A Clinical and Cost-Effectiveness Analysis

PREIMPLANTATION GENETIC DIAGNOSIS FOR ANEUPLOIDY IN WOMEN OF ADVANCED MATERNAL AGE: A CLINICAL AND COST-EFFECTIVENESS ANALYSIS

Evelyn Lee

BA, MA (Econ)

A thesis in fulfilment of the requirements for the degree of

Doctor of Philosophy

Centre for Big Data Research in Health and School of Women’s and Children’s Health Faculty of Medicine, University of New South Wales Sydney, Australia

31 July 2018

Declaration of Originality

Signature

iii

Abstract

Significant improvements in the treatment of infertility have been achieved through advances in assisted reproductive technologies (ART). However, achieving acceptable success rates in women of advanced maternal age using their own oocytes remains an intractable challenge, primarily due to increasing rates of aneuploidy in embryos created from women towards the end of their reproductive lives.

Preimplantation genetic diagnosis for aneuploidy (PGD-A) which screens all 24- chromosomes, is increasingly used in clinical practice to select euploid embryos for transfer. However, there is a dearth of evidence regarding its effectiveness beyond a single

ART cycle.

This thesis is constructed around four interrelated studies, providing new evidence on the clinical and cost-effectiveness of PGD-A in women of advanced maternal age over repeated ART cycles.

Study One provides a systematic review of relevant literature published up to 2017. Study

Two is a retrospective cohort analysis of 2,093 ART-naive women aged 37-years or older who commenced either PGD-A or conventional morphological assessment for the selection of embryos for transfer. Although the per-cycle live-birth rate was higher in the

PGD-A group, the cumulative live-birth rates (CLBR) for up to three ‘complete ART cycles’ (fresh plus frozen embryo transfers) were not statistically different. However, the

PGD-A group had a lower rate of pregnancy loss, required less time and fewer ART cycles to achieve a live-birth. Study Three extended the analysis by performing a patient- level cost-effectiveness analysis, concluding that PGD-A is likely to be considered cost- effective from a healthcare perspective, with an incremental cost-effective ratio per live- birth of 28,103 Australian dollars. The final study used a Markov model to evaluate the

vi cost-effectiveness of standard ART treatment compared to PGD-A, ‘social egg freezing’, and donor ART. Notwithstanding the limitations of modelling studies, the analysis suggests that PGD-A is the most cost-effectiveness strategy from a healthcare perspective.

In conclusion, this doctoral research program found that in women of advanced maternal age undergoing ART treatment, PGD-A resulted in a similar CLBR compared to standard

ART treatment but had a lower rate of pregnancy loss, required shorter time to achieve a live-birth and fewer cycles needed-to-treat. PGD-A is also likely to be considered cost- effective in this patient group undergoing ART treatment from a public-funding perspective.

vii

Acknowledgements

At the end of this long journey, I would like to extend my appreciation to my supervisor, A/P Georgina Chambers for her guidance over and above her professional duty of supervision. Thank you for your incisive comments on the drafts which I believe has raised the quality of this research.

I would like to thank Dr Michael Costello for his guidance on clinical issues despite his busy schedule.

Many thanks, to Dr Willings Botha for his valuable input on the economics aspect and continual encouragement. A special thanks to Limin for her good advice and support.

Finally, for my family whose love and support has kept me steadfast in resolve throughout the candidature. Thank you.

viii

Table of content

Abstract...... vi Acknowledgements ...... viii List of Tables ...... xii List of Figures ...... xiii List of abbreviations ...... xiv List of publications arising from this thesis ...... xv Chapter 1- Introduction...... 2 1.1 Background ...... 2 1.2 Aim and contribution ...... 7 1.3 Thesis outline ...... 11 Chapter 2: Epidemiology of Infertility ...... 15 2.1 Introduction ...... 15 2.2 Fecundity and fertility ...... 15 2.3 Global estimates of infertility ...... 21 2.4 Country-specific estimates ...... 22 2.5 Temporal trends in infertility rates ...... 24 2.6 Causes of infertility ...... 26 2.7 Overview of commonly used non ART treatments ...... 31 2.8 Summary ...... 51 Chapter 3– Overview of assisted reproductive technology and intrauterine insemination ...... 53 3.1 Introduction ...... 53 3.2 Role of enhanced ART techniques and protocols in women of advanced maternal age……...... …………………………………….. …………………………. 61 3.3 Role of ART in women of advanced maternal age ...... 71 3.4 Summary ...... 75 Chapter 4–Measuring ART success ...... 78 4.1 Introduction ...... 78 4.2 Commonly used numerators in measuring ART success ...... 79 4.3 Commonly used denominators in measuring ART success ...... 83 4.4 Current challenges in measuring ART success...... 88 4.5 Measuring success rate in older women ...... 93 4.6 Summary ...... 93

Chapter 5- Economics and ART ...... 96 5.1 Introductoin ...... 96 5.2 Economic evaluation in healthcare...... 96 5.3 Model based economic evaluation ...... 107 5.4 Other elements in economic evaluation ...... 109 5.5 Summary ...... 110 Chapter 6- The clinical effectiveness of pre-implantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A): systematic review (updated) ...... 113 6.1 Introduction ...... 113 6.2 Methods ...... 115 6.3 Results ...... 116 6.4 Discussion ...... 124 6.5 Summary ...... 126 Chapter 7-ART cumulative live-birth rates using preimplantation genetic diagnosis to screen for embryo aneuploidy (PGD-A): a cohort analysis ...... 128 7.1 Introduction ...... 128 7.2 Background ...... 128 7.3 Aim of the study ...... 130 7.4 Methods ...... 130 7.5 Results ...... 135 7.6 Discussion ...... 147 7.7 Summary ...... 152 Chapter 8- Longitudinal analysis of PGD-A in women of advanced maternal age: a cost-effectiveness analysis Introduction ...... 154 8.1 Introduction ...... 154 8.2 Background ...... 154 8.3 Aim of the study ...... 157 8.4 Methods ...... 157 8.5 Results ...... 167 8.6 Discussion ...... 176 8.7 Summary ...... 180 Chapter 9: Which ART treatment strategy is the most clinically and cost-effective for women of advanced maternal age? A Markov Model ...... 182 9.1 Introduction ...... 182 9.2 Background ...... 182 9.3 Aim of the study ...... 184

9.4 Choosing the appropriate decision model ...... 184 9.5 Methods ...... 185 9.6 Results ...... 204 9.7 Discussion ...... 214 9.8 Summary ...... 218 Chapter 10- Discussion ...... 221 10.1 Overview ...... 221 10.2 Main findings of the study...... 221 10.3 Direction of future research ...... 229 10.4 Concluding remarks ...... 232 References ...... 233 Appendix ...... 267 Appendix 1: The clinical effectivenss of PGD-A in all 24 chromosomes: systematic review…………...... 267 Appendix 2: ART CLBR following PGD-A or morphological assessment of embryos: a cohort analysis ...... 278 Appendix 3: Downs and Black checklist ...... 286 Supplementary Table 1: Detail of study included in the updated systematic review .. 289 Supplemnetary Table 2 :Demographic and outcomes according to per-protocol analysis...... 307 Supplementary Table 3: Clinical outcomes for up to three ‘complete cycles” ...... 309

List of Tables

Table 1: Self-reported causes of infertility among women in Scotland, 2009 ...... 26

Table 2: Demographic characteristic and outcomes of first three ‘single cycles’ (fresh or frozen/thaw cycles) according to intention-to-treat analysis ...... 140

Table 3: Clinical outcomes for up to three ‘complete ART cycles’ according to intention- to-treat analysis ...... 142

Table 4: PBS item code and costs associated with an ART treatment cycle, 2015 ...... 159

Table 5: Medicare Benefits Schedule (MBS) item numbers, MBS benefits and out-of- pocket expenses associated with an ART treatment cycle, 2015 ...... 163

Table 6: Cost of treatment and number of ART treatment and additional procedures undertaken by the PGD-A group and the morphology assessment group, 2015 ...... 165

Table 7: Incremental cost-effectiveness ratio (ICER) for the PGD-A group versus the morphology assessment group (base case results) ...... 169

Table 8: Incremental cost-effectiveness ratio (ICER) for the PGD-A group versus the morphology assessment group (patient perspective) ...... 174

Table 9: Clinical data sources ...... 195

Table 10: Cost associated with an ART treatment cycle and distribution, 2015 ...... 200

Table 11: Base case estimates (healthcare perspective) ...... 206

Table 12: Base case estimates (patient perspective) ...... 213

xii

List of Figures

Figure 1: Trends in ART treatment, embryo transfer cycles and live-birth and multiple- birth rates, Australia 2002–2015 ...... 3 Figure 2: Relationship between studies carried out for this doctoral research program . 10 Figure 3: Rate of aneuploidy relative to age of female partner...... 30 Figure 4: Time to clinical pregnancy from date of first OPU to conception resulting in a live birth for PGD-A group and morphology assessment group...... 87 Figure 5: Search flow diagram based on PRISMA ...... 117 Figure 6: Pathway profile of women in the study ...... 137 Figure 7: Intention-to-treat analysis: Cumulative live-birth rates (CLBR) of first and subsequent ‘complete cycles’ for all women and among women who had an embryo transfer procedure in their first fresh cycle ...... 144 Figure 8:Intention-to-treat analysis: Time to clinical pregnancy from date of first oocyte pick-up to clinical pregnancy resulting in a live birth for the PGD-A group and the morphology assessment group ...... 146 Figure 9: Cost-effectiveness plane showing the relationship between the incremental cost and effect between the PGD-A group and the morphology assessment group from a healthcare perspective ...... 171 Figure 10: Cost-effectiveness acceptability curve (CEAC) from a healthcare perspective ...... 172 Figure 11: Cost-effectiveness plane showing the relationship between the incremental cost and effect for the PGD-A group and the morphology assessment group from a patient perspective...... 175 Figure 12: Types of ART strategies used in women of advanced maternal age ...... 189 Figure 13: Transition of patients undertaking different ART strategies in the model .. 193 Figure 14: Cost-effectiveness plane ...... 203 Figure 15:Cost-effectiveness plane for PGD-A versus reference autologous ART strategy ...... 208 Figure 16: Cost-effectiveness plane for social egg freezing strategy versus reference autologous ART strategy...... 209 Figure 17: Cost-effectiveness plane for donor ART strategy versus reference autologous ART strategy ...... 210 Figure 18: Cost-effectiveness acceptability curves over a range of willingness to pay thresholds ...... 211

xiii

List of abbreviations

aCGH array comparative genomic hybridisation ANZARD Australian and New Zealand Assisted Reproductive Database ART assisted reproductive technologies ASRM American Society for Reproductive Medicine AUD Australian dollar CDC Centres for Disease Control and Prevention CI Confidence interval CLBR Cumulative live-birth rate COH Controlled ovarian hyperstimulation DET double embryo transfer ESHRE European Society of Human Reproduction and Embryology FET frozen embryo transfer FISH Fluorescence in Situ hybridization FSH follicle stimulating hormone HTA Health Technology Assessment ICER Incremental cost-effectiveness ratio International Committee Monitoring Assisted Reproductive ICMART Technologies ICSI intracytoplasmic sperm injection IUI Intrauterine insemination IVF In vitro fertilisation MBS Medical benefits scheme NICE National Institute for Health and Care Excellence NSFG National Survey of Family Growth OHSS ovarian hyperstimulation syndrome OOP out-of-pocket OPU oocyte pickup PBAC Pharmaceutical Benefits Advisory Committee PBS Pharmaceutical benefits scheme PGD-A Preimplantation genetic diagnosis for aneuploidy PRISMA Preferred Reporting Items for Systematic Reviews and Meta–analyses PSA Probabilistic Sensitivity Analysis QALY Quality-adjusted life year RCT Randomised controlled trial SET single embryo transfer TTP time to pregnancy UK United Kingdom US United States USD United States Dollar WHO World Health Organisation WTP willingness to pay

xiv

List of publications arising from this thesis

Peer-reviewed articles published

1. Lee E, Illingworth P., Wilton L., Chambers G. M. The clinical effectiveness of preimplantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A): systematic review. Human reproduction 2014: 3(2), 473-483

2. Lee, E., Chambers, G. M., Hale, L., Illingworth, P., & Wilton, L. Assisted reproductive technology (ART) cumulative live birth rates following preimplantation genetic diagnosis for aneuploidy (PGD‐A) or morphological assessment of embryos: A cohort analysis. Australian and New Zealand Journal of Obstetrics and Gynaecology. 2017.

Oral Presentation

Lee E and Chambers G.M. Cumulative live birth rates (CLBR) following preimplantation genetic diagnosis for aneuploidy (PGD‐A): A cost-effectiveness analysis. 7th Congress of the Asia Pacific Initiative on Reproduction (ASPIRE). Kuala Lumpur, Malaysia. March 2017.

Awarded Best Presentation 2017

Poster Presentation

Lee E and Chambers G.M. The clinical effectiveness of preimplantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A): systematic review. Poster presentation. Australian Centre for Perinatal Science (ACPS) Inauguaral Symposium. UNSW Sydney, Australia. September 2013.

Chapter 1

Introduction

Chapter 1- Introduction

1.1 Background

Over the past 40 years, assisted reproductive technology (ART), such as in vitro fertilisation (IVF), has revolutionised the treatment of infertility, allowing many couples to have families who otherwise would have remained childless. Latest global estimates indicate that ART utilisation has increased fourfold over the last decade, with more than

1.8 million ART cycles undertaken in 2010 and over 6.5 million children born worldwide as a result of ART (Adamson, et al., 2006, Dyer, et al., 2016).

The latest published International Committee for Monitoring Assisted Reproductive

Technology (ICMART) world report: assisted reproductive technology revealed that

Australia had the second highest level of ART utilisation in the world, after Israel, with

2,337 ART cycles per million population performed in 2010 (Dyer, et al., 2016). Data from Australian and New Zealand Assisted Reproductive Technology Database

(ANZARD) reported that 67,602 autologous ART cycles were undertaken in Australia in

2015, which represented a greater than twofold increase from 29,393 ART cycles undertaken in 2002 (Bryant, et al., 2004, Fitzgerald, et al., 2017).

Together with this increasing utilisation rate, the success and safety of ART treatment is also improving. The ANZARD reported that the live-birth rate per embryo transfer cycle increased by more than 30% between 2002 and 2015, from 19.2% to 25.3%, whereas the multiple birth rate declined from 18.9% to 4.4% over the same period (Bryant, et al.,

2004, Fitzgerald, et al., 2017). Figure 1 summarises the trend in ART treatment in

Australia between 2002 and 2015.

Figure 1: Trends in ART treatment, embryo transfer cycles and live-birth and multiple-birth rates, Australia 2002–2015

Sources: (Bryant, et al., 2004, Fitzgerald, et al., 2017, Macaldowie, et al., 2012)

The reasons behind the relatively high ART utilisation levels in Australia are multifactorial; however supportive public funding through Australia’s universal healthcare insurance scheme, Medicare, is an important factor (Chambers, et al., 2013).

In Australia, almost all ART treatment cycles receive partial reimbursement from the

Medicare Medical Benefits Scheme (MBS) and Pharmaceutical Benefits Scheme (PBS), with no restrictions based on female age, number of previous ART cycle attempts or number of existing children (Medicare Australia, 2015).

Therefore, Australia has a unique population for evaluating the effectiveness of ART treatment in women of advanced maternal age. In 2015, one in four (24.8%) ART cycles performed in Australia were in women aged 40 years and over, increasing from one in

3 five in 2005 (19.6%), with this age group continuing to be the fastest growing group for

ART use (Fitzgerald, et al., 2017).

Despite the growing use of ART and increasing success rates overall, infertile women of advanced maternal age remain the most challenged in terms of achieving a live birth, particularly women who prefer to use their own eggs for ART (autologous treatment).

For women aged 40 years or over, less than one in ten (7.5%) initiated autologous cycles result in a live birth compared to 26.2% of cycles undertaken by women aged below 30 years, 24.3% in women aged 30–34 years, and 18.4% in women aged 35–39 years following ART (Fitzgerald, et al., 2017).

A similar low rate of ART success in women of advanced maternal age is reported worldwide. For example, recent data generated from European registries by the European

Society of Human Reproduction and Embryology (ESHRE) reported that for women aged

40 years or over initiating IVF, less than one in ten achieved a delivery (8.1%) compared to almost one in four women aged below 34 years (22.7%%) and 18.1% for women aged between 35 and 39 years (Calhaz-Jorge, et al., 2016)

With the continuing trend towards delayed childbearing and the growing need for fertility treatment to improve the chance of having a child, the ART utilisation rate among older women is likely to increase considerably. In fact, in Australia, the proportion of first-time mothers aged 35–39 years almost trebled from 4.5% to 11.7% between 1991 and 2011, and the proportion of first births for women aged 40–44 years quadrupled from 0.6% to

2.6% over the same period (Lancaster, et al., 1994, Li, et al., 2013).

The majority of all embryos — whether conceived naturally or through ART — are lost preclinically and a major reason is the chromosome number imbalance (or aneuploidy)

(Hassold, et al., 2007). While embryos resulting from women of all ages can present with

4 aneuploidy, the rate of aneuploidy increases almost exponentially after the age of 36–37 years (Franasiak, et al., 2014) and this has been shown to be largely responsible for the low success rates in older women (Fragouli, et al., 2011, Wells and Levy, 2003).

Based on the hypothesis that the selection of chromosomal balanced (euploid) oocytes and embryos will lead to better ART outcomes, preimplantation genetic diagnosis for aneuploidy (PGD-A) was developed and incorporated into clinical practice. Although the earlier version of PGD-A which used Fluorescence in Situ hybridization (FISH) of cleavage stage embryos, was limited by its capacity to assess only a subset of chromosomes, current evidence on more recent technologies that allow testing of all 24 chromosomes, from four small prospective randomised controlled trials, mainly in young women with a good prognosis, suggests that PGD-A is associated with an improved ART success rate in a single ART cycle (Chen, et al., 2015, Dahdouh, et al., 2015).

However, there remains a paucity of high quality evidence to support the clinical effectiveness of PGD-A in older women, and whether the cumulative success rates over multiple ART cycles favours the incorporation of PGD-A into clinical practice (Lee, et al., 2014).

ART is a costly intervention with each stimulated cycle costing approximately 11,000

Australian dollars (AUD). Although Medicare reimburses almost two-thirds of the cost of an ART cycle, patients are required to pay the remaining costs as an out-of-pocket expense. However, patients with private health insurance are usually eligible to partially or fully claim charges incurred for admitted hospital procedures and for some non-PBS medications. In particular, same-day hospital accommodation, theatre charges and anaesthetists’ fees for oocyte retrieval procedures are usually covered for members of private health insurance schemes. PGD-A is an additional expense to the patient as it is

5 not covered by Medicare (Medicare Australia, 2015). Although it is likely that the cost of

PGD-A will decrease as technology matures (Munné and Cohen, 2017), currently PGD-

A adds approximately AUD 2,550 to the cost of a standard ART treatment cycle.

Despite the high cost and lack of evidence on its clinical effectiveness, PGD-A is increasingly used in the clinical setting in Australia. For example, of the 67,250 treatments cycles performed in Australia and New Zealand in 2015 where fertilisation occurred, or an embryo was thawed, 8.5% (compared to 2.4% in 2005) were subjected to

PGD-A (Fitzgerald O, et al., 2017)..There has been only one study that has assessed the cost-effectiveness of PGD-A (Collins, et al., 2017). This study, which was based on an economic model and inputs from the literature, reported that PGD-A is cost-effective in women aged over 37 years. However, the model outcome was based on a single-cycle analysis, and thus the impact of the cost and cumulative live-birth rate (CLBR) resulting from repeated ART cycles was not evaluated (Collins, et al., 2017).

An economic evaluation of PGD-A is important because it is a costly procedure that potentially has wide applicability. Furthermore, with increasing pressures on healthcare budgets, full economic evaluation methods are increasingly required and applied by decision makers to assess whether value for money exists and to determine the trade-offs between costs and benefits from emerging technologies (Devlin and Parkin, 2003,

Sculpher, et al., 2006).

Although current evidence has shown that PGD-A is associated with higher pregnancy and live-birth rates in single ART cycles among younger women (Chen, et al., 2015,

Dahdouh, et al., 2015, Rubio, et al., 2017), whether PGD-A improves the cumulative success rate and reduces the overall time and number of ART cycles to reach a live birth

6 in older women and whether it is cost-effective from an individual or healthcare perspectives is largely unknown.

1.2 Aim and contribution

The overarching aim of this doctoral research program was to contribute to the evidence on the clinical and cost-effectiveness of PGD-A for infertile women of advanced maternal age. The findings reported in this doctoral thesis will add to the contemporary evidence on the role of PGD-A in clinical practice and provide economic evidence about PGD-A to assist in making well-informed funding decisions.

The doctoral research program comprises four related studies that are presented in Figure

2 below.

1.2.1 Study One: The clinical effectiveness of PGD-A in all 24 chromosomes: systematic review

The aim of this study was to systematically review current evidence regarding the clinical effectiveness of the comprehensive PGD-A techniques specifically array comparative genomic hybridisation (aCGH), single nucleotide polymorphism arrays, and real time quantitative polymerase chain reaction. This review identified gaps in the evidence which provided the impetus for the analytical studies that followed.

The findings of the systematic review were published in Human Reproduction in 2014

(Lee, et al., 2014). The full manuscript can be found in Appendix 1. An update on studies on PGD-A since the published systematic review is included in this thesis in Chapter 6.

1.2.2. Study Two: ART cumulative live-birth rates following PGD-A or morphological assessment of embryos: a cohort analysis

The aim of this study was to evaluate the cumulative live-birth rates (CLBR) in infertile women aged 37 years or over who commenced their first ART cycle using PGD-A or morphological assessment of embryos alone. This study used the intention-to-treat and per-protocol analytical principles to evaluate multiple clinical outcomes of PGD-A. This is the first study to assess the clinical value of PGD-A in terms of CLBR over multiple cycles and average time and number of ART cycles needed to achieve a live birth.

This study was published in the Australian and New Zealand Journal of Obstetrics and

Gynaecology in 2017 (Lee, et al., 2017). The full manuscript can be found in Appendix

1.2.3. Study Three: Longitudinal analysis of PGD-A in women of advanced maternal age: a cost-effectiveness analysis

The aim of this study was to use patient-level data to assess the cost-effectiveness of

PGD-A compared to morphological assessment of embryos alone. This study extended the clinical findings of Study Two by applying a health technology assessment method to determine whether undertaking PGD-A over multiple ART cycles represented value for money compared to morphological assessment of embryos alone, from an individual and healthcare perspective.

1.2.4 Study Four: Which ART treatment strategy is the most clinically and cost- effective for women of advanced maternal age? A Markov model

The aim of this study was to compare the clinical and cost-effectiveness of PGD-A and three other commonly used ART strategies in women of advanced maternal age. This study used a Markov decision-analytic model to incorporate and synthesise the most 8 contemporary published data from the literature to assess the four ART strategies ―

PGD-A, social egg freezing, ART using donated oocytes, and standard ART in women aged 40 years following 6–12 months of infertility.

Figure 2: Relationship between studies carried out for this doctoral research program

1.3 Thesis outline

The thesis provides a contemporary assessment of the clinical and cost-effectiveness of

PGD-A in older women. The thesis is structured into the following chapters:

Chapter 1 is an introductory chapter providing a background, rationale and the overarching aim of this doctoral research program. An overview of the aims and contributions of four related studies conducted in this research program is provided. An outline of the thesis structure including a summary of Chapters 2–10 is given.

Chapter 2: Epidemiology of infertility provides an overview of the epidemiology of infertility fecundity and fertility. This is followed by the common definitions and estimates of infertility from a clinical, epidemiological and demographic perspective. The chapter concludes with a description of the common causes of infertility, with a focus on women of advanced maternal age.

Chapter 3: Overview of ART and intrauterine insemination (IUI) presents a review of the empirical evidence on commonly used medically assisted fertility treatments. This is followed by a comparison of the clinical effectiveness of ART with other alternative fertility treatments, focusing on women of advanced maternal age. The second part of the chapter discusses enhanced ART techniques and protocols including PGD-A. The chapter concludes with a discussion on the role of ART in women of advanced maternal age and the associated risk of congenital anomalies and multiple pregnancies following ART treatment.

Chapter 4: Measuring ART success discusses the alternative measures of treatment success using ART and the challenges associated with these measures. It emphasises that there is no single measure that can adequately capture treatment effectiveness and safety as well as assess the cost-effectiveness of ART. The chapter concludes with a discussion on the importance of taking a longitudinal perspective over multiple ART cycles when measuring ART success.

Chapter 5: Economic evaluation of ART begins with an introduction to the study of health economics. This is followed by an overview of the economic evaluation framework and how it is applied to inform the rational allocation of healthcare resources. Throughout the chapter, economic concepts that relate to ART are emphasised, to provide context for the economic studies conducted in this doctoral research program.

Chapter 6: The clinical effectiveness of PGD-A in all 24 chromosomes: systematic review provides a comprehensive review of the clinical studies regarding the effectiveness of PGD-A published up until November 2017. In total, 26 studies were identified and critically appraised based on the principles of the Preferred Reporting Items for Systematic Reviews and Meta–Analyses (PRISMA) statement. In addition, the methodological quality of the included studies was assessed using a modified version of the Downs and Black checklist.

Chapter 7: ART cumulative live-birth rates following PGD-A or morphological assessment of embryos: a cohort analysis is the first of three empirical studies to assess the clinical and cost-effectiveness of PGD-A relative to the morphological assessment of embryos alone. This is the first study to measure the CLBRs in women aged 37 years or over commencing their first ART cycle with either PGD-A or morphological assessment of embryos alone. The longitudinal analysis of outcomes based on observational data

12 provided an assessment of the clinical effectiveness of PGD-A over multiple cycles in a real-world setting.

Chapter 8: Longitudinal analysis of PGD-A in women of advanced maternal age: a cost- effectiveness analysis extends the analysis undertaken in Chapter 7 by assessing the cost- effectiveness of PGD-A. The study includes a ‘bottom-up’ costing approach to identify, measure and value direct resource consumption associated with ART treatment and relates the costs to the outcome (i.e. CLBR) to derive measures of cost-effectiveness from an individual and healthcare perspective.

Chapter 9: Which ART treatment strategy is the most clinically and cost-effective for women of advanced maternal age: A Markov model extends the cost-effectiveness analysis described in Chapter 8 by constructing a Markov model to incorporate two alternative treatment strategies – social egg freezing and the use of donated oocytes.

These four commonly used ART strategies are reflected in the model to represent the likely clinical pathways. A significant part of this chapter is the estimation of model parameters for the transition probabilities, the associated health outcomes, and costs.

Chapter 10: Discussion brings together the key findings from the four studies conducted for this doctoral research, concluding with suggested directions for further research.

Chapter 2

Epidemiology of Infertility

Chapter 2: Epidemiology of Infertility

2.1 Introduction

It is estimated that approximately 15% of women of reproductive age are affected by infertility, causing much personal suffering to an estimated of 72.4 million couples globally (Boivin, et al., 2007)

Despite its significant impact, there remains no clear consensus on the definition of

‘infertility’, nor a standard set of measures to estimate its incidence or prevalence (Gnoth, et al., 2005, Mascarenhas, et al., 2012a). This is largely due to the differing stakeholders’ perspectives across disciplines and institutions engaged with such a fundamental human function (Gurunath, et al., 2011).

The following section begins with the key definitions and measures of fecundity and fertility, then the common definitions and estimates of infertility from a clinical, epidemiological or demographic perspective. Section 2.6 concludes with a discussion of the common causes of infertility, with a focus on women of advanced maternal age.

2.2 Fecundity and fertility

The International Committee Monitoring Assisted Reproductive Technologies

(ICMART) defines ‘fecundity’ as the capacity to have a live birth based on the probability of pregnancy during a single menstrual cycle in a woman with adequate exposure to sperm and no contraception, culminating in a live birth (Zegers-Hochschild, et al., 2017).

According to Leridon and Slama (Leridon and Slama, 2008), the degree of fecundity depends on (a) fecundability (i.e. the monthly probability of pregnancy among sexually active non-contraceptive couples), (b) the rate of foetal mortality (i.e. the probability that

15 a pregnancy does not result in a live birth) and (c) the probability of being permanently sterile (i.e. unable to conceive).

Therefore, depending on the couple’s timing and frequency of sexual intercourse and biological parameters, a couple’s monthly per-cycle probability of conception can vary from zero to an estimated upper limit of 60% (Louis, 2011, Smarr, et al., 2017).

Fecundity is often used interchangeably with fertility although, classically, ‘fertility’ refers to the reproductive performance as measured by actual live births rather than capacity to reproduce (Habbema, et al., 2004, Schmidt, et al., 2012). Hence at a population level, ‘fertility rate’ refers to the average number of babies born per woman

(or per 1000 women) throughout reproductive life. According to the Australian Bureau of Statistics’ latest data, Australia’s total fertility rate in 2015 was 1.8 children per woman, which is well below the replacement total fertility rate of 2.1 (Australian Bureau of

Statistics, 2015).

2.2.1 Measure of fecundity

Measuring fecundity at an individual or population level is complex but is necessary to allow the counterfactual – infertility – to be diagnosed at an individual level and estimated at a population level. The denominator for fecundity is either a unit of time (e.g. month) or a menstrual cycle. Based on the chosen denominator, the probability of conception among sexually active, non-contraceptive couples is estimated. A particular challenge with measuring fecundity is to identify, reliably, the period of exposure to pregnancy

(Leridon, 2007). The following section discusses three common measures of fecundity at the population level.

a) Time to pregnancy

A couple’s fecundity can be measured using time to pregnancy (TTP), which quantifies the number of calendar months or menstrual cycles leading to a pregnancy (Keiding, et al., 2012). The underlying premise of this approach is that a shorter time or fewer cycles is indicative of higher fecundity.

The utility of the TTP approach is that it can be assessed prospectively (i.e. where couples are interviewed periodically, starting from when they first attempt to become pregnant), or retrospectively (i.e. where couples are interviewed retrospectively about the time taken to conceive) (Joffe, et al., 2005). Previous estimates of fecundity using the TTP approach in sexually active, non-contraceptive couples found that the cumulative probability of pregnancies at 12 months ranged from between 81% and 95%. This means that, on average, up to 95% of sexually active, non-contraceptive couples would be expected to conceive within 12 months (Gnoth, et al., 2003, Wang, et al., 2003).

The TTP approach has proven useful in descriptive epidemiology, particularly in identifying trends and the effect of environmental and occupational exposures on human reproduction (Joffe, 2000, Joffe, et al., 2005, Scheike and Keiding, 2006).

b) Cross-sectional design

The use of a cross-sectional study design offers an alternative approach for estimating fecundity in a population. The often-cited United States (US) National Survey of Family

Growth (NSFG) uses a cross-sectional design with a nationally representative sample of women aged 15–44 to provide national estimates of pregnancy, infertility, fecundity and information on family-related topics such as relationship status, sexual activity, contraception use and pregnancy (Lepkowski, et al., 2010). The latest NSFG data, for the period 2006–2010, showed that the fecundity rate was 51% among married women who

17 were not surgically sterile, or did not have any problems conceiving or carrying a pregnancy to live birth in the previous 36 months (Chandra, et al., 2013).

c) Current duration design

The current duration approach utilises a TTP-like distribution from a cross-sectional sample to determine the current duration of unprotected intercourse (i.e. the time elapsed from stopping contraception, or first attempt to become pregnant) among women who are currently at risk of pregnancy at the time of their interview. This duration is then used to estimate the proportion of couples who are not pregnant after a given number of months of unprotected intercourse (Keiding, et al., 2002). Unlike the TTP approach (described in

Section 2.1.1 a), which excludes couples who do not conceive, the current duration method includes all women at risk of pregnancy without any bias against couples who are involuntarily childless or have pregnancy intentions but did not achieve a live birth

(Slama, et al., 2012, Thoma, et al., 2013). For example, using the current duration approach, Slama and colleagues found that of 867 eligible women who were engaging in regular sexual intercourse with a male partner, were not using any method of contraception, and had not delivered in the previous three months, approximately 24%

(95%CI, 19–30%) of the women had not conceived by 12 months and another 11% of the women (95%CI, 8–14%) had not conceived by 24 months (Slama, et al., 2012).

2.2.2 Infertility

Estimating the true incidence and prevalence of infertility is challenging due to definitional differences, the instruments used, and the population being studied

(Gurunath, et al., 2011). The three most common definitions of ‘infertility’ from a clinical, epidemiological or demographic perspective are described below.

a) Clinical definition

The clinical definition of ‘infertility’ aims to provide clinical guidance on when patients warrant being investigated for infertility and is designed to be sensitive enough to identify couples who may benefit from fertility treatment (Steures, et al., 2004).

The clinical definition of ‘infertility’ is usually the failure to achieve conception after 12 months of being exposed to pregnancy. This period originates from the biological observation that 85–90% of couples of normal reproductive health and not using contraception, conceive within 12 months (Cramer, et al., 1979, Guttmacher, 1956). More recently, it has been suggested that the clinical definition of ‘infertility’ should be based on the length of time spent trying to conceive, adjusted for female age, so that women of advanced maternal age receive earlier investigation and treatment (Gurunath, et al.,

2011).

The ICMART defines infertility1 as ‘failure to achieve a clinical pregnancy after 12 months or more of regular, unprotected sexual intercourse or due to an impairment of a

1 The revised Glossary on Fertility and Fertility care 2017, which includes the definitions of ‘infertility’ and ‘fecundity’ was led by ICMART in partnership with representatives from several collaborative organisations, including the American Society for Reproductive Medicine (ASRM), European Society of Human Reproduction and Embryology (ESHRE), the American College of Obstetricians and Gynecologists (ACOG), the International Federation of Fertility Societies (IFFS), Red Latinoamericana de Reproducción Asistida (REDLARA), African Fertility Society (AFS), Groupe Inter-africain d'Etude, de Recherche et d'Application sur la Fertilité (GIERAF), Asia Pacific Initiative on Reproduction (ASPIRE) and Middle East Fertility Society (MEFS). 19 person's capacity to reproduce either as an individual or with his/her partner’ (Zegers-

Hochschild, et al., 2017).

Similarly, the National Institute for Health and Care Excellence (NICE) in the United

Kingdom (UK) defines infertility as ‘failure to conceive after one year of unprotected vaginal sexual intercourse in the absence of any known cause of infertility’ (NICE, 2013).

The American Society for Reproductive Medicine (ASRM) Committee revised the clinical definition of infertility in 2013 to incorporate age-specific thresholds. Currently, the ASRM defines infertility as ‘the failure to achieve a successful pregnancy after 12 months or more of appropriate, timed unprotected intercourse or therapeutic donor insemination or after six months for women over age 35 years’ (Practice Committee of the American Society for Reproductive Medicine, 2013).

b) Epidemiological definition

The epidemiological definition of ‘infertility’ aims to provide a standardised definition of infertility for surveillance studies and for reporting population indicators of reproductive health (World Health Organization, 2014).

The World Health Organization (WHO) defines infertility as the ‘inability of women of reproductive age (15–49) who are at risk of becoming pregnant (i.e. sexually active, not using contraception and not lactating) and report trying unsuccessfully to conceive for 24 months or more’ (World Health Organization, 2014). By using 24 months of unprotected intercourse without conceiving, it minimises the risk of false positives, which may overestimate the infertility rate (Dunson, et al., 2004, Te Velde, et al., 2000).

c) Demographic definition

The demographic definition of ‘infertility’ aims to provide a population-based estimate of reproductive outcomes. The population-based estimate of infertility is primarily used for comparison between countries and to track critical changes in the distribution and trends of infertility (Mascarenhas, et al., 2012a).

