Article Text
PDF
Abstract
Objective This paper uses health economics methods to discuss the cost-effectiveness value of long protocol and antagonist protocol for in vitro fertilisation and embryo transfer (ET) in the Chinese population.
Design Health economic evaluation study.
Setting The data needed to construct the model for this study were derived from published studies and other secondary sources in China.
Participants No patients participated in the study.
Measures The main outcomes were live birth rate (LBR) and cost. From the societal perspective, we considered the direct and indirect costs over the course of the treatment cycles. A cost-effectiveness was measured using the incremental cost-effectiveness ratio and the probability that a protocol has higher net monetary benefit. Sensitivity analysis was carried out to verify the reliability of the simulation results.
Results For the Chinese population, the long protocol resulted in a higher LBR than the antagonist protocol (29.33% vs 20.39%), but at the same time, it was more expensive (￥29 146.26 (US$4333.17) vs ￥23 343.70 (US$3470.51)), in the case of considering only one fresh ET cycle. It was the same when considering subsequent frozen ET (FET) cycles (51.78% vs 42.81%; ￥30 703.02 (US$4564.62) vs ￥24 740.95 (US$3678.24)). The results of most subgroups were consistent with the results of the basic analysis. However, for certain populations, the long protocol was the inferior protocol (less effective and more expensive).
Conclusion For the Chinese population, when the monetary value per live birth was greater than ￥65 420 (US$9726) and ￥66 400 (US$9872), respectively, considering only one fresh cycle and considering subsequent frozen cycles, the long protocol is the preferred protocol. This threshold also varies for women of different ages and ovarian response capacities. For women in POSEIDON (Patient-Oriented Strategies Encompassing IndividualizeD Oocyte Number) group 2, group 3 and group 4, antagonist protocol is recommended as the preferred protocol. The results of this study need to be verified by further large-scale randomised controlled trials.
- health economics
- health economics
- reproductive medicine
- health policy
- subfertility
Data availability statement
Data are available upon reasonable request. All data can be obtained by contacting the corresponding author.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
This research considered two cases: (1) only fresh embryo transfer (ET) and (2) subsequent frozen ETs.
This study conducted subgroup analyses based on female age and ovarian response capacity.
This study did not set a specific willingness-to-pay threshold but sought an alternative method of cost-effectiveness threshold.
The model parameters of this study were derived from other clinical studies.
Model construction relied on too many assumptions and the scenario was relatively ideal.
Introduction
Infertility is a disease defined by the American Society for Reproductive Medicine as the failure to achieve a successful pregnancy after 12 months or more of regular, unprotected sexual intercourse or due to an impairment of a person’s capacity to reproduce either as an individual or with her/his partner.1 One study showed that the prevalence of infertility is as high as 25% among couples of childbearing ages in China.2 In 1978, traditional in vitro fertilisation (IVF) treatment was successfully applied for the first time and assisted reproductive technology (ART) has become an important part of modern medicine playing a significant role in family planning.3 In 2017, the International Committee for Monitoring Assisted Reproductive Technologies defined ART as all interventions involving in vitro manipulation of human oocytes and sperm or embryos for reproductive purposes,4 mainly including artificial insemination (AI), IVF and embryo transfer (IVF-ET), intracytoplasmic sperm injection (ICSI), preimplantation genetic testing, gamete or zygote intrafallopian transfer, and gamete and embryo cryopreservation.4 Assisted reproductive treatment is too expensive, the duration of treatment is long, the outcome is uncertain, and in addition, assisted reproductive treatment is excluded from the medical insurance system in China, which has led to many families being reluctant to try this treatment.5 6 Under this background, the economic evaluation of ART is of great importance from both economic and health perspectives. Moreover, the population structure in China may benefit immensely from the economic evaluation of ART.
