Jaroslaw Wechowski, M.D., Ph.D., a Mark Connolly, M.Sc., b Dirk Schneider, M.D., b Philip McEwan, Ph.D., c and Richard Kennedy, M.D.

Transcription

1 Cost-saving treatment strategies in in vitro fertilization: a combined economic evaluation of two large randomized clinical trials comparing highly purified human menopausal gonadotropin and recombinant follicle-stimulating hormone alpha Jaroslaw Wechowski, M.D., Ph.D., a Mark Connolly, M.Sc., b Dirk Schneider, M.D., b Philip McEwan, Ph.D., c and Richard Kennedy, M.D. d a Health Economics, Cardiff Research Consortium, Cardiff, United Kingdom; b Health Economics, Ferring International Center, St. Prex, Switzerland; c School of Mathematics, Cardiff University, Cardiff; and d Centre for Reproductive Medicine, University Hospital, Coventry, United Kingdom Objective: To assess the cost-effectiveness of two gonadotropin treatments that are available in the United Kingdom in light of limited public funding and the fundamental role of costs in IVF treatment decisions. Design: An economic evaluation based on two large randomized clinical trials in IVF patients using a simulation model. Setting: Fifty-three fertility clinics in 13 European countries and Israel. Patient(s): Women indicated for treatment with IVF (N ¼ 986), aged 18 38, participating in double-blind, randomized controlled trials. Intervention(s): Highly purified menotropin (HP-hMG, Menopur) or recombinant follitropin alpha (rfsh, Gonal-F). Main Outcome Measure(s): Cost per IVF cycle and cost per live birth for HP-hMG and rfsh alpha. Result(s): HP-hMG was more effective and less costly versus rfsh for both IVF cost per live birth and for IVF cost per baby (incremental cost-effectiveness ratio was negative). The mean costs per IVF treatment for HP-hMG and rfsh were 2408 (95% confidence interval [CI], 2392, 2421) and 2660 (95% CI 2644, 2678), respectively. The mean cost saving of 253 per cycle using HP-hMG allows one additional cycle to be delivered for every 9.5 cycles. Conclusion(s): Treatment with HP-hMG was dominant compared with rfsh in the United Kingdom. Gonadotropin costs should be considered alongside live-birth rates to optimize outcomes using scarce health-care resources. (Fertil Steril Ò 2009;91: Ó2009 by American Society for Reproductive Medicine.) Key Words: Economic evaluation, cost-effectiveness, modeling, gonadotropins, in vitro fertilization, menotropin, recombinant FSH In industrialized countries, an estimated 17% of couples seek medical advice for infertility (1). The lifetime prevalence of infertility in the United Kingdom (UK), defined as at least 1 Received October 8, 2007; revised and accepted January 8, 2008; published online March 12, This study was funded by Ferring Pharmaceuticals, St. Prex, Switzerland. Dr. Jaroslaw Wechowski and Dr. Philip McEwan are academic consultants who provide health-consulting services to pharmaceutical and medical device companies and governments. Dr. Jaroslaw Wechowski is the owner of Pharmarchitecture Limited, a consultancy for pharmaceutical companies not involved in the current study. Dr. Richard Kennedy is clinical director of the University Hospital Centre for Reproductive Medicine in Coventry and has worked as a nonremunerated advisor to this project. Mark Connolly and Dirk Schneider are employees of Ferring Pharmaceuticals. The results of this study were presented in an abbreviated format at the European Society for Human Reproduction and Embryology meeting, which was held in Lyon, France, in July Reprint requests: Mark Connolly, Director Market Access, Ferring International Center SA, Chemin de Vergognausaz 50, CH-1162, St. Prex, Switzerland (FAX: ; mark@gmasoln.com). year of unsuccessful attempts to conceive, has been reported to range between 17.3% and 26.4% (2, 3). Treatment for infertility has a very good prognosis, with 80% 90% of couples being successful after 1 year and 95% after 2 years (4). In addition to proven effectiveness, in many markets, to qualify for reimbursement, treatments must meet rigorous economic criteria of cost-effectiveness. In light of the increasing demand for health care across all areas, healthcare providers often resort to rationing as a means of controlling costs; such rationing often disregards evidence on relative cost-effectiveness of treatment options. In many markets, including the UK, treatment of infertility is an easy target for budget cutting because it is often viewed as a low health priority (5). Based on best available efficacy and cost-effectiveness evidence, the National Institute for Health and Clinical Excellence (NICE) recommends that up to three cycles should be reimbursed (6). Despite NICE s recommendation, in practice approximately only 25% of cycles are funded by the National Health Service (NHS) (7) /09/$36.00 Fertility and Sterility â Vol. 91, No. 4, April doi: /j.fertnstert Copyright ª2009 American Society for Reproductive Medicine, Published by Elsevier Inc.

