A cost utility analysis of microwave and thermal balloon endometrial ablation techniques for the treatment of heavy menstrual bleeding

Transcription

1 BJOG: an International Journal of Obstetrics and Gynaecology May 2004, Vol. 111, pp DOI: /j x A cost utility analysis of microwave and thermal balloon endometrial ablation techniques for the treatment of heavy menstrual bleeding Ruth Garside, a Ken Stein, a Katrina Wyatt, b Ali Round, a Martin Pitt a Objective To assess the cost effectiveness of the second-generation surgical treatments for heavy menstrual bleeding (microwave and thermal balloon endometrial ablation) compared with existing endometrial ablation techniques (transcervical resection and rollerball, alone or in combination) and hysterectomy. Design A state transition (Markov) cost utility economic model. Population Women with heavy menstrual bleeding. Methods A Markov model was developed using spreadsheet software. Transition probabilities, costs and quality of life data were obtained from a systematic review of effectiveness undertaken by the authors, from published sources, and expert opinion. Cost data were obtained from the literature and from a NHS trust hospital. Indirect comparison of thermal balloon endometrial ablation versus microwave endometrial ablation or either second-generation endometrial ablation method versus hysterectomy, and comparison of second-generation versus first-generation techniques were carried out from the perspective of health service payers. The effects of uncertainty were explored through extensive one-way sensitivity analyses and Monte Carlo simulation. Main outcome measures Incremental cost effectiveness ratios based on cost per quality adjusted life year (QALY) gained, and cost effectiveness acceptability curves. Results Compared with first-generation techniques, both microwave and thermal balloon endometrial ablation cost less and accrued more QALYs. Hysterectomy was more expensive, but accrued more QALYs than all endometrial ablation methods. Baseline results showed that differences between microwave endometrial ablation and thermal balloon endometrial ablation were slight. Sensitivity analyses showed that small changes in values may have a marked effect on cost effectiveness. Probabilistic simulation highlighted the uncertainty in comparisons between different endometrial ablation options, particularly between secondgeneration techniques. Conclusions Despite limitations in available data, the analysis suggests that second-generation techniques are likely to be more cost effective than first-generation techniques in most cases. Hysterectomy, where a woman finds this option acceptable, continues to be a very cost effective procedure compared with all endometrial ablation methods. INTRODUCTION One in 20 women aged consult their GP for heavy menstrual bleeding each year in the UK and while first and second line medical treatments may be effective, many cases require surgical intervention. Transcervical resection and rollerball endometrial ablation (alone or in combination) have been offered as treatments since the a Peninsula Technology Assessment Group (PenTAG), Peninsula Medical School, Universities of Exeter and Plymouth, UK b Research and Development Support Unit, Haldon View Terrace, Exeter, UK Correspondence: Ms R. Garside, Peninsula Technology Assessment Group (PenTAG), Peninsula Medical School, Universities of Exeter and Plymouth, Dean Clarke House, Southernhay East, Exeter, EX1 1PQ, UK. D RCOG 2004 BJOG: an International Journal of Obstetrics and Gynaecology 1980s. Since the 1990s, newer techniques (second generation) such as microwave and thermal balloon ablation have come into practice. An estimated 26,000 hysterectomies and about 16,000 endometrial ablations were performed for menorrhagia in We carried out a cost utility analysis of microwave endometrial ablation and thermal balloon endometrial ablation compared with each other, to first-generation endometrial ablation and to hysterectomy. The perspective of the analysis is the NHS and it was produced to inform policy making. METHODS A systematic review of the effectiveness of endometrial ablation was undertaken by the authors and is reported elsewhere. 2 This included a previous systematic

2 1104 R. GARSIDE ET AL. Fig. 1. Influence diagram for patients undergoing endometrial ablation. review of first-generation techniques of endometrial ablation compared with hysterectomy, 3 and RCTs of microwave endometrial ablation or thermal balloon endometrial ablation compared with first-generation techniques (transcervical resection/rollerball), alone or in combination. No head-to-head comparison of microwave endometrial ablation versus thermal balloon endometrial ablation, or of either technique compared with hysterectomy has been published. Our calculations are therefore based on studies carried out in different populations and should be interpreted with caution. The systematic review included two RCTs of microwave endometrial ablation compared with transcervical resection/rollerball and eight RCTs or non-randomised controlled trials of thermal balloon endometrial ablation compared with transcervical resection/rollerball. The trials included a total of 1561 women, with sample sizes of 20 to 322 (median 143). Overall, there were few differences identified between first- and second-generation techniques of endometrial ablation over a range of outcomes being reported. However, second-generation techniques are quicker to perform and may be safer. to the use of different technologies and the difference in outcome and costs between cohorts yields estimates of cost effectiveness. The endometrial ablation model examines the progress of six hypothetical cohorts of women with heavy menstrual bleeding who are treated separately by either thermal balloon endometrial ablation, microwave endometrial ablation, transcervical resection or rollerball, transcervical resection combined with rollerball endometrial ablation or hysterectomy. The model calculates incremental cost utility (cost per quality adjusted life year [QALY]) between options. A QALY is a year of life adjusted for its value (quality) where a value of one is full health, and a value of zero is death. The value of a year in ill health is therefore weighted according to the severity of the health state relative to perfect health. One year in full health is equivalent to two years spent in a health state with a utility of 0.5. Negative values indicate that a state is considered to be worse than death. Uncertainty was explored using one-way sensitivity analyses and Monte Carlo simulation. Cost effectiveness model A state transition (Markov) model was developed by the authors in Microsoft Excel with the support of a clinical advisory group. In this method, hypothetical cohorts of patients move between discrete health states simulating the clinical course of the condition, with each state having associated costs and quality of life. Transition probabilities, costs and quality of life vary according Fig. 2. Influence diagram for those undergoing hysterectomy.

