 13 centres in the UK and Canada hasbeen reported. 72 The series included all patientsfrom 1994, when the first experimental procedurewas undertaken, to 1999. Only one majorcomplication (a visceral burn) was reported, givinga serious complication rate of 0.7/1000. Results areshown in Table 23.A prospective study of 296 women undergoingTBEA between 1994 and 1996 in 15 centres inCanada and Europe assessed complications ofthermal balloon ablations after 1 year; 10112 months’ data were available for 163 women. Nointraoperative complications were reported. Minorpostoperative ablations were reported as one caseof cystitis, six cases of febrile morbidity (diagnosedas low-grade endometritis), two haematometra andone hospitalisation for pain. The minorcomplication rate was therefore 3% (30/1000).A European survey of clinicians by Rogerson andDuffy reported on complications with TBEA in5800 women. 71 The study used the outcomesdescribed in the MISTLETOE study. Thesurvey achieved a 33% response rate fromgynaecologists thought to be actively usingThermachoice. Reported adverse effects are shownin Table 24.

Health Technology Assessment 2004; Vol. 8: No. 3TABLE 24 Complications with TBEA reported by Rogerson and Duffy 71Intraoperative complication Incidence (%) (n = 5859) Rate/1000Haemorrhage 0.03 0.003Uterine perforation 0.17 0.017Cardiovascular system/respiratory complication 0.02 0.002Visceral burn 0.02 0.002Equipment failure 0.2 0.02Total (excluding equipment failure) 0.23 0.023Caution should be exercised when comparingacross uncontrolled observational studies as it isnot possible to assess the existence and effect ofpossible biases. In addition, when consideringadverse effects, the definition and method of datacollection may be different for the different studies.SummaryChapter 4: Effectiveness●●●●●Two systematic reviews and 10 controlledtrials were included in the review. Thesystematic reviews were of good quality andthe controlled trials were of variable quality.Two trials were of MEA and eight of TBEAand the comparators were either TCRE orRB or combined technique.Overall, there were few significant differencesbetween the outcomes of first- and secondgenerationtechniques including bleeding,satisfaction and QoL measures and repeatsurgery rates. Significant differences werereported most often by Pellicano and colleagues,which was a relatively poor quality study.Second-generation techniques hadsignificantly shorter operating and theatretimes.There appear to be fewer perioperative adverseeffects with second-generation techniques andpostoperative effects are similar.There are no studies directly comparingsecond-generation techniques andhysterectomy and so this comparison canonly be indirectly inferred from studies offirst-generation techniques and hysterectomy.Compared with hysterectomy, TCRE and RBare quicker to perform and result in shorterhospitalisation and faster return to work.Hysterectomy results in more adverse effectsand is more expensive, although the need forretreatment leads this difference to decreaseover time. Satisfaction with hysterectomy isinitially higher, but there is no significantdifference after 2 years.Economic evaluation ofmicrowave and thermal balloonablationAssumptions used in the modelTable 25 shows the assumptions for transitionprobabilities between states, costs of procedures,discounting and utilities used in the model andtheir source. Many of these values may be subjectto uncertainty, for example the utility values usedhave been taken from the only publishedcost–utility study found to have elicited valuesfrom women with menorrhagia. However, asdiscussed in the section ‘Measuring the impact ofHMB’ (p. 4), these values may be problematic. Wehave addressed such uncertainty throughsensitivity analysis (see Table 33). Similarly,estimates of re-intervention rates are hampered byshort-term follow-up and the different ways inwhich different studies report postoperativebleeding. Again, sensitivity analysis has beenperformed to explore the impact of uncertainty inthese parameters.CostsDetails of resource use that informed thecalculation of costs in the model are shown inTable 26. As seen in Table 17, the operating andtheatre times are not clearly reported in the trialsincluded in this review. Operating times havetherefore been examined in sensitivity analysis (seeTable 33) as this is likely to impact on the overallestimated costs. Costs of managing complicationshave not been included and this is a limitation ofthe model. This may underestimate the actualcosts of procedures, hysterectomy in particular.Costs of the equipment for microwave andthermal balloon ablation are shown in Table 27.The two sets of costs for the microwave system arebased on different systems of supply. One involvespurchase of the system and the other, under whichthe majority of UK centres using MEA operate, isa placement arrangement. Under thisarrangement, the list price is £375 per treatment.59© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

ResultsTABLE 25 Assumptions used in the modelAssumptions Value Source Justification for sourceTransitionsBackground death rate (death) 0.001234 Life tables UK figures – starting age 42 yearsas given in the studies included inthis assessment, and increasing yearon yearComplications after hysterectomy 0.035 VALUE study 48 Large UK observational studyDeath after hysterectomy (direct cause) 0.00025 VALUE study 48 Large UK observational studyMedian length of complications after 2 months Clinician estimatehysterectomyLength of convalescence period post 2 months Lethaby et al., 2002 52 Mean time of return tohysterectomywork/normal activities in systematicreview of hysterectomyWaiting time – mean (median) 94 (54) days HES, 2001 73 Table 5, Q07 UK data setComplications after TCRE + RB 0.0398 MISTLETOE study 54 Large UK observational studyDeath after TCRE + RB (direct cause) 0.0002 MISTLETOE study 54 Large UK observational studyComplications after RB 0.0200 MISTLETOE study 54 Large UK observational studyDeath after RB (direct cause) 0 MISTLETOE study 54 Large UK observational studyComplications due to TCRE alone 0.0606 MISTLETOE study 54 Large UK observational studyDeath after TCRE alone 0.0003 MISTLETOE study 54 Large UK observational studyMedian length of complications following 1 month Professional estimate1st-generation techniquesComplications due to MEA 0.0007 Case series, 1433 women 72 Large UK observational studyDeath after MEA (direct cause) 0 Case series, 1433 women 72 Large UK observational studyComplications due to TBEA 0.0023 See Table 24 European survey of complicationsin 5800 womenDeath after TBEA (direct cause) 0 Adverse effect evidence Systematic review of controlledin this report (Tables 17 trial evidenceand 86)Median length of complications after 1 month Professional estimate2nd-generation techniquesTBEA treatment failure (recurrent 0.11 Gervaise data (immediate Controlled trial. Only data availablemenorrhagia) post-operative) 90 for immediate postoperative failureratesTBEA treatment failure years 2 and 3 0.1 See Table 20 RCTs in this assessmentProportion of women with recurrent 0.6 5-year follow-up of women Long-term RCT data for TCREmenorrhagia who undergo hysterectomyundergoing TCRE (vsmedical management) 43Proportion of women with recurrent 0.4 5-year follow-up of women Long-term RCT data for TCREmenorrhagia who repeat ablationundergoing TCRE (vsmedical management) 43Proportion of women with second EA 0.9 Professional estimatefailure who undergo hysterectomywithin 6 monthsComplications after repeat TCRE or Twice the Maclean-Fraser et al., 70 Comparative case series study ofRB ablation rate after 2002 and professional primary and repeat ablations. Only1st ablation estimate data on complications after repeatablation60continued

Health Technology Assessment 2004; Vol. 8: No. 3TABLE 25 Assumptions used in the model (cont’d)Assumptions Value Source Justification for sourceDeath after repeat TCRE/RB ablation 0.0003 MISTLETOE study 54 Large UK auditFirst-year return of menorrhagia post- 0.11 Effectiveness data median RCT data, best available evidenceTCRE/RB at 12 months (Table 9)Second- and third-year return of 0.1 Effectiveness data median RCT data, best available evidencemenorrhagia following TCRE/RBat 24 months (Table 9) plusrepeat surgery rate(Table 20)First-year return of menorrhagia post- 0.11 Effectiveness data median RCT data, best available evidenceTBEA/MEA at 12 months (Table 9)Second- and third-year return of 0.1 Effectiveness data median RCT data, best available evidencemenorrhagia following TBEA/MEAat 24 months (Table 9) plusrepeat surgery rate(Table 20)Discount ratesCosts 6% NICE As recommended by NICEBenefits 1.5% NICE As recommended by NICEHealth state utilitiesChronic statesMenorrhagia 0.55 Sculpher, 1998 30 Median value based on interviewswith 60 women with menorrhagiaPremenopausal following recovery from 0.9 Sculpher, 1998 30 Median value based on interviewssuccessful TCREwith 60 women with menorrhagiaPremenopausal following recovery from 0.95 Sculpher, 1998 30 Median value based on interviewshysterectomywith 60 women with menorrhagiaDead 0 Usual valueTemporary statesComplications after hysterectomy 0.55 Assumption Same as menorrhagiaHysterectomy 0.63 Assumption One-third less than recovery afterhysterectomyConvalescence after hysterectomy 0.95 Sculpher, 1998 30 Median value based on interviewswith 60 women with menorrhagiaMEA/convalescence after MEA 0.85 Sculpher, 1998 30 Convalescent states postablationassumed to be the same for alltypes of ablation. Based on theSculpher score for TCRE 30TBEA/convalescence after TBEA 0.85 Sculpher, 1998 30 Convalescent states postablationassumed to be the same for alltypes of ablation. Based on theSculpher score for TCRE 30TCRE and RB/convalescence after TCRE 0.85 Sculpher, 1998 30 Median value based on interviewsand RBwith 60 women with menorrhagia61© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

ResultsTABLE 26 Surgical management: assumptions used in the cost-effectiveness for modelProcedure Data Source JustificationAbdominal hysterectomyLength of stay (median) 4 days Local median waiting time (Mid- UK data based on all women,Devon PCT residents) and expert uncomplicated menorrhagia will beopinionshorterDay cases 0% HES 2000/01 Table 5, Q07 UK data setDuration of surgery 59 minutes Lethaby et al., 2000 52 Good quality systematic review% under GA 100% Assumed1st-generation EAWaiting time – mean (median) 79 (45) days HES, 2001, 73 Table 5, Q17 UK data setLength of stay – weighted mean 2.0 days Lethaby et al., 2000 52 Good-quality systematic reviewDay cases 60% HES, 2001, 73 Table 5, Q17 UK data setDuration of surgery – TCRE 40.9 minutes Median from effectiveness data in RCT data – best available evidencethis report (Table 17)Duration of surgery – RB 50 minutes Effectiveness data in this report RCT data – best available evidence(Table 17)Duration of surgery – TCRE/RB 31.6 minutes Median from Effectiveness data inthis report (Table 17)RCT data – best available evidence% under GA 78 Lethaby and Hickey, 2002 9 Systematic review2nd-generation EAWaiting time – mean (median) 80 (50) days HES, 2001, 73 Table 5, Q16 UK data setLength of stay – mean (median) 1.6 (1) days HES, 2001, 73 Table 5, Q16 UK data setDay cases 65% HES, 2001, 73 Table 5, Q16 UK data setDuration of surgery – MEA 31.3 minutes Effectiveness data for theatre in Median from RCT data – bestthis report (Table 17)available evidenceDuration of surgery – TBEA 18.6 minutes Effectiveness data for theatre in Median from RCT data – bestthis report (Table 17)available evidence% under GA 52 Bain et al., 2001 67 Partially randomised study of LA versusGA among 98 women in the UKFifty-one UK centres operate this arrangement inall, of which six are in Scotland (informationsupplied by Microsulis Medical). However, thecosts may be subject to other types of arrangementand this uncertainty has been addressed throughsensitivity analysis (see Table 33).Staff costs are shown in Table 28, costs of GAand LA in Table 29 and total procedure costs inTable 30.62

Health Technology Assessment 2004; Vol. 8: No. 3TABLE 27 Microwave and thermal balloon equipment costsEquipment Cost (£) Life-time Source NotesThermal balloonCavaterm control unit 3990 10 years ManufacturerCavaterm disposable balloon 280 Single use ManufacturercatheterThermachoice generator 6000 10 years Manufacturer Cost from manufacturer, life timeassumedThermachoice disposable 335–350 Single use Manufacturer The list price is £350; manufacturerballoon catheterinforms that owing to variousdiscounts, £335 is the UK averagepriceThermachoice cost of surgical 290 Per patient Manufacturer Calculated from cost given in EurosMicrowaveMEA system 39950 ManufacturerMaintenance contract for 5000 Annual ManufacturerMEA systemPlacement arrangement 375 Price per Manufacturer According to the manufacturer, thistreatmentarrangement is used by 51 UKcentres.TABLE 28 Staff costsStaff Cost/minute (£) SourceSurgeon (consultant) 0.77 Southampton University HospitalAnaesthetist (consultant) 0.77 Southampton University HospitalAnaesthetist nurse (Grade H) 0.28 Southampton University HospitalInstrument nurse (Grade G) 0.25 Southampton University HospitalTrolley nurse (Grade G) 0.25 Southampton University HospitalCirculating nurse (Grade G) 0.25 Southampton University HospitalRecovery nurse 0.25 Southampton University HospitalSenior house officer 0.29 Southampton University HospitalRegistrar 0.26 Southampton University HospitalNurse practitioner 0.28 Southampton University HospitalTABLE 29 Costs of anaesthesia, ward costsResource Cost £ SourceGA 1.08 per minute Microsulis submissionLA 7.7 per minute Microsulis submissionInpatient bed 231 per day Southampton University Hospital – estimated from own cost +50%TABLE 30 Total procedure costsProcedure Baseline price (£)Hysterectomy 2096TCRE 1110TCRE/RB 1027RB 1190MEA 942TBEA 82663© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

ResultsTABLE 31 Summary of cost–utility analysis for MEA at 10 yearsProcedure Total cost (£) Total QALYs Incremental Incremental ICERcost QALY vs MEA (£/QALY)MEA – baseline 1448470 8360.70 – – –TBEA 1323925 8360.77 124545 –0.06 TBEA dominatesTCRE 1731734 8357.03 –283264 3.67 MEA dominatesTCRE + RB 1785045 8357.99 –336574 –2.71 MEA dominatesRB 1752359 8359.92 –303889 0.78 MEA dominatesHysterectomy 2320512 8774.34 –872042 –413.63 2108QALY, quality-adjusted life-year; ICER, incremental cost-effectiveness ratio.TABLE 32 Summary of cost–utility analysis for TBEA at 10 yearsProcedure Total cost (£) Total QALYs Incremental Incremental ICERcost QALYs (£/QALY)TBEA – baseline 1323925 8360.77 – – –MEA 1448470 8360.70 –124545 0.06 TBEA dominatesTCRE 1731734 8357.03 –407809 3.73 TBEA dominatesTCRE + RB 1785045 8357.99 –461119 2.78 TBEA dominatesRB 1752359 8359.92 –428434 0.85 TBEA dominatesHysterectomy 2320512 8774.34 –996587 –413.57 241064Baseline resultsThe total costs for the modelled cohort of 1000women over 10 years are presented. Table 31 showsthe cost-effectiveness of MEA compared with eachof the other procedures and Table 32 shows thecost-effectiveness of TBEA compared with each ofthe other procedures.With MEA, similar QALYs are accrued for aslightly higher cost compared to TBEA. Comparedto TCRE, TCRE combined with RB, and RBalone, MEA accrues more QALYs and costs less.Compared to hysterectomy, MEA is cheaper, butaccrues fewer QALYs.Compared with MEA, TBEA costs slightly less andaccrues very slightly more QALYs. Compared withTCRE, TCRE combined with RB, and RB alone,TBEA costs less and accrues more QALYs.Compared with hysterectomy, TBEA costs less andaccrues fewer QALYs.Although TBEA is seen to dominate MEA, inreality absolute differences in both cost and QALYsare small and are necessarily based on an inferredcomparison. It is possible that the methods used tocalculate costs are not sensitive enough to identifysuch small differences accurately.Sensitivity analysesSensitivity analysis was used to assess the effect ofaltering input values; this is particularly importantwhere the accuracy of initial inputs is uncertain.The sensitivity of the results to changes in variousmodel parameters was examined by varying theseparameters from the base case assumption across arange of values. Parameters tested through suchsensitivity analyses together with the values usedare shown in Table 33. Each variable was variedindependently.In order to investigate the sensitivity of the modelto these various parameters, graphs showing theICER for TBEA and MEA versus each other, firstgenerationEA and hysterectomy are shown inAppendix 8.In comparing TBEA and MEA head to head,relatively small changes have greater effect, insome cases changing the direction of effect. Thissuggests that the initial findings should be treatedwith caution, particularly as we have had to rely oninferred comparison for these interventions. Costassociated with each procedure and the proceduretime of each procedure were important. Inaddition, the model is sensitive to aspects thataffect the total QALYs accrued, such as relativepercentage of women having complications,length of complications and death rate.Compared with first-generation ablation andhysterectomy, the model was found not to besensitive to the following variables:●complication rate of treatment (in either first orrepeat ablations)

Health Technology Assessment 2004; Vol. 8: No. 3TABLE 33 Inputs varied in sensitivity analysesAssumptions Values used in Source Justification for sourcesensitivityanalysesTransitionsComplications following MEA 0.0001–0.0023 Upper value based on Upper from large UK audit ofnumbers for TBEA. Lower TBEA, lower on RCTson rate in RCTsDeath following MEA – direct cause 0–0.0002 Values for EA reported in Minimum and maximum deaththis report, Table 25 rates reported for EA proceduresincluded in this reviewComplications following TBEA 0.001–0.005 Effectiveness evidence in Based on RCTs – best availablethis report, Tables 18 and evidence19Death following TBEA – direct cause 0–0.0003 Values for EA reported in Minimum and maximum deaththis report, Table 25 rates reported for all proceduresincluded in this reviewProportion of complications lasting 0.1–0.9 Authors’ assumption Values give wide range to test tomore than 1 month for TBEA/MEAsensitivityComplication rate with repeat Same rate as MacLean-Fraser et al., Minimum assumed the same as firstablation first ablation to 2002 70 and assumption ablation, upper limit based on case4 times that in series study of first and secondfirst ablationablation complication ratesFirst-year return of menorrhagia 0.05–0.02 Effectiveness data median RCT datapost-TBEA/MEA at 12 months (Table 9)Second- and third-year return of 0.05–0.2 Total return of menorrhagia Menorrhagia assumed to include allmenorrhagia post-TBEA/MEA at 3 years 21–51% (Tables 9 those reporting menorrhagia at aand 20)given follow-up plus those whohave had a repeat EA orhysterectomy in that time periodPercentage of women with 0.2–0.8 Expert opinion and Upper limit based on expertrecurrent menorrhagia receiving assumption opinion, lower limit assumedhysterectomy over repeat ablationUtilitiesMenorrhagia 0.5–0.8 Sculpher, 1998 30 and Lowest value from mean reportedassumptionin interviews with women withmenorrhagia. Upper valueestimated in comparison to otherhealth state utilitiesTBEA and MEA 0.5–0.9 Authors’ assumption Lower limit same as menorrhagiamean – varies amount ofdiscomfort and adverse effectsWell following EA 0.75–0.99 Authors’ assumption Lower limit half way betweenmenorrhagia and well, allowing forsome long-term adverse effects,upper limit close to full healthCosts (£)LA 0–100% Author’s assumption Full range of none to all proceduresunder anaestheticProportion of second-generation 0–100% Authors’ assumption Full range of none to all proceduresprocedures done in an office settingdone in an office/non-theatresettingcontinued65© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

ResultsTABLE 33 Inputs varied in sensitivity analyses (cont’d)Assumptions Values used in Source Justification for sourcesensitivityanalysesLength of hospital stay 0.5–1.0 Lower level clinician Input from clinical experience andopinion, upper level from national dataHES UK averageProcedure time 20–42 minutes This report, Table 17 Lowest and highest recordedtheatre timesEquipment costs MEA 187–562 Author’s assumption Cost ±50%Equipment costs TBEA 158–474 Authors’ assumption Cost ±50%ModelDuration of model 3–10 years Authors’ assumption66●●●●length of complication statepercentage of those being treated for recurrentmenorrhagia who are treated by hysterectomyversus repeat EAutility for menorrhagiautility for TBEA and MEA state.The model is slightly sensitive to:●●●●●●percentage recurrence of menorrhagiapostablationcost of equipment per treatmentprocedure timepercentage of operations performed underLAlength for which the model is rundeath rate as direct result of treatment.The model is highly sensitive to●utility value for ‘well’ postablation.For MEA versus TBEA, the cost of the procedureis very important in assessing incremental costeffectiveness.The length of the procedure, whichis related to theatre cost, is also important. Themodel is sensitive to the length of time for whichthe model is run. As absolute costs and QALYs forMEA and TBEA are very similar, changes in thesenumbers lead to large effects in the modeloutputs.There must be considerable uncertainty aroundthese results given that the model is sensitive tothe utility state ‘well’ and there are few data forthis parameter. In addition, the difference in costand utility between TBEA and MEA is small, sosmall changes in these change the marginal costeffectiveness.Economic analyses supplied byindustryThree economic analyses were submitted to NICEby industry sponsors of MEA and TBEA:●●●SummaryChapter 4: Cost-effectiveness●●●●The economic model suggests that secondgenerationtechniques are more cost-effectivethan first-generation techniques of EA forHMB.The model is sensitive to utility values afterrecovery around which there is considerableuncertainty.Indirect comparisons should be viewed withcaution. However, there appears to be littledifference in costs or utilities between TBEAand MEA and small changes in these affectthe relative cost-effectiveness.Both TBEA and MEA appear to be less costlythan hysterectomy, although the latter resultsin more QALYs.a cost–utility analysis of MEA submitted byMicrosulis Medicalcost minimisation and cost-effectivenessanalyses of the Thermachoice TBEA devicesubmitted by Gynecarea cost-effectiveness analysis of the CavatermTBEA device submitted by Wallesten Medical.The analyses are of variable quality. Details of theappraisals of each analysis, carried out within theframeworks proposed by Drummond andcolleagues 75 and Sculpher and colleagues, 74 aregiven in Appendix 8.

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.PatientpopulationRB + TCRETCRERBTBEAAComplicationsNot deadNo complicationsDead(repeat structure from Node A)(repeat structure from Node A)(repeat structure from Node A)ConvalescenceConvalescenceRepeat EAHysterectomyPost-convalescence(repeat structurefrom Node B)MEA(repeat structure from Node A)FIGURE 15 Microsulis model structureHysterectomyDeadNot deadBComplicationsNo complicationsConvalescenceConvalescencePost-convalescencePost-convalescenceHealth Technology Assessment 2004; Vol. 8: No. 367

ResultsTABLE 34 Total discounted costs and QALYs by procedure and cost-effectiveness at 5 years: Microsulis modelProcedure Total costs (£) Total QALYs Incremental Incremental ICER (MEA)costs (vs MEA) QALY (vs(£) MEA)MEA 1238 3.76 – – –TBEA 1611 2.97 373 –0.79 DominatedRB 1550 3.54 312 –0.21 DominatedRB + TCRE 1441 3.48 203 –0.28 DominatedTCRE 2032 3.56 793 –0.20 DominatedHysterectomy 2728 4.08 1489 0.32 £459468Microsulis modelThe Microsulis model was carried out as an‘independent analysis’ by the York HealthEconomics Consortium. The model structure is ofhigh quality, and includes comparisons between allthe options addressed in this assessment. Themodel is a decision-tree design (Figure 15), buthandles the time to events by weighting the QALYcalculation associated with each possible paththrough the tree. The design allows preciseaccount to be taken of time spent withcomplications of the procedures. A 5-year timehorizon is taken, justified on the grounds thatalmost all repeat procedures would be carried outwithin this period in the event of initial treatmentfailure. The increasing risk of second operation ismodelled using a logarithmic function. A singlerepeat procedure is permitted. For MEA this isassumed to be TCRE + RB whereas for otherablation techniques the original option is repeated(i.e. TCRE, TCRE + RB, RB or TBEA). Thesources for estimates used in the model arepredominantly taken from the literature. A rangeof one- and two-way sensitivity analyses and aMonte Carlo simulation, in which all parameterestimates are varied (sampling from triangulardistributions), were carried out.The Microsulis submission concludes that MEA isa cost-saving treatment. Total discounted costs at5 years are estimated at £1238. Cost savings of14–55% over other treatments are estimated. Totaldiscounted costs and QALYs according to theMicrosulis model are shown in Table 34.Costs of complications, which differ according totechnology, are listed but methods for theirestimation are not detailed. However, ascomplication rates are low for all procedures, thisis unlikely to have a major effect on the overallfindings.A key parameter determining difference incost–utility is the utility weight attached to thehealth state of post-convalescence. This isestimated as being 0.86 following hysterectomy,0.73 for TCRE and TCRE + RB, 0.74 followingRB, 0.79 following MEA and 0.57 following TBEA.Methods of calculating these values are describedin Appendix 9. Counter-intuitively, the utilityweights for post-convalescent states after alltechnologies except hysterectomy and MEAappear to be lower than the convalescent states.The utility weight associated with HMB is 0.50.Sources for the estimates of repeat operation arenot detailed. Comparison with the repeat surgeryrates reported in the available comparative trialsof EA techniques suggests that these are takenfrom studies not included in the systematic reviewreported in this assessment. The repeat surgeryrate is important as a determinant of the overallcost of ablation procedures. The repeathysterectomy rate is important as the postconvalescentstate following hysterectomy carries ahigher utility weight than the health statefollowing other operations. A higher repeat ratefor one ablation technique will therefore lead tomore time spent in a health state valued morehighly than that experienced by women opting foralternative ablation techniques.The sensitivity analysis included a scenario inwhich post-convalescent utilities were assumed tobe equal. MEA no longer dominated TCRE, whichcould yield additional QALYs at additional cost of£35,000. Hysterectomy provides additional QALYsat additional cost at all levels of post-convalescentutility weight, with a maximum ICER of £35,213per QALY when this parameter is at its lowestlevel assumed (0.73).The incremental cost-effectiveness of hysterectomyis estimated at a level that has been considered bymany decision makers as representing acceptablevalue for money under most assumptions. Thismay be related to the time horizon of the model.This option results in women entering the healthstate with highest value and spending most time init. Even under the assumption that all post-