Among the established methods, IVF-ET remains the most popular assisted reproductive therapy for infertile patients in terms of population coverage and patient selection.7 The treatment process of IVF-ET is divided into the ovulation cycle and transplant cycle, with at least two basic elements of successful pregnancy after IVF-ET treatment: a certain number of high-quality embryos and good endometrial receptivity. Obtaining high-quality embryos, which is critical to promoting pregnancy, relies on obtaining a certain number of high-quality oocytes, which involves controlled ovarian hyperstimulation (COH), a key process in IVF-ET treatment. A study revealed that hormonal stimulation accounted for a significant proportion of the per-cycle expenses.8
Ovulation induction protocols mainly include gonadotropin-releasing hormone agonist (GnRH-a) long protocol, super long protocol, short protocol, super short protocol, GnRH-antagonists (GnRH-A) protocol and micro stimulus protocol. Among these protocols, the GnRH-a long protocol and GnRH-A protocol are currently the most widely used in clinical practice and applicable for a wide range of people.9 GnRH-A long protocol is also known as the long luteal phase protocol, starting from the middle luteal phase. In this protocol, ovulation induction is relatively long, with an average of about 4 weeks. This protocol easily controls and monitors the growth of oocytes and helps patients create more mature and high-quality oocytes. While the protocol has a lower risk of failure, the body is more susceptible to the ovarian hyperstimulation syndrome (OHSS).10 GnRH-A protocol starts ovulation from the first 2–3 days of menstruation and usually takes about 10 days of medication, which is relatively short and affects the body less, therefore a low risk of OHSS.11 However, the development of follicles is poor and ovulation is relatively rare. Both of these protocols have their advantages and disadvantages, and the efficiency of pregnancy aid is still highly controversial. Therefore, this study fully considered various possibilities in clinical practice and used a pharmacoeconomic evaluation model to evaluate the cost-effectiveness of long and antagonist protocols during IVF-ET treatment. The study is strictly reported in accordance with the Consolidated Health Economic Evaluation Reporting Standards 2022 checklist.12
Materials and methods
Model
This research employed TreeAge Pro software (TreeAge Pro Healthcare 2022, R1.2, Williamstown, Massachusetts, USA) to construct a decision tree model and decision tree combined with Markov model for cost-effectiveness analysis with the live birth rate (LBR) as an index to measure the effect. Our models were based on a hypothetical population of women suffering from infertility. Both models simulated the whole process of ovarian stimulation, IVF/ICSI, ET, pregnancy and live birth in patients receiving IVF-ET treatment. Among them, the decision tree model simulated the situation in which patients received fresh ET. For the subsequent frozen ETs (FETs) process, a Markov model was added to the basis of the decision tree model while the decision tree model of ET remained unchanged. In this case, the evaluation index was cumulative LBR. Figure 1 illustrates the health states transition diagram in the decision tree combined with Markov model, which consists of four different health states: ‘fresh cycle’, ‘frozen cycle’, ‘live birth’ and ‘drop-out’. The schematic representation of cycle pathways is depicted in figure 2. The whole process considered only one ovarian stimulation and oocyte retrieval. All patients started from the fresh cycle during which patients underwent ovarian stimulation, oocyte retrieval, IVF/ICSI and fresh ET, and automatically withdrew from the model when obtaining no oocytes or available embryos. When available embryos were obtained, patients could choose to either directly undergo fresh ET or cryopreserved embryos for subsequent frozen cycle for FETs. In case of no successful live birth, the action could be dropping out. If there were previously cryopreserved embryos, the FET could be carried out until a successful live birth or using up the available embryos. Each patient would exit the model automatically after having a live birth (ie, each patient had at most one live birth). In the model, patients underwent a maximum of four cycles of attempts (ie, one fresh cycle and three frozen cycle attempts).13
Health states transition diagram.
Cycle pathways in the model. ET, embryo transfer; FET, frozen FT; ICSI, intracytoplasmic sperm injection; IVF, in vitro fertilisation.
Model assumptions
In this study, the specific assumptions were as follows: (1) The average assisted reproductive treatment cycle is about 3 months.13 14 In other words, a couple could make up to four attempts in a year. (2) The whole process only considers one ovarian stimulation and oocyte retrieval, and each patient could only perform one fresh cycle and three frozen cycles at most. (3) Three frozen cycles have the same state transition probabilities.