3 COST UTILITY OF MEA AND TBEA FOR HMB 1105 Table 1. Model assumptions. Assumptions Value Source Justification for source Transition rates between states Background death rate (death) Life Tables UK figures starting age 42 as given in the studies included in this assessment, and increasing year on year. Complications after hysterectomy VALUE study 5 Large UK observational study Death after hysterectomy VALUE study 5 Large UK observational study (direct cause) Complications after TCRE þ RB MISTLETOE study 6 Large UK observational study Death after TCRE þ RB MISTLETOE study 6 Large UK observational study (direct cause) Complications after RB MISTLETOE study 6 Large UK observational study Death after RB (direct cause) 0 MISTLETOE study 6 Large UK observational study Complications due to TCRE alone MISTLETOE study 6 Large UK observational study Death after TCRE alone MISTLETOE study 6 Large UK observational study Complications due to MEA Case series 1433 women 9 Large UK observational study Death after MEA (direct cause) 0 Case series 1433 women 9 Large UK observational study Complications due to TBEA Case series of 5800 women European survey of complications in 5800 women Death after TBEA (direct cause) 0 Systematic review, Garside et al. 2 Systematic review of controlled trial evidence TBEA treatment failure (recurrent menorrhagia) 0.11 RCT 147 women (immediate post-op) 16 Controlled trial. Only data available for immediate post-operative failure rates TBEA treatment failure years 2 and Drug Therapeutic Bulletin Systematic review of controlled trial evidence Proportion of women with recurrent menorrhagia who undergo hysterectomy Proportion of women with recurrent menorrhagia who repeat ablation. Proportion of women with second EA failure who undergo hysterectomy within 6 months Complications after repeat TCRE or rollerball ablation Death after repeat TCRE/RB ablation to First year return of menorrhagia post TCRE/RB Second and third year return of menorrhagia following TCRE/RB First year return of menorrhagia post TBEA/MEA Second and third year return of menorrhagia following TBEA/MEA Length of time spent in different states Median length of complications after hysterectomy Length of convalescence period post hysterectomy Waiting time for hysterectomy, mean (median) Median length of complications following first-generation techniques Median length of complications after second generation techniques year follow up women undergoing TCRE (vs medical management) year follow up women undergoing TCRE (vs medical management) Professional estimate Twice the rate after 1st ablation MacLean-Fraser et al. 7 and professional estimate MISTLETOE study 6 Large UK audit 0.11 Median for trials at 12 months from trials included in systematic review by Garside et al Median at 24 months plus repeat surgery rate from trials included systematic review Garside et al Median at 12 months from trials included in systematic review Garside et al Median at 24 months plus repeat surgery rate from trials included systematic review Garside et al. 2 Long term RCT data for TCRE Long term RCT data for TCRE Comparative case series study of primary and repeat ablations. Only data on complications after repeat ablation. RCT data, best available evidence RCT data, best available evidence RCT data, best available evidence RCT data, best available evidence 2 months Clinician estimate No published data 2 months Lethaby et al. 3 Mean time of return to work/normal activities in systematic review of hysterectomy 94 (54) days HES 2000/01 1 Table 5 Q07 UK data set 1 month Professional estimate No published data 1 month Professional estimate No published data

4 1106 R. GARSIDE ET AL. Table 1. (continued) Assumptions Value Source Justification for source Discount rates Costs 6% HM Treasury Guidance As recommended by NICE Benefits 1.5% HM Treasury Guidance As recommended by NICE Health state utilities Chronic states Menorrhagia 0.55 Sculpher 10 Median value based on interviews with 60 women with menorrhagia Premenopausal following recovery from successful TCRE 0.9 Sculpher 10 Median value based on interviews with 60 women with menorrhagia Premenopausal following recovery from hysterectomy 0.95 Sculpher 10 Median value based on interviews with 60 women with menorrhagia Dead 0 Usual value Temporary states Complications after hysterectomy 0.55 Assumption Same as menorrhagia Hysterectomy 0.63 Assumption One third less than recovery after hysterectomy Convalescence after hysterectomy 0.95 Sculpher 10 Median value based on interviews with 60 women with menorrhagia MEA/convalescence after MEA 0.85 Sculpher 10 Convalescent states post-ablation assumed to be the same for all types of ablation. Based on the Sculpher 10 score for TCRE TBEA/convalescence after TBEA 0.85 Sculpher 10 Convalescent states post-ablation assumed to be the same for all types of ablation. Based on the Sculpher 10 score for TCRE TCRE