Health Technology Assessment 2004; Vol. 8: No. 3convalescent states have the same utility,hysterectomy has implicit advantage as there is noprobability of HMB or the disutility associatedwith further procedures. However, this assumesthat women prefer amenorrhoea to eumenorrhoeaor lighter menstrual loss.Following sensitivity analysis and Monte Carlosimulation, it is concluded that MEA wouldcontinue to dominate RB and TCRE in over 95%of cases and that hysterectomy would continue toyield additional benefits for extra cost. The costeffectivenessof hysterectomy over MEA is subjectto considerable uncertainty with a range from£2000 to over £130,000 per QALY. MEA is shownto dominate TBEA under almost all scenariosconsidered, although Monte Carlo simulationshowed that this may not be the case in as many as95% of circumstances, and therefore the costeffectivenesslevels of MEA and TBEA are notdissimilar.Thermachoice modelThe industry submission from Gynecare providestwo pieces of evidence regarding the economics ofThermachoice:1. a cost analysis of Thermachoice versus TCREand vaginal hysterectomy2. a crude cost-effectiveness analysis based on costsrequired to achieve several outcomes of interest:amenorrhoea, eumenorrhoea, satisfaction, andavoidance of surgical re-intervention.Cost analysisThe cost analysis is not a complete economicanalysis, as is acknowledged by the industrysubmission to NICE. The study, which has not yetbeen published elsewhere, was carried out in1995–7 in Paris, based on 147 people undergoingthermal ablation (n = 47), hysteroscopicelectroresection (n = 50) and vaginal hysterectomy(n = 50). Limited methodological details arereported and it is therefore difficult to determinethe usefulness of the study in judging the costsand, relatedly, cost-effectiveness of Thermachoice.The analysis is restricted to in-hospital costsaccruing to each technology, with differences intime in operating theatre accounting for almost allthe difference in technology costs. The resultssuggest that vaginal hysterectomy has a highercost than thermal ablation or TCRE (€2799 versus€1424 and €1508, respectively).Although the study is reported as a micro-costingstudy, methods are not clearly reported and it istherefore impossible to judge, comprehensively,the validity of the results. Particularly importantissues that cannot be addressed include thefollowing:●●●●●the methods of calculating resourceconsumption are not reportedthe completeness of resource use ascertainmentand how missing data were handled are notreportedpotential uncertainty in the estimates has notbeen addressedthe base year for cost estimates is not reportedthe methods for allocating overheads is notreported and overheads are not applied to thecosts of follow-up or of ward-based care.These methodological limitations, which may beaddressed by more comprehensive reporting inthe final published version of this study, make itdifficult to comment on the results. However, thefollowing observations can be made on thereported data:●●●●●●The summary measures are not defined(assumed to be means) and no measures ofspread in the data are reported.Resource use and costs in France are likely to bedifferent from those in the UK, limiting theapplicability of the findings to the UK. Forexample, the cost of a hospital bed day isconsiderably lower than in the UK. In addition,vaginal hysterectomy makes up a minority ofUK hysterectomies and the length of operation,length of hospital stay and complication ratesare different to those for abdominalhysterectomy.The assumption that there are no overnight staysas a result of complications arising from EA mayreflect the experience of the small populationstudied. However, such overnight stays mayoccur and therefore the reported cost differencemay be biased in favour of Thermachoice,although by an uncertain amountThe analysis is predominantly driven bydifference in time spent in operating theatre,calculated per minute for a range ofprofessionals. This unit of measurement doesnot reflect the opportunity cost of the resources.The analysis does not include priorhysteroscopy in patients as part of the work-upfor Thermachoice.Five follow-up visits are recorded for womenafter hysterectomy, compared with one in eachof the ablation techniques. The number offollow-up visits seems high, particularly forvaginal hysterectomy. If the number of follow-69© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Results1st-line 2nd-line 3rd-line 4th-lineHysterectomyOral therapy1st-generation EA2nd-generation EARepeat ablationHysterectomyHysterectomyNo further treatmentNo further treatmentFIGURE 16 Flow chart of treatment pathway considered in the Cavaterm model70up visits in the UK is less than that reported,then the difference in costs betweenhysterectomy and ablation will have beenoverestimated. However, since the costs offollow-up are based only on surgeon’s time (andmethods for calculating this are not given), theamount remains very small.Cost-effectiveness analysisThe analysis is acknowledged in the industrysubmission to NICE to be simplistic.Thermachoice TBEA is compared with TCRE andhysterectomy at 3 years. It has been appraisedusing the framework by Drummond andcolleagues 75 and this is shown in Appendix 8. Theanalysis reports the following outcomes:●●●●Cost per additional case of amenorrhoea. TheICERs for TCRE and hysterectomy, comparedwith Thermachoice TBEA, are €1736 and€1378, respectively.Cost per additional case of eumenorrhoea orless. The ICERs for TCRE and hysterectomy,compared with Thermachoice TBEA, are€19,789 and €16,751, respectively.Cost per reintervention case avoided.Thermachoice TBEA dominates TCRE. TheICER for hysterectomy, compared withThermachoice, is estimated as €16,994.Cost per additional satisfied patient. TheICERs for TCRE and hysterectomy, comparedwith Thermachoice TBEA, are €14,135 and€26,650, respectively.The analysis has a number of weaknesses inaddition to those reported for the cost analysis onwhich the cost-effectiveness estimates are based.The model does not allow the timing of events tobe taken into account within the overall timeframeof the analysis. The outcome measures used do notallow the disbenefits of treatments to be taken intoaccount, for example, adverse effects and thedifferent times to convalesce following hysterectomyand ablation procedures. Importantly, no account istaken of the uncertainty in the probabilities ofoutcome or costs in the analysis and so no estimateof the likelihood of the estimates of costeffectivenessbeing achieved in practice is possible.Cavaterm modelThe Cavaterm model is a decision-tree model witha 3-year time horizon. The comparisons arebetween hysterectomy and first-generation andsecond-generation EA. Use of the two types ofTBEA equipment is considered separately in theanalysis. Since the economic evaluation uses adecision analytic approach, its quality isconsidered in detail using the frameworkproposed by Sculpher and colleagues. 74 Thisappraisal is reported in detail in Appendix 9.The estimates for effectiveness in the Cavatermmodel are based on a meta-analysis of all reportedstudies of Cavaterm. The authors suggest that theinclusion of a large number of patients (over 2000)from over 30 trials will “override anomalies andexperimental differences” and that this approachis preferable to restricting effectiveness data tothat reported from RCTs. This position is open todebate: the quality of a meta-analysis depends onthe quality of the studies that it includes and willbe valid only if there is not significant clinical andstatistical heterogeneity between the includedstudies. The Cavaterm analysis includes asensitivity analysis in which only data from RCTsare included, noting that no RCTs have shown asignificant difference in the effectiveness of firstand second-generation techniques, although theprobability of a type II error remains.The effectiveness analysis is flawed in that it doesnot take account of the different timing of the

Health Technology Assessment 2004; Vol. 8: No. 3underlying trials with respect to treatment failure.Cavaterm is assumed to be successful in 90% ofcases (100 minus the repeat procedure rate), basedon data up to 1 year, whereas Thermachoice issuccessful in only 83.3% of cases, based on aweighted mean of data up to 3 years. Sincetreatment failure appears to be time dependent,the analysis appears to be biased in favour ofCavaterm.The model estimates that Cavaterm is the mostcost-effective option based on cost per treatmentsuccess (based on RCT data) of £767 versus £828for Thermachoice, £865 for first-generation EAand £2050 for hysterectomy. Sensitivity analysisidentified initial procedure cost and failure rate asimportant sources of uncertainty. The results ofone-way sensitivity analyses incorporating a widerange of values for these parameters suggest thatCavaterm would produce a lower cost pertreatment success at failure rates of up to 74% oran initial procedure cost of less than £1910. Thecorresponding thresholds for Cavaterm versusTCRE were estimated as 18% and £800.The analysis also reports potential savings fromCavaterm on resource use (operating theatre time,anaesthetics, length of hospital stay), mortality andlabour market productivity (based on reducedconvalescence time). In these analyses, the impactof Cavaterm is considered as a replacement for allhysterectomies currently performed for HMB.This is not a realistic scenario given the evidencefor patient preferences in a proportion of womenseeking treatment for HMB.Comparison between costeffectivenessanalysesThe four models included in the NICE appraisalprocess for EA differ in design, inputs and,consequently, outputs.The models of TBEA alone, produced by themanufacturers, do not estimate cost–utility. TheThermachoice analysis shows that TBEA is likelyto have lower surgical costs than hysterectomy orTCRE. The cost-effectiveness analysis is relativelysimplistic and suggests that additional benefitsmay be realised with comparator treatments, butat additional cost. The model has significantmethodological shortcomings.The Cavaterm analysis is derived from a decisiontreemodel run over 3 years. Results suggest thattreatment success using Cavaterm would cost lessthan with all alternatives, and considerably lessthan with hysterectomy. Considerable resourcesavings are postulated, as are the avoidance ofsome deaths from hysterectomy and a reducedburden of morbidity through reducedcomplications. Uncertainty is addressed throughthe use of limited sensitivity analysis and MonteCarlo simulation. The distributions chosen for theMonte Carlo simulations in this, and theMicrosulis model, are simple and may notadequately model the uncertainty in theparameters concerned.The Microsulis analysis is the most sophisticated ofthose submitted by industry sponsors to NICE.Although a decision-tree design, the model takesaccount of the timing of events and allows forincreasing risk of repeat procedures over time.Structurally, this is the most robust of the modelssupplied by industry sponsors. Importantly,only this model, and that constructed by theauthors of this assessment, provide estimates forcost–utility.The Microsulis model concludes that MEA is likelyto be more effective and less costly than allalternatives except hysterectomy under most of theassumptions modelled. The model incorporatesuncertainty in the parameter values throughsensitivity analysis and Monte Carlo simulation. Inthis respect, the method provides somereassurance of the robustness of the results.However, the model has a number of weaknesses,arising mainly from the quality of the data used toinform all cost-effectiveness analyses in this area.In particular, the available utility estimates and theway in which they are used in the model may giverise to some concern about the validity ofestimates of cost-effectiveness.Table 35 highlights some of the key differencesbetween the modelling studies of EA. Appendix 10provides a more detailed comparison of thedifferences between parameters included in themodels. Only the study carried out for Microsulisprovided a very detailed breakdown of individualcost elements that informed their procedure costsand these have been omitted from the table forthe sake of brevity. All Thermachoice figures wereprovided in Euros and have been converted topounds sterling based on €1 = £0.635 based onconversion rate in December 2002.The results of the cost-effectiveness analyses vary.This is to be expected given the differences inmodelling approaches and the complexity of theanalyses. All models show that EA is less resourceintensive than hysterectomy. There is therefore apotential for resource savings arising from more71© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

ResultsTABLE 35 Comparison between four economic analyses of EA techniquesPenTAG Microsulis Cavaterm ThermachoiceType of model State transition Decision tree Decision tree (a) Cost analysis(Markov)(b) Simple cost-effectiveness analysisOutput Cost per QALY Cost per QALY Cost per Cost per:treatmentsuccessTime horizon 10 years 5 years 3 years 3 yearsProcedure costs:Hysterectomy £2096 £2644 £2050 £1777TCRE £1110 £1129 £593 £958Cavaterm £826 £712 £584 –Thermachoice £826 £712 £905 £904MEA £942 £674 £793 –Additional case of amenorrhoeaAdditional case of eumenorrhoea or lessper reintervention rate avoidedAdditional satisfied patientProbability ofhysterectomy:After TBEA 0.248 (year 5) 0.321 (year 5) – 0.077 (year 3 Thermachoice)0.0595 (year 3 Cavaterm)After MEA 0.248 (year 5) 0.208 (year 5) – 0.0252After RB 0.248 (year 5) 0.368 (year 5) – 0.065–0.195Utility values:Convalescence after 0.8 0.76 N/A N/ATBEAConvalescence after 0.8 0.76MEAConvalescence after 0.8 0.76TCRE or RB orTCRE + RBConvalescence after 0.63 0.74hysterectomyPost-convalescence 0.9 0.57after TBEAPost-convalescence 0.9 0.79after MEAPost-convalescence 0.9 0.73/0.74after TCRE or RBor TCRE + RBPost-convalescence 0.95 0.86after hysterectomyUtility in 0.55 0.5menorrhagia72widespread use of EA as first-line surgicalmanagement in cases where there is not a strongpreference for amenorrhoea. However, the size ofany savings remains uncertain owing to thedifficulty in estimating costs accurately.Based on the available evidence, secondgenerationEA techniques appear to offeradvantages over first-generation techniques interms of value for money to the NHS. Analysessuggest that cost advantages may be accompaniedby effectiveness and safety gains, leading to thedominance of second-generation techniques inboth our analysis and that submitted byMicrosulis. However, the differences in both costsand effects are not large and are subject toconsiderable uncertainty. A major source ofuncertainty in estimating cost–utility is the valuethat should be placed on the relevant healthstates.Although some of the analyses submitted toNICE suggest a difference between secondgenerationtechniques, decision-makers shouldbear in mind that the evidence base for clinicaleffectiveness in this area is small, and in some

Health Technology Assessment 2004; Vol. 8: No. 3respects very weak, depending on indirectcomparisons. Any consideration of the costeffectivenessbetween second-generationalternatives is further complicated by limiteddetailed data on costs during the entire clinicalcourse of a patient and should therefore be viewedwith great caution.Impact on NHS budgetThe economic models supplied by this assessmentteam and by industry assess the relative cost perQALY of each treatment, assuming that a womanwith HMB may follow any treatment path.However, the impact of second-generation EAtechniques on the NHS budget will depend on anumber of factors, such as:●SummaryChapter 4: Cost-effectivenessinformation supplied by industry●●●●●The quality of economic analyses submittedby industry is variable and results areuncertain.Only one provides a cost–utility analysis.All find that second-generation EAtechniques offer value for money comparedwith first-generation EA techniques. The sizeof the savings is uncertain owing to thedifficulty in estimating costs accurately.Each industry submission found their ownproduct to be the cheapest or the most costeffectivetreatment.The only other cost–utility analysis alsofound their model to be very sensitive toutility values and there is uncertainty aroundthis value.Women’s preferences for different treatmentsoffered, which will depend on an individual’sdesire for such aspects as amenorrhoeaas an outcome and avoidance of majorsurgery.●●The number of women with HMB who areeligible for each treatment (for example, largerand abnormal uteri are a contraindication forTBEA and thin Caesarean scars are acontraindication for MEA).The existing diffusion of the technologies in theUK (for example, the number of surgeonsperforming TCRE/RB ablation and the numberof centres that have second-generation ablationequipment).There are currently nearly 26,000 hysterectomiesin the UK for HMB and a further 16,000 EAs, ofwhich about 2000 are second-generationtechniques (see the section ‘Surgical treatment’,p. 8). Table 36 below shows the effects of changingthe balance of the current 42,000 surgicalprocedures for women with HMB. The cost offirst-generation EA is the cost of TCRE combinedwith RB as this is the most usual technique in theUK and the cost of hysterectomy is based onabdominal hysterectomy as this accounts for 80%of hysterectomies undertaken in the UK. Initialcosts have been calculated assuming secondgenerationablation is equally divided betweenTBEA and MEA.If hysterectomies were replaced by EA, overallcosts would be reduced. If half were replaced byfirst-generation techniques, costs would bereduced by £13,546,000 (Table 37). If half of allhysterectomies were replaced by secondgenerationtechniques (equally split between thetechnologies) costs would be reduced by£15,405,000 (Table 38).If all first-generation techniques were replaced bysecond-generation techniques, a saving of£2,002,000 would be made (Table 39).If all hysterectomies were replaced by EA, costsavings would be £28,951,000 if half went to firstgenerationtechniques and the remaining halfwere equally split between second-generationTABLE 36 Estimate of current costs to the NHS of surgical procedures for HMBProcedure Cost per procedure (£) No. of procedures Total cost (£)Hysterectomy 2069 26000 53794000TCRE/RB 1027 14000 14378000MEA 942 1000 942000TBEA 826 1000 826000Total 6994000073© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Resultstechniques (Table 40), and £32,812,000 if all werereplaced by second-generation techniques(Table 41). It is unlikely, however, that allhysterectomies for HMB could be replaced by EAas some women will prefer this treatment or it willbe the only available option.The largest cost savings are therefore to be madethrough replacing some hysterectomies for HMBwith EA. Although replacing current levels of firstgenerationablation with second-generationablation techniques also results in savings, theseare smaller.TABLE 37 Costs to NHS if half of current hysterectomies were replaced by first-generation EAProcedure Cost per procedure (£) No. of procedures Total cost (£)Hysterectomy 2069 13000 26897000TCRE/RB 1027 27000 27729000MEA 942 1000 942000TBEA 826 1000 826000Total 56394000TABLE 38 Costs to the NHS if half of current hysterectomies were replaced by second-generation EAProcedure Cost per procedure (£) No. of procedures Total cost (£)Hysterectomy 2069 13000 26897000TCRE/RB 1027 14000 14378000MEA 942 7500 7065000TBEA 826 7500 6195000Total 54535000TABLE 39 Costs to the NHS if first-generation techniques were replaced by second-generation techniquesProcedure Cost per procedure (£) No. of procedures Total cost (£)Hysterectomy 2069 26000 53794000TCRE/RB 1027 0 0MEA 942 8000 7536000TBEA 826 8000 6608000Total 67938000TABLE 40 Costs to the NHS if all hysterectomies were replaced by EAProcedure Cost per procedure (£) No. of procedures Total cost (£)Hysterectomy 2069 0 0TCRE/RB 1027 27000 27729000MEA 942 7500 7065000TBEA 826 7500 6195000Total 40989000TABLE 41 Costs to the NHS if all hysterectomies were replaced by second-generation EAProcedure Cost per procedure (£) No. of procedures Total cost (£)74Hysterectomy 2069 0 0TCRE/RB 1027 0 0MEA 942 21000 19782000TBEA 826 21000 17346000Total 37128000

Health Technology Assessment 2004; Vol. 8: No. 3Chapter 5DiscussionMain resultsHMB is a common condition that results in aconsiderable burden of ill health among women.Surgical intervention is frequently soughtfollowing failure of medical intervention, and arange of options are now available.Hysterectomy is an established and effectivetreatment for HMB. However, it is more expensivethan newer alternatives, is a more complexprocedure and may result in more seriouscomplications. Although the guarantee ofamenorrhoea as a treatment outcome may bepreferred by some women, this may notcompensate others for having to undergo majorsurgery with its associated risks and recovery timeor for the loss of their womb. Because of this, newminimally invasive surgical techniques have beendeveloped.Clinical effectivenessIn this assessment we have carried out a systematicreview of the effectiveness of MEA and TBEAagainst first-generation EA and hysterectomy. Thisincluded nine trials, of which eight were RCTs.However, there are no studies comparing MEA orTBEA to hysterectomy, so effectiveness and costeffectivenesscompared with hysterectomy havehad to be inferred through indirect comparison. Agood-quality systematic review of first-generationtechniques compared with hysterectomy showedthat hysterectomy was more effective (as measuredby improvement in HMB and patient satisfaction),but was associated with greater consumption ofhealthcare resources and more adverse effects.Satisfaction rates and effectiveness with firstgenerationtechniques and hysterectomy were highand the reviewers concluded that first-generationtechniques are an alternative surgical treatmentfor HMB.The systematic review carried out for thisassessment included 10 studies comparing MEA(two studies) and TBEA (eight studies) with firstgenerationablation techniques. Duration offollow-up was limited (range 3–60 months, median24 months). One trial included 5-year follow-upbut with 46% LTFU. Overall, the quality of theRCTs is moderate, and limited by the impossibilityof blinding operators and subjects. All studieshave some methodological limitations. The trialsof MEA included more participants than those ofTBEA and were of higher quality and greaterapplicability to the UK.The included studies of MEA and TBEA did notshow a significant difference betweenamenorrhoea rates after first-generation comparedwith second-generation techniques. Only onestudy showed a first-generation technique (RB) tobe significantly superior to a second-generationtechnique (TBEA) for the outcome ofamenorrhoea measured at 2 years, although thisstudy had much loss to follow-up. The medianproportion of women with the outcome ofamenorrhoea appears higher among those treatedwith MEA (46%) than those with TBEA (14%),although the ranges overlap (MEA 36–55%, TBEA10–40%) and the amenorrhoea rates in the MEAtrials were also higher for the control group. Nodifferences in amount or pattern of continuingmenstrual loss were shown in studies thatexamined these outcomes. No differences weredemonstrated for dysmenorrhoea or PMSsymptoms in the included studies.Patient satisfaction is reported in eight of the 10included studies and was high in all cases, despitedifferences in methods of outcome measurement,and showed no difference in satisfaction withdifferent technologies in most of the comparisons.Two studies show a significant difference insatisfaction favouring TBEA over control whencategories of satisfaction are collapsed into adichotomous variable (satisfied/not satisfied).However, in both cases this was seen only at onepoint of follow-up and not all follow-up points (at12 months but not at 24 months, for example).One of these studies was not randomised and theother was not of high quality.One study each of MEA and TBEA investigatedeffects on QoL. One MEA study, by Cooperand colleagues, 86 used the SF-36 outcome measureand showed improvements across the majority ofdomains for both MEA and TCRE over baseline.Only one comparison between groups wassignificant in an analysis of covariance: physicalrole was significantly improved in the TCRE group75© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Discussion76compared with the MEA group. The clinicalsignificance of this isolated finding is uncertain.Meyer and colleagues 82 investigated QoL using aglobal question of impact and found no significantdifference between TBEA and RB. Both first- andsecond-generation ablation techniques have apositive impact on ability to work/pursue normalactivities, although neither study that examinedthis outcome showed a difference betweentechniques.All studies reporting operating time showed thatsecond-generation techniques require significantlyless operating or theatre time than first-generationtechniques. Differences in approaches to definingthe time of interest make interpretation of theresults difficult and preclude pooling the results ofindividual studies. Whether the difference in timeto complete the procedure would be sufficient topermit staff redeployment for other purposes ispossible but not certain. No differences in lengthof hospital stay have been shown. Equipmentfailure was reported in only one trial (Cooper andcolleagues 86 ) and was significantly more frequentwith MEA (9%) than TCRE (2%). However, thistrial used a prototype machine and the samecentre has since undertaken nearly 1000 furtherMEA treatments with no further equipment failure(Cooper K, Aberdeen Royal Infirmary: personalcommunication, 2002).The adverse effect profiles of second- and firstgenerationablation techniques reported in RCTsare similar at around 3–4% overall. Adverse effectsinclude uterine perforation, haemorrhage, pain,haematometra, post-tubal sterilisation syndrome,endometritis and pregnancy. Second-generationtechniques were associated with fewerintraoperative complications in the RCTs (1 versus5% for MEA versus TCRE and 0 versus 3% forTBEA versus RB) and are not prone to theproblem of fluid overload – a potentially seriouscomplication possible with hysteroscopictechniques. The small size of RCTs limits statisticalpower to demonstrate whether such differencesare significant. Two large uncontrolledobservational studies of MEA and TBEA providefurther evidence for low rates of complications.Repeat surgery rates provide some indication oftreatment failure. Reoperation rates appear toincrease with time. In the longest duration study,25% of the group allocated to TBEA and 16% ofthose allocated to RB in the trial by Meyer andcolleagues 82 had had repeat surgery by 5 years offollow-up. This figure is based on the mostconservative estimate of success – ITT figures are11 and 7%, respectively. Most women who neededfurther surgery had hysterectomy. Adverse effectrates in repeat ablations may be higher than whenablation is the primary procedure.Costs and cost-effectivenessThe costs of MEA and TBEA procedures aresimilar, although the methods used to assess thesecosts may not be sensitive enough to measure sucha small difference precisely. MEA was found to beslightly more expensive at £942 per treatmentcompared with £826 for TBEA. Compared withcombined TCRE and RB, which is the mostcommon type of first-generation EA in the UK,both methods are cheaper, by £85 for MEA and by£201 for TBEA per treatment. Operation time isimportant and the data available were unclear, butthis was examined in sensitivity analysis.The cost-effectiveness analysis necessarily dependson inferred comparisons – between MEA andTBEA and between both second-generationtechniques and hysterectomy. Such comparisonsare prone to bias and confounding and should beviewed with caution. However, in the absence ofdirect evidence, such analyses are deemednecessary by decision-makers. The cost–utilityanalysis carried out for this assessment suggeststhat TBEA may be slightly less costly and veryslightly more effective than MEA at 10 years,although differences in costs and utilities are verysmall and subject to considerable uncertainty.Both second-generation techniques similarlydominate TCRE, RB and TCRE/RB combined.Hysterectomy yields additional benefits foradditional cost, with cost–utility ratios of around£2400 per QALY against both MEA and TBEA.These findings are subject to considerableuncertainty. In particular, the absence of evidenceof clinical benefit between second-generationoptions and between first- and second-generationtechniques suggests that these results should betreated with great caution and may beinsufficiently robust to guide highly specific policydecisions.Assumptions, limitations anduncertaintiesQuality of studiesThe quality of the studies was variable, asdiscussed in the section ‘Quality assessment ofRCTs’ (p. 29). This may limit the validity of thefindings. In addition, the included studiescontained varied populations – women withmenorrhagia as measured in different ways,

Health Technology Assessment 2004; Vol. 8: No. 3inclusion or exclusion of women with fibroids, theinclusion of women who were postmenopausal inone study and two studies that did not give detailsabout the included population. The comparatorwas either TCRE alone, RB alone or TCRE andRB combined, and these procedures may not havebeen consistent among the patients in the controlgroups of some studies. RB and TCRE are knownto have different adverse effect rates, as shown inthe MISTLETOE study (Table 21). As a result, nometa-analyses were possible. The studypopulations and techniques may not reflect clinicalpractice in the UK.Both MEA trials use GnRH as a thinning agent forall participants. However, the TBEA trials vary intheir approach to prethinning of theendometrium. While no chemical prethinning isadvised by the manufacturers, two trials usedGnRH in both arms of the study. Three used apreoperative D&C for both arms, two did notreport any prethinning and one used aprethinning agent only in the control arm. It isnot known what effect prethinning has on theeffectiveness of second-generation EA. Asystematic review of prethinning agents in firstgenerationEA found that prethinning improvedthe operating conditions for the surgeon andshort-term clinical outcomes, although the effecton amenorrhoea and repeat surgery decreasedover time. 55Outcome measuresAs outlined in the section ‘Measurement of bloodloss’ (p. 3), outcome measurement in HMB isproblematic. It is not clear which outcome shouldbe considered as the most important in assessingthe success of EA techniques given preferences fortype of treatment and outcome. Whileamenorrhoea is an objective measure, it isarguably not the most appropriate measure forwomen who wish treatment to lessen menstrualbleeding but do not necessarily requiremenstruation to be eliminated. In clinical practice,where women are offered a choice of treatment,women who privilege amenorrhoea as anoutcome may prefer to have a hysterectomy fromthe start.As detailed in the section ‘Measurement of bloodloss’ (p. 3), there are a number of methods formeasuring actual menstrual blood volume.However, these are rarely used in routine clinicalpractice and other measures are not usedconsistently across the studies. Women who do nothave clinical HMB but subjectively regard theirbleeding as unacceptably heavy are likely to be lesssatisfied with their treatment for HMB. 17 Thosetrials included in this review that have stringentlymeasured HMB as an inclusion criterion forwomen entering the trial may show higher successrates than will be seen in normal clinical practice.Only the MEA trial by Cooper and colleagues 86used self-defined HMB as an entry criterion,which mirrors clinical practice in the UK.Those trials using the PBAC method ofmeasuring HMB have different levels for inclusionof women, as well as for defining success oftreatment.As the major effect of HMB on sufferers isdecreased QoL, this is an important outcomemeasure. Of the included studies, only Cooperand colleagues 86 used a recognised QoL measure(the SF-36) and no trials used a condition-specificmeasure of QoL. The validity of using genericmeasures of QoL alone in studies of HMB hasbeen questioned (see the section ‘Measuring theimpact of HMB’, p. 4). A range of surrogatemeasures of impact on QoL, such as ability towork outside the home or impact on life havebeen used. Satisfaction, another importantpatient-relevant outcome measure, is measuredon different scales in the studies and no clearreports of the method of obtaining these dataare given. It is difficult to draw conclusionsfrom an outcome such as satisfaction, which isrelated both to processes and outcomes oftreatment. For both QoL and satisfaction, thevariety of measures used makes comparisonbetween studies difficult.Availability of evidenceOnly two studies of MEA and eight of TBEA wereidentified that met the inclusion criteria.There is also little evidence in the literature forlong-term follow-up of women who haveundergone MEA or TBEA for HMB. Therefore,longer term rates of recurrent HMB, andassociated further surgery, are not known. It is alsonot known what adverse effects may beexperienced in the longer term.There is some evidence that in the long term,women who have undergone hysterectomy (for anyindication) may be at increased risk of symptomssuch as urinary incontinence, 50 vasomotorsymptoms and some psychological symptoms. 51However, women with HMB as a group will alsohave more psychological symptoms than women ofthe same age without HMB. In addition, in clinicalstudies, satisfaction with hysterectomy isreportedly very high. 5277© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Discussion78Cost–utility analysisThere is little difference in the costs and utilitiesfor TBEA and MEA, and these are difficult toestimate precisely. In addition, the opportunitycosts of freeing senior staff, bed and theatre timeif second-generation techniques are increasinglydone by junior staff and under LA have not beenexamined.The economic model is very sensitive to the utilityvalues used, especially the value for women whoare ‘well’ following recovery from an EA procedureor hysterectomy. Little published evidence isavailable for this, leaving the results of the costeffectivenessmodel necessarily uncertain. Acost–utility study by Sculpher 30 has provided mostof the utility values used in this report. Values wereobtained using the TTO method in interviewswith 60 women who had been referred tosecondary care by their GP and haduncomplicated HMB. Other methods of valuinghealth states, such as standard gamble or the EQ-5D, may have generated different values, and inturn different costs/QALY.The value for the state of menorrhagia was ratedat a median of 0.55 (mean 0.5, SE 0.04) by thewomen interviewed in the Sculpher study. 30 Thisseems low (see Table 2 for examples of utilityvalues for other health states). A mean value of 0.5using the TTO method as here suggests thatwomen would be prepared to trade 50% of theirfuture life expectancy to avoid it. The range ofscores for menorrhagia was zero (as bad as beingdead) to 0.95 (where 1.0 is best possible health).Clearly, even among women suffering from HMB,the impact of the condition is valued verydifferently by different individuals. A single utilityvalue must therefore be regarded as uncertain. Inthe same study, women were asked to rate theirown current health state, which had a mean of0.65 (SE 0.04) and a median of 0.75 (range 0–1.0),much higher than the state of menorrhagia, whichthe author ascribes to most women notmenstruating at the time of the interview. Theauthor acknowledges that there are problemseliciting values for chronic health states thatmay affect QoL on a daily basis but for which theworst effects are episodic. In addition, for HMB,effects are not life-long, but will disappear atmenopause.Further health states, such as the utility value forpost-convalescence (‘well’) after treatment forHMB may be particularly difficult to interpret.After hysterectomy, there is no possibility of HMBor other menstrual symptoms returning.Hysterectomy also prevents the possibility of somegynaecological cancers. In contrast, hysterectomymay cause premature ovarian failure and earlymenopause, in addition to having some longerterm adverse effects such as urinary incontinence.An ablation procedure cannot guaranteeamenorrhoea, and there is the possibility ofrecurrent HMB. In the cost–utility analysis bySculpher, 30 women rated the ‘well’ state followinghysterectomy more highly than that following EA(median 0.95 versus 0.90, respectively). This maybe influenced by individual women’s preferencefor a particular treatment. Sculpher suggests that“further analysis is required to explore whetherpreference-based treatment allocation has thepotential to be cost-effective”. 30SubgroupsThe suitability of women with a complaint of HMBfor the different treatments discussed in thisassessment is likely to depend both on thewoman’s expectations and personal requirements(such as family completion or presence of othermenstrually related symptoms) and the preferenceof her GP. For example, women who stronglyprefer amenorrhoea as an outcome, or who havesevere associated menstrual symptomatology(severe PMS, for example) may not be suitablecandidates for EA techniques but be better treatedwith hysterectomy, whereas those preferring toavoid a GA may be better suited to secondgenerationEA techniques. Other aspects of EAthat are known to appeal to women are theavoidance of major surgery, shorter hospitalisationand quicker return to work. 62 However, as womenmay desire conflicting aspects of surgery (such aswishing to stop periods but also wanting to avoidhospitalisation), full information about theprocedures on offer and careful counselling maybe needed. 63TBEA is not suitable for women with largeruterine cavities (>10–12 cm) or those with uterinepathology or abnormalities, who will need tochoose another method of treatment. Pathologymay account for 20–60% of women with HMB 14,15although the review was unable to obtaininformation about the percentage of women withHMB with abnormally shaped uterine cavities orthose with cavities >12 cm in length.Practical considerationsResource savings may be possible with secondgenerationtechniques if more junior medical staffor nurse practitioners were able to carry out theprocedures. The MEA operations reported byCooper and colleagues 86 were all performed by