2 As a consequence of rationing within the NHS, the UK ranks among the lowest in Europe for provision of IVF when compared with countries such as Belgium and Denmark, which have more generous state funding (8). Furthermore, because of the relationship between a couple s ability to pay and access to infertility treatment, it is likely that many infertile couples do not seek or discontinue treatment because of limited financial resources (9, 10). When the macro aspects of infertility and assisted reproductive technologies (ART) treatment are considered, as in two recent studies, it is conceivable that access barriers, financial or legislative, which prohibit couples from seeking treatment can have a small but meaningful long-term demographic impact (11, 12). A critical component in economic evaluations is the clinical outcomes data on which cost-effectiveness claims are made. Accumulating data now show that treatment outcome differs significantly depending on the gonadotropin chosen. As previously reported by the European and Israeli Study Group (EISG), patients treated with either highly purified (HP) hmg or recombinant FSH (rfsh) achieved similar ongoing pregnancy rates (13). A subsequent analysis of the EISG IVF cohort demonstrated that patients receiving HP-hMG achieved a statistically significant (P¼.037) higher ongoing pregnancy rate (31%) than patients receiving rfsh (20%) (14). An integrated analysis of the EISG and menotrophin versus recombinant FSH in vitro fertilization trial (MERiT) studies that was reported in an article by Sørensen et al. and in an earlier article by Platteau et al. demonstrated significantly higher live-birth rates of 27% with HP-hMG compared with 21% with rfsh (odds ratio [OR] ¼ 1.36; 95% confidence interval [CI], ; P¼.04) (15). Additionally, a recent meta-analysis by Al-Inany et al. (16) has shown that there is a statistically significant increase in live-birth rate with patients treated with hmg compared with rfsh (P<.05). In clinical terms, this translates into a number needed to treat (NNT) to get an additional live birth with hmg versus rfsh of 23 (95% CI, ) for a 25% baseline chance of live birth with IVF. Because of the need to optimize outcomes with limited financial resources, it is important to consider the costs of all elements of the treatment options. One possible variation in IVF costs can be attributed to different gonadotropin regimes. Previous studies have explored the cost-effectiveness of ART using a variety of different clinical parameters and assumptions (17). Cost-effectiveness of recombinant and urinary FSH has been compared (18 24), and one study included urinary hmg as an additional comparator (21). Urinary hmg and its highly purified preparation were analyzed jointly in comparison with rfsh for ovulation induction (25). HP-hMG alone was compared with rfsh, but the analysis combined IVF and intracytoplasmic sperm injection (ICSI) patients (26). Of all identified economic studies involving HP-hMG, only one was based on patient-level data from a randomized trial (27), and in one, efficacy data were extracted from a meta-analysis (24). While many studies reported cost per clinical or ongoing pregnancy (18, 19, 22 26, 28), cost per live birth or per baby was addressed less frequently (29) and mainly in studies not involving comparison of gonadotropins (27). No report of cost per live birth after stimulation with HP-hMG has been identified in published sources. Only one study involving HP-hMG and rfsh was conducted in the UK setting, although it was not specific to IVF patients (26). Our previous study comparing HP-hMG and rfsh in IVF patients in the UK setting was a cost-minimization analysis (27) based on one clinical trial only (30). Pooling of two large randomized controlled trials (RCTs) allowed us to conduct the first analysis of live-birth rates after stimulation with the two gonadotropins and of the related costs in IVF patients in the UK. Existing economic evidence suggests that cost savings could be achieved by using HP-hMG. Based on a randomized clinical trial, menotropin was found to be less expensive per cycle in the UK when compared with rfsh, allowing 13% more cycles to be offered (13, 26). A recent study based on a simulation model using outcomes from an RCT demonstrated cost-savings resulting from using HP-hMG instead of rfsh (27, 30). Menotropin was also found to be cost saving compared with rfsh in a model based on a meta-analysis of clinical trials (25). As new clinical data become available, it is common practice to update economic evaluations to incorporate new efficacy results, while adjusting resource use and changes in costs. Therefore, our study is based on the two RCTs involving IVF patients treated with HP-hMG or rfsh with cost per live birth as the primary outcome in the analysis. Since new adequately powered trials of gonadotropins in subfertile patients are unlikely to be conducted, pooling of results from existing trials remains a valuable and informative technique (31). Decision analytic models are commonly used to evaluate treatments for which sufficient cost data have not been collected alongside clinical trials (32). Ideally, prospective clinical trials should be conducted to determine both efficacy and costs, but the value and feasibility of such approach is limited in multiple multistep interventions, such as assisted reproduction (28); therefore modeling enables representation of each stage and cycle of the treatment process and unbiased assessment of health outcomes and resource use. We employed state-of-the-art simulation modeling to combine best available costs and efficacy evidence. MATERIALS AND METHODS Overview An economic evaluation comparing HP-hMG (Menopur; Ferring Pharmaceuticals, St. Prex, Switzerland) and rfsh (follitropin alpha; Gonal-F, Merck Serono, Geneva, Switzerland) was conducted based on a discrete event simulation (DES) model. The objective of the study was to determine the following: 1068 Wechowski et al. Economic evaluation of HP-hMG versus rfsh Vol. 91, No. 4, April 2009