and RB/convalescence after TCRE and RB 0.85 Sculpher 10 Median value based on interviews with 60 women with menorrhagia TCRE ¼ transcervical resection; TBEA ¼ thermal balloon endometrial ablation; HES ¼ Hospital Episode Statistics; NICE ¼ National Institute for Clinical Excellence; MISTLETOE ¼ Minimally Invasive Surgical Techniques: Laser, Endothermal or Endoresection; RB ¼ Rollerball; MEA ¼ microwave endometrial ablation; RCT ¼ randomised controlled trials; VALUE ¼ Vaginal, Abdominal and Laparoscopic Uterine Excision. Main assumptions Structure of the economic model The structure of the Markov model used is shown in detail in Fig. 1 (pathway for patients undergoing any type of endometrial ablation) and Fig. 2 (pathway for patients undergoing hysterectomy). Health states are shown in boxes, and arrows show the transitions that can occur in each monthly cycle of the model. For example, from hysterectomy, patients can either move to a state of convalescence (recovery from the operation in the absence of complications), have complications or die through direct or other causes. The health states and pathways are the same for all types of endometrial ablation. A cohort of 1000 women eligible for each procedure are modelled for each operation. The starting age of women in the model is 42, based on the median age of women in the clinical trials of endometrial ablation included in our systematic review. 2 The model runs for a total of 10 years. The model assumes that all women become menopausal at age 52, which is the average age of menopause in the UK. Each cycle is one month long. In reality, complications following a second-generation ablation may be experienced for less than one month. The death rate from causes other than procedure is based on values for women in the Life Tables of England and Wales for the years 1998 to 2000 starting at age Clinical procedure assumptions Hysterectomy is abdominal hysterectomy as two-thirds of UK hysterectomies are by this route. 1 Only peri-operative complications are modelled. Recurrent menorrhagia occurs in the first three years after operation. 2 If endometrial ablation of any type fails, repeat ablation or hysterectomy is offered. The model assumes that 90% of women with recurrence will have a repeat procedure, with 60% having repeat endometrial ablation and 30% having a hysterectomy. This further procedure takes place within six months of recurrence. Only one repeat ablation is offered and this is of the same

5 COST UTILITY OF MEA AND TBEA FOR HMB 1107 Table 2. Clinical procedures assumptions. Procedure Data Source Justification Abdominal hysterectomy Length of stay (median) 4 days Local median waiting time (Mid Devon PCT residents) and expert opinion UK data based on all women, uncomplicated menorrhagia will be shorter Day cases 0% HES 2000/01 Table 5 Q07 UK data set Duration of surgery 59 minutes Lethaby et al. 3 Good quality systematic review % under general anaesthetic 100% Assumed First generation endometrial ablation Waiting time, mean (median) 79 (45) days HES 2000/01 Table 5 Q17 UK data set Length of stay, weighted mean 2.0 days Lethaby Good quality systematic review Day cases 60% HES 2000/01 Table 5 Q17 UK data set Duration of surgery TCRE 40.9 minutes Median from trials included in systematic RCT data best available evidence review by Garside et al. 2 Duration of surgery RB 50 minutes Median from trials included in systematic RCT data best available evidence review by Garside et al. 2 Duration of surgery TCRE/RB 31.6 minutes Median from trials included in systematic RCT data best available evidence review by Garside et al. 2 % under general anaesthetic 78% Lethaby and Hickey 14 Systematic review Second generation endometrial ablation Waiting time, mean (median) 80 (50) days HES 2000/01 Table 5 Q16 UK data set Length of stay, mean (median) 1.6 (1) days HES 2000/01 Table 5 Q16 UK data set Day cases 65% HES 2000/01 Table 5 Q16 UK data set Duration of surgery MEA 31.3 minutes Theatre time from trials included in systematic review by Garside et al. 2 in this report Median from RCT data best available evidence Duration of surgery TBEA 18.6 minutes Theatre time from trials included in systematic review by Garside et al. 2 in this report Median from RCT data best available evidence % under general anaesthetic 52% Bain et al. 13 Partially randomised study of LA vs GA among 98 women in the UK EA ¼ endometrial ablation; TCRE ¼ transcervical resection; MEA ¼ microwave endometrial ablation; RCT ¼ randomised controlled trials; LA ¼ local anaesthetic; RB ¼ rollerball; TBEA ¼ thermal balloon endometrial ablation; HES ¼ Hospital Episode Statistics; PCT ¼ primary care trust; GA ¼ general anaesthetic. type as the original. Following this only hysterectomy is available. Based on clinical opinion, it is assumed that 90% of women with recurrence following repeat endometrial ablation have a hysterectomy within six