Demographers typically define infertility as the inability of a sexually active, non- contraception woman to achieve a live birth, without specifying a timeframe of exposure

(Wilson and Pressat, 1985). More recently, Mascarenhas and colleagues proposed a new demographic definition of infertility: ‘the absence of a live birth for couples that have been in a union for at least five years, during which neither partner used contraception, and where the female partner expresses a desire for a child’. The proposed five-year exposure timeframe seeks to further reduce false positives and provides a consistent measure of the magnitude, distribution and underlying trends of infertility at a population level for comparison between surveys and countries (Mascarenhas, et al., 2012a).

2.3 Global estimates of infertility

Global estimates of infertility are frequently based on two large-scale studies. Boivin and colleagues pooled data from 172,413 married women using 25 population-based surveys and estimated that the prevalence of clinical infertility ranged from 3.5% to 26.4% in developed countries and from 1.3% to 25.7% in developing countries. They concluded that the global estimate of clinical infertility was 9% (or 72.4 million women who are involuntarily childless), with slightly more than half seeking medical care for infertility

(Boivin, et al., 2007).

A more recent study supported by the WHO and the Bill and Melinda Gates Foundation, as part of the 2010 Global Burden of Disease Study, analysed 277 demographic and reproductive health surveys from 190 countries and territories during 1990–2010

(Mascarenhas, et al., 2012).

This study, which used live birth as the outcome and a five-year exposure period, estimated that 1.9% of women aged 20–44 (or 48.5 million couples) worldwide were infertile in 2010. However, it is estimated that 121 million women will be classified as infertile (i.e. a two-fold increase from the 48.5 million using the demographic definition) if the duration of infertility used in the demographic-based study is adjusted using WHO’s epidemiological definition of ‘infertility’, i.e. reducing the time period from five years to two years (World Health Organization, 2014).

2.4 Country-specific estimates

Few country-specific estimates of infertility are available in the literature. Using a cross- sectional population-based survey, a recent study in the UK, comprising 15,162 men

(6,293) and women (8,869) aged 16–74, reported infertility prevalence rates of 12.5% among women and 10.1% among men (Datta, et al., 2016). The estimates are similar to earlier national estimates of infertility in the UK (Bhattacharya, et al., 2009, Cabrera-

León, et al., 2015, Oakley, et al., 2008).

Data from the latest NSFG survey, for the period 2006–2010, reported that approximately

6.0% of married women residing in the US, aged between 15 and 44, were infertile according to the clinical definition of infertility (i.e. women who were not surgically sterile and had at least 12 consecutive months of unprotected sexual intercourse without becoming pregnant). Similar rates were found among cohabiting women in a similar age range (Chandra, et al., 2013).

In Canada, population-based data from a 2009–2010 health survey of 29,858 married and common-law couples with a female partner aged 18–49 revealed that the prevalence of current infertility (defined in the survey as couples who reported no pregnancy, did not use any form of birth control, reported having sexual intercourse during the previous 12 months and had tried at some point to become pregnant with their current partner) was

11.5% (Bushnik, et al., 2012).

Country-specific estimates on infertility in developing countries are limited. However, a small number of dated studies found that infertility affects more than 20% of people in

Gambia, Ethiopia and Nigeria, and that infertility in these countries is often secondary in nature (i.e. following a successful birth) (Adetoro and Ebomoyi, 1991, Ebomoyi and

Adetoro, 1990, Larsen, 2000).

In Australia, there is no single source for estimating the prevalence of infertility across the population. Most of the estimates are limited to few small-scale cohort studies

(Bruinsma, et al., 1998, Marino, et al., 2010, Venn, et al., 1993, Webb and Holman, 1992).

For example, in a cohort study of 974 women, who were born between January 1973 and

December 1975 at the Queen Elizabeth Hospital in Adelaide, South Australia, almost one in four survey respondents reported difficulty becoming pregnant (24.2%) and more than half of these women had sought medical assistance to achieve pregnancy (57.9%)

(Marino, et al., 2010).

The Australian Longitudinal Study on Women’s Health (ALSWH) provides the largest epidemiological examination of female infertility (Mishra, 2014). The ALSWH comprises cross-sectional panel surveys of randomly selected participants from the general population recruited in 1996, who are invited to complete a mailed survey every three to four years. The study found that among Australian women aged between 28 and

33 (n=9,145), approximately one in five (17.3%) reported a history of difficulty in conceiving for more than 12 months, and of these women, one-third accessed fertility treatment (Herbert, et al., 2009, Jensen, et al., 2008). The infertility rate reported in the

ALSWH (17.3%) is lower than that reported by Marino and colleagues (24.2%).

However, the differences may be due to survey design and the format of the survey questions (Marino, et al., 2010).

2.5 Temporal trends in infertility rates

There is conflicting evidence regarding whether the infertility rate is increasing. A number of studies indicate that infertility is increasing, primarily due to the increasing trend towards delayed childbearing, the negative impact of lifestyle factors such as obesity, and the increased prevalence of abnormal semen parameters. There is also the possibility that the perceived increase in infertility may be due to an improved awareness among patients and the clinicians, as well as better surveillance methods for estimating the incidence and prevalence of infertility in the general population (Ledger, 2009,

Levine, et al., 2017, Rooney and Domar, 2014, Sharma, et al., 2016). Conversely, it has been suggested that the declining fertility rates are likely due to fewer people trying to have children as the population growth has slowed (Mascarenhas, et al., 2012). However, some have argued that the inherent bias in current estimation methods and definitional differences in the population-based studies may have underestimated the ‘true’ prevalence of the global infertility problem (Boivin, et al., 2007, Gurunath, et al., 2011).

A recent systematic review and meta-regression analysis of 185 studies on 244 unique sperm count samples from 42,935 men reported a significant decline (52.4%) in the sperm counts between 1973 and 2011 in Western countries (North America, Europe, Australia and New Zealand), with no evidence of a ‘leveling off’ in recent years. The findings

24 indicate that a high proportion of men from Western countries have sperm counts below the threshold value for infertiliy and reproduction (Levine, et al., 2017). This result is consistent with an earlier observational study that reported a decrease in sperm counts by almost 50% between 1938 and 1990 (Carson III, 2013).

Conversely, a recent review of 277 population-based demographic and health surveys concluded that the prevalence of infertility has not changed significantly over the last two decades. In 2010, the proportion of women aged 20–44 reporting difficulty in conceiving was 1.9% (or 3.8 million) compared to 2.0% (or 4.2 million) in 1990 (Mascarenhas, et al., 2012).

Chandra and colleagues reported a decline in the infertility rate in the US based on the five-yearly NSFG surveys (Chandra, et al., 2013). Between 1965 and 2010, the authors reported that the 12-month infertility rate among married women aged 15–44 declined from 11.2% to 6.0%. Furthermore, primary infertility estimates for most regions in sub-

Saharan Africa showed that infertility rates had also decreased, from 2.7% in 1990 to

1.9% in 2010 (Mascarenhas, et al., 2012).

The estimation of the incidence and prevalence of infertility is complex. However, consensus on the definition of ‘infertility’ and method for measuring the incidence and prevalence of infertility is the first step in improving the estimation of the national and global rates of infertility, in order to better inform public health policies and clinical decision making.

2.6 Causes of infertility

Earlier epidemiological studies have found that among couples seeking treatment for infertility, at least one-third were infertile due to male factor, 14–22% due to tubal factor,

10–27% due to ovulatory dysfunction, and a smaller proportion due to unexplained infertility and endometriosis (Hull, et al., 1985, Templeton, et al., 1990). These estimates are consistent with a population-based cohort study of 4,522 women aged 31–50 in

Scotland, which found common self-reported causes of infertility included ovulation problems, sperm quality problems and unexplained infertility (Table 1) (Bhattacharya, et al., 2009).

Table 1: Self-reported causes of infertility among women in Scotland, 2009 Diagnosis Primary infertile Secondary infertile women

women % %

Ovulation problems 32.3 23.2

Sperm quality problems 29.3 23.8

Unexplained infertility 29.3 29.8

Blocked fallopian tubes 12.0 13.9

Endometriosis 10.7 10.0

Others 13.8 21.2

Source: (Bhattacharya, et al., 2009)

However, it is clear that other factors, such as physiological and negative lifestyle factors, also contribute to the incidence of infertility, including body mass index, negative lifestyle factors and increasing maternal age (Rittenberg, et al., 2011).

a) Body mass index (BMI)

A higher BMI (typically ⩾25 kg/m2) is of a particular concern for women who wish to conceive (Mantakas and Farrell, 2010, Sharma, et al., 2013). A systematic review and meta-analysis of 33 studies on the effect of being overweight on assisted reproductive technology (ART) outcomes reported that, compared to women with normal BMI (18.5

≤ BMI <25 kg/m2), women with BMI ⩾25 kg/m2 had a 10% reduced chance of a clinical pregnancy rate per cycle (RR 0.90, 95%CI 0.85–0.94, 25 studies). Furthermore, overweight and obese women (with BMI ⩾25 kg/m2) had a 30% increased risk of miscarriage compared to women with normal BMI (relative risk (RR) 1.31, 95%CI 1.18–

1.45, 22 studies) (Rittenberg, et al., 2011).

b) Negative lifestyle factors

Lifestyle behaviours, such as smoking and alcohol consumption, have been shown to be detrimental to a couple’s fertility (Anderson, et al., 2010, Eggert, et al., 2004, Gormack, et al., 2015). A recent systematic review and meta-analysis of 21 studies found that patients who smoked at the time of fertility treatment have a 50% reduced chance of a live-birth rate per cycle (odds ratio (OR) 0.54, 95% Confidence Interval (CI) 0.30–0.99,

4 studies), and a 2.6-fold higher risk of miscarriage compared to patients who did not smoke (OR 2.65, 95%CI 1.33–5.30, 7 studies) (Rittenberg, et al., 2011).

Regular and high alcohol consumption in women also increases the risk of a longer time to natural conception (Axmon, et al., 2006, Juhl, et al., 2003). A recent prospective cohort study, which followed up 6210 women attempting to conceive (median age of 28), found

27 that the highest quartile consumption of alcohol (≥14 servings a week) was associated with an 18% decrease in fecundability compared with women who did not consume alcohol (OR 0.76, 95%CI 0.51–1.11) (Mikkelsen, et al., 2016).

c) Female age-related infertility

Pregnancies in women older than age 35 are increasingly common in developed countries

(Balasch and Gratacós, 2012, Matthews and Hamilton, 2009). For example, in the US, the latest report from the National Vital Statistics, for 2015, reveals that the first-birth rate for women aged 35–39 has increased by almost ten-fold, from 1.7 in 1973 to 10.9 in 2006.

Although the first-birth rate for women aged 35–39 decreased slightly to 10.4 in 2010, it rose again to 11.0 in 2012. A similar upward trend has also been reported for women aged

40–44, where the first-birth rate has increased by more than four-fold from 0.5% in 1985 to 2.3% in 2012 (Matthews and Hamilton, 2009).

Australia is not exempt from the increasing trend of delayed childbearing. Between 1991 and 2011, the proportion of first-time mothers aged 35–39 almost trebled from 4.5% to

11.7%. Over the same period, first births for women aged 40–44 rose from 0.6% to 2.6%

(Lancaster, et al., 1994, Li, et al., 2013).

The reasons for the delayed childbearing are complex. However, it has been shown that factors such as higher education attainment, career opportunity and increased availability of fertility treatment influences the decision on when to have children (Mills, et al., 2011).

It is well established that fertility declines with increasing maternal age, with advanced female age being the most important predictor of fertility and reproductive outcomes

(ESHRE Capri Workshop Group, 2005, van Loendersloot, et al., 2010). However, there is no universal definition of ‘advanced maternal age’, in part because the effects of increasing age occur as a continuum rather than a threshold effect, and declining fertility

28 varies between individuals (Lass, et al., 1998, Te Velde and Pearson, 2002). It is generally accepted that the decline of fecundity and fertility rates is significant in the early thirties, with a more rapid decline after the age of 35 (ESHRE Capri Workshop Group, 2005,

Practice Committee of the American Society for Reproductive Medicine, 2008, Sauer,

2015, Schwartz and Mayaux, 1982).

The decline in fertility with increasing female age corresponds to the natural biological attrition in ovarian reserve. It is estimated that the maximum complement of oocytes is approximately 6–7 million at 20 weeks of gestation, decreasing progressively to approximately 1–2 million oocytes at birth, and 25,000 by age 37. At the time of menopause, the supply is reduced to about 1,000 follicles (Baker, 1963, Te Velde and

Pearson, 2002).

In addition to the declining number of oocytes with female age, the presence of chromosomal abnormalities (i.e. aneuploidy) in the oocytes also increases with female age (Hassold, et al., 2007, Taylor, et al., 2014). This is illustrated in a recent study by

Franasiak and colleagues who performed 15,169 trophectoderm biopsies and found that the rate of aneuploidy in embryos increased steadily from age 31 through to age 43, plateauing at approximately 85% (Figure 3).

Figure 3: Rate of aneuploidy relative to age of female partner Source: (Franasiak, et al., 2014)

The consequence of delaying childbearing results in a reduced risk of natural conception and conception using ARTs. There is also an increased risk of miscarriage and babies with chromosomal abnormalities such as Down syndrome (trisomy 21) (Fragouli, et al.,

2011, Wells and Levy, 2003). It has been estimated that for women aged 40 or older, approximately 50–80% of all first trimester pregnancy losses are due to chromosomal abnormalities in the developing foetus (Hollier, et al., 2000, Nasseri, et al., 1999).

To summarise, the most important predictor of fertility and reproductive outcomes is the woman’s age. Therefore, the choice of fertility treatment for older women nearing the end of their reproductive years requires careful consideration due to the limited window of opportunity to become pregnant. Over the years, there have been significant developments in reproductive technologies and protocols to improve the success rate in older women. The next section‒Section 2.7 provides an overview of commonly used fertility treatments, with a focus on women of advanced maternal age.

2.7 Overview of commonly used non ART treatments

Although ART is usually the most successful treatment option available for infertile couples, it is not the only fertility treatment. Other treatment modalities, ranging from expectant management to surgery, have played key roles in improving many infertile couples’ chance of success. These treatments include testing and monitoring of ovulation and semen parameters, medical support of natural conception, and medications to stimulate ovulation before retrieval of oocytes (Marino, et al., 2010).

Selection of the appropriate treatment for infertility is a challenging process. A diagnostic evaluation for infertility is indicated for couples who fail to achieve a successful pregnancy after 12 months or more of regular unprotected intercourse, although earlier assessment is often recommended, particularly for women aged over 35 (Practice

Committee of the American Society for Reproductive Medicine, 2012).

The choice of treatment is dependent on several factors such as maternal age, cause and duration of infertility, and medical history (Smith, et al., 2003). A number of clinical practice guidelines, such as the NICE clinical guidelines on assessment and treatment for people with fertility problems (NICE, 2013) and the ASRM Guidelines for Practice, are available to guide the treatment process (Practice Committee of the American Society for

Reproductive Medicine, 2012). However, the availability of health insurance, government subsidies and the patient’s preference also play important roles in the decision-making process (Chambers, et al., 2013, Hamilton and McManus, 2012).

Typically, infertility treatments are performed in a ‘stepwise’ fashion, progressing from more minimal therapies such as ovulation induction using oral agents to more invasive and complex treatments such as in vitro fertilisation (IVF) and intracytoplasmic sperm injection (ICSI) (Armstrong and Akande, 2013).

2.7.1 Types of fertility treatments used in treating common causes of infertility

The following section discusses commonly used non-ART treatments in clinical practice.

In this thesis, unless otherwise specified, non-ART treatments include expectant management, ovulation induction, ovarian stimulation, intrauterine insemination and surgery.

For couples with a better prognosis, such as unexplained infertility, mild male-factor infertility or minimal-to-mild endometriosis, treatment modalities, including expectant management (EM), intrauterine insemination (IUI) and controlled ovarian hyperstimulation (COH, or also known as ovarian stimulation [OS], where the general aim is to induce development of multiple follicle growth in women with a normal ovulatory menstrual cycle), are commonly used first-line treatments to achieve the desired outcomes (Macklon, et al., 2006, Meyer, et al., 2009, van den Boogaard, et al., 2013).

a) Expectant management/timed intercourse

Natural cycle with expectant management (NC+EM) or timed intercourse (NC+TI) is usually defined as watchful waiting without any active clinical or therapeutic intervention. In this approach, the woman needs to be aware of when she will ovulate and the best time for unprotected intercourse to increase her chances of pregnancy. Other factors influencing the success rate in the expectant approach include the woman’s age, duration of infertility and pregnancy history in the same relationship. Natural cycle with

EM or TI potentially eliminates the need for unnecessary early treatment, removes the

32 risk of associated complications such as multiple pregnancies and has lower treatment cost (Custers, et al., 2011, Sadeghi, 2015).

Two observational studies reported a high probability of spontaneous pregnancy in women with unexplained infertility and a good prognosis for spontaneous pregnancy

(Brandes, et al., 2010, Steures, et al., 2006). Steures and colleagues randomised 253 couples with unexplained infertility and an intermediate prognosis (i.e. 30–40% chance of a spontaneous ongoing pregnancy within the next 12 months based on the Hunault predictive model2) (Hunault, et al., 2004) to six months of COH with IUI (COH +IUI) or

NC+EM. The study found no evidence of a difference in the ongoing pregnancy rate per couple (23% with COH +IUI vs 27% with NC+EM, relative risk (RR) 0.85, 95%CI 0.63–

1.1) or live-birth rate per couple (20.4% with COH +IUI vs 23.8% with NC+EM, RR

0.86, 95%CI 0.54–1.4) between the two treatments (Steures, et al., 2006).

Brandes and colleagues followed up 437 couples with unexplained infertility (mean age

31.8) over five years in a secondary or tertiary care setting. The study found that among women with unexplained infertility and a good prognosis (i.e. at least 30% chance of a spontaneous ongoing pregnancy within the next six to twelve months based on the

Hunault predictive model), almost two in three ongoing pregnancies were spontaneous

(including NC+EM) (64.2%)) compared to 9.3% with COH +IUI, and 10.3% with ART

(Brandes, et al., 2011). This finding is consistent with another prospective observational study which reported a cumulative live-birth rate of 33.3% (95%CI 27.6–39.0) at 36 months following NC+EM in a cohort of 562 untreated couples with unexplained infertility registered at a secondary care setting (Collins, et al., 1995).

2 Hunault and colleagues used pooled-patient data to develop predictive models to estimate the probability of spontaneous conception based on patient and treatment characteristics including female age and duration of infertility. 33

b) Intrauterine insemination

The first paper on Intrauterine insemination (IUI) was published in the International

Journal of Fertility in 1962 (Cohen, 1962). IUI has since evolved, with progress in sperm preparation techniques, cycle monitoring and induced ovulation with human chorionic gonadotrophin (hCG) in COH cycles (Johnson, et al., 2013).

IUI is a relatively non-invasive technique commonly performed for couples with unexplained infertility, mild male infertility and minimal-to-mild endometriosis, or as an empirical treatment for a broad range of fertility indications (ESHRE Capri Workshop

Group, 2009). The rationale behind IUI is to bypass the cervical mucus barrier and increase the density of male gametes at the site of fertilisation in the fallopian tube in a natural or stimulated cycle (Puri and Aggarwal, 2015). The aim of combining IUI with

COH is to increase the number of available oocytes for fertilisation to maximise the chance of conception (Veltman‐Verhulst, et al., 2016).

The European IVF Monitoring (EIM) program reported that in 2012, 175,028 IUI cycles using the husband/partner’s sperm were performed in 24 countries, resulting in an overall mean delivery per cycle of 8.5%, with 9.0% twin deliveries and 0.4% triplet deliveries

(Calhaz-Jorge, et al., 2016).

However, the application of IUI to a broad range of indications, such as unexplained infertility, has resulted in an ongoing debate as evidence on the effectiveness compared to other treatments, such as ART, has yielded mixed results (Bensdorp, et al., 2015,

Bhattacharya, et al., 2008, ESHRE Capri Workshop Group, 2009, Reindollar, et al., 2010,

Steures, et al., 2006, Tjon-Kon-Fat, et al., 2015).

In the UK, NICE explicitly recommends against the use of IUI with or without COH because of the associated risk of multiple gestations, ovarian hyperstimulation syndrome

(OHSS) and a lack of robust evidence on the effectiveness of IUI in improving the live- birth rate. For all couples with unexplained infertility, low sperm count or poor-quality sperm and mild endometriosis, NICE recommends ART as a first-line treatment after extended NC+EM (NICE, 2013).

Despite NICE’s recommendation, a recent online survey showed that the majority of UK fertility clinics continue to offer IUI to couples with unexplained infertility (Nandi, et al.,

2015). The debate about the role of IUI in clinical practice was also recently reinforced with emerging evidence from publications that suggests ART with single embryo transfer and ART in a modified natural cycle were non-inferior and more costly than COH +IUI for couples with unexplained infertility or mild male infertility, or couples where the woman had an unfavourable prognosis (defined as less than 30% chance of natural conception within the next 12 months based on the Hunault model or failure to conceive within at least three years of unprotected intercourse) (Bensdorp, et al., 2015, Tjon-Kon-

Fat, et al., 2015).

i) IUI and the success rate in male factor infertility

Male infertility is a common condition among infertile couples and has been estimated to be directly responsible for 20% of infertile couples and a contributor in another 30% of infertile couples (Carson III, 2013). In the latest ANZARD report, which collects data on all ART treatments cycles and outcomes undertaken in Australia and New Zealand,

15.1% of all initiated fresh autologous and recipient cycles in 2015 reported male factor infertility factors as the only cause of infertility (Fitzgerald, et al., 2017). IUI is commonly used as the first-line treatment in patients with male factor infertility, although evidence on the clinical and cost-effectiveness of IUI (with or without COH) for male factor

35 infertility is unclear (Chambers, et al., 2010, Cissen, et al., 2016, Goverde, et al., 2000,

Robinson, et al., 1995, Tournaye, 2006, Wordsworth, et al., 2011).

A recent systematic review and meta-analysis of randomised controlled trials (RCTs) that compared natural cycle with IUI (NC+IUI) with NC+TI for male infertility found no evidence of a difference in the cumulative pregnancy rate per couple (odds ratio (OR)

4.57, 9.5%CI 0.21–102, 2 RCTs, n=62 couples). Neither RCT reported on the live-birth or multiple pregnancy rates (Cissen, et al., 2016).

Three RCTs compared COH +IUI with COH + timed intercourse (COH +TI) for male infertility, and the meta-analysis found no evidence of a difference in the pregnancy rate per couple (OR 1.51, 95%CI 0.74–3.07, 3 RCTs, n=202 couples), live-birth rate per couple (OR 0.89, 95%CI 0.30–2.59, 1 RCT, n=81 couples) or multiple pregnancy rate per couple (OR 3.15, 95%CI 0.12–79.69, 1 RCT, n=81 couples) (Cissen, et al., 2016).

There were five RCTs that compared NC+IUI with COH +IUI. There was no evidence of a difference in the pregnancy rate per couple (OR 1.68, 95%CI 1.00–2.82, 4 RCTs, n=

399 couples) and live-birth rate per couple (OR 1.34; 95%CI 0.77–2.33, 3 RCTs, n=346 couples) between the two treatments. None of the RCTs reported on multiple pregnancy rates (Cissen, et al., 2016).

A single RCT compared COH+IUI with NC+TI in couples with oligo and/or astheno and/or teratozoospermia (OAT) and immunologic infertility. The trial found no evidence of a difference in the pregnancy rate per couple (OR 3.14, 95%CI 0.12–81.35, n=44 couples) or the live-birth rate per couple (OR 3.14, 95%CI 0.12–81.35) between the two treatments. The trial did not report on the multiple pregnancy rate (Francavilla, et al.,

2009)

ii) IUI and the success rate in couples with unexplained infertility

In Australia and New Zealand, one in four initiated fresh autologous and recipient cycles in 2015 were in couples with unexplained infertility (25.9%) (Fitzgerald, et al., 2017).

Couples were classified as having unexplained infertility when they tried to conceive for at least one year and the fertility work-up showed patent fallopian tubes, an ovulatory menstrual cycle and a normal semen analysis.

IUI can be used with or without COH as a first-line treatment for couples with unexplained infertility. The two commonly used drugs for COH include clomiphene citrate, which is taken orally, and gonadotropins, administered by subcutaneous injection

(Puri and Aggarwal, 2015).

There is conflicting evidence on whether IUI with or without COH improves pregnancy and live-birth rates for couples with unexplained infertility. While an earlier review showed COH+IUI to be superior to NC+TI in couples with unexplained infertility

(Hughes, 1997), a recent systematic review and meta-analysis of 14 RCTs comparing IUI with EM/TI (both with or without COH) concluded that there was insufficient evidence to establish a difference in the live-birth and multiple pregnancy rates in most of the comparisons of the two treatments in couples with unexplained infertility (Veltman‐

Verhulst, et al., 2016).

A single RCT compared NC+IUI with NC+EM in women with unexplained infertility and found no evidence of a difference in the pregnancy rate per couple (OR 1.53, 95%CI

0.88–2.64, n=334 couples), live-birth rate per couple (OR 1.60, 95%CI 0.92–2.78, n=334 couples) or multiple pregnancy rate per couple (OR 0.50, 95%CI 0.04–5.53, n=334 couples) between the two treatments (Bhattacharya, et al., 2008).

There were seven RCTs comparing COH+IUI with COH+TI, and meta-analysis showed a higher pregnancy rate per couple using COH+IUI (OR 1.69, 95%CI 1.14–2.53; 6 RCTs, n=517 couples), but no evidence of a difference in the live-birth rate per couple (OR 1.59,

95%CI 0.88–2.88, 2 RCTs, n=208 couples) or multiple pregnancy rate per couple (OR

1.46, 95%CI 0.55–3.87, 4 RCTs, n=316 couples) between the two treatments (Veltman‐

Verhulst, et al., 2016).

There were four RCTs comparing NC+IUI with COH +IUI. There was no evidence of a difference in the pregnancy rate for the two treatments (OR 0.16, 95%CI 0.01–1.77, 1

RCT, n=26 couples), but the live-birth rate per couple was higher for women who were treated with COH + IUI compared to those who had NC+IUI (OR 0.48, 95%CI 0.29–

0.82, 4 RCTs, n=396 couples). The meta-analysis found no evidence of a difference in the multiple pregnancy rate per couple (OR 0.33, 95%CI 0.01–8.70, 2 RCTs, n=65 couples) between the two treatments (Veltman‐Verhulst, et al., 2016).

Two RCTs compared COH+IUI with NC+TI /EM, but only one RCT reported on the live-birth rate (Steures, et al., 2006). In that study, the authors reported no evidence of a difference in the live-birth rate per couple between COH+IUI with NC+EM (20.4% with

COH+IUI vs 23.8% with NC+EM, RR 0.86, 95%CI 0.54–1.4). Custers and colleagues followed up the remaining couples in the RCT conducted by Steures and colleagues for up to three years and reported no evidence of a difference in the ongoing pregnancy rate per couple between the two treatments (73% with COH+IUI vs 72% with NC+EM, RR

0.99, 95%CI 0.85–1.1) (Custers, et al., 2011, Steures, et al., 2006). The meta-analysis of the two RCTs found no evidence of a difference in the pregnancy rate (OR 1.0, 95%CI

0.59–1.67, 2 RCTs, n=304 couples) and multiple pregnancy rate per couple (OR 2.00,

95%CI 0.18–22.34, 2 RCTs, n=304 couples) between the two treatments (Veltman‐

Verhulst, et al., 2016). 38

There was a single RCT comparing NC+IUI with COH+TI. This trial reported that

NC+IUI improved the pregnancy rate per couple (OR 1.77, 95%CI 1.01–3.08, n=342 couples) and live-birth rate per couple (OR 1.95, 95%CI 1.10–3.44, n=342 couples) when compared to COH+TI. There was no evidence of a difference in the multiple pregnancy rate per couple (OR 1.05, 95%CI 0.07–16.90, n=342 couples) (Bhattacharya, et al., 2008).

iii) IUI and the success rate in couples with minimal-to-mild endometriosis

‘Endometriosis’ is defined as an inflammatory disease process, characterised by lesions of endometrial-like tissue outside the uterus that are associated with pelvic pain and/or infertility. Some suffers have no symptoms at all (Kennedy, et al., 2005). It is estimated that more than 176 million women worldwide are affected by endometriosis (Adamson, et al., 2010). Endometriosis can be classified into four stages of severity using the revised

American Society for Reproductive Medicine (r-ASRM) classification system: minimal

(stage I), mild (stage II), moderate (stage III) or severe (stage IV). Other classification systems for endometriosis include the Enzian classification system, which was developed to supplement the r-ASRM score for the description of deeply infiltrating endometriosis and the endometriosis fertility index (EFI) for predicting fertility outcomes following surgical staging endometriosis (Johnson, et al., 2017).

A single RCT randomised 103 infertile women with minimal-to-mild endometriosis to

COH+IUI or NC+EM and found that the cumulative live-birth rate was five-fold higher following COH+IUI (OR 5.6, 95%CI 1.18–17.4). The trial reported that three couples had multiple pregnancies following COH+IUI. The multiple pregnancy rate following

NC+EM was not provided (Tummon, et al., 1997).

iv) IUI and the success rate in couples with cervical hostility

Cervical hostility and the testing of such by the post-coital test is controversial (Practice

Committee of the American Society for Reproductive Medicine, 2015). However, a previous systematic review and meta-analysis of evidence for isolated cervical factor, defined as a repeated negative post-coital test despite a normal semen quality and adequate timing, located five trials (three were randomised and two were non-RCTs) that compared IUI with TI (both with or without COH) in women with cervical hostility.

However, data from these trials was not pooled for meta-analysis due to poor methodological quality and a high level of heterogeneity across the studies. Nonetheless, the review found IUI (with or without COH) not to be effective in terms of reproductive outcomes for women with cervical hostility (Helmerhorst, et al., 2006).

Steures and colleagues randomised 264 couples with an isolated cervical factor and a prognosis of more than 30% chance of a spontaneous ongoing pregnancy within the next

12 months based on the Hunault predictive model to six months of either IUI (with or without COH) or NC+EM. In the study, couples in the IUI group undertook three cycles of IUI without COH, and if they failed to achieve conception, subsequent IUI cycles were performed with COH. The trial found no evidence of a difference in the ongoing pregnancy rate per couple between IUI (with or without COH) and NC+EM (43% with

IUI with or without COH vs 27% with NC+EM) (RR 1.6, 95%CI 0.91–2.8). There was one twin pregnancy with IUI (with or without COH) and none with NC+EM (Steures, et al., 2007).

c) Ovulation induction

Ovulation induction (OI) has been one of most widely used treatments for infertility for more than four decades (Franks, et al., 1985, Holzer, et al., 2006, Kistner, 1967). The

40 general aim of OI is to stimulate and select a single follicle that will be able to reach the pre-ovulatory size and rupture. This treatment is commonly used in women with anovulatory infertility (i.e. when ovaries fail to release an oocyte during the menstrual cycle), particularly in women with polycystic ovary syndrome (PCOS, which is characterised by chronic anovulation and hyperandrogenism) (The Thessaloniki

ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group, 2008). National registries for OI do not exist, but in some countries such as Finland, it is estimated that

2.5% of the children in the late 1990s were born after OI and 2.3% of Danish children were born after OI in 2002 (Nyboe Andersen and Erb, 2006).

i) Clomiphene citrate

Clomiphene citrate (CC), a selective estrogen receptor modulator with both estrogenic and antiestrogenic properties, has been used for OI for more than four decades. It was first introduced as an agent to treat anovulatory infertility (Greenblatt, 1961). CC is the initial treatment of choice for most anovulatory or oligo-ovulation infertility (i.e. where there are irregular ovulatory cycles) (Ethics Committee of the American Society for

Reproductive Medicine, 2013). CC is effective only in the presence of normal circulating estrogen levels, with the potential for estrogen negative feedback on gonadotrophin production. The recommended CC treatment typically begins with a single daily dose of a 50mg tablet increasing to 150mg/day, with a maximum of 250mg (Dickey, et al., 1997,

Homburg, 2005). Treatment is usually continued for between six and nine cycles, although in the UK, NICE recommends up to 12 cycles. In successful treatment, one or more dominant follicles will develop and mature. However, it has been found that clomiphene resistance (failure to ovulate after taking a 150mg dose of CC) occurs in approximately 15–40% of women with PCOS (Kousta, et al., 1997, Pritts, 2002).

Clomiphene citrate and the success rate in women with anovulatory disorder

A recent systematic and meta-analysis of three crossover RCTs comparing CC with a placebo or no treatment in women with anovulatory infertility showed that CC improved the clinical pregnancy rate per couple (OR 5.91, 95%CI 1.77–19.68, 3 RCTs, n=133 women). None of the studies reported on live-birth or multiple pregnancy rates (Brown and Farquhar, 2016).

Two RCTs have compared CC with gonadotropins (follicle-stimulating hormone, FSH) in therapy-naive women with anovulatory infertility due to PCOS (Homburg, et al., 2012,

López, et al., 2004). A meta-analysis of these two RCTs showed a significantly higher clinical pregnancy rate per couple (OR 0.61, 95%CI 0.4–0.93, 2 RCTs, n=378 women) and ongoing pregnancy/live-birth rate per couple (OR 0.64, 95%CI 0.41–0.98, 2 RCTs, n=378 women) in favour of gonadotropins (FSH). There was no evidence of a difference in the multiple pregnancy rate (OR 0.26, 95%CI 0.06–1.06, 3 RCTs, n=696 women)

(Brown and Farquhar, 2016).

Clomiphene citrate and the success rate in women with unexplained infertility

The aim of using CC COH (with or without IUI) for women with unexplained infertility is to increase the number of ovulation oocytes and/or correct subclinical ovulatory dysfunction and a ‘luteal phase defect’ (Dickey and Holtkamp, 1996). The use of CC in women with unexplained infertility is usually limited to three cycles but may continue for up to six cycles (Dickey, 2015). Compared to COH with gonadotropin, CC offers the advantage of a lower incidence of multiple pregnancies, OHSS and cost, and a lesser need for cycle monitoring (Dickey, 2009).

However, the clinical benefit of CC for women with unexplained infertility is unclear.

Hughes and colleagues performed a systematic review and meta-analysis of seven RCTs

42 to assess the benefits of using CC with or without IUI to improve pregnancy outcomes for women with unexplained infertility (Hughes, et al., 2010). Four RCTs compared CC

(with or without IUI) to a placebo or no treatment (including NC+EM) in women with unexplained infertility, and meta-analysis found no evidence of a difference in the clinical pregnancy rate per woman between the two treatments when CC was used with IUI

(CC+IUI) (OR 2.40, 95%CI 0.70–8.19, 2 RCTs, n=76 women), CC alone (OR 1.03,

95%CI 0.64–1.66, 2 RCTs, n=458 women) and CC + hCG (OR 1.66, 95%CI 0.58–4.80,

2 RCTs, n=105 women) (Hughes, et al., 2010).

Only one RCT reported on the live-birth rate. Bhattacharya and colleagues randomised women with unexplained infertility to either using six cycles of CC COH or six months of NC + EM. The trial found no evidence of a difference in the live-birth rate per woman

(OR 0.79, 95%CI 0.45–1.38, n=385 women) or the multiple pregnancy rate per ongoing pregnancy (OR 1.01, 95%CI 0.14–7.19, n=385 women) between the two treatments

(Bhattacharya, et al., 2008).