3 1. Efficacy expressed as number of live births and number of babies born per patient initiating treatment. Results of the simulation replicate the trials results presented in the earlier paper, however, small discrepancies in point estimates and CIs may arise from random process in the simulation (15). 2. Average IVF cost per treatment based on each of the gonadotropin treatment options. The simulation results are reported as per patient treated, rather than per cycle, as it is the patients who were randomized in trials (33). However, results for one cycle treatment can be interpreted both as per cycle and per treatment. 3. Cost per live birth for each treatment option. 4. Incremental cost-effectiveness expressed as the difference in cost of the two treatment options divided by the difference in number of deliveries and babies born per treated patient. 5. Impact of inclusion of maternal and neonatal costs arising from successful pregnancies, considering costs of multiple pregnancies. Clinical Evidence The modeled population was obtained by pooling patients from two prospective, randomized, multinational trials: EISG (IVF stratum, n ¼ 255) and MERiT (all patients, n ¼ 731) (13, 14). Both studies were conducted in accordance with the Declaration of Helsinki on good clinical practice, and ethics approval was obtained in all participating centers. Women aged were recruited at 53 fertility clinics in 13 European countries and Israel. The entire modeled cohort was treated with either HP-hMG (n ¼ 491) or rfsh (n ¼ 495) using a long GnRH agonist stimulation protocol. The primary causes of infertility in both studies were tubal and unexplained infertility; the baseline characteristics of patients in both trials, eligibility criteria, treatment protocols, and primary endpoints were similar, justifying the pooled analysis. Cost Data Resource use data were based on established UK treatment practices or items collected in the two randomized clinical trials. Unit costs per item were obtained from published UK sources. Treatment doses of HP-hMG and rfsh were sampled based on dosage reported in the two trials. The unit costs of per vial of rfsh containing 75 IU of FSH activity and per vial of HP-hMG containing 75 IU of FSH activity and 75 IU of LH activity were obtained from the British National Formulary (34). In the UK centers, GnRH agonists were used according to the trial center standard practice, and therefore, the costs of down-regulation were based on buserelin (Superfact, Sanofi-Aventis, Paris, France) 250 mg/day for an average of 14 days. To reflect UK treatment practices, the costs of choriongonadotropin alpha (Ovitrelle, Merck Serono, Geneva, Switzerland) based on a single dose of 250 mg at a cost of were included in the cost calculations. Similarly, to reflect UK treatment practices, the costs of vaginal P gel (Crinone, Merck Serono) were included in the costs based on treatment ranging from 13 to 42 days after embryo transfer (ET). Since daily application is indicated for 30 days after laboratory evidence of pregnancy, three packages with 15 dosages each per successful pregnancy were assumed at per package (34). Based on expert advice, all moderate and severe cases of ovarian hyperstimulation syndrome (OHSS) were assumed to lead to hospitalization at an average cost of 1032 per diagnosis. Frequency of OHSS was extracted from trial data (1.81% for FSH and 1.83% for HP-hMG; not significant). The cost of patient visits was assumed to be included in costs of retrieval, fertilization, and ET. Specialist consultation costs were added during stimulation and for pregnancy determination. Costs of delivery and neonatal care were considered separately for singletons, twins, and triplets (35). Cost and outcome consequences of failed treatments and the probability of adoption were not considered. The perspective of the study was that of the NHS payer and nonmedical costs, and costs of time and loss of productivity were not incorporated in the analysis. Since fewer cycles would be required on HP-hMG than on rfsh to achieve one delivery and one birth (baby), the exclusion of social costs had a conservative effect likely to underestimate the cost advantage of HP-hMG. Costs were indexed to 2006, and no discounting was applied because of the short time horizon employed. All input variables for costs used in the model are summarized in Table 1. The Simulation Model A DES was designed to model clinical outcomes and related costs and coded using visual basic for applications in Excel (Microsoft Excel; Microsoft Corp., Redmond, WA). In DES, probabilities based on individual patient data from the combined trial cohort are simulated for each clinical event, using computer-generated random numbers representing chance events (36). In such simulations, also known as Monte Carlo simulations, results from numerous runs are averaged over all patients, approximating the results that would be obtained if deterministic probabilities, identical for each patient, were used. As a result, small insignificant variations in point estimates and CIs can be observed between the deterministic and simulated analysis. In DES, the events do not need to be synchronized as in Markov models, allowing greater flexibility and more accurate representation of the patient pathway, thus reflecting clinical reality with greater accuracy. The following treatment events (stages) were modeled from the combined clinical cohort: stimulation, oocyte retrieval, ET, pregnancy, and live births. Modeled treatment was consistent with the procedures performed in the clinical trials, and the patient transitions between stages of the model were determined by pooled patient-level data from the two clinical trials. The number of oocytes retrieved and number of twins and triplets were also obtained from the trial data. Probabilities of fertilization and implantation success were accounted for by subsequent events and therefore not considered explicitly in the model. The logical flow Fertility and Sterility â 1069