Health Technology Assessment 2004; Vol. 8: No. 3experienced registrars rather than consultants. Inaddition, first-generation techniques are skilledoperations that require training and experience.Not all consultants are therefore currently able toperform them.Need for further research●●●Head-to-head comparisons of secondgenerationEA techniques should beconsidered.Longer term follow-up for all methods of EA inRCTs will provide better information aboutfailure rates and repeat procedures, in additionto checking whether longer term complicationsare an issue.More sophisticated modelling studies mayimprove estimates of cost-effectiveness, takinginto account population heterogeneity, andwould permit exploration of issues relevant toimplementation such as waiting times anddetailed budget impact.●●●●Given the importance of the utility values indetermining the cost-effectiveness of treatmentsfor HMB, further research to establish utilitiesfor the states of HMB, its surgical treatment,convalescence and complications of treatmentwould be valuable.Future studies of HMB should use validatedQoL measures and established modes ofmeasuring patient satisfaction both with theprocedure and with the outcomes.Further research into the effect of theconstellation of symptoms associated withmenstruation (such as pain, bloating and breasttenderness) and the part that these symptomsplay in women’s perceptions of bleeding andthe effect of its treatment could help toestablish which women will find treatment ofbleeding alone acceptable.Alternative models of care for EA should befurther investigated, including differentoperators (non-consultant medical staff andspecialist nurses) and different settings (officeversus operating theatre).79© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Chapter 6ConclusionsBoth MEA and TBEA techniques appear tooffer effective alternatives in the surgicaltreatment of women with HMB.Second-generation techniques are quicker toperform and appear to provide similar outcomesto first-generation approaches. First-generationtechniques are associated with fewer adverseeffects than hysterectomy and there is evidence infavour of greater safety for second- over firstgenerationtechniques. In trials between first- andsecond-generation techniques, there were very fewsignificant differences in the main clinicaloutcomes.In essence, there seems to be little discernibledifference between second-generation techniqueson the basis of currently available data. However,TBEA may be suitable for fewer women as it hasmore restrictions on uterine size, abnormality andpathology. Both MEA and TBEA appear to offersimilar outcomes to older ablation techniques atsimilar or lower costs. It is not possible to predictwhich patients will become amenorrhagic and thedifferences are small. If amenorrhoea is thepreferred outcome, hysterectomy is the mosteffective technology, but with higher costs. Thecost–utility ratio for hysterectomy versus EA iswithin the range considered by decision-makers torepresent acceptable value for money.The potential exists for reducing costs of ablationfurther by using non-consultant operators or forincreasing access by carrying out ablation in othersettings, such as outpatient suites or communityhospitals. The impact of such developmentscannot currently be estimated with certainty.Finally, the value of increasing the range oftreatment choices available to women has not beenconsidered in this health technology assessment,but may form an important consideration fordecision-makers.81© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3AcknowledgementsThanks are due to Dr Hakan Brodin for hisadvice on the health economics sections ofthis report and to those who provided usefulcomments through peer review. Liz Hodson, whoobtained the papers used in this report, is alsothanked.Expert advisory panelMr N Amso, Senior Lecturer, ConsultantGynaecologist, University Hospital of Wales,Cardiff; Dr S Bhattacharya, Senior Lecturer(Honorary Consultant), Aberdeen MaternityHospital; Dr K Cooper, Consultant Gynaecologist,Aberdeen Royal Infirmary; Mr N Liversedge,Consultant Gynaecologist, Royal Devon andExeter Hospital; Dr M-A Lumsden, Reader, RoyalInfirmary, Glasgow; Mr NC Sharp, ConsultantGynaecologist, Royal United Hospital, Bath.Declared competing interests of theexpert advisory panelDr Nazar Amso has been a principal investigatorand first author in the first large observationalreport of Thermachoice EA (1998) and its followup(accepted for publication). He has hadtravel/accommodation expenses paid for byGynecare in the past when presenting data.Dr Kevin Cooper has undertaken trials onmicrowave ablation and has hadtravel/accommodation to conferences paid for byMicrosulis in the past. Mr Nicholas Sharp is coinventorof the microwave technology and has afinancial interest. Microsulis Medical Ltd havesponsored a Research Fellowship at the RoyalUnited Hospital for the last 7 years and havesponsored conference attendance for Mr Sharpand for the Research Fellows.Contributions of authorsRuth Garside drafted the protocol and finalreport, assessed studies for inclusion, extracteddata, populated the economic model andundertook sensitivity analyses. Ken Steinundertook project management, contributed tothe writing of the report, critiqued the industrysubmissions and provided editorial comment ofthe report and the model. Ali Round designedthe economic model and provided comment ondrafts of the report. Katrina Wyatt assessedstudies for inclusion and extracted data,contributed to writing the report and providededitorial comment on drafts of the report.Alison Price commented on the protocol, carriedout all searches and applied inclusion criteriaand provided editorial comment on the draftreport.83© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3References1. National Statistics Online. Census 2001.http://www.statistics.gov.uk/census2001/pop2001/england_wales.asp.2002.2. Coulter A, Bradlow J, Agass M, Martin-Bates C,Tulloch A. Outcomes of referrals to gynaecologyout-patient clinics for menstrual problems: anaudit of general practice records. Br J ObstetGynaecol 1991;98:789–96.3. Coulter A, Jenkinson C. Quality of life and patientsatisfaction following treatment for menorrhagia.Fam Pract 1994;11:394–401.4. Rybo G. Clinical and experimental studies onmenstrual blood loss. Acta Obstet Gynecol Scand1966;45:1–23.5. Royal College of Obstetricians and Gynaecologists.The initial management of menorrhagia. Evidencebased guidelines No. 1. London: Royal College ofObstetricans and Gynaecologists; 1998.6. Chimbira TH, Anderson ABN, Turnbull AC. Studyof menstrual blood loss. Br J Obstet Gynaecol1980;87:603–9.7. Fraser IS, McCarron G, Markham R. Apreliminary study of factors influencing perceptionof menstrual blood loss volume. Am J Obstet Gynecol1984;149:788–93.8. Hallberg L, Hogdahl A, Nilsson L, Rybo G.Menstrual blood loss – a population study. ActaObstet Gynecol Scand 1966;45:320–51.9. Lethaby A, Hickey M. Endometrial destructiontechniques for heavy menstrual bleeding. CochraneDatabase Syst Rev 2002;CD001501. Oxford: UpdateSoftware.10. Kochli OR. Endometrial ablation in the year 2000– do we have more methods than indications?Contrib Gynecol Obstet 2000;20:91–120.11. Turnbull AC, Rees MC. Gestrinone in thetreatment of menorrhagia. Br J Obstet Gynaecol1990;97:713–15.12. Andersson JK, Rybo G. Levonorgestrel-releasingintrauterine device in the treatment ofmenorrhagia. Br J Obstet Gynaecol 1990;97:690–4.13. Royal College of Obstetricians and Gynaecologists.The management of menorrhagia in secondarycare. Evidence based guidelines No. 5. London:RCOG Press; 1999.14. Vercellini P, Cortesi I, Oldani S, Morchetta M, DeGiorgi O, Crosignani PG. The role of transvaginalultrasonography in the evaluation of patients withexcessive menstrual bleeding. Hum Reprod 1997;12:1768–71.15. Bain C, Parkin DE, Cooper KG. Is outpatientdiagnostic hysteroscopy more useful thanendometrial biopsy alone for the investigation ofabnormal uterine bleeding in unselectedpremenopausal women? A randomisedcomparison. Br J Obstet Gynaecol 2002;109:805–11.16. Hallberg L, Nilsson L. Determination of menstrualblood loss. Scand J Clin Lab Invest 1964;16:244–8.17. Gannon MJ, Day P, Hammadieh N, Johnson N. Anew method for measuring menstrual blood lossand its use in screening women before endometrialablation. Br J Obstet Gynaecol 1996;103:1029–33.18. Higham JM, O’Brien PM, Shaw RW. Assessment ofmenstrual blood loss using a pictorial chart. Br JObstet Gynaecol 1990;97:734–9.19. Reid PC, Coker A, Coltart R. Assessment ofmenstrual blood loss using a pictorial chart: avalidation study. Br J Obstet Gynaecol2000;107:320–2.20. Wyatt KM, Dimmock PW, Walker TJ, O’BrienPMS. Determination of total menstrual blood loss.Fertil Steril 2001;76:125–31.21. Coulter A, Kelland J, Long A, Melville A,O’Meara S, Sculpher M, et al. The management ofmenorrhagia. Effective Health Care 1995;9.22. Hurskaienen R, Teperi J, Turpeinen U, GrenmanS, Kivelä A, Kujansuu E, et al. Combinedlaboratory and diary method for objectiveassessment of menstrual blood loss. Acta ObstetGynecol Scand 1998;77:201–4.23. Royal College of Obstetricians and Gynaecologists.Recommendations arising from the 39th studygroup: disorders of the menstrual cycle. Disordersof the menstrual cycle. London: RCOG, 2000.24. Warner P, Critchley OD, Lumsden MA, Campbell-Brown M, Douglas A, Murray G. Referral formenstrual problems: a cross sectional survey ofsymptoms, reasons for referral and managment.BMJ 2001;323:24–8.25. Clark TJ, Khan KS, Foon R, Pattison H, Bryan S,Gupta JK. Quality of life instruments in studies ofmenorrhagia: a systematic review. Eur J ObstetGynecol Reprod Biol 2002;104:96–104.26. Lampling DL, Rowe P, Clarke A, Lethaby A.Development and validation of the menorrhagiaoutcomes questionnaire. Br J Obstet Gynaecol1998;105:776–9.85© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

References8627. Shaw RW, Brickley M, Evans L, Edwards M.Perceptions of women on the impact ofmenorrhagia on their health using multi-attributeutility assessment. Br J Obstet Gynaecol 1998;105:1155–9.28. Jenkinson C, Peto V, Coulter A. Measuring changeover time: a comparison of results from a globalsingle item of health status and the multidimensionalSF-36 health status surveyquestionnaire in patients presenting withmenorrhagia. Qual Life Res 1994;3:317–21.29. Jenkinson C, Peto V, Coulter A. Making sense ofambiguity: evaluation in internal reliability andface validity of the SF-36 questionnaire in womenpresenting with menorrhagia. Qual Health Care1996;5:9–12.30. Sculpher M. A cost–utility analysis of abdominalhysterectomy versus transcervical endometrialresection for the surgical treatment of menorrhagia.Int J Technol Assess Health Care 1998;14:302–19.31. Weinstein MC. Estrogen use in post-menopausalwomen – costs, risks and benefits. N Engl J Med1980;303:308–16.32. Carter KJ, Ritchey NP, Castro F, Caccamo LP,Kessler E, Erickson BA. Treatment of early stagebreast cancer in the elderly: a health outcomebasedapproach. Med Decis Making 1998;18:213–19.33. Hillner BE, Smith TJ. Efficacy and costeffectiveness of adjuvant chemotherapy in womenwith node-negative breast cancer: a decisionanalysismodel. N Engl J Med 1991;324:160–8.34. Hillner BE, Hollenberg JP, Pauker SG.Postmenopausal estrogens in prevention ofosteoporosis: benefit virtually without risk ofcardiovascular effects are considered. Am J Med1986;80:1115–27.35. Tsevat J, Goldman L, Lamas GA, Pfeffer MA,Chapin CC, Connors KF, et al. Functional statusversus utilities in survivors of myocardialinfarction. Med Care 1991;29:1153–9.36. Brickley M, Kay E, Shepherd JP, Armstrong RA.Decision analysis for lower-third-molar surgery.Med Decis Making 1995;15:143–51.37. Weinstein MC, Coley CM, Richter JM. Medicalmanagement of gallstones: a cost-effectivenessanalysis. J Gen Intern Med 1990;5:277–84.38. Bass EB, Steinberg EP, Pitt HA, Saba GP, LillemoeKD, Kafonek DR. Cost effectiveness ofextracorporeal shock-wave lithotripsy versuscholecystectomy for symptomatic gallstones.Gastroenterology 1991;101:189–99.39. Dolan P, Kind P. Inconsistency and health statevaluations. Soc Sci Med 1996;42:609–15.40. Crow R, Gage H, Hampson S, Hart J, Kimber A,Storey L, et al. The measurement of satisfactionwith health care: implications for practice from asystematic review of the literature. Health TechnolAssess 2002;6(32).41. Coulter A, Kelland J, Peto V, Rees MCP. Treatingmenorrhagia in primary care: an overview of drugtrials and a survey of prescribing practice. Int JTechnol Assess Health Care 1995;11:456–71.42. Prentice, A. AMES study. Presented at the JointBSGE/RCOG conference, Conservative surgery formenorrhagia, London, 17 September 2002.43. Cooper KG, Jack SA, Parkin DE, Grant AM. Fiveyearfollow up of women randomised to medicalmanagement or transcervical resection of theendometrium for heavy menstrual loss: clinicaland quality of life outcomes. Br J Obstet Gynaecol2001;108:1222–8.44. Department of Health. Hospital episode statistics2000–1. London: Office for National Statistics; 2002.45. Bridgman SA, Dunn KM. Has endometrialablation replaced hysterectomy for the treatmentof dysfunctional uterine bleeding? Nationalfigures. Br J Obstet Gynaecol 2000;107:531–4.46. Farquhar C, Naoom S, Steiner CA. The impact ofendometrial ablation on hysterectomy rates inwomen with benign uterine conditions in theUnited States. Int J Technol Assess Health Care2002;18:625–34.47. Vessey MP, Villard-Mackintosh L, McPherson K,Coulter A, Yeates D. The epidemiology ofhysterectomy: findings in a large cohort study. Br JObstet Gynaecol 1992;99:402–7.48. Maresh MJA, Metcalfe MA, McPherson K, OvertonC, Hall V, Hargreaves J, et al. The VALUE nationalhysterectomy study: description of the patients andtheir surgery. Br J Obstet Gynaecol 2002;109:302–12.49. Farrell SA, Kieser K. Sexuality after hysterectomy.Obstet Gynecol 2000;95:1045–51.50. Brown JS, Sawaya G, Thorn DH, Grady D.Hysterectomy and urinary incontinence: asystematic review. Lancet 2000;356:535–9.51. Kuh D, Wadsworth M, Hardy R. Women’s health inmidlife: the influence of the menopause, socialfactors and health in earlier life. Br J ObstetGynaecol 1997;104:923–33.52. Lethaby A, Shepperd S, Cooke I, Farquhar C.Endometrial resection and ablation versushysterectomy for heavy menstrual bleeding.Cochrane Database Syst Rev 2002;CD000329.53. Weber AM. Endometrial ablation. Obstet Gynecol2002;99:969–70.54. Overton C, Hargreaves J, Maresh M. A nationalsurvey of the complications of endometrialdestruction for menstrual disorders: theMISTLETOE study. Br J Obstet Gynaecol1997;104:1351–9.

Health Technology Assessment 2004; Vol. 8: No. 355. Sowter MC, Lethaby A, Singla AA. Pre-operativeendometrial thinning agents before endometrialdestruction for heavy menstrual bleeding(Cochrane review). Cochrane Database Syst Rev2002;CD001124.56. Vancaillie TG. Electrocoagulation of theendometrium. In Lewis BV, Magos AL, editors.Endometrial ablation. Edinburgh: ChurchillLivingstone; 1993, pp. 133–42.57. Davis JA, Miller CD. Fluid infusion duringhysteroscopic surgery. In Lewis BV, Magos AL,editors. Endometrial ablation, Edinburgh:Churchill Livingstone; 1993, pp. 41–56.58. Garry R. Evidence and techniques in endometrialablation: consensus. Gynaecol Endosc 2002;11:5–17.59. Coulter A, Peto V, Doll H. Patients’ preferencesand general practitioners’ decisions in thetreatment of menstrual disorders. Fam Pract1994;11:67–74.60. Marshall S, Hardy RJ, Kuh D. Socioeconomicvariation in hysterectomy up to age 52: national,population based, prospective cohort study. BMJ2000;320:1579.61. Cooper KG, Grant AM, Garratt AM. The impactof using a partially randomised patient preferencedesign when evaluating alternative managementsfor heavy menstrual bleeding. Br J Obstet Gynaecol1997;104:1367–73.62. Nagele F, Rubinger T, Magos A. Why do womenchoose endometrial ablation rather thanhysterectomy? Fertil Steril 1998;69:1063–6.63. Sculpher M, Dwyer N, Browning J, Horsley S,Cullimore J. A survey of women’s preferencesregarding alternative surgical treatments formenorrhagia. Health Expect 1998;1:96–105.64. Sharp NC, Cronin N, Feldberg I, Evans M,Hodgson D, Ellis S. Microwaves for menorrhagia:a new fast technique for endometrial ablation.Lancet 1995;346:1003–4.65. Hodgson DA, Feldberg IB, Sharp N, Cronin N,Evans M, Hirschowitz L. Microwave endometrialablation: development, clinical trials and outcomesat three years. Br J Obstet Gynaecol 1999;106:684–94.66. Bongers MY, Mol B-WJ, Brolmann H-AM.Prognostic factors for the success of thermalballoon ablation in the treatment of menorrhagia.Obstet Gynecol 2002;99:1060–6.67. Bain C, Cooper KG, Parkin DE. A partiallyrandomized patient preference trial of microwaveendometrial ablation using local anaesthesia andintravenous sedation or general anaesthesia: apilot study. Gynaecol Endosc 2001;10:223–8.68. Moher D, Cook DJ, Eastwood S, Olkin I, RennieD, Stroup DF, et al. Improving the quality ofreports of meta-analyses of randomised controlledtrials: the QUOROM statement. Lancet 1999;354:1896–900.69. The Government Actuary Department. Interim lifetables – expectation of life for England and WalesFemales 1998–2000. London: The GovernmentActuary Department; 2002.70. MacLean-Fraser E, Penava D, Vilos GA.Perioperative complication rates of primary andrepeat hysteroscopic endometrial ablations. J AmAssoc Gynecol Laparosc 2002;9:175–7.71. Rogerson L, Duffy S. A European survey of thecomplications of a uterine thermal balloon systemin 5800 women. Gynaecol Endosc 2002;11:171.72. Parkin DE. Microwave endometrial ablation(MEA): a safe technique? Complication datafrom a prospective series of 1400 cases. GynaecolEndosc 2000;9:385–8.73. Department of Health. Hospital episode statistics2000–1. London: Office for National Statistics;2002.74. Sculpher M, Fenwick E, Claxton K. Assessingquality in decision analytic cost-effectivenessmodels: a suggested framework and example ofapplication. Pharmacoeconomics 2000;17:461–77.75. Drummond MF, O’Brien B, Stoddart GL,Torrance GW. Methods for the economicevaluation of health care programmes. Oxford:Oxford University Press; 1997.76. Iyer V, Farquhar C, Jepson R. Oral contraceptionpills for heavy menstrual bleeding. CochraneLibrary 2002;4.77. Lethaby A, Irvine G, Cameron IM. Cyclicalprogestogens for heavy menstrual bleeding(Cochrane review). Cochrane Library 2002;4.Oxford: Update Software.78. Beaumont H, Augood C, Duckitt K, Lethaby A.Danazol for heavy menstrual bleeding (Cochranereview). Cochrane Library 2002;4. Oxford: UpdateSoftware.79. Lethaby A, Augood C, Duckitt K. Nonsteroidal antiinflammatorydrugs for heavy menstrual bleeding(Cochrane review). Cochrane Library 2002;4. Oxford:Update Software.80. Lethaby A, Farquhar C, Cooke I. Antifibrinolyticsfor heavy menstrual bleeding (Cochrane review).Cochrane Database Syst Rev 2002. Oxford: UpdateSoftware.81. Lethaby AE, Cooke I, Rees M. Progesterone/progestogen releasing intrauterine systemsversus either placebo or any other medicationfor heavy menstrual bleeding. Cochrane DatabaseSyst Rev 2000;CD002126. Oxford: UpdateSoftware.82. Meyer WR, Walsh BW, Grainger DA, Peacock LM,Loffer FD, Steege JF. Thermal balloon and87© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Referencesrollerball ablation to treat menorrhagia: amulticenter comparison. Obstet Gynecol 1998;92:98–103.83. Romer T. Therapy of recurrent menorrhagia –Cavaterm balloon coagulation versus roller-ballendometrium coagulation – a prospectiverandomized comparative study. Zentralbl Gynakol1998;120:511–14.84. Grainger DA, Tjaden BL, Rowland C, Meyer WR.Thermal balloon and rollerball ablation to treatmenorrhagia: two-year results of a multicenter,prospective, randomized, clinical trial. J Am AssocGynecol Laparosc 2000;7:175–9.85. Loffer FD. Three-year comparison of thermalballoon and rollerball ablation in treatment ofmenorrhagia. J Am Assoc Gynecol Laparosc 2001;8:48–54.86. Cooper KG, Bain C, Parkin DE. Comparison ofmicrowave endometrial ablation and transcervicalresection of the endometrium for treatment ofheavy menstrual loss: a randomised trial. Lancet1999;354:1859–63.87. Bain C, Cooper KG, Parkin DE. Microwaveendometrial ablation versus endometrial resection:a randomized controlled trial. Obstet Gynecol2002;99:983–7.88. Microsulis Americas. Microwave endometrialablation. FDA trial. Coral Springs, FL: MicrosulisAmericas; 2002.89. Bongers MY, Mol BW, Dijkhuizen FP, BrolmannHA. Is balloon ablation as effective as endometrialelectroresection in the treatment of menorrhagia?J Laparoendosc Adv Surg Tech A 2000;10:85–92.90. Gervaise A, Fernandez H, Capella-Allouc S, TaylorS, La Vieille S, Hamou J, et al. Thermal balloonablation versus endometrial resection for thetreatment of abnormal uterine bleeding. HumReprod 1999;14:2743–7.91. Soysal ME, Soysal SK, Vicdan K. Thermal balloonablation in myoma-induced menorrhagia underlocal anesthesia. Gynecol Obstet Invest 2001;51:128–33.92. Pellicano M, Guida M, Acunzo G, Cirillo D,Bifulco G, Nappi C. Hysteroscopic transcervicalendometrial resection versus thermal destructionfor menorrhagia: a prospective randomized trialon satisfaction rate. Am J Obstet Gynecol 2002;187:545–50.93. Zon-Rabelink I. Rollerball versus Thermachoice.10th Congress of the European Society forGynaecological Endoscopy, Proceedings 2001;11–14.94. Loffer FD,.Grainger D. Five-year follow-up ofpatients participating in a randomized trial ofuterine balloon therapy versus rollerball ablationfor treatment of menorrhagia. J Am Assoc GynecolLaparosc 2002;9:429–35.95. Loffer FD. Three-year comparison of thermalballoon rollerball ablation in treatment ofmenorrhagia (vol 8, pg 48, 2001). J Am AssocGynecol Laparosc 2001;8:330.96. Brun JL, et al. French randomised comparativemulticentre study of Cavaterm versus resectionin patients with meorrhagia: interim results.Morges, Switzerland: Wallsten Medical SA; 2002.97. Pinion SB, Parkin DE, Abramovich DR, Naji A,Alexander DA, Russell IT, et al. Randomised trialof hysterectomy, endometrial laser ablation, andtranscervical endometrial resection fordysfunctional uterine bleeding. BMJ 1994;309:979–83.98. Bae IH, Pagedas AC, Perkins HE, Bae DS.Postablation-tubal sterilization syndrome. J AmAssoc Gynecol Laparosc 1996;3:435–8.99. McCausland AM, McCausland VM. Frequency ofsymptomatic cornual hematometra andpostablation tubal sterilization syndrome aftertotal rollerball endometrial ablation: a 10-yearfollow-up. Am J Obstet Gynecol 2002;186:1274–80.100. Overton C, Hargreaves J, Maresh M. A nationalsurvey of the complications of endometrialdestruction for menstrual disorders: theMISTLETOE study. Contemp Rev Obstet Gynaecol1998;10:209–16.101. Amso NN, Stabinsky SA, McFaul P, Blanc B,Pendley L, Neuwirth R. Uterine thermal balloontherapy for the treatment of menorrhagia: the first300 patients from a multi-centre study.International Collaborative Uterine ThermalBalloon Working Group. Br J Obstet Gynaecol1998;105:517–23.102. O’Connor H, Broadbent JA, Magos AL,McPherson K. Medical Research Councilrandomised trial of endometrial resection versushysterectomy in management of menorrhagia.Lancet 1997;349:897–90188