The authors of the Cochrane review on CC concluded that there is no evidence to suggest that CC improves reproductive outcomes for women with unexplained infertility

(Hughes, et al., 2010).

Wordsworth and colleagues (Wordsworth, et al., 2011) conducted a cost-effectiveness analysis based on the results of the trial that compared CC COH + TI/EM with NC+EM in women with unexplained infertility (Bhattacharya, et al., 2008). This economic study found that CC COH + TI/EM as a first-line treatment for women with unexplained infertility was not cost-effective compared to NC+EM (cost per live birth for NC+EM was £72, 95%CI £0–£206 compared to £2611, 95%CI £1870–£4166 with CC COH +

TI/EM). The results suggest that NC+EM plays an important role when a couple’s financial resources are limited (Wordsworth, et al., 2011).

ii) Letrozole

Letrozole is a third-generation aromatase inhibitor that was originally used for the treatment of breast cancer. Letrozole inhibits the aromatase enzyme by competitively binding to the heme of the cytochrome P450 subunit of the enzyme, resulting in a blockade of androgen conversion into estrogen and a subsequent increase in intraovarian androgens (Haas and Casper, 2017).

The beneficial use of letrozole for women with anovulatory infertility due to PCOS has been reported in a number of systematic reviews. A systematic review and pair-wise meta-analysis of RCTs comparing letrozole and CC in all PCOS women (therapy-naive,

CC-resistant, CC status not reported and mixed PCOS population) found a significantly higher live-birth rate (OR 1.78, 95%CI 1.37–2.32, 5 RCTs, n=1275 women), pregnancy rate (OR 1.56, 95%CI 1.22–2.00, 13 RCTs, n=2474 women) and ovulation rate (OR 2.01,

95%CI 1.47–2.76, 9 RCTs, n=1752 women) with letrozole (PCOS Australian Alliance).

Similar favourable outcomes are also reported in a recently published systematic review and network meta-analysis of RCTs comparing the effectiveness of letrozole and CC. In this network meta-analysis, letrozole was shown to be more effective than CC in terms of the pregnancy rate (OR 1.58, 95%CI 1.25–2.00), live-birth rate (OR 1.67, 95%CI 1.11–

2.49) and ovulation rate (OR 1.99, 95%CI 1.38–2.87) for all WHO Group 2 (including

PCOS) anovulatory women and therapy-naive WHO Group 2 (including PCOS) anovulatory women (Wang, et al., 2017).

iii) Gonadotrophins

Ovarian induction with gonadotropins began in the 1960s and is commonly used as a second-line therapy, particularly for women with anovulatory infertility with PCOS who are resistant to CC after a maximum dose or CC failure (failure to conceive despite ovulation in six or more cycles of treatment) (Guzick, 2007, Practice Committee of the

American Society for Reproductive Medicine, 2008). Gonadotropins are associated with higher costs and an increased risk of OHSS and multiple gestations (Kafy and Tulandi,

2007).

Gonadotrophins and the success rate in women with anovulatory infertility

The clinical benefit of gonadotropins over CC for first-line OI therapy in anovulatory therapy-naive women with PCOS has been examined in two RCTs (Homburg, et al.,

2012, López, et al., 2004). Homburg and colleagues randomised 302 therapy-naive women with anovulatory infertility associated with PCOS to receive three cycles of either gonadotropin (low-dose FSH) or CC (Homburg, et al., 2012). The results based on the intention-to-treat analysis demonstrated no evidence of a difference in the clinical pregnancy rate per woman (50.3% with FSH vs 41.3% with CC, p-value=0.1) or live- birth rate per woman (45.3% with FSH vs 37% with CC, p=0.12) between the two treatments. However, the results based on per-protocol analysis found a higher clinical pregnancy rate per woman (58% with FSH vs 44% with CC, p=0.03) and live-birth rate per woman (52% with FSH vs 39% with CC, p=0.04) with a low dose of FSH. There was no evidence of a difference in the multiple pregnancy rate between the two treatments

(3.4% with FSH and 0% with CC, p-value was not provided). The cost of treatment was also similar for CC and FSH treatments. The authors of the RCT concluded that OI with low-dose FSH protocol improves reproductive outcomes for treatment-naive women with

PCOS (Homburg, et al., 2012).

In the second trial conducted by López and colleagues, 76 anovulatory therapy-naive women, due to PCOS, were randomised to receive three cycles of either gonadotropin

(low-dose recombinant FSH) or CC. The trial found no statistical difference in the pregnancy rate per woman (24% with CC vs 42% with recombinant FSH, p=0.09), live- birth rate per woman (16% with CC vs 29% with recombinant FSH, p=0.17) or multiple pregnancy rate (0% with CC vs 18% with recombinant FSH, p-value was not provided).

The authors concluded that gonadotropin (low-dose recombinant FSH) is an equally effective alternative to CC as a first-line treatment for anovulatory PCOS women (López, et al., 2004).

A meta-analysis of these two RCTs showed a significantly higher clinical pregnancy rate per couple (OR 0.61, 95%CI 0.4–0.93, 2 RCTs, n=378 women) and ongoing pregnancy/ live-birth rate per couple (OR 0.64, 95%CI 0.41–0.98, 2 RCTs, n=378 women) in favour of gonadotropins (FSH). There was no evidence of a difference in the multiple pregnancy rate (OR 0.26, 95%CI 0.06–1.06, 3 RCTs, n=696 women) (Brown and Farquhar, 2016).

iii) Laparoscopic ovarian drilling

Laparoscopic ovarian drilling (LOD) was first described by Gjönnaess in 1984

(Gjönnaess, 1984) and is a second-line therapy in CC-resistant PCOS women. A recent systematic review and meta-analysis identified 19 RCTs comparing LOD (with or without OI) and OI with other treatments (CC, CC and metformin, CC and tamoxifen, gonadotrophins and aromatase inhibitors) in clomiphene-citrate resistant PCOS women.

The meta-analysis found no evidence of a difference in the pregnancy rate per woman

(OR 0.94, 95%CI 0.78–1.14, 18 RCTs, n=1,930 women) and live-birth rate per woman

(OR 0.77, 95%CI 0.59–1.01, 8 RCTs, n=1034 women) in clomiphene-citrate resistant

PCOS women undergoing LOD compared to other medical treatments. The rate of

46 multiple pregnancies was significantly lower in the LOD (with or without OI) compared to trials using gonadotrophins (OR 0.13, 95%CI 0.03– 0.52, 5 trials, n=166 women)

(Farquhar, et al., 2012).

d) Controlled ovarian hyperstimulation (COH)

Early IVF pregnancies have been achieved from unstimulated ovulatory cycles, but the success rate was low (Pelinck, et al., 2002). Dodson and colleagues first proposed the use of COH+IUI in 1987 as an alternative therapeutic option for intractable, unexplained infertility (Dodson, et al., 1987). Since then, COH has emerged as a potential therapeutic modality for women with anovulatory infertility or unexplained infertility and is a common protocol in the preparation for IVF procedures

(McClamrock, et al., 2012).

The aim of COH depends on the strategy used. COH with TI or IUI is intended to induce development of two or three pre-ovulatory follicles, whereas COH with IVF aims to yield many oocytes to compensate for inefficiencies in laboratory procedures that limit fertilisation and subsequent embryo development in vitro (Fauser, et al., 2005). However, the development of multiple follicles in COH cycles increases the risk of multiple gestations and OHSS compared to natural conception cycles (Dickey, 2007, Practice

Committee of the American Society for Reproductive Medicine, 2012). Earlier studies estimated that the combined contribution of COH+IUI and OI to the national multiple birth cohort increased from 18.9% in 1997 to 22.8% in 2003, which is more than six times higher than published estimates for ART singleton births in the US (McClamrock, et al.,

2012) (Schieve, et al., 2009, Wright, et al., 2007).

In a narrative review of 10 selected series of prospective RCTs that reported on gestational plurality associated with COH+IUI (gonadotropins), the authors found twin and high-order pregnancy rates after COH+IUI were as high as 28.6% and 9.3% respectively (McClamrock, et al., 2012).

A systematic review and meta-analysis identified 43 RCTs that compared the outcomes of different COH protocols (anti-oestrogens, gonadotrophins with or without gonadotropin-releasing hormone (GnRH) agonists/antagonists) with IUI in couples with unexplained infertility, male-factor infertility and minimal-to-mild endometriosis. The review found no evidence of a difference in the pregnancy rates between women who used 150 IU gonadotrophins per day and those who used 75 IU gonadotrophins per day

(OR 1.2, 95%CI 0.69–1.9, 2 RCTs, n=297). The review findings suggest that a low-dose gonadotropin protocol can potentially be used to reduce the risk of multiple gestations without compromising the reproductive outcome of women with unexplained infertility, male factor infertility and women with minimal-to-mild endometriosis (Cantineau and

Cohlen, 2007).

e) Surgical treatments

Surgical treatment, such as laparoscopic surgery, is considered for women with endometriosis, tubal factor infertility and uterine abnormalities (Strathy, et al., 1982). In some cases, the purpose of surgical treatment is to restore the anatomic and functional integrity of the reproductive organs so that the woman is able to conceive spontaneously

(Coccia, et al., 2008).

i) Surgical treatment and the success rate in women with minimal-to-mild endometriosis

The general aim of surgical treatment using laparoscopic excision is to destroy or remove all visible endometriotic lesions and repair the damage to organs and other sites caused by endometriosis. However, surgery does not completely reverse the chronic inflammatory state or repair severe anatomical distortion (Garry, 2004, Koch, et al.,

2012).

There is conflicting evidence on whether laparoscopic surgery or diagnostic laparoscopy for early-stage endometriosis improves reproductive outcomes for women with minimal- to-mild endometriosis. In 1997, Marcoux and colleagues randomised 341 women with minimal-to-mild endometriosis to either laparoscopic surgery or diagnostic laparoscopy.

These women were followed up for 36 weeks after the procedure and results showed that laparoscopic surgery increased the cumulative probability of a pregnancy up to 20 weeks gestation by 73% (30.7% with surgery vs 1.7% with diagnostic laparoscopy, 95%CI 1.2–

2.6) (Marcoux, et al., 1997). The second RCT conducted by Gad and colleagues randomised 20 women to laparoscopic surgery or diagnostic laparoscopy and reported a pregnancy rate of 28.0% in the laparoscopic surgery group compared to 23.8% in the diagnostic laparoscopy group after an 18-month follow-up. However, this study would have been underpowered to detect any differences in outcomes and must be viewed with caution (Gad and Badroui, 2012). Moini and colleagues reported a similar pregnancy rate for women who had laparoscopic surgery (24.0%) and the diagnostic laparoscopy group

(18%) (p=0.49). The trial did not report on live-birth or multiple pregnancy rates (Moini, et al., 2012). The meta-analysis of these RCTs found a higher clinical pregnancy rate per woman (OR 1.89, 95%CI 1.25–2.86, 3 RCTs, n=528 women) and ongoing

49 pregnancy/live-birth rates per couple with laparoscopic surgery compared to diagnostic laparoscopy (OR 1.94, 95%CI 1.20–3.16, 2 RCTs, n=382) (Duffy, et al., 2014).

It is less clear if surgical treatment improves the live-birth rate in women with minimal- to-mild endometriosis prior to COH+IUI. One small retrospective study has assessed laparoscopic treatment for minimal-to-mild endometriosis prior to attempting COH+IUI.

Werbrouck and colleagues found no difference in the cumulative live-birth rates after up to four cycles with COH+IUI between women who recently had surgical treatment for minimal or mild endometriosis and women with unexplained infertility (70.2% in women with minimal endometriosis, 68.2% in women with mild endometriosis, 66.5% in women with unexplained infertility (p-value was not reported)) (Werbrouck, et al., 2006).

ii) Surgical treatment and the success rates for tubal factor infertility

Disease of the tubal fallopian tubes accounts for 25–35% of reported cases of female infertility (Hunter and Hart, 2009). The latest ANZARD report indicates that tubal disease accounted for 11.3% of all initiated autologous fresh ART cycles in 2015 (Fitzgerald, et al., 2017).

Tubal surgery is a viable treatment option for tubal factor infertility. It offers the advantages of being a one-time and usually minimally invasive outpatient procedure.

Women treated with tubal surgery may attempt conception every month without further intervention. More importantly, tubal surgery avoids the adverse outcomes of OHSS and multiple gestations associated with alternative treatment strategies such as ART (Chua, et al., 2017, Practice Committee of the American Society for Reproductive Medicine,

2012). Chua and colleagues conducted a meta-analysis combining 22 observational studies of women (n=2,810) undergoing salpingostomy for tubal infertility (due to hydrosalpinges) and reported a cumulative pregnancy rate of 20.0% (95%CI 17.5–22.8)

50 at one year and 25.5% (95% CI 22.2–29.4) at two years. The authors concluded that the clinical effectiveness in terms of live-birth rate between tubal surgery and NC+EM or

ART for women with tubal infertility is still unclear (Chua, et al., 2017).

Nonetheless, the Practice Committee of American Society for Reproductive Medicine advises clinicians to consider factors such as patient preference, maternal age, ovarian reserve, prior fertility, number of children desired, site and extent of the tubal disease, presence of other infertility factors, surgeon’s experience, and cost and success rates of the ART treatment when deciding on the optimal treatment for women with tubal infertility (Practice Committee of the American Society for Reproductive Medicine,

2012).

2.8 Summary

Although non-ART treatments are often considered the first-line treatment for infertility in relatively good prognosis patients, they play a limited role in the management of women of advanced maternal age, generally considered to be aged 37 or older. Due to the diminishing number and quality of oocytes with maternal age, advanced ART treatments are generally considered the best approach to treatment. Chapter 3 describes the process of ART treatment and review the contemporary literature on the indications for ART in older women.

Chapter 3

Overview of assisted reproductive technology and intrauterine insemination

Chapter 3– Overview of assisted reproductive technology and intrauterine insemination

3.1 Introduction

The in vitro fertilisation (IVF) procedure was first developed to assist couples in which the woman had irreparable tubal damage (Rock and Menkin, 1944) Until 1978, when the successful first live born baby using IVF was reported, women without functioning fallopian tubes were considered by physicians to have no chance of conception. Methods such as reparative surgery or tuboplasty to re-establish a conduit for gametes to transit were the only options for women with damaged tubes. However, the success rate of these approaches was limited, with less than one in four women achieving pregnancy (Hull and

Fleming, 1995, Wang and Sauer, 2006).

Over the last three decades, there have been significant developments in the reproductive technologies and protocols leading to improved success rates in infertile couples with a range of indications. Today, the term ‘ART’ (assisted reproductive technology) replaces

‘IVF’ as ART covers a broader spectrum of techniques and protocols (such as embryo transfer, preimplantation genetic diagnosis, and gamete and embryo cryopreservation), which are being increasingly used in clinical practice to optimise treatment outcomes.

Therefore, ART procedures comprise all interventions that involve the in-vitro handling of both human oocytes and sperm, or of embryos for the purpose of reproduction (Zegers-

Hochschild, et al., 2017). In line with this definition, artificial insemination (where, following laboratory preparation, sperm are placed into the woman’s reproductive tract) is considered a non-ART procedure in this thesis.

This thesis refers to ART treatment and procedures as IVF (where oocytes and sperm are joined outside the body prior to transfer of embryos into the uterus) and other enhanced

53 techniques and protocols, including embryo transfer, preimplantation genetic diagnosis, and gamete and embryo cryopreservation. Non-ART treatment refers to expectant management, ovulation induction, ovarian stimulation, intrauterine insemination and surgery.

3.1.1 ART treatment steps

The steps in a standard fresh ART cycle with fresh embryo(s) transfer include the following:

1. Controlled ovarian stimulation (COH) during which follicle stimulation

hormone (FSH) is administered to a woman to induce multiple follicle growth

2. Monitoring (mainly ultrasound), sometimes hormone measurement), is carried

out to assess the growth of the follicles.

3. Trigger to induce oocyte maturation

4. Transvaginal oocyte pick-up (OPU)

5. Insemination or Intracytoplasmic sperm injection (ICSI)

6. In vitro culture, during which a fertilised oocyte is cultured for 2–3 days to form

a cleavage-stage embryo (6–8 cells) or 5–6 days to create a blastocyst-stage

embryo (60–100 cells).

7. Embryo transfer usually at cleavage or blastocyst stage

8. Luteal phase support includes administration of progesterone, estrogen (E2) or

human chorionic gonadotropin (hCG).

9. Freezing of surplus embryos if of suitable quality

During an ART cycle, treatment may be discontinued at any stage for reasons including an inadequate ovarian response to stimulation, failure to obtain oocytes, failure of oocyte

54 fertilisation, inadequate embryo growth or patient choice (please refer to Chapter 4.4.1 for reasons that couples discontinue treatment)..

3.1.2 Effectiveness of ART versus non-ART treatments

The effectiveness of ART must be considered in comparison to other fertility treatments for various indications. Treatment for ART-naive infertile couple, where the female partner is under age 36, typically starts with the least invasive and less costly interventions using non-ART treatments such as expectant management (EM) and intrauterine insemination (IUI), before progressing to more invasive, costly and burdensome treatments such as ART. However, there are differences in opinion as to whether this progressive treatment sequence is optimal for indications such as couples with unexplained infertility or mild male-factor infertility (Armstrong and Akande, 2013,

NICE, 2013, Practice Committee of the American Society for Reproductive Medicine,

2006). This is reflected in the conflicting recommendations in the clinical guidelines for women under age 36 with unexplained infertility or with a partner who has mild male- factor infertility: the current American Society for Reproductive Medicine (ASRM) guideline recommends a progressive approach starting with six cycles of IUI before IVF, while the National Institute for Health and Care Excellence (NICE) guideline recommends against the use of IUI, and encourage women in the United Kingdom (UK) to undertake IVF as the first-line treatment after two years of natural cycle with EM due to the lack of robust evidence on the role of IUI (with or without COH) (Bhattacharya, et al., 2008, NICE, 2013, Steures, et al., 2006).

For women aged 36 or over, there is a general consensus to start investigating treatments for infertility after six months to one year of infertility (Johnson and Tough, 2012,

National Institute for Health and Care Excellence, 2013, Practice Committee of the

American Society for Reproductive Medicine, 2008).

The following section reviews the existing evidence on the effectiveness of ART in relation to other commonly used non-ART treatments in couples with unexplained infertility, men with mild male-factor infertility and women of advanced maternal age3. a) Couples with unexplained infertility or men with mild male-factor infertility

i) ART versus expectant management (EM)/timed intercourse (TI)

The effectiveness of ART compared to natural cycles with expectant management

(NC+EM) or timed intercourse (NC+TI) in couples with unexplained infertility remains unclear. A recent systematic review and meta-analysis of randomised controlled trials

(RCTs) identified two RCTs that compared IVF and NC+EM for couples with unexplained infertility, but only one RCT reported on the live-birth rate. This trial reported a higher live-birth rate per woman after one cycle of IVF compared to three months of NC+EM (odds ratio (OR) 22.0, 95%CI 2.56–189.37, n=139 women) (Hughes, et al., 2004). However, a recent longitudinal cohort study of 437 women with a mean age of 32, who were referred to a fertility clinic for a mean duration of unexplained infertility of 21 months, reported that most pregnancies were conceived spontaneously (73.9%) compared to using ART (13.5%) in the five-year follow-up period (Brandes, et al., 2011).

ii) ART versus IUI alone

Two RCTs have compared IVF and IUI alone in couples with unexplained infertility or mild male-factor infertility. There was no evidence of a difference in the clinical pregnancy rate between the two treatments (OR 4.83, 95%CI 0.94–24.95, 1 RCT, n=44

3 Unless otherwise stated, this thesis defines women of advanced maternal age as >36. 56 women), but the live-birth rate per woman was higher for IVF compared to IUI alone (OR

2.47, 95%CI 1.19–5.12, 2 RCTs, n=156 couples). There was no evidence of a difference in the multiple pregnancy rate between the two interventions (OR 1.03, 95%CI 0.04–

27.29, 1 RCT, n=44 women) (Pandian, et al., 2015). A cost analysis conducted alongside the RCT by Goverde and colleagues found that although the live-birth rate was higher with IVF, it cost more than five times that of IUI alone (€3350 vs €623) (Goverde, et al.,

2000).

iii) ART versus COH+IUI

A recent systematic review and meta-analysis identified five RCTs comparing IVF and

COH+IUI in women with unexplained infertility (Pandian, et al., 2015). The trials were analysed by treatment status due to a significant heterogeneity between the studies. The meta-analysis found that among treatment-naive women, the clinical pregnancy rate per woman was higher after IVF compared to those who were given IUI + gonadotrophins

(OR 1.45, 95%CI 1.03–2.03, 3 RCTs, n=627 women), but there was no evidence of a difference in the live-birth rate per woman (OR 1.27, 95%CI 0.94–1.73, 4 RCTs, n=745 women) or the multiple pregnancy rate per woman (OR 0.79, 95%CI 0.45–1.39, 4 RCTs, n=745 women).

The review identified one RCT that compared IVF and IUI + gonadotrophins in pre- treated women (i.e. these women received CC+IUI before receiving IUI+ gonadotrophins or IVF). The trial reported a higher pregnancy rate (OR 14.13, 95%CI 7.57–26.38, n=280 women) and live-birth rate per woman (OR 3.90, 95%CI 2.32–6.57, n=280 women) in the IVF group than the IUI + gonadotrophins group (Pandian, et al., 2015).

It is noteworthy that in the Cochrane review, only one RCT followed up couples for 12 months after randomisation. In this trial, 602 couples with unexplained infertility or male-

57 factor infertility were randomised to receive one of three treatment protocols: (a) three cycles of IVF with elective single embryo transfer (eSET) plus any subsequent frozen embryo transfers (IVF+eSET); (b) six cycles of IVF in a modified natural cycle (i.e. a cycle in which mono-follicular growth results in one oocyte at follicular aspiration and in one embryo after fertilisation (modified natural IVF)); or (c) six cycles of COH+IUI.

There was no evidence of a difference in the cumulative live-birth rate (CLBR) in the three treatment protocols (IVF+eSET (59%), COH+IUI (56%) and modified natural IVF

(51%) (Bensdorp, et al., 2015).

The authors of the Cochrane review concluded that IVF is associated with a higher live- birth rate than IUI alone. In women who were pre-treated with clomiphene + IUI, IVF was associated with a higher live-birth rate compared to COH (gonadotropins) + IUI.

However, there was no evidence of a difference in the live-birth rate between IVF and

COH+IUI in treatment-naive women (Pandian, et al., 2015). b) Women of advanced maternal age

This following section discusses the existing evidence on the effectiveness of ART in relation to other commonly used non-ART treatments in women of advanced maternal age.

i) ART versus expectant management (EM)/timed intercourse (TI)

As fertility decreases with increasing female age, NC+EM or TI is generally not recommended for women aged 36 or over. Therefore, there is a general agreement to encourage older women to commence investigation into treatment of infertility after six months to one year of infertility (NICE, 2013, Practice Committee of the American

Society for Reproductive Medicine, 2008).

ii) ART versus IUI alone

There is a lack of prospective studies comparing IUI alone and IVF in women of advanced maternal age. One retrospective study reported that the live-birth rate after IUI alone was

4.8% (1 live birth/21 cycles) in women aged 40 or over (Haebe, et al., 2002). This lies within the range of the average live-birth rate of 4–10% per fresh ART cycle reported for this patient group (Fitzgerald, et al., 2017, Luke, et al., 2012, Smith, et al., 2015).

However, factors such as time to pregnancy and costs should be considered when deciding on the appropriate treatment.

ii) ART versus COH+IUI

The beneficial effect of IVF compared to COH+IUI in women of advanced maternal age has been reported in two studies. A single RCT, the Forty and Over Treatment Trial

(FORT-T), randomised 154 women aged 38–42 with ≥6 months of unexplained infertility to receive two cycles of one of three treatment protocols: (a) CC+IUI (n=51 women), (b)

FSH+IUI (n=52 women) or (c) Immediate IVF (n=51 women) (Goldman, et al., 2014).

The trial found that immediate treatment with two cycles of IVF for older women (aged

38–42) resulted in a higher cumulative clinical pregnancy rate per couple (44.7% with

IVF vs 14.0% with either CC+IUI or FSH+IUI, p=0.0001) and CLBR per couple (27.7% with IVF vs 9.7% with either CC+IUI or FSH+IUI, p=0.008) compared to two cycles of

COH+IUI (CC+IUI and FSH+IUI). However, the mean time to pregnancy was shorter with COH+IUI, with 2.1 ± 0.1 months with CC+IUI, 3.0 ± 0.1 months with FSH+IUI and

5.7 ± 0.2 months with immediate IVF (Goldman, et al., 2014).

This finding is consistent with a cohort study of 247 women which reported a higher live- birth rate with IVF (13.7%) compared to both COH+IUI (2.6%) and in vitro maturation

(0%) in women aged 40 or older with diminished ovarian reserve or unexplained infertility (Wiser, et al., 2012).

3.1.3 Effectiveness of IVF versus in vitro maturation

iv) ART versus in vitro maturation (IVM)

It is estimated that polycystic ovary syndrome (PCOS) occurs in 5–10% of all women of reproductive age and 50% of infertile women (Azziz, et al., 2005). It has been shown that women with PCOS are at an increased risk of developing ovarian hyperstimulation syndrome (OHSS) during ART. In vitro maturation (a sequence of laboratory procedures that enable extracorporeal maturation of immature oocytes into fully mature oocytes that are capable of being fertilised with potential to develop into embryos) offers a useful alternative for earlier retrieval of immature oocytes from small antral follicles, with no or low doses of gonadotrophins to stimulate the follicles, and less pain experienced during oocyte retrieval (relative to IVF) (Reinblatt, et al., 2011, Seyhan, et al., 2014). However, a recent systematic review and meta-analysis did not locate any RCTs that have compared

IVM followed by IVF (with or without ICSI) and conventional IVF (with or without

ICSI), and hence the benefits of using IVM in terms of the live-birth rate for women with

PCOS remains unclear (Siristatidis, et al., 2015). Previous studies have found that IVM protocol provides the advantage of eliminating the risk of OHSS in women with PCOS compared to other treatment protocols such as IVF (Gremeau, et al., 2012, Walls, et al.,

2014).

The effectiveness of IVM in older women was assessed in a cohort study by Wiser and colleagues. In the study, they compared the outcomes of IVM in women of different ages and found the pregnancy rate declined with maternal age, with almost one in three women

60 aged 26–35 achieving an ongoing pregnancy after IVM (30.0% per woman) and none in women aged 40–45 (0%) (Wiser, et al., 2011).

3.2 Role of enhanced ART techniques and protocols in women of advanced maternal age

Although IVF may represent the best chance for some infertile couples to conceive a child, it is limited in its ability to fully compensate for the age-related decline in the quantity and quality of oocytes (Practice Committee of the American Society for

Reproductive Medicine, 2008). Therefore, advanced maternal age remains a significant challenge to achieving pregnancy and a live birth following IVF.

The latest report by the Australian and New Zealand Assisted Reproduction Database

(ANZARD) on all ART treatment cycles undertaken in Australia and New Zealand in

2015 indicates that 28.2% of all fresh cycles were undertaken by women aged 40 or over.

However, of these initiated cycles, 13% were cancelled before oocyte retrieval, and 39% of all the initiated cycles (excluding ‘freeze-all’ cycles) did not advance to embryo transfer, resulting in a per-cycle clinical pregnancy rate of 8.0% and live-delivery rate of

4.5% in women aged 40 or over (Fitzgerald, et al., 2017).

Several population-based studies also reported similar low live-birth rates in older women. Based on the data of all ART cycles performed in the UK between 2003 and

2012, Smith and colleagues reported that the live-birth rate in the first fresh cycle in women aged 40–42 was 12.3% and this declined to 3.7% in women aged over 42 (Smith, et al., 2015). In the United States (US), Luke and colleagues used linked cycles of ART obtained from the Society for ART (SART) database for 2004–2008 and reported that the live-birth rate per initial fresh cycle in women aged 38–40 was 21.7%, declining to 11.4% in women aged 41–42 and 4.0% in women aged 43 (Luke, et al., 2012). 61

Over the years, several enhanced ART techniques and protocols have been developed and introduced into the clinical setting, and some have been used in older women to improve their chance of success. The following section discusses five commonly used enhanced

ART techniques and protocols and the current evidence on their effectiveness in improving the chance of success in older women.

3.2.1 Intracytoplasmic sperm injection (ICSI)

ICSI is a technique to inject a single sperm directly into the oocyte. This technique was first introduced as a fertilisation technique for patients with severe male-factor infertility who could not be treated by standard IVF treatment (Mahadevan, et al., 1983).

The first successful application of ICSI was reported in 1992 (Palermo, et al., 1992).

Since then, the use of ICSI has expanded to include other indications such as mild and moderate male-factor infertility, advanced maternal age and unexplained infertility

(Babayev, et al., 2014, Practice Committee of the American Society for Reproductive

Medicine, 2012). This is reflected in the International Committee Monitoring Assisted

Reproductive Technologies (ICMART) world report which indicates an increase in the use of ICSI from 60.6% of all ART cycles in 2004 to 67.4% in 2010, despite a stable rate of male-factor infertility diagnosis (<40%) over the same period (Dyer, et al., 2016).

A similar trend has been reported in the US. Data from the National ART Surveillance

System (NASS) has shown that among fresh cycles with non-male factor infertility, the use of ICSI has increased from 15.4% of all ART cycles in 1996 to 66.9% in 2012 (Boulet, et al., 2015, Kissin, et al., 2015). Chambers and colleagues examined ICSI use in Australia from 2002 to 2013 and noted an increase from 55.7% of all ART cycles to 64.8%, although male-factor infertility remained stable at ~20% over the same period (Chambers, et al., 2010).

The expansion of ICSI use for couples with non-male factor infertility, such as women of advanced maternal age, may be due to its perceived benefits in achieving a better fertilisation rate than for a standard IVF procedure (Babayev, et al., 2014). As oocytes retrieved from older women are often of lower quality compared to those from younger women, ICSI use may have the potential to improve fertilisation rates and result in better clinical outcomes (Korkmaz, et al., 2015, Tannus, et al., 2017).

However, several studies have found that ICSI offered no added advantage over standard

IVF in terms of reproductive outcomes for non-male factor infertility regardless of maternal age (Bhattacharya, et al., 2001, Boulet, et al., 2015, Chambers, et al., 2016,

Tannus, et al., 2017).

A recent retrospective cohort study that compared conventional IVF (n=255 women, mean age=41.1) vs ICSI (n=490 women, mean age=41.2) reported no difference in the clinical pregnancy rate (21.1% vs 16.7% respectively, p=0.82) or live-birth rate (11.9% vs 9.6% respectively, p=0.71) per cycle started between conventional IVF and ICSI. A similar outcome was also reported in the subanalysis where IVF and ICSI techniques were compared in women with poor prognosis (≤3 oocytes retrieved) (Tannus, et al., 2017).

The Practice Committees of the American Society for Reproductive Medicine and the

Society for Assisted Reproductive Technology (SART) have also concluded that there is insufficient evidence to support the use of ICSI in the absence of male-factor infertility

(NICE, 2014, Practice Committee of the American Society for Reproductive Medicine,

2012). Previous US studies suggested that the provision of insurance cover for fertility treatments may have contributed to the increased use of ICSI for non-male factor infertility (Bitler and Schmidt, 2012, Jain, et al., 2002).

Questions of safety have been raised about the use of ICSI for non-male factor infertility

(Fauser, et al., 2014, Rubino, et al., 2015). Previous meta-analyses have been largely reassuring regarding the risk of adverse perinatal outcomes following ICSI use (Hansen, et al., 2013, Lie, et al., 2004, Wen, et al., 2012). However, a recent population-based study by Davis and colleagues found an increased risk of birth defects in ICSI infants (OR 1.55,

95%CI 1.24–1.94) (Davies, et al., 2012).

Some have argued that the increased risk observed in ART children may be due to a patient’s infertility (Rimm, et al., 2004, Rimm, et al., 2011), although it is likely that the causes of adverse perinatal outcomes are multifactorial with risk factors such as patient’s infertility and the intervention contributing to the increased risk of birth defects (ESHRE

Capri Workshop Group, 2014, Hansen, et al., 2013).

3.2.2 Preimplantation genetic diagnosis (PGD)

PGD is a technique to remove one or more cells from the embryo for analysis of chromosome aneuploidy, other chromosomal disorders or genetic diseases. The PGD technique was first incorporated into ART to screen embryos of women who were known to carry chromosomal structural abnormalities (Munné, et al., 1995), sex-linked disorders

(Handyside, et al., 1990) and monogenetic disorders (Verlinsky, et al., 1990). This is referred to as preimplantation genetic diagnosis (PGD). Based on European Society of

Human Reproduction and Embryology (ESHRE) PGD data, collected from cycles carried out in 2011–2012, with follow-up of pregnancies and babies born until October 2013, showed the overall clinical pregnancy rate per transfer after PGD (for chromosomal abnormalities, sexing for x-linked diseases and single gene disorder) was 29% (De Rycke, et al., 2017).

Over the last decade, the indications for PGD have expanded to diagnose numerical chromosomal abnormalities (aneuploidy) in women with no known genetic disorders but predisposed to having aneuploidy in their embryos due to increasing maternal age, recurrent implantation failure or recurrent miscarriage (Rubio, et al., 2003, Voullaire, et al., 2007). This is referred to as PGD for aneuploidy (PGD-A) or preimplantation genetic screening (PGS).

For both PGD and PGD-A, the diagnosis can be performed on the polar bodies, blastomeres from cleavage-stage embryos or trophectoderm cells from the blastocyst

(Gianaroli, et al., 2012, Harton, et al., 2013).

PGD-A, using comprehensive aneuploidy screening of all 24 chromosomes, has been shown in observational studies to improve rates of implantation, clinical pregnancy and live births in women of advanced maternal age (Chen, et al., 2015, Dahdouh, et al., 2015,

Lee, et al., 2015).

To date, only one RCT of PGD-A in older women (aged 38–41) has been published. This study reported a significantly higher live-birth rate using blastomere array comparative genomic hybridisation (array CGH) analysis for blastocyst transfer compared to standard morphologic embryo selection at the blastocyst stage (44.0% vs 24.8% respectively). A cost analysis conducted alongside the RCT reported a higher mean cost per live birth with

PGD-A compared to cycles using standard morphologic embryo selection (€23,895 vs

€21,968 respectively, p-value not reported in the trial) (Rubio, et al., 2017).

The cost and clinical outcomes from this RCT are similar to results from our cohort study

(reported in Chapters 7 and 8), comparing PGD-A with morphological assessment for selection of embryos in ART-naive women aged 37 or over. This cohort study found that although PGD-A for selection of embryos during ART is a costly procedure, it led to a

65 higher live-birth rate per ‘single cycle’, and women undertook almost half the number of initiated cycles, embryo transfers and time to achieve a live birth compared to those who used standard morphological assessment of embryos alone during ART.

3.2..3 ‘Freeze-all’ strategy

The ‘freeze-all’ strategy is where all available fresh embryos are cryopreserved for subsequent frozen/thawed transfers (FET) in a non-stimulation cycle to optimise the success rate for infertile couples. There has been significant progress in cryopreservation methods and the vitrificaton technique has been shown to improve cryosurvival rates of vitrified blastocysts (95% of all blastocysts) and birth outcomes. Consequently, the freeze-all strategy has been increasingly used for women at risk of developing OHSS or undergoing preimplantation genetic testing cycles (Cobo, et al., 2012, Shapiro, et al.,

2014).