4 TABLE 1 Cost variables used in simulation economic evaluation. Variable Value Uncertainty interval Source Comments Cost of Gonal-F (rfsh) per IU, Based on price of per vial containing 75 IU of rfsh. Cost of Menopur (HP-hMG) per IU, Based on price of per vial containing 75 IU of FSH and 75 IU of LH activity Median dose rfsh, IU Median dose HP-hMG, IU Sample trial data Sample , 30 14, 30 Dose was bootstrapped from pooled MERiT and EISG trial data due to right skewness of the trial data distribution (median 2250 IU). Cost of Buserelin, , 34 Superfact (Sanofi-Aventis), unit cost of per 5.5mg vial. Dosing based on local practice, 250 mg/day for 14 days. Cost of choriongonadotropin alpha, , 34 Choriongonadotropin alpha (Ovitrelle, Merck Serono) was costed as a single dose at 250 mg. Luteal support, Vaginal P gel (Crinone, Merck Serono) was costed after ET during days. Since daily application is indicated for 30 days after laboratory evidence of pregnancy, three packages with 15 dosages each per successful pregnancy were assumed at per package. Cost of retrieval, UHCW cost inflated to 2006 price. Cost of ET, UHCW cost inflated to 2006 price. Cost of fertilization, UHCW cost inflated to 2006 price. Hospitalization cost for OHSS, Cost of clinical pregnancy determination, Cost of ongoing pregnancy determination, All moderate and severe OHSS was assumed to lead to hospitalization , 47 Based on HCG test cost of and specialist consultation at , 47 Based on ultrasound cost of and specialist consultation at Cost of pregnancy loss, Costs of early and late pregnancy loss were not differentiated. Note: UHCW ¼ University Hospitals of Coventry and Warwickshire. Wechowski. Economic evaluation of HP-hMG versus rfsh. Fertil Steril of the model is illustrated in Figure 1; the process is briefly described below. Each patient entered the simulated treatment pathway in the stimulation state. Treatment with gonadotropins was administered in this state, accounting for prior pituitary down-regulation, as described elsewhere (EISG, MERiT). Ovulation was induced, and patients entered the oocyte retrieval state. After fertilization, for those patients with viable embryos retrieved, ET was performed, which led to the stage of pregnancy that resulted in either pregnancy loss or live birth, single or multiple delivery Wechowski et al. Economic evaluation of HP-hMG versus rfsh Vol. 91, No. 4, April 2009

5 FIGURE 1 Flow diagram representing events simulated in the model. Wechowski. Economic evaluation of HP-hMG versus rfsh. Fertil Steril Statistical Analysis Efficacy results were simulated using available case intention-to-treat data (37). To test the robustness of the results, the last observation carried forward (LOCF) method for imputing missing values was also used. Only the available case results are reported, as the LOCF-based results were not significantly different and did not affect the conclusions. Simulation output cost variables were not normally distributed; hence the Wilcoxon rank sum test was used to test the significance of mean and median results. In addition, the results were verified using a permutation test on means and medians of samples obtained by bootstrapping the cost results from the model. Confidence intervals were produced by bootstrapping. Bootstrapping estimates the sampling distribution of a statistic through a large number of simulations, employing sampling with replacement from the original data (38). Samples of sizes equal to the trial groups were drawn from the simulated parent population of 50,000 patients. Mean values were used for summary reporting, except for costs inclusive of maternal and neonatal costs, where medians were Fertility and Sterility â 1071

6 inappropriate owing to skewed irregular distribution attributed to costs associated with twins and triplets. Two-sided 95% CIs were obtained based on cost distributions from bootstrap as the 2.5th and the 97.5th percentiles. Considerable skewness of the gonadotropin dose distribution was addressed by sampling the original trial data. Comparative cost-effectiveness of gonadotropins was expressed using an incremental cost-effectiveness ratio (ICER). The ICER is a comparative metric used to represent the additional cost required to achieve an additional unit of clinical effectiveness between two different treatment options (39). Probabilistic sensitivity analysis (PSA) was performed to address uncertainty in the parameters used in the model. Based on accepted practices, values for parameters were sampled from 95% CIs based on probabilities of success at each stage of simulation and from uncertainty intervals for costs, judged to conservatively approximate the 95% CIs. Results from the PSA were summarized using cost-effectiveness acceptability curves (CEACs) (40). CEACs present the probability that a given intervention is more cost-effective than the alternative treatment as a function of ceiling ratio (CR) or willingness to pay for one unit of effectiveness, typically life year adjusted for quality (QALY). RESULTS Efficacy The live-birth rate obtained from the DES after one fresh IVF cycle was 26.6% (CI, 22.7% 30.5%) for HP-hMG and 20.6% (CI, 17.0% 24.1%) for rfsh (OR ¼ 1.40; CI, ; P¼.03). When adjusted for multiple births, the respective values were 338 (95% CI, ) and 268 (95% CI, ) babies per 1000 cycles