months. There is no convalescence state after ablation as all women are assumed to have fully recovered within one month (the cycle length). Convalescence following ablation is therefore captured in the utility value for the initial operative health state for endometrial ablation. Source of estimates The initial search for this assessment was broad in scope. In populating the model, a hierarchy of evidence was used. Firstly, data from good quality systematic reviews of RCTs were sought. If these were not available, data from good quality individual RCTs were sought. Where these were not available large prospective, observational studies conducted in the UK were used. Finally, if no published evidence could be found, clinical opinion was used. The exception to this hierarchy were data for perioperative complications and death. The infrequency of these events means that the small RCTs provide imprecise estimates. Large national audits of hysterectomy and firstgeneration endometrial ablation were used the VALUE and the MISTLETOE studies. 5,6 These provide best estimates of rare events. For complications following repeated ablation, data were taken from a prospective cohort study of 800 primary and 75 repeat ablations. 7 For second-generation techniques, large cohort studies investigating complication rates were used 8,9 (see Table 1). Health state utility values were taken from a published cost utility analysis of surgery for menorrhagia, 10 in which preferences were elicited from 60 women with heavy menstrual bleeding on health state descriptions using the time tradeoff technique. 11 We have assumed that initial and recurrent heavy menstrual bleeding have the same utility value. The state of being well is less than 1.0 as it encompasses general health values for women of this age. In the absence of alternative data, the utility value of convalescence after hysterectomy is assumed by the authors to be one-third less than the state of well following recovery following hysterectomy. This gives a reasonable value that falls between the health state values for complications after hysterectomy, and recovery after hysterectomy (see Table 1).

6 1108 R. GARSIDE ET AL. Table 3. Equipment costs. Equipment Cost ( ) Lifetime Source Notes Thermal balloon Cavaterm control unit years Manufacturer Cavaterm disposable balloon catheter 280 Single use Manufacturer Thermachoice generator years Manufacturer Cost from manufacturer, life time assumed. Thermachoice disposable balloon catheter Single use Manufacturer The list price is 350, manufacturers informs that due to various discounts, 335 is the UK average price Thermachoice cost of surgical devices 290 Per patient Manufacturer Calculated from cost given in Euros Microwave Microwave EA system 39,950 Manufacturer Maintenance contract for MEA system 5000 Annual Manufacturer Placement arrangement 375 Price per treatment Manufacturer According to the manufacturer, this arrangement is used by 51 UK centres. EA ¼ endometrial ablation; MEA ¼ microwave endometrial ablation. Resource use and costs Aspects of care in the model The costs for each type of surgical procedure were calculated based on estimates of resource use described below. Data for costs were taken from the literature and from the costings unit at Southampton University Hospital. The cost of procedures include endometrial thinning agents prior to microwave endometrial ablation, anaesthetic, dedicated equipment, operating time and inpatient stay. Pre-operative treatment It is assumed that all women with heavy menstrual bleeding will have the cause investigated in secondary care in line with current professional recommendations. 12 The economic model assumes that all women with heavy menstrual bleeding receive these investigations as a routine and prior to being offered any treatment. Such costs are not, therefore, relevant to the comparison between treatments. All patients undergoing first-generation ablations and microwave endometrial ablation are assumed to receive four to five weeks pretreatment with thinning agents: oral danazol (200 mg daily) if undergoing local anaesthetic treatment or the LHRH analogue Zoladex if undergoing general anaesthetic. Danazol is preferred where local anaesthetic is used as it causes less cervical resistance. 13 Surgical procedures Details for average length of stay in hospital and waiting time for hysterectomy are taken from the Hospital Episode Statistics (Code Q07 abdominal hysterectomy) for the UK. 1 Mean duration of surgery for hysterectomy was taken from a systematic review carried out in Details of resource use for firstgeneration endometrial ablation were taken from the review by Lethaby and Hickey. 14 Hospital episode statistics for were used to obtain waiting times for surgery. 1 All hysterectomies were assumed to be undertaken under general anaesthetic. Data on the proportion of first-generation endometrial ablation procedures using local anaesthetic were taken from Lethaby and Hickey. 