This strategy also provides a viable option for women with poor prognosis to accumulate embryos from a series of oocyte retrievals without immediate fresh embryo transfer

(embryo banking). This practice prevents the problem of premature discarding of ‘poor quality’ embryos when they can be cryopreserved and transferred in FET cycles to optimise their chance of conception (Garcia-Velasco and Fauser, 2016, Munné, et al.,

2016, Werner, et al., 2014). While embryo banking is not common in Australia, it is common in countries, such in the US, where the cost of ART and PGD-A makes embryo banking financially attractive.

The latest ANZARD report, for 2015, indicates that the rate of ‘freeze-all’ cycles (where all oocytes and embryos were cryopreserved for potential future use during ART) more than trebled over the five years to 2014 (5.0% of all initiated fresh cycles in 2011 to 17.1% in 2015) (Fitzgerald, et al., 2017).

The beneficial use of freeze-all cycles for subsequent FET has been reported. A systematic review and meta-analysis of 11 observational studies that compared the obstetrics and perinatal outcome of FET vs fresh embryo transfer concluded that singleton pregnancies after FET were associated with better perinatal outcomes compared to those after fresh embryo transfer during IVF: perinatal mortality (relative risk (RR) 0.68,

95%CI 0.48–0.96), small for gestational age (RR 0.45, 95%CI 0.30–0.66), preterm birth

(RR 0.84, 95%CI 0.78–0.90), low birth weight (RR 0.69, 95%CI 0.62–0.76) and antepartum haemorrhage (RR 0.67, 95%CI 0.55–0.81) (Maheshwari, et al., 2012).

Roque and colleagues conducted a systematic review and meta-analysis of three RCTs that assessed clinical outcomes after fresh and FET cycles. The meta-analysis found a higher ongoing pregnancy rate per embryo transfer among women who had FET cycles

(RR 1.32, 95%CI 1.10−1.59, n=316 women), although there was no evidence of a difference in the miscarriage rate between the fresh and FET groups (RR 0.83, 95%CI

0.43−1.60, n=316 women). Roque and colleagues concluded that the higher pregnancy rate could be a result of the embryo-endometrium synchrony during the FET cycles

(Roque, et al., 2013).

The effectiveness of the freeze-all strategy in older women was assessed in a recent RCT, which randomised women (mean age=36) who were using PGD-A (using next generation sequencing) to either a freeze-all cycle (n=91 women) or a fresh day 6 ET during the stimulated cycle (n=88 women) (Coates, et al., 2017). The trial reported that the freeze- all strategy with known euploid embryos led to better clinical outcomes than fresh embryo transfer cycles with known euploid embryos, in terms of a higher ongoing pregnancy rate

(62.6% in freeze-all group vs 40.9% in fresh ET group, p<0.01) and live-birth rate per transfer (61.5% in ‘freeze-all’ group vs 39.8% in fresh group, p<0.01) (Coates, et al.,

2017). 67

However, there are concerns that the freeze-all strategy may not be applicable to older women, as sequential transfer of cryopreserved embryos in multiple transfer FET cycles increases time to pregnancy and may not be practical in women who are nearing the end of their reproductive years (Kissin, et al., 2015, Lee, et al., 2015). Furthermore, it is important to consider that this strategy requires more treatment cycles and adds an emotional, physical and financial cost to the patients (or third-party payers) depending on the funding arrangement. Older women, particularly those with poor prognosis, often have no or very few supernumerary euploid embryos available for the freeze/thawing process. This is shown in the RCT conducted by Coates and colleagues

(Coates, et al., 2017) who randomised 179 women (mean age= 36 years) to either freeze-all or fresh embryo transfer after PGD-A to determine which embryo transfer strategy improve clinical outcomes. The RCT reported that only half of the women

(52%) in the fresh embryo transfer group compared to 67% in the ‘freeze-all’ group has at least one any expanded blastocyst for PGD-A. Overall, 40% of the total sample did not have any expanded blastocyst available at day 5 for biopsy.

3.2.4 Oocytes cryopreservation

Cryopreservation of oocytes for fertility preservation using the conventional slow- freezing technique was initially used to help preserve fertility in women and girls at risk of future ovarian failure due to chemotherapy or radiotherapy (Akar and Oktay, 2005,

Nakayama and Ueno, 2006, Wallace, et al., 2005). Since the improvement in vitrification techniques and the associated clinical outcomes, this strategy has is increasingly offered as a preventive measure for age-related fertility decline (Petropanagos, et al., 2015).

However, factors such as the optimal age for oocyte cryopreservation and the number and

68 quality of oocytes are critical in influencing the clinical efficiency of oocyte cryopreservation (Mesen, et al., 2015, Stoop, 2010).

A few small studies have reported the outcome of women who vitrified their oocytes to prevent age-related infertility (Hammarberg, et al., 2017, Hodes-Wertz, et al., 2013). The largest study of women who returned to use their stored oocytes was reported by Cobo and colleagues (Cobo, et al., 2015). Of the 120 women who vitrified their oocytes (at a mean age of 37.7) and returned to use their stored oocytes at a mean age of 40, 35% of the women achieved a clinical pregnancy and one-fifth achieved a live birth (Cobo, et al.,

2015).

There is a paucity of data on the long-term health of children born using oocyte cryopreservation with the vitrification technique. However, a recent review of 58 reports conducted between 1986 and 2008 found no significant difference in the incidence of congenital abnormalities between children born after vitrification and spontaneously conceived children (Noyes, et al., 2009).

3.2.5 Oocyte/embryo donation/recipient program

An oocyte/embryo donation/recipient program is where a woman receives oocytes or embryos donated from another woman. Initially developed as a treatment for premature ovarian failure in young women, the oocyte recipient program has become an important and viable alternative for couples with inherited conditions, female partner with premature ovarian failure, female partner with advanced reproductive age and diminished ovarian reserve, or with unexplained recurrent implantation failure affecting the chance of achieving a live birth (Marinakis and Nikolaou, 2011).

The 2015 ANZARD report indicates that women aged 40 or over who used donated oocytes are five times more likely to achieve a live delivery compared to women using

69 their own oocytes (21.9% vs 4.5%) (Fitzgerald, et al., 2017). An earlier population-based study using data from the ANZARD found that the donor’s age had the largest impact on the live-delivery outcome. It reported that cycles with a donor age of less than 30 was associated with a live-delivery rate of 27.4% compared to 13.5% in cycles with a donor age of ≥40 (Wang, et al., 2011). The higher success rate following the use of donated oocytes is similar to that reported in a recent US study which found more than one in four donor oocyte cycles using fresh embryos resulted in a term singleton live birth (27.5% of donor oocyte cycles) (Kawwass, et al., 2013).

However, a recent systematic review and meta-analysis of seven studies comparing the perinatal outcomes of oocyte donation and autologous oocytes in IVF pregnancies found an increased risk of preterm birth (OR 1.45, 95%CI 1.20–1.77, 6 studies), low birthweight

(OR 1.34, 95%CI 1.12–1.60, 5 studies), early preterm birth (OR 2.14, 95%CI 1.40–3.25,

3 studies) and very low birth weight (OR 1.51, 95%CI 1.51–1.95, 4 studies) in oocyte donation compared to autologous oocyte IVF pregnancies (Mascarenhas, et al., 2017).

Jeve and colleagues identified 11 studies that compared the pregnancy complications of donated oocyte and autologous oocytes in IVF (Jeve, et al., 2016). The meta-analysis found that the risk of developing hypertensive disorder in pregnancy is higher with donor oocyte pregnancy compared to autologous oocyte pregnancy (OR 3.92, 95%CI 3.21–

4.78, 5 studies). The chance of caesarean delivery (OR 2.71, 95%CI 2.23–3.30, 6 studies), development of a small-for-gestation-age baby (OR 1.81, 95%CI 1.26–2.60, 6 studies) and the preterm delivery rate (OR 1.34, 95%CI 1.08–1.66, 9 studies) are also higher with donor oocyte pregnancies.

3.3 Role of ART in women of advanced maternal age

3.3.1 ART and risk of congenital anomalies in women of advanced maternal age

It is well established that the rate of chromosomal abnormalities (aneuploidy) increases with female age, resulting in an increased risk of implantation failure, pregnancy loss or birth of a child with chromosome abnormalities such as Down syndrome (trisomy 21)

(Franasiak, et al., 2014, Mégarbané, et al., 2009). It is estimated that for women aged 40 or over, approximately 50–80% of all first-trimester pregnancy losses are due to chromosomal abnormalities in the developing foetus (Hodes-Wertz, et al., 2012,

Nagaoka, et al., 2012, Nasseri, et al., 1999).

Aneuploidic conceptions that result in a live birth typically result in serious congenital malformation and/or cognitive abnormalities (Hassold, et al., 2007, Mégarbané, et al.,

2009). The most common chromosomal abnormalities associated with older women are trisomies 21, 18 and 13, with trisomy 21 occurring in 1 in 400 women at age 35, 1 in 105 at age 40 and 1 in 12 at age 45 (Hollier, et al., 2000).

Questions have been raised about the role of ART in congenital anomalies. Hansen and colleagues reviewed 45 studies and reported an increased risk of birth defects in ART children compared to naturally conceived children (relative ratio 1.32, 95%CI 1.24–1.42).

This is equivalent to <2% of the 5 million children born through ART worldwide (Hansen, et al., 2013). Previous reviews and meta-analyses have also reached the same conclusion: there is a higher risk of birth defects following ART (Hansen, et al., 2013, Rimm, et al.,

2004, Wen, et al., 2012).

While factors such as increased maternal age, multiple births and preterm births are known risk factors for congenital anomalies (Lampi, et al., 2012), it remains unclear if

71 the increased risk of birth defects in ART infants is the result of underlying infertility or the unique aspect of ART techniques or protocols such as ICSI (Hansen, et al., 2013).

Previous studies have attempted to tease out the contribution of different risk factors such as the use of culture media, medications, ICSI technique or embryo cryopreservation on adverse perinatal outcomes (Halliday, et al., 2009, Pinborg, et al., 2013). There is growing evidence that suggests that the transfer of frozen-thaw embryos can reduce the risk of common adverse perinatal outcomes including preterm delivery (Dumoulin, et al., 2010).

A recent study by Davis and colleagues assessed the contribution of maternal age and lifestyle factors such as smoking in early pregnancy to the risk of birth defects in women undergoing IVF and ICSI and those who conceived naturally (Davies, et al., 2017). The study was based on a large Australian cohort of 2,211 IVF infants, 1,399 ICSI infants and

301,060 naturally conceived infants born between 1986 and 2002. The authors found that increasing maternal age (≥35) is not a risk factor for having a child with a birth defect in women undergoing ART. The study reported that within the IVF group, there was no significant difference in the risk of birth defects between the reference group aged 30–34 and older women aged 35–39 (adjusted OR 1.28, 95%CI 0.86–1.92) or those aged 40 or older (adjusted OR 0.63, 95%CI 0.24–1.69). Interestingly, when IVF and ICSI data are combined, the authors found a lower risk of defects in births to women aged 40 or over compared to the reference group aged 30–34 (adjusted OR 0.45, 95%CI 0.22–0.92).

As the number of ART-conceived children born to older women continues to grow, more research based on careful surveillance data is needed to obtain a better understanding of the risk factors of congenital anomalies following ART to minimise the risk of pregnancies resulting in babies born with birth defects (Barnhart, 2013).

3.3.2 ART and multiple pregnancies in women of advanced maternal age

The overall incidence of multiple births has increased over the last three decades, largely due to the increasing use of ART and an increase in spontaneous pregnancies in women of advanced maternal age (Black and Bhattacharya, 2010, Kulkarni, et al., 2013, Practice

Committee of the American Society for Reproductive Medicine, 2012). It is estimated that in the US in 2013, ART accounted for 1.5% of all babies born, but constituted 19% of all multiple birth infants, including 29.6% of all triplet or higher order infants

(Kulkarni, et al., 2017). In Europe in 2010, 19.2% of all pregnancies after ART were multiple pregnancies (European IVF-monitoring Consortium, et al., 2017). Australia has one of the lowest multiple pregnancy rates following ART globally, with the most recent

ANZARD data for treatments performed in 2015 reporting a multiple birth rate of less than 4.4% (Fitzgerald, et al., 2017). However, this is still almost three times higher than the national average for multiple births for 2015 (4.2% vs 1.5% of all conceptions)

(Australian Institute of Health and Welfare, 2017).

Multiple pregnancies following ART and non-ART treatments are a major cause of concern as these pregnancies, particularly those occurring in older women, are associated with increased maternal and perinatal risks as well as higher health care costs (Chambers, et al., 2014, Pinborg, 2005, Practice Committee of the American Society for Reproductive

Medicine, 2012). Women with multiple pregnancies have been shown to have higher risks of preeclampsia, antepartum and postpartum haemorrhage and caesarean delivery compared with women with singleton pregnancies (Duckitt and Harrington, 2005).

Although the mortality is low, the risk of morbidity for both the mother and infant increases with multiple pregnancies (Murray and Norman, 2014). Most twins and higher order multiples births are found to be more vulnerable than singleton children as they have a lower birthweight, suffer more complications at birth and are more often born 73 prematurely – all of which are associated with long-term health problems (Delobel-

Ayoub, et al., 2009). There is evidence that stillbirth and infant mortality rates are higher among multiple births than among singleton (Joseph, et al., 2005, Usta and Nassar, 2008).

In ART, multiple pregnancies are a common outcome usually resulting from the transfer of more than one embryo. The latest ANZARD report, for 2015, shows that 19% of double embryo transfers (DET) resulted in multiple pregnancies compared to 2% following single embryo transfers (SET) (Fitzgerald, et al., 2017). In recent years, clinical guidelines have been revised to encourage fertility centres to adopt elective single embryo transfer (eSET) into routine use for selected patients (Human Fertilisation Embryology

Authority, 2009, Practice Committee of the American Society for Reproductive

Medicine, 2012).

However, in a published systematic review and meta-analysis of RCTs, it was shown that eSET reduced pregnancy and live-birth rates in mostly young women with a number of good quality embryos available compared to DET (Pandian, et al., 2014). As additional fresh or FET cycles will be needed to achieve similar live-birth rates to DET, this can lead to higher treatment costs from undergoing multiple cycles, longer time to pregnancy and additional medical risk from these procedures (Min, et al., 2010, Thurin, et al., 2004).

Female age may be considered an indication for DET despite the increased risk of multiple pregnancies in this high-risk obstetric group (Miller, et al., 2016). This is because the chance of producing chromosomally abnormal embryos increases with female age, most of the transferred embryos will likely not implant (Kushnir and Frattarelli, 2009).

Therefore, in older women, it has been argued that transferring more than one embryo at a time, maximises the chance of success in older women (Gleicher, 2009; Gleicher, 2012).

There have been a number of published observational studies comparing eSET with DET in older women. A recent retrospective cohort study of 310 women (mean age=41) found a comparable clinical pregnancy rate per cycle (OR 1.04, 95%CI, 0.62–1.75) and live- birth rate per cycle (OR 1.43, 95%CI, 0.78–2.64) between women who had an elective single blastocyst transfer (eSBT) and those who had elective double blastocyst transfer

(eDBT). The multiple birth rate was lower in the eSBT group compared to the eDBT group (0% vs 16%, p=0.02) (Tannus, et al., 2017). This finding is consistent with an earlier study comparing eSET and DET in 628 women aged 40−44 (Niinimäki, et al.,

2012). The study reported a similar clinical pregnancy rate per fresh cycle (23.5% in eSET vs 19.5% in DET, p=0.22) and live-birth rate per fresh cycle (13.6% in eSET vs 11.0% in DET, p=0.32). The multiple birth rate (twins) was lower in the eSET group compared to the DET group (0% vs 7.5%, p=0.14) (Niinimäki, et al., 2012). The 2015 ANZARD report showed that the live-delivery rate per transfer was similar following SET and DET in women aged over 40 or over (8.2% vs 8.6% respectively) (Fitzgerald, et al., 2017).

3.4 Summary

In most developed countries, there is a trend towards delaying childbearing to after age

35. However, female fertility declines with age. Despite the significant progress in the clinical and laboratory aspect of ART, standard ART does not compensate for the age- related decline in the quality and quantity of oocytes. The live-birth rate in older women remains low compared to their younger counterparts. Older women are also at increased risk of pregnancy loss or birth of a baby with congenital abnormalities due to an increasing rate of aneuploidy in embryos. Although the use of donor oocytes has allowed many women of advanced maternal age to achieve a live birth, other alternative treatment strategies such as PGD-A are increasingly offered to older women to improve their chance of conceiving a genetically related child. The current evidence on the effectiveness of 75

PGD-A is reviewed in Chapter 6. The findings on the clinical and cost-effectiveness of

PGD-A versus standard ART cycles using morphological assessment for selection of embryos in ART-naive women aged 37 or older are presented in Chapters 7 and 8.

Chapter 4

Measuring ART success

Chapter 4–Measuring ART success

4.1 Introduction

‘Success’ can be defined as the achievement of a favourable or desired outcome. In ART, the concept of success is not universally agreed upon and has been described (Wilkinson, et al., 2016). This was indeed evident in a recent appraisal of 142 randomised controlled trials (RCTs) of ART, which identified more than 800 different combinations of measure of ART outcomes (Wilkinson, et al., 2016, Williams, et al., 2015).

The lack of uniform and widely accepted measures of ART success is largely due to the complex, multi-level nature of ART treatment. For example, at the ‘single’ cycle level, there are a series of stages, including ovarian stimulation, oocyte pick-up (OPU) and embryo transfer (ET), which can occur in a treatment cycle. At a single ‘complete cycle’ level, there is a fresh and a possible multiple frozen/thaw embryo transfer (FET) following a stimulation cycle; and at a patient level, multiple ‘complete cycles’ may be undertaken over the entire course of treatment (Wilkinson, et al., 2016).

The numerous clinical and procedural events that occur at different stages in a treatment cycle and over the entire course of treatment mean that a plethora of different numerators

(e.g. implantation rate, pregnancy rate, live-birth rate) and denominators (e.g. initiated single cycle, OPU, number of women) are available to develop different measures of ART success (Griesinger, 2016).

Furthermore, the different perspectives of stakeholders such as embryologists, clinicians and couples, play important roles in the definition and measure of ART success (Legro, et al., 2014). For example, embryologists may view an implanted embryo as a preliminary

ART success while infertile couples consider ART success as the birth of a healthy baby.

Although intermediate (surrogate) outcomes are informative for different stakeholders, the ultimate goal of fertility treatment is to provide safe medical treatment for male and female patients, resulting in a healthy live birth with minimum short and long-term adverse outcomes (Braakhekke, et al., 2015, Legro and Myers, 2004).

4.2 Commonly used numerators in measuring ART success

Typically, measure of outcome success includes a numerator, which represents a clinical parameter of interest (e.g. live birth), and a denominator, which provides the clinical context in which the outcome is measured (e.g. total number of patients or embryo transfer). This section discusses four numerators that are commonly used in measuring

ART outcomes.

4.2.1 Implantation outcome

Implanted embryos are those embryos that have successfully implanted in the uterus resulting in the formation of a gestation sac. The implantation outcome is commonly expressed as the percentage of embryos transferred (Zegers-Hochschild, et al., 2017).

Although implantation rate, as an outcome measure, provides the advantage of an immediate assessment of treatment effectiveness, it remains an intermediate outcome of

ART success (Griesinger, 2016, Land and Evers, 2004). Additionally, questions have been raised about the validity of implantation rate as a measure of ART success under any circumstance. These questions have centred around the ‘cohort effects’ where the implantation of one embryo predicts the implantation outcome of the remaining embryos in the cohort (Romanski, et al., 2017), and the statistical problem of pooling embryos by groups of women, which creates a sample of interrelated observations and thus violates the independence necessary for statistical testing and the intention-to-treat principle of analysis (Griesinger, 2016, Matorras, et al., 2005). The statistical problem is of particular

79 concern when assessing the effectiveness of embryo screening methods, such as preimplantation genetic diagnosis for aneuploidy (PGD-A), which often results in fewer or no embryos available for transfer. This phenomenon is shown in the retrospective cohort study comparing the clinical effectiveness of PGD-A and morphological assessment of embryos in ART-naive women aged 37 or over, reported in Chapter 6.

In this study, cohorts were assigned based on the embryo selection methods – PGD-A or morphological assessment of embryos used in their first fresh cycle [(PGD-A, n=110 women (PGD-A group); morphological assessment of embryos, n=1,983 women

(morphology assessment group)]. The study found that although fewer embryos were transferred in the PGD-A group compared to the morphology assessment group (1.11 vs

1.36 embryos transferred per ET respectively, p<.001), the PGD-A group achieved a higher implantation rate than the morphology assessment group (36.75% vs 13.75% respectively, p<0.001); however, cumulative live-birth rates (CLBRs) for up to three complete cycles showed no significant difference between the two study groups (30.90% vs 26.77%, p=0.34).

4.2.2 Pregnancy outcome

Pregnancy outcome is one of the most common numerator measures used to evaluate the effectiveness of ART treatment (Dapuzzo, et al., 2011, Wilkinson, et al., 2016). Although pregnancy rate has the practical advantage of a shorter lapsed time for measuring treatment outcome compared to using live birth, it is still an intermediate outcome for infertile couples seeking a healthy live birth (Practice Committee of American Society for Reproductive Medicine, 2012). Pregnancy does not guarantee a live birth. This is shown in the retrospective cohort study (reported in Chapter 6) which found that one in five pregnancies following PGD-A (19.5%) and one in three pregnancies in women who

80 used morphological assessment of embryos alone (34.8%) resulted in miscarriage, ectopic pregnancy or termination. Furthermore, ‘pregnancy’ can be variously defined and reported as biochemical, ongoing or viable and these variations can have significant implications when assessing the comparative effectiveness between interventions (Legro, et al., 2014, Wilkinson, et al., 2016).

4.2.3 Live birth outcome

Live birth has traditionally been the preferred primary outcome for ART success (Legro, et al., 2014, Sharma, et al., 2002). Live birth is generally defined as the birth of one or more live-born babies, with twins or higher order multiples (HOM) counted as a single live-birth delivery (Macaldowie, et al., 2012, Zegers-Hochschild, et al., 2017).

Compared to measures such as implantation rate or pregnancy rate, live birth provides a more quantitative and temporal measure of treatment benefit that is clinically meaningful for patients who want to know the probability of achieving a live birth from that cycle, and clinicians who want to compare treatment outcomes (Barnhart, 2014).

Moreover, all live births, regardless of the gestational age (e.g. 20, 22 or 28 weeks), used for birth registrations in different countries are reported as vital record events. Therefore, the collection of data on live births is more complete and less open to interpretation than other measures such as pregnancy outcome (Barfield, 2016). However, live birth outcome as a definitive measure of ART success has limitations because it does not measure the health outcomes of the infant and can mask important trends such as multiple birth rates

(as twins and HOM are counted as a single live-birth delivery). As a result, the focus on pregnancy or live-birth rates as a measure of success alone can incentivise unsafe ET practices, where more than one embryo is transferred to improve pregnancy and live-birth rates, but at the expense of a higher multiple birth rate (Stern, et al., 2012). Furthermore,

81 the ‘live birth’ outcome necessitates waiting nine months until an outcome can be recorded. Therefore, given the risk of multiple pregnancy following ART, the choice of success measure requires consideration as it has significant implications for the safety of the mother and the health of babies born from these technologies (Practice Committee of

American Society for Reproductive Medicine, 2012).

4.2.4 Singleton live birth at term

In an effort to incorporate the multiple pregnancy rate into the measure of ART success,

Min and colleagues proposed the ‘birth of a healthy singleton baby at full gestation per cycle initiated’ (known as Birth Emphasising a Successful Singleton at Term, BESST) as the primary measure of ART success (Min, et al., 2004).

The BESST outcome, or other similar measures, that shift the emphasis from pregnancy or live-birth rates to a primary endpoint that focuses on achieving the birth of a healthy baby, has been regarded by some as the ‘gold standard’ for ART success (Heijnen, et al.,

2004, Hughes, 2015, Wennerholm and Bergh, 2004).

However, there remains little agreement about combining both effectiveness and safety into a single measure, as proposed in the BESST index (Braakhekke, et al., 2015).

Braakhekke and colleagues have argued that a composite measure of ART hinders the ability to assess treatment effect (e.g. singleton live birth) and safety (e.g. risk of preterm birth and congenital anomalies) separately and this could potentially lead to erroneous conclusions about treatment effect. This is illustrated by Braakhekke and colleagues who reviewed the findings of one of the most highly cited RCTs that compared double embryo transfer (DET) and single embryo transfer (SET) following ART (Thurin, et al., 2004).

Braakhekke and colleagues noted that although the live-birth rate was comparable between SET and DET, the rate of multiple pregnancy rate was higher after DET

82 compared to SET (33.0% vs 0.8% respectively) and pre-term rate was 29% after DET and 12% after SET. However if the outcome was measured based on a composite measure

- ‘singleton live birth’, it would have shown a marginal advantage of SET compared to

DET [relative risk (RR) 1.3 (95%CI 1.1–1.7)], and inadvertently failed to highlight the increased risk of poorer outcomes associated with DET (Braakhekke, et al., 2015).

4.3 Commonly used denominators in measuring ART success

The selection of an appropriate denominator is important as it provides the clinical context and captures the exposure–risk relationship (Wilkinson, et al., 2017). The five most commonly used denominators in measuring ART success are described below.

4.3.1 Per initiated single cycle

Measuring outcomes per initiated single cycle (stimulated fresh or FET) is arguably the most relevant measure as it informs patients and clinicians of the probability of success

(e.g. live birth) from a single cycle. In most studies and ART registries, success rate per initiated single cycle is stratified by maternal age, cause of infertility and treatment modalities to provide a better assessment of the probability of success based on an intention-to-treat principle (Fitzgerald, et al., 2017).

The disadvantage of measuring outcomes on a per initiated single cycle basis is that it does not provide information about the chance of success beyond that cycle. This is an important consideration because, after an unsuccessful cycle, couples often want to know the chance of success if they return for subsequent cycles (Luke, et al., 2012, Maheshwari, et al., 2015, Malizia, et al., 2009, McLernon, et al., 2016). Furthermore, with increasing cryopreservation of embryos for subsequent FET cycles, ART success based on a per initiated single cycle basis does not provide information about the total chance of ART success over the entire treatment course (Maheshwari, et al., 2015).

4.3.2 Per embryo transfer

Success expressed per embryo transfer is calculated by dividing the selected numerator

(e.g. live births) by the number of transfers (fresh, FET or combined). Therefore, initiated cycles that fail to reach ET are excluded from the denominator in the calculation of the success rate.

Using embryo transfer as the denominator inflates the success rate, particularly when intermediate numerators such as clinical pregnancy are used. This is illustrated in the latest Australian and New Zealand Assisted Reproduction Database (ANZARD) report, which shows the clinical pregnancy rate for women aged 40–44 when expressed per initiated single cycle was 8.5%, but ‘increases’ to 15.5% when expressed as clinical pregnancy per embryo transfer (Fitzgerald, et al., 2017).

It should be noted that the numbers of cycles excluded from the success rate are not trivial, particularly for older women. The 2015 ANZARD report reveals that the number of initiated cycles which failed to reach OPU or embryo transfer accounted for 28.4% of initiated cycles (excluding ‘freeze-all’ cycles discussed below in section 4.4.2), with the majority of the excluded cycles among women aged 35 or over (74.0%) (Fitzgerald, et al., 2017).

However, an argument in favour of measuring ART success on a per embryo transfer basis is that it removes confounding factors due to differences in the patient’s prognostic characteristics so that ART outcome is assessed solely on the quality of embryos transferred (Abdalla, et al., 2009).

4.3.3 Per complete cycle

Over the last decade, the number of initiated fresh cycles resulting in the cryopreservation of oocytes and embryos for subsequent transfer in FET cycles has grown steadily 84

(Fitzgerald, et al., 2017, Wong, et al., 2014). This shift has been driven by recent advances in cryopreservation methods, particularly the vitrification technique, an increasing shift towards SET, and an improved understanding of the embryo-endometrium synchrony

(Penzias, et al., 2017, Shapiro, et al., 2011).

Estimates from the latest (2015) ANZARD report reveal that the number of autologous

FET cycles rose by 23% over the five years to 2015, with a corresponding increase in the live delivery rate per FET cycle from 19.1% (in 2011) to 25.3% (in 2015) (Fitzgerald, et al., 2017).

Therefore, to capture the increasing use of FET, and to provide a more holistic measure of ART treament success that includes both fresh and FET cyles from a stimulation cycle, cumulative success rate from a ‘complete cycle’ perspective is increasingly used for reporting ART success. In the per ‘complete cycle’ perspective, the numerator is the number of live births achieved from all ETs using embryos resulting from an initial stimulation cycle, and the denominator is the stimulated cycle (Chambers, et al., 2017,

McLernon, et al., 2016, Stern, et al., 2010).

This approach provides the advantage of measuring the patient’s total reproductive potential from the start of the stimulation cycle, and thus offers an all-inclusive success rate which is arguably more relevant and meaningful than a measure that provides a single cycle estimate to inform a patient of her chance of success in subsequent cycles. In addition, the ‘complete cycle’ perspective of measuring success is more likely to incentivise SET as it minimises the emphasis on the pregnancy rate or live-birth rate resulting from a single cycle (Doody, 2014, Stern, et al., 2012).

However, there is concern as to whether this approach of using a ‘complete cycle’ could potentially lead to a more aggressive ovarian stimulation for the purpose of maximising oocyte numbers for more FET cycles (Doody, 2014, Stern, et al., 2012).

4.3.4 Time to pregnancy

Time to pregnancy (TTP) offers an alternative approach to measuring ART success. This approach involves calculating the time that lapses between the commencement of treatment and achieving a pregnancy or a live birth (Sozou and Hartshorne, 2012). This approach has the advantage of comparing the probability of a live birth for different treatment protocols or interventions within the same timeframe (Daya, 2005,

Maheshwari, et al., 2015).

The Kaplan–Meier estimate is a common method used in calculating the TTP over a period of time. It involves computing probabilities of the occurrence of an event (e.g. pregnancy) at certain points in time. The cumulative probability of events is usually expressed in a graphical format with the log-rank test for comparison of TTP between two treatment protocols or comparison groups (Daya, 2005).

The use of TTP is illustrated in the cohort study of PGD-A in Chapter 7. In this study, women in both study groups (PGD-A group versus morphology assessment group) were followed up longitudinally for up to three ‘complete cycles’. The Kaplan–Meier method was used to determine TTP by calculating the time between the dates of the first OPU to the date of the first conception resulting in a live birth, with the log-rank test used for comparison of aggregate TTP between the study groups (Figure 4).

The analysis shows that it took women in the PGD-A group (i.e. women who commenced their first ART cycles using PGD-A) a significantly shorter time to achieve a clinical pregnancy resulting in a live birth compared to the women who used morphological

86 assessment of embryos alone (i.e morphology assessment group) (104.8 days vs 140.6 days, log-rank test p<0.05).

Figure 4:Time to clinical pregnancy from date of first OPU to conception resulting in a live birth for PGD-A group and morphology assessment group (Chapter 7)

4.3.5 Per woman as a denominator for measuring success

As most infertile women undertake more than one ART cycle over the entire course of treatment, there has been an increasing shift towards measuring and reporting CLBR over multiple single or ‘complete cycles’ per woman. The ‘complete cycle’ CLBR is defined as the cumulative chance of success from all fresh and FET cycles after a given number of ‘complete cycles’ per woman (Chambers, et al., 2017, McLernon, et al., 2016).

The reporting of ART success based on a per woman rather than per cycle or transfer basis provides a more accurate and realistic representation of the chance of success over an entire course of treatment (Chambers, et al., 2017, Maheshwari, et al., 2015,

McLernon, et al., 2016, Moragianni and Penzias, 2010). Moreover, as women are followed up over multiple cycles, this measure can be used to inform the patient of the 87 potential number of single or ‘complete cycles’ needed to reach a live birth, and to enable the comparison of treatment modalities and in making the decision whether to initiate treatment.

This is illustrated in the retrospective cohort study comparing PGD-A and morphological assessment of embryos alone reported in Chapter 7. The cohort found that although the

CLBR was comparable between the two study groups (i.e PGD-A versus morphology assessment group), women in the PGD-A group undertook significantly fewer number of

ART cycles to reach a live birth compared to women in the morphology assessment group

(6.06 cycles in the PGD-A group vs 10.98 cycles in the morphology assessment group; p<.01).

However, measuring ART success using ‘complete cycle’ CLBR has limitations. Because

CLBR requires longitudinal reporting of clinical outcomes, collection of the relevant data for reporting may become logistically difficult. Another limitation is that the variation in the number of treatment cycles between patients may affect the validity of the measure of the outcomes (Abdalla, et al., 2009)

4.4 Current challenges in measuring ART success

As mentioned in previous sections, the selection of an appropriate measure of ART success is less than straightforward. A suite of ART success measures would ideally incorporate clinical effectiveness, acceptability and safety. They would also be reliable, valid and easy to generate and understand (Abdalla, et al., 2009, Tiitinen, et al., 2004).

However, there are several factors that may affect the accuracy of the success rate calculation. The following section discusses three challenges in measuring ART success.

4.4.1 Impact of discontinuation rate

Previous studies have found that a large proportion of couples starting ART treatment

(23–60%) discontinue treatment before a live birth is achieved (Chambers, et al., 2017,

Custers, et al., 2013). While some couples stop treatment due to medical advice, others discontinue due to financial difficulties, or the physical and/or emotional burden of treatment (Brandes, et al., 2009, Gameiro, et al., 2012, Lande, et al., 2015, Rajkhowa, et al., 2005).

The high rate of discontinuation affects the estimate of the cumulative probability of success following ART because assumptions are made about the probability of success of women who discontinue treatment. Life table analysis (including the Kaplan–Meier estimate) assumes that women who discontinue treatment have the same chance of achieving a live birth as those who continue treatment. This method is likely to overestimate the prognosis of women who discontinue treatment and their chance of achieving a pregnancy resulting in a live birth. This is referred to as an ‘optimistic’ or

‘optimal’ estimate (Doody, 1993, Stolwijk, et al., 2000). Conversely, a method that assumes women who discontinue treatment have no chance of achieving a pregnancy is likely to underestimate their success rate. This approach of measuring CLBR is referred to as a ‘conservative’ estimate (Troude, et al., 2012, Verhagen, et al., 2008). In reality, it is likely that the true estimate of the CLBR falls between the optimal and conservative estimates. This approach of estimating success has been increasingly used in counselling patients about their likelihood of success over multiple ART cycles (Chambers, et al.,

2017, Maheshwari, et al., 2015, McLernon, et al., 2016, Smith, et al., 2015, Verhagen, et al., 2008).

4.4.2 ‘Freeze-all’ strategy

The trend of increasing use of ‘freeze-all’ cycles creates new dilemmas in the reporting of ART outcomes in the national ART registry reports and intervention studies (Doody,

2014, Kulak, et al., 2016). ‘Freeze-all’ cycles are fresh cycles where the intention is to cryopreserve all oocytes or embryos for potential future use. The increasing use of the

‘freeze-all’ strategy is likely to be multifactorial, including recent improvements in cryopreservation methods, the increasing use of SET in preference to DET, and the growing evidence of the superior endometrial receptivity associated with FET cycles. The growing use of embryo screening methods during ART, such as PGD-A, has also resulted in the increased use of freeze-all cycles (Doody, 2014, Shapiro, et al., 2014). The estimates from the latest ANZARD report (2015) indicate that the proportion of ‘freeze- all’ cycles in Australia and New Zealand has more than trebled over the last five years

(from 5% in 2011 to 17.1% in 2015) (Fitzgerald, et al., 2017, Macaldowie, et al., 2012).