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 1Pictorial blood loss assessment chartAn assessment chart is illustrated in Figure 17.Name: Ann OtherDay start: 1 July 2002Towel Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8Clots/floodingPain50p x 1p x 3Tampon Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8Clots/floodingPainFIGURE 17 Pictorial blood loss assessment chart. From Higham JM, O’Brien PM, Shaw RW. Assessment of menstrual blood loss usinga pictorial chart. Br J Obstet Gynaecol 1990;97:734–9, reproduced with permission of the authors.89© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 2Research protocolTechnology assessments for theNHS HTA programmeFinal protocol: microwave and thermalballoon endometrial ablation for heavymenstrual bleeding: a systematicreviewDetails of the research teamCorrespondence to: Ms Ruth Garside, ResearchFellow, Peninsula Technology Assessment Group,Dean Clarke House, Southernhay East, ExeterEX1 1PQ, UK.Dr Ken Stein, Senior Lecturer in Public Health,Peninsula Technology Assessment Group (Lead).Dr Katrina Wyatt, Lecturer in Health ServicesResearch, University of Exeter.Mrs Kim Dalziel, Research Fellow, PeninsulaTechnology Assessment Group.Dr Ali Round, Senior Lecturer in Public Health,Peninsula Technology Assessment Group.Ms Alison Price, Information Specialist,Southampton Health Technology AssessmentCentre.Full title of research questionWhat are the effectiveness and cost-effectiveness ofMEA and TBEA techniques for HMB comparedwith transcervical resection and RB ablation andhysterectomy?Clarification of the research question and scopeHMB (menorrhagia) can have a major impact onwomen’s lives. Objective menorrhagia is defined astotal blood loss of more than 80 ml permenstruation over several consecutive cycles. 1However, since objective measurement is difficult,other subjective methods of estimating blood loss,such as flooding, passing of clots, the numbers ofpads or tampons used and haemoglobin levels, arelikely to be used in clinical practice. Subjectiveassessment of a woman’s periods and the effectthat they have on her lifestyle should be taken intoconsideration when looking at treatment efficaciesfor HMB.Menorrhagia without major pathology is acondition that affects many otherwise healthywomen, with one in 20 women aged 30–49 yearsconsulting her GP each year with menorrhagia. 2First-line treatment is usually with drugs, althoughonly 58% of women receive medical therapybefore referral to a specialist. 3 Once referred to agynaecologist, 60% of women with menorrhagiawill have a hysterectomy within 5 years. One infive women in the UK have a hysterectomy beforethe age of 60 years (Coulter, 1991, in RCOGguidelines for menorrhagia in secondary care,1998) and about half of these are for a patientcomplaint of menorrhagia. 4 It has beenestimated that up to half of all women presentingwith menorrhagia will have blood loss withinthe normal range defined by populationstudies. 5 Hysterectomy is the only operationcarried out without a routine assessment of theorgan. 6In 2000–1, 51,858 hysterectomies were performed,of which 82% were abdominal and the remaindervaginal. 7 Of these operations, at least half mightbe expected to be performed for menorrhagia. 8Hysterectomy is a radical solution for HMB, andthere are risks of peri- and postoperativecomplications and, in some cases, significantemotional implications. Since the 1980s, EAtechniques have been developed as alternative,less invasive treatments for menorrhagia. Allmethods of endometrial destruction aim to destroythe inner lining of the uterus (endometrium). Theendometrium is capable of regeneration andtechniques must cause necrosis of the endometrialcells in order to suppress menstruation. Thisincludes removing the full thickness of the uterinelining together with the superficial myometrium(underlying muscular layer), and the basal glandsthought to be the focus of endometrial growth.First-generation techniques such as resection, RBand laser ablation require direct visualisation ofthe endometrium using a hysteroscope.A Cochrane review comparing endometrialresection and ablation techniques withhysterectomy has been undertaken and wasupdated in 1999. 9 This review considers five RCTs,four comparing TCRE and hysterectomy and onewith a three-way comparison including laser EA.This will be reviewed and an updated search forrelevant RCTs undertaken in order to provideadditional information for the appraisal to offer a91© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 292more complete overview of the ablation techniquesand hysterectomy.The Cochrane review concluded that endometrialdestruction offered an alternative surgicaltreatment for menorrhagia to hysterectomy. Bothtypes of procedure were considered as effectiveand had high satisfaction rates from women. Thepermanent relief that hysterectomy offers is offsetby longer operating time, longer recovery periodand higher rates of postoperative complications.The initial cost of endometrial destruction issignificantly lower than for hysterectomy but, as aproportion of women require further surgery, thiscost difference lessens over time. 9It has been suggested that newer EA techniques(such as MEA and TBEA) have fewercomplications than resection. Whereas older styleEA techniques require specialist training andrequire a high level of technical skill, newermethods are regarded as quick and easy to learn. 10Technologies to be appraisedMEA uses high-frequency microwave energy toheat and destroy the endometrium rapidly.Microwaves at a frequency of around 9 GHz areused and these are absorbed by the endometrialtissue to a depth of 3 mm. The heat that isgenerated is conducted deeper into theendometrium so that tissue is destroyed to amaximum depth of 5–6 mm, aiming at sufficientEA without risk to adjacent organs.An applicator inserted into the uterine cavitythrough the dilated cervix delivers themicrowaves. The applicator is slowly withdrawnwith a sweeping movement to ensure that all ofthe endometrium is treated. The temperature ismonitored and controlled through an externalcontrol unit. Treatment takes 5–10 minutes tocomplete and can be carried out under GA or LA.Medication is given to minimise cramping duringand after the procedure.Thermal ablation uses a silicone or latex ballooncatheter, which is inserted into the uterus throughthe vagina. A sterile liquid is used to inflate theballoon to fit the uterine cavity and is then heatedto about 87°C and circulated within the balloonfor about 8 minutes, causing thermal ablation ofthe endometrial lining. Either LA or GA may beused. Medication is given to minimise crampingduring and after the procedure.A preliminary literature review found 52references relating to RCTs of hysterectomy versusvarious methods of EA, comparing types of EA orpreparatory techniques used during EA. Thirteenof these are RCTs of MEA or TBEA versus firstgenerationtechniques. However, there is likely tobe repeat reporting of the same trials among thesereferences.ScopeAll RCTs and non-RCTs of MEA or TBEA versusany removal and ablation of endometrium (byresection or RB) or hysterectomy will be included.Head-to-head comparisons of MEA and TBEA willbe sought. Uncontrolled studies will be excluded.The existing Cochrane systematic review ofendometrial resection and hysterectomy will bereviewed. An updated search to locate any recentRCTs of this comparison will be undertaken.PopulationAll women were recruited from family planningclinics, primary care or specialist clinics.Inclusion criteriaStudies including premenopausal women withregular heavy periods measured objectively orsubjectively.Exclusion criteriaStudies including women with the followingcriteria will be excluded if these women cannot beseparately identified:●●●●postmenopausal bleeding (>1 year from the lastperiod)irregular menses and intermenstrual bleeding(metrorrhagia)pathological causes of menorrhagia (e.g. uterinecancer)iatrogenic causes of menorrhagia (e.g. IUD).InterventionsME or TBEA versus any removal and ablation ofendometrium (including TCRE and EA byelectrocautery or laser) or hysterectomy (by openabdominal, vaginal or laparoscopic routes).Outcomes● QoL: women’s perceived change in QoL● menstrual bleeding: amenorrhoea, objective orsubjective assessment of improvement inmenstrual blood loss● duration of surgery● length of hospital stay● time to return to normal activities/work● rate of satisfaction: at years after surgery 1, 2, 3and 4

Health Technology Assessment 2004; Vol. 8: No. 3●●●requirement for further surgery for menstrualsymptoms: at years after surgery 1, 2,3 and 4.adverse events: including uterine perforation,bleeding, haematometra, laceration, airembolism, intra-abdominal injury, fluidabsorption, infection, cyclical pain, pregnancyand deathresource use/cost.Patient preferencesInformation about patient preferences formethods or treatment for menorrhagia will betaken from included studies. We will extract dataon the number of women approached toparticipate, the number taking part and thenumber who expressed a preference for aparticular surgery.Report methodsThe report will include a systematic review of theevidence for clinical effectiveness and costeffectivenessbased on clinical review and cost datafrom published sources. The review will beundertaken systematically following the generalprinciples outlined in NHS CRD Report 4. Theresearch protocol will be updated as necessary as theresearch programme progresses. Any changes to theprotocol will be reported to NCCHTA and NICE.Search strategy and inclusion criteriaSearches for clinical efficacy will start with theCochrane library. Where good-quality relevantsystematic reviews are found these will form thecore of the assessment of effectiveness.Preliminary searches show that a Cochranereview for hysterectomy versus TCRE and RBexists and searches for this comparison will berestricted to the years since the existing reviewwas written.For the main research question, all publicationsthat describe trials of MEA or TBEA techniquesversus other EA techniques or versus hysterectomywill be obtained using the search strategydescribed below. Preliminary searches have shownthat a Cochrane review of endometrial destructiontechniques also exists. Where appropriate, anymeta-analyses will be updated.Only studies with a comparison arm will beconsidered for inclusion. Where RCT evidencedirectly addressing the questions of interest andsufficient to reach a conclusion is obtained thennon-randomised studies will not be included. Ifinsufficient RCT evidence is available, nonrandomisedstudies will be included.Titles and abstracts will be examined for inclusionby two independent reviewers and disagreementwill be resolved by consensus.DatabasesElectronic databases: including MEDLINE(Silver Platter); PubMed (previous 6 months forlatest publications); EMBASE; The CochraneLibrary including the Cochrane SystematicReviews Database, Cochrane Controlled TrialsRegister, DARE, NHS EED and HTA databases;NRR (National Research Register); Web of ScienceProceedings; Current Controlled Trials; ClinicalTrials.gov.Bibliographies of included studies will be assessedfor relevant studies.Contacting research groups and industry.Inclusion● systematic reviews● RCTs● controlled clinical trials.Exclusion● animal models● preclinical and biological studies● narrative reviews, editorials, opinions● non-controlled studies● non-English language papers● reports published as meeting abstracts only.Review methodsData extraction strategyData will be extracted by one researcher andchecked by another.Quality assessmentAssessments of quality will be performed using theindicators shown below. Owing to the nature of theintervention, the presence of blinding of treatmentand treatment concealment are not applicablemeasures of quality except possibly in head-toheadcomparisons.Trial characteristics:1. Appropriate method of randomisation ofRCTs.2. Blind assessment of outcomes.3. Numbers of women randomised, excluded andLTFU.4. Whether ITT analysis is performed.5. Whether a power calculation was done.6. Timing, duration and location of the study.Study participants:1. Age and any other recorded characteristics ofwomen in studies.93© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 22. Inclusion criteria.3. Exclusion criteria.Interventions used:1. Type of EA technique and route ofhysterectomy surgery.2. Endometrial thinning agents used.Outcomes:1. Methods used to evaluate women’s satisfactionand QoL post-surgery.2. Methods used to measure menstrual loss.3. Methods used to evaluate resource and patientscosts.4. Length of follow up.Methods of analysis/synthesisWhere appropriate, meta-analysis methods will beemployed to estimate a summary measure ofeffect, otherwise information will be synthesised bynarrative methods.Methods for evaluating QoL, costs and costeffectivenessand/or QALYsQoL measures, costs for treatments and savingswill be taken from published work. Estimates ofresource costs from individual trusts or groups oftrusts may be used, if time permits, wherepublished data are not available.If an economic analysis for microwave or thermalablation already exists, we will provide a critiqueof this. If no economic analysis already exists,a cost-effectiveness model will be undertakenof microwave and thermal ablation techniquesversus TCRE and RB ablation andhysterectomy.Handling the industry submissionWhere information provided by industry meetsour inclusion criteria, this will be included in thereview.Project managementTimetableDraft protocol: 30 July 2002Finalised protocol: 20 August 2002Progress report: 13 November 2002Draft final report: 22 January 2003Competing interestsNone.External reviewersA group is currently being formed. This group willact as an expert resource to guide the process ofthe review. At least two separate experts will beidentified as peer reviewers of the completed draftreview.References1. Royal College of Obstetricians and Gynaecologists.The initial management of menorrhagia. Evidencebased guidelines No. 1. London: Royal College ofObstetricians and Gynaecologists; 1998.2. Vessey MP, Villard-Mackintosh L, McPherson K,Coulter A, Yeates D. The epidemiology ofhysterectomy: findings in a large cohort study. Br JObstet Gynaecol 1992;99:402–7.3. Coulter A, Bradlow J, Agass M, Martin-Bates C,Tulloch A. Outcomes of referrals to gynaecologyout-patients clinics for menstrual problems: anaudit of general practice records. Br J ObstetGynaecol 1991;98:789–96.4. Chimbira TH, Anderson ABN, Turnbull AC. Studyof menstrual blood loss. Br J Obstet Gynaecol1980;87:603–9.5. Fraser IS, McCarron G, Markham R. A preliminarystudy of factors influencing perception of menstrualblood loss volume. Am J Obstet Gynecol1984;149:788–93.6. Wyatt KM, Dimmock PW, Walker TJ, O’Brien PMS.Determination of total menstrual blood loss. FertilSteril 2001;76:125–31.7. Hospital Episode Statistics. London: Department ofHealth, 2001.8. Coulter A, Kelland J, Long A, Melville A, O’MearaS, Sculpher M, et al. The management ofmenorrhagia. Effective Health Care 1995;9.9. Lethaby A, Shepperd S, Cooke I, Farquhar C.Endometrial resection and ablation versushysterectomy for heavy menstrual bleeding.Cochrane Database Syst Rev 2000.10. Cooper KG, Bain C, Parkin DE. Comparison ofmicrowave endometrial ablation and transcervicalresection of the endometrium for treatment ofheavy menstrual loss: a randomised trial. Lancet1999;354:1859–63.94

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 3Search strategyTwo separate searches were undertaken for thisproject. One searched specifically for researchevidence on MEA and TBEA for all years, theother looked for research comparing hysterectomywith the first-generation EA techniques of RBablation and TCRE from 1999 onwards to updatean existing Cochrane review. The followingdatabases were searched for published studies andrecently completed and ongoing research.Search 1. Microwave and thermalballoon endometrial ablationCochrane Library (Issue 3, 2002)Includes the Cochrane Systematic ReviewsDatabase, Cochrane Controlled Trials Register,DARE, NHS EED and HTA databases.#1 (((MENORRHAGIA or BLEEDING) orBLOOD) or MENSTRUAL)#2 MICROWAVE*#3 MICROWAVES*.ME#4 THERMAL OR BALLOON#5 (ENDOMETRI* near ((ABLAT* orRESECT*) or DESTRUCTION))#6 DIATHERMY*.ME#7 BALLOON-DILATATION*.ME#8 CATHETER-ABLATION*.ME#9 #2 OR #3 OR #6#10 #9 AND #5#11 #4 OR #7 OR #8#12 #11 AND #5#13 #10 OR #12.National Research Register (Issue 2,2002)As for the Cochrane Library (above).MEDLINE (WebSPIRS) (1966–August2002)(((‘Menorrhagia-’ / all subheadings inMIME,MJME) or (menorrhagia) or (bleeding orblood or menstrual)) and (((microwave near(endomet* ablat*)) or (explode ‘Diathermy-’ / allsubheadings in MIME,MJME) or (microwave*)) or((thermal balloon) or (Catheter-Ablation-methodsin MIME) or (‘Catheter-Ablation’ / all subheadingsin MIME,MJME) or (Balloon-Dilatation-methodsin MJME) or (Catheter-Ablation-methods inMJME) or (thermal near (balloon* or ablat*)))))and ((explode ‘Hysterectomy-’ / all subheadings inMIME,MJME) or (hysterectom*)).PubMed (Internet version for recentstudies) (last 180 days)endometrial and (ablation or resection ordestruction)EMBASE (WebSPIRS) (1980–August2002)((ENDOMETRI* near ((ABLAT* or RESECT*) orDESTRUCTION)) and ((microwave*) or(‘microwave-irradiation’ / all subheadings) or(‘microwave-radiation’ / all subheadings) or(‘diathermy-’ / all subheadings))) or((ENDOMETRI* near ((ABLAT* or RESECT*) orDESTRUCTION)) and ((thermal near balloon) or(‘balloon-dilatation’ / all subheadings) or (‘ballooncatheter’/ all subheadings) or (‘catheter-ablation’ /all subheadings))).Web of Science Proceedings (all yearsfrom 1980)(endometrial or endometrium) and (ablation orresection or destruction) and (microwave* orthermal balloon).Clinical Evidence (Issue 7, September)2002Endometrial and (destruction or resection orablation).Search 2. Endometrial ablation(TCRE/RB) versus hysterectomy(from 1999 to August 2002)Cochrane Library (Issue 3, 2002)(from 1999–August 2002)#1 (((MENORRHAGIA or BLEEDING) orBLOOD) or MENSTRUAL)#2 HYSTERECTOMY*:ME#3 HYSTERECTOM*#4 (ENDOMETRI* near ((ABLAT* or RESECT*)or DESTRUCTION))#5 ((#2 or #3) or #4)#6 #1 and #5#7 #6 Publication date from 1999 to 2002.95© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 3National Research Register (Issue 2,2002)#1 (ENDOMETRI* NEAR ((ABLAT* ORRESECT*)OR DESTRUCTION))#2 HYSTERECTOM*#3 HYSTERECTOMY*.ME#4 (((MENORRHAGIA OR BLEEDING) ORBLOOD) OR MENSTRUAL)#5 #1 OR #2 OR #3#6 #5 AND #4.MEDLINE (WebSPIRS) (1999–August2002)(((‘Menorrhagia-’ / all subheadings inMIME,MJME) or (menorrhagia) or (bleeding orblood or menstrual)) and ((explode ‘Hysterectomy-’ / all subheadings in MIME,MJME) or(hysterectom*) or (endometr* near (ablat* orresect* or destruction)))) and (((((‘Menorrhagia-’ /all subheadings in MIME,MJME) or(menorrhagia) or (bleeding or blood ormenstrual)) and ((explode ‘Hysterectomy-’ / allsubheadings in MIME,MJME) or (hysterectom*)or (endometr* near (ablat* or resect* ordestruction)))) and (English in la) and(LA=ENGLISH) and (PT=RANDOMISED-CONTROLLED-TRIAL)) or ((((‘Menorrhagia-’ /all subheadings in MIME,MJME) or(menorrhagia) or (bleeding or blood ormenstrual)) and ((explode ‘Hysterectomy-’ / allsubheadings in MIME,MJME) or (hysterectom*)or (endometr* near (ablat* or resect* ordestruction)))) and (LA=ENGLISH) and(PT=META-ANALYSIS)) or ((systematic near(review or overview)) or meta-anal* or metaanal*)or (random*)).PubMed (Internet version) (last180 days)(endometrial or endometrium) and (ablation orresection or destruction).EMBASE (WebSPIRS) (1999–July 2002)((((‘menorrhagia-’ / all subheadings) or(menorrhagia or bleeding or blood or menstrual))and ((‘endometrium-ablation’ / all subheadings) or(endometr* near (ablat* or resection ordestruction)) or (explode ‘hysterectomy-’ / allsubheadings) or (hysterectom*))) and (random* ormeta-anal* or metaanal* or (systematic* near(review* or overview*)))) and (English in la).Web of Science Proceedings(1999–August 2002)(endometrial or endometrium) and (ablation orresection or destruction).An updated search of MEDLINE and EMBASEwas run for both search strategies on 4 December2002 to cover the intervening months from Augustto November 2002 before the report was drafted.96

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 4Excluded studiesExcluded studies are illustrated in Figure 18.Microwave and thermalballoon ablation277 abstractsidentifiedTCRE/rollerball vs hysterectomyfrom 1999120 abstractsidentified (40 ofwhich duplicatedthose from T/MEAsearch)68 full text articlesacquired(17 potential controlledtrials, 24 relevant caseseries for background,15 backgroundinformation, 7 with noabstract that may berelevant, 3 systematicreviews) plus 3supplied by industryExclusions:35 narrative reviews/editorials/opinions/letters,7 preclinical/biological studies,16 case studies, 11 conf.abstracts, 7 pt group notmenorrhagic, 59 interventionnot thermal balloon ormicrowaves EA, 8 only non-ptrelevant outcomes reported, 1animal model, 11 non-Englishlanguage, 5 not TCRE/rollerball comparator13 full textarticles acquired(7 relevant trials,6 background papers)Exclusions:13 narrative reviews/editorials/opinions/letters, 3preclinical/biological studies,13 case series studies, 1 conf.abstract, 2 pt group notmenorrhagic, 44 interventionnot TCRE or rollerball, 33comparator not hysterectomy,1 non-English language8 potential trials excluded at fulltext stage – reasons below10 controlled trialsincluded in review and2 systematic reviews0 studies includedin reviewFIGURE 18 Excluded studies97© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 4List of excluded studies from search strategyStudyReason for exclusion atfull-text stageUterine balloon to avoid hysterectomy. J Women’s Health 1997;6:401–2Bongers MY, Mol BWJ, Fernandez H, Gervaise A. Thermal balloon ablation versusendometrial resection for treatment of abnormal uterine bleeding. Hum Reprod2000;15:1424–5Garuti G, Cellani F, Colonnelli M, Luerti M. Endometrial thermal ablation to treatdysfunctional menorrhagia; a clinical experience using two different techniques.Ital J Gynaecol Obstet 2001;13:160–5Genolet PM, Gerber S, De Grandi P, Friberg B, Ahlgren M. Endometrial ablationfor dysfunctional uterine bleeding in the perimenopause, clinical results of amulticentre trial with the Cavaterm thermal balloon. 9th InternationalMenopause Society World Congress on the Menopause 1999;315–20Opinion pieceLettersComparison of HTA and thermalballoon ablationAbstract onlyLoffer FD, Grainger D, Kung RC, Stabinsky SA. Endometrial ablation for the treatment of Abstract onlymenorrhagia: a randomised trial comparing uterine balloon therapy with rollerball.Acta Obstet Gynecol Scand 1997;76: 23Parkin D. A randomised controlled trial comparing transcervical endometrial resectionwith microwave endometrial ablation in the treatment of dysfunctional uterinebleeding: 2 year follow up. 9th Annu Congress of Int Soc for Gynecol Endoscopy/10thAnnu in Mtg of Australian Gynaecological Endoscopy Soc, 16–19 April 2000;140Romer T. The treatment of recurrent menorrhagia – Cavaterm-balloon-coagulationversus RB-endometrial ablation – a prospective randomised comparative study.Zentralbl Gynakol 1998;120:511–14Wortman, M. Thermal balloon and rollerball ablation to treat menorrhagia:a multicenter comparison Obstet gynecol 1998; 92:1057Abstract onlyExcluded because in German butlater included when an Englishtranslation was supplied byWallesten, the makers ofCavatermLetterStudies supplied by industry, exclusionsStudyHawe J, Abbott J, Hunter D, Phillips G, Garry R. Pre-publication copy of a doubleblind randomised controlled trial comparing the Cavaterm endometrial ablationsystem with the Nd:YAG laser for the treatment of dysfunctional uterine bleeding.Br J Obstet Gynaecol, in pressReason for exclusion atfull-text stageWrong comparator98

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 5Included systematic reviews – QUOROM checklist1. Lethaby et al., 2002. 52 Endometrial resectionand ablation versus hysterectomy for heavymenstrual bleeding.1. Title: Identify the report as a systematic review?Yes – Cochrane Review.2. Abstract: Uses a structured format?Yes. Organised as:BackgroundObjectivesOutlines the clinical problemThe clinical question states thatthe review will compare EAtechniques but is not explicit instating that clinical effectivenessor cost-effectiveness is to beevaluatedSearch strategy Databases and additional sourcessearched are listedSelection Describes the population,criteria intervention and study designData collection Describes outcomes extracted,and analysis methods of data extraction, andquantitative data synthesis insufficient detail to permitreplication. Methods for validityassessment not describedMain resultsReviewers’conclusions3. IntroductionCharacteristics of included trialsnot reported. Description offindings presented but not pointestimates or CIsReports the main resultsYes. Describes the clinical problem, biologicalrationale for the intervention.4. MethodsSearchingDatabases searched are listed, handsearching listed. No restrictions ofpublication status, language oryear of publication are statedDichotomous data expressed asPeto OR and 95% CI, meta-analysis using RevMan, continuousdata shown as weighted meandifference and 95% CI. Stated thatfixed approach used unlesssignificant heterogeneity, althoughin fact all outcomes use fixed-effectmodels. Where only medians andranges were available, the medianwas regarded as identical with themean and estimate of SDcalculated from the range (range ×0.95/4). Not clear by what methodthe studies are weighted in themeta-analysis for continuousoutcomes as not stated – it doesnot appear to be samplesize, and is not consistent acrossoutcomesSelectionValidityassessmentDataabstractionStudycharacteristicsQuantitativedatasynthesisInclusion criteria are given thatinclude description of includedpopulation, intervention studydesign and outcomesMethodological quality isdescribed in relation toadequate concealment priorto randomisation, powercalculations for sample size,ITT analysis and attrition rates.Sensitivity analyses areundertakenIndependently by two reviewers.Additional information about trialmethodology and results soughtfrom corresponding author oftrials where necessaryStudy design, patientcharacteristics, interventiondetails, outcome definitions,length of follow-up are assessed.Heterogeneity was examined byinspecting the scatter in the datapoints on the graphs and theoverlap in their CI, and bychecking the results of chisquaredtests99© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 51005. ResultsTrial flowStudycharacteristicsQuantitativedata synthesis6. DiscussionNot includedStudy design, patientcharacteristics, interventiondetails, outcome definitions,length of follow-up are tabulatedAgreement on selection andvalidity assessment is notreported. Results of meta-analysispresented from RevMan.The discussion summarises key findings, clinicalinferences based on internal and external validityare not discussed, the results are interpreted basedon the total evidence included in the review,potential biases are not discussed. The studyaddresses the problem of heterogeneity betweenthe studies. Sensitivity analyses were performed asa result of this; it is stated that there was nochange in the direction of results although pointsestimates did change which are not stated. Futureresearch agenda is suggested.2. Lethaby and Hickey, 2002. 9 Endometrialdestruction techniques for heavy menstrualbleeding1. Title: Identify the report as a systematic review?Yes – Cochrane Review.2. Abstract: Uses a structured format?Yes – organised as:Background Outlines the clinic problemObjectives Expresses the clinical questionexplicitlySearch strategy The databases searched andother search methods are listedSelection Selection criteria – type of trials,criteria population, intervention andoutcomes are listedData collection Methods for inclusion, qualityand analysis assessment and data extractionare described. Methods of datasynthesis not describedMain resultsReviewers’conclusionsCharacteristics of RCTs includedand excluded are not described.Point estimates and 95% CIs aregivenMain results given3. IntroductionYes. Describes the clinical problem, biologicalrationale for the intervention.4. MethodsSearchingSelectionValidityassessmentDataabstractionStudycharacteristicsQuantitativedata synthesis5. ResultsTrial flowStudycharacteristicsDetails of databases searchedgiven, search terms listed,registers searched listed, handsearching listedInclusion and exclusion criteriagiven. Titles and abstractsscreened by one reviewer.Uncertainty at full-script stageresolved by discussion withcolleagueQuality of included trials assessedindependently by two reviewersData extraction performedindependently by two reviewers.Additional information about trialmethodology and results soughtfrom corresponding author oftrials where necessaryStudy characteristics aredescribedDichotomous data expressed asORs with 95% CI meta-analysiswith RevMan using Peto-modifiedMantel–Haenszel method.Continuous outcomes shown asweighted mean difference with95% CI. Heterogeneity assessed byinspecting the scatter in the datapoints on the graphs and theoverlap of their CIs, and bychecking results of Chi-squaredtests. Fixed-effects model usedunless there was significantheterogeneity (one outcome only– use of LA). No subgroupanalysis planned. A priorisensitivity analyses plannedTrial flow diagram not includedDescriptions for each trialtabulated, including patientcharacteristics, method ofrandomisation, inclusion/exclusioncriteria, outcomes. Interventionsdescribed but not referenced

Health Technology Assessment 2004; Vol. 8: No. 3Quantitativedata synthesisAgreement on selection andvalidity assessment is notreported. Results of meta-analysispresented from RevMan. However,in the section comparing all firstgenerationand all secondgenerationmethods of EA, thefigures given in the text and thosepresented in the graphs differ. Inthe case of results foramenorrhoea, this leads the textto suggest the difference issignificant, whereas the graphdoes not.6. DiscussionThe discussion summarises key findings. Internaland external validity (e.g. study differences inactual menstrual blood loss among participants,inclusion of patients who had not failed medicalmanagement) are discussed. The results arediscussed in the light of total available evidence.Potential biases in the review process are notdiscussed. Future research agenda is suggested.101© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 6Included systematic reviews103© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