14 For second-generation techniques, these data were taken from Bain et al. s 13 study of microwave endometrial ablation in which 63% of women had a preference about type of anaesthetic, of which 52% chose local anaesthetic (see Table 2 for details of Clinical assumptions). Equipment cost Two thermal balloon ablation systems are available in the UK: Cavaterm and Thermachoice; and one type of microwave equipment (Microsulis Medical) Equipment costs were based on details provided by the manufacturers of these devices. The cost of thermal balloon is the mean cost of the two devices. Costs of the equipment for microwave endometrial ablation and thermal balloon endometrial ablation are shown Table 4. Costs of anaesthesia, ward costs. Resource Cost Source General anaesthetic 1.08 per minute Microsulis submission Local anaesthetic 7.7 per minute Microsulis submission Inpatient bed 231 per day Southampton University Hospital estimated from own cost þ 50%

7 COST UTILITY OF MEA AND TBEA FOR HMB 1109 Table 5. Total procedure costs. Procedure Baseline price ( ) Hysterectomy 2096 TCRE 1110 TCRE/RB 1027 RB 1190 MEA 942 TBEA 826 TCRE ¼ transcervical resection; TBEA ¼ thermal balloon endometrial ablation; RB ¼ rollerball; MEA ¼ microwave endometrial ablation. in Table 3. The two sets of costs for the microwave system are based on different systems of supply. One involves purchase of the system and the other, under which the majority of UK centres using microwave endometrial ablation operate, is a placement arrangement, in which centres pay a list price of 375 per treatment (see Table 3). Staff costs It is assumed that all hysterectomy and all first endometrial ablation techniques are undertaken by a consultant. Staff involved are assumed to include a junior anaesthetist, a trolley nurse, instrument nurse and circulating nurse. Given the relative simplicity of second-generation ablation techniques, the costs were also calculated using a more junior surgeon (registrar) (Table 4). Total procedure costs are shown in Table 5. Discounting Costs were discounted at 6% and benefits at 1.5% in line with UK Treasury guidance. Analyses An incremental analysis of costs and benefits was performed for each of the following comparisons (Table 5): microwave endometrial ablation versus thermal balloon endometrial ablation microwave endometrial ablation versus transcervical resection microwave endometrial ablation versus transcervical resection and rollerball microwave endometrial ablation versus rollerball microwave endometrial ablation versus hysterectomy thermal balloon endometrial ablation versus transcervical resection thermal balloon endometrial ablation versus transcervical resection and rollerball thermal balloon endometrial ablation versus rollerball thermal balloon endometrial ablation versus hysterectomy Dealing with uncertainty The effects of uncertainty in individual parameters were explored using extensive one-way sensitivity analyses, in which values of inputs were varied and the effects on incremental cost effectiveness noted. A probabilistic simulation was also developed to explore the effects on cost effectiveness of uncertainty in the model inputs. In this approach, the model is run 1000 times with input values sampled from individual probability density functions. Values for transition probabilities and health state utilities within the model were sampled from beta distributions while treatment cost values were sampled from log-normal distributions. The parameters and type of distribution in the stochastic model follow accepted norms Table 6. Summary of cost utility analysis for TBEA and MEA at 10 years. Procedure Total costs ( ) Total QALYS Incremental costs Incremental QALYs ICER Baseline results TBEA 1,323, MEA 1,448, Incremental results for TBEA vs MEA TBEA vs MEA 124, TBEA dominates Incremental results for TBEA vs first-generation techniques and hysterectomy TCRE 1,731, , TBEA dominates TCRE þ rollerball 1,785, , TBEA dominates Rollerball 1,752, , TBEA dominates Hysterectomy 2,320, , ,410 Incremental results for MEA vs first-generation techniques and hysterectomy TCRE 1,731, , MEA dominates TCRE þ rollerball 1,785, , MEA dominates Rollerball 1,752, , MEA dominates Hysterectomy 2,320, , ,108