Model parameters
Transition probabilities
In this study, common databases were systematically searched from the database establishment to September 2022. The search terms “GnRH-a long protocol”, “GnRH-A protocol” and “Assisted Reproductive Technology” were used as search terms, and the search method was combined with subject term and free term. Literature inclusion criteria: (1) Population: patients with low fertility treated with IVF/ICSI-ET; (2) Intervention and Comparison: long protocol and antagonist protocol; (3) Outcomes: clinical pregnancy rate, LBR; (4) Study design: randomised controlled trial, cohort study. Literature exclusion criteria: (1) The assisted pregnancy protocol was not IVF/ICSI-ET; (2) The interventions did not involve a long or antagonist protocol; (3) The patient had additional comorbidities, such as endometriosis, endometrial cancer, etc; (4) Systematic reviews, meta-analyses, case reports, etc and (5) Duplicate literature, lack of access to full text, missing information. The flow chart of literature screening is shown in online supplemental figure S1. The characteristics of the included studies are shown in online supplemental table S1. Six studies were finally selected to determine the transition probability parameters of the basic model9 15–19 (table 1).
Supplemental material
Transition probability parameter table
Costs
This study measured costs involving direct costs and indirect costs from the societal perspective. The direct medical costs in this study were obtained from literature and generally included preoperative examination, ovarian stimulation, oocyte retrieval, fertilisation, embryo culture, ET, pregnancy test, abortion, delivery and other related costs. Due to the differences in the incidence of complications between the two ovulation induction protocols, particularly OHSS, the treatment cost will also be greatly different, which can directly affect the total costs. So the treatment cost of OHSS was included as a direct cost in this study. Direct non-medical costs included transportation costs, which were calculated based on the average fare from Chongqing’s various districts and counties to Chongqing on the China Railway 12306.com. Indirect costs included the cost of lost work, which was composed of the working days lost and the per capita disposable income, according to data from the National Bureau of Statistics (table 2 and online supplemental table S2).
Cost parameter table
This study converted all cost data into monetary values in RMB in 2022 (￥1=US$0.14867; ￥1=€0.14136) based on the survey year of the literature data, deflated by the annual growth rate of the Consumer Price Index in China and provided corresponding information on US dollar. This research determined the cost parameters referring to the following eight studies.9 20–27 This model disregards the discount rate due to the short time span of the sample period.
Cost-effectiveness analysis
In cost-effectiveness analysis, this study used LBR to measure the effectiveness while applying cumulative LBR in case of considering subsequent FETs. As some studies28 29 have shown that willingness-to-pay (WTP) value for fertility is highly uncertain, and there is currently no determination of the WTP value for fertility in China. Therefore, this study did not set a WTP threshold. Instead, we used the incremental cost-effectiveness ratio (ICER) and the probability that a protocol has higher net monetary benefit (NMB) as a measure of cost-effectiveness. The ICER is calculated as the ratio of the cost difference and outcome difference between long protocol and antagonist protocol, and represents an increase in cost per additional live birth for long protocol compared with antagonist protocol. The NMB value represents the benefit of each protocol in terms of live birth using the monetary value, which is the monetary benefit of the protocol in terms of birth outcomes minus the cost. Assuming that per live birth has a certain monetary value, the LBR is multiplied by this monetary value, and the total cost is subtracted to obtain the NMB value of each protocol. Monte Carlo simulations were performed on the model, and each simulation could determine which of the two protocols had higher NMB within a range of monetary values and then calculate the proportion that each protocol had higher NMB. The result could be interpreted as the probability that a protocol had higher NMB. In addition to the basic cost-effectiveness analysis, we also performed subgroup analyses based on ovarian response capacity and age. Six subgroups were assigned according to the diagnostic criteria of ovarian response ability and POSEIDON (Patient-Oriented Strategies Encompassing IndividualizeD Oocyte Number) criteria (online supplemental table S3). The POSEIDON criterion is a stratified criterion for poor ovarian responders. The model parameters are shown in online supplemental table S4, referring to four studies.11 30–32
Sensitivity analysis
First, a one-way sensitivity analysis (OWSA) was carried out for all parameters. One parameter in the model changes item by item according to the set upper and lower limits while other model parameters remain unchanged. In this way, the influence degree of each parameter on the simulation results is evaluated. Second, this research conducted probabilistic sensitivity analysis (PSA) and 10 000 Monte Carlo simulations for each protocol. This approach allows multiple parameters to vary simultaneously within a reasonable range to estimate the 95% CI of the ICER. In this way, this study tested the comprehensive influence of multiple parameters changing simultaneously on the simulation results. Then, the results were represented by the ICER scatter plot and NMB curves under a range of live birth monetary values. Also, this study considered upper and lower limits for all model parameters, and let each parameter range within ±10% of the baseline value. In addition, the PSA assumes that the cost parameter and transition probability parameter follow gamma and beta distributions, respectively, per capita disposable income follows triangular distributions, and working day lost follow uniform distributions (online supplemental tables S5 and S6).