initiated (OR ¼ 1.40; 95% CI, ; P¼.02). Furthermore, the required NNT to achieve one additional live birth when using HPhMG instead of rfsh was 17 for deliveries and 14 for babies. Costs Total mean IVF costs per person starting the treatment after one fresh IVF treatment cycle for HP-hMG and rfsh were 2408 (95% CI, 2392, 2421) and 2660 (95% CI, 2644, 2678), respectively (P<.001; Table 2). The mean per-cycle cost saving resulting from using HP-hMG instead of rfsh was 253, which would allow financing of an additional cycle for every 9.5 HP-hMG cycles delivered. The total cost of gonadotropin was 36% lower for HP-hMG than for rfsh, that is, 471 ( ) versus 736 ( ; P<.001). TABLE 2 IVF cycle costs for HP-hMG and rfsh (follitropin alpha) cycles. Parameter Total costs per patient treatment, Gonadotropin costs per patient treatment Mean cost per live birth excluding maternal and neonatal costs, Median cost per live birth excluding maternal and neonatal costs, Total costs per patient treatment including maternal and neonatal costs, Mean cost per live birth including maternal and neonatal costs, Median cost per live birth including maternal and neonatal costs, HP-hMG 2408 a ( ) 471 a ( ) 9,058 a ( ,676) Follitropin alpha 2660 a ( ) 736 a ( ) 12,944 a (11,124 15,788) Cost difference ICER, per live birth ICER, per baby b 12,620 b c 3886 c ,804 a (12,753 17,592) 18,909 a (15,854 23,230) b 12,945 b 3796 a Differences between HP-hMG and rfsh significant at P<.001 (Wilcoxon rank sum test), despite instances of overlap of CIs. In permutation test CIs for the difference in means and medians did not contain 0, indicating significance in all comparisons. b Confidence intervals not provided for medians as they have no clinical interpretation when the median value lies outside the interval; CIs for differences obtained in permutation tests did not include zero, indicating significance. c Difference not statistically significant (P>.05, Wilcoxon rank sum test). Wechowski. Economic evaluation of HP-hMG versus rfsh. Fertil Steril Wechowski et al. Economic evaluation of HP-hMG versus rfsh Vol. 91, No. 4, April 2009

7 Remaining costs related to IVF procedures, inclusive of OHSS treatment costs, were comparable ( 1487 vs. 1492, respectively), as were other costs related to stimulation ( 344 vs. 327). Mean total IVF costs per live birth after one treatment cycle for HP-hMG and rfsh were 9058 ( ,676) and 12,944 ( 11,124 15,788), respectively (P<.001). The corresponding median costs per live birth per patient were 8995 and 12,620, respectively (P<.001). When maternal and neonatal costs were applied to livebirth data, including costs of multiple pregnancies, the median costs per patient starting the treatment with HP-hMG and rfsh were 9149 and 12,945 (P<.001). The corresponding mean costs per live birth per patient were 14,804 ( 12,753 17,592) and 18,909 ( 15,854 23,230), respectively (P<.001). The mean IVF costs per baby after one treatment cycle for HP-hMG and rfsh were 7,133 ( ) and 9944 ( ,219), respectively (P<.001). When maternal and neonatal costs were added, the median costs per baby with HP-hMG and rfsh were 7205 and 9946, respectively (P<.001). Incremental Cost-Effectiveness ICER was negative for both IVF cost per live birth and IVF cost per baby, indicating that HP-hMG is dominant (less costly and more effective) versus rfsh alpha (Table 2). When maternal and neonatal costs were added to the cost of IVF, ICER was 824 and 710 per delivery and per baby, respectively. Changes in Drug Acquisition Costs The impact of varying the acquisition costs of rfsh alpha on comparative costs per live birth after one cycle was tested in the model. The results indicate that an equivalent cost per live birth for both gonadotropins cannot be achieved, regardless of the cost of rfsh. To achieve an equivalent cost per live birth, the price of rfsh would have to be negative (see solid line in Fig. 2). An equivalent total IVF cost per cycle would be achieved at the cost of rfsh alpha 2.65 per vial (data not shown). This represents a price reduction to around oneeighth of the published British National Formulary price for rfsh follitropin alpha. FIGURE 2 Cost per live birth as a function of rfsh price. Wechowski. Economic evaluation of HP-hMG versus rfsh. Fertil Steril deliveries (CR ¼ 0), HP-hMG was preferred in 100% of cases because of cost savings. DISCUSSION Our results demonstrate that treatment with HP-hMG after one fresh cycle of IVF offers significantly higher live-birth rates that were achieved at lower cost compared with treatment with rfsh. The live-birth rates simulated in the model are comparable with outcomes in the previously reported pooled analysis (15). The reported OR by Platteau et al. was 1.36 (95% CI, ), and the slight discrepancy observed in this study and the previous study are due to random processes in the simulation. Additionally, these results are in agreement with a previously reported meta-analysis of FIGURE 3 Cost-effectiveness plane with the scatter of PSA results for HP-hMG versus FSH. Probabilistic Sensitivity Analysis The PSA for one IVF cycle revealed that the mean and median value of ICER was negative both for cost per delivery and for cost per baby, which suggests cost saving. The scatter output from the PSA results is plotted on the costeffectiveness plane as shown in Figure 3. To quantify the results, CEACs were plotted. The CEAC showed that the probability of being cost-effective for HP-hMG was 98.4% for the value of CR of 20,000 per live birth both exclusive and inclusive of neonatal and maternal costs. For all values of CR smaller than 20,000, the probability was greater than 98.4%, and was equal to 100% for CR ¼ 6720 and less. If no amount was assumed to be allocated for additional Wechowski. Economic evaluation of HP-hMG versus rfsh. Fertil Steril Fertility and Sterility â 1073