8 Table 7. Inputs varied in sensitivity analyses. Assumptions Values used in sensitivity analyses Source Justification for source Transitions Complications following MEA Upper value based on numbers for TBEA. Upper from large UK audit of TBEA, lower on RCTs Lower on rate in RCTs. Death following MEA direct cause Values for EA reported in this review Minimum and maximum death rates reported for EA procedures included in this review Complications following TBEA Effectiveness evidence in this report Based on RCTs best available evidence Death following TBEA direct cause Values for EA reported in this review Minimum and maximum death rates reported for all procedures included in this review Proportion of complications lasting more than one month for TBEA/MEA Authors assumption Values give wide range to test to sensitivity. Complication rate with repeat ablation First year return of menorrhagia post TBEA/MEA Second and third year return of menorrhagia after TBEA/MEA Percentage of women with recurrent menorrhagia receiving hysterectomy over repeat ablation Same rate as first ablation to 4 times that in first ablation MacLean-Fraser et al. 7 and assumption Effectiveness data median at 12 months RCT data Total return of menorrhagia at 3 years, 21 51% Minimum assumed the same as first ablation, upper limit based on case series study of first and second ablation complication rates. Menorrhagia assumed to include al those reporting menorrhagia at a given follow up plus those who have had a repeat EA or hysterectomy in that time period Expert opinion and assumption Upper limit based on expert opinion, lower limit assumed R. GARSIDE ET AL. Utilities Menorrhagia Sculpher 10 and assumption Lowest value from mean reported in interviews with women with menorrhagia. Upper value estimated in comparison to other health state utilities. TBEA and MEA Authors assumption Lower limit same as menorrhagia mean varies amount of discomfort and adverse effects. Well following EA Authors assumption Lower limit half way between menorrhagia and well, allowing for some long term adverse effects, upper limit close to full health. Costs ( ) Local anaesthetic 0 100% Author s assumption Full range of none to all procedures under anaesthetic Proportion of second generation procedures done in an office setting 0 100% Authors assumption Full range of none to all procedures done in an office/non-theatre setting. Length of hospital stay Lower level clinician opinion, upper Input form clinical experience and national data. level form HES UK average Procedure time minutes Systematic review Garside et al. 2 Lowest and highest recorded theatre times Equipment costs MEA Author s assumption Cost plus and minus 50% Equipment costs TBEA Authors assumption Cost plus and minus 50% Model Duration of model 3 10 years Authors assumption

9 COST UTILITY OF MEA AND TBEA FOR HMB 1111 and are derived from empirical data where possible. The resulting 1000 cost effectiveness ratios indicate the range of outcomes that might be expected given the inherent uncertainty in the underlying data. Net-benefit of each therapeutic option was calculated and the results depicted on a cost effectiveness acceptability curve which demonstrates, at different levels of willingness to pay for an additional QALY, the probability that each option is the most cost effective. BASELINE RESULTS The total costs for the cohorts of 1000 women over 10 years are shown below. Table 6 shows the cost effectiveness of microwave endometrial ablation and thermal balloon endometrial ablation compared with each of the other procedures. With microwave endometrial ablation, very slightly fewer QALYs are accrued for a very slightly higher cost compared with thermal balloon endometrial ablation. This difference is negligible. Compared with transcervical resection, transcervical resection combined with rollerball, and rollerball alone, microwave endometrial ablation accrues more QALYS and costs less. Compared with hysterectomy, microwave endometrial ablation is cheaper, but accrues fewer QALYS. Compared with transcervical resection, transcervical resection combined with rollerball and rollerball alone, thermal balloon endometrial ablation costs less and accrues more QALYS. Compared with hysterectomy, thermal balloon endometrial ablation costs less and accrues fewer QALYs. Sensitivity analyses Table 7 shows the inputs of the one-way sensitivity analyses. Relatively small changes in input values have a marked effect on cost effectiveness, in some cases, changing the direction of effect in the comparison between second-generation techniques. Procedure cost and time, frequency and duration of complications and death rate are particularly important. In the comparison of second-generation techniques to first-generation or hysterectomy, changes to the following variables had little or no effect on cost effectiveness: Complication rate of treatment (in either first or repeat ablations). Length of complication state. Percentage of those being treated for recurrent menorrhagia who are treated by hysterectomy versus repeat endometrial ablation. Utility for menorrhagia. Utility for thermal balloon endometrial ablation and microwave endometrial ablation state. Recurrence rates, equipment cost, procedure time, proportion using local anaesthetic, death rate and model Fig. 3. Simulation output (1000 trials) showing probability of hysterectomy or different endometrial ablation amounts being the most cost effective at different levels of willingness to pay amounts.