Patient and public involvement
No patients were involved in this study.
Results
Cost-effectiveness
The long protocol resulted in a higher LBR than the antagonist protocol (29.33% vs 20.39%), but at the same time, it was more expensive (￥29 146.26 (US$4333.17) vs ￥23 343.70 (US$3470.51)), in the case of considering only one fresh ET cycle. Compared with the antagonist protocol, the cost of the long protocol increased by ￥64 899.47 (US$9648.60) per additional live birth. Considering the subsequent FET cycles, the long protocol was still more effective (51.78% vs 42.81%) and more expensive (￥30 703.02 (US$4564.62) vs ￥24 740.95 (US$3678.24)) than the antagonist protocol. The corresponding ICER was ￥66 477.97 (US$9883.28)/live birth (table 3).
Cost-effectiveness analysis (basic model)
Subgroup analysis
The results of most subgroups were consistent with the results of the basic analysis, with the long protocol being more effective but more expensive than the antagonist protocol. However, there were cases where the results were different. For POSEIDON group 4, women when considering only one fresh cycle, and POSEIDON group 2, group 3 and group 4, women when considering subsequent frozen cycles, the long protocol was the inferior protocol (less effective and more expensive) (online supplemental table S7).
Sensitivity analysis
One-way sensitivity analysis
According to the OWSA tornado diagrams (online supplemental figure S2–S8), all models indicated that the two cost parameters with the greatest influence on the results were the costs of long and antagonist protocols ovulation stimulating drugs. For POSEIDON women in groups 2 and 4 considering only one fresh cycle, and POSEIDON women in group 3 considering subsequent frozen cycle, some transition probability parameters had a great impact on the model, and the uncertainty was relatively high.
Probabilistic sensitivity analysis
The ICER scatter plot showed the uncertainty of the results (online supplemental figures S9–S15). For women in POSEIDON group 4 considering only one fresh cycle, and for women in POSEIDON group 2, group 3 and group 4 considering subsequent frozen cycles, most of the scatter points were distributed in the north-west quadrant, the long protocol was considered to be the inferior protocol. In other cases, most of the scatters were distributed in the north-east quadrant, and the long protocol was considered more effective and more expensive. For women in POSEIDON group 2, when only one fresh cycle was considered, baseline analysis showed that the long protocol was more effective and more expensive than the antagonist protocol, which only occurred in 51.12% of replications (north-east quadrant), and 48.88% of results considered the long protocol was less effective and more expensive. This was reflected in the high ICER (￥21 496 984.54 (US$3195 956.69)/live birth) in the baseline analysis.
The NMB curves for a range of monetary values per live birth are shown in online supplemental figures S16–S22. In general, the probability of having a higher NMB was higher for the antagonist protocol when the monetary value per live birth ranged from 0 to a specific value; when the monetary value per live birth reached this specific value, the probability of having higher NMB became equal for the two protocols; when the monetary value per live birth exceeded this specific value, the probability of having a higher NMB was higher for the long protocol. However, in some cases, the curves did not go the same way, and for women in POSEIDON group 4 considering only one fresh cycle, and for women in POSEIDON group 2, group 3 and group 4 considering subsequent frozen cycles, the probability of having higher NMB was consistently lower in the long protocol than in the antagonist protocol. For women in POSEIDON group 2, when only one fresh cycle was considered, the probability of having higher NMB was higher in the long protocol than in the antagonist protocol after monetary value per live birth reached a specific value, but the difference was not significant.
