Economic evaluations of assisted reproductive technologies in high-income countries: a systematic review.

As education and career opportunities have expanded for women, many women in high-income countries have elected to delay starting a family Chua et al. (2020) . With increasing age, there is attrition in oocyte number and increased risk of aneuploidy in the remaining oocytes ( Yatsenko and Rajkovic, 2019 ; Chua et al. , 2020 ). Concurrently, endometriosis, tubal disease, and increased BMI may coincide as impediments to fertility ( Broughton and Moley, 2017 ; Zondervan et al. , 2020 ). Infertile women are likely to experience symptoms of anxiety more than others ( Kiani et al. , 2020 ). Combinations of all these factors together lead many women to consider assisted reproduction ( Somigliana et al. , 2016 ). The term ART encompasses any technique that includes ‘the in vitro (outside of body) handling of human oocytes (eggs) and sperm or embryos for the purposes of establishing a pregnancy’ ( Newman and Chambers, 2021 ). An audit from The International Federation of Fertility Societies found a total of 6977 clinics in 2022, an increase of 776 clinics worldwide since the previous report in 2019. IVF is the mainstay of work in these clinics: a current estimate of 940 503 IVF cycles were managed in Europe in 2017, an increase of 2.4% since 2014 ( The European IVF Monitoring Consortium for the European Society of Human Reproduction et al. , 2021 ). A linear rise in the number of treatment cycles is also evident in the USA, with more than 160 000 cycles annually in the 2016 update ( Toner et al. , 2016 ). In response to high, and increasing, demand for care, IVF has become a burgeoning industry in all high-income settings ( International Federation of Fertility Societies’ Surveillance (IFFS), 2019 , 2022 ). The economic burden of infertility as well as approaches to reduce costs and increase access have been identified in the top 10 research priorities for future infertility research ( Duffy et al. , 2021a ). As of 2020, 19 American states require insurers to offer infertility treatment coverage, but IVF is sometimes excluded from the infertility treatments offered, or the number of funded IVF cycles is limited ( Kawwass et al. , 2021 ). The European IVF-Monitoring Consortium notes there is no public funding available in Switzerland, and that of the 39 countries providing some funding, there is a three-cycle limit in 16 countries ( Calhaz-Jorge et al. , 2020 ). The provision or proportion of government funding can also vary with maternal age, existence of older children, and female BMI ( Calhaz-Jorge et al. , 2020 ; Kawwass et al. , 2021 ). The impact of high cost on individuals or couples in an otherwise publicly funded healthcare system may be one reason why only half of surveyed British adults sought treatment for infertility ( Datta et al. , 2016 ). Furthermore, over one-third of couples surveyed from eight different countries acknowledged cost was a barrier to accessing or proceeding with assisted reproduction ( Domar et al. , 2021 ). There is also sometimes a gulf between preparedness to pay and capacity to pay for help with infertility treatments ( Dyer et al. , 2017 ). IVF access has socioeconomic associations: access is easier for those with higher incomes and higher educational achievement ( Tierney and Cai, 2019 ). Ultimately both health care funders and those seeking care for infertility require treatments to be supported by high-quality evidence demonstrating improved outcomes. Better outcomes might justify additional cost. Over the past decade, there has been an increasing number of ART interventions implemented in routine clinical practice globally, resulting in a shift towards single embryo transfer (SET), blastocyst transfer, freeze-all transfer strategy, the use of preimplantation genetic testing for aneuploidy (PGT-A), and ICSI, as well as other IVF add-ons ( Kushnir et al. , 2017 ; Gleicher et al. , 2019 , 2021 ; Bahadur et al. , 2020 ; de Geyter et al. , 2020 ; Lensen et al. , 2021 ). In addition to the concerns about the lack of evidence on the effectiveness and/or safety for some interventions, other concerns have also been raised about the lack of robust assessments of economic evaluations before implementing most interventions in clinical practice ( ESHRE Capri Workshop Group, 2015 ). Given that the last systematic review on the cost-effectiveness of ART interventions was conducted in 2002 ( Garceau et al. , 2002 ) and ART clinical practice has evolved, it is important to systematically map and assess the latest economic evaluations of ART interventions available in current clinical practice to inform health policy and future research. This systematic review aims to provide an overview of economic evaluations of all ART interventions performed in high-income countries in order to assist both couples and clinicians, and to synthesize the economic evaluations and the reporting quality of these studies.