10 1112 R. GARSIDE ET AL. Fig. 4. Simulation output (1000 trials) showing probability of different methods of endometrial ablation being at different levels of willingness to pay. duration had small effects on cost effectiveness. However, the model was highly sensitive to the utility value for the well health state following ablation. The probabilistic simulation shows the range of outcomes that might be expected in practice, given the underlying uncertainty in available data. Figure 3 shows the probability that each of the technologies is the most cost effective treatment at various different levels of willingness to pay for an additional QALY. Up to very low levels of willingness to pay (about 2000 per QALY gained), it appears that thermal balloon endometrial ablation is likely to be most cost effective although the figures at these low levels are affected by the absolute, although small, differences in procedure cost. As willingness to pay increases, hysterectomy is very likely to be considered the most cost effective within the constraints of this analysis. Given that, for a variety of reasons, some women will not wish to have a hysterectomy, we repeated this analysis excluding hysterectomy. It appears that thermal balloon endometrial ablation is likely to be most cost effective, although differences between it and microwave endometrial ablation are very small beyond a willingness to pay of 5000 per QALY gained and negligible over 20,000 (Fig. 4). DISCUSSION This study has a number of strengths. It has been produced by an independent scientific team, with clinical support, and was informed by an exhaustive systematic review of the effectiveness literature. It is the first evaluation of the cost effectiveness of second-generation techniques and uses a probabilistic approach to handle the inevitable uncertainty in the data available on these new technologies. There are, however, some remaining areas of uncertainty. The analysis necessarily depends on inferred comparisons between microwave endometrial ablation and thermal balloon endometrial ablation and between both secondgeneration techniques and hysterectomy. Such comparisons are prone to bias and confounding and should be viewed with caution, although the probabilistic analysis takes as full account as possible of imprecision in the underlying data. In the absence of direct comparisons, such analyses provide the only available approach to assessing value for money. In particular, in the absence of a direct comparison, we believe it is impossible, on the basis of current evidence, to reach a general conclusion on cost effectiveness between different second-generation techniques. The costs and benefits associated with different secondgeneration techniques are similar, as demonstrated by the large degree of overlap in the outputs of the probabilistic analysis. The costing methods used in this analysis were necessarily imprecise and distinguishing between these two approaches on the basis of current evidence is therefore difficult. Nevertheless, both techniques appear less costly than the most commonly used first-generation techniques and appear, at least, to be no less effective. Both secondgeneration techniques therefore dominate transcervical resection, rollerball and transcervical resection/rollerball combined. Hysterectomy yields additional benefits for additional cost, with cost utility ratios of around 2400 per QALY against both microwave endometrial ablation and thermal

11 COST UTILITY OF MEA AND TBEA FOR HMB 1113 balloon endometrial ablation. This is because hysterectomy, despite its greater initial cost, carries no risk of recurrence of menorrhagia. Women choosing this option therefore spend longer in the post-treatment well state and accrue more QALYs as a result. The economic model is very sensitive to utility values used, especially the value for women who are well following recovery from any procedure, on which little published evidence is available. The study by Sculpher 10 provided most of the utility values used in this report. The value for the state of menorrhagia was rated at a median of 0.55 (mean 0.5, SE 0.04) by the women interviewed in the study. This seems low. A value of 0.5 for a health state using the time trade off method means that women would be prepared to trade 50% of their future life expectancy to avoid it. The range of scores obtained for menorrhagia was zero (as bad as being dead) to 0.95 (where 1.0 is best possible health.) Clearly, even among women suffering from heavy menstrual bleeding, the impact of the condition is valued very differently by different individuals. A single utility value must therefore be regarded as imprecise, although this imprecision is taken into account in the probabilistic analysis. In the same study, women with heavy menstrual bleeding were asked to rate their own current, pretreatment health state. This had a mean of 0.65 (SE 0.04) and a median of 0.75 (range 0 1.0), much higher than the state of menorrhagia, which may have been because most women were not menstruating at the time of the interview. There are problems eliciting values for chronic health states in which the worst effects are episodic. In addition, for heavy menstrual bleeding, effects are not lifelong, but will disappear at menopause. Further health states, such as the utility value for postconvalescence ( well ) after treatment for heavy menstrual bleeding, may be particularly difficult to interpret. After hysterectomy, there is no possibility of heavy menstrual bleeding or other menstrual symptoms returning. Hysterectomy also prevents the possibility of some gynaecological cancers. In contrast, hysterectomy is associated with premature ovarian failure and early menopause, as well as carrying the risk of longer term adverse effects such as urinary incontinence. 