Discussion
IVF-ET remains the most common assisted reproductive therapy for infertility patients.7 COH is an important part of IVF-ET treatment and hormonal stimulation accounted for a significant proportion of the per-cycle expenses.8 Among the ovulation induction protocols, the long and antagonist protocols are the most widely used, despite their controversial effects.10 33–35 However, these existing studies only considered the differences in effectiveness and safety between protocols, limiting the significance as the economic aspect were ignored. This study used health economics methods to evaluate the cost-effectiveness value of long protocol and antagonist protocol in IVF-ET via two scenarios. In general, the long protocol was more effective but more expensive than the antagonist protocol. Combined with the NMB curves, it was determined which protocol had higher probability of having higher NMB when the monetary value per live birth was within a certain range. However, when considering subsequent frozen cycles, for women in POSEIDON group 2, group 3 and group 4, the long protocol became the inferior protocol, that is, less effective and more expensive. This result was supported by most of the scatters of the ICER scatter plot. The NMB curves showed that regardless of the monetary value per live birth, the probability of having higher NMB in the long protocol was always lower than that in the antagonist protocol. For women in POSEIDON group 4, the probability of having higher NMB in the long protocol was always lower than that in the antagonist protocol, regardless of whether only one fresh cycle or the subsequent frozen cycles were considered. In the OWSA, results suggest many strategies to effectively improve the cost-effectiveness value of these two protocols including a reduction in the cost of ovulation stimulating drugs in assisted reproductive therapy or considering them in the list of medical insurance and improving the success rate of the treatment process. In addition, improving the success rate of treatment in women in POSEIDON group 2, group 3 and group 4 will greatly enhance the cost-effectiveness value of both protocols.
Since the ultimate goal of assisted reproductive therapy is to obtain live birth, this study used LBR as an effect indicator. Other than this study, there were few studies on the economic evaluation of long protocol and antagonist protocol with inconsistent results. Jing et al found that the antagonist protocol was more cost-effective in the fresh ET cycle.9 The effect indicator was the sustained pregnancy rate in Jing’s study9 while it was LBR in this study, which might be responsible for the difference in the results. Pan et al also took the LBR as one of the research indicators.27 Cheng et al took quality-adjusted life-years (QALYs) as the outcome indicator.14 However, QALYs are an accurate indicator to capture the health state of patients, and the value of infertility treatment is not easily reflected in QALYs.36 Jing et al used patient data from their medical institutions9 while parameters in this study were derived from published literature. Therefore, different sources of data may account for the divergent findings observed between the two studies. Besides, Jing et al9 did not perform subgroup analysis of patients. Different medical levels, prices and physical signs of patients in different regions and medical institutions may affect the selection of the most cost-effective treatment plan. Therefore, the specific situation of each region and medical institution should be taken into consideration when performing a targeted cost-effectiveness analysis.
One of the advantages of this study is that we considered two cases in the economic evaluation of the long and antagonist protocols: (1) only fresh ET and (2) subsequent FETs. This strategy fully considered various possibilities in clinical practice, so that the constructed models could be closer to the clinical setting and increase the reference value of the simulation results. In addition, female age and ovarian response capacity are key factors affecting the fertility treatment success rates. This study conducted subgroup analyses based on female age and ovarian response capacity to further explore whether the more cost-effective protocol changed in women with different ages and ovarian response capacities.