104Reference and design Research question and search strategy Inclusion and quality criteria• Author: Lethaby et al., 2002 52• Study topic: EA andhysterectomy for HMB• Aim: To determine the effectiveness of endometrialresection and ablation techniques vs hysterectomyto reduce menstrual blood flow• Search strategy (databases searched): Trial Registerof the Cochrane Menstrual Disorders andSubfertility Group, MEDLINE, EMBASE, CurrentContents Biological Abstracts, Social SciencesIndex, PsycLIT and CINAHL. Relevant journalswere hand searched and citation lists of includedtrials, conference abstracts and review articles alsosearched• Search terms: menorrhagia, excessive menstrualblood loss, dysfunctional uterine bleeding, irondeficient anaemia, heavy menstrual bleeding,hysterectomy, vaginal hysterectomy, totalabdominal hysterectomy, subtotal abdominalhysterectomy, laparoscopic hysterectomy,transcervical resection of the endometrium, TCRE,endometrial, laser ablationInclusion criteria:• Study design: RCTs• Interventions: Resection, RB, laser or other ablations of the endometrium• Population: Women of reproductive years with regular heavy periods measured eitherobjectively or subjectively• Setting: Primary care, family planning or specialist clinics• Outcome measures: Objective or subjective improvement in menstrual blood loss,women’s perceived change in QoL (recorded in a reproducible and validate format),length of stay in hospital, time to return to work, duration of surgery, rate ofsatisfaction at 1, 2, 3, 4 years, mortality• Quality criteria: trial characteristics – method of randomisation, presence of blinding oftreatment allocation, quality of allocation concealment, number of women randomised,excluded or LTFU, whether ITT analysis done, whether power calculation was done,duration timing and location of study. Participant characteristics – age and any otherrecorded characteristics, other inclusion criteria, exclusion criteria. Interventions – typeof endometrial destruction techniques and route or hysterectomy. Outcomes –methods used to measure blood loss, to evaluate resource and patient costs and toevaluated participant satisfaction and change in QoL post-surgery.• Application of methods: Trials were selected for inclusion by 2 reviewers, assessment ofquality was independently assessed by 2 reviewers using forms designed to CochraneguidelinesResults• Quantity of included studies: 5 RCTs, total of 752 participants• Quality of included studies: 4 out of 5 had an allocation score of A based on adequate concealment prior to allocation. The other gave no indication of method of concealmentalthough randomisation was by sealed envelope. No trial was blinded – patients and surgeons knew what operation was performed. Power calculations were performed for 4/5studies and analysis was by ITT. 4 studies were single centre and the 5th had 9 UK centres but no imbalances were seen in baseline prognostic factors. Withdrawals afterrandomisation and prior to surgery were 8, 2, 6, 13 and 3%. At longer follow-up, additional losses were 9, 21, 0, 39 and 9%. Two trials calculated cost per participant based onresource use. A third summed the average costs of variable resources and then added a factor of 100% to allow for fixed costs (this method did not permit estimates of varianceto be calculated)• Combined treatment effect (including point estimates, CI, p values, etc.): Satisfaction: at 1 year odds of satisfaction higher with hysterectomy (Peto OR 0.46, 95% CI 0.24 to 0.88;p = 0.02), at 2 years (OR = 0.31, 95% CI 0.16 to 0.59; p = 0.00). However, no difference at 3 (OR 0.32, 95% CI 0.08 to 1.37; p = 0.12) and 4 years (OR 0.52 to 95% CI 0.21to 1.26; p = 0.15).• Improvement in MBL: At 1 year odds of greater proportion with improved MB favoured hysterectomy (OR 0.12, 95% CI 0.06 to 0.25), at 2 years no difference (OR 0.10, 95%CI 0.00 to 5.41, p = 0.3); at 4 years no difference (OR 0.15, 95% CI 0.01 to 2.38, p = 0.18). ORs at 2 and 4 years based on 1 study.• QoL: GR inventory scores (based on 1 study) no difference at 1 year [WMD 0.000 (95% CI –1.750 to 1.750, p = 0.00)].• All the following SF-36 scores at 2 years – Role limitation (physical), no difference (WMD –1.426, 95% CI –10.310 to 7.458, p = 0.8); role limitation (emotional), no difference(WMD –7.272, 95% CI –15.741 to 1.196, p = 0.09); social functioning higher scores with hysterectomy (WMD –7.182, 95% CI –12.387 to –1.97, p = 0.01); mental health, nodifference (–2.935, 95% CI –7.386 to –1.516, p = 0.20); energy, no difference (WMD –5.026, 95% CI –10.373 to 0.322, p = 0.07); pain, better with hysterectomy (WMD–8.709, 95% CI –15.034 to –2.38, p = 0.01); general health perception, better with hysterectomy (WMD –6.697, 95% CI –12.203, to –1.19, p = 0.02); physical functioning, nodifference (WMD –2.756, 95% CI –7.188 to 1.676, p = 0.20).• Change in Euroqol score from baseline at 4 months (1 study), no difference (WMD –7.0000, 95% CI –17.286 to 3.286, p = 0.18), at 2 years (1 study), no difference (WMD–1.5000, 95% CI –6.287 to 3287, p = 0.50).Appendix 6continued

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.Results (cont’d)• SSR score 2 years after surgery (1 study), no difference (WMD –3.700, 95%CI –11.169 to 3.769, p = 0.30).• Total HAD score 2 years after surgery (1 study), no difference (WMD 1.500, 95% CI –1.329 to 4.319, p = 0.30), anxiety HAD scores 2 and 4 years after surgery, no difference(WMD 0.669, 95% CI –0.302 to 1.641, p = 0.18); depression HAD scores 2 and 4 years after surgery, no difference (WMD 0.002, 95% CI –0.092 to 0.096, p = 1.00).• The following 4 measures each based on 1 study: proportion with improvement in QoL at 2 years, no difference (1 study) (OR 0.54, 95% CI 0.15 to 1.98, p = 0.40);proportion with improvement in general health at 1 year, better for hysterectomy (OR 0.26, 95% CI 0.11 to 0.63); proportion with improvement in general health 4 years aftersurgery, no difference (0.36, 95% CI 0.13 to 1.01, p = 0.05); proportion with improved symptoms at 1 year, no difference (OR 0.43 95% CI 0.15 to 1.28, p = 0.13).• Duration of surgery: Shorter with TCRE/ablation (WMD –23.062, 95% CI –23.799 to –22.324, p = 0.00)• Duration of hospital stay: Shorter with TCRE/ablation (WMD –4.907, 95% CI 4.948, to –4.866, p = 0.00)• Time to return to work: Shorter with TCRE/ablation (WMD –4.641, 95% CI –4.853 to –4.430, p = 0.00)• Adverse effects• Immediate:• Sepsis fewer with TCRE/ablation (OR 0.16, 95% CI 0.10 to 0.24, p = 0.00); haemorrhage, no difference (OR 0.59, 95% CI 0.20 to 1.74, p = 0.30); blood transfusion, fewer withTCRE/ablation (OR 0.22, 95% CI 0.08 to 0.57, p = 0.00); urinary retention, fewer with TCRE/ablation (OR 0.13, 95% CI 0.04 to 0.44, p = 0.00); anaemia (1 study), fewer withTCRE/ablation (OR 0.12, 95% CI 0.03 to 0.43, p = 0.00); pyrexia (1 study), fewer with TCRE/ablation (OR 0.12, 95% CI 0.06 to 0.27, p = 0.00); vault haematoma, fewer withTCRE/ablation (OR 0.14, 95% CI 0.06 to 0.34, p = 0.00); wound haematoma (1 study), fewer with TCRE/ablation (OR 0.11, 95% CI 0.04 to 0.32, p = 0.00); anaesthetic, nodifference (1 study) (OR 0.12, 95% CI 0.01 to 1.99, p = 0.14); fluid overload, more likely with TCRE/ablation (OR 5.57, 95% CI 1.82 to 17.12, p = 0.00); perforation (1study),no difference (OR 6.85, 95% CI 0.14 to 346.18, p = 0.30), GI obstruction, ileus (1 study), no difference (OR 0.47, 95% CI 0.05 to 4.57, p = 0.50); laparotomy (1 study),no difference (OR 0.33, 95% CI 0.05 to 2.41, p = 0.30); cautery of hypergranulation, fewer with TCRE/ablation (OR 0.12, 95% CI 0.02 to 0.94)• Assessment of heterogeneity: Through examining the scatter in data points on graphs and their overlap in CI and by checking the results of chi-squared tests• Adverse effects after discharge:• Sepsis (1 study), fewer with TCRE/ablation (OR 0.19, 95% CI 0.08 to 0.47, p = 0.00); haematoma, no difference (OR 0.55, 95% CI 0.13 to 2.4, p = 0.4); diarrhoea (1 study), nodifference (OR 0.13, 95% CI 0.00 to 6.68, p = 0.3); haemorrhage (1 study), no difference (OR 7.24, 95% CI 0.14 to 365.04, p = 0.3)• Further surgery for HMB• Within 1 year, more likely with TCRE/ablation (OR 7.33, 95% CI 4.18 to 12.86, p = 0.00).• At 2 years, more likely with TCRE/ablation (OR 7.5, 95% CI 4.20 to 13.42, p = 0.00).• At 3 years (1 study), more likely with TCRE/ablation (OR 4.45, 95% CI 1.78 to 11.15, p = 0.00).• At 4 years (1 study), more likely with TCRE/ablation (OR 9.84, 95% CI 4.93 to 19.67, p = 0.00).Methodological comments• Search strategy: OK• Participants: OK• Inclusion/exclusion criteria: OK• Quality assessment of studies: Good• Method of synthesis: Differences between groups for continuous data outcomes using weighted mean difference. A fixed-effects model used unless significant heterogeneityshown, in which case results were confirmed with a random effects model. Median regarded as identical to the mean where this was the only measure available and an estimateof SD calculated from the range. Some outcomes were reported by only one included studyGeneral comments• Generalisability: High• Appropriate outcome measures used?: Yes• Any differences in baseline characteristics of patients and controls?: None reported• Appropriate analysis?: Yes• Funding?: None statedGR, Golombok Rust Inventory of Marital State; HAD, hospital Anxiety and Depression Scale; MB, menstral bleeding; MBL, menstral blood loss; SSR, Sabbatsberg Sexual Rating Scale.continuedHealth Technology Assessment 2004; Vol. 8: No. 3105

106Reference and design Research question and search strategy Inclusion and quality criteria• Authors: Lethaby and Hickey,2002 a• Study topic: EA techniques forHMB• Aim: To compare the efficacy, safety and acceptability ofmethods used to destroy the endometrium to reduceHMB in premenopausal women• Search strategy (databases searched): Regular 6-monthlysearches of the Trials Register for the Menstrual Disordersand Subfertility Cochrane Group (most recent July 2001),also MEDLINE (1966–Sept. 2001), EMBASE (1980–Aug.2001), Current contents (1993–week 38, 2001), BiologicalAbstracts (1980–June 2001), PsychINFO (1967–Aug.2001), CINAHL (1982–July 2001)• Search terms: menorrhagia, hypermenorrhagia, (excessive)menstrual blood loss, dysfunctional uterine bleeding, irondeficient anaemia, heavy menstrual bleeding, dysfunctionaluterine bleeding, transcervical resection of theendometrium, TCRE, endometrial ablation, laser ablation,hysteroscopy, electrosurgery, rollerball, (thermal) balloon,hypertherm(ia), thermotherapy, photodynamic therapy,phototherapy, cryoablation, microwave ablation,radiofrequency, saline irrigation, laser interstitial,Thermachoice, Cavatherm, ELITT, Vesta, Novasure,Microsulis, Cryogen, bipolarIn addition, the National Research Register issue 3, 2001,MRC clinical trials register and NHS CRD were searchedusing the search terms menorrhagia and endometrial ablation,and hand-searching of journals, conference abstracts andreview articles was undertaken. Experts, manufacturers andauthors were also contactedInclusion criteria• Study design: RCT and comparative studies• Interventions: TCRE, laser, RB, saline irrigation, microwave, radiofrequency,heated balloon, photodynamic therapy, cryoablation and any other endometrialdestruction techniques compared with each other or grouped into categories(1st- or 2nd-generation techniques) to reduce HMB• Population: Women of reproductive years with regular heavy periods measuredobjectively or subjectively• Setting: Primary care, family planning or specialist clinics• Outcome measures: Primary – objective or subjective assessment ofimprovement in MBL, QoL, improvement of menstrual symptoms such asamenorrhoea and PMS. Secondary – length of hospital stay, time to return towork, duration of surgery, operative difficulties, rate of satisfaction withprocedure, complication rate, resource use/cost, requirement for furthersurgery for HMB, mortality• Quality criteria: Trial characteristics – method of randomisation, blinding, qualityof allocation concealment, number randomised, excluded, LTFU, ITT analysis,power calculation included, duration, timing, location of study, source offunding. Study characteristics – age and other recorded characteristics ofwomen, other inclusion criteria, exclusion criteria. Interventions used – type ofEA technique. Outcomes – methods used to measure blood loss, to evaluateresource and patient costs and to evaluate satisfaction, change in QoL andmenstrual symptoms• Application of methods: Data extracted independently by 2 reviewers usingforms according to Cochrane guidelines. Authors of 4 trials contacted forfurther information but only one response receivedAppendix 6continued

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.Results• Quantity of included studies: 8 studies, 1595 participants• Quality of included studies: All had parallel group design, 3 multicentre. 5 had adequate randomisation procedures, 3 did not report if randomisation was concealed. Blinding notreported and unlikely in all. Two trials did not report ITT, 2 had no drop-outs, 4 reported ITT but 2 of these did not in fact include drop-outs in final analysis. 2 studies did notreport power calculations. Five had funding from large pharmaceutical companies• Combined treatment effect (including point estimates, CI, p values, etc.): Significant differences only shown below – all other outcomes no significant differences• Laser vs TCRE – laser surgery average 9 minutes longer (WMD = 9.15, 95% CI 7.2 to 11.1, p = 0.00); OR of equipment failure (OR = 6, 95% CI 1.7 to 20.9, p = 0.01) andfluid overload (OR 5.2, 95% CI 1.5 to 18.4, p = 0.01) greater with laser• Vaporising electrode versus TCRE – Odds of ‘difficult’ surgery higher with TCRE (OR = 0.25, 95% CI 0.09 to 0.73, p = 0.01); with TCRE fluid deficit greater (WMD = 258 ml,95% CI 173.9 to 342.1, p = 0.00); duration of surgery longer with TCRE (WMD = 1.5 minutes, 95% CI 0.35 to 2.65, p = 0.01)• Balloon versus RB – With RB, amenorrhoea more likely at 12 months (OR 0.55, 95% 0.31 to 0.99, p = 0.05) and 36 months (OR = 0.5, 95% CI 0.25 to 0.97, p = 0.04) notsignificantly different at 24 and 60 months. Greater likelihood of repeat surgery with RB at 24 months (OR = 0.35, 95% CI 0.12 to 0.99, p = 0.05) but effect not seen at 12 and36 months. At 5 years, odds of satisfaction greater with RB (OR = 0.13, 95% CI 0.02 to 0.94, p = 0.04) but not at other years• Vesta versus TCRE – Duration of procedure longer for TCRE (WMD = 16.2 minutes, 95% CI 12.9 to 19.6, p = 0.00). Women with Vesta more likely to have LA (OR = 20.5,95% CI 10.7 to 39.3, p = 0.00)• Microwave versus TCRE – Odds of haemorrhage higher with TCRE (OR = 0.14, 95% CI 0.02 to 0.8, p = 0.03). Odds of equipment failure higher with microwave (OR = 4.07,95% CI 1.1 to 15.0, p = 0.03)• HTA versus RB – HTA more likely to have LA (OR 2.85, 95% 1.6 to 5.1, p = 0.00) and less likely to have haematometra (OR = 0.18, 95% CI 0.03 to 0.93, p = 0.04) but morelikely to have abdominal pain at 2 weeks (OR 1.85, 95% CI 1.1 to 3.1, p = 0.02) and less likely to have nausea vomiting after surgery (OR 3.7, 95% CI 1.5 to 9.0,p = 0.01)• 2nd- versus 1st-generation techniques overall – 1st generation takes longer (WMD = –10.6, 95% CI –18.6 to –2.5, p = 0.01) and has better chance of amenorrhoea at12 months (OR = 0.76, 95% CI 0.6 to 1.0, p = 0.04). More chance of equipment failure with 2nd generation (OR 4.1, 95% CI 1.1 to 14.9, p = 0.03) and LA (OR = 7.6, 95%CI 1.1 to 52.7, p = 0.04). NB: text and graph data disagree• Adverse effects: 2nd generation techniques less likely to have cervical lacerations (OR = 0.08, 95% CI 0.01 to 0.49, p = 0.01), hematometra (OR = 0.14, 95% CI 0.04 to 0.57,p = 0.01), haemorrhage (OR = 0.14, 95% CI 0.02 to 0.80, p = 0.03). 1st-generation techniques less likely to have nausea and vomiting (OR = 2.94, 95% CI 1.52 to 5.70, p = 0.00)• Assessment of heterogeneity: Significant heterogeneity found when comparing 1st- and 2nd-generation techniques overall for use of LA and time taken for procedure. Randomeffects model confirmed significant differences between the techniquesMethodological comments• Search strategy: OK• Participants: OK• Inclusion/exclusion criteria: All methods of ablation were included; in many cases this leads to only one trial for each intervention• Quality assessment of studies: Good• Method of synthesis: Good – dichotomous data outcomes pooled unless ratio of mean to SD

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 7Included controlled study detailsReference and design Intervention Subjects Outcome measures• Author: Cooper et al.,1999 86• Study design: RCT• Recruitment dates: Sept.1996–Feb. 1998• Setting: Single UKgynaecologicaloutpatient dept• Treatment: MEA.Control TCRE bycombinationelectrocauterytechnique – fundusand cornual regionsablated with RB• Surgeon experience: 2surgeons with at least50 prior TCREs, MEAtraining and at least5 MEAs• Surgery pretreatment:3.6 mg goserelin 5weeks prior to op.• Type of anaesthesia:100% GA• Total number of patients: 263randomised, 129 assigned MEA(123 received), 134 assignedTCRE (132 received)• Indication for surgery: DUB• Inclusion criteria:Premenopausal women,completed their families,uterine size equiv. to 10 weekpregnancy or less, gaveinformed consent• Exclusion criteria:Histopathological abnormalitiesof endometrium• Participant characteristics: Meanage, MEA 41.1 years (SD 6.7),TCRE 41.0 years (SD 8.4).Described their periods asheavy – 83 (65%) MEA, 80(60%) TCRE. 60% in botharms had the problem for 3+years, fibroids >2 cm in 14(11%) MEA, 18 (14%) TCRE• Primary and secondaryoutcome measures used:Primary – patients’satisfaction with andacceptability of procedures.Secondary effect onmenstrual status, healthrelatedQoL, operativedetails and morbidity• Method of assessingoutcomes: Patientquestionnaire including QoLmeasure SF-36, operatingdetails reported by surgeonquestionnaire. Bleeding andpain score calculated using afive-point scale• Length of follow up:12 monthsMEATCREpreop. postop. preop. postop. 95% CIResults: (n = 129) (n = 116) (n = 134) (n = 124) for difference (p)• SymptomsAmenorrhoea – 46 (40%) – 49(40%) –14 to 20 (0.23)Irregular periods 66 (51%) – 76 (57%) – –3–7 days bleeding 58 (45%) 49 (42%) 54 (40%) 51 (41%) –11 to 13 (0.23)>7 days bleeding 70 (54%) 6 (5%) 80 (60%) 9 (7%) –17 to 35 (0.23)>3 days heavy bleeding 88 (69%) 8 (7%) 82 (64%) 7 (6%) –10 to 31 (0.79)Dysmenorrhoea 91 (73%) 24 (20%) – 90 (68%) 22 (18%) – –11 to 20 (0.62)same/worsesame/worse2× sanitary protection 111 (86%) 14 (12%) 113 (84%) 16 (14%) –17 to 21 (0.98)Bleeding score 27 (22–36) 3 (0–8) 27 (21–34) 3 (0–10) –3.2 to1.2 (0.37)Pain score 19 (11–26) 1 (0–9) 16 (7–25) 1 (0–7) –2.7 to 1.8 (0.7)Bloating 107 (87%) 75 (65%) 115 (87%) 63 (51%) 1 to 26 (0.03)Breast discomfort 94 (76%) 64 (55%) 103 (79%) 61 (49%) –6 to 18 (0.64)Irritability 105 (86%) 67 (58%) 117 (87%) 65 (52%) –6 to 19 (0.4)Headaches 89 (75%) 56 (48%) 93 (72%) 54 (44%) –7 to 17 (0.46)Depression 71 (57%) 42 (36%) 79 (61%) 49 (40%) –9 to 17 (0.5)>2 days work absence 46 (36%) 4 (3%) 49 (37%) 8 (7%) –Menstruation unchanged – 9 (8%) – 11 (9%) –14 to 26 (0.98)or worsecontinued109© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7preop. postop. preop. postop. 95% CI(n = 116) (n = 116) (n = 124) (n = 124) (ANCOVA p)• SF-36 score, mean (SD)Physical functioning 84.6 (19.2) 0.7 (18.9) 82.2 (23.3) 2.4 (16.8) –6.4 to 2.9 (0.58)Social functioning 60.1 (23.0) 20.6 (26.5) 60.1 (22.9) 16.2 (24.4) –2.1 to 10.9 (0.12)Role – physical 56.5 (42.2) 23.9 (49.4) 62.9 (41.7) 11.3 (41.7) –1.0 to 24.3 (0.03)Role – emotional 61.8 (42.5) 17.0 (48.5) 62.6 (43.2) 13.7 (47.9) –9.1 to 15.6 (0.38)Mental health 44.3 (22.6) 6.3 (19.5) 63.8(21.7) 6.0 (22.2) –4.9 to 5.7 (0.83)Energy/fatigue 63.6 (18.8) 12.8 (21.7) 43.3 (24.3) 12.1 (23.0) –4.9 to 6.5 (0.58)Pain 55.4 (28.2) 14.8 (31.0) 63.7(26.1) 7.2 (31.1) –0.2 to 15.5 (0.54)General health 69.7 (21.7) 2.4 (20.3) 73.0 (19.4) –2.9 (20.0) – 0.2 to10.5 (0.06)MEA (n = 116) TCRE (n = 124) 95% CI (p)• SatisfactionTotally or generally satisfied 89 (77%) 93 (75%) –12 to 17 (0.88)Cure or acceptable improvement 91 (78%) 94 (76%) –11 to 18 (0.76)Treatment acceptable 109 (94%) 112 (90%) –11 to 35 (0.34)Would recommend treatment 105 (91%) 110 (89%) –16 to 25 (0.8)• Operation details (n = 129) (n = 134)Mean operating time (minutes) (SD) 11.4 (10.5) 15.0 (7.2) –5.7 to 1.4 (0.001)Mean theatre time (minutes) (SD) 20.9 (11.3) 26.2 (8.7) –7.7 to 2.8 (

Health Technology Assessment 2004; Vol. 8: No. 3Reference and design Intervention Subjects Outcome measures• Authors: Bain et al.,2002 87• Study design: RCT• Recruitment dates: Notstated• Setting: One UK hospitalobstetric andgynaecologicaldepartment• Treatment: MicrowaveEATCRE control usingRB at the fundus andcornual areas• Surgeon experience:Two surgeons eachwith 50 TCREexperience, trainingand 5 MEAs• Surgery pretreatment:Subcutaneousgoserelin 3.6 mg5 weeks beforeoperation• Type of anaesthesia:GA• Total number of patients: 263(129 MEA, 134 TCRE)• Indication for surgery: Referredby gynaecological dept for EA• Inclusion criteria: Benignendometrial histological samplewithin 6 months, uterine size≥ 10 week pregnancy. Womenwith fibroids and irregularcavities not excluded.• Exclusion criteria:Perimenopausal (FSH >30 U/l),adnexal pathology, furtherpregnancy contemplated• Participant characteristics: MEAmean age 41.4 years (SD 5.4),TCRE mean age 42.2 years (SD5.8). For baseline SF-36measures see below. TCRE hadsignificant (p = 0.03) higherpain than MEA group atbaseline• Primary and secondary outcomemeasures used: Satisfaction,acceptability of menstrualimprovement. QoL, furthersurgery• Method of assessing outcomes:Satisfaction, acceptability ofmenstrual improvement bydirect questioning. SF-36 forQoL. Subsequent surgeryfrom hospital database.Bleeding and pain scoresobtained using a 5-point scalefor each day of period,maximum score 50• Length of follow-up: hospitalreview at 4 months. Mailfollow-up at 12 and24 monthsMEA (n = 120) TCRE (n = 129)95% CI forResults: preop. postop. preop. postop. difference (p)• SymptomsIrregular periods 60 (50%) n/s 70 (54%) n/s –>7 days bleeding 64 (53%) n/s 74 (57%) n/s –>3 days heavy bleeding 81 (67.5%) 2 (2%) 81 (63%) 7 (5%) 9 to 1.3% (p = 0.33)Dysmenorrhoea 84 (70%) 22 (18%) – 83 (64%) 29 (22%) – –14 to 5% (p = 0.78)same/worsesame/worse2× or more sanitary 103 (86%) 9 (14%) 109 (84%) 17 (22%) –13 to 2% (p = 0.36)protection Median – MedianMean bleeding score 28.1 (SD 9.4) 1 (0,6 25th, 27.8 (SD 9.1) 3 (0,10 25th, –1 to 0 (p = 0.06)75th percentile)75th percentile)Mean pain score 18.9 (SD 11.4 ) 0 (0,7 25th, 16.4 (SD 12.4) 1 (0, 8 25th, 0 to 0 (p = 0.22)75th percentile)75th percentile)Unchanged or heavier – 8 (7%) 14 (11%) –11 to 3% (p = 0.10 )amenorrhoea – 57 (47%) 53 (41%) –9 to 15% (p = 0.19)(n = 120) (n = 120) (n = 129) (n = 129)postop. Change in score postop. Change in score 95% CI (p)• QoLSF-36 scoreMean (SD)Physical functioning 83.9 (19.8) 2.3 (21.3) a 82.5 (22.9) 0.9 (20.4) –3.8, 6.6 (0.28)Social functioning 59.9 (22.6) 10.1 (27.5) b 60.4 (22.8) 6.2 (23.7) c –2.5,10.3 (0.33)Role – physical 56.1 (43.1) 18.5 (53.7) b 63.7 (41.4) 6.1 (43.8) –0.2,24.6 (0.06)Role – emotional 61.3 (42.3) 17.8 (47.5) b 63.0 (42.9) 4.2 (40.1) a –3.6, 23.5 (0.17)Metal health 63.3 (18.8) 6.0 (21.6) c 63.3 (20.8) 4.1 (19.8) c –3.3, 6.9 (0.44)Energy/fatigue 43.6 (22.6) 11.4 (25.1) b 43.3 (24.4) 11.8 (22.6) b –6.4, 5.5 (0.90)Pain 55.7 (28.3) 13.5 (31.7) b 63.4 (26.0) 3.0 (29.8) 2.9, 18.2 (0.02)General health 70.2 (21.6) 0.0 (24.4) 73.0 (19.2) –2.9 (19.0) –2.5, 8.4 (0.29)(change frombaseline significant)continued111© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7MEA (n = 120) TCRE (n = 129)95% CI forResults: preop. postop. preop. postop. difference (p)• SatisfactionCompletely or generally satisfied 79% 67% 7 to 22 (0.02)Recommend to friend 90% 90%Menstrual loss acceptable 96% 88% 0.6 to 14 (0.03)• Further surgeryHysterectomy rate 11.60% 12.7%Laparoscopy plus hysteroscopy 2 2Diagnostic hysteroscopy 1 1Repeat ablation 0 0• Adverse effectsPregnancy 1 0Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: Uncertain• Method of randomisation: By telephone with secretary opening the next in a series of sealed, opaque, sequentiallynumbered envelopes showing treatment code, determined by computer-generated random number squares• Power calculation?: A sample size of 80% power to detect a 15% absolute difference in treatment satisfaction at a 5%significance level (p < 0.05)• All patients given same intervention? Yes• Loss to follow-up?: Yes: 249/263 LTFU at 2 years• Method of data analysis: Analysis by ITT, continuous variables with normal distribution analysed using independent andpaired t-tests, Mann-Whitney U-test for ordinal or non-parametric continuous variables. Independent nominal data wereanalysed using chi-squared or Fischer’s exact test. Paired categorical data that were related or consisted of dichotomousvariables were analysed with Wilcoxon’s signed rank test and McNemar’s test, respectivelyGeneral comments• Generalisability: High• Inter-centre variability: Not applicable• Conflicts of interest: Microsulis provided equipment and part-time financial support for one author to undertake theresearchn/s, not significant.a p < 0.05.b p < 0.001.c p < 0.01.112

Health Technology Assessment 2004; Vol. 8: No. 3Reference and design Intervention Subjects Outcome measures• Authors: Microsulis,2002 88• Study design: RCT• Recruitment dates: April2000 – Sept. 2001• Setting: 8 sites in UK andUSA• Treatment: MEA RB• Surgeon experience:Not stated• Surgery pre-treatment:single leuprolideacetate depot3–5 weeks prior toprocedure.• Type of anaesthesia: At7 centres (data herewith one siteremoved) – MEA GA37%, i.v. sedation62%, regional 185• Exclusion criteria: Not stated• Participant characteristics: 22%of patients had fibroids