Typically, ‘freeze-all’ cycles, which have no immediate transfer or measurable outcome in the reporting period, are excluded from the denominator of the live-birth rate because the cycle does not represent a risk of pregnancy. However, there has been considerable debate as to whether this approach inflates the success rate because the stimulated cycles that resulted in a ‘freeze-all’ cycle are not counted in either the fresh or FET denominators

(Kulak, et al., 2016).

Kushnir and colleagues used the 2013 ART data from the Centers for Disease Control and Prevention (CDC) to examine the effect of excluding freeze-all cycles in the current reporting of ART success outcomes by the CDC and the Society for Assisted

Reproductive Technology (SART) registry. The authors found that the live-birth rate was inflated by 15–56.3% when ‘freeze-all’ cycles were excluded from the success rate

90 calculation (1.6% per fresh cycle vs 0.7% per initiated fresh cycle), an indicative that this approach of success rate calculation is potentially misleading (Kushnir, et al., 2016).

It has been suggested that CLBR, which takes into account the contribution of all fresh and frozen embryo transfer cycles over repeated complete cycles, is a better way of measuring ART success because the ‘freeze-all’ cycles are counted in the CLBR estimates. Consequently, this addresses the issue of whether the transfer is made in the initial fresh or subsequent frozen embryo transfer cycles. This approach of measuring

ART success also reflects contemporary clinical practice and encourages safe embryo transfer practices (Chambers, et al., 2017, Doody, 2014).

4.4.3 Economic evaluation

ART is an expensive intervention (approximately 11,000 Australian dollars (AUD) per fresh cycle) and is characterised by a wide variation in the level of public and third-party reimbursement of treatment costs worldwide (Chambers, et al., 2009). The cost per live birth can be estimated by relating the cost of treatment to the delivery of a live birth.

Previous estimates have reported that the cost per live birth in Australia ranges from AUD

13,951 to AUD 27,000, with the cost increasing significantly with maternal age and the number of failed treatment cycles (Griffiths, et al., 2010).

As clinical and laboratory techniques for ART continue to evolve and improve, new and often more costly treatment options will increasingly be introduced into the clinical setting. However, with increasing pressure to meet higher demand for medical services with constraints in healthcare budgeting, cost-effectiveness analysis of medical interventions has become a necessary complement to inform stakeholders about whether the intervention is good value for money (Drummond, et al., 2015).

The most common unit of measurement in the health economic assessment is the quality- adjusted life-year (QALY), which combines the quality of life and the length of time spent in a particular health state into a single measure. QALYs provide a convenient metric in the cost-effectiveness analysis of interventions across diseases/conditions and thus support choices for resource allocation within an overall healthcare budget. As a consequence, QALY is the preferred outcome measure for most government bodies and authorities that require economic assessment of interventions prior to considering funding via public resources (Neumann, 2011).

However, the use of the QALY to quantify the benefits of fertility treatment has been contentious (McGregor and Caro, 2006). QALYs are designed to capture gains in the quality and quantity of life of those living, whereas fertility treatment is primarily judged by the ability to create new life. There is also the challenge of incorporating foetal events, such as stillbirths and miscarriages, into QALYs, because QALYs assume life commences at birth (Devlin and Parkin, 2003). Furthermore, reproductive medicine measures other significant non-health benefits that are not captured adequately by

QALYs. These include family formation, life meaning, and even providing a sense of closure for infertile couples having attempted all reasonable approaches to conception

(Ryan, 1999). Finally, as there are at least three potential patients involved in the ART treatment process, infertile woman or man and the unborn child, it raises the question of whose QALYs should be considered: the mother, father or the baby?

4.5 Measuring success rate in older women

An essential part of the management of older women is to provide them with realistic expectations of their chances of success with ART treatment so that they can make a well- informed decision about initiating treatment, treatment choice and continuing treatment

(De Neubourg, et al., 2015).

This is important because compared to younger women, success rates following ART decline rapidly with maternal age. Furthermore, as the rate of aneuploidy increases with maternal age, older women are more likely to experience a cancelled cycle from poor ovarian response, have no euploid embryos (i.e. chromosomally normally embryos) available for transfer, and are more likely to discontinue treatment before a pregnancy is achieved. Undertaking additional treatment without a sufficient incremental chance of success can cause undue emotional distress and financial cost to both the patients and taxpayers (Klipstein, et al., 2005, Verhaak, et al., 2010)

CLBR over a number of complete cycles has been suggested as an appropriate measure of success in older women. The longitudinal analysis of outcomes provides the advantage of assessing the woman/patient’s total reproductive potential on a per complete cycle basis and over the entire course of the treatment (Maheshwari, et al., 2015).

4.6 Summary

This chapter describes the advantages and disadvantages of common approaches to measuring ART success. Because of the complexity of ART treatment and the different perspectives of stakeholders, there is no single measure that can adequately capture treatment effectiveness and safety and assessed the cost-effectiveness of treatment at the same time. However, it is important that the measures chosen to evaluate ART success

93 should reflect the questions being asked by the different stakeholder and by whom to determine the best metric of ART success to employ.

However, with rapidly changing clinical and laboratory practices (e.g. increasing use of

PGD-A and oocyte/embryo cryopreservation), and the importance of presenting the success rate from a patient’s perspective, CLBR over multiple complete cycles, along with a measure of safety (e.g. multiple birth rate), is arguably the most appropriate measure of ART success to facilitate a well-informed decision about initiating treatment, treatment choice and continuing treatment. The cohort study reported in Chapter 7 illustrates the multiple measures of ART success, including CLBR over multiple complete cycles and multiple birth rates following PGD-A or morphological assessment for selection of embryos in ART-naive women aged 37 or over.

Chapter 5

Economics and ART

Chapter 5- Economics and ART

This chapter begins with an introduction to the study of health economics, followed by an overview of the economic evaluation framework and how it is used to inform the rational allocation of resources. Throughout the chapter, the economic concepts that relate to assisted reproductive technologies (ART) are discussed to provide context for the economic studies conducted in this doctoral research program.

5.1 Introductoin

Economics is based on the premise of scarcity. In healthcare, there exists unlimited demand for improved health, but limited public resources to meet these demands. This means that choices have to be made on who receives the treatment or intervention

(McIntosh and Luengo-Fernandez, 2006). To maximise allocative efficiency, the treatment or intervention that yields the greatest health gains for society per unit of resource is deemed to be the most cost effective. Furthermore, as heath budgets are generally fixed, spending on one healthcare intervention results in potential benefits foregone for an intervention not funded (opportunity cost) (Buchanan, 1991). These concepts of scarcity, opportunity cost and allocative efficiency are fundamental to the study of health economics and underpin the guiding principles used to inform healthcare resource allocation decision-making.

5.2 Economic evaluation in healthcare

Economic evaluation in healthcare is defined as the comparative analysis of the costs and outcomes of alternate treatments or interventions. The aim of economic evaluation is to inform and guide decisions for the optimal allocation of scarce healthcare resources so that societal health gain and wellbeing are maximised, while opportunity costs are minimised (Drummond, et al., 2015). 96

Therefore, in economic evaluation, two major components must be present: first, information on both the costs and outcomes of treament or intervention and, second, comparison with an alternative treatment or intervention to assess the relative opportunity costs and benefits to maximise the health gained per unit of resource used (Drummond, et al., 2015).

This formal economic evaluation approaches, also known as health technology assessment (HTA) are increasingly used by government agencies worldwide to make resource allocation decisions. These include, the Pharmaceutical Benefits Scheme (PBS) and the Medical Services Advisory Committee (MSAC) in Australia, the National

Institute for Health and Clinical Excellence (NICE) in the United Kingdom (UK)

(Claxton, et al., 2002) and the Canadian Agency for Drugs and Technologies in Health in

Canada (CADTH, 2006).

However, economic evaluation of fertility treatments such as ART is challenging for health economists and decision makers alike. Despite the relative maturity and widespread use of HTA frameworks, the benefits afforded from fertility treatments are not well captured or quantified by traditional economic evaluation methods. This is principally because fertility treatments are judged by their ability to generate life, whereas almost all other forms of medical care are judged by their ability to save, extend or improve the quality of existing life (Devlin and Parkin, 2003, Klitzman, 2017). The challenges of conducting economic evaluation of fertility treatments are briefly discussed in Section 5.2.2d.

5.2.1 Information on cost for economic evaluations

Economic evaluation is a study that compares the costs and outcomes of alternative treatment or intervention. Therefore, when conducting an economic evaluation, it is important to show how costs and outcomes are measured and valued. A costing process for economic evaluation typically involves three steps: (a) identifying, (b) measuring, and

a) Identifying costs

Cost in health economic evaluations represents the resource consumption required, or a consequent to the treatment or intervention (McIntosh, 2010). Broadly, the range of costs can be categorised into direct, indirect and intangible costs.

Direct costs are those consumed by the treatment or intervention and can be borne either by the health system and/or patients. If they are borne by the latter, they are generally referred to as out-of-pocket (OOP) expenses. For example, direct health costs in fertility treatments such as ART include resources use in the delivery of ART services such as consultations, hormones and medications, and laboratory, embryology and counselling services.

Indirect costs include those not directly related to the intervention, and include productivity losses from hours of lost work, and under some definitions extend to the consequences of illness (e.g. morbidity or mortality costs). In the context of ART, the most significant indirect cost relates to the care of multiple gestation pregnancies during the perinatal period and beyond. As ART treatment poses an increased risk of multiple gestation pregnancy that is largely due to the transfer of more than one embryo during

98 treatment, this results in substantial indirect health costs of caring for both infants and their mothers (Connolly, et al., 2010).

Intangible costs represent the psychological cost of pain or reduction in quality of life that patients experience from a treatment or illness (e.g. infertility). Intangible costs are more difficult to quantify and allocate a monetary value to it (Tarricone, 2006). As there is no opportunity cost incurred, many economists do not consider intangible cost as cost in the strictest sense. However intangible cost can be captured on the outcome side in an economic evaluation as part of the quality-adjusted-life years (QALY).

The decision on the type of cost included in an economic evaluation is often driven by the perspective of the economic evaluation (Simoens, 2009). For example, an economic evaluation undertaken from a healthcare perspective would consider the direct costs associated with the resources consumed in the delivery of the intervention, whereas a societal perspective would include direct, indirect and intangible costs. The different perspectives of study are discussed in section 5.4.1.

b) Measuring costs

After the costs relevant to the study have been collected, the next step is to measure the resource use associated with the intervention so that the unit cost can be assigned to each unit of resource consumption (Chapko, et al., 2009, McIntosh, 2010). There are two approaches that are generally used to measure resource consumption: (a) the top-down approach (macro-costing) which uses estimate of resource use at the aggregate level divided by the number of items of interest (e.g. number of inpatient days), (b) the bottom- up approach (micro-costing) identifies and measures all the individual components of a treatment and sums them to obtain the total cost of treatment or intervention.

The ‘bottom-up’ approach is considered the gold standard in healthcare economic evaluation because all relevant cost components associated with the intervention are identified and valued at the most detailed level (Oostenbrink, et al., 2002). This allows for the identification of the distribution of costs among patients and resource items. This approach thus provides a better insight into patient subgroups and the range of cost items that might potentially drive the results of an economic evaluation. However, the ‘bottom- up’ approach is time consuming, especially when resource use and unit costs are unavailable or inaccurately recorded (Jacobs and Barnett, 2017). Chapter 8 presents a cost-effectiveness study of preimplantation genetic diagnosis for aneuploidy (PGD-A) and illustrates the use of the ‘bottom-up’ costing approach in identifying and collecting costs associated with ART treatment among women who commenced ART treatment with PGD-A and those who used morphological assessment of embryos alone.

c) Valuing costs

The final step is to quantify the category of costs by multiplying each unit cost by the volume consumed, and then tallying across all categories (McIntosh, 2010). For example, in the cost-effectiveness analysis reported in Chapter 8, the number and types of treatment cycles (e.g. cancelled cycles, initiated fresh cycle with and without an embryo transfer) undertaken for each study group were multiplied by the corresponding unit cost of each cycle type, and then summed across all treatment cycles undertaken for the two study groups over the study period.

The costing processes in economic evaluation that involve identifying, measuring and valuing the resources utilised in a treatment or intervention are similar for all economic evaluation methods. However, economic evaluation methods are distinguished from each other based on the way that outcomes are measured and valued (Drummond, et al., 2015).

100

The following section discusses five economic evaluation methods used in healthcare, based on the definition of outcomes used.

5.2.2 Methods of economic evaluation in healthcare

The choice of economic evaluation method primarily depends on the available information on outcomes of the interventions to be evaluated. The following section describes five main economic evaluation methods: cost minimisation analysis (CMA), cost–consequence analysis (CCA), cost–effectiveness analysis (CEA), cost-utility analysis (CUA) and cost-benefit analysis (CBA) (Briggs and O'Brien, 2001, Drummond, et al., 2015, Lorgelly, et al., 2010)

a) Cost minimisation analysis

CMA is used when the outcome (e.g. proportion of live-births) of competing alternatives have been unambiguously demonstrated to be clinically equivalent. This is the case where clinical studies have ruled out significant clinical differences between the new and existing treatments, and the least costly treatment option is preferred.

However, CMA is rarely undertaken as there is always a level of uncertainty surrounding estimates of treatment effects. When the level of uncertainty in treatment effect are combined with the differences and uncertainty in treatment costs, a more efficient intervention may still be identified (Alderson and Chalmers, 2003, Briggs and O'Brien,

2001). For this reason, most modern practitioners of heath economics do not consider

CMA as an economic evaluation method for informing resource allocation decision- making (Briggs and O'Brien, 2001)

101

b) Cost–consequence analysis

CCA is a form of economic evaluation where disaggregated costs and a range of outcomes are presented in a balance-sheet format to allow decision makers to choose which resource, costs and health outcomes are relevant in their decision-making context

(Mauskopf, et al., 1998). This approach allows policy makers to choose the costs and outcomes that are important to them; however, it lacks an explicit decision-making framework and hence, hinders its usefulness and applicability (Lorgelly, et al., 2010).

c) Cost-effectiveness analysis

In CEA, the outcome of an intervention is measured in natural units, e.g. live births for fertility treatment (Culyer, 2015, Drummond, et al., 2015). Although CEA is a valuable tool for evaluating competing interventions that report the same outcome, this technique cannot compare the cost-effectiveness of interventions that have different clinical outcomes (e.g. fertility treatment versus vaccinations). As benchmarks do not exist for what society considers value for money for health outcomes other than QALYS (see next section), operationalising cost-effectiveness analysis for informing resource allocation questions is challenging. The cost-effectiveness analyses undertaken in this doctoral research program (Chapters 8 and 9) uses incremental cost per live-birth as the outcome measure and provides a discussion on how this metric is used and interpreted.

d) Cost utility analysis

CUA is an extension of a CEA that measures the quality and duration of health produced or forgone by an intervention. The most common outcome used in CUA is the incremental cost per QALY gained (Culyer, 2015).

102

The use of QALY incorporates two health dimensions –the health-related quality of life

(typically expressed as a health state utility weight ranging between 0, which is the worst state, and 1 which is the best state) combined with the length of time in a particular state to estimate the change in QALYs associated with the intervention (Edlin, et al., 2015).

In contrast to CEA, CUA provides a generic measure for healthcare intervention that enables comparison of cost and outcome to be made across different diseases and interventions. Therefore, CUA is generally recommended by national funding bodies such as the Pharmaceutical Benefits Advisory Committee (PBAC) in Australia (Henry, et al., 2005) and NICE in the UK (NICE, 2013).

However, in the context of valuing ART treatment, the application of QALYs as an outcome measure is challenging, because QALYs are designed to value existing life and not an unborn child, or a child not yet conceived. QALYs also do not capture events such as stillbirths and miscarriages (Devlin and Parkin, 2003). Furthermore, even if QALYs might have place in valuing ART treatment there are at least three potential stakeholders that might warrant their quality of life being valued, infertile women and men as well as the unborn child or children (Chambers, et al., 2013, Devlin and Parkin, 2003).

Despite the difficulty with the QALY framework as a measure of outcome in fertility treatment, NICE and an earlier study by Scotland and colleagues adopted CUA to estimate the QALY gained from ART treatment. However, the QALY values gained was for infertile women (NICE, 2013, Scotland, et al., 2011)

e) Cost-benefit analysis

CBA expresses the benefits of the intervention in monetary terms, i.e. a dollar value is placed on the life-years gained or even a live-born baby. For this reason, CBA is considered to be the broadest form of economic evaluation as it allows comparisons with

103 other interventions within the healthcare systems, as well as other sectors of the economy.

The ratio of monetary benefits to overall costs determines whether the monetary benefit produced by an intervention is worth the costs, with the intervention said to be cost- beneficial if the benefits exceed the costs (Morris, et al., 2007)

The CBA method uses several techniques to estimate the value of an intervention. The value can be expressed by a stakeholder’s (usually society’s) willingness to pay (WTP) for the outcome. The WTP is considered the ‘shadow price’ for the outcome and this estimate can be determined using either a revealed preference approach, or a stated preference approach (Lancsar and Louviere, 2008).

The revealed preference approach elicits an individual’s WTP for goods or services by observing the actual choices people make when health risks are traded-off against money e.g. higher salaries for riskier jobs (Fujiwara and Campbell, 2011). The stated preference approach explicitly asks individuals to state how much they would be willing to pay to benefit from a treatment or intervention. There are a number of techniques to obtain the stated preference WTP values. This includes using hypothetical survey to ask the respondents directly of their willingness to pay values [contingent valuation (CV)] or ask respondents to choose between two or more hypothetical treatment options with varying attributes or characteristics (Ryan, 2004) If one of the attributes is cost, the marginal willingness to pay for different attributes of a treatment have a whole can be ascertained.

The earliest application of the CV method in ART was undertaken by Neumann and

Johannnesson (Neumann and Johannesson, 1994) who explored survey respondents’

WTP for IVF treatment in the event they were infertile. The study found that the implied

WTP for a baby was 177,730 US dollars (USD) for potential child bearers and USD 1.8 million for society to pay for insurance to allow couples access to ART. Since this study,

104 a small number of other studies have been conducted to elicit the monetary value of ART treatment (Landfeldt, et al., 2012, Poder, et al., 2014, Ryan, 1996). For example, a survey of 294 infertile women undertaking fertility treatment found that respondents placed a high value on the characteristic of ovarian stimulation dosage and were willing to pay 530

Euro (95% confidence interval €500- €570) (from the current cost of €200) to reduce the dose variability from 10–20% to 1–2% (Landfeldt, et al., 2012)

Although there is growing recognition and use of the stated preference methods in outcome measurements, there remains methodological and practical challenges that might limit the validity of its application (Ryan and Farrar, 2000). For example, a review of observational studies found that the dollar estimates for the same commodity vary considerably with study design and survey formats (Venkatachalam, 2004). There was concern that the biased estimate associated with hypothetical validity (i.e. respondent is unable to adequately value the treatment that they have not yet experienced) and income effect (i.e. WTP is strongly associated with ability to pay) might unduly affects outcome estimates. Furthermore, the DCE method can be cognitively demanding for respondents, which might also affect study outcomes (Lancsar and Louviere, 2008).

An alternative approach to quantify the economic value of ART is to measure the net lifetime tax contribution to society of an ART-conceived child. For example, Svensson and colleagues (Svensson, et al., 2008) applied the generational accounting method to quantify the fiscal impact of an ART-conceived child on society and reported a net return of 24% on government investment in ART, suggesting a strong economic basis for publicly subsidising ART treatment.

105

5.2.3 Analysis and interpretation of economic evaluation results

Incremental cost-effectiveness ratio

In CEA and CUA methods, both additional gains (e.g. additional live-births or QALYs) and additional costs from alternate treatments or interventions are summarised and divided to generate an incremental cost-effectiveness ratio (ICER). Specifically, the denominator of the ICER reflects the additional gain in outcome from the alternate intervention and the numerator reflects the additional cost of obtaining the additional gain

(Drummond, et al., 2015). This definition of cost depends on the decision maker’s perspective i.e. societal, healthcare, or patient perspective as discussed in Section 5.4.1.

The ICER is calculated as follows, where A refers to the alternate treatment or intervention and B is the comparator:

Cost − Cost ICER = A B Effectiveness A − EffectivenessB

The ICER is compared to a cost-effectiveness threshold value (or ceiling ratio) to determine whether the alternate intervention represents ‘value for money’ relative to the comparator. In other words, the threshold value serves as a benchmark or ‘shadow price’ that distinguishes health interventions that come at an ‘acceptable’ additional cost for a unit of outcome (e.g. live-birth) from those that are excessively costly (Neumann, et al.,

2016).

There is no cost-effectiveness threshold value (or range of values) that represents society’s (or the government’s) WTP for an additional live-birth in Australia or elsewhere. However, in Australia a general level of 50,000 Australian dollars (AUD) per

QALY is assumed based on submission to the PBAC and funding decisions (Harris, et al., 2008, Shiroiwa, et al., 2010). In the UK, the NICE uses a threshold of £20 000 to

106

£30 000 per QALY (Culyer, et al., 2007). If the ICER falls below this threshold, it is considered cost–effective and likely to be eligible for public funding. However, other factors such as social values, severity of the condition and budgetary impact, alongside safety and clinical evidence play a significant role in influencing funding decisions

(Devlin and Parkin, 2003, Rawlins and Culyer, 2004).

In this thesis, the ICER for an additional live birth from PGD-A compared to morphological assessment of embryos alone (Chapter 8) and other competing alternative

ART strategies (Chapter 9) are reported. Whether or not the additional cost for a live birth represents value for money is difficult to determine because it depends on society’s willingness to pay for a live birth from fertility treatments. This challenge is addressed in greater detail in Chapters 8 and 9.

5.3 Model based economic evaluation

Decision-analytic models such as the state transition Markov model provides a useful framework for conducting economic evaluations, particularly when modelling conditions involves events that may occur repeatedly over time, as in the case with multiple ART cycles. In a Markov model, patients are at any time in a specific ‘health state’ that indicates their position in their treatment cycle e.g. oocyte retrieval or embryo transfer.

Patient progress along an ART treatment cycle through a series of these health states, each with an associated probability (i.e. transition probability) for that particular treatment strategy. In Markov model, outcomes and transition probability for all stages of a treatment cycle (e.g. cancelled cycle, successful oocyte retrieval, embryo transfer) are obtained from several sources such as randomised controlled trial, National IVF registries and expert opinions. Therefore, Markov modelling allows synthesising a range of evidence regarding the costs and effectiveness of alternate strategies in the structured

107 model and thus provides a ‘natural’ framework and includes all important clinical pathways and outcomes for assessing both the economic and clinical effect. However, as the validity of such decision-analytic models depends on the accuracy of the model structure and data source, models are often criticised for the lack of transparency and replication in real world settings (Cooper, et al., 2007).

Chapter 9 reports the results of a Markov model constructed as part of this doctoral research program to evaluate the cost-effectiveness of standard ART treatment compared to PGD-A, social freezing and donor ART in women of advanced maternal age.

5.3.1 Uncertainty in model based economic evaluations

Similar to other statistical measures, the results of a decision-analytic model are subject to influences of variability, uncertainty and heterogeneity, which that might affect the overall model estimates of measures of cost-effectiveness (Briggs, et al., 2006). The two main types of uncertainty relate to structural uncertainly where the model structure does not represent real-world clinical pathways, and parameter uncertainty where the model inputs are not accurate or relevant to the model structure (Briggs, et al., 2006). Therefore, it is important to assess the extent of uncertainty to reduce the risk of bias. The next section discusses ways of dealing with uncertainty as they were applied in both Chapters

8 and 9.

a) Deterministic sensitivity analysis

The influence of parameter uncertainty can be systematically evaluated using deterministic sensitivity analysis to test the sensitivity of the model results to specific parameters or a set of parameters (Briggs, et al., 2006). For example, in the decision- analytic Markov model presented in Chapter 9, the number of oocyte retrieval procedures was varied from one to two to test the uncertainty around the optimal number of procedure

108 required to generate sufficient oocytes to perform two frozen embryo transfer cycles and determine if changes in the number of oocyte retrieval procedures affected ICER estimates.

Deterministic sensitivity analysis provides the advantage of identifying which parameter value influences the overall results and can also be used to estimate threshold values above or below which one strategy becomes preferred over another (Cohen and Reynolds,

2008). The key limitation of deterministic sensitivity analysis is the assumption that there is no causal relationship between the value taken by one parameter and the other parameter values in the model (Petrou and Gray, 2011).

a) Probabilistic sensitivity analysis

Probabilistic sensitivity analysis (PSA) is stochoastic method used to assess the extent of uncertainty and its influence on the model results (Briggs, et al., 2006). In PSA, the individual model parameters are replaced by their probability distributions to reflect the expected values and the uncertainty around each value is explored. Some probability distributions are more suitable for certain parameters due to their attributes. For example, as cost parameters are always positive and right skewed, a Gamma distribution is usually applied. Similarly, as Beta distribution takes the range between 0 and 1, live birth rates from ART are usually modelled as Beta distributions (Briggs, et al., 2006).

5.4 Other elements in economic evaluation

5.4.1 Perspective of the study

As described previously in this chapter, the perspective used in economic evaluations determines which costs and benefits are included. A societal approach is often preferred as all effects (outcomes) and costs associated with the use of the treatment or intervention

109 are incorporated into the analysis regardless of who pays the costs or who receives the benefits (Morris, et al., 2007).

In this thesis, both a healthcare and patient perspective were undertaken for all analyses.

A healthcare perspective includes all costs incurred by the healthcare system and patient’s

OOP expenses. However, a previous review of 85 published economic evaluation studies of ART found that slightly more than half of the economic studies (58%) reported from a partial healthcare perspective, with few studies (2%) included OOP expenses in their analysis (Moolenaar, et al., 2013). This approach fails to consider the financial barrier created by the high OOP expenses associated with ART treatment.

5.4.2 Discounting cost and outcomes

Discounting is an economic method that captures an individual's preference to consume good or services now rather than in the future (Jo, 2014). It reveals an individual’s rate of time preference i.e. how much they prefer current consumption of resource over the same amount of consumption in the future (Claxton, et al., 2011). In ART, the discount rate is not usually applied due to the short time horizon for fertility treatment as a baby is conceived within one year of treatment.

5.5 Summary

This chapter summarised how health economic evaluations provide methods to synthesis clinical outcomes and costs to derive a metric, an ICER, to inform the rational investment in healthcare treatments or interventions whilst incorporating the impact of uncertainty.

As clinical and laboratory techniques for ART continue to evolve and improve, clinicians, funders and infertile couples are increasingly presented with treatment options that are often more-costly than those that are conventionally available. Therefore, economic evaluations provide a useful tool for policy-makers to balance the trade-off between costs 110 and outcomes to inform decisions around the allocation of scarce healthcare resources

(Drummond, et al., 2015).

Unlike other areas of healthcare intervention, ART presents unique challenges that make the interpretation of economic evaluation challenging and complex. The usual metric used in economic evaluations, QALYs, are not well suited to the evaluation of fertility treatment which creates life, rather than saving or extending existing life. This inherent difficulty in valuing fertility treatment alongside other medical interventions is perhaps one of the reasons ART treatment is subject to such variation in public funding arrangement throughout the world (Chambers, et al., 2013, Connolly, et al., 2010).

Furthermore, from a patient perspective, ART is a costly intervention without government subsidy. For most patients, treatment affordability is an important consideration

(Chambers, et al., 2014). Economic evaluation methods provide a useful tool to rationally evaluate the cost and effective between treatment options to inform decision makers whether there is good value in the alternative treatment option. However, a consensus view on what constitutes ‘good–value’ in ART treatment would need to be a priority for future research if management of infertility is to become an integral part of public and private funded healthcare systems.

111

Chapter 6

The clinical effectiveness of preimplantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A): systematic review (updated)

112

Chapter 6- The clinical effectiveness of pre-implantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A): systematic review (updated)

6.1 Introduction

As part of the doctoral research programme, a systematic review of the clinical effectiveness of preimplantation genetic diagnosis for aneuploidy (PGD-A) in all 24 chromosomes was conducted (Lee, et al., 2014). The review identified 19 relevant studies

[three randomised controlled trials (RCTs) and 16 observational studies] and concluded that preimplantation genetic diagnosis for aneuploidy (PGD-A) improves implantation and clinical pregnancy rates in young and good prognosis women. However, there remains a paucity of high quality evidence to support the clinical effectiveness of PGD-

A in older women. The findings of this review have been published in Human

Reproduction (second highest ranked journal in Reproductive Biology) in 2014 and was featured as one of the most cited articles (2017 Release of journal citation reports®). The abstract of the published systematic review is presented below, and the full manuscript can be found in the Appendix 1.

113

114

Since the original systematic review, two systematic reviews and meta-analyses have been published by Chen and colleagues (Chen, et al., 2015) and Dahdouh and colleagues

(Dahdouh, et al., 2015). The two published systematic reviews and meta-analyses included studies published up until 2015 and their conclusions were broadly consistent with the original systematic review (Lee, et al., 2014). This section provides an update of studies on PGD-A that were published after the two published systematic reviews and meta-analyses (Chen, et al., 2015, Dahdouh, et al., 2015). In this review, a meta-analysis was not performed because of the clinical heterogeneity between the studies such as the population age groups (young and older women), types of biopsy (blastomere or trophectoderm biopsy) and study design (RCT and observational studies).

6.2 Methods a) Search Strategy and Study Selection

The search strategy and study selection for this updated review is similar to the approach used in the first review (Lee, et al., 2014). A computer assisted search to identify relevant articles was performed using the OVID interface to MEDLINE, EMBASE, SCOPUS,

Cochrane Library, NHS Economic Evaluation Database and EconLit. The search was limited to studies of humans, English language text and covered all eligible articles published from 2014 to 2017. The comprehensive PGD-A techniques included in the review, refers to CGH (array and metaphase), single nucleotide polymorphism arrays and real time quantitative polymerase chain reaction (qPCR).

115 b) Assessment of Study Quality

Critical appraisal of the articles was conducted using the Preferred Reporting Items for

Systematic Reviews and Meta–analyses (PRISMA) Statement (Moher, et al., 2015).

Additionally, the methodological quality of the studies that met the inclusion criteria were assessed using a modified version of Downs and Black checklist which is considered valid and reliable for assessing randomised and nonrandomised studies of healthcare interventions and covers reporting, external validity, internal validity (bias and confounding) and statistical power (Downs and Black, 1998). Consistent with the previous systematic review, a modified Downs and Black checklist was developed using

22 of the 27 items where each item in the checklist was scored either 0 or 1, with a score of less than 8 considered a poor-quality study, a score of 8-15 a moderate quality study, and a score of more than 15 a high -quality study (Lee, et al., 2014). The Downs and

Black checklist can be found in Appendix 3.

6.3 Results

Figure 5 outlines the results of the search strategy which identified seven studies meeting the inclusion criteria for this systematic review. Studies were excluded if they were designed primarily to validate technical aspects of comprehensive PGD-A, used predominately donated oocytes or lacked clinical outcome measures. Details of each study are included in Supplementary Table 1.

116

Figure 5:Search flow diagram based on PRISMA

117

In this updated systematic review, one RCT (Rubio, et al., 2017) and six observational studies were identified (Chang, et al., 2016, Greco, et al., 2014, Kushnir, et al., 2016, Lee, et al., 2015, Ubaldi, et al., 2015, Ubaldi, et al., 2017). Of these studies, six studies included a control group where embryos were selected based on morphology or have not used

PGD-A (Chang, et al., 2016, Greco, et al., 2014, Kushnir, et al., 2016, Lee, et al., 2015,

Rubio, et al., 2017, Ubaldi, et al., 2015). All the observational studies were cohort studies and the RCT by Rubio and colleagues reported outcomes from the first ‘complete ART cycle’ i.e. fresh and any subsequent frozen/thawed embryos originating from the oocyte pick-up (OPU) (Rubio, et al., 2017). a) Study characteristic

Supplementary Table 1 summarises the characteristics and outcomes of the seven eligible studies. The updated review includes studies that were published between 2014 and 2017. There was little geographical spread in the studies, with four studies performed in Europe (Italy and Spain) (Greco, et al., 2014, Rubio, et al., 2017, Ubaldi, et al., 2015,

Ubaldi, et al., 2017) and the remaining three studies in the United States (US) (Chang, et al., 2016, Kushnir, et al., 2016, Lee, et al., 2015).

Four studies reported the number of study participants undertaking PGD-A (Greco, et al.,

2014, Lee, et al., 2015, Rubio, et al., 2017, Ubaldi, et al., 2017) and the remaining three studies reported based on the number of cycles started or embryo transfer with PGD-A

(Chang, et al., 2016, Kushnir, et al., 2016, Ubaldi, et al., 2015). In this review, the number of study participants undertaking PGD-A across the studies ranged from 43 to 170 patients and the number of PGD-A cycles ranged from 328 to 5471 cycles.

Overall, the mean age of study patients in the review ranged from 32.8 years to 44.7 years.

Most studies (n=5) reported outcomes of patients of advanced maternal age (AMA) or

118 patients with a poor prognosis, i.e. recurrent implantation failure (Greco, et al., 2014, Lee, et al., 2015, Rubio, et al., 2017, Ubaldi, et al., 2015, Ubaldi, et al., 2017). Two population- based studies stratified the national data into three age categories (less than <35 years, aged between 35 and 37 years and more than 37 years) (Chang, et al., 2016, Kushnir, et al., 2016)

Four studies assessed trophectoderm biopsy (Greco, et al., 2014, Lee, et al., 2015, Ubaldi, et al., 2015, Ubaldi, et al., 2017) and only one study assessed blastomere biopsy (Rubio, et al., 2017). Two studies reported outcomes based on fresh transfers after PGD-A

(Chang, et al., 2016, Kushnir, et al., 2016), three studies reported both fresh and FET cycles (Greco, et al., 2014, Lee, et al., 2015, Rubio, et al., 2017) and the remaining two studies reported outcomes of FET (Ubaldi, et al., 2015, Ubaldi, et al., 2017). All but one of the seven studies reported on live birth or delivery rates (Chang, et al., 2016, Kushnir, et al., 2016, Lee, et al., 2015, Rubio, et al., 2017, Ubaldi, et al., 2015, Ubaldi, et al., 2017).

Four studies used the number of initiated cycles as the denominator in line with intention- to-treat principles (Kushnir, et al., 2016, Rubio, et al., 2017, Ubaldi, et al., 2015, Ubaldi, et al., 2017). b) Quality evaluation:

The Downs and Black scores for the nonrandomised studies ranged from 11 to 16 (mean:

12) indicating that the studies were of moderate quality whilst the single RCT scored 18 indicating it was of high quality. The main weaknesses of the seven included studies included lack of randomisation, reporting outcomes based on per embryo transfer, failure to include the outcome of subsequent FET cycles in order to calculate cumulative live birth rates (CLBR) and failure to control for confounding variables either within the study

119 design or within the analysis (e.g. variable number of embryos transferred between case and control groups). c) Clinical outcomes for PGD-A on women of advanced maternal age

A single RCT and another five observational studies examined the use of PGD-A in women of AMA (mean age >35 years old). The results of the studies (one RCT and four observational studies) which included a control group where embryos were selected based on morphology or have not used PGD-A, reported that PGD-A leads to a higher live delivery or birth rate in older women (Lee, et al., 2015, Rubio, et al., 2017, Ubaldi, et al.,

2015). An uncontrolled observational study assessing the use of PGD-A in women aged

44 years or over reported that female age and number of metaphase II (MII) oocytes collected were highly correlated with the chance of achieving a delivery (Ubaldi, et al.,

2017).