Of the 505 papers identified in database searching, 40 papers fulfilled our eligibility criteria of an economic evaluation of an intervention in assisted reproduction in the last decade ( Fig. 1 ). PRISMA flowchart for identifying studies included in a systematic review of economic evaluations of ART in high-income countries . The key characteristics of included studies are provided in Table 1 . Two-thirds of included papers were published after 2016 and half were from European settings. Studies addressed either a general (unselected) population (n = 26), poor prognosis patients (n = 7), unexplained infertility (n = 5), male infertility (n = 1), or predicted high responders (n = 1). The ART interventions studied included controlled ovarian stimulation (n = 15), IVF (n = 9), PGT-A (n = 7), SET (n = 5), ICSI (n = 3), and freeze-all embryo transfer (n = 1). The details of these interventions are visualized in Fig. 2 . Live birth was the most reported primary outcome (n = 27), followed by ongoing pregnancy (n = 6) and quality-adjusted life years (n = 3). In 26/40 studies, the economic evaluation was by way of a model. The clinical efficacy inputs were calculated from at least some trial data in 10/26 studies, while inputs in the other 16 modelling studies were based on observational data. The perspective was solely the health funder in 32 of the 40 papers. A cost-effectiveness analysis (rather than cost-utility or cost minimization) was provided in 34 papers. Sunburst diagram: research landscape of interventions in economic evaluations of ART. SET, single embryo transfer; EM, expectant management; PGT-A, preimplantation genetic testing for aneuploidy; COS, controlled ovarian stimulation. COS n = 15 studies, IVF n = 9, PGT-A n = 7, SET n = 5, ICSI n = 3, freeze-all n = 1. Overview of 40 included studies of an ART intervention together with a cost analysis. Gizzo et al. (2017) examines outcomes in Italy and Spain, and Gizzo et al. (2018) examines outcomes in Italy, Spain, and Germany. The papers by Mennini et al. (2018) and Scotland et al. (2011) measured both cost per additional birth and quality-adjusted life years per additional birth. PGT-A, preimplantation genetic testing for aneuploidy. Comprehensive details of the CHEERS assessment results are provided in Supplementary Table S2 . Reporting of both the target population and time horizon of the study was noted in 32 of 40 studies whereas no study reported on patient preferences for outcomes. Those studies comparing immediate IVF with expectant management, and the group of studies reporting embryo transfer number, had greater adherence to the checklist relative to studies on other interventions. The perspective of the economic evaluation was noted in 38 of 40 included papers. The source of funding considered in the economic evaluation was explicitly stated in 31 papers. Of the 10 studies comparing agents for ovarian hyperstimulation, nine acknowledged their funding source, and eight of these nine were funded by industry. Among 10 included studies on IVF, nine evaluated IVF versus non-IVF interventions (four on expectant management and five on IUI) and the other study compared different IVF funding policies ( Fig. 2 ). Consideration of a delay in IVF start for those with unexplained infertility was explored in four studies ( Supplementary Table S3 ). Pham et al. (2018) assessed the cost-effectiveness of a 6-month delay for women with unexplained infertility aged <40 years, from both the health funder and the patient perspective. Eijkemans assessed the cost-effectiveness of a 12-month delay for couples from the health funder perspective only ( Eijkemans et al. , 2017 ). van Eekelen assessed the cost-effectiveness of ovarian hyperstimulation and IUI as an intermediary step between expectant management