Appendix 7Intervention MEA (n = 215) Comparison RB (n = 107)Results: preop. postop. preop. postop. p• Operation details: (n = 209) (n = 106)Anaesthesia time 39.26 (SD 25.44) 47.10 (SD 23.4) 0.007Anaesthesia time (excluding the 41.67 (SD 26.21) 50 (SD 22.96) 0.009study with all GA)Treatment time 3.45 (SD 1.02) 20.26 (15.60) 0.000• Further surgeryRepeat ablation 0 0 –Hysterectomy 1 1• Adverse effects – – –Methodological comments• Prospective?: Not stated• Consecutive patients enrolled?: Not stated• Method of randomisation 2: 1 ratio of MEA to RB treatments. Methods of allocation and concealment not stated• Power calculation?: None stated• All patients given same intervention?: Not stated. All receive same pretreatment• Loss to follow-up?: 13 (6%) MEA and 9 (8%) RB patients LTFU• Method of data analysis: ITT data supplied only for amenorrhoea and treatment success measures, otherwise evaluablepatient data given only. Subgroup analyses are given for women with and without fibroids, cavity length and BMI>30 kg/m 2General comments• Generalisability: Low. Few details of patient characteristics given and no exclusion criteria given• Main outcome measured independently: Yes• Inter-centre variability: Amenorrhoea rates between centres were assessed and showed a significant difference betweentreatments in only 1 of 8 studies. One study gave all patients GA and data about anaesthetic are provided with andwithout study included• Conflicts of interest: Conducted by the manufacturer of MEA as part of their application for FDA approval in the USA.Unpublished, therefore not peer reviewedBMI, body mass index.114

Health Technology Assessment 2004; Vol. 8: No. 3Reference and design Intervention Subjects Outcome measures• Authors: Bongers et al.,2000 89• Study design: Prospectivecohort study comparingTBEA and TCRE• Recruitment dates: Allwomen undergoingTCRE in 1992–4, TBEA1995–7• Setting: General teachinghospital in TheNetherlands• Treatment: TBEA(Thermachoice) TCRE• Surgeon experience:Not stated• Surgery pretreatment:TCRE group GnRH for8–12 weeks. TBEAD&C prior toprocedure• Type of anaesthesia:GA or spinalanaesthetic• Total number of patients: 152(77 TBEA; 75 TCRE)• Indication for surgery:Menorrhagia unresponsive tomedical treatment• Inclusion criteria: Patientdecided on EA, eligible resultsof hysteroscopy andendometrial sampling• Exclusion criteria: Uterussounded >12 cm, a separateuterus, submucous fibroids,intrauterine adhesions.• Participant characteristics:TBEA:Age 42.5 years (SD 6.3)Loss of clots 0 (0–10)Total endometrial thickness8 (3–37)Uterine length 8.8 (SD 1.1)TCRE:Age 43.2 years (SD 6.4)Loss of clots 4 (0–10)Total endometrial thickness 8.5(1–32)Uterine length 8.1 (SD 1.5)• Primary and secondary outcomemeasures used:Surgical re-interventionDuration of menstruation,dysmenorrhoea, patients’satisfaction at 3, 6, 12 and24 months.• Method of assessing outcomes:During a 20-minuteoutpatient visit. Satisfactionon a 4-point scale – perfectlysatisfactory, satisfactory, notreatment effect, complaintsworsened• Length of follow-up: 24 monthsIntervention TBEA (n = 77) Comparison TCRE (n = 75)95% CI forResults: preop. postop. preop. postop. difference (p)• SymptomsDysmenorrhoea – – – 3 (4%) aNone 38/77 31/75 No differencein baseline groupsMild 28/77 30/75Moderate 1/77 2/75Severe 1/77 0Unknown 9/77 12/75Mean duration of 8 (1–30) 8 (2–25) TCRE shorter at 3menstruation months (p = 0.01)days (range)No differencedetected at 6, 12or 24 months• Bleeding patternsAmenorrhoea –3 months 17% 36%6 months 15% 22%12 months 16% 26%24 months 13% 17%Hypomenorrhoea 68/77 – 65/75 –Polymenorrhoea 2/77 – 0/75 –Metrorrhagia 7/77 2 (3%) a 10/75 7 (9%) aMenorrhagia – 7 (9%) a – 9 (12%) a• QoLcontinued115© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7Intervention MEA (n = 215) Comparison RB (n = 107)Results: preop. postop. preop. postop. p116• SatisfactionPerfectly satisfied:3 months 51 (66%) 60 (80%) No differences between6 months 39 (63%) 39 (57%) the groups, however,12 months 40 (63%) 30 (52%) significant interaction24 months 28 (60%) 20 (43%) between changesSatisfied:over time – satisfaction3 months 15 (20%) 8 (11%) decreased (p = 0.001)6 months 6 (10%) 5 (7%) and decrease stronger12 months 8 (13%) 1 (2%)0 in TCRE than TBEA24 months 2 (4%)0 3 (6%)0No treatment effect:3 months 8 (10%) 6 (8%)06 months 10 (16%) 24 (35%)12 months 6 (10%) 21 (37%)24 months 5 (11%) 8 (17%)Complaints worsened:3 months 3 (4%) 1 (1%)06 months 7 (11%) 1 (1%)012 months 9 (14%) 5 (9%)024 months 12 (25%) 16 (34%)• Operation detailsProcedure abandoned 8 (10%) 13 (17%) RR 1.7(converted to TCRE) (9 no distention, (95% CI 0.73 to 3.8)4 technical problems)Learning curve effect None None• Further surgeryYear 1 TCRE – 4 (5%)Year 1 Hysterectomy 8 (10%) 12 (16%)Year 2 TCRE – 4 (5%)Year 2 Hysterectomy 9 (12%) 15 (20%)3 year cumulative 13% 26% p = 0.11re-intervention rate RR 0.36(95% CI 0.05 to 2.5)• Adverse effects – Perforation 1(hysterectomy next day)>2000 ml intravasation 3 (4%)>1000 ml intravasation 20 (26%)Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: Yes for TCRE, uncertain for TBEA• Method of randomisation: None• Power calculation?: Yes – assuming a 9% re-intervention rate after TCRE, a series of 150 patients would be needed toshow balloon ablation was equally effective. Note: this is only stated in the abstract, not in the body of the text• All patients given same intervention?: Uncertain – likely that other aspects of care changed over time. In the first half of thestudy, women undergoing TBEA were treated for 8 minutes, whereas those in the second half were treated for 16 minutes• Loss to follow-up?: Uncertain – this is not stated but TBEA group has no satisfaction data for 29 patients (38%) at24 months, TCRE for 28 (37%)• Method of data analysis: Differences in baseline tested using chi-squared or Student’s t-test. Tested for learning curveeffect by looking at the number of totally ablated endometrium at cases 1–20, 2–21, 3–22, etc. Analysis on ITT basis(although see data above – not all outcomes are reported as ITT, and some outcomes only percentages are given, makingit impossible to calculate ITT). Kaplan–Meier curves constructed for re-interventions and compared using log-rank test.RR for re-intervention calculated using Cox regression analysis, univariate and multivariate. Repeated measures if variance(ANOVA) was used to establish time effects, treatment effect and time by treatment effect. For repeated measure data,patients with missing measurements included if they had data for at least 2 data points. Differences considered significantat p < 0.05 level. Student’s t-test used to examine differences between groups at specific timesGeneral comments• Generalisability: Moderate• Main outcome measured independently: Yes• Inter-centre variability: N/A• Conflicts of interest: None statedRR, relative risk; ANOVA, analysis of variance.a These are postoperative symptoms leading to surgical re-intervention.

Health Technology Assessment 2004; Vol. 8: No. 3Reference and design Intervention Subjects Outcome measures• Authors: Brun et al.,2002 96• Study design: RCT• Recruitment dates:Aug. 1999–Oct 2001• Setting: 7 centres inFrance• Treatment: TBEA(Cavaterm)Control TCRE• Surgeon experience:TCRE experiencedsurgeons• Surgery pretreatment:Conventional D&Cperformed just beforeoperation start• Type of anaesthesia:Not stated• Total number of patients: 51randomised(29 Cavaterm, 21 TCRE, 1 LTFU)• Indication for surgery:Menorrhagia ormenometrorrhagia (>80 PBAC)• Inclusion criteria: Normal uterus,no wish for future pregnancy, noclinical suspicious malignancy• Exclusion criteria: Menopause• Participant characteristics: TBEA:mean age 45.5 years (SD 6.04,range 35–59), mean weight69 kg (SD 16.1, range 43–105),mean height 164 cm (SD 5.6,range 153–175)TCRE: mean age years 46.7 (SD6.0, range 33–46), mean weight68.6 kg (SD 20.9, range 42–111),mean height 160.9 cm (SD 5.6,range 145–168)• Primary and secondary outcomemeasures used: Bleedingstatus, satisfaction, adverseeffects• Method of assessing outcomes:Bleeding recording cardbased on Higham and Janssen(PBAC). Patient assessed owncondition at follow-up,satisfaction rated as‘excellent, good, moderate orbad’• Length of follow-up: 3 monthsTBEATCRE95% CI forResults: preop. n = 29 postop. preop. n = 21 postop. difference• SymptomsMenorrhagia 13 (45%) 12 (57%)Meno-metrorrhagia 14 (48%) 7 (33%)Metrorrhagia 2 (7%) 2 (10%)Bleeding score At 3 months At 3 monthsMean (SD) 459 (237) 44 (48) 273 (107) 75 (78)Median (range) 365 (132–1000) 33 (0–154) 266 (81–467) 64 (0–259)Mean change (SD) –413 (242) –199 (157) In both groups(95% CI) (–284, –542) (–107, –289) significantly changefrom baseline(p < 0.001)• QoL – – – – –• SatisfactionExcellent 12 (41%) 8 (38%)Good 15 (52%) 11 (52%)Moderate 2 (7%) 2 (9%)• Operation details• Further surgery – –• Adverse effectsBurns in the cervical channel1 (3%) 0Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: Unclear• Method of randomisation: 1:1 randomisation done centrally (does this tally with 29:21 patients in the 2 groups?)• Power calculation?: 80 patients per treatment arm needed for a 15% difference in efficacy with a power of 90% at the 5%significance level. This study reports on

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Gervaise et al.,1999 90• Study design: Controlledstudy. Controls takenfrom records of TCREpatients during sametime period as theintervention group• Recruitment dates: Nov.1994–April 1998• Setting: Single centre inFrance• Treatment: Thermalballoon ablation(Thermachoice)TCRE using 1.5%glycine• Surgeon experience:Not stated• Surgery pretreatment:None• Type of anaesthesia: LAused where medicallynecessary, or desiredby patient in TBEAgroup – 28 (38%)TBEA• Total number of patients: 147(73 BEA, 74 TCRE)• Indication for surgery: Abnormaluterine bleeding• Inclusion criteria: 40+ years,excessive menstrual blood loss(as measured by no. ofpads/cycle), premenopausalwomen had to have failedmedical therapy (progestins) orbe unwilling to continue withthem, postmenopausal womenwere not willing to discontinueHRT• Exclusion criteria: Fibroids,polyps, premalignant lesions,uterine cavity >12 cm, thosewishing to retain fertility• Participant characteristics:TBEA: Age 46.3 ± 1.3 years(34–66); menopausal status 5(6.8%); parity 2.4 ± 0.3 (0–9);pads/cycle 86 ± 40.4;anteverted: retroverted 59:14;uterine cavity depth 8.9 ± 0.3(6–12)TCRE: Age 47.4 ± 1.4 years(34–65); menopausal 20 (27%);parity 1.9 ± 0.2 (0–4);pads/cycle 81 ± 41.7;anteverted: retroverted 63:11;uterine cavity 9.1 ± 0.2 (7–12)Differences in parity andmenopause significantTCRE• Primary and secondary outcomemeasures used: Amenorrhoeaor eumenorrhoea orhypomenorrhoea.Elimination of dysmenorrhoea• Method of assessing outcomes:Telephone interview• Length of follow-up: TBEAmedian 18.3 ± 2.7 (range3–44) months,TCRE median 19.2 ± 2.3(range 3–36) monthsn = 73 n = 44 n = 74 n = 47Immediate 24 months Immediate 24 monthsResults: preop. postop. preop. postop. pAt 24 months• SymptomsAmenorrhoea 18 (24.7%) 16 (36.4%) 28 (37.8%) 18 (38.3%) n/sHypomenorrhoea 16 (21.9%) 7 (15.9%) 23 (31.1%) 13 (27.7%) n/sEumenorrhoea 28 (38.4%) 15 (34.1%) 10 (13.5%) 8 (17.0%) 0.0006Menorrhagia 8 (11.0%) 4 (9.1%) 9 (12.2%) 7 (14.9%) n/sMetrorrhagia 3 (4.1%) 2 (4.5%) 4 (5.4%) 1 (2.1%) n/s• QoL – – – – –• Satisfaction – – – – –118continued

Health Technology Assessment 2004; Vol. 8: No. 3• Operation detailsMean operating time 20.3 minutes 44.8 minutes

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Meyer et al.,1998 82• Study design: RCT• Recruitment dates:Jan.–Sept. 1996• Setting: 12 centres inUSA and Canada• Treatment:Thermachoice thermalballoonControl – RB• Surgeon experience: Allhad “extensiveexperience of RB EA”• Surgery pretreatment:None• Type of anaesthesia:GA TBA 53%,RB 84%• Total number of patients: 275• Indication for surgery:Menorrhagia• Inclusion criteria: Aged30+ years, premenopausal,have normal Pap smear andendometrial biopsy within thepast 6 months, 3 months’documented history of HMB,failed medical therapy, uterinecavity sounded between 4 and10 cm, no further desire forchildbearing, willing tocontinue with contraceptionfor 3 years postablation• Exclusion criteria: Women withsubmucous myomas orsuspected genital tract infectionor malignancy, those who hadundergone previous EA• Participant characteristics:mean (SD) range:TBEA – Age 40.2 years (4.9)30–51; BMI 24.0 (6.5)14.4–52.7; age at onset ofmenorrhagia 29.6 years (9.7)10.0-47.0; years withmenorrhagia 9.9 (8.5)0.5–37.0; uterine cavity8.6 cm (1.1) 4.0–10.0.RB – Age 40.9 years (5.2)29–50; BMI 22.9 (5.5)15.7–39.6; age at onset ofmenorrhagia 29.8 years (9.6)11.0–49.0; years withmenorrhagia 10.0 (8.9)1.0–35.0; uterine cavity8.6 cm (1.2) 4.0–10.5• Primary and secondary outcomemeasures used: Satisfaction,menstrual bleeding, PMS,ability to work outside thehome• Method of assessing outcomes:MB measured throughpictorial diary system – scoresof >150 for menorrhagia.Examination at 3, 6 and12 months• Length of follow-up: 12 monthsIntervention TBEAComparison RBpreop. postop. preop. postop.Results: (n = 128) (n = 125) (n = 117) (n = 114) p(n = 245 – completed 6-month follow-up)• SymptomsPMS 115 (89.8%) 106 (90.6%)PMS moderate/severe 101 (78.6%) a 41 (32.8%) 90 (76.6%) a 33 (29.0%) a 0.05Mild 52 (40.6%) Same 37 (31.6%) SameModerate 45 (35.2%) 31(24.8%) 26(22.8%)Severe Increased 54 (46.2%) Increased

Health Technology Assessment 2004; Vol. 8: No. 3Intervention TBEAComparison RBpreop. postop. preop. postop.Results: (n = 128) (n = 125) (n = 117) (n = 114) pScore decreased – 77 (61.6%) – 78 (68.4%)by 90%≥ 50% reduction – At least – At least112 (90%+) 103 (90%+)Haemoglobin 12.7 (±1.4) 12.5 (±1.6)values (g/dl)Reduction in the ~75 (~60%) ~68 (~60%)number of womenwith anaemiaMenorrhagia has At least 4 (3.2%) At least 2 (1.8%)severe impact 90 (70%+) a 82 (70%+)• QoLInability to work 51 (39.8%) 5 (4.0%) 45 (38.5%) 3 (2.7%)outside the home• Satisfaction (n = 125) (n = 114)Very satisfied 107 (85.6%) 99 (86.7%)Satisfied 13 (10.4%) 14 (12.4%)Not satisfied 5 (4.0%) 1 (0.9%)• Operational detailsProcedure time

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Graingeret al., 2000 84• Study design: RCT• Recruitment dates:Jan.–Sept. 1996• Setting: 14 universityaffiliatedor privatepractice centres inUSA and Canada• Treatment:Thermachoice thermalballoon. Control – RBelectrosurgical ablation• Surgeon experience: Allexperienced in RB andtrained in balloonablation• Surgery pretreatment:No drugpretreatment.3-minute curettageusing 5 mm curetteprior to ablation• Type of anaesthesia:Not stated• Total number of patients: 255• Indication for surgery: Excessivemenstrual bleeding• Inclusion criteria: Aged 30+ years,premenopausal, documentedhistory of 3 months’ HMB(measured by pictorial diarysystem as 80 ml or more), uterinecavity between 6 and 10 cm, nofurther fertility desired, continuecurrent contraception for 3 years• Exclusion criteria: Women withsubmucous myomas, suspectedgenital tract infection ormalignancy, history of EA• Participant characteristics: Nonestated• Primary and secondary outcomemeasures used: Amount ofuterine bleeding, effect onQoL. Secondary complicationsand adverse effects• Method of assessing outcomes:Pictorial diary system forbleeding, questionnaire forother menstrual symptoms,impact on life and satisfactionwith treatment. Adverseeffects documented andrecorded.• Length of follow-up: 48 months.Following-up telephonecontact at 24 h, examined at1 week, 3 and 6 months, 1and 2 yearsThermal BalloonRBpreop. postop. preop. postop.(n = 131) (n = 122) (n = 124) (n = 105) 95% CI forResults: 1 yr 2 yr 1 yr 2 yr difference• SymptomsNo PMS n/s 34 36 32 37(27.2%) (29.2%) (28.1%) (35.2%) n/sPMS moderate or severe 103 (78.6%) 41 35 95 (76.6%) 33 31(32.8%) (28.6%) (29.0%) (29.5%) n/sUnable to work outside 52 (39.8%) 5 1 48 (38.5%) 3 3home (4.0%) (0.8%) (2.7%) (2.9%) n/sMean menstrual diary score At 1 year decreased At 1 year decreasedby 85.5% by 91.7% n/spostop. yr 1 postop. yr 2 postop. yr 1 postop. yr 2• Satisfaction 85.60% 105 (86.1%) 86.70% 91 (86.7%) n/sVery satisfied 10.40% 12 (9.8%) 12.40% 12 (11.4%) n/sNot satisfied 4% 5 (4.1%) 0.90% 2 (1.9%) n/sRecommend procedure 119 (97.5%) 103 (99%) n/sAge (years)Age (years)Menstrual symptoms at 2 years 40 40Amenorrhoea 13 (11%) 8 (15%) 19 (18%) 26 (25%) n/sSpotting 18 (15%) 13 (11%) 23 (22%) 14 (13%) n/sHypomenorrhoea 44 (36%) 55 (45%) 43 (41%) 31 (30%) n/sEumenorrhoea 30 (25%) 26 (21%) 13 (12%) 22 (21%) n/sMenorrhagia 16 (13%) 11 (9%) 8 (8%) 13 (12%) n/s• Further surgery 4 (3%) 11 (8.9%)• Adverse effects 1 (0.8%) pregnancy2.5 years after ablationcontinued122

Health Technology Assessment 2004; Vol. 8: No. 3Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: Uncertain• Method of randomisation: Randomised by blocks in 1:1 allocation• Power calculation?: Assuming an 85% response rate for RB, 108 evaluable patients per treatment required to detect ifballoon therapy was at least 20% less effective ( =0.05, 90% power)• All patients given same intervention?: Yes• Loss to follow-up?: 16 discontinued before 1 year 227/255 on study at 2 years• Method of data analysis: Paired t-tests, chi-squared probabilities and a repeated measures analysis of variance to comparedemographics and outcomes. For most variables, numbers are not given so it is not possible to check whether ITT hasbeen done; this seems unlikely. One variable at 1 year is definitely not ITT [no PMS at 1 year (n = 34) 27.2% at 2 years(n = 35) (29.2%) TBEA; no PMS at 1 year (n = 32) 28.1%, at 2 years (n = 37) 35.2% (RB)]General comments• Generalisability: Poor• Main outcome measured independently: Unclear – questionnaires used• Inter-centre variability: Not examined• Conflicts of interest: Supported by Gynaecare123© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Loffer, 2001 85• Study design: RCT• Recruitment dates:Jan.–Sept. 1996• Setting: 12 US and 2Canadian university andprivate practice sites.• Treatment: TBEA• Surgeon experience: Allinvestigators wereexperienced in RB andtrained in TBEA• Surgery pretreatment:No drugs used. Timed3-minute suctioncurettage given to allprior to ablation• Type of anaesthesia:LA, LA with sedationand GA. GA morefrequent with RB• Total number of patients: 275enrolled. 255 treated underprotocol (131 TBEA, 124 RB).At 3 years data available forTBEA 114 and RB 100• Indication for surgery:Menorrhagia• Inclusion criteria: Aged30+ years, premenopausal,normal Pap smear andendometrial biopsies, at least3 months’ documented historyof excessive bleedingunresponsive to medicaltherapy measured byminimum threshold score ondaily pictorial record ofbleeding, normal uterinecavity, 4–10 cm sound, nodesire for further fertility,willing to continue for 3 yearson current contraceptive• Primary and secondary outcomemeasures used: Patientsreported menstrual flow.Also menstrual symptoms,adverse effects, impact ofmenorrhagia on QoL• Method of assessing outcomes:Validated pictorial diarymethodPatient questionnaire• Length of follow-up: Telephonecontact within 24 h.Examined at 1 week, 2, 6 and12 months. Interviewed at 2and 3 years• Exclusion criteria: Submucousmyomas, suspected genitalurinary tract infection ormalignancy, those withprevious ablation• Participant characteristics:None statedTBEARBpreop. postop. preop. postop.at 3 years at 3 years 95% CIResults: (n = 114) (n = 99) difference• SymptomsAmenorrhoea 17 (14.9%) 26 (26.3%)Spotting 11 (9.6%)0 16 (16.2%)Hypomenorrhoea 45 (39.5%) 26 (26.3%)Eumenorrhoea 33 (29.0%) 25 (25.2%)Menorrhagia 8 (7.0%) 6 (6.0%)(n = 137) (n = 114) (n = 138) (n = 99)No PMS symptoms 12 (8.8%) 36 (31.6%) 12 (8.7%) 37 (37.4%)• QoL (n = 137) (n = 114) (n = 138) (n = 99)Menorrhagia having 96 (70.3%) a 2 (1.8%) 108 (78.6%) a 2 (2.0%)Severe impact on lifeModerate impact 38 (28.1%) a 9 (7.9%) 28 (20.5%) a 8 (8.0%)Minor impact 2 (1.6%) a 103 (90.3%) 1 (0.9%) a 89 (90.0%)(n = 136) (n = 112) (n = 136) (n = 98)Not able to workoutside the home 54 (39.7%) 5 (4.5%) 57 (41.9%) 5 (5.1%)124continued

Health Technology Assessment 2004; Vol. 8: No. 3• Satisfaction(n = 114) (n = 100)Very satisfied or satisfied 109 (95.6%) 97 (94%)• Further surgery (n = 114) (n = 99)9 (7.9%) 14 (14%)(1 repeat EA, hysterectomies8 hysterectomies)• Adverse effects 0 2 (1.6%)Fluid overload (Perioperative)1 (0.8%) cervical lacerationPostoperative 3 (2.3%) 1 (0.8%) each endometritis,possible endometritis hematometra, postablation1 (0.8%) UTI sterilisation syndromeMethodological comments• Prospective?: Yes• Consecutive patients enrolled?: Uncertain• Method of randomisation: Using a 1:1 allocation ratio at each centre• Power calculation?: Assuming RB response rate of 85%, 108 women required in each arm to provide a 0.9 power todetect if the test procedure is at least 20% less effective at preventing menorrhagia ( = 0.05)• All patients given same intervention?: Yes• Loss to follow-up?: Yes – 20 patients were randomised and not entered into the study – 11 withdrew voluntarily, 8 werenot eligible and 1 RB aborted because of uterine perforation secondary to cervical dilation. At 3 years, 17 from the TBEAand 24 from the RB group were LTFU• Method of data analysis: ITT not performedGeneral comments• Generalisability: Low• Main outcome measured independently: No. Using patients’ completed pictorial diaries• Inter-centre variability: Not reported• Conflicts of interest: Supported by Gynaecarea In a number of cases, only percentages, not actual numbers, are provided in the text. Actual numbers have been calculatedusing this percentage of the number of people reported as followed up (n in the table). In a number of cases, the resultantnumber is uncertain. For those marked with a it is not possible to ascertain a whole number of people from the data given.The number provided is the nearest estimate. It is suspected that additional missing data for individual variables have beenexcluded without comment (changing the denominator) causing this anomaly.125© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Loffer andGrainger, 2002 94• Study design: RCT• Recruitment dates:1996–7• Setting: 14 NorthAmerican centres, 12 ofwhich provided data forthis 5-year follow-up,which was not plannedin original protocol• Treatment: Thermalballoon(Thermachoice)Control group – RB• Surgeon experience: Allexperienced with RBablation and trained inTBEA• Surgery pretreatment:3-minute suctioncurettage• Type of anaesthesia:Not stated• Total number of patients:255 treated (131 TBEA, 124RB), 147 (76 TBEA, 71 RB)follow-up for 5 years but 122(61 TBEA, 61 RB) analysed forbleeding patterns – thoseundergoing repeat procedureexcluded• Indication for surgery:Menorrhagia• Inclusion criteria: Desiring nofuture fertility• Exclusion criteria: menopause,evidence of cervical or uterinemalignancy, uterine anatomicalabnormalities• Patient characteristics: At studyrecruitment mean age TBEA40.4 years, RB 40.9 years. At5-year follow-up mean ageTBEA 45.7, RB 46.1 year. BMI,duration menorrhagia beforesurgery, uterine size similarbetween groups• Primary and secondaryoutcome measures used:Menstrual status,dysmenorrhoea, pelvic pain,satisfaction, additionalgynaecological treatmentsand conditions• Method of assessing outcomes:Patient questionnaireadministered by telephone bythe physician’s office.Bleeding status self-reportedas none, spotting, light,normal or excessive. Severityof dysmenorrhoea and pelvicpain or cramping notassociated with mensesreported as none, mild,moderate or severe. Successwas calculated as the numberof women with normal orless bleeding without furtherprocedure (successes)divided by successes plus allknown treatment failures(excessive bleeding or repeatprocedure at 5 years)• Length of follow-up: 5 years(±3 months)TBEA TBEA RB RBat 3 years at 5 years at 3 years at 5 yearsResults: (n = 14) (n = 61) (n = 98) (n = 61) p• SymptomsAmenorrhoea 17 (15%) 14 (23%) 26 (26%) 20 (33%)Spotting 11 (10%) 6 (10%) 16 (16%) 7 (11%)Hypomenorrhoea 45 (39%) 23 (38%) 26 (26%) 15 (25%)Eumenorrhoea 33 (29%) 15 (25%) 25 (25%) 17 (28%)Menorrhagia 8 (7%) 3 (5%) 6 (6%) 2 (3%)Dysmenorrhoea:None 52% 52%Mild 21% 26%Moderate 21% 13%Severe 5% 8%Non-menstrual pelvic pain:None 42 (69%) 49 (80%)Mild 13 (21%) 7 (11%)Moderate 3 (5%) 5 (8%)Severe 3 (5%) 0Success 58/85 (68%) 59/85 (69%) 0.87• SatisfactionSatisfied with procedure 57 (93%) 61 (100)Of those who had a further 22 (88%)procedure – satisfied(n = 25)126continued