Among the included studies which reported a higher delivery or live-birth rate after PGD-

A in older women, two studies used the 2011-2012 Centres for Disease Control (CDC) population-based data to compare the clinical outcomes of fresh non-donor ART cycles with PGD-A and cycles that did not report the use of PGD-A (i.e. non PGD-A cycles)

(Chang, et al., 2016, Kushnir, et al., 2016). Chang and colleagues reported a higher live- birth delivery rate per embryo transfer after PGD-A compared to cycles without reported use of PGD-A (Chang, et al., 2016). In response to Chang and colleagues’ findings,

Kushnir and colleagues reanalysed the data but reported outcomes based on per cycle started. Kushnir and colleagues reported that PGD-A leads to a higher per-cycle started live-birth rate of 17.7% compared to 12.7% in cycles that did not report the use of PGD=A in women aged over 37 years. Although both studies showed similar outcome in terms of a higher live-birth delivery rate in older women, Kushnir and colleagues commented on

120 the higher proportion of women aged over 37 years whose cycles reached embryo transfer after PGD-A compared to the proportion of non PGD-A cycles that reached embryo transfer (CDC population-based dataset) suggests favourable selection of good prognosis patients for PGD-A. Therefore, the reporting of outcomes based on per embryo transfer in Chang and colleagues’ study, is likely to be biased and misleading (Kushnir, et al.,

2016).

d) Clinical outcomes of PGD-A in young women

Three observational studies compared the clinical outcomes of PGD-A in young women aged < 35 years (Chang, et al., 2016, Greco, et al., 2014, Kushnir, et al., 2016). Two studies used the 2011 and 2012 Centres for Disease Control (CDC) population-based data to compare clinical outcomes between fresh non-donor PGD-A cycles and non PGD-A cycles in patients aged <35 years. Both studies reported a lower live-birth delivery rate based on per transfer and per started cycle after PGD-A compared to non PGD-A cycles

(Chang, et al., 2016, Kushnir, et al., 2016).

Greco and colleagues assessed the clinical outcomes of single blastocyst transfer with

PGD-A (PGD-A group) and blastocyst transfer of one or two morphological-assessed embryos (control group) in young women (mean age 32.8 years) with repeated implantation failure (mean = 4.9 failures). The study included another control group of good prognosis patient (ART naïve women without any detected problem of ovarian reserve and uterine receptivity or sperm quantity or quality) for single blastocyst transfer with PGD-A. This study found a higher implantation rate per transfer (presence of a gestational sac) and clinical pregnancy rate per transfer (presence of an intrauterine gestational sac at seven weeks of gestation) after single blastocyst transfer with PGD-A

121 compared to using morphological assessment of embryos alone in young women with recurrent implantation failure (Greco, et al., 2014)

e) Effect of cell biopsy stage:

Four studies revealed a significant benefit associated with trophectoderm biopsy at the blastocyst stage with PGD-A (Greco, et al., 2014, Lee, et al., 2015, Ubaldi, et al., 2015,

Ubaldi, et al., 2017) and one study used cleavage-stage blastomere biopsy and reported favourable outcomes in terms of a higher implantation rate, lower miscarriage rate and a higher live-birth rate per patient started after day-3 blastomere biopsy (Rubio, et al.,

2017). Two population-based observational studies did not provide information on whether blastomere or trophectoderm biopsy was performed on the PGD-A cycles

(Chang, et al., 2016, Kushnir, et al., 2016). All studies reporting on trophoectoderm biopsy showed an increase in the ongoing/clinical pregnancy rate although significant differences between patient characteristics and clinical practices were not adjusted for, such as the quality of embryos or the number of embryos transferred. f) Implantation rate

Three studies including the RCT, compared the implantation rate between PGD-A and morphological assessment of embryos alone (Greco, et al., 2014, Lee, et al., 2015, Rubio, et al., 2017). Although the three studies varied in study design, biopsy stage, type of PGD-

A technique, age and prognosis of patients, all three studies reported a higher implantation rate after PGD-A compared to morphological assessment of embryos alone.

122 g) Pregnancy Loss

All seven studies reported on pregnancy loss after PGD-A (variously defined as miscarriage and spontaneous abortion) as an outcome measure (Chang, et al., 2016,

Greco, et al., 2014, Kushnir, et al., 2016, Lee, et al., 2015, Rubio, et al., 2017, Ubaldi, et al., 2015, Ubaldi, et al., 2017). The rates of pregnancy loss after PGD-A reported in the studies ranged from 0% to 16.8%. All three observational studies which reported pregnancy loss found no statistical difference in the rates of pregnancy loss after PGD-A in younger women (< 35 years). Five studies (one RCT and four observational studies), all of which included a control group of patients, reported a significantly lower miscarriage rates after PGD-A in older women (>35 years) (Lee, et al., 2015, Rubio, et al., 2017, Ubaldi, et al., 2015) h) Frozen embryo transfer.

The introduction of modern vitrification techniques offers a more efficient method to cryopreserve euploid embryos for subsequent FET. Four studies reported the outcomes of FET cycles after PGD-A (Greco, et al., 2014, Lee, et al., 2015, Rubio, et al., 2017,

Ubaldi, et al., 2017). One study reported that FET cycles with euploid embryos resulted in a higher implantation rate and live-delivery rate per-embryo transferred compared to standard FET cycles using morphologically assessed embryos (Lee, et al., 2015). Two studies that reported CLBR (which takes into account the contribution of both fresh and subsequent FET cycles) found similar live-delivery rates for PGD-A and morphological assessment of embryos alone (Rubio, et al., 2017, Ubaldi, et al., 2015).

123

6.4 Discussion

The strengths of this systematic review include a comprehensive search of the literature and quality appraisal of all included studies. The review found a consistent effect of PGD-

A in improving implantation rate and per-cycle live-birth/delivery rate compared to morphological assessment of embryos alone, specifically for older women (>35 years).

The finding that PGD-A improves clinical outcomes is in line with the results from previous reviews of observational studies and existing meta-analyses assessing PGD-A in older women (Chen, et al., 2015, Dahdouh, et al., 2015, Lee, et al., 2014).

However, most of the older women (>35 years) included in the studies were women with good prognosis which may account for the better outcomes associated with PGD-A

(Chang, et al., 2016, Kushnir, et al., 2016, Lee, et al., 2015, Rubio, et al., 2017). The numbers of poor prognosis women whose cycles were excluded from the analyses are not trivial. For example, in the retrospective cohort study by Lee and colleagues, 36.6% of the PGD-A cycles (n=44) were excluded from the analysis as none of the biopsied embryos in these cycles were euploid (Lee, et al., 2015). Almost 30% of women who initiated an ovarian stimulation cycle in the RCT by Rubio and colleagues which compared PGD-A and morphological assessment of embryos were excluded from the analysis due to low MII oocyte number (Rubio, et al., 2017). It is possible that the process of excluding women of poor prognosis who have a lower chance of achieving a euploid blastocyst from the analysis, may introduce selection bias.

Although the current review found a higher live birth rate in the initial fresh cycle after

PGD-A, the CLBR was comparable for PGD-A and morphological assessment of embryo alone (Rubio, et al., 2017, Ubaldi, et al., 2015). While some may argue for sequential transfer of untested embryos over multiple cycles, there is evidence that PGD-A reduces

124 pregnancy loss, time to pregnancy and the number of ART cycles needed to achieve a live birth compared to morphological assessment of embryo alone (Lee, et al., 2017,

Rubio, et al., 2017). These characteristics of the PGD-A strategy are relevant and important to older women who have a higher risk of miscarriage and limited reproductive time for repeated treatment failures. Furthermore, the recurring cycle of hope and despair from repeated ART cycles in which untested aneuploid embryos are transferred should not be underestimated (Boden and Boden, 2007, Lande, et al., 2015)

However, because of the potential for false positives associated with PGD-A, and the growing recognition of the incidence and nature of chromosomal mosaicism affecting successful outcomes it has been suggested that for older women particularly, among those with mosaic embryos with no euploid embryos, the embryo should be transferred to increase the chance of establishing a pregnancy (Fiorentino, et al., 2014, Maxwell, et al.,

2016, Munne, et al., 2010), This has been demonstrated in recent studies which reported that euploid/diploid mosaic embryos hold the potential to implant and result in the birth of healthy babies (Greco, et al., 2015, Spinella, et al., 2018).

There remains a paucity of studies evaluating the cost-effectiveness of PGD-A. Although

Rubio and colleagues included a cost-outcome analysis conducted alongside the RCT, the incremental cost of a live birth after PGD-A was not evaluated, and therefore it is not considered to be a full cost-effectiveness analysis (Rubio, et al., 2017).

Another study that used a decision-analytic model reported that PGD-A was more cost- effective compared to morphological assessment of embryos in women aged over 37 years who had at least one blastocyst for transfer (Collins, et al., 2017). However, the outcome of the model was based on a single-cycle analysis, which is of limited value as women undertake more than one ART cycle to improve their chance of achieving a live

125 birth (Maheshwari, et al., 2015, Malizia, et al., 2009). Chapter 8 presents the cost- effectiveness analysis of PGD-A in infertile women aged 37 years or over for up to three

‘complete ART cycles’.

The main limitation of the synthesis of the current evidence is the paucity of high-quality published studies on PGD-A. According to the Downs and Black scoring system, in the current review, only the single RCT was scored as high quality, with the remaining studies scored as moderate quality.

6.5 Summary

This updated review assessed the most recently published studies on PGD-A and confirmed that PGD-A is associated with a higher implantation rate and per-cycle started live-birth rate particularly for older women (> 35 years). PGD-A has also been shown to reduce the time-to-pregnancy and the number of ART cycles needed to achieve a live birth. However, current evidence suggests that the CLBR is similar for PGD-A and morphological assessment of embryos alone.

The conflicting evidence on the effectiveness of PGD-A on important outcomes such as a higher per-cycle live birth rate but a comparable CLBR makes the decision of whether to use PGD-A and for which indications e.g. women of advanced maternal age difficult. In addition, chromosomal mosaicism will continue to present more challenges for patients and clinicians as PGD-A is increasingly used to select and transfer embryos in an attempt to improve live-birth rate.

126

Chapter 7

ART cumulative live-birth rates using preimplantation genetic diagnosis to screen for embryo aneuploidy (PGD-A): a cohort analysis

127

Chapter 7-ART cumulative live-birth rates using preimplantation genetic diagnosis to screen for embryo aneuploidy (PGD-A): a cohort analysis

7.1 Introduction

This chapter uses a cohort analysis to assess the clinical effectiveness of PGD-A with morphological assessment of embryos alone in women of advanced maternal age. The primary aim of the analysis was to evaluate the cumulative live-birth rate (CLBR) of

PGD-A. Data were analysed according to intention-to-treat and per-protocol principles.

The CLBR and other clinically important outcomes for PGD-A and morphological assessment of embryos alone are presented in Section 7.5. A summary of the findings concludes the chapter.

7.2 Background

The aim of PGD-A during ART is to improve success rates by selecting and transferring euploid embryos that have a normal number of chromosomes. It is well established that aneuploid embryos, which have an abnormal number of chromosomes, are less likely to implant and more likely to miscarry than euploid embryos (Fragouli and Wells, 2012,

Hassold, et al., 1996, Vialard, et al., 2011, Wilton and Wells, 2013). The rate of aneuploidy increases with maternal age with almost three quarters of oocytes from women aged over 40 years affected by aneuploidy compared to one quarter in women in their early 30s (Fragouli, et al., 2011, Wells and Levy, 2003). Increased aneuploidy rates are also more likely to be found in women who suffer from recurrent miscarriages and repeated implantation failure (Rubio, et al., 2013, Voullaire, et al., 2007).

128

Over the last two decades, considerable research has focused on developing and incorporating diagnostic techniques to enhance the selection of euploid embryos for transfer. Fluorescent in situ hybridization (FISH) analysis of 8−10 chromosomes in biopsied blastomeres was initially used to identify euploid embryos for transfer.

However, several randomised controlled trials (RCTs) failed to show any benefit of FISH, most likely because of the limited number of chromosomes that could be assessed using this approach (Blockeel, et al., 2008, Debrock, et al., 2010, Schoolcraft, et al., 2009). One notable exception to this was the study by Rubio and colleagues (Rubio, et al., 2013) who demonstrated significantly improved outcomes after PGD-A using FISH.

The advent of comprehensive molecular cytogenetic techniques, such as array comparative genomic hybridization (array CGH), single nucleotide polymorphism arrays, and real time quantitative polymerase chain reaction, which allow testing of all 24 chromosomes in embryos have shown to result in higher implantation and clinical pregnancy rates compared to morphological assessment of embryos alone (Chen, et al.,

2015, Dahdouh, et al., 2015, Lee, et al., 2014)

However, whether PGD-A using these advanced techniques improves live birth rate over multiple cycles, particularly in women of advanced maternal age (generally considered to be 37 years or over), is not clear (Lee, et al., 2014; Chen, et al., 2015; Dahdouh, et al.,

2015). Most published studies on PGD-A reported outcomes in terms of per ‘single cycle’ rather than CLBR that account for outcomes over multiple cycles per patient (Fishel, et al., 2011, Keltz, et al., 2013, Schoolcraft, et al., 2010, Sher, et al., 2009). Rubio and colleagues’ study (Rubio, et al., 2017) is the only RCT that reported outcomes from the first ‘complete ART cycle’ i.e. fresh and any subsequent frozen/thawed embryos originating from the oocyte pick-up (OPU). This RCT reported that the CLBR was

129 comparable between PGD-A and morphological assessment of embryo alone in women of advanced maternal age (mean age 39 years) with relatively good prognosis.

CLBRs are particularly relevant in assessing the effectiveness of PGD-A compared to embryo selection by morphology alone, as PGD-A usually results in fewer embryos available for both fresh transfer and cryopreservation (Mastenbroek and Repping, 2014).

Furthermore, counselling patients regarding their chance of a live birth after multiple cycles is arguably a more relevant measure of ART treatment success and encourages a holistic approach to treatment which promotes single embryo transfer and realistic expectations of treatment success (Maheshwari, et al., 2015, McLernon, et al., 2016).

7.3 Aim of the study

The primary aim of this cohort study was to assess the CLBR of PGD-A compared to morphological assessment of embryos alone in women of advanced maternal age (≥37 years old). This analysis used a number of analytical principles and outcome measures including CLBR, average number of cycles and time to reach a pregnancy resulting in a live birth, to provide a more comprehensive assessment of the clinical value of PGD-A in older women.

7.4 Methods

7.4.1 Clinical data source

The study cohort comprised infertile women aged 37 years or over who commenced their first fresh ART cycle in a large private fertility clinic in Melbourne, Australia between

January 2011 and June 2013. The treatment outcomes from their first and subsequent fresh and frozen/thaw cycles were followed up until March 31 2014 or the first clinical pregnancy leading to a live born baby.

130

The study cohort was stratified into women whose first OPU used either PGD-A or morphological assessment for selection of embryos alone (hereafter referred to as ‘PGD-

A group’ and ‘morphology assessment group’ respectively). An initiated fresh cycle was defined as commencing with the administration of follicle stimulating hormone (FSH) and an initiated frozen/thaw cycle was defined as the thawing of one or more embryos.

Exclusion criteria included women who had PGD testing for monogenic disease, chromosomal translocation or rearrangements, or sex selection. Women were also excluded if they used donor oocytes or initiated a cycle with the intent of freezing all embryos for later use.

7.4.2 Embryo development and evaluation a) Ovarian stimulation

Cycles of ovarian stimulation were programmed using the oral contraceptive pill to regulate the number and timing of cases to be processed through the laboratory. Protocols of ovarian hyperstimulation were down regulation with GnRH agonist (Synarel®, Pfizer

Australia Pty Ltd) and recombinant FSH or antagonist protocol with Orgalutran® (Merck

Sharp & Dohme Australia Pty Ltd) and recombinant FSH. Choice of cycle protocol and

FSH dosage was clinician directed. FSH dosage was chosen with regard to anti-Mullerian hormone assay and antral follicle count with a maximum dosage used of 300iu. b) Embryo culture and biopsy

All oocytes in cycles where PGD-A was planned were fertilised with a single sperm using intra-cytoplasmic sperm injection (ICSI). In the morphology assessment group, fertilisation was achieved by co-incubation of up to three oocytes with approximately

100,000 motile sperm, except where ICSI was indicated, usually because of poor sperm quality. Embryos were cultured in Quinn’s Sage Advantage Medium. 131

In the PGD-A group, embryo biopsy was performed on day 3 of embryo development. A single nucleated blastomere was biopsied from all embryos that had at least 6 cells and less than 30% fragmentation on the morning of day 3.

Embryo biopsy was performed by breaching the zona pellucida with a series of laser pulses and removal of a single nucleated cell. c) Array comparative genomic hybridization

Chromosome enumeration in single blastomeres was determined using array CGH

(24Sure, Illumina, Inc., San Diego, CA) according to the manufacturer’s instructions. In brief, blastomeres were loaded into 4l PBS and whole genome amplification was performed using Sureplex (Illumina, Inc.). Amplified DNA was labelled with either Cy3 or Cy5 fluorochrome and applied to a 24sure microarray chip and hybridised overnight.

Known normal male DNA labelled with either Cy3 or Cy5 served as a reference.

Fluorescence ratios, and hence relative copy number of individual chromosomes, were determined using the latest available version of BlueFuse software (Illumina, Inc.). The results of the day 3 biopsy were available on day 5 for fresh transfer of euploid embryo(s) and vitrification of surplus embryos. d) Morphological assessment of embryos

Morphological assessment of cleavage stage embryos was as previously described

(Edgar, et al., 2009). Briefly, in addition to cell number, the degree of cellular fragmentation was evaluated using a grading system with grade 1 (no fragmentation), grade 2 (≤10% fragmentation), grade 3 (11-30% fragmentation), grade 4 (31-50% fragmentation, and grade 5 (> 50% fragmentation).

132 e) Embryo transfer

In the morphology assessment group, embryo transfer was performed predominantly on day 2, but occasionally at later stages (i.e. days 3 or 5). Embryos with the highest morphological score were selected for transfer, which was performed using standard trans-cervical approach. In the PGD-A group embryo transfer was performed predominantly on day 5, but occasionally on day 6, after chromosome enumeration using array CGH, and one or two euploid embryos were transferred. f) Embryo cryopreservation

Supernumerary embryos of good quality (based on cell number and less than 30% fragmentation) in the morphology assessment group and supernumerary euploid embryos in the PGD-A group were frozen using liquid nitrogen and cryoprotectant for possible use in subsequent frozen/thawed embryos transfer (FET) cycles. Embryos in the morphology assessment group were cryopreserved using slow-freeze technique and euploid biopsied embryos in the PGD-A group were cryopreserved by vitrification using a commercial vitrification kit (Cook Medical, Australia) according to the manufacturer’s instructions

(Edgar, et al., 2009).

7.4.3 Definition of outcome measures

The CLBR was defined as the number of live births achieved by women who commenced treatment, i.e. the number of live births divided by the total number of women who started treatment in either the PGD-A group or morphology assessment group.

Cycle outcomes included implantation rate (number of gestational sacs or fetal heartbeats on ultrasound divided by the number of embryos transferred), pregnancy (presence of a heart beat on ultrasound) rate per initiated cycle or per embryo transfer) and live birth

(birth of at least one live born baby) rate per initiated cycle or per embryo transfer 133 procedure. The mean duration of the follow up period was calculated from the date of the first OPU through to 31 March 2014 or the first clinical pregnancy leading to a live born baby. Treatment was considered to be discontinued if women did not achieve a live birth and did not proceed with a subsequent cycle during the study period.

In this study, a ‘single cycle’ referred to the outcomes of a discrete fresh or frozen/thaw cycle and a ‘complete ART cycle’ considered outcomes from all embryos created from an OPU including those from any subsequent frozen/thaw embryo transfers.

7.4.4 Statistical analysis

All data were analysed based on the intention-to-treat and per-protocol principles. The intention-to-treat analysis assigned all outcomes from the first and subsequent treatment cycles to the strategy they received in the first fresh cycle (PGD-A or morphological assessment alone). For example, if embryo selection in a woman’s first cycle was based on morphological assessment (i.e. the morphology assessment group) but PGD-A was used in her second cycle, all results would be allocated to the morphology assessment group. The use of the intention-to-treat principle therefore reflects the actual experience and clinical pathway of women undergoing treatment.

The per-protocol analysis excluded women from the analysis once they crossed over to the alternate treatment strategy during the study period. This analytical approach provided an assessment of effectiveness of PGD-A and morphological assessment without contamination of the results by the alternative strategy.

The demographic and treatment characteristics of the women up to their first three ‘single cycles’ (fresh or frozen/thaw cycles) were compared between the PGD-A group and the morphology assessment group. Cycle outcome measures included number of oocytes collected, proportion of cycles reaching embryo transfer procedures, number of embryos

134 cryopreserved, and the embryo quality. Chi-square or Fisher’s exact test were used for categorical variables and t-test for continuous variables. A p value of less than 0.05 was considered to indicate statistical significance.

Time to clinical pregnancy leading to the first live born baby for each woman was also calculated from the date of their first OPU and compared between the PGD-A group and the morphology assessment group. A Kaplan Meier survival plot was generated with log rank test performed to test for differences in the time to pregnancy resulting in a live birth between the two study groups.

All analyses were conducted using Stata software version 11.2 (Stata Corp, College

Station, TX).

7.4.5 Ethical approval

The study was approved by the University of New South Wales, Human Research Ethics

Advisory (HREA) Panel 1 and the Melbourne IVF Research Ethics Committee.

7.5 Results

7.5.1 Overall ‘single cycle’ success rate

During the study period, the entire study cohort undertook 288 PGD-A cycles where euploid embryos were selected for transfer and 5,771 cycles where morphological assessment was used to select embryos for transfer. The live birth rate per initiated cycle

(fresh or FET) was higher following PGD-A compared to cycles (fresh or FET) using morphological assessment alone (44 live births/288 initiated cycles [15.28%] versus 521 live births/5771 initiated cycles [9.03%] respectively, p<.001). The live birth rate per embryo transfer procedure (fresh or frozen/thaw) was almost three times higher following

PGD-A than morphological assessment alone (44 live births/137 embryo transfer

135 procedures [32.12%] versus 521 live births/4622 embryo transfer procedure [11.27%] respectively, p<.001).

The average follow-up time in the PGD-A group was 503.62 ± 335 days compared to an average of 560.05 ± 338 days in the morphology assessment group (p=0.08). The proportion of women who discontinued treatment during the study period was similar between the two study groups i.e. of the women who did not achieve a live birth, 61.72% of the women in PGD-A group and 55.97% of women in morphology assessment group did not undertake three ‘complete ART cycles’ during the study period (p=0.31).

7.5.2 Intention-to-treat analysis: Cycle characteristics for up to the first 3 ‘single cycles’

Table 2 presents the demographic data on the 2,093 infertile women and the characteristics for up to first 3 ‘single cycles’ (fresh or FET). Of these women, 110 had their embryos assessed using PGD-A in their first fresh cycle (PGD-A group) and another

1983 women had their embryos selected based on morphological assessment alone in their first fresh cycle (morphology assessment group). Not all women undertook three

‘single cycles’ because some had a live birth or discontinued treatment before their third

‘single cycle’. The pathway profile of women who met the inclusion criteria is also presented in Figure 6.

136

Figure 6: Pathway profile of women in the study

Synopsis of study using intention-to-treat analysis of first and subsequent ‘complete cycles’ of women who used PGD-A for assessment of embryos or morphological assessment alone in their initial fresh and subsequent ‘complete cycles’ Outcomes from treatments in the dashed boxes were based on per-protocol analysis

137

7.5.3 Intention-to-treat analysis: Clinical outcomes for up to the first 3 ‘single cycles’

Table 2 presents the clinical outcomes for up to the first three ‘single cycle’ according to intention-to-treat analysis. Women in the PGD-A group and morphology assessment group were similar in their mean age at the initial OPU (40.06 years in PGD-A group versus 40.05 years in morphology assessment group). Compared to the morphology assessment group, PGD-A patients underwent fewer fresh or frozen/thaw ‘single cycles’

- (2.00 cycles in PGD-A group versus 2.31 cycles in morphology assessment group, p<.001), had more oocytes collected per OPU (10.83 oocytes in PGD-A group versus

6.89 oocytes in morphology assessment group, p<.001), achieved better fertilisation rate

(61.77%in PGD-A group versus 56.61% in morphology assessment group, p<.001), and had higher average number of embryos created per OPU (6.70 embryos in PGD-A group versus 3.90 embryos in morphology assessment group, p<.001)

The embryo quality was similar between the two study groups. In the PGD-A group, the mean grade of days 2– 4 embryos were 2.28 ± 0.75 compared to 2.02 ± 0.81 in the morphology assessment group. For days 5–6, the mean grade was 3.47± 1.53 in the PGD-

A group compared to 3.46 ± 1.27 in the morphology assessment group.

The PGD-A group had fewer embryos available for transfer or cryopreservation per OPU than the morphology assessment group (0.84 embryos in PGD-A group versus 2.57 embryos in the morphology assessment group, p<.001). The mean number of embryos per transfer was lower in the PGD-A group compared with the morphology assessment group for both fresh and frozen/thaw transfers (1.11 embryos per transfer in PGD-A group versus 1.36 embryos per transfer in morphology assessment group, p<.001).

138

The PGD-A group achieved a higher implantation rate (36.75% in PGD-A group versus

13.75% in morphology assessment group, p<.001) and pregnancy rate per embryo transfer procedure – (per fresh embryo transfer procedure: 36.76% in PGD-A group versus 17.87% in morphology assessment group, p<.001; per frozen/thaw embryo transfer procedure: 53.33% in PGD-A group versus 17.52% in morphology assessment group, p<.01). The PGD-A group achieved a higher per initiated cycle live-birth rate

(14.47% versus 9.12%, p<0.01), and had a significantly lower pregnancy loss rate than the morphology assessment group (19.51% versus 34.78%, p<0.05).

Importantly, women in the PGD-A group underwent fewer ‘single cycles’ than the morphology assessment group to achieve a live birth [6.91 cycles (221 ‘single cycles’ initiated/32 live births) in PGD-A group versus 10.96 cycles (4,593 ‘single cycles’ initiated/419 live births) in morphology assessment group; p<.01]. The number of embryo transfer procedures needed to reach a live birth in the PGD-A group was 3.28 (105 embryo transfer procedures/32 live births) compared to 8.65 (3625 embryo transfer procedures/419 live births) in the morphology assessment group; (p<.001)

To account for the fact that most cycles from the PGD-A group transferred blastocysts

(94.6%) compared to the morphology assessment group (6.7%), a subanalysis was performed to compare single embryo blastocyst transfer between the two study groups

(PGD A: n= 88 vs morphology assessment group: n=155). The live-birth rate per ET procedure remained higher in the PGD-A group compared to the morphology assessment group (34.09% vs 12.26%, p<0.001), indicating that the type of embryo transferred in the morphology assessment group (day 2/3 or day 5/6), did not alter the interpretation of the live-birth rate. The results are shown in Table 2.

139

Table 2: Demographic characteristic and outcomes of first three ‘single cycles’ (fresh or frozen/thaw cycles) according to intention-to-treat analysis1 Morphology Total number of women at start of PGD-A assessment p-value first fresh cycle 2 n= 110 n= 1983

Age at first oocyte pick-up (OPU), mean ± SD, range 40.06 ± 1.97, 37-45 40.05 ± 2.34, 37-49 p=0.96

‘Single cycles’3 initiated n, mean per woman ± SD, 221, 2.00 ± 0.81 4593, 2.31 ± 0.83 p<0.001

Fresh cycle initiated 4, n (% of total ‘single cycles’ initiated) 203 (91.86) 3312 (72.13) p<0.001

Frozen/thaw cycle initiated 4, n (% of total ‘single cycles’ initiated) 18 (8.14) 1281 (27.87)

OPU cycles, n, mean per woman ± SD 196,1.78 ± 0.79 3194,1.61 ± 0.72 p<0.05

Oocytes collected, mean per OPU ± SD 10.83 ± 6.60 6.89 ± 5.67 p<0.001

Fertilisation rate (two-pronuclei/oocytes collected), % 61.77 56.61 p<0.001

Embryos created per OPU, mean ± SD 6.69 ± 4.58 3.90 ± 3.75 p<0.001

Embryos numbers and quality

Day 2-4 embryos, mean grade ± SD, (1=good, 5=poor) 2.28 ± 0.75 2.03 ± 0.81 p=0.12

Day 5 & 6 embryos, mean grade ± SD, (1=good, 8=poor) 3.47 ± 1.53 3.46 ± 1.27 p=0.55

Embryos available for transfer or cryopreservation, mean per cycle with OPU ± SD 0.84 ± 1.19 2.57 ± 2.64 p<0.001

Embryo transfer procedures

Embryo transfer (ET) procedures, n (% per cycle started) 105 (47.51) 3,625 (78.94) p<0.001

Embryos transferred, mean per ET procedure ± SD, range 1.11 ± 0.32, 1-2 1.36 ± 0.48,1-3 p<0.001

Implantation rate, (gestational sacs/total embryos transferred), % 36.75 13.75 p<0.001

140

Total number of women at start of Morphology first fresh cycle 2 (continue) PGD-A group assessment group n= 110 n= 1983 p-value

Pregnancy rate 5

Per cycle initiated, % 18.55 14.02 p=0.06

Per fresh ET procedure, % 36.76 17.87 p<0.001

Per frozen/thaw ET procedure, % 53.33 17.52 p<0.01

Percentage of pregnancies that failed to reach birth 6, % 19.51 34.78 p<0.05

Live-birth7, rate

Per cycle initiated, % 14.47 9.12 p<0.01

Per ET procedure, % 30.47 11.56 p<0.001 Multiple live-birth rate, % (% of live-birth delivery) 6.25 7.16 p=0.81 Total number of SET on day 5/6, n (% of total number of single embryo transferred) 88 (94.62) 155 (6.68) p<.001

Proportion of live births per single embryo transferred on day 5/6 (%) 34.09 12.26 p<.001

Discontinue rate8, % (% of women without a live born) 71.80 50.00 p<0.001

1 The intention-to-treat analysis assigned all outcomes from the first and any subsequent fresh or frozen/thaw ‘single cycles’ to the embryo selection strategy that the woman received in her first fresh cycle (PGD-A or morphological assessment alone). 2 The study cohort is stratified into women whose first OPU used either PGD-A or morphological assessment for selection of embryos. 3 A ‘single cycle’ refers to the outcomes of a discrete fresh or frozen/thaw cycle. 4 An initiated fresh cycle is defined as commencing with the administration of follicle stimulating hormone. An initiated frozen/thaw cycle is defined as the thawing of one or more embryos. 5 Pregnancy rates is defined as the presence of a heart beat per initiated cycle or embryo transfer procedure 6 Pregnancies that failed to reach birth include miscarriages, missed abortions, blighted pregnancies, ectopic pregnancies and terminations. 7 A live-birth is defined as the birth of at least one live born baby, with twins and triplets counted as one live-birth. 8 Discontinue rate refers to women who discontinue with current treatment without a live birth after up to 3 ‘single cycles’ A p-value of less than 0.05 is considered statistically significant

141

7.5.4 Intention-to-treat analysis: Cumulative live-birth rates (CLBR)

According to intention-to-treat analysis, the CLBR of up to three ‘complete ART cycles’

(i.e. where all subsequent frozen/thaw cycles were considered part of a ‘complete ART cycle’ from which the embryo initially originated) was comparable between the two study groups (30.90% in PGD-A group versus 26.77% in morphology assessment group, p=0.34). Table 3 and Figure 7 present the CLBR for up to three ‘complete ART cycles’.

Table 3: Clinical outcomes for up to three ‘complete ART cycles’ 1 according to intention-to-treat analysis 2 Morphology Total number of women at the start of first PGD-A assessment group fresh cycle 3, n= 110 n= 1983 p-value

Cumulative live-birth rate (CLBR),4 % 30.90 26.77 p=0.34

Proportion of women who did not have an embryo transfer (ET) procedure in their first fresh cycle, (% of first fresh cycle initiated) 58.18 20.32 p<0.001

CLBR of women with an ET procedure in first fresh cycle, % 50.00 30.89 p<0.01

Discontinue rate 5 (% of women without a live-born) 61.72 55.97 p=0.31

Follow-up time 6, mean ± SD (days) 503.62 ± 335.53 560.05 ± 338.73 p=0.08

1 A ‘complete cycle’ refers to the outcomes from all embryos created from an oocyte pick-up (OPU), including fresh and any subsequent frozen/thaw embryo transfer (FET) cycles. 2 The intention-to-treat analysis assigned all outcomes from the first and subsequent ‘complete cycle’ (fresh and any FET cycles) to the embryo selection strategy that the woman received in her first fresh cycle (PGD-A or morphological assessment alone). 3 The cohort is stratified into women whose first OPU have used either PGD-A or morphological assessment for selection of embryos. 4 Cumulative live-birth rate (CLBR) is defined as the number of live births divided by the total number of women who started treatment in either the PGD-A group or the morphology assessment group. 5 Discontinue rate refers to women who did not undertake up to 3 ‘complete cycles’ during the study period. 6 Mean follow-up time is based on the date of first OPU through to 31 Mar 2014 or to the first clinical pregnancy leading to a live born baby. A p-value of less than 0.05 is considered statistically significant

142

To account for the high proportion of women who did not reach an embryo transfer procedure in the PGD-A group compared to the morphology assessment group in the first fresh cycle (58.18% in the PGD-A group versus 20.32% in the morphology assessment group), the study also undertook a restricted analysis of the CLBR including only women who reached an embryo transfer procedure in their first fresh cycle. In this analysis, the

CLBR of up to three ‘complete ART cycles’ was significantly higher in the PGD-A group than in the morphology assessment group (50.0%; versus 30.89%, p<0.01) (Figure 7).

143

1 Cumulative live-birth rate (CLBR) is defined as the number of live-births divided by the total number of women who started treatment in either the PGD-A group or the morphology assessment group (i.e. 110 women in PGD-A group and 1983 women in the morphology assessment group).

2 A ‘complete cycle’ refers to the outcomes from all embryos created from an oocyte pick-up (OPU), including fresh and any subsequent frozen/thaw embryo transfer (FET) cycles.

3 CLBR of woman with an embryo transfer (ET) procedure in the first fresh cycle refers to the number of live births achieved by the number of women who commenced treatment and have an ET procedure in the first fresh cycle (i.e 46 women in the PGD-A group and 1580 women in the morphology assessment group have an ET procedure in their first fresh cycle ** P<0.01 ***P<0.001 PGD-A group versus morphology assessment group

Figure 7: Intention-to-treat analysis: Cumulative live-birth rates (CLBR) 1 of first and subsequent ‘complete cycles’ 2 for all women and among women who had an embryo transfer (ET) procedure in their first fresh cycle 3

144

7.5.5 Per-protocol analysis: Clinical outcomes for up to first 3 ‘single ART cycles’

The clinical outcomes of up to 3 ‘single cycles’ (fresh or frozen/thaw) were also analysed using per-protocol analysis i.e. outcomes of women were excluded from the analysis once they crossed over to the alternate treatment strategy during the study period. The results were similar to the results obtained from the intention-to-treat analysis and are presented in Supplementary Table 2.