and IVF ( van Eekelen et al. , 2020 ). Finally, Busnelli assessed the cost of IVF rather than expectant management in a group of women pre-defined to be poor responders ( Busnelli and Somigliana, 2018 ). All papers used the outcome measure of cost per livebirth. An 18-month time frame was used to model the impact of a 6-month period of expectant management in women <40 years of age in Australia. A reduction in livebirth rate of 0.1–0.2% was associated with a delayed start of IVF treatment ( Pham et al. , 2018 ). Eijkemans et al. noted that IVF becomes more cost-effective compared to expectant management as women age. van Eekelen noted that expectant management for 1–2 years before beginning IVF lowers the average cost per livebirth compared to an immediate IVF start ( van Eekelen et al. , 2020 ). Four papers used a health funder perspective to evaluate IUI as an alternative to IVF ( Supplementary Table S4 ) ( van Rumste et al. , 2014 ; Yu et al. , 2014 ; Tjon-Kon-Fat et al. , 2015 ; Almaslami and Aljunid, 2020 ). All found IVF to be slightly more effective than IUI to achieve an ongoing pregnancy, livebirth, or healthy child. Three of the four studies found a higher cost associated with IVF, resulting in ICERs of an additional US $50 492 per additional ongoing pregnancy, an additional $37 925 per additional live birth or an additional $50 490 per additional healthy child, respectively. A Canadian study modelled the impact of provincial funding strategies for IVF. Results were stratified for embryo transfer number and maternal age ( Vaidya et al. , 2015 ). Cost-effectiveness was greater when double embryo transfer (DET) was permitted at higher maternal ages. Three papers examined the cost-effectiveness of use of ICSI for fertilization in IVF cycles where male factor infertility was not present ( Supplementary Table S5 ). In models designed from a health funder perspective in the USA and the Netherlands, ICSI had a similar ICER per additional live birth gained for unexplained infertility ($66 742) and mild male factor infertility ($72 000) ( Vitek et al. , 2013 ; Moolenaar et al. , 2015 ). Rescue ICSI in the setting of total fertilization failure (as an alternative to cycle cancellation) resulted in 11% more live births at lower cost in the third study ( Shalom-Paz et al. , 2011 ). Seven papers included in this review assessed the economics of using PGT-A as an adjunct to IVF, with the outcome measure of live birth ( Table 2 ). In five studies, the funder perspective was chosen, but three of these also calculated costs from the patient perspective ( Lee et al. , 2019 , 2021 ; Neumann et al. , 2020 ). Comparison of IVF with and without preimplantation genetic testing for aneuploidy. PGT-A With a single blastocyst at maternal age (years): RCT, randomized controlled trial; ICER, incremental cost-effectiveness ratio. In two studies, PGT-A lessened the live birth rate and increased costs ( Somigliana et al. , 2019 ; Lee et al. , 2021 ). In the three studies using a health funder perspective and reporting higher live birth rates with PGT-A, the ICER to achieve an additional live birth ranged from an additional $20 507 to $607 833 ( Lee et al. , 2019 ; Neumann et al. , 2020 ). Among 15 studies on controlled ovarian stimulation, three evaluated individualized controlled ovarian stimulation, one evaluated the downregulation protocol, one evaluated natural cycle versus stimulated IVF, and the other 10 studies reported medication for ovarian stimulation ( Fig. 2 ). Three papers assessed ovarian reserve testing from a health funder perspective with a view to individualizing the gonadotrophin dose for ovarian hyperstimulation ( Supplementary Table S6 ). Moolenaar modelled both IVF with standard dosing and IVF with individualized dosing compared to no