Health Technology Assessment 2004; Vol. 8: No. 3TBEARBat 5 yearsat 5 yearsResults: (n = 61) (n = 61) p• Further surgeryBetween year 3 andyear 5 follow-up:Hysterectomy 13 7Repeat ablation 2 2D&C 0 1At 5 years follow-up:Hysterectomy 21 21Repeat ablation 3 2D&C 0 1Reason for hysterectomy: (n = 21) (n = 21)Bleeding 9 (43%) 7 (33%)Pelvic pain 3 (14%) 10 (48%)Bleeding and pelvic pain 5 (24%) 1 (5%)Myomas 1 (5%) 1 (5%)Ovarian cysts 1 (5%) 0Mood swings /depression 0 1 (5%)Uterine prolapse 2 (9%) 0Endometrial hyperplasia 0 1 (5%)Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: Not stated• Method of randomisation: 1:1 allocation• Power calculation?: None stated• All patients given same intervention?: Yes but techniques may vary between centres• Loss to follow-up?: 53/131 (40%) TBEA, 53/124 (43%) RB LTFU. The paper also excludes from analysis of outcomes afurther 25 patients (10%) who underwent a repeat procedure between years 3 and 5• Method of data analysis: Descriptive statistics. Logistic regression performed using a stepwise selection for gravidity, parity,baseline Higham score, uterine position, years of menorrhagia, sound measurement, procedure duration, age and BMI.No characteristic strongly predicted treatment outcomeNote that 6/14 patients reporting amenorrhoea at 5 years were over 50 and/or experiencing hot flushesData for dysmenorrhoea have been extracted from presented graph; data in the text do not concur with the graph –indicating much less moderate to severe dysmenorrhoea than shownGeneral comments• Generalisability: Moderate – baseline characteristics not provided although they are reported in other papers relating tothis trial• Main outcome measured independently: Uncertain• Inter-centre variability: None stated• Conflicts of interest: Supported in part by Gynaecare127© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Pellicano et al.,2002 92• Study design: RCT• Recruitment dates: May1998–June 1999• Setting: Single centre inItaly• Treatment: TBEA(Cavaterm)Control: TCRE + RB(2.7% sorbitol and0.54% mannitoldistention solution. RBfor corneal area,fundus and isthmus)• Surgeon experience:“Proficient” in TCRE• Surgery pretreatment:TCRE group depotGnRH (Enantone 3.75)6 and 2 weeks beforesurgery.No pretreatment priorto TBEA• Type of anaesthesia:Spinal anaesthesia• Total number of patients:96 randomised (50 TCRE,46 TBEA)82 treated (42 TCRE,40 TBEA)• Indication for surgery:Menorrhagia unresponsive tomedical treatment• Inclusion criteria:

Health Technology Assessment 2004; Vol. 8: No. 3TCRE/RB (n = 42) TBEA (n = 40)Results: preop. postop. preop. postop. p• SymptomsIrregular periods 26 (62%) 24 (60%) n/sPeriod >7days 33 (79%) 34 (85%) n/sCycle

Appendix 7TCRE/RB (n = 42) TBEA (n = 40)Results: preop. postop. preop. postop. p• Adverse effectsIntraoperative:Fluid overload 5 (12%) – n/sCervical tear 1 (2%) – n/sConversion to 2 (5%) – n/shysterectomy (dueto severe uterineperforation)Postoperative pain: VAS (SD) 3.8 (±0.6) 3.2 (±0.7) n/sPostoperativeFever 2 (5%) 1 (2.5%) –UTI/retention 1 (2%) 0 –Haemorrhage 4 (10%) 5 (12.5%) –Blood transfusions – 2 (5%) –Pain at discharge (VAS) 1.5 (±0.6) 1.9 (±0.3) 0.01Pain at 3 days (VAS) 0.5 (±0.2) 0.4 (±0.1) n/sPain at 7 days (VAS) 0 0 n/sUrinary incontinence at 3 (9%) 2 (6%)2 years (n = 33, 35)CIN grade 1 (year 2) 1 (3%) 1 (3%)Postoperative vaginal bleeding (days) 7.8 (±1) 5.2 (±1.8) 0.05Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: All invited to participate• Method of randomisation: Computer-generated random number sequence• Power calculation?: No• All patients given same intervention?: Yes• Loss to follow-up?: 105 eligible patients consented, 9 withdrew before randomisation. 96 randomised and 14 refusedallocated treatment (8/50 TCRE, 6/46 TBEA, 15%). 4 TCRE and 3 TBEA LTFU at 1 year (7%) and 9 TCRE and 5 BEALTFU at 2 years (15%). Total LTFU = 28/96 (29%)• Method of data analysis: ITT not used. Test for differences in characteristics between the groups using 2-tailed Student’st-test for unpaired data, preoperative basal differences using Students t-test for paired data. Chi-squared test used forpostoperative details and satisfaction between the groups. Wilcoxon rank sum test for operative times, blood loss,duration of symptoms, discharge timeGeneral comments• Generalisability: High• Main outcome measured independently: Yes• Inter-centre variability: N/A• Conflicts of interest: Surgical equipment supplied by Wolf Germany and Wallsten Medical130

Health Technology Assessment 2004; Vol. 8: No. 3Reference and design Intervention Subjects Outcome measures• Authors: Romer, 1998 83• Study design: ProspectiveRCT• Recruitment dates: Notgiven• Setting: Not given• Treatment: Thermalballoon (Cavaterm) vsRB ablation• Surgeon experience:Not given• Surgery pretreatment:2× 4 weekly injectionsof GnRH (leuprolide3.75 mg) operationperformed 2 weeksafter injection• Type of anaesthesia:GA for bothinterventions• Total number of patients: 20(10 intervention, 10 control)• Indication for surgery:Recurrent therapy refractorymenorrhagia (not assessed)• Inclusion criteria: Menorrhagia• Exclusion criteria: Internaluterine cavity length 10 cm,incomplete family planning,intrauterine abnormalities,myomas, glandular–cystic,adenomyosis hyperplasia,carcinoma• Participant characteristics: Age,RB 40 (35–50) years, TBEA 42(37–52) years; hormonetherapy attempts, RB 3 (2–5),TBEA 3 (1–6); curettage, RB2.5 (2–4), TBEA 2.2 (2–5)• Primary and secondaryoutcome measures used:SatisfactionBleeding patterns• Method of assessing outcomes:Not stated• Length of follow-up:9–15 monthsInterventionComparison95% CI forResults: preop. postop. preop. postop. difference• Symptoms• Bleeding patternsAmenorrhoea 4 (40%) 3 (30%) DichotomousHypomenorrhoea 5 (50%) 6 (6%) data presentedEumenorrhoea 1 (10%) 1 (1%)Hypermenorrhoea 0 0• QoLNot reported• Satisfaction All patients satisfied with treatment No CI givenoutcome• Operation details –• Further surgery• Adverse effectsNo treatmentfailures reported at9–15 monthsAuthor does acomparison of prosand cons of eachtechnique but notbased on trial dataMethodological comments• Prospective?: Yes• Consecutive patients enrolled?: Not clear• Method of randomisation: Not clear• Power calculation?: Not stated• All patients given same intervention?: 10 given RB; 10 given Cavaterm• Loss to follow-up?: None• Method of data analysis: dichotomous dataGeneral comments• Generalisability: Low• Main outcome measured independently: No• Inter-centre variability: N/A• Conflicts of interest: None stated131© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Soysal et al.,2001 91• Study design: RCT• Recruitment dates: Sept.1997–Feb. 1999• Setting: Universitymedical centre in Turkey• Treatment: TBEAControl – RB ablation,glycine distentionmedium• Surgeon experience: RBperformed by oneexperienced surgeon,TBEA by staffsurgeons orsupervised residents• Surgery pretreatment:Two monthlyinjections of depotGnRH analogue(3.6 mg goserelinacetate)• Type of anaesthesia:All TBEA LA, all RBGA• Total number of patients: 96(48 TBEA, 48 RB)• Indication for surgery: Myomainducedmenorrhagia• Inclusion criteria: Age40+ years, completedchildbearing, PBACdocumented menorrhagia,myomatous uterus diagnosedby ultrasound examination,uterine size 12 weeks or less,at clinical evaluation or 380 mlor less at ultrasound or amyoma less than 5 cmdiameter. All patients had aphysical exam., diagnostichysteroscopy, suction biopsyand cervical smear• Exclusion criteria: Active PID,any submucous myoma largerthan 3 cm or with 150= menorrhagia) at 3, 6 and12 months. Haemoglobinvalues recordedpreoperatively and at12 months.Operating time (frominsertion of operating tool toremoval), intraoperativecomplications, postoperativepain score recorded 12 hafter surgery using 10-pointlinear pain score. Successdefined as eumenorrhoea orPBAC

Health Technology Assessment 2004; Vol. 8: No. 3• SatisfactionNot very satisfied 33% 39% n/s• Operation detailsOperation time (minutes) 11.5 ± 0.8 37.3 ± 7.5 0.0001• Further surgeryHysterectomy 4 4 n/s• Adverse effectsLinear pain score at 12 h 3.1 ± 1.7 3.2 ± 2.1 n/sIntraoperatively:Fluid overload – 2 0.05Haemorrhage – 2Cervical injury – 1Postoperative:Haematoma 1 2 n/sEndometritis 2 1Methodological comments• Prospective?: Yes• Consecutive patients enrolled?: Uncertain• Method of randomisation: Computer-generated randomisation using opaque, sealed envelopes• Power calculation?: None stated• All patients given same intervention?: Yes• Loss to follow-up?: 96 patients recruited, 3 patients allocated to TBEA lost before procedure, no other LTFU• Method of data analysis: SPSS for tests such as Student’s t-test for independent samples and paired samples, theMann–Whitney U-test, Fisher’s exact test, chi-squared test and others were used. Baseline characteristics give a mean,SD and a range – if the data were believed to be non-parametric, median and range should be given; if not, mean and SDwould sufficeGeneral comments• Generalisability: High• Main outcome measured independently: Yes• Inter-centre variability: N/A• Conflicts of interest: None statedPID, pelvic inflammatory disease.133© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 7Reference and design Intervention Subjects Outcome measures• Authors: Zon-Rabelink,2001 93• Study design: RCT• Recruitment dates: Notstated• Setting: The Netherlands.Number of centres notgiven• Treatment: TBEA,Control RB• Surgeon experience:Not stated• Surgery pretreatment:All patients pretreatedwith zoladex 6 and2 weeks prior tosurgery• Type of anaesthesia:Not stated• Total number of patients: 139(77 TBEA, 62 RB), 2 from RBgroup excluded afterrandomisation• Indication for surgery:Menorrhagia• Inclusion criteria: PBAC score>184, DUB according to TVSand hysteroscopy• Exclusion criteria: None stated• Patient characteristics: Nodifferences found in age,parity, uterine cavity,endometrial thickness and Hband preoperative FSH levels• Primary and secondaryoutcome measures used:PBAC score, adverse effects,success rate, QoL• Method of assessing outcomes:Success defined as PBACscore

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 8Graphs showing sensitivity analyses forMEA and TBEASensitivity analyses are illustrated in Figures 19–27.–13838000–4435801–2661653–1920046–4591070–1920046750978–1649562–19200462576741–4016292–1920046–1920046–2680780–3441515–1226932–1920046–2579349–1285736–1920046–2297936–2727361–1921002–1920046–1919727–1922004–1920046–1910310–1766683–1920046–2288722–1734596–1920046–2057348–1480344–1920046936533–1818221–1920046–2164250–1394942–1920046–10915682–192004684687–16000000–14000000–12000000–10000000–8000000–6000000–4000000–2000000FIGURE 19 Sensitivity analysis: cost per QALY for TBEA versus MEA0200000040000003 years5 years7 yearsModel runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting =100LA = 0LA = 48LA = 100Utility for “well” = 0.75Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for TBEA = 0.5Utility for TBEA = 0.8Utility for TBEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8% recurrence = 21%% recurrence = 31%% recurrence = 51%% repeat menorrhagia having hysterectomy = 0.3% repeat menorrhagia having hysterectomy = 0.6% repeat menorrhagia having hysterectomy = 0.8% of complications lasting >1 month = 0.1% of complications lasting >1 month = 0.5% of complications lasting >1 month = 0.9Inflation factor for complications following repeat EA =1Inflation factor for complications following repeat EA = 2Inflation factor for complications following repeat EA = 4Complication rate = 0.001Complication rate = 0.0023Complication rate = 0.005Death rate = 0Death rate = 0.0002135© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 8–192799–126713–103748–88283–77173–116608–77173–89422–77173–98982–77173–77173–94501–111828–77173–85837–77173–64316–56751–77190–77173–77167–77209–77173–76996–74632–77173–82291–72580–77173–80546–77183–77173–77116–77174–77173–77171–77192–77173–77122–77173–132801–32360–33555–54723 years5 years7 yearsModel runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting = 100LA = 0LA = 48LA = 100Utility for “well” = 0.75Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for MEA = 0.5Utility for MEA = 0.8Utility for MEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8Percentage recurrence overall = 21%Percentage recurrence overall = 31%Percentage recurrence overall = 51%% of those with recurrent menorrhagia who have hysterectomy = 0.3% of those with recurrent menorrhagia who have hysterectomy = 0.6% of those with recurrent menorrhagia who have hysterectomy = 0.8% complications last >1 month = 0.1% complications last >1 month = 0.5% complications last >1 month = 0.9Inflation factor for complications after repeat EA = 1Inflation factor for complications after repeat EA = 2Inflation factor for complications after repeat EA = 4Complication rate = 0.0001Complication rate = 0.0007Complication rate = 0.0023Death rate = 0Death rate = 0.0002–250000 –200000 –150000 –100000 –50000 0136FIGURE 20 Sensitivity analysis: cost per QALY MEA versus TCRE

Health Technology Assessment 2004; Vol. 8: No. 3–260400–181719–147843–125271–109175–155558–109175–62792–104478–109175–145577–109175–109175–122385–135596–97139–109175–120624–109175–91468–80963–109200–109175–109167–109226–109175–108919–105727–109175–116160–101790–109175–114612–109032–109175–110623–109146–109175–109233–108994–109175–109559–109175–185528–310873 years5 years7 yearsModel runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting =100LA= 0LA = 48LA = 100Utility for “well” = 0.75Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for TBEA = 0.5Utility for TBEA = 0.8Utility for TBEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8% recurrence = 21%% recurrence = 31%% recurrence = 51%% repeat menorrhagia having hysterectomy = 0.3% repeat menorrhagia having hysterectomy = 0.6% repeat menorrhagia having hysterectomy = 0.8% of complications lasting > 1 month = 0.1% of complications lasting > 1 month = 0.5% of complications lasting > 1 month = 0.9Inflation factor for complications following repeat EA =1Inflation factor for complications following repeat EA = 2Inflation factor for complications following repeat EA = 4Complication rate = 0.001Complication rate = 0.0023Complication rate = 0.005Death rate = 0Death rate = 0.0002–300000 –250000 –200000 –150000 –100000 –50000 0FIGURE 21 Sensitivity analysis: cost per QALY TBEA versus TCRE137© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 8–211314–170041–143038–388700–573844–388700–52075–446207–388700–178310–421090–388700–388700–470051–551402–183920–388700–4293763 years5 years7 yearsModel runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting = 100LA = 0LA = 48LA = 100–2194686Utility for “well” = 0.75–388700–305031–260220–388766–388700–388678–388835–388700–388028–382486–388700–401056–35978–388700–410055–386290–388700–412185–388245–388700–38615–384085–388700–401729–388700Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for MEA = 0.5Utility for MEA = 0.8Utility for MEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8Percentage recurrence overall = 21%Percentage recurrence overall = 31%Percentage recurrence overall = 51%% of those with recurrent menorrhagia who have hysterectomy = 0.3% of those with recurrent menorrhagia who have hysterectomy = 0.6% of those with recurrent menorrhagia who have hysterectomy = 0.8% complications last >1 month = 0.1% complications last >1 month = 0.5% complications last >1 month = 0.9Inflation factor for complications after repeat EA = 1Inflation factor for complications after repeat EA = 2Inflation factor for complications after repeat EA = 4Complication rate = 0.0001Complication rate = 0.0007Complication rate = 0.0023Death rate = 0402687–2500000 –2000000 –1500000 –1000000 –500000 0 500000 1000000Death rate = 0.0002138FIGURE 22 Sensitivity analysis: cost per QALY MEA versus RB ablation

Health Technology Assessment 2004; Vol. 8: No. 3–764219–649863–567759–506020–710653–5060203 years5 years7 yearsModel runs for 10 yearsCost –50%Cost–301386Cost +50%–485298–506020Procedure time = 20 minsProcedure time = 18.1 mins–16511Procedure time = 42 mins–666618Hospital stay = 0.5 day–506020–506020–564302–622583–452919–506020–556531Hospital stay = 1 dayOffice setting = 0Office setting = 50Office setting = 100LA = 0LA = 48LA = 100–182054Utility for “well” = 0.75–506020–407856–353063–506118–506020–505987–506221–506020–505016–495578–506020–526965–466544–506020–535173–496060–506020–622515–504119–506020–509874–493433–506020–534796–506020Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for TBEA = 0.5Utility for TBEA = 0.8Utility for TBEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8% recurrence = 21%% recurrence = 31%% recurrence = 51%% repeat menorrhagia having hysterectomy = 0.3% repeat menorrhagia having hysterectomy = 0.6% repeat menorrhagia having hysterectomy = 0.8% of complications lasting > 1 month = 0.1% of complications lasting > 1 month = 0.5% of complications lasting > 1 month = 0.9Inflation factor for complications following repeat EA = 1Inflation factor for complications following repeat EA = 2Inflation factor for complications following repeat EA = 4Complication rate = 0.001Complication rate = 0.0023Complication rate = 0.005Death rate = 0621023Death rate = 0.0002–1000000 –800000 –600000 –400000 –200000 0 200000 400000 600000 800000FIGURE 23 Sensitivity analysis: cost per QALY TBEA versus RB ablation139© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 8–2879343 years–2297045 years–192147–1659447 yearsModel runs for 10 years–229241Cost –50%–165944Cost–103728Cost +50%–160130–165944Procedure time = 20 minsProcedure time = 18.1 mins–60831Procedure time = 42 mins–214878Hospital stay = 0.5 day–165944–165944–183703–201461–149764–165944–181335Hospital stay = 1 dayOffice setting = 0Office setting= 50Office setting = 100LA = 0LA = 48LA = 100–348925Utility for “well” = 0.75–165944–141252–126227–165985–165944–165930–166027–165944–165531–160562–165944–176958–153436–165944–175168–165339–165944–171908–165828–165944–166178–165176–165944–167587–165944Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for TBEA = 0.5Utility for TBEA = 0.8Utility for TBEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8% recurrence = 21%% recurrence = 31%% recurrence = 51%% repeat menorrhagia having hysterectomy = 0.3% repeat menorrhagia having hysterectomy = 0.6% repeat menorrhagia having hysterectomy = 0.8% of complications lasting > 1 month = 0.1% of complications lasting > 1 month = 0.5% of complications lasting > 1 month = 0.9Inflation factor for complications following repeat EA = 1Inflation factor for complications following repeat EA = 2Inflation factor for complications following repeat EA = 4Complication rate = 0.001Complication rate = 0.0023Complication rate = 0.005Death rate = 0–371343Death rate = 0.0002–400000 –350000 –300000 –250000 –200000 –150000 –100000 –50000 0140FIGURE 24 Sensitivity analysis: TBEA versus combined TCRE and RB ablation

Health Technology Assessment 2004; Vol. 8: No. 3–274827–286006–124019–177345–124019–27045–14058–124019–32360–153515–124019–124019–147454–170890–65027–124019–135737–124019–104842–93300–124048–124019–124009–124079–124019–123718–120008–124019–132188–115203–124019–130514–123925–124019–124918–124001–124019–124056–123838–124019–124508–124019Model runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting = 100LA = 0LA = 48LA = 100Utility for “well” = 0.75Utility for “well” = 0.9Utility for “well” = 0.95Utility for “well” = 0.99Utility for MEA = 0.5Utility for MEA = 0.8Utility for MEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8Percentage recurrence overall = 21%Percentage recurrence overall = 31%Percentage recurrence overall = 51%% of those with recurrent menorrhagia who have hysterectomy = 0.3% of those with recurrent menorrhagia who have hysterectomy = 0.6% of those with recurrent menorrhagia who have hysterectomy = 0.8% complications last >1 month = 0.1% complications last >1 month = 0.5% complications last >1 month = 0.9Inflation factor for complications after repeat EA = 1Inflation factor for complications after repeat EA = 2Inflation factor for complications after repeat EA = 4Complication rate = 0.0001Complication rate = 0.0007Complication rate = 0.0023Death rate = 0Death rate = 0.0002–350000 –300000 –250000 –200000 –150000 –100000 –50000 0FIGURE 25 Sensitivity analysis: MEA versus combined TCRE and RB ablation141© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 8–36175174370729102410282924101991236724101704273824102410252926482301241025136382410226024102464238024102568273624101761256824102368241424102373241024102408241524102399241023993 years5 years7 yearsModel runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting = 100LA = 0LA = 48LA = 100Utility for “well” = 0.75Utility for “well” = 0.932418 Utility for “well” = 0.95Utility for “well” = 0.99Utility for TBEA = 0.5Utility for TBEA = 0.8Utility for TBEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8% recurrence = 21%% recurrence = 31%% recurrence = 51%% repeat menorrhagia having hysterectomy = 0.3% repeat menorrhagia having hysterectomy = 0.6% repeat menorrhagia having hysterectomy = 0.8% of complications lasting > 1 month = 0.1% of complications lasting > 1 month = 0.5% of complications lasting > 1 month = 0.9Inflation factor for complications following repeat EA = 1Inflation factor for complications following repeat EA = 2Inflation factor for complications following repeat EA = 4Complication rate = 0.001Complication rate = 0.0023Complication rate = 0.005Death rate = 0Death rate = 0.0002–10000 –5000 0 5000 10000 15000 20000 25000 30000 35000FIGURE 26 Sensitivity analysis: cost per QALY for TBEA versus hysterectomy142

Health Technology Assessment 2004; Vol. 8: No. 3–31664527324325462108245821081472221721081710230221082108226224161721210821855582108197721082156208221082246244521081440225821082064210921082097210821082108211121082102210821013 years5 years7 yearsModel runs for 10 yearsCost –50%CostCost +50%Procedure time = 20 minsProcedure time = 18.1 minsProcedure time = 42 minsHospital stay = 0.5 dayHospital stay = 1 dayOffice setting = 0Office setting = 50Office setting = 100LA = 0LA = 48LA = 100Utility for “well” = 0.75Utility for “well” = 0.928316 Utility for “well” = 0.95Utility for “well” = 0.99Utility for MEA = 0.5Utility for MEA = 0.8Utility for MEA = 0.9Utility for menorrhagia = 0.5Utility for menorrhagia = 0.55Utility for menorrhagia = 0.8Percentage recurrence overall = 21%Percentage recurrence overall = 31%Percentage recurrence overall = 51%% of those with recurrent menorrhagia who have hysterectomy = 0.3% of those with recurrent menorrhagia who have hysterectomy = 0.6% of those with recurrent menorrhagia who have hysterectomy = 0.8% complications last >1 month = 0.1% complications last >1 month = 0.5% complications last >1 month = 0.9Inflation factor for complications after repeat EA = 1Inflation factor for complications after repeat EA = 2Inflation factor for complications after repeat EA = 4Complication rate = 0.0001Complication rate = 0.0007Complication rate = 0.0023Death rate = 0Death rate = 0.0002–5000 0 5000 10000 15000 20000 25000 30000FIGURE 27 Sensitivity analysis: cost per QALY MEA versus hysterectomy143© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 9Quality assessment of industry-submittedeconomic analysesAssessment of economic model supplied by Microsulis Medical (basedon the Sculpher framework)1. StructureIs there a clear statement of the decisionproblem, the context and the perspective?Is a theory of the underlying disease detailed?Are the underlying assumptions involved inthe model clearly specified? Are they justified?Are the implications of relaxing theseassumptions described?2. Disease statesIs the chosen model type appropriate forthe time dimension of the disease process?Is a justification of the choice of stateswithin the model provided? If so, does thisaccord with the theory of disease process?Is any empirical evidence provided on thesuitability of the states (e.g. sensitivity tochange in the underlying disease)?Have any important disease states beenomitted from the model?3. OptionsIs there a clear statement of the optionsbeing evaluated?Do these appear to cover the range oflogical and feasible options?The model aims to determine the costs andconsequences of MEA, balloon ablation, RB ablation, RBwith resection, resection only and hysterectomytreatments for menorrhagia in the UK. MEA is thetechnology appraised and is compared to first- andsecond-generation EA techniques. An incremental costeffectivenessanalysis is used to estimate additional costsand benefits of using MEA rather than the othertreatments.Background information is provided about menorrhagiaand existing surgical treatments.Assumes probability of further procedures over timefollows a logarithmic distribution.Two-stage pathway in a decision tree. Initially, there arenine health states (pre-operation, operation, death,complication, convalescence, postoperative,menorrhagia, further surgery and hysterectomy). Stage 2is slightly different for women having MEA than with theother ablation methods, as those with recurrentmenorrhagia have a TCRE/RB procedure orhysterectomy, not a repeat procedure.Not directly but the states do appear to adequatelydescribe the states involved in menorrhagia and itstreatment.No evidence is given although the states do appear tomap the progress of condition.NoYes, the model evaluates first- and second-generation EAmethods.Yes.145© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 94. Time horizonIs the time horizon of the analysis stated?Yes – model duration is 5 years.If so, is this justified in terms of theThis time horizon is justified based on the majorityunderlying disease and the effect ofof further procedures being undergone by the end ofinterventions? year 5.5. Cycle length (if relevant)If relevant, is the cycle length used inthe model stated.Is justification offered on the choice of cyclelength? If so, does the justification relate tothe disease process?Not relevant.No.1466. Data identificationAre the sources of parameter values in themodel clearly stated?Most transition probabilities are from the literature.Cumulative probabilities of repeat resection orhysterectomy at 1, 2 and 3 years following initialresection were obtained from a life table analysis. Theprobability of a further procedure within each year is afunction of the probability of undergoing the procedureat year 1 and a growth rate corresponding to the timesince the initial procedure and follows a logarithmicdistribution.Utilities are taken from published literature. Those formenorrhagia, convalescence and post-convalescence forresection and hysterectomy are taken from a publishedcost–utility analysis 30 derived from a time trade ofanalysis with 60 women with menorrhagia. Owing to thesimilarity of descriptions for convalescence with TCREbeing similar to the other methods of ablation, this valuewas also assigned to them.Resource costs are estimated from the perspective of theNHS in pounds sterling. Theatre overhead costs arecalculated from information received from a singleScottish NHS trust. Source of staff costs is not stated.Other costs come from Chartered Institute of PublicFinance Accountancy (CIPFA) and the RoyalPharmaceutical Society of Great Britain. Operationdetails are taken from the literature.The cost of TBEA equipment is taken from the full listprice, plus cost of umbilical cable. All other proceduresare assumed to require standard operating equipment,which is assumed to be included in the theatreoverheads.Is reasonable empirical justification, from No. Most data come from the literature.earlier iterations of the model, offered that these The utility values for post-convalescence are calculated asdata are optimal?the ratio of ‘bleeding and pain’ scores for eachprocedure and TCRE. The ‘bleeding and pain’ score wasthe summation of the proportion of women withamenorrhoea and with dysmenorrhoea at 12 months,based on data in RCTs. This method of calculating autility score is not sourced or justified. In addition,amenorrhoea may not be the best measure of success asmany women do not seek this as a treatment aim. Thosewho do may be more likely to seek hysterectomy forHMB.