8 randomized trials comparing menotropins with rfsh, which showed that using the former produced a clinically relevant higher pregnancy rate (27%) than using the latter (22%) (41). A recent update of the meta-analysis with three additional studies included and the majority of the patients treated with the highly purified preparation HP-hMG showed a statistically significant higher live-birth rate (relative risk [95% CI]: 1.18 [1.02, 1.38]; P¼.03). The advantage of using HP-hMG is also evidenced by relatively low values of NNT. Typically, an NNT of less than 20 is required to justify preference for the more effective treatment in a routine clinical setting (42, 43). Several previous studies have investigated the costs of IVF delivery in the UK. For example, NICE guidelines published in 2004 suggest an average IVF cost per cycle of 2771 (6).A recently published assessment of costs based on the MERiT clinical data alone suggested that the costs of HP-hMG and rfsh treatment were 2396 and 2633, respectively (27). In addition, Lloyd et al. established that the cost for one IVF cycle was 2423 with HP-hMG and 2745 with rfsh; however, these estimates are based on a combined IVF and ICSI population (26). The results presented here are consistent with previous cost per cycle estimates. Based on our analysis, when IVF costs were considered alongside efficacy results, treatment with HP-hMG generated cost savings with approximately 9% savings per cycle relative to rfsh. Based on the savings achieved using HP-hMG, this would suggest that within a fixed budget, additional IVF cycles could be delivered through the NHS. When maternal and neonatal costs were included in the analyses, additional costs accrued to HP-hMG attributed to the increased live-birth rate. This resulted in a change in the ICER from negative to positive for the incremental cost per delivery and incremental cost per baby analyses of 824 and 710, respectively. Slightly greater cost savings per cycle were found using HP-hMG versus rfsh when compared with our previous analysis based on the MERiT trial data (27). Importantly, current analysis of combined trial data shows significance of higher success rates with HP-hMG (15). Greater efficacy inevitably contributed to higher average maternal and neonatal costs per treatment initiated, thus reducing the incremental advantage of HP-hMG relative to the previous single-trial analysis; nevertheless, HP-hMG remains highly costeffective. Additionally, the analysis presented here indicates that some of the costs accrued to those patients treated with HP-hMG because of additional costs attributed to maternal and neonatal costs arising from nonsignificant differences in multiple birth rates per treatment initiated (P¼.59). By including costs differences arising from nonsignificant differences in multiple birth rates, we believe this approach to be conservative. In our analysis, we report the results both exclusive and inclusive of maternal and neonatal costs. It may be argued, however, that if neonatal costs are to be considered in the analysis, utility of future discounted life years of babies born should also be included. If the costs related to treatment for infertility should be considered only up the point when the patient is no longer infertile, that is, up to clinical or ongoing pregnancy, then neither neonatal costs nor utility of the babies would be counted. If, however, cost per live birth is reported, the neonatal utility should also be estimated. Even with adjustment of average lifetime morbidity, each live birth would increase joint mother and child utility considerably. Additional utility gain reported by a successfully treated couple would augment the advantage of HP-hMG even further. Costs of treatment are largely fixed costs (staff, facilities), but drugs, mainly gonadotropins, make up 20% (HP-hMG) and 28% (rfsh) of the total IVF fresh cycle costs (Table 2). To test the sensitivity of our cost per cycle and cost per live birth results, we varied the drug acquisition costs of rfsh to establish at what point the two products were equivalent. The results indicate that, regardless of the available discount on the published price of rfsh, a cost per live birth equivalent to HP-hMG cannot be achieved (cost of rfsh would have to be negative). An equivalent total IVF cost would be achieved with an rfsh price reduction of 8.5 times (11.9%) the published British National Formulary price. Even though it is common practice for fertility units in the UK to negotiate drug costs directly with companies, the results from this analysis show little possibility for the two drugs to be comparably cost-effective. Apart from drug acquisition costs, the dosage of gonadotropins used in stimulation is a significant factor in treatment costs and, as a result, affects cost-effectiveness results. In this study gonadotropin dosage was extracted from the patientlevel data from the clinical trials; the values were consistent with those reported for HP-hMG and rfsh in the combined analysis (15). A recent trend in some markets is to use lowdose gonadotropin stimulation protocols (44). In our evaluation we did not assess the impact of low-dose stimulation protocols as this would tend to lower gonadotropin costs for both treatments, thus influencing the cost-saving benefits of HP-hMG, but the impact on relative efficacy of the two drugs could not be estimated. Until low-dose protocols become the norm, we believe this approach to be valid for informing current treatment practices as applied in the MERiT and EISG studies. One of the key issues in conducting economic evaluations is the handling of uncertainty around input parameters. To address issues of uncertainty there are well-established guidelines available to instruct investigators regarding best treatment practices. For example, NICE suggests the use of PSA, with simultaneous variation of all input parameters because the results obtained using this method are most robust (31, 32). This is in contrast to previous economic evaluations of gonadotropins that did not include any sensitivity analysis (20, 23) and only addressed the use of one-way sensitivity analysis (19, 21, 25) or two-way sensitivity analysis (18). Furthermore, when efficacy input was varied in simulations, the costs were often kept at mean values (22, 28, 29), or no details on their variation were reported (24). Also in the study 1074 Wechowski et al. Economic evaluation of HP-hMG versus rfsh Vol. 91, No. 4, April 2009