treatment, and concluded that both choices were likely to be considered to be cost-effective ( Moolenaar et al. , 2011 ). Yates found individualized dosing resulted in additional ongoing pregnancies with an ICER of an additional $6067 per ongoing pregnancy ( Yates et al. , 2011 ). van Tilborg et al. (2017) subsequently conducted a multicentre prospective cohort study of 1515 women and found that individualized dosing resulted in fewer pregnancies at greater expense. A cost-minimization study based on a retrospective cohort examined progesterone as a downregulation agent ( Supplementary Table S7 ) ( Evans et al. , 2019 ). The cost per livebirth was measured from the perspective of the health funder over a 2-month time horizon. The GnRH antagonist downregulation cost $6676 less to achieve a live birth compared to progesterone, and the GnRH agonist downregulation was also $9753 cheaper than progesterone to achieve a live birth. A retrospective cohort study was used to evaluate the cost-effectiveness of six cycles of modified natural cycle for ovulatory women against a single IVF cycle with controlled ovarian hyperstimulation ( Supplementary Table S8 ) ( Groen et al. , 2013 ). The cost per live birth was measured from the health funder perspective with a 10-month time horizon. A single IVF cycle with ovarian hyperstimulation was cheaper and as effective as six cycles of modified natural cycle. An economic evaluation of agents for ovarian hyperstimulation was the topic of 10 included papers ( Supplementary Table S9 ). A cost-effectiveness evaluation was described in 9 of 10 papers. The perspective was solely that of the health funder in eight papers, whereas the remaining two reported from both the perspective of the patient and health funder ( Fragoulakis et al. , 2012 , 2016 ; Gizzo et al. , 2017 , 2018 ; Barrenetxea et al. , 2018 ; Barriere et al. , 2018 ; Grynberg et al. , 2018 ; Mennini et al. , 2018 ; Xue et al. , 2019 ; Robins et al. , 2020 ). Four studies were cost-effectiveness analyses of use of a biosimilar compared to recombinant FSH (rFSH). The incremental cost to use rFSH over a comparator varied from an additional $506 to $8474 ( Gizzo et al. , 2018 ; Xue et al. , 2019 ). Six papers assessed the cost policies of transferring one or more embryos when more than one is available ( Scotland et al. , 2011 ; Hernandez Torres et al. , 2015 ; Vaidya et al. , 2015 ; van Heesch et al. , 2016 ; van Loendersloot et al. , 2017 ; Sitler et al. , 2020 ) ( Supplementary Tables S10 and S11 ). The extreme variation in time horizon chosen for economic evaluation impacts on comparability of these papers. Transfer of two embryos was found to be increasingly cost-effective with increasing maternal age in three cohort papers( Scotland et al. , 2011 ; Vaidya et al. , 2015 ; van Loendersloot et al. , 2017 ). The incremental cost to attain an additional live birth with transfer of two embryos rather than one fell from an additional $43 000 for a 32-year-old mother to $24 000 for a 39-year-old mother ( Scotland et al. , 2011 ). Dutch and Canadian studies, also from a funder perspective, showed the same trends ( Vaidya et al. , 2015 ; van Loendersloot et al. , 2017 ). Cost-effectiveness varied with perspective: in the USA, SET is the dominant strategy from the health funder perspective, whereas an incremental cost of $27 145 is associated with SET for the patient. This is explained by the patient paying for IVF cycles, and the funder paying for obstetric and paediatric care ( Sitler et al. , 2020 ). A single paper was included addressing this topic for cost-effectiveness from a funder perspective ( Supplementary Table S12 ). Based on observational evidence, a freeze all policy was more effective and less costly than fresh embryo transfer ( Papaleo et al. , 2017 ).