Health Technology Assessment 2004; Vol. 8: No. 3For the first iteration of the model, hassatisfactory justification been offered that dataare based on a search of all the low-cost datasources (e.g. MEDLINE, DARE, Cochranelibrary)?Are ranges specified for parameters?Is there evidence to suggest selective use ofdata?If some parameter estimates are based onelicitation of expert opinion, have the methodsused for this purpose been adequately described(e.g. inclusion criteria, sample size, elicitationmethods)?Are the claims made about the modelresults tempered by the limitations of the data?7. Data incorporationFor each parameter value, is there clear andreasonable justification of how data havebeen incorporated into the model?The utility calculation gives a low post-convalescencevalue for TBEA which has relatively low levels ofamenorrhoea, 0.57, while the other EA methods rangefrom 0.73 to 0.79 and hysterectomy 0.86. In addition,this utility value of 0.57 for TBEA, 0.73 for RB andTCRE and TCRE alone and 0.74 for RB ablation duringpost-convalescence, is lower than the figure of 0.76 thatthese methods all receive during convalescence, which iscounter-intuitive. It would be expected that utility ofconvalescence was lower than that for post-convalescence(‘well’).There was no indication in the literature to ascertain theduration of recurrent menorrhagia prior to undergoinga repeat procedure. In the base case analysis it wastherefore assumed that a woman would havemenorrhagia for 50% of the time between the end ofconvalescence and the time of a further procedure. Nojustification for this figure is given.Yes. MEDLINE and EMBASE were searched for relevantliterature. Search limits were RCTs, English language,published after 1994 and human studies.Yes.Some. It is assumed that 50% of women undergoingTBEA and MEA received LA in an office setting and67% of the remaining women had LA in an operatingtheatre. Although this latter figure is based in publishedevidence, 67 unpublished evidence from the same centrehas concluded that post-operative pain and nausea makeMEA unsuitable as an outpatient, rather than day-case,procedure. In addition, the estimation of first-generationprocedures undertaken under LA is 14%, taken from aUK RCT. 102 However, this may be an underestimate assystematic review 9 evidence showed that 23% of womenundergoing RB ablation had LA. This will underestimatethe cost of MEA and TBEA compared with firstgenerationtechniques.Not applicable.The authors discuss limitations of the data available forseveral parameters.In the absence of post-convalescence utility values, thevalue available for resection was multiplied by a factorrepresenting relative severity of bleeding and pain. Thiswas calculated by summing the proportion of womenwith amenorrhoea and the proportion of women withdysmenorrhoea at 12 months. No reference is given forthis technique, which gives a value of 0.57 for balloonablation and of 0.79 for MEA owing to the relatively low147© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 9Has a stochastic analysis been undertaken?If so, do the distributions in parametervalues reflect second-order uncertainty?Have appropriate distributions been selectedfor each parameter?Have interval rates been translated intotransition probabilities using the appropriateformula?If appropriate, has a half-cycle correctionbeen applied to adjust time-related estimatesin the model?8. Internal consistencyIs there a statement about the tests ofinternal consistency that were undertaken?9. External consistencyAre any relevant studies and/or modelsidentified by the analyst for purpose ofcomparison?Have any comparisons of the outputs of themodel with independent external sourcesbeen reported?If so, are the conclusions justified? Havediscrepancies been investigated and explained?level of amenorrhoea with TBEA and therefore biases infavour of MEA.Women experiencing repeat menorrhagia are assumedto spend half the time between with the postconvalescenceutility value and half with the value formenorrhagia.Uncertainty has been examined by one- and two-waysensitivity analyses and a Monte Carlo simulation wasused to vary all parameter simultaneously. Parametersvaried are listed and the range used for each given.Triangular distribution is used in Monte Carlosimulation.Not applicable.Not applicable.Not applicable.Not applicable.No statement is made about tests of internal consistencythat were undertaken.No.No.Not applicable.Quality assessment of economic analysis supplied by makers ofThermachoice (using the Drummond framework)148Was a well-defined question posed in answerableform?Was a comprehensive description of competingalternatives given?Yes, the comparison is between thermal balloon, TCREand hysterectomy.The viewpoint of the analysis is not stated. Cost data aretaken from the French healthcare system and are notcomprehensive. A 3-year time horizon is taken, whichmay underestimate re-intervention rates and bias theanalysis in favour of EA.Competing alternatives are described, although someaspects of care are not included in the comparison.

Health Technology Assessment 2004; Vol. 8: No. 3Was the effectiveness of the programme or servicesestablished?Were all important costs and consequencesidentified?Were costs and consequences measuredaccurately in appropriate units?Were costs and consequences valued credibly?Were costs and consequences adjusted fordifferential timing?Was an incremental analysis performed?Was allowance made for uncertainty in theestimates of costs and consequences?Did the presentation and discussion of resultsinclude all issues of concern to users?Effectiveness data are taken from the report of 3-yearfollow-up in the Meyer trial. Estimates for effects are notcalculated on an ITT basis and no account is taken ofthe precision of results. For example, the difference inamenorrhoea between thermal balloon and TCRE wasnot statistically significant.No.Costing study was acknowledged as not beingcomprehensive, focusing on surgical component.Outcome measurement in relation to EA is discussedelsewhere in this assessment report.Resources were identified and costed in the Frenchhealthcare system – some difficulty in extrapolating theseto the UK. Base year for costings not stated.Consequences are reasonably maintained in naturalunits.No, although time horizon is short (3 years).Yes.No – a major shortcoming of the analysis.No. The analysis is acknowledged to be limited.Quality assessment of economic analysis supplied by the makers ofCavaterm (using the Sculpher framework)1. StructureIs there a clear statement of the decisionproblem, the context and the perspective?Is a theory of the underlying disease detailed?Are the underlying assumptions involved inthe model clearly specified? Are they justified?The comparisons are clearly stated. The perspective isnot well defined but is predominantly that of the NHS,and in particular the secondary care sector. However,number of days absent from work is included, whichincorporates an element of patient or societalperspective.The condition process is described elsewhere in theindustry submission to NICE and is relatively simple.The treatment pathway is clearly described. The model'sbaseline is current practice, i.e. the proportion of womenreceiving each of the competing technologies. Thecurrent utilisation of different second-generationtechniques was estimated from expert opinion. Notjustified (methods not stated).All second ablations are repeats of the originaltechnique. Justified – unlikely that women will move toanother ablation technique and no information on thisavailable.149© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 9150Are the implications of relaxing theseassumptions described?2. Disease statesIs the chosen model type appropriate forthe time dimension of the disease process?Is a justification of the choice of stateswithin the model provided?If so, does this accord with the theory ofdisease process?Is any empirical evidence provided on thesuitability of the states (e.g. sensitivity tochange in the underlying disease)?Have any important disease states beenomitted from the model?3. OptionsIs there a clear statement of the options beingevaluated?Do these appear to cover the range of logicaland feasible options?4. Time horizonIs the time horizon of the analysis stated?If so, is this justified in terms of theunderlying disease and the effect ofinterventions?It is assumed that all women who undergo anunsuccessful second ablation will have hysterectomy. Thiswill represent a slight overestimate of the number of womeneventually undergoing hysterectomy. It is likely that somewomen will reject hysterectomy for a variety of reasons. Thisgroup may have a further ablation or continue with medicaltreatment. Some will reach the menopause before hysterectomy iscarried out. The increase in the number of hysterectomiesperformed for failure of ablation will bias the model againstablation.The sensitivity analysis examines the effect of relaxingassumptions regarding differential effectiveness ofsecond-generation technologies, using different sourcesof effectiveness data and varying other key inputs in onewaysensitivity analyses. The impacts of relaxing morefundamental assumptions regarding the treatmentpathway are not explored.The time horizon of 3 years is justified as the extent ofcurrent data from RCTs. However, a longer timeframemay be appropriate given the importance of the failurerate and its potential relationship with time beyond thisperiod.The modelling approach does not permit a cost–utilityanalysis.The modelling approach does not allow for thedifferential timing of events and associated discounting.Yes.Not relevant.No.No.Yes.Yes.Yes – 3 years.No. The average age of women in the RCTs of EA was42 years. Since the menopause occurs on average around10 years later and failure rates may be time dependent,it is likely that the 3-year time horizon may haveunderestimated cumulative failure rate.

Health Technology Assessment 2004; Vol. 8: No. 35. Data identificationAre the sources of parameter values in themodel clearly stated?Is reasonable empirical justification, fromearlier iterations of the model, offered thatthese data are optimal?For the first iteration of the model, hassatisfactory justification been offered that dataare based on a search of all the low-cost datasources (e.g. MEDLINE, DARE, Cochranelibrary)?Are ranges specified for parameters?Is there evidence to suggest selectiveuse of data?If some parameter estimates are based onelicitation of expert opinion, have the methodsused for this purpose been adequately described(e.g. inclusion criteria, sample size, elicitationmethods)?Are the claims made about the model resultstempered by the limitations of the data?Yes.No – this is the first iteration of the model.Yes. The model is informed by a review of theeffectiveness of the technologies concerned.Yes.Possibly.No – as noted above.Not in all cases. The assumption that Cavaterm is moreeffective than the alternative balloon ablationtechnology, Thermachoice, is given undue weight giventhe nature of the underlying empirical data. This comesfrom an indirect comparison, based on trials carried outon small numbers of women over different follow-uptimes. Failure rates are similar for the two technologiesat 12 months.Some sweeping claims for Cavaterm are made, forexample, relating to the complete replacement ofexisting technologies with Cavaterm and potentialimpact on operating theatre time and bed days. It isunlikely that such a complete technological transferwould be achieved because (a) some women will have astrong preference for hysterectomy, based on their highvaluation of amenorrhoea over eumenorrhoea, and (b)not all women with menorrhagia are candidates forballoon ablation owing to variation in uterinemorphology and pathology. Similar claims are made forthe potential impact of Cavaterm use on hospital bedday capacity and the labour market.6. Data incorporationFor each parameter value, is there clear andreasonable justification of how data have beenincorporated into the model?Not in all cases. There is limited justification for thechoice of one source for data over another.Failure rates are acknowledged to be a key parameter.However, the method for incorporating data is weak,mainly because of the way that primary research hasbeen reported. In the industry submission, data fromstudies carried out at different times are combined in ameta-analysis and compared across the different EA151© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 9Has a stochastic analysis been undertaken?If so, do the distributions in parameter valuesreflect second-order uncertainty?Have appropriate distributions been selectedfor each parameter?Have interval rates been translated intotransition probabilities using the appropriateformula?If appropriate, has a half-cycle correctionbeen applied to adjust time-related estimatesin the model?7. Internal consistencyIs there a statement about the tests of internalconsistency that were undertaken?8. External consistencyAre any relevant studies and/or modelsidentified by the analyst for purpose ofcomparison?Have any comparisons of the outputs of themodel with independent external sourcesbeen reported?If so, are the conclusions justified? Havediscrepancies been investigated and explained?technologies. The most appropriate statistical analysiswould be a survival analysis, including time to failure, asthis outcome is likely to be highly time dependent. Suchdata are lacking, which undermines attempts to comparedifferent EA technologies.Yes. The model.No. A uniform distribution for parameter values isassumed in each case.No.Not relevant.Not relevant.No. The model as received does not permit closeexamination of the underlying calculations being carriedout as two key spreadsheets are not included oraccessible.None were available.No.See elsewhere in this assessment report.152

Health Technology Assessment 2004; Vol. 8: No. 3Appendix 10Parameters used in the industry andPenTAG economic modelsParameter PenTAG Microsulis Thermachoice Cavatermvalue value value valueProcedure cost hysterectomy 2096 2644 1778 a 2050Procedure cost TCRE 1110 1129 958 593Procedure cost RB 1190 624 – 593Procedure cost TCRE/RB 1027 545 – 593Procedure cost Cavaterm 826 712 – 584Procedure cost Thermachoice 826 712 905 581Procedure cost MEA 942 674 – 798Success rate following repeat EA – – – 0.5 × thatof primary EAsuccess rateMean cost of a complication following balloon ablation (£) – 770 – –Mean cost of a complication following hysterectomy (£) – 647 – –Mean cost of a complication following MEA (£) – 695 – –Mean cost of a complication following RB+TCRE (£) – 641 – –Mean cost of a complication following RB (£) – 408 – –Mean cost of a complication following resection (£) – 614 – –Discount rate for benefits (% expressed as decimal) 0.015 0.015 – –Discount rate for costs (% expressed as decimal) 0.06 0.06 – –Failure rate 1st-generation EA 0.31 – – 0.1–0.3Failure rate MEA 0.31 – – 0.12Failure rate Thermachoice 0.31 – – 0.14Failure rate Cavaterm 0.31 – – 0.07Probability of hysterectomy following balloon ablation 0.088 0.016 – –at year 1Probability of hysterectomy following MEA at year 1 0.088 0.078 – –Probability of hysterectomy following RB at year 1 0.088 0.015 – –Probability of hysterectomy following RB + TCRE 0.088 0.11 – –at year 1Probability of hysterectomy following resection at 0.088 0.11 – –year 1Probability of hysterectomy following balloon 0.248 0.321 – 0.077ablation at year 5Thermachoice0.0595CavatermProbability of hysterectomy following MEA 0.248 0.208 – 0.0252Probability of hysterectomy following RB 0.248 0.368 – 0.065–0.195Probability of hysterectomy following RB + TCRE 0.248 0.250 – 0.065–0.195Probability of hysterectomy following resection 0.248 0.269 – 0.065–0.195Probability of stopping treatment after failure 0 – – 0.0168following MEAProbability of stopping treatment after failure 0 – – 0.021following TBEAThermachoice0.073 CavatermProbability of stopping treatment after failure 0 – – 0.005–0.015following 1st-generation EAProportion of patients receiving LA 0 0.14 – 0.0.001–0.025(vs GA; RB, RB + TCRE, resection)a Vaginal hysterectomy.continued153© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Appendix 10Parameter PenTAG Microsulis Thermachoice Cavatermvalue value value valueProportion of patients receiving LA (vs GA; MEA, 0.5200 0.6300 – 0.4–0.6 TBEAballoon ablation)0.4–0.0 MEAProportion of MEAs and balloon ablations performed 0 0.500 – –in office (vs theatre)Probability of a surgical complication of balloon ablation 0.0023 0.032 – 0.03–0.04Probability of a surgical complication of hysterectomy 0 0.129 – 0.2–0.5Probability of a surgical complication of MEA 0.0007 0.020 – 0.07Probability of a surgical complication of RB + TCRE 0.0606 0.130 – 0–0.15Probability of a surgical complication of RB 0.0200 0.106 – 0–0.15Probability of a surgical complication of resection – 0.111 – 0–0.15Time required to perform balloon ablation (minutes) 27.400 – 20–30Time required to perform hysterectomy (minutes) 66.500 – 50–135Time required to perform MEA (minutes) 21 20.900 – 20–30Time required to perform RB ablation (minutes) 20 39.600 – 25–36Time required to perform RB + TCRE (minutes) 26.2 28.400 – 25–36Time required to perform resection (minutes) – 51.200 – 25–36Probability of repeat surgery following balloon ablation 0.11 0 – –at year 1Probability of RB+TCRE following MEA at year 1 – 0.009 – –Probability of repeat surgery following RB at 0.11 0 – –year 1Probability of repeat surgery following RB+TCRE 0.11 0.029 – –at year 1Probability of repeat surgery following resection 0.11 0.110 – –at year 1Probability of repeat surgery following balloon ablation 0.31 0.011 – –at year 5Probability of RB + TCRE following MEA at year 5 0.31 0.046 – –Probability of repeat surgery following RB 0.31 0.000 – –at year 5Probability of repeat surgery following RB +TCRE 0.31 0.317 – –at year 5Probability of repeat surgery following resection 0.11 0.127 – –at year 1Duration of complications from balloon ablation (years)

Health Technology Assessment 2004; Vol. 8: No. 3Parameter PenTAG Microsulis Thermachoice Cavatermvalue value value valueUtility in menorrhagia (

Health Technology Assessment 2004; Vol. 8: No. 3Health Technology AssessmentProgrammeMembersPrioritisation Strategy GroupChair,Professor Tom Walley, Director,NHS HTA Programme &Professor of ClinicalPharmacology,University of LiverpoolProfessor Bruce Campbell,Consultant Vascular & GeneralSurgeon, Royal Devon & ExeterHospitalProfessor Shah Ebrahim,Professor in Epidemiologyof Ageing, University ofBristolDr John Reynolds, ClinicalDirector, Acute GeneralMedicine SDU, RadcliffeHospital, OxfordDr Ron Zimmern, Director,Public Health Genetics Unit,Strangeways ResearchLaboratories, CambridgeMembersHTA Commissioning BoardProgramme Director,Professor Tom Walley, Director,NHS HTA Programme &Professor of ClinicalPharmacology,University of LiverpoolChair,Professor Shah Ebrahim,Professor in Epidemiology ofAgeing, Department of SocialMedicine, University of Bristol,Canynge Hall, WhiteladiesRoad, BristolDeputy Chair,Professor Jenny Hewison,Professor of Health CarePsychology, Academic Unit ofPsychiatry and BehaviouralSciences, University of LeedsSchool of Medicine, LeedsProfessor Douglas Altman,Professor of Statistics inMedicine, Centre for Statisticsin Medicine, Oxford University,Institute of Health Sciences,Cancer Research UK MedicalStatistics Group, Headington,OxfordProfessor John Bond, Professorof Health Services Research,Centre for Health ServicesResearch, University ofNewcastle, School of HealthSciences, Newcastle upon TyneProfessor John Brazier, Directorof Health Economics, SheffieldHealth Economics Group,School of Health & RelatedResearch, University ofSheffield, ScHARR RegentCourt, SheffieldDr Andrew Briggs, PublicHealth Career Scientist, HealthEconomics Research Centre,University of Oxford, Instituteof Health Sciences, OxfordDr Christine Clark, MedicalWriter & Consultant Pharmacist,Cloudside, Rossendale, LancsandPrincipal Research Fellow,Clinical Therapeutics in theSchool of Pharmacy, BradfordUniversity, BradfordProfessor Nicky Cullum,Director of Centre for EvidenceBased Nursing, Department ofHealth Sciences, University ofYork, Research Section,Seebohm Rowntree Building,Heslington, YorkDr Andrew Farmer, SeniorLecturer in General Practice,Department of Primary HealthCare, University of Oxford,Institute of Health Sciences,Headington, OxfordProfessor Fiona J Gilbert,Professor of Radiology,Department of Radiology,University of Aberdeen, LilianSutton Building, Foresterhill,AberdeenProfessor Adrian Grant,Director, Health ServicesResearch Unit, University ofAberdeen, Drew Kay Wing,Polwarth Building, Foresterhill,AberdeenProfessor Alastair Gray, Director,Health Economics ResearchCentre, University of Oxford,Institute of Health Sciences,Headington, OxfordProfessor Mark Haggard,Director, MRC ESS Team, CBUElsworth House, Addenbrooke’sHospital, CambridgeProfessor F D Richard Hobbs,Professor of Primary Care &General Practice, Department ofPrimary Care & GeneralPractice, University ofBirmingham, Primary Care andClinical Sciences Building,Edgbaston, BirminghamProfessor Peter Jones, Head ofDepartment, UniversityDepartment of Psychiatry,University of Cambridge,Addenbrooke's Hospital,CambridgeProfessor Sallie Lamb, ResearchProfessor in Physiotherapy/Co-Director, InterdisciplinaryResearch Centre in Health,Coventry University, CoventryDr Donna Lamping, SeniorLecturer, Health ServicesResearch Unit, Public Healthand Policy, London School ofHygiene and Tropical Medicine,LondonCurrent and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)Professor David Neal, Professorof Surgical Oncology, OncologyCentre, Addenbrooke's Hospital,CambridgeProfessor Tim Peters, Professorof Primary Care Health ServicesResearch, Division of PrimaryHealth Care, University ofBristol, Cotham House, CothamHill, BristolProfessor Ian Roberts, Professorof Epidemiology & PublicHealth, Intervention ResearchUnit, London School ofHygiene and Tropical Medicine,LondonProfessor Peter Sandercock,Professor of Medical Neurology,Department of ClinicalNeurosciences, University ofEdinburgh, Western GeneralHospital NHS Trust, BramwellDott Building, EdinburghProfessor Martin Severs,Professor in Elderly HealthCare, Portsmouth Institute ofMedicine, Health & Social Care,St George’s Building,PortsmouthDr Jonathan Shapiro, SeniorFellow, Health ServicesManagement Centre, ParkHouse, Birmingham165© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Health Technology Assessment ProgrammeMembersDiagnostic Technologies & Screening PanelChair,Dr Ron Zimmern, Director ofthe Public Health Genetics Unit,Strangeways ResearchLaboratories, CambridgeDr Paul Cockcroft, ConsultantMedical Microbiologist/Laboratory Director, PublicHealth Laboratory,St Mary’s Hospital,PortsmouthProfessor Adrian K Dixon,Professor of Radiology,Addenbrooke’s Hospital,CambridgeDr David Elliman, Consultant inCommunity Child Health,LondonDr Andrew Farmer, SeniorLecturer in General Practice,Institute of Health Sciences,University of OxfordDr Karen N Foster, ClinicalLecturer, Dept of GeneralPractice & Primary Care,University of AberdeenProfessor Jane Franklyn,Professor of Medicine,University of BirminghamProfessor Antony J Franks,Deputy Medical Director, TheLeeds Teaching Hospitals NHSTrustMr Tam Fry, HonoraryChairman, Child GrowthFoundation, LondonDr Susanne M Ludgate, MedicalDirector, Medical DevicesAgency, LondonDr William Rosenberg, SeniorLecturer and Consultant inMedicine, University ofSouthamptonDr Susan Schonfield, CPHMSpecialised ServicesCommissioning, CroydonPrimary Care TrustDr Margaret Somerville,Director of Public Health,Teignbridge Primary Care Trust,DevonMr Tony Tester, Chief Officer,South Bedfordshire CommunityHealth Council, LutonDr Andrew Walker, SeniorLecturer in Health Economics,University of GlasgowProfessor Martin J Whittle,Head of Division ofReproductive & Child Health,University of BirminghamDr Dennis Wright, ConsultantBiochemist & Clinical Director,Pathology & The KennedyGalton Centre, Northwick Park& St Mark’s Hospitals, HarrowMembersPharmaceuticals PanelChair,Dr John Reynolds, ClinicalDirector, Acute GeneralMedicine SDU, OxfordRadcliffe HospitalProfessor Tony Avery, Professorof Primary Health Care,University of NottinghamProfessor Iain T Cameron,Professor of Obstetrics &Gynaecology, University ofSouthamptonMr Peter Cardy, ChiefExecutive, Macmillan CancerRelief, LondonDr Christopher Cates, GP andCochrane Editor, Bushey HealthCentre, Bushey, Herts.Mr Charles Dobson, SpecialProjects Adviser, Department ofHealthDr Robin Ferner, ConsultantPhysician and Director, WestMidlands Centre for AdverseDrug Reactions, City HospitalNHS Trust, BirminghamDr Karen A Fitzgerald,Pharmaceutical Adviser, Bro TafHealth Authority, CardiffProfessor Alastair Gray,Professor of Health Economics,Institute of Health Sciences,University of OxfordMrs Sharon Hart, ManagingEditor, Drug & TherapeuticsBulletin, LondonDr Christine Hine, Consultant inPublic Health Medicine,Bristol South & West PrimaryCare TrustProfessor Robert Peveler,Professor of Liaison Psychiatry,Royal South Hants Hospital,SouthamptonDr Frances Rotblat, CPMPDelegate, Medicines ControlAgency, LondonMrs Katrina Simister, NewProducts Manager, NationalPrescribing Centre, LiverpoolDr Ken Stein, Senior Lecturer inPublic Health, University ofExeterProfessor Terence Stephenson,Professor of Child Health,University of NottinghamDr Richard Tiner, MedicalDirector, Association of theBritish Pharmaceutical Industry,LondonProfessor Dame Jenifer Wilson-Barnett, Head of FlorenceNightingale School of Nursing& Midwifery, King’s College,London166Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)

Health Technology Assessment 2004; Vol. 8: No. 3MembersTherapeutic Procedures PanelChair,Professor Bruce Campbell,Consultant Vascular andGeneral Surgeon, Royal Devon& Exeter HospitalDr Mahmood Adil, Head ofClinical Support & HealthProtection, Directorate ofHealth and Social Care (North),Department of Health,ManchesterProfessor John Bond, Head ofCentre for Health ServicesResearch, University ofNewcastle upon TyneMr Michael Clancy, Consultantin A & E Medicine,Southampton General HospitalDr Carl E Counsell, SeniorLecturer in Neurology,University of AberdeenDr Keith Dodd, ConsultantPaediatrician, DerbyshireChildren’s Hospital, DerbyProfessor Gene Feder, Professorof Primary Care R&D, Barts &the London, Queen Mary’sSchool of Medicine andDentistry, University of LondonMs Bec Hanley, FreelanceConsumer Advocate,Hurstpierpoint, West SussexProfessor Alan Horwich,Director of Clinical R&D, TheInstitute of Cancer Research,LondonDr Phillip Leech, PrincipalMedical Officer for PrimaryCare, Department of Health,LondonMr George Levvy, ChiefExecutive, Motor NeuroneDisease Association,NorthamptonProfessor James Lindesay,Professor of Psychiatry for theElderly, University of LeicesterDr Mike McGovern, SeniorMedical Officer, Heart Team,Department of Health, LondonDr John C Pounsford,Consultant Physician, NorthBristol NHS TrustProfessor Mark Sculpher,Professor of Health Economics,Institute for Research in theSocial Services, University ofYorkDr L David Smith, ConsultantCardiologist, Royal Devon &Exeter HospitalProfessor Norman Waugh,Professor of Public Health,University of AberdeenCurrent and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)167

Health Technology Assessment ProgrammeMembersExpert Advisory NetworkMr Gordon Aylward,Chief Executive,Association of British Health-Care Industries, LondonMs Judith Brodie,Head of Cancer SupportService, Cancer BACUP, LondonMr Shaun Brogan,Chief Executive, RidgewayPrimary Care Group, Aylesbury,BucksMs Tracy Bury,Project Manager, WorldConfederation for PhysicalTherapy, LondonMr John A Cairns,Professor of Health Economics,Health Economics ResearchUnit, University of AberdeenProfessor Howard Stephen Cuckle,Professor of ReproductiveEpidemiology, Department ofPaediatrics, Obstetrics &Gynaecology, University ofLeedsProfessor Nicky Cullum,Director of Centre for EvidenceBased Nursing, University of YorkDr Katherine Darton,Information Unit, MIND – TheMental Health Charity, LondonProfessor Carol Dezateux,Professor of PaediatricEpidemiology, LondonProfessor Martin Eccles,Professor of ClinicalEffectiveness, Centre for HealthServices Research, University ofNewcastle upon TyneProfessor Pam Enderby,Professor of CommunityRehabilitation, Institute ofGeneral Practice and PrimaryCare, University of SheffieldMr Leonard R Fenwick,Chief Executive, Newcastleupon Tyne Hospitals NHS TrustProfessor David Field,Professor of Neonatal Medicine,Child Health, The LeicesterRoyal Infirmary NHS TrustMrs Gillian Fletcher,Antenatal Teacher & Tutor andPresident, National ChildbirthTrust, Henfield, West SussexMs Grace Gibbs,Deputy Chief Executive,Director for Nursing, Midwifery& Clinical Support Servs., WestMiddlesex University Hospital,Isleworth, MiddlesexDr Neville Goodman,Consultant Anaesthetist,Southmead Hospital, BristolProfessor Robert E Hawkins,CRC Professor and Director ofMedical Oncology, Christie CRCResearch Centre, ChristieHospital NHS Trust, ManchesterProfessor F D Richard Hobbs,Professor of Primary Care &General Practice, Department ofPrimary Care & GeneralPractice, University ofBirminghamProfessor Allen Hutchinson,Director of Public Health &Deputy Dean of ScHARR,Department of Public Health,University of SheffieldProfessor Rajan Madhok,Medical Director & Director ofPublic Health, Directorate ofClinical Strategy & PublicHealth, North & East Yorkshire& Northern Lincolnshire HealthAuthority, YorkProfessor David Mant,Professor of General Practice,Department of Primary Care,University of OxfordProfessor Alexander Markham,Director, Molecular MedicineUnit, St James’s UniversityHospital, LeedsDr Chris McCall,General Practitioner, TheHadleigh Practice, CastleMullen, DorsetProfessor Alistair McGuire,Professor of Health Economics,London School of EconomicsDr Peter Moore,Freelance Science Writer,Ashtead, SurreyDr Andrew Mortimore,Consultant in Public HealthMedicine, Southampton CityPrimary Care TrustDr Sue Moss,Associate Director, CancerScreening Evaluation Unit,Institute of Cancer Research,Sutton, SurreyProfessor Jon Nicholl,Director of Medical CareResearch Unit, School of Healthand Related Research,University of SheffieldMrs Julietta Patnick,National Co-ordinator, NHSCancer Screening Programmes,SheffieldProfessor Chris Price,Visiting Chair – Oxford, ClinicalResearch, Bayer DiagnosticsEurope, CirencesterMs Marianne Rigge,Director, College of Health,LondonProfessor Sarah Stewart-Brown,Director HSRU/HonoraryConsultant in PH Medicine,Department of Public Health,University of OxfordProfessor Ala Szczepura,Professor of Health ServiceResearch, Centre for HealthServices Studies, University ofWarwickDr Ross Taylor,Senior Lecturer, Department ofGeneral Practice and PrimaryCare, University of AberdeenMrs Joan Webster,Consumer member, HTA –Expert Advisory Network168Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)

FeedbackThe HTA Programme and the authors would like to knowyour views about this report.The Correspondence Page on the HTA website(http://www.ncchta.org) is a convenient way to publishyour comments. If you prefer, you can send your commentsto the address below, telling us whether you would likeus to transfer them to the website.We look forward to hearing from you.The National Coordinating Centre for Health Technology Assessment,Mailpoint 728, Boldrewood,University of Southampton,Southampton, SO16 7PX, UK.Fax: +44 (0) 23 8059 5639 Email: hta@soton.ac.ukhttp://www.ncchta.org ISSN 1366-5278