9 comparing HP-hMG and rfsh, uncertainty was addressed by bootstrapping, but not all parameters were varied (26). To comply with NICE guidelines and to best inform medical decision makers, we have applied an all-inclusive PSA to address uncertainty in the analysis presented here (31, 32). The results of our study are likely to have underestimated the true cost-effectiveness of HP-hMG relative to rfsh. There are two factors that could likely influence the costeffectiveness findings presented here: [1] the treated population and [2] assumptions regarding cancellation rates. Randomized trials comparing gonadotropins almost always include patients with favorable ovarian response profiles, which is likely to favor the least effective treatment option, here rfsh (45). Based on the trial data, the cancellation rate before oocyte retrieval of 5.9% for the rfsh group and 4.3% for the HP-hMG group (not significant) was assumed (15). Engmann et al. previously noted a cancellation rate of 3.5% of patients starting treatment (46). Using the latter value in the model slightly increases cost-effectiveness of HP-hMG with no effect on costs per delivery and per baby and no impact on conclusions. In conclusion, our results demonstrate that higher success rates can be achieved with HP-hMG compared with rfsh and can be achieved at lower cost. HP-hMG is cost saving when only IVF treatment costs are considered. The inclusion of maternal and neonatal costs in the analysis adversely influenced the results for HP-hMG because of the increased live births. Despite the additional maternal and neonatal costs, HP-hMG remained highly cost-effective. Furthermore, sensitivity analysis revealed that HP-hMG was superior regardless of the rfsh unit price, a direct consequence of both the lower price of HP-hMG and the higher success rate achieved using this gonadotropin. It is envisaged that the results of our study will assist medical decision makers and clinicians by highlighting that clinical choice has significant direct implications for treatment success where gonadotropins differ in efficacy and that improved outcomes can be achieved at lower cost using HP-hMG. REFERENCES 1. Cahill DJ, Wardle PG. Management of infertility. BMJ 2002;325: Gunnell DJ, Ewings P. Infertility prevalence, needs assessment and purchasing. J Pub Health 1994;16: Buckett W, Bentick B. The epidemiology of infertility in a rural population. Acta Obstet Gynecol Scand 1997;70: Brosens I, Gordts S, Valkenburg M, Puttemans P, Campo R, Gordts S. Investigation of the infertile couple: when is the appropriate time to explore female infertility? Hum Reprod 2004;19: Gloucestershire NHS. The future of healthcare in Gloucestershire: proposals for developing sustainable NHS services A public consultation National Institute for Health and Clinical Excellence. Clinical guideline 11. Fertility: assessment and treatment for people with fertility problems Kennedy R, Kingsland C, Rutherford A, Hamilton M, Ledger W. Implementation of the NICE guideline Recommendations from the British Fertility Society for national criteria for NHS funding of assisted conception. Hum Fertil 2006;9: Andersen AN, Goossens V, Gianaroli N, Felberbaum R, de Mouzon J, Nygren KG. Assisted reproductive technology in Europe, Results generated from European registers by ESHRE. Hum Reprod 2007;22: Rajkhowa M, McConnell A, Thomas GE. Reasons for discontinuation of IVF treatment: a questionnaire study. Hum Reprod 2005;21: Collins J. An international survey of the health economics of IVF and ICSI. Hum Reprod Update 2002;8: Sobotka T, Hansen MA, Jensen TK, Pedersen AT, Lutz W, Skakkebaek NE. Will fertility of Danish women remain stable due to assisted reproduction? Assessing the role of assisted reproduction in sustaining cohort fertility rates. Annual Meeting of the Population Association of America, March Hoorens S, Gallo F, Cave J, Grant J. Can assisted reproductive technologies help to offset population ageing? An assessment of the demographic and economic impact of ART in Denmark and UK: case report. Hum Reprod 2007;22: The European and Israeli Study Group on Highly Purified Menotropin versus Recombinant Follicle-Stimulating Hormone. Efficacy and safety of highly purified menotropin versus recombinant follicle-stimulating hormone in in vitro fertilization/intracytoplasmic sperm injection cycles: a randomized, comparative trial. Fertil Steril 2002;78: Platteau P, Smitz J, Albano C, Sørensen P, Arce J-C, Devroey P. Exogenous luteinizing hormone activity may influence the treatment outcome in in vitro fertilization but not in intracytoplasmic sperm injection cycles. Fertil Steril 2004;81: Platteau P, Nyboe Andersen A, Loft A, Smitz JEJ, Danglas P, Devroey P. Highly purified hmg versus recombinant FSH for ovarian stimulation in IVF Cycles. Reprod Med Online Accepted for publication ( 09). 