This systematic review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist ( Page et al. , 2021 ). The review protocol was registered with the international prospective register of systematic reviews (PROSPERO) (CRD42021261537). Economic evaluation studies on any type of ART interventions were included in the review, including cost-effectiveness, cost-benefit, cost-utility, and cost-minimization studies. Studies were included if they reported results from high-income countries as defined by the World Bank ( World Bank, 2021 ), and were published in English language, peer reviewed journals, and since January 2011. We included studies irrespective of the outcome measure assessed (e.g. live birth). Studies were excluded if they assessed ART for indications other than infertility (e.g. egg freezing). The following databases were searched for eligible studies ( Supplementary File S1 ): MEDLINE, PUBMED, CINAHL, ECONLIT, COCHRANE, EMBASE, and SCOPUS. The reference lists of all included studies were also checked to identify additional eligible studies ( Supplementary Table S1 ). The last search was conducted on 22 June 2022. Two reviewers (E.O. and C.B.) independently screened titles and abstracts in Covidence systematic review software ( Veritas Health Innovation, 2021 ). The same two reviewers then independently assessed full-text articles for inclusion. Where there were disagreements about article inclusion, reviewers first attempted to resolve this through discussion to reach consensus. Where consensus could not be reached, a third reviewer (E.C. or R.W.) adjudicated the final decision. The professional society for health economics and outcomes research (ISPOR) recommends that health economic studies be evaluated according to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) ( Husereau et al. , 2022 ). This checklist is a reporting rather than quality evaluation and contains 24 items to be included for thorough reporting. Documentation of adherence to the CHEERS reporting criteria is included in Supplementary Table S2 . Data were extracted to an Excel spreadsheet by one author (E.O.) was reviewed by C.B. and E.C. for accuracy. Extracted information included: author, year, population studied, perspective taken, time horizon, intervention and comparator, costs associated with the intervention and comparator, incremental costs, and discounting of costs. The key effect measure studied was the incremental cost-effectiveness ratio (ICER). We converted all cost outcomes to 2021 costs using https://www.inflationtool.com . To enable comparisons across studies, we converted all cost data into US Dollars (USD) using https://www.xe.com . The key characteristics of studies were tabulated descriptively. Types of interventions investigated in economic evaluations were visualized in a sunburst. Findings of different studies were grouped according to the type of interventions evaluated. A synopsis of the findings of economic evaluations for each intervention studied is provided in the Results section below.

In economic evaluations, stated gains in effectiveness should be based on high-quality data to avoid a potentially biased economic assessment. In this systematic review, we found that not all economic evaluations are based on such evidence on effectiveness. In interventions with no proven clinical benefits, economic evaluations could be of limited value. Outcomes of ART studies should ideally be reported with gestational age at delivery and birthweight data stratified by IVF intervention. There are long-term health and educational costs associated with early delivery and small for gestational age birthweight. Economic evaluation time horizons should be long enough to capture childhood outcomes ( Bentley et al. , 2018 ). Use of quality evidence on effectiveness and measurement of obstetric and neonatal outcomes should pave the way for improved cost-effective use of ART.

This review identifies and describes studies reporting a contemporary economic evaluation of ART interventions in high-income countries. These evaluations involved a limited number of interventions, including IVF, ICSI, PGT-A, controlled ovarian stimulation, SET, and freeze-all embryo transfer, but most other IVF add-ons were not reported in any economic evaluation study. Dominant interventions are rare in the ART sphere, as incremental gains in clinical efficacy are relatively small while new technologies are relatively expensive for most interventions. Interventions such as use of biosimilars or HMG for ovarian stimulation are cheaper but compromise clinical effectiveness. IVF per se is more effective and more expensive than expectant management or IUI. High-cost interventions, including PGT-A or ICSI for non-male factor infertility, seem unlikely to be clinically effective, and are therefore not likely to be cost-effective. Cost-effectiveness around the number of embryos transferred depends on the group studied, the outcomes measured, and the time horizon used. In future work the preferences of those affected should be reported, as this impacts on the feasibility of practice change implementation. Economic evaluations are dependent on the quality of evidence on effectiveness. Interpretation of these findings should therefore be cautious. Our review showed that not all included studies used robust evidence from clinical trials and systematic reviews on clinical efficacy to inform their economic evaluations. Some of the claimed