Some studies investigated the WTP range of ART in different countries, showing that the WTP values were highly uncertain.28 29 As the WTP threshold for fertility treatment has also not been determined in China and there is currently an evidence gap in the WTP guidelines for fertility, this study did not set a specific WTP threshold but sought an alternative method of cost-effectiveness threshold,37 38 which is suggested to be a more appropriate method to assess the economic value of infertility treatments. In this study, assuming that per live birth has a certain monetary value, both the costs and benefits of the intervention are monetised and can be directly compared. Multiply the LBR by the monetary value per live birth and subtract the total cost to get the NMB of each protocol. A positive NMB indicates value for money, and a protocol with higher NMB is considered more cost-effective. 10 000 Monte Carlo simulations were performed on the model, and each simulation could determine which of the two protocols had higher NMB within a range of monetary values and then calculate the proportion that each protocol had higher NMB. The result could be interpreted as the probability that a protocol had higher NMB. The final result was expressed as NMB curves, giving the probability that each protocol had higher NMB in a range of monetary values per live birth. This may to some extent compensate for the uncertainty caused by the lack of threshold.
This study has the following limitations. First, this study could only rely on published clinical research data to build the models due to the lack of a multicentre, large-sample open database in China. Other studies carried out clinical research according to data from their medical institutions and the data collection strategy is the most appropriate to produce results to represent their own institutions. This study used their data to build models, the simulation results of the models may be biased and the universality is insufficient. However, the use of simulated data demonstrates that data from other published clinical studies can be valuable in providing timely reference in settings where large clinical studies are complex to implement. In addition, model construction relies on too many assumptions and the scenario is relatively ideal, which is different from clinical practice. The results could only be used as one of the references for clinicians to choose treatment strategies. Finally, we included both IVF and ICSI patients, even though these patient groups can differ in terms of diagnosis, indications, etc. The choice of using IVF versus ICSI may have an impact on the outcome. Due to the limited literature available, most published clinical studies do not distinguish between IVF and ICSI, the effects of IVF and ICSI on the study results could not be explored at this time. However, a recent large study has shown no significant difference in clinical efficacy between IVF and ICSI.39
At present, the economic evaluation research of ART in China faces many challenges. The lack of large public databases and the evidence gap in fertility WTP guidelines are the main reasons for the slow progress in related research. The Society For Assisted Reproductive Technology (SART) has been established in the USA,40 and the success of ART in the USA is largely attributable to SART.41 Therefore, China should establish a large ART database and conduct relevant research on the scope of WTP of ART. Although using the model is convenient and quick to conduct economic evaluation research, the idealisation of the model scenario will make the results different from the clinical practice. Future prospective studies, long-term follow-up of patients, large samples and multicentre randomised controlled trials are needed to explore more cost-effective pregnancy assistance protocols in a more comprehensive and systematic way.
Conclusion
In conclusion, there is strong uncertainty about the cost-effectiveness value of long and antagonist protocols in the absence of a defined monetary value per live birth. For the Chinese population, when only one fresh cycle is considered, the long protocol is the preferred option when the monetary value per live birth is greater than￥65 420 (US$9726). When subsequent frozen cycles are considered and the monetary value per live birth is greater than ￥66 400 (US$9872), the long protocol is the preferred option. This threshold also varies for women of different ages and ovarian response capacities. For women in POSEIDON group 2, group 3 and group 4, antagonist protocol is recommended as the preferred protocol. The results of this study need to be verified by further large-scale randomised controlled trials.
Data availability statement
Data are available upon reasonable request. All data can be obtained by contacting the corresponding author.
Ethics statements
Patient consent for publication
Ethics approval
Not required.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
YS and CC contributed equally.
Contributors All authors conceived the study. YS, CC, YT and MZ chose the relevant parameters for the model. YS, YT and MZ built and ran the model. YS, JT and KP analysed and interpreted the data. YS and CC drafted the manuscript. All authors contributed critical revision to the paper and approved the final manuscript. KP is the guarantor of this article.
Funding This study was supported by grants from (1) National Social Science Fund (22XGL012); (2) Future Medical Research Innovation Team Project of Chongqing Medical University (W0081); (3) Philosophy and Social Sciences Innovation Team Project of Chongqing Medical University (ZX190101) and (4) The Special research and development plan of smart medicine for Postgraduates of Chongqing Medical University (YJSZHYX202202).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.