15 An ablation procedure cannot guarantee amenorrhoea, and there is the possibility of recurrent heavy menstrual bleeding. Although hysterectomy emerges as the most cost effective option at levels of willingness to pay which appear to be well within that tolerated by the NHS, there are other important considerations in deciding whether secondgeneration techniques should be available. Firstly, there are advantages to the second-generation techniques that may not be captured in our analysis. As they are conducted without direct visualisation, there is no chance of fluid overload, associated with serious potential adverse effects of first-generation techniques. The new techniques are quicker and easier to learn and use. They may also be better suited to use under local anaesthetic, and therefore office use. Thermal balloon endometrial ablation is not suitable for women with irregularly shaped uterine cavities, a disadvantage not apparent for microwave endometrial ablation. Secondly, women express strong preferences in choosing treatment for menorrhagia, and the impact of this factor is uncertain. The willingness of policymakers to pay in order for there to be choice of treatments is not known, nor are the benefits that are associated with the exercise of choice. In Sculpher s 10 study, women rated the well state following hysterectomy more highly than that following endometrial ablation (median 0.95 vs 0.90, respectively). This may indicate the influence of preferences for particular treatments. Sculpher 10 suggests that further analysis is required to explore whether preference based treatment allocation has the potential to be cost effective. Currently, only about 16,000 endometrial ablation operations are performed annually in the UK. This compares to about 26,000 hysterectomies for heavy menstrual bleeding. 1 The MISTLETOE study found that 83% of NHS hospitals were providing endometrial ablation in It is not yet clear how fast or extensive newer secondgeneration techniques will be taken up. However, the introduction of first-generation endometrial ablation techniques has not yet caused a large reduction in the number of hysterectomies being performed. It is possible that minimally invasive techniques merely lower the threshold for surgery and so savings to the NHS may be less than indicated by a simple transfer of activity from hysterectomy to endometrial ablation. CONCLUSIONS Both microwave endometrial ablation and thermal balloon endometrial ablation appear to offer cost effective alternatives to first-generation endometrial ablation techniques. Compared with hysterectomy, both are cheaper, but hysterectomy is associated with greater QALYs. Differences between thermal balloon endometrial ablation and microwave endometrial ablation are slight and it is possible that the methods used here are not sensitive enough to reliably detect such small differences. No studies directly comparing thermal balloon endometrial ablation with microwave endometrial ablation have been undertaken nor has hysterectomy been tested against secondgeneration techniques. Comparison between second-generation techniques is a priority for further research as is longer term follow up of endometrial ablation. More extensive research into the impact of heavy menstrual bleeding, and the impact of different surgical treatments on quality of life would be valuable. In particular, more extensive studies of treatment and outcome preferences would assist future cost utility analyses.

12 1114 R. GARSIDE ET AL. References 1. Department of Health. Hospital Episode Statistics London: Office for National Statistics, Garside R, Stein K, Wyatt K, Round A, Price A. The effectiveness and cost-effectiveness of microwave and thermal balloon endometrial ablation for heavy menstrual bleeding: a systematic review and economic modelling. Health Technol Assess 2004;8. 3. Lethaby A, Shepperd S, Cooke I, Farquhar C. Endometrial resection and ablation versus hysterectomy for heavy menstrual bleeding. Cochrane Database Syst Rev 2000:CD The Government Actuary Department. Interim Life Tables Expectation of Life for England and Wales Females London: The Government Actuary Department, Maresh MJA, Metcalfe MA, McPherson K, et al. The VALUE national hysterectomy study: description of the patients and their surgery. Br J Obstet Gynaecol 2002;109: Overton C, Hargreaves J, Maresh M. A national survey of the complications of endometrial destruction for menstrual disorders: the MISTLETOE study. Br J Obstet Gynaecol 1997;104: MacLean-Fraser E, Penava D, Vilos GA. Perioperative complication rates of primary and repeat hysteroscopic endometrial ablations. JAm Assoc Gynecol Laparosc 2002;9: Rogersen L, Duffy S. A European survey of the complications of a uterine thermal balloon system in 5800 women. Gynaecol Endosc 2002;11: Parkin DE. Microwave endometrial ablation (MEA): a safe technique? Complication data from a prospective series of 1400 cases. Gynaecol Endosc 2000;9: Sculpher M. A cost utility analysis of abdominal hysterectomy versus transcervical endometrial resection for the surgical treatment of menorrhagia. Int J Technol Assess Health Care 1998;14: Torrance GW. Utility approach to measuring health-related quality of life. J Chronic Dis 1987;40: Royal College of Obstetricians and Gynaecologists. The Management of Menorrhagia in Secondary Care. Evidence Based Guidelines No. 5. London: RCOG Press, Bain C, Cooper KG, Parkin DE. A partially randomized patient preference trial of microwave endometrial ablation using local anaesthesia and intravenous sedation or general anaesthesia: a pilot study. Gynaecol Endosc 2001;10: Lethaby A, Hickey M. Endometrial destruction techniques for heavy menstrual bleeding. Cochrane Database Syst Rev 2002:CD Clarke A, Black N, Rowe P, Mott S, Howle K. Indications for and outcome of total abdominal hysterectomy for benign disease: a prospective cohort study. Br J Obstet Gynaecol 1995;102: Gervaise A, Fernandez H, Capella-Allouc S, et al. Thermal balloon ablation versus endometrial resection for the treatment of abnormal uterine bleeding. Hum Reprod 1999;14: Which operation for menorrhagia? Drug Ther Bull 2000;38: Cooper KG, Jack SA, Parkin DE, Grant AM. Five-year follow up of women randomised to medical management or transcervical resection of the endometrium for heavy menstrual loss: clinical and quality of life outcomes. Br J Obstet Gynaecol 2001;108: Accepted 27 April 2004