16. Al-Inany HG, Abou-Setta AM, Aboulghar MA, Mansour RT, Serour GI. Efficacy and safety of human menopausal gonadotrophins versus recombinant FSH: a meta-analysis. Reprod Biomed Online 2008;16: Garceau L, Henderson J, Davis JL, Petrou S, Henderson LR, McVeigh E, et al. Economic implications of assisted reproductive techniques: a systematic review. Hum Reprod 2002;17: Mantovani LG, Belisari A, Szucs TD. Pharmaco-economic aspects of invitro fertilization in Italy. Hum Reprod 1999;14: Van Loon J, Liaropoulos L, Mousiama T. Economic evaluation of a recombinant follicle-stimulating hormone (Follitropin Beta, Puregon) in infertile women undergoing in vitro fertilisation in Greece. Clin Drug Invest 2000;19: Ola B, Papaionnou S, Afnan MA, Hammadieh N, Gimba S. Recombinant or urinary follicle-stimulating hormone? A cost-effectiveness analysis derived by particularizing the number needed to treat from a published meta-analysis. Fertil Steril 2001;75: Sykes D, Out HJ, Palmer SJ, van Loon J. The cost-effectiveness of IVF in the UK: a comparison of three gonadotrophin treatments. Hum Reprod 2001;16: Silverberg K, Daya S, Auray JP, Duru G, Ledger W, Wikland M, et al. Analysis of the cost-effectiveness of recombinant versus urinary follicle-stimulating hormone in in vitro fertilization/intracytoplasmic sperm injection programs in the United States. Fertil Steril 2002;77: Gerli S, Casini ML, Unfer V, Costabile L, Bini V, Di Renzo GC. Recombinant versus urinary follicle-stimulating hormone in intrauterine insemination cycles: a prospective, randomized analysis of costeffectiveness. Fertil Steril 2004;83: Hatoum HT, Keye WR, Marrs RP, Walton SM, Marshall DC. A Markov model of the cost-effectiveness of human-derived follicle-stimulating hormone (FSH) versus recombinant FSH using comparative clinical trial data. Fertil Steril 2005;83: Al-Inany HG, Abou-Setta AM, Aboulghar MA, Mansour RT, Serour GI. HMG versus rfsh for ovulation induction in developing countries: a cost-effectiveness analysis based on the results of a recent metaanalysis. Reprod Biomed Online 2006;12: Fertility and Sterility â 1075

10 26. Lloyd A, Kennedy R, Hutchinson J, Sawyer W. Economic evaluation of highly purified menotropin compared with recombinant follicle-stimulating hormone in assisted reproduction. Fertil Steril 2003;80: Wechowski J, Connolly M, McEwan P, Kennedy R. An economic evaluation of HP-hMG and rfsh based on a large randomised trial. Reprod Biomed Online 2007;15: Daya S, Ledger W, Auray JP, Duru G, Silverberg K, Wikland M, et al. Cost-effectiveness modeling of recombinant FSH versus urinary FSH in assisted reproduction techniques in the UK. Hum Reprod 2001;16: Revelli A, Poso F, Gennarelli G, Moffa F, Grassi G, Massobrio M. Recombinant versus highly-purified, urinary follicle-stimulationg hormone (r-fsh vs. HP-uFSH) in ovulation indication: a prospective, randomized study with cost-minimization analysis. Reprod Biol Endocrinol 2006; 4: Andersen AN, Devroey P, Arce JC, MERIT Group. Clinical outcome following stimulation with highly purified hmg or recombinant FSH in patients undergoing IVF: a randomized assessor-blind controlled trial. Hum Reprod 2006;21: Johnson NP, Proctor M, Farquhar CM. Gaps in the evidence for fertility treatment an analysis of the Cochrane Menstrual Disorders and Subfertility Group database. Hum Reprod 2003;18: Buxton MJ, Drummond MF, Van Hout BA, Prince RL, Sheldon TA, Szucs T, et al. Modelling in economic evaluation: an unavoidable fact of life. Health Econ 1998;7: Daya S. Pitfalls in the design and analysis of efficacy trials in subfertility. Hum Reprod 2003;18: British Medical Association, Royal Pharmaceutical Society of Great Britain. British National Formulary Available at: Ledger WL, Anumba D, Marlow N, Thomas CM, Wilson EC. The costs to the NHS of multiple births after IVF treatment in the UK. Br J Obstet Gynaecol 2006;113: Mandelblatt JS, Fryback DG, Weinstein MC, Russell LB, Gold MR, members of the Panel on Cost-Effectiveness in Health, and Medicine. Assessing the effectiveness of health interventions for cost-effectiveness analysis. J Gen Intern Med 1997;12: The Cochrane Collaboration. The Cochrane manual issue 3, Briggs AH, Wonderling DE, Mooney CZ. Pulling cost-effectiveness analysis up by its bootstraps: a non-parametric approach to confidence interval estimation. Health Econ 1997;6: Gold MR, Siegel JE, Russell LB. Cost-effectiveness in health and medicine. New York: Oxford University Press, Fenwick E, O Brien BJ, Briggs A. Cost-effectiveness acceptability curves facts, fallacies and frequently asked questions. Health Econ 2004;13: van Wely M, Westergaard L, Bossuyt P, van der Veen F. Effectiveness of human menopausal gonadotropin versus recombinant folliclestimulating hormone for controlled ovarian hyperstimulation in assisted reproductive cycles: a meta-analysis. Fertil Steril 2003;80: Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment effect. BMJ 1995;310: Riegelman R, Schroth WS. Adjusting the number needed to treat: incorporating adjustments for the utility and timing of benefits and harm. Med Decis Making 1993;13: Marci R, Senn A, Dessole S, Chanson A, Loumaye E, De Grandi P, et al. A low-dose stimulation protocol using highly purified folliclestimulating hormone can lead to high pregnancy rates in in vitro fertilization patients with polycystic ovaries who are at risk of a high ovarian response to gonadotropins. Fertil Steril 2001;75: Balasch J, Barri PN. Reflections on the cost-effectiveness of recombinant FSH in assisted reproduction. The clinician s perspective. J Assist Reprod Genet 2001;18: Engmann L, Maconochie N, Bekir JS, Jacobs HS, Tan SL. Cumulative probability of clinical pregnancy and live birth after multiple cycle IVF package: a more realistic assessment of overall and age-specific success rates? Br J Obstet Gynaecol 1999;106: Curtis L, Netten A. Unit costs of health and social care Personal Social Services Research Unit, University of Kent, Wechowski et al. Economic evaluation of HP-hMG versus rfsh Vol. 91, No. 4, April 2009