The PGD-A group achieved a higher live birth rate per ‘single cycle’ than the morphology assessment group (16.49% in PGD-A group versus 9.10% in morphology assessment group, p<.01). On average women in the PGD-A achieved a live birth with fewer number of initiated ‘single cycles’ and embryo transfer procedures than those in the morphology assessment group. The number of initiated ‘single cycles’ to reach a live birth was 6.06

(188 ‘single cycles’/31 live births) for the PGD-A group and 10.98 (4559 ‘single cycles’/415 live births) for the morphology assessment group (p<.01). The number of embryo transfer procedures needed to reach a live birth in the PGD-A group was 2.94 (91 embryo transfer procedures/31 live births) compared to 8.71 (3615 embryo transfer procedures/415 live births) in the morphology assessment group; (p<.001)

7.5.6 Per-protocol analysis: Cumulative live-birth rates

From a per-protocol perspective, the CLBR for up to 3 ‘complete ART cycles’ was similar between the two study groups (30.00% in PGD-A group versus 26.17% in morphology assessment group, p=0.37). However, caution is needed when interpreting these finding because of the small number of women in the PGD-A group who returned for their 3rd

‘complete ART cycle’. Supplementary Table 3 presents the CLBR for up to three

‘complete ART cycles’.

145

7.5.7 Time to pregnancy leading to a live birth

Figure 8 presents the Kaplan-Meier curves for the time taken to reach a clinical pregnancy leading to a live birth. The intention-to-treat analysis revealed that on average it took women in the PGD-A group 25% less time (35.8 fewer days) than the morphology assessment group to achieve a clinical pregnancy leading to a live birth (104.8 days versus

140.6 days, log rank test p<0.05).

Similar results were found using the per-protocol analysis where women in the morphology assessment group took an average of 32 more days (p<.001) to reach clinical pregnancy leading to live birth than PGD-A group.

Figure 8: Intention-to-treat analysis: Time to clinical pregnancy from date of first oocyte pick-up (OPU) to clinical pregnancy resulting in a live birth for the PGD-A group and the morphology assessment group

146

7.6 Discussion

This is the first study to comprehensively analyse the cumulative effectiveness of PGD-

A over repeated cycles in women aged 37 years or over attending a fertility clinic for routine care.

This study found that in women aged 37 years or over, a strategy of PGD-A led to a higher live-birth rate per initiated ‘single cycle’ (fresh or frozen/thaw cycle), fewer ART cycles to achieve a live-birth, a shorter average time to a clinical pregnancy resulting in a live- birth and a lower rate of pregnancy loss compared to morphological assessment of embryos. However, the CLBR over one or multiple ‘complete ART cycles’ (all ETs from one OPU) was comparable for the two study groups.

The higher live-birth rate per ‘single cycle’ and lower rate of pregnancy loss after PGD-

A compared to morphological assessment of embryos alone observed in this study is consistent with findings from previous studies for women of advanced maternal age that only assessed per cycle results (Fishel, et al., 2011, Forman, et al., 2012, Sher, et al.,

2009). In particular, this study results are similar to the recently published RCT of PGD-

A by Rubio and colleagues (Rubio, et al., 2017) that was conducted in women of advanced maternal age, which used a similar platform of cleavage-stage biopsy and aCGH. Rubio and colleagues’ study has been the only RCT of PGD-A in women of advanced maternal age and was undertaken in women with a relatively good prognosis.

While this is a retrospective cohort study that cannot confirm causality, the findings from this present study complement this previous RCT, by providing a longitudinal perspective over multiple ‘complete ART cycles’ performed in an unrestricted population of ART- naïve older women attending a fertility clinic. Such real-world data are necessary for generalising RCT findings to clinical practice (Hershkop, et al., 2017, Rothwell, 2005).

147

Due to the embryo biopsy screening process, the mean number of embryos available for transfer or cryopreservation per OPU was significantly lower in the PGD-A group than in the morphology assessment group (0.84 embryos vs. 2.57 embryos). Therefore, it is not surprising that the CLBRs for the two groups converged over time because women in the morphology assessment group had more cryopreserved embryos available for subsequent FET cycles.

Given the apparent similarity in CLBRs, some have argued that sequential transfer of untested single embryos (Mastenbroek and Repping, 2014, Wong, et al., 2014) over repeated cycles is preferable to PGD-A particularly for older women with limited and marginal quality embryos available for transfer (Gleicher, et al., 2008). It is argued that this approach obviates the possibility of false negative results or chromosomal mosaicism where embryos that are falsely considered to be non-viable after PGD-A could otherwise have been cryopreserved and transferred in subsequent frozen/thaw cycles (Garcia-

Velasco and Fauser, 2016, Munné, et al., 2016, Werner, et al., 2014). This was also shown in a recent study by Greco and colleagues where some embryos with low level chromosomal mosaicism were implanted and developed into healthy newborns (Greco, et al., 2015).

However, while sequentially transferring untested cryopreserved embryos might be a viable option, undergoing multiple transfer cycles wastes reproductive time particularly in older women whose fertility potential decreases with age (Speroff, 1994, Wyndham, et al., 2012). This is exemplified in the present study where women in the morphology assessment group undertook an average of 11 cycles compared to 7 cycles in the PGD-A group to reach a live birth. Undergoing infertility treatment is physically and psychologically demanding for couples, particularly the waiting period between the embryo transfer procedure and a pregnancy result (Rajkhowa, et al., 2005, Verberg, et al., 148

2008). This is especially the case for couples experiencing repeated treatment failure and pregnancy loss (Verhaak, et al., 2005). The recurring cycle of hope and despair from repeated ART cycles in which untested aneuploid embryos are transferred should therefore not be underestimated.

Furthermore, several studies have demonstrated that aneuploidies are only subtly related to the morphological appearance of the embryos and high-quality embryos have been shown to have a high frequency of aneuploidy (Mertzanidou, et al., 2013, Voullaire, et al., 2007). Therefore, while the approach to select embryos for transfer solely based on their morphological appearance could potentially avoid the, albeit small, risk of discarding viable embryos, it does not reduce the risk of aneuploidic conception.

Aneuploidic conceptions that result in a live birth typically result in serious congenital malformation and/or cognitive abnormalities (Hassold, et al., 2007, Hassold and Hunt,

2001).

This present study reported on outcomes of cleavage stage biopsy and analysis of a single cell as it was the preferred approach during the study period. In recent years, many PGD-

A laboratories have favoured the use of days 5 or 6 blastocyst biopsy where 5−10 cells are removed from each embryo (Schoolcraft and Katz-Jaffe, 2013, Scott, et al., 2013).

Analysis of multiple cells is less technically challenging for the testing laboratory, although chromosomal mosaicism between cells within the biopsy sample can make interpretation problematic (Munné, et al., 2016). Blastocyst biopsy has been reported to result in higher implantation and pregnancy rates per transfer compared to cleavage stage biopsy (Scott et al, 2013) although some slightly slower-growing embryos that do not reach the blastocyst stage in time will not be biopsied. Although the developmental potential of these embryos is likely to be lower, some will be fully viable and would have resulted in live births if they had been transferred (Kovalevsky, et al., 2013). Some 149

patients have very few embryos that reach the blastocyst stage in vitro and for them, cleavage stage biopsy might be the best option.

The present study found a significantly lower rate of pregnancy loss among women who used PGD-A compared to those who used embryo selection by morphology alone

(19.51% in PGD-A versus 34.78% in the morphology assessment group, p<0.05). Similar outcomes of fewer pregnancy losses with PGD-A have been reported in other studies

(Chang, et al., 2016, Forman, et al., 2012, Rubio, et al., 2017)

The effect of PGD-A should also be considered in the context of its cost effectiveness.

The cost of PGD-A diagnostic tests will likely reduce as technology matures; however, presently, PGD-A is an expensive procedure, adding approximately $2,550 Australian dollars to the cost of a standard ART treatment cycle (approximately $11,000) in

Australia (Chambers, et al., 2012; Handyside et al 2014). However, this study has shown that women who used embryo selection by morphology alone undertook almost one and half times as many cycles to reach a similar CLBR than women who used PGD-A.

Therefore, the additional expenses associated with PGD-A could potentially be balanced against the savings from fewer repeated treatment cycles (Munné and Cohen, 2017).

7.6.1 Strength of the study

This present study has several strengths that distinguishes it from earlier studies. This is the first study to undertake a longitudinal reporting approach to assess the effectiveness of PGD-A. This study has chosen clinically important endpoints i.e. CLBR, number of cycles needed to treat, and time to pregnancy leading to a live birth and denominators such as per ‘single cycle’ and ‘complete ART cycles’ to assess the effectiveness of PGD-

A compared to using morphological assessment alone.

150

7.6.2 Limitation of the study

However, several aspects of the study warrant caution when interpreting the results.

While this is a large study, it remains a retrospective observational study that can only assess associations rather than causal relationships. The effectiveness of PGD-A over successive cycles would be best investigated by means of a RCT. Furthermore, although women in the two study groups were similar in mean age (40 years old) and embryo quality (what was the data on embryo quality), and were ART-naïve patients in the clinic, the study did not morphology assessment for the duration or type of their subfertility, or other important prognostic factors such as ovarian reserve.

Women in the PGD-A group had significantly higher number of oocytes collected than the morphology assessment group (10.83 versus. 6.89, respectively). While this might indicate a better prognostic in the PGD-A cohort, the advice of the treating clinicians is that this was protocol driven and was, based on achieving sufficient embryos for testing in the PGD-A group, and therefore may not confound the findings.

Another potential limitation of this study is the duration of follow up between the PGD-

A and morphology assessment groups. Although not statistically significant, women in the PGD-A group had a shorter mean time of follow-up than the morphology assessment group (a difference of 57 days; 503 days in PGD-A versus 560 days in the morphology assessment group), potentially underestimating and biasing against the PGD-A CLBR compared to the morphology assessment group.

151

7.7 Summary

The present study complements efficacy research obtained from RCTs and provides important information of the effectiveness of PGD-A over repeated cycles in a ‘real- world’ clinic setting. The study findings showed that, despite similar CLBR in women of advanced maternal age who have either PGD-A or morphological assessment of embryos in the first cycle, the PGD-A group achieved a higher live-birth rate per ‘single cycle’, undertook almost half the number of initiated and embryo transfer procedures, and took significantly less time to achieve a live-birth, suggesting that the additional expenses associated with PGD-A could potentially be balanced against the savings from fewer repeated treatment cycles. Chapter 8 extends the clinical findings reported in this chapter by assessing the cost-effectiveness of PGD-A to determine whether PGD-A represents

‘value for money’ from an individual and healthcare perspectives.

152

Chapter 8

Longitudinal analysis of PGD-A in women of advanced maternal age: a cost-effectiveness analysis

153

Chapter 8- Longitudinal analysis of PGD-A in women of advanced maternal age: a cost-effectiveness analysis

8.1 Introduction

This chapter extends the clinical findings reported in Chapter 7 by evaluating the cost- effectiveness of PGD-A versus morphological assessment of embryos alone in woman of advanced maternal age. The chapter begins with an overview of the health economic methods, including clinical and cost data sources, the costing approach applied to estimate the direct healthcare cost associated with ART treatment cycles, and the specific method used for the cost-effectiveness analysis (Section 8.4). The results of the cost-effectiveness analyses are then presented from both a patient and healthcare perspective (Section 8.5).

The chapter concludes with a summary of the findings and interpretation of the economics cost-effectiveness measures obtained.

8.2 Background

Chapter 7 assesses the clinical effectiveness of PGD-A and morphological assessment of embryos alone in ART-naïve women (i.e. first treatment with ART) aged 37 years or over using a retrospective cohort study design. The primary outcome measure from the cohort study was the cumulative live-birth rate (CLBR) for up to three ‘complete ART cycles’.

A ‘complete ART cycle’ refers to the outcomes from all embryos created from an oocyte pick-up (OPU). The cohort study found that, compared to ART-naïve women who commenced treatment with a morphological assessment of embryos (morphology assessment group), ART-naïve women who commenced treatment using PGD-A (PGD-

A group) achieved a higher per-cycle live-birth rate (14.47% in the PGD-A group versus

9.12% in the morphology assessment group, p<0.01), a lower rate of pregnancy loss

(19.50% in the PGD-A group versus 34.78% in the morphology assessment group,

154

p<.05), took a shorter mean time to reach a clinical pregnancy leading to a live birth

(104.8 days in the PGD-A group versus 140.6 days in the morphology assessment group, p<0.05) and required fewer ART cycles to achieve a live birth (6.91 cycles in the PGD-

A group versus 10.96 cycles in the morphology assessment group, p<0.01). However, a key finding from the cohort study was that after three ‘complete ART cycles’, the CLBR was comparable for the two study groups (30.90% in the PGD-A group versus 26.77% in the morphology assessment group, p=0.34).

These clinical findings are broadly in line with current evidence regarding the clinical role of PGD-A during ART treatment, which suggests that PGD-A results in higher implantation, pregnancy and per-cycle live-birth rates. However, when successive cycles are considered, the CLBR is comparable to the morphological assessment of embryos alone (Chen, et al., 2015, Dahdouh, et al., 2015, Rubio, et al., 2017). Chapter 6 provides a detailed summary of current evidence on the clinicial effectiveness of PGD-A compared to the morphological assessment of embryos alone.

Although systematic reviews have assessed the clinical effectiveness of PGD-A, the cost- effectiveness of PGD-A is largely unknown. Determining whether or not PGD-A is a cost-effective strategy is important from both a patient and a healthcare perspective.

Although the CLBR is similar for PGD-A and morphological assessment of embryos alone, fewer cycles are needed with PGD-A, meaning that there is a potential trade-off between the cost of PGD-A and the cost of additional ART cycles needed to achieve a live birth. PGD-A is an expensive procedure, adding approximately $2,550 Australian dollars (AUD) to the cost of a standard ART treatment cycle (approximately AUD

11,000) in Australia. There is also the potential risk of misdiagnosis of aneuploidy (false positives), which might result in patients undertaking additional costly fresh ART cycles to obtain euploid embryos for transfer (Greco, et al., 2015). Therefore, the key question 155

is whether the clinical benefit of PGD-A compared to morphological assessment of embryos alone is worth the cost (i.e. if it represents ‘value for money’) from an individual, government and societal perspective.

To date, only two studies have assessed the economic value of PGD-A, both of which involved women of advanced maternal age (Collins, et al., 2017, Rubio, et al., 2017).

Rubio and colleagues performed a cost-outcome analysis of PGD-A alongside a randomised controlled trial (RCT) that compared PGD-A and morphological assessment of embryos alone in women of advanced age (mean age 39 years). A cost-outcome analysis estimates the ratio of the cost of treatment to the outcome (in this case the live- birth rate); however, it does not compare this ratio to alternative treatment strategies.

Therefore, because the incremental cost of PGD-A for a live birth was not evaluated in the study by Rubio and colleagues (Rubio, et al., 2017), the analysis cannot be considered a full economic evaluation.

Collins and colleagues developed a decision-analytic model to assess the cost- effectiveness of PGD-A in women aged over 37 years who had at least one blastocyst for transfer. The decision-analytic model, based on the United States (US) healthcare perspective, reported an incremental cost-effectiveness ratio (ICER) – a measure of the additional cost for a live birth with PGD-A of $105,489 US dollars (USD) relative to using morphological assessment of embryos alone. An overview of ICER is discussed in chapter 5, which covers the economics and ART. As the ICER from the model was below their stated cost-effectiveness threshold of USD 145,063 for an ART-conceived child, the authors concluded that PGD-A is cost-effective for women aged over 37 years. The cost- effectiveness threshold value of USD 145,063 used by the authors was based on an average live-birth rate of 13.4% and an average cost of USD 19,415 for an ART cycle obtained from the literature. This is a novel approach to establishing a cost-effectiveness 156

threshold, and it has not been used in studies to determine the cost-effectiveness of other fertility treatment strategies. Furthermore, the model outcomes were based on a single- cycle analysis, and thus the impact of the cost and CLBR resulting from subsequent fresh and frozen/thaw cycles was not evaluated in the study (Collins, et al., 2017).

Evaluating the cost-effectiveness of PGD-A over successive cycles is important because most women undertake more than one ART cycle over the course of a treatment program.

A longitudinal analysis of multiple ART cycles provides a more comprehensive and meaningful assessment of the cost-effectiveness of PGD-A.

8.3 Aim of the study

The aim of the economic evaluation in the present study was to assess the cost- effectiveness of PGD-A in older women using the patient-level data obtained from the clinical effectiveness study of PGD-A reported in Chapter 7.

8.4 Methods

8.4.1 Clinical data sources

The treatment and outcome data used in the cost-effectiveness analysis are described in detail in Chapter 7. Briefly, the cohort study comprised infertile women aged 37 years or over (n=2,093) commencing their first fresh ART cycle at a large private fertility clinic in Melbourne, Australia between January 2011 and June 2013. The study cohort was stratified into women whose first OPU cycle used PGD-A (n=110 women, PGD-A group) or morphological assessment of embryos alone (n=1,983 women, morphology assessment group). In the PGD-A group, embryo biopsy was performed on day 3 of embryo development, tested using array comparative genomic hybridisation (array CGH), and transfer was predominantly performed on day 5. In the morphology assessment group, the majority of morphological assessed embryos was transferred predominantly on day 2. 157

Treatment outcomes from their first and subsequent fresh and frozen/thaw cycles were collected through to March 2014, or to the first live birth. All treatment data and outcome data were sourced from the fertility clinic’s patient and scientific information systems.

8.4.2 Cost data sources

The primary cost-effectiveness analysis was undertaken from a healthcare perspective in which the direct costs associated with ART treatment cycles were included, and reflected the payments made by the government, private health insurers and patients. Direct costs in this study consist of the resources consumed in the delivery of ART services, including consultations, medications, laboratory and embryology services, counselling services and healthcare personnel. Costs are expressed in 2015 AUD.

In Australia, the direct costs of ART services are covered by three payers – Medicare, private health insurers and patients. Australia’s universal insurance scheme, Medicare, provides standard rebates for ART services through its Medicare Benefits Schedule

(MBS), and for most ART medications through its Pharmaceutical Benefits Scheme

(PBS). Private health insurance (PHI) provides rebates for eligible scheme members for

ART procedures performed in a private hospital setting. Patients usually incur out-of- pocket (OOP) expenses to cover the gap between the MBS rebate and physician fees, the cost of procedures not covered by the MBS or PHI, and the cost of some medications not listed on the PBS. Indirect costs, such as productivity losses and travel times, were not included in the cost-effectiveness analysis because these costs are generally not considered by funding bodies when making decisions about the allocation of healthcare resources.

Additionally, the cost-effectiveness analysis was performed from a patient perspective where only the OOP expenses were considered. This is important because PGD-A is not

158

funded through the MBS or PHI, and there are significant OOP expenses, generally around AUD 4,000 for a standard ART treatment cycle (Chambers, et al., 2012).

The following sections describe in greater detail the cost component sources for this study. a) Pharmaceutical Benefits Scheme and ART

The majority of prescribed medications used in ART are covered under Section 100 of the PBS (s100) of the specialised medicines scheme. The PBS item codes and the costs for the commonly prescribed gonadotropin-releasing hormones analogues, follicle- stimulating hormones, human chorionic gonadotropin and progesterone were sourced from the 2015 PBS Schedule and are presented in Table 4. It is assumed that women in both study groups receive similar PBS-prescribed medications for a fresh ART treatment cycle.

Table 4: PBS item code and costs associated with an ART treatment cycle, 2015 PBS item Description and the prescription unit Unit cost code (AUD) 9584K Ganirelix 250 microgram/0.5 mL injection, 5 x 0.5 mL 230.40 syringes

8871X; Follitropin alfa 900 units one cartridge (adjustable 1305.45 6433N dose). Estimated cost based on average 220 units for 11 days = 2420 units,

6182J Choriogonadotropin Alfa 250 micrograms – 1500 units 54.80 3 ampoules 6366C Crinone progesterone 8% vaginal gel, 15 applications 148.50 Total 1739.15 Source: Schedule of Pharmaceutical Benefits report for 1 Jan 2015 to 31 Jan 2015; (Australian Government Department of Health and Ageing (DoHA), 2015) PBS: Pharmaceutical Benefit Scheme. Cost is expressed in 2015 AUD

159

b) Private health insurance and ART

Private health insurance (PHI) in Australia provides cover for in-hospital services, a number of ancillary medical services and some medications that are not covered under the PBS. As part of an ART treatment cycle, the same-day hospital accommodation charge for an OPU procedure is partially or fully covered by PHI if a patient is an eligible scheme member. The costs associated with an OPU procedure generally include charges associated with a Theatre Band-3 procedure, same-day hospital accommodation charges

(AUD 1,350) and anaesthetist fees (AUD 500). However, if a patient does not have PHI, the total hospital cost for an OPU procedure is assumed to be AUD 1,850.

c) Medicare Benefits Schedule and ART

In Australia, citizens and some non-permanent residents are eligible for partial reimbursement of ART services that are ‘medically necessary’ to treat infertility with no age or cycle limit imposed on rebates. The Medicare Benefits Schedule (MBS), administered and published by the Australian Government’s Department of Health, lists eligible MBS service items that attract a subsidy through Medicare. Each ART service item listed on the MBS has a ‘recommended’ provider fee (schedule fee) and a benefit amount (rebate) that the Australian Government reimburses for the service (Medicare

Australia, 2015).

In general, the Australian Government reimburses 75% of the schedule fee for ART service items performed for admitted patients in a hospital or day hospital facilities, and

85% for services performed as an outpatient service. For example, for MBS item 13200, which is used to claim for the first stimulated ART cycle reaching OPU performed in a calendar year, the schedule fee in 2015 was AUD 3,110.75 and the MBS benefit was

AUD 3,032 (85% of the schedule fee). For MBS item 13212, which is used to claim an

160

OPU, the schedule fee in 2015 was AUD 354.45 and the MBS benefit was AUD 265.85

(75% of the schedule fee).

A combination of MBS items is used to represent the sequential steps for different ART treatment cycles and adjunct procedures. For example, benefits paid for MBS items 13209

(planning and management for ART treatment), 13200 (initial fresh ART cycle performed in a calendar year), 13212 (oocyte retrieval) and 13215 (fresh or frozen embryo transfer) represent the total Medicare benefits payable for one fresh ART transfer cycle. In 2015, the MBS benefits paid for one fresh ART transfer cycle was AUD 3,819.95 (Medicare

Australia, 2015). The MBS items and associated fees for ART services relevant to this study were sourced from the MBS for 2015 and presented in Table 5. d) Out-of-pocket expenses and ART

Because provider fees are not regulated by the Australian Government, fertility clinics and specialists are free to charge patients a higher fee than the MBS benefit (and the schedule fee) set by the Australian Government. The difference between the actual fee charged by the specialists or clinic and the MBS benefit, coupled with any additional expenses such as medications not listed on the PBS, adjunct ART procedures such as

PGD-A, and hospital ‘gaps’, must be covered directly by the patient. For inpatient services such as OPU, patients receive a partial or full rebate for their OOP expenses through their PHI scheme.

The actual OOP expenses a patient pays are complicated by the existence of a number of

MBS safety nets that the government has introduced to reduce the financial burden on patients with substantial OOPs resulting from out-of-hospital services such as specialist attendances or services provided in private clinics.

161

The most important safety net for ART is the ‘Extended Medicare Safety Net’ (EMSN), which was introduced in 2004 to provide financial relief and assist individuals and families who incur high OOP expenses. Under this safety net, the Australian Government pays 80% of the OOP expenses for MBS services for the remainder of a calendar year once an annual threshold for an individual or family is reached. The annual threshold value for 2015 was AUD 2,000 per patient for non-concession patients (i.e. those who do not receive additional subsidies from the government due to financial hardship). For high- cost treatments that are principally conducted as an out-of-hospital service – such as ART treatment – the EMSN greatly reduced OOP expenses. However, in 2010, ‘caps’ were introduced to a number of MBS items, including all ART service items, to reduce the inflationary effect from the EMSN, which has led many fertility treatment providers to increase the fees that they charge for ART services (Centre for Health Economic Research and Evaluation, 2011).

To estimate the OOP expenses for the cost-effectiveness analysis undertaken in this doctoral research program, a comprehensive website survey of all Australian fertility clinics that published their ART fee schedule was conducted in 2015. Because most

Australian fertility clinics advertise their ART services (including PGD-A) and the associated clinic fees, it is possible to estimate the average OOP expenses borne by patients for different ART services. Of the 84 fertility clinics operating in Australia in

2015, 64 published their clinic fee schedule online. Based on the survey, the average OOP expense incurred by the patient for a fresh ART transfer cycle in the calendar year was

AUD 3,204.63. Table 5 presents the MBS benefits and OOP expenses for the different

ART services covered in this study.

162

Table 5: Medicare Benefits Schedule (MBS) item numbers, MBS benefits and out-of- pocket expenses associated with an ART treatment cycle, 2015

Treatment type MBS item number Unit cost (AUD)

MBS Out-of-pocket

benefits* expenses** Fresh cycle cancelled before oocyte retrieval a 13202; 13209 467.75 2,685.00

Fresh cycle proceeding to oocyte 13200; retrieval but cancelled before 13201f;13209; embryo transfer b 13212 3,370.20 3,204.63

13200; 13201f;13209; 13212; 13215; Completed fresh ART cycle c 13251 3,819.95 3,204.63

Failed frozen/thaw embryo transfer 13209; 13218 787.15 1,439.48

Completed frozen/thaw embryo 13209; 13218; transfer d 13215 881.60 1,439.48

PGD-A (per cohort of embryos) e NA - 2,550.00

Embryo cryopreservation and storage NA - 180.00 a Treatment includes planning and management, semen preparation, use of drugs to induce superovulation, ultrasound examinations. b Treatment includes planning and management, semen preparation, use of drugs to induce superovulation, ultrasound examinations, laboratory services, procedure for oocyte retrieval. c Treatment includes planning and management, use of drugs to induce superovulation, ultrasound examinations, intracytoplasmic sperm injection (ICSI) and preparation of semen, laboratory services, procedure for oocyte retrieval, embryo transfer. d Treatment includes planning management, preparation and transfer of frozen embryos. e PGD-A procedure involves using array comparative genomic hybridisation (array CGH) to select and transfer euploid embryos at fresh and frozen/thaw embryo transfer (FET) cycle. f 13201 applies to patient’s second or subsequent fresh cycle in a calendar year. Item 13200 apples to the initial fresh cycle proceeding to oocyte retrieval. *MBS: Medicare Benefits Schedule taken from the 2015 fee. OOP expenses exclude the cost of same-day hospital charge, theatre and anaesthetist fee for oocyte retrieval procedure (AUD 1,850). Source: (Medicare Australia, 2015). **OOP expenses are based on a website survey carried out in 2015. PGD-A= preimplantation genetic diagnosis for aneuploidy.

163

8.4.3 Costs estimates used in the cost-effectiveness analysis

The cost estimates for the PGD-A group and the morphology assessment group are outlined in Table 6. A ‘bottom-up’ costing approach was used to identify individual resource items associated with the treatment cycles undertaken by the patients in each study group (i.e. the PGD-A and morphology assessment groups in the cohort study described in Chapter 7). This costing approach is considered the gold standard in economic evaluations because it provides a more accurate and detailed unit cost than the

‘top-down’ costing approach which uses high-level summaries of cost items (Federowicz, et al., 2010, McIntosh, 2010). The two costing approaches were discussed in Chapter 5 on the economics and ART.

The total cost of treatments for each study group was estimated by multiplying the number and type of cycles reaching each cycle stage (e.g. cancelled cycle, initiated fresh cycle without an embryo transfer, ‘complete ART cycle’) with the associated unit cost, and then summing across all treatment cycles undertaken in each study group over the study period.

164

Table 6: Cost of treatment and number of ART treatment and additional procedures undertaken by the PGD-A group and the morphology assessment group, 2015

Cycle type Unit cost – Unit cost – No. of ART No. of ART direct patient cycles cycles healthcare f perspectiveg (PGD-A (morphology AUD AUD group) assessment group) Fresh cycle cancelled before oocyte retrieval a 4,743 2,685 7 136

Fresh cycle proceeding to oocyte retrieval but cancelled before embryo transfer b 10,015 3,204 106 809

Completed fresh ART cycle c 10,614 3,204 93 2,884

Failed frozen/thaw embryo transfer 2,226 1,440 4 237

Completed frozen/thaw embryo transfer d 2,321 1,440 18 1,765

PGD-A (per cohort of embryos) e - 2,550 175 89

Embryo cryopreservation and storage (1 year) - 180 30 1,675 a Treatment includes planning and management, semen preparation, use of drugs to induce superovulation, ultrasound examinations. b Treatment includes planning and management, semen preparation, use of drugs to induce superovulation, ultrasound examinations, laboratory services, procedure for oocyte retrieval. c Treatment includes planning and management, use of drugs to induce superovulation, ultrasound examinations, intracytoplasmic sperm injection (ICSI) and preparation of semen, laboratory services, procedure for oocyte retrieval, embryo transfer. d Treatment includes planning management, preparation and transfer of frozen embryo. e PGD-A procedure involves using array CGH to select and transfer euploid embryos during fresh and frozen/thaw embryo transfer (FET) cycle. f Direct healthcare cost include MBS benefit for relevant ART services, PBS subsidy for prescribed medication for a fresh ART cycle and the charges for theatre and accommodation and anaesthetist fee paid by PHI and patient’s OOP expenses. Unit costs are rounded off to the nearest tenth for calculation. All costs are expressed in 2015 AUD. PGD-A = preimplantation genetic diagnosis for aneuploidy MBS= Medicare Benefits Schedule, PBS=Pharmaceutical Benefits Scheme, PHI = private health insurance. g Patient perspective includes patient’s OOP expenses for ART treatment services undertaken during the study period. The MBS benefit in this table is based on the initial fresh ART treatment cycle performed in that calendar year.

165

8.4.4 Cost-effectiveness analysis

Chapter 5 provides a detailed discussion of key health economic concepts and the economic evaluation methods including the cost-effectiveness analysis used in this chapter. Consistent with the cohort study approach in Chapter 7, the cost-effectiveness analysis was undertaken and analysed according to intention-to-treat principles. The cost and primary outcome (i.e. CLBR per woman) for the two study groups are expressed as an incremental cost-effectiveness ratio (ICER). The ICER is calculated as the ratio of incremental cost and incremental outcome between the PGD-A group and the morphology assessment group, expressed by the following formula:

퐶표푠푡 − 퐶표푠푡 ICER = 푃퐺퐷−퐴 푔푟표푢푝 퐶표푛푡푟표푙 푔푟표푢푝 퐸푓푓푒푐푡 푃퐺퐷−퐴 푔푟표푢푝− 퐸푓푓푒푐푡퐶표푛푡푟표푙 푔푟표푢푝

Thus, in this study, the ICER reflects the additional cost needed to obtain an additional live birth from adopting a strategy of selecting embryos using PGD-A compared to selecting embryos based on morphologic assessment of embryos only.

8.4.5 Sensitivity analyses

The non-parametric technique of re-sampling from the original dataset (bootstrapping) was used to generate confidence intervals (CIs) for the mean treatment cost difference and estimates of cost–effect pairs for the calculation of the ICER (Drummond MF, et al.,

2005). The ICER was expressed with 95% CIs obtained from 5,000 bootstrap replications and plotted on a cost-effectiveness plane. A cost-effectiveness acceptability curve

(CEAC) was then constructed, plotting the probability that the PGD-A strategy was cost- effective over a range of cost-effectiveness thresholds. The cost-effectiveness threshold is often called the willingness-to-pay (WTP) threshold and represents the values at which the ICER is deemed cost-effective.

166

The impact of changing the cost and effectiveness parameters of PGD-A was explored using a one-way sensitivity analysis. Specifically, the cost of the PGD-A procedure was reduced by 10% to assess how sensitive the results were to changes in the cost of PGD-

A, because it is likely that as this technology matures, the cost of PGD-A will decrease.

All statistical, economic and sensitivity analyses were performed using STATA software version 11.2 (Stata Corp, College Station, TX).

8.5 Results

8.5.1 Clinical effectiveness

The cohort study in Chapter 7 reported that the CLBR for up to three’ complete ART cycles’ was 30.90% for the PGD-A strategy compared to 26.77% for the morphology assessment group. The difference was not statistically different (p-value=0.34). This lack of statistical difference in the clinical outcome between the two study groups might suggest that a cost-minimisation analysis is a possible economic evaluation method.

However, it is important to perform a full cost-effectiveness analysis when non-statistical difference is present because of the uncertainty around the estimates and the potential trade-offs between the costs and outcomes that can only be assessed with a cost- effectiveness analysis (Briggs and O'Brien, 2001, Dakin and Wordsworth, 2013). This is particularly relevant in this study as the PGD-A group took a significantly fewer average number of ART cycles than the morphology assessment group to reach a pregnancy resulting in a live birth.

167

8.5.2 Cost-effectiveness analysis a) Base-case estimate (healthcare perspective)

Table 7 summarises the base-case result of the cost-effectiveness analysis. Based on the costing approach described in section 8.4.3, the mean treatment cost per woman in the PGD-A group and the morphology assessment group was AUD 22,962 and AUD

21,801 respectively. Given that the CLBR for up to three complete ART cycles was

30.90% with PGD-A and 26.77% in the morphology assessment group, this yielded an ICER of AUD 28,103 (95%CI AUD 2,722,116 to being dominated by the morphology assessment group). The ICER of AUD 28,103 represents the cost of an additional live birth with PGD-A compared to morphological assessment of embryos alone.

The wide 95%CI for ICER implies that there is a considerable level of uncertainty with some probability that PGD-A is more-costly and less effective compared to the morphology assessment group. This uncertainty is also illustrated in the cost- effectiveness plane in Figure 9 which shows the results of 5,000 simulations of PGD-

A group versus the morphology assessment group.

The majority of the simulations are located in right upper quadrant of the plane. This represents that PGD-A is more-costly and more effective relative to the morphology assessment. However, a smaller proportion (17%) of the simulations were located on the left upper quadrant. This means that PGD-A is likely to be more-costly and less effective relative to the morphology assessment. Therefore, it is dominated by morphology assessment. The following section 8.5.2b on “Sensitivity analysis from a healthcare perspective” discussed the simulation results in greater detail.

168

Table 7: Incremental cost-effectiveness ratio (ICER) for the PGD-A group versus the morphology assessment group (base case results)

Treatment group Mean treatment CLBR (%) Mean difference Effect difference, % cost (AUD) cost (AUD) (95%CI) a (95%CI) a

PGD-A group 22,962 30.90

Morphology assessment group 21,801 26.77 1,161 (–585 to 2,908) 4.13 (–4.7 to 13.0) ICER (AUD) (95%CI) 28,103 (2,722,116 to dominated*) a The 95% confidence interval (95%CI) is calculated based on non-parametric bootstrap estimation using 5,000 replications. * Dominated: PGD-A group was more costly and less effective than the morphology assessment group. Costs are expressed in 2005 AUD.

169

b) Sensitivity analysis from a healthcare perspective

Figure 9 presents the cost-effectiveness plane depicting the scatterplot of the 5,000 replications of the mean differences in the costs and CLBR for the PGD-A group and the morphology assessment group. Most of the simulations (73%) lie in the right upper quadrant, indicating that although PGD-A has a high probability of being more effective than the morphology assessment group, it is also more costly compared to the morphology assessment group. The small proportion of simulation in the left upper quadrant indicated that in around 17% of the cases, undertaking PGD-A is more costly and less effective than the morphological assessment. Therefore, PGD-A is dominated by morphological assessment group.