efficacy seems to be exaggerated when compared to the latest high-quality evidence from clinical trials and Cochrane reviews. In these cases, the new intervention is either not effective or there is a lack of evidence on effectiveness, and therefore consensus has not been reached prior to conducting economic evaluations. Nevertheless, these economic evaluations were conducted and published, which results in research waste and may misinform practice and policy. An example was the single included study of a freeze-all strategy showing greater efficacy and lower cost than fresh transfer, but with results based on a relatively small observational dataset on effectiveness. We note that included cohort studies find that DET is increasingly cost-effective in two instances: in populations of advanced maternal age and also when cost-effectiveness is assessed with a time horizon extending until adulthood of the resultant children ( Scotland et al. , 2011 ; Vaidya et al. , 2015 ). In 2014, two embryos were transferred in 16.6% of cycles in Australia but 54.5% of fresh cycles in Europe ( Harris et al. , 2016 ; de Geyter et al. , 2018 ). Offering DET to older women aims to create more singleton pregnancies, and to lessen the proportion of couples who would otherwise remain childless. No randomized trial of embryo transfer number in women of advanced maternal age is available to support or refute the findings of the included cohort studies. Use of PGT-A varies from 2% in the UK to 27% in the USA in 2016. Trials published subsequently have shown no improvement in the live birth rate related to the addition of PGT-A to a standard IVF cycle. ( Ozgur et al. , 2019 ; Theobald et al. , 2020 ; Yan et al. , 2021 ). The apparent lack of improvement in clinical efficacy is the driver for the high ICERs noted in this review ( Collins et al. , 2017 ; Neal et al. , 2018 ; Lee et al. , 2019 , 2021 ; Somigliana et al. , 2019 ; Neumann and Griesinger, 2020 ; Neumann et al. , 2020 ). ICSI for non-male infertility does not increase live birth rates ( Boulet et al. , 2015 ; American Society for Reproductive Medicine, 2020 ; Dang et al. , 2021 ) and is therefore not a cost-effective intervention. Both PGT-A and ICSI for non-male factor infertility had, however, become accepted options in some centres prior to availability of high-quality data reporting outcomes ( Peipert et al. , 2021 ). National spending on ART is valuable to society as well as family units ( ESHRE Capri Workshop Group, 2015 ). The 2021 Australian Intergenerational Report notes that fertility rates are directly associated with national productivity. Aged care spending is anticipated to almost double between 2020–2021 and 2060–2061 while the proportion of the working age population falls ( The Commonwealth of Australia, 2021 ). National spending on ART therefore appears supported for the benefit of individuals but also state funders. As there will always be competing demands for funding from a national health budget, careful outcome measurement helps substantiate cost-effectiveness. To this end, adherence to a consistent optimal outcome measure (healthy term infant) and consistent gestation thresholds matter ( Duffy et al. , 2021b ). Whereas infertile women who conceive without ART have an odds ratio (OR) 1.38 (95% CI 1.25–1.54) for preterm birth compared to those who conceive spontaneously, those with an ART conception are even more likely to deliver preterm, with an OR 2.46 (1.46–4.18) compared to women with natural conception ( Messerlian et al. , 2012 ; Cavoretto et al. , 2022 ). Strategies to improve these outcomes in turn improve cost-effectiveness of the ART investment. It is noteworthy that no included study included reference to the preferences of infertile patients. Recent research has noted that women are prepared to accept an increased risk of hyperstimulation if they thought it would increase their likelihood of conception ( Stormlund et al. , 2019 ; Abdulrahim et al. , 2021 ). At present there are too few evaluations of the preferences of women for infertility treatments which might result in a more complex pregnancy and higher risks of preterm birth or low birthweight. Decisions around effective use of allocated funds to treat infertility require incorporation of the informed preferences of those affected and the willingness of the payer to pay a premium for small incremental gains. Perceptions of cost-effectiveness will vary, and families should be entitled to spend more if they wish in pursuit of small gains. As care providers, our duty is to provide accurate information on which to base these decisions. The strength of this paper is the inclusion and comparison of cost-effectiveness studies from countries with different health funding structures, enabling direct financial comparisons. Some included studies use measures of effectiveness that may be thought invalid or exaggerated in the wider literature ( Shields and Elvidge, 2020 ). Variation in the populations studied alter cost-effectiveness figures as clinical efficacy may change considerably depending on the patient cohort studied. In addition, limiting the review to high-income countries may be seen as a weakness given the enormous growth of ART in emerging economies. Economic analysis to allow selection of clinically effective interventions and to allow selection of the most cost-effective strategy is arguably even more valid in middle- and low-income countries ( Chiware et al. , 2021 ).