The cost-effectiveness plane also indicated that there is uncertainty around the estimate with cost–effect pairs appearing in other quadrants. The dispersion of simulations above and below the horizontal axis and the wide 95% CI around the ICER indicates that there is some uncertainty about whether the incremental effect (i.e. additional live births) from the PGD-A is achieved at a higher or lower cost.

170

Figure 9: Cost-effectiveness plane showing the relationship between the incremental cost and effect between the PGD-A group and the morphology assessment group from a healthcare perspective

To explore how the ICER might be further interpreted from a decision maker’s perspective, a CEAC was constructed for different cost-effectiveness threshold values for a live birth with PGD-A. That is, the CEAC summarises the probability of PGD-A being cost-effective relative to morphological assessment of embryos alone for a range of thresholds that society is willing to pay for a live birth (Figure 10). For example, if society is willing to pay AUD 50,000 for an ART-conceived child, the results suggest that a strategy of PGD-A in older women is more cost-effective compared to the morphology assessment group in 80% of cases.

171

Figure 10: Cost-effectiveness acceptability curve (CEAC) from a healthcare perspective

To assess the impact of changing the cost and effectiveness parameters of PGD-A on the cost-effectiveness result, the cost of the PGD-A procedure was reduced by 10% while keeping other treatment cost and parameter estimates constant. This one-way sensitivity analysis yielded an ICER of AUD 19,225 per additional live birth, which is almost 31% less than the base-case estimate of AUD 28,103 per additional live-birth with PGD-A.

This finding, that a 10% increase in cost of PGD-A results in a 31% increase in ICER, suggests that the cost-effectiveness of PGD-A is very sensitive to the cost of the PGD-A procedure.

172

c) Base-case estimate (patient perspective)

In the context of the present study, a cost-effectiveness analysis that only considers OOP expenses incurred by an individual represents the patient’s perspective. Therefore, this analysis was repeated only including OOP expenses, and is presented in Table 8. The results demonstrate that the PGD-A group incurred a higher average OOP expenses than the morphology assessment group (AUD 9,913 versus AUD 7,340 respectively). The

ICER was AUD 62,270 for an additional live birth with PGD-A compared to morphological assessment of embryos alone from a patient’s perspective. This is perhaps not surprising given that PGD-A is a relatively costly procedure that is not covered by

Medicare, and more fresh ART cycles were undertaken by the PGD-A group than the morphology assessment group during the study period.

Overall, the cohort study found that the PGD-A group undertook fewer ART cycles to reach a live birth than the morphology assessment group; however, a higher proportion of these cycles were fresh ART cycles compared to the morphology assessment group that undertook a higher proportion of frozen/thaw embryo transfer (FET) cycles.

Consequently, the cost of fewer ART cycles undertaken in the PGD-A group did not compensate for the additional cost of fresh ART cycles, or for the cost of the PGD-A procedure in the PGD-A group compared to the morphology assessment group.

173

Table 8: Incremental cost-effectiveness ratio (ICER) for the PGD-A group versus the morphology assessment group (patient perspective)

Treatment group Mean treatment CLBR (%) Mean difference in cost Effect difference (%) cost (AUD) (AUD) (95%CI) a (95%CI) a

PGD-A group 9,913 30.90

Morphology assessment group 7,340 26.77 2,572 (1,862–3,282) 4.13 (–4.6 to 12.9)

ICER (AUD) (95%CI) 62,270 (7,457,774 to dominated*) a 95% confidence interval (95%CI) calculated by non-parametric bootstrap estimation using 5,000 replications. * Dominated: PGD-A group was more costly and less effective than the morphology assessment group. Cost are expressed in 2015 AUD.

174

Figure 11: Cost-effectiveness plane showing the relationship between the incremental cost and effect for the PGD-A group and the morphology assessment group from a patient perspective

Figure 11 presents the cost-effectiveness plane from a patient perspective based on the scatterplot of 5,000 replications of the mean differences in the costs and CLBR for the

PGD-A group and the morphology assessment group. The majority of the simulations lie in the right upper quadrant of the cost-effectiveness plane, indicating that although PGD-

A has a higher probability of being more effective, it is also more costly than the morphological assessment of embryos alone in 82% of the cases.

175

8.6 Discussion

This is the first study to use real-world patient-level data to assess the cost-effectiveness of PGD-A relative to the morphological assessment of embryos alone. The cost-outcome analysis performed alongside the RCT by Rubio and colleagues (Rubio, et al., 2017) did not calculate the incremental cost of PGD-A for a live birth (i.e. ICER), and it is not considered a cost-effectiveness study. The other cost-effectiveness study by Collins and colleagues (Collins, et al., 2017) relied on economic modelling, and model parameters were sourced from the literature. From a healthcare funding perspective, the present study found an incremental cost of AUD 28,103 (95%CI AUD 2,722,116 to being dominated by morphology assessment group) for an additional live birth with PGD-A. Whether this represents good value for money depends on society’s cost-effectiveness (WTP) threshold for a live birth conceived through fertility treatment.

In Australia, there is no cost-effectiveness threshold value (or range of values) that represents society’s (or the government’s) WTP for an additional live birth. The most common outcome measure for health economic assessment is the quality-adjusted life- year (QALY), which combines the quality of life and the length of time spent in a health state. For most government bodies and authorities such as the Pharmaceutical Benefits

Advisory Committee (PBAC), the QALY is the preferred outcome measure of interpreting the value for money that an intervention represents before considering funding through public resources (Neumann, 2011).

However, the application of QALY to quantify the benefits of fertility treatments such as

ART is highly questionable. Unlike other areas of healthcare, fertility treatment is uniquely judged based on its ability to create new life. This is diametrically different to

176

how other types of medical treatments are judged, which are based on their ability to save, extend or improve the quality of life, and are all better suited for QALY measurement.

Economic evaluations of fertility treatment are also conceptually difficult because foetal events such as stillbirths and miscarriages are not captured by QALY, because QALY assumes life commences at birth (Devlin and Parkin, 2003, Klitzman, 2017, McGregor and Caro, 2006).

However, medical interventions that result in a ICER of up to AUD 50,000 per QALY gained are considered cost-effective within the Australian setting (Harris, et al., 2008).

Therefore, the ICER reported in the present cost-effectiveness study (AUD 28,103) suggests that from a healthcare perspective, a strategy of PGD-A for up to three ‘complete

ART cycles’ in women aged 37 years or over is likely to be cost-effective compared to using conventional morphological assessment of embryos alone.

This finding is consistent with the result of the decision-analytic model developed by

Collins and colleagues (Collins, et al., 2017), which concluded that PGD-A is cost- effective compared to morphological assessment of embryos alone in women aged over

37 years with at least one blastocyst for transfer from a healthcare perspective. However, the analysis in this chapter is based on real-world patient-level data, which captured treatments undertaken by women for up to three ‘complete ART cycles’. Therefore, this result provides a more realistic and comprehensive assessment of the overall cost- effectiveness of PGD-A.

While the study found that PGD-A is cost-effective from a healthcare perspective, the

ICER of AUD 62,270 for an additional live birth with PGD-A is likely to be considered not cost-effective from a patient perspective. The higher OOP expenses in the PGD-A group were largely explained by the additional cost of the PGD-A procedure

177

(AUD2,550), and a higher proportion of fresh ART cycles undertaken in the PGD-A group compared to the morphology assessment group over the study period. This is not surprising given that PGD-A usually results in fewer embryos available for transfer, particularly in older women, and thus more fresh cycles are undertaken to obtain euploid embryos for transfer.

The higher expenses associated with PGD-A mean that PGD-A would be prohibitive for some patients. A US study examining the factors associated with an individual’s decision to initiate ART treatment with PGD-A found that the cost of PGD-A (approximately USD

3,500 per PGD-A procedure) is the most significant determinant in a patient’s decision whether to use PGD-A, followed by factors such as clinic provider’s education on PGD-

A (25.6%) and partner’s support (21.9%) (Gebhart, et al., 2016). However, it is likely that as technology matures, the cost of PGD-A will decrease and become more affordable for patients (Munné and Cohen, 2017). This was demonstrated in the sensitivity analysis where the incremental cost of a live birth with PGD-A was reduced by 30% when the cost of PGD-A decreased by 10%.

With the significant improvement in the cryopreservation methods using modern vitrification techniques, it has been suggested that the sequential transfer of untested embryos over multiple transfers is less costly (Papaleo, et al., 2017, Roque, et al., 2015), and may just as effective an alternative to PGD-A (Mastenbroek and Repping, 2014,

Wong, et al., 2014). However, other non-economic factors such as the time to pregnancy, maternal age, the number of viable embryos obtained, and patient preferences should also be considered when deciding on treatment modality (Lee, et al., 2014, Munné and Cohen,

2017).

178

8.6.1 Strength of the study

This is the first study to use real-world patient-level data to assess the cost-effectiveness of PGD-A compared to the morphological assessment of embryos alone. Therefore, this analysis distinguishes itself from the earlier cost-effectiveness study of PGD-A which relied on economic modelling techniques and estimates of clinical outcomes from the literature (Collins, et al., 2017). Furthermore, the cohort study in Chapter 7 undertook a longitudinal perspective of patients undertaking PGD-A and thus provides a more accurate and reliable estimate of the cost-effectiveness of PGD-A.The bottom-up costing approach, which provides detailed unit costs incurred for different treatment cycle types, allows the cost-effectiveness analysis to be generalised to other healthcare setting.

8.6.2 Limitation of the study

Although the cost-effectiveness analysis of PGD-A was based on a carefully designed and executed retrospective cohort analysis, observational studies are vulnerable to bias and confounding factors (Grimes and Schulz, 2002, Ioannidis, et al., 2001). The cohort study was based on patients who underwent ART treatment in a single fertility clinic in an Australian setting, which might not be applicable to other patient populations, different biopsy methods and timing (polar body and trophectoderm), and funding arrangements beyond those analysed. Finally, in recent years, many PGD-A laboratories have favoured the use of days 5 or 6 blastocyst biopsy where 5−10 cells are removed from each embryo, thus the results of this study may not be relevant to the contemporary practice.”

179

8.7 Summary

The present study suggests that a strategy of commencing ART treatment with PGD-A in women of advanced maternal age is cost-effective from a healthcare perspective but not from a patient perspective.

Given the trend towards delayed childbearing and the growing dependence on fertility treatment to improve the chances of having a child, the financial burden of ART on the healthcare system is likely to continue to increase and, therefore, cost-effectiveness analysis such as the one described in this chapter is necessary and important (Clement, et al., 2009, Taylor, et al., 2004).

In recent years, there has been an increasing trend to use other treatment strategies to treat women of advanced maternal age. These include ‘social freezing’, where women cryopreserve their oocytes at a younger age and return for ART treatment at a later time to use their stored frozen oocytes, and ART using donated oocytes from younger women whose oocytes have a higher reproductive potential (Baldwin, et al., 2015, Hodes-Wertz, et al., 2013, Kawwass, et al., 2013).

Although several studies have assessed the clinical effectiveness of these treatment strategies in women of advanced maternal age individually (Cobo, et al., 2015, Kawwass, et al., 2013, Kushnir, et al., 2015, Wang, et al., 2012), no published studies to date have systematically compared the clinical effectiveness and cost-effectiveness of these strategies in a decision-analytical model. Therefore, Chapter 9 extends the cost- effectiveness analysis described in this chapter by using a Markov model to compare four commonly used treatment strategies – PGD-A, social oocyte freezing, ART using donor oocytes, and standard ART treatment – in women of advanced maternal age.

180

Chapter 9

Which ART treatment strategy is the most clinically and cost-effective for women of advanced maternal age? A Markov model

181

Chapter 9: Which ART treatment strategy is the most clinically and cost-effective for women of advanced maternal age? A Markov Model

9.1 Introduction

This chapter extends the cost-effectiveness analysis described in Chapter 8 by constructing a Markov model to incorporate two other alternate treatment strategies – social egg freezing and the donated oocytes. The chapter begins with the development of the model structure and the description of four ART strategies commonly used in women of advanced maternal age (Section 9.5.2). A major part of the chapter is the estimation of model parameters, including transition probabilities, costs and health outcomes to populate the model (Section 9.5.4). The base-case result and probabilistic sensitivity analysis are presented in Section 9.7. This chapter concludes with a summary of the results.

9.2 Background

Despite the mainstream acceptance of ART treatment and incremental improvements in overall success rates, women of advanced maternal age (generally considered to be 37 years or over), remain a challenge for fertility physicians (Marquard, et al., 2010, Sunkara, et al., 2011).

According to the latest report from Australian and New Zealand Assisted Reproduction

Database (ANZARD) on all ART treatment performed in 2015, less than one in ten

(7.5%) initiated cycles result in a live delivery for women aged 40 years or over using their own eggs for standard ART, compared to 26.2% of cycles undertaken by women aged below 30 years, 24.3% in women aged 30–34 years, and 18.4% in women aged 35–

39 years following ART (Fitzgerald, et al., 2017). 182

Chapters 7 and 8 reported the clinical and cost effectiveness of preimplantation genetic diagnosis for aneuploidy (PGD-A) in women aged 37 years or over. However, two other

ART strategies − social egg freezing (at a younger age for potential future use) and the use of donated oocytes are commonly used for improving the success rates in older women.

Social egg freezing, where autologous oocytes are collected and cryopreserved at a younger age for potential future use, has been increasingly used as an option to counter the delving in fertility as women age (Cobo, et al., 2015, Stoop, 2016). Social egg freezing strategy has become an increasingly viable option for extending reproductive lifespan due to modern vitrification methods. The largest series of women who returned to use their stored oocytes was reported by Cobo and colleagues (Cobo, et al., 2015). Among the 120 women who vitrified their oocytes at a mean age of 37.7 years and returned to use their stored oocytes at about 40 years of age, slightly more than one third of them (35%) achieved clinical pregnancy and one fifth achieved a live-birth.

The most traditional approach to improving ART success in women of advanced maternal age is the use of donated oocytes from younger women. For example, rates of donor ART continue to increase in the United States (US), with 10,801 donor cycles performed in

2000 to around 20,000 in 2013, accounting for 10.5% of all ART cycles and 60% of cycles in women aged over 40 years (Centres for Disease Control and Prevention, 2015,

Kawwass, et al., 2013). Success rates following donor ART remain relatively good in women aged over 40 years, with more than one in four cycles resulting in a term singleton live-birth (Kawwass, et al., 2013).

183

9.3 Aim of the study

The aim of the present study described in this chapter was to develop a decision-analytic model to assess the relative clinical and cost-effectiveness of the four most common ART strategies in women aged 40 years following 6 to 12 months of infertility.

9.4 Choosing the appropriate decision model

There are different types of modelling techniques that are applicable to the economic evaluation of healthcare interventions. The appropriate model type depends on the nature, time frame and complexity of the decision problem. The most common decision models are decision trees and the Markov model although there are alternative modelling approaches such as patient-level simulation and discrete event models (Briggs, et al.,

2006).

Decision trees are the simplest and most straightforward type of decision-analytic modelling in economic evaluation. This type of model is most appropriate when analysing a one-off event which has a short and fixed time horizon with no recurring events e.g. single ART cycle.

In comparison, Markov models utilise stochastic (random) processes where patients move from one health state to another or stay in the same state based on transition probabilities.

Markov models are most suited to scenarios where an intervention is associated with an ongoing risk e.g. the probability of achieving a live birth over multiple ART cycles. This technique allows patients to move between health states over a specific timeframe that are divided into equal time (or ART cycles) until the study is completed (Briggs, et al.,

2006).

184

In the present study, the Markov model was the preferred choice of modelling technique because most women undertake more than one ART cycle over the entire treatment course, and consequently, treatment pathways can become very complex and difficult to be represented by a simple decision tree.

9.5 Methods

9.5.1 Study population

The hypothetical study population was ART-naïve (i.e. first time seeking ART treatment) women aged 40 years following 6 to 12 months of infertility.

Although the retrospective cohort study described in Chapters 7 and 8 evaluated the clinical and cost‒effectiveness of PGD-A and morphological assessment of embryos alone for women aged 37 years or over, this Markov model was developed for women aged 40 years because most published data describes women by 5-year age groupings, and the definition of women of advanced maternal age often relates to this older age group.

9.5.2 Model strategies and structure

The Markov model was developed to represent each of the four main treatment strategies

− autologous ART, social egg freezing, PGD-A and donor ART. The schematic representation of the four main treatment strategies is presented in Figure 12 and summarised below. Each strategy is configured to represent the likely clinical pathway in terms of the number of embryos available for cryopreservation and transfer.

185

1. Autologous ART

Autologous ART is the reference strategy and involves ART naïve women undertaking two ‘complete autologous ART cycles’ at age 40. A ‘complete autologous ART cycle’ in the reference strategy is defined as a fresh autologous cycle followed by two subsequent frozen/thaw embryo transfer (FET) cycles resulting from one ovarian stimulation episode. The model begins with infertile ART naïve women commencing treatment by entering the ‘fresh cycle’ state where she can either proceed to an oocyte pick-up (OPU) or the cycle is cancelled (e.g. due to poor response or hyperstimulation). If the cycle is cancelled, she can either end her treatment or undertake a second fresh cycle. If no live-birth is achieved after her first embryo transfer, she can end her treatment, commence a second fresh cycle or undertake a FET cycle. The strategy pathway ends when two ‘complete autologous ART cycles’ have been undertaken.

2. Preimplantation genetic diagnosis for aneuploidy

The PGD-A strategy involves undertaking two ‘complete autologous cycles’ with PGD-

A at 40 years of age. In this strategy, a ‘complete PGD-A autologous cycle’ is defined as a fresh autologous cycle followed by one subsequent FET cycle resulting from one ovarian stimulation episode. The model begins with infertile 40-year-old ART naïve women entering the model in a ‘fresh cycle’ state leading to either an OPU procedure or a cancelled cycle. If the initial fresh cycle is cancelled, the woman can either end her treatment or undertake a second fresh cycle. If no live-birth is achieved after her first embryo transfer, she can end her treatment, commence a second fresh cycle or undertake a FET cycle. Consistent with the protocol for array comparative genomic hybridization

(aCGH) applied in Chapters 7 and 8, it is assumed that the present study used the same

186

aCGH protocol and blastomere biopsy to select euploid embryos for transfer during fresh and FET cycles.

3. Social egg freezing

The social egg freezing strategy involves elective oocyte cryopreservation at age 35 with women returning at age 40 for ART using their stored oocytes. In this strategy, the model starts with 35-year-old women entering the ‘fresh cycle’ leading to either an OPU procedure or a cancelled cycle. If the fresh cycle is cancelled, the women can either end her treatment or undertake another stimulation cycle to retrieve oocytes for cryopreservation. The model assumes that one OPU procedure will retrieve sufficient oocytes for two autologous FET cycles at age 40. The strategy pathway ends when two

FET cycles using vitrified oocytes have been performed at age 40.

4. Donor ART

The donor ART strategy starts with infertile 40-year-old ART naïve women entering the model to undertake two ‘complete autologous ART cycles’ followed by two donor ART cycles. In this strategy, a ‘complete autologous ART cycle’ refers to a fresh autologous cycle followed by two subsequent FET cycles resulting from one episode of ovarian stimulation. If the woman is unsuccessful in achieving a live-birth after two ‘complete autologous ART cycles’, she can either end her treatment or undertake up to two FET cycles using donated oocytes (donor ART cycles). The donor ART strategy pathway ends when two ‘complete autologous ART cycles’ followed by two donor ART cycles have been undertaken.

187

In all strategies, ‘live-birth’ and ‘treatment ends’ are absorbing states which means that women who enter these states cannot move to another health state. The model keeps track of treatment costs and live-birth rates associated with each health state so that as the model simulates the cohorts progressing through each strategy, the expected mean costs, cumulative live-birth rate (CLBR) and cost-effectiveness are calculated. The cycle time of the Markov model has been defined as a single treatment cycle [i.e. a discrete fresh or FET cycle].

It should be highlighted that for all strategies in the model, women can end their treatment when (i) a cycle is cancelled (e.g. due to poor response or hyperstimulation),

(ii) no embryos are available for transfer, or (iii) a live-birth is not achieved after embryo transfer. In the model, treatment strategy ends (i.e. an absorbing state is entered) when (i) a live-birth is achieved and (ii) the treatment pathway is completed.

188

Figure 12: Types of ART strategies used in women of advanced maternal age

189

9.5.3 Model parameters

The Markov model was informed by a range of data sources from the literature including

RCTs, ART registries and observational studies. Live-birth rates using autologous fresh,

FET and donated ART cycles, cycle cancelation, transition probabilities and discontinuation rates were sourced from the ANZARD held at the National Perinatal

Epidemiology and Statistics Unit of the University of New South Wales, Sydney (Harris, et al., 2016). ANZARD collects information on all ART treatment cycles undertaken in

Australia and New Zealand, including the resulting treatment and pregnancy outcomes.

The model inputs are detailed below and in Table 9, including the source of the data inputs and the distribution of the data used.

1. Autologous ART

All transition probabilities and live-birth rates in the autologous ART strategy were sourced from ANZARD for women who commence ART treatment at age 40. For example, the ANZARD reported that, for women aged 40 years, the proportion of OPU cycles reaching embryo transfer was 68% resulting in a live-birth rate of 16.0%. The proportion of FET cycles reaching embryo transfer was 91% resulting in a live-birth rate of 13.0% (Table 9)

2. PGD-A

The fresh embryo transfer and live-birth rates after PGD-A were taken from a recently published multicentre RCT comparing PGD-A using day-3 blastomere biopsy and aCGH with morphological assessment of embryos alone in relatively good prognosis older women (mean age 39.1 years) (Rubio, et al., 2017). In the RCT, a total of 100 women had

PGD-A in their first fresh cycle. Among them, 68% of the fresh cycles started reached embryo transfer resulting in a delivery rate per fresh embryo transfer of 52.9%. As the

190

RCT reported the outcome of only one ‘complete cycle’ (i.e. fresh and any subsequent

FET cycles), the Markov model in the present study assumed that there was a 20% reduction in the live-birth rate during the second fresh PGD-A cycle. Live-birth rates in

PGD-A FET cycles were assumed to be similar to the estimate used in the fresh PGD-A cycle (Table 9).

3. Social egg freezing

The embryo transfer and live-birth rates following social egg freezing were sourced from the retrospective multicentre study by Cobo and colleagues (Cobo, et al., 2015). This study was based on 1,382 women (mean age of 37.2 years) who electively vitrified their oocytes to prevent age-related infertility. In the study, an average of 12.7 oocytes was retrieved and 9.7 metaphase II oocytes were vitrified. The study reported that 120 women returned to use their stored oocytes after average oocytes storage of 2.2 years. As not all oocytes survived following vitrification and thawing (survival rate was 80.5% in this

RCT), 85% of women who returned, had an embryo transfer resulting in a live birth rate of 23.5% per frozen embryo transfer (Cobo, et al., 2015). The model inputs are presented in Table 9.

4. Donor ART

The live-birth rate of ART donor cycles was sourced from ANZARD. The donor programme in Australia and New Zealand accounted for approximately 5% of all ART cycles in the five years to 2014 (Harris, et al., 2016). In the model, the live-birth rate of

27.0% per FET was based on the average age of donors (i.e. 32 years) for recipients at age 40 (Table 9).

191

5. Transition probabilities

Transition probabilities refer to the likelihood of patients moving from one health state to another. The Markov model applied in the present study modelled transition between four main health states in each strategy: ‘fresh cycle’, ‘FET’, ‘live-birth’ and ‘treatment ends’.

The donor ART strategy has an additional health state ‘donor’ as patients may undertake

FET cycles using donor oocytes if previous two ‘complete autologous cycles’ failed to achieve a live birth.

Figure 13 below illustrates the transition of patient undertaking different ART strategies in the Markov model. The oval represents a health state and the arrows represent transitions between the health states in each strategy.

192

Figure 13: Transition of patients undertaking different ART strategies in the model

Patients were modelled as being in one of these health states during a cycle and transitioning to other health states based on assigned health state-specific transition probabilities (Table 9).

193

Using the PGD-A strategy as an example, Figure 13b shows that the transition probability of patients without a live birth moving from fresh PGD-A cycle to PGD-A

FET cycle is 0.13, which indicates that 13% of patients progress from a PGD-A fresh cycle to a PGD-A FET cycle whilst the remaining 87% of patients undertake another

PGD-A fresh cycle or end their treatment.

Figure 13b: Diagram to represent four health states of a Markov model for PGD-A strategy

194

Table 9: Clinical data sources Probability estimates Distributionc References ART Cycle 1 ART Cycle 2 Autologous ART Fresh cycle Initiated cycles reaching oocyte pick-up (OPU) 0.88 0.88 Beta (Harris, et al., 2016) OPU cycles reaching embryo transfer (ET) 0.68 0.68 Beta (Harris, et al., 2016) (Harris, et al., 2016, Kushnir, et al., 2016, Luke, et al., 2012, Smith, et al., 2015, Sunkara, et al., 2016, Live-birth per ET 0.16 0.14 Beta Ubaldi, et al., 2015)

Frozen-thawed transfer (FET) cycle Embryo transfer per initiated FET cycle 0.91 0.91 Beta (Harris, et al., 2016) (Harris, et al., 2016, Luke, et al., 2012, McLernon, et al., 2016, Smith, et al., 2015, Ubaldi, et al., Live-birth per FET 0.13 0.13 Beta 2015)

Transition probabilities after failed treatment a Progression to fresh cycle - 0.41 Beta (Chambers, et al., 2017) Progression to FET cycle 0.42 0.21 Beta (Chambers, et al., 2017)

(De Neubourg, et al., 2015, Harris, et al., 2016, Luke, et al., 2012, McLernon, et al., 2016, Smith, Drop-out (fresh and FET cycle) 0.17 0.24 Beta et al., 2015)

195

Table 9 (Continue) Probability estimates Distribution c References ART cycle 1 ART cycle 2 PGD-A strategy Fresh cycle Initiated cycle reaching OPU 0.88 0.88 Beta (Harris, et al., 2016) OPU cycles reaching embryo transfer 0.68 0.68 Beta (Rubio, et al., 2017) Live-birth per ET 0.53 0.42 Beta (Rubio, et al., 2017)

PGD-A FET cycle Embryo transfer per initiated FET cycle 0.88 0.88 Beta (Harris, et al., 2016) Live-birth per FET 0.53 0.42 Beta (Rubio, et al., 2017)

Transition probabilities after failed treatment a Progression to PGD-A FET cycle 0.13 0.13 Beta (Lee, et al., 2017) (De Neubourg, et al., 2015, Harris, et al., 2016, Luke, et al., 2012, McLernon, et al., 2016, Drop-out (fresh and FET cycle) 0.17 0.24 Beta Smith, et al., 2015)

Social egg Freezing Fresh cycle Initiated cycle reaching OPU 0.92 0.92 Beta (Harris, et al., 2016)

FET cycle Embryo transfer per initiated FET cycle 0.85 0.85 Beta (Cobo, et al., 2015) (Cobo, et al., 2015, Doyle, et al., 2016, Garcia- Live-birth per FET 0.23 0.23 Beta Velasco, et al., 2013, Hammarberg, et al., 2017) Transition probabilities after failed treatment a Progression to FET cycle 0.83 - Beta (Chambers, et al., 2017) (De Neubourg, et al., 2015, Harris, et al., 2016, Luke, et al., 2012, McLernon, et al., 2016, Drop-out (fresh and FET cycle) 0.17 - Beta Smith, et al., 2015)

196

Table 9 (Continue) Probability estimates Distribution c References

Donor ARTb ART cycle 1 ART cycle 2

FET cycle (using donor oocytes)

Embryo transfer per initiated FET cycle 0.84 0.84 Beta (Harris, et al., 2016) (Harris, et al., 2016, Hasson, et al., 2016, Live-birth per FET (using donated oocytes) 0.27 0.27 Beta Kawwass, et al., 2013, Wang, et al., 2011)

Transition probabilities after failed treatment a

Progression to FET cycle 0.42 - Beta (Chambers, et al., 2017) (De Neubourg, et al., 2015, Harris, et al., 2016, Luke, et al., 2012, McLernon, et al., Drop-out (FET cycle) 0.17 - Beta 2016, Smith, et al., 2015) a Failed treatment includes cancelled cycle (e.g. due to poor response), no embryo transferred or fails to achieve a live-birth b For donor ART strategy, the probability estimates for two ‘complete autologous ART cycles’ are similar to the estimates used in the autologous ART cycles strategy c Beta distribution is a type of probability function to model the uncertainty about the probability of success (e.g. live birth rate after PGD-A) of an intervention. It is used in this modelling study as beta probability is bounded by 0% and 100% (which is appropriate in the probability estimates used in the study. In the probability sensitivity analysis, the probability estimates of each parameter (e.g. 53% used in the live birth rate per ET after PGD- A) was replaced by beta distribution. During each of the 10,000 simulations, a value is randomly selected from the distribution. These values were used to compute the ICER and cost-effectiveness acceptability curve to determine if which ART strategy is most cost-effective relative to the standard ART strategy for a given willingness to pay.

197

9.5.4 Cost data sources

The direct cost estimates used in the Markov model are based on the cost-effectiveness analysis described in Chapter 8 Section 8.4.3. In Australia, the direct costs of ART treatment are covered by three payers - Medicare, private health insurers and patients.

Australia’s universal insurance scheme, Medicare, provides partial rebates for most

ART services through its Medicare Benefits Schedule (MBS), and ART medications through its Pharmaceutical Benefits Scheme (PBS). Private health insurance (PHI) provides rebates for eligible scheme members for ART procedures performed in a private hospital setting.

Patients usually incur out-of-pocket (OOP) expenses to cover the gap between MBS rebate and physician fees, the cost of procedures not covered by the MBS or PHI (e.g.

PGD-A), and the cost of some medications not listed on the PBS. A summary of cost estimates used in the model is listed in Table 10, including the unit costs and the distribution of the costs. In this present study, costs are expressed in 2015 Australian dollars (AUD).

198

9.5.5 Perspective of the study

One of the most important considerations in an economic evaluation is the perspective of the study because this determines the costs and benefit that are included in the study, and also who the decision maker is. An overview of the different ‘perspectives of study’ was addressed in Chapter 5 Section 5.4.1 which covers the economics and ART.

In this study, the economic evaluation was undertaken from a patient and healthcare perspectives. In the healthcare perspective, all direct costs associated with the ART strategies in the model were included, and reflected the payments made by the government, private health insurers and patients.

In the patient perspective, only OOP expenses associated with the use of treatment were included to evaluate which ART strategy represented value for money from a patient perspective.

199

Table 10: Cost associated with an ART treatment cycle and distribution, 2015

Cycle Type Distribution Unit cost (AUD) Unit cost (AUD) Direct healthcare g Patient perspective h Fresh cycle cancelled before oocyte retrieval a e Gamma 4,743 2,685

Fresh cycle proceeding to oocyte retrieval but cancelled before embryo transfer b Gamma 10,015 3,204

Completed fresh ART cycle c Gamma 10,614 3,204

Failed frozen/thaw embryo transfer Gamma 2,226 1,440

Completed frozen/thaw embryo transfer d Gamma 2,321 1,440

PGD-A (per cohort of embryos) f Gamma - 2,550

Embryo cryopreservation and storage Gamma - 180 a Cost includes planning and management, semen preparation, use of drugs to induce superovulation, ultrasound examinations b Cost includes planning and management, semen preparation, use of drugs to induce superovulation, ultrasound examinations, laboratory services, procedure for oocyte retrieval, c Cost includes planning and management, use of drugs to induce superovulation, ultrasound examinations, intracytoplasmic sperm injection (ICSI) and preparation of semen, laboratory services, procedure for oocyte retrieval, embryo transfer d Cost includes planning management, preparation and transfer of frozen embryo e Cancelled cycle refers to a ‘superovulated’ cycle that is cancelled prior to oocyte retrieval. f PGD-A procedure involves using blastomere and array CGH to select and transfer euploid embryos during fresh and FET cycle g Direct healthcare cost include MBS benefit for relevant ART services, PBS subsidy for prescribed medication for a fresh ART cycle and the charges for theatre and accommodation and anaesthetist fee paid by PHI and patient’s out of pocket expenses. Unit costs are rounded off to the nearest tenth for calculation. All costs are expressed in 2015 Australian dollar (AUD). PGD-A= Preimplantation genetic diagnosis for aneuploidy MBS= Medical Benefit Schedule, PBS=Pharmaceutical Benefits Scheme, PHI-Private Health Insurance. The MBS benefit in this table is based on the initial fresh ART treatment cycle performed in that calendar year h Patient perspective includes patient’s out-of-pocket expenses for ART treatment

200

9.5.6 Discounting

Discounting is the adjustment that is made in economic evaluations to ensure that the costs and or outcomes which tend to occur at different points in time are expressed on a common basis in terms of their net present value (Drummond, et al., 2015). As per the

Australian Pharmaceutical Benefits Advisory Committee (PBAC) guidelines, a discount rate of 5% was applied to both costs and outcomes to provide the net present value at age 40 years (PBAC Guidelines, 2016).

9.5.7 Clinical and cost-effectiveness analysis

The primary clinical outcome measure is CLBR per strategy. The economic evaluation measures are the mean costs and CLBR per strategy, and the incremental cost- effectiveness ratio (ICER). The ICER is calculated as the ratio of incremental costs and outcomes between alternate strategies (i.e. PGD-A, social egg freezing and donor) and the autologous ART strategy (reference strategy) and reflects the extra cost needed to obtain an additional live-birth by adopting the alternative strategy over the reference strategy, as expressed in the formula below (Drummond, et al., 2015).

퐶표푠푡 − 퐶표푠푡 ICER = 푎푙푡푒푟푛푎푡푒 푡푟푒푎푡푚푒푛푡 푟푒푓푒푟푒푛푐푒 푠푡푟푎푡푒푔푦 퐸푓푓푒푐푡 푎푙푡푒푟푛푎푡푒 푡푟푒푎푡푚푒푛푡− 퐸푓푓푒푐푡푟푒푓푒푟푒푛푐푒 푠푡푟푎푡푒푔푦

9.5.8 Sensitivity analysis

The purpose of economic evaluation is to inform decision making about the efficient allocation of resources. Like any statistical measure, the point estimate obtained from

ICER will carry a level of uncertainty (Briggs, et al., 2006).

Parameter uncertainty reflects the uncertainty and imprecision surrounding the model inputs values such as transition probabilities, costs and effects (Petrou and Gray, 2011).

The level of uncertainty relates to the precision and applicability of the source data.

201

For example, results of a RCT typically have high internal validity, however the results may not be reproducible in a real-world clinical setting (i.e. generalisability) (Kunz and

Oxman, 1998)

a) Probabilistic sensitivity analysis (PSA)

Sensitivity analysis helps to quantify this degree of uncertainty. In the present study, probabilistic sensitivity analysis (PSA) was conducted for the base case results (PGD-A, social egg freezing and donor ART strategies versus autologous ART strategy). PSA involves characterising parameter uncertainty using prior distribution on the input values to capture the degree of uncertainty around the point estimates. In the present study, Beta distributions were fitted around the probabilities because they can take values between 0 and 1 and, cost parameters were fitted with Gamma distributions to represent the positive and right-skewed values typically in healthcare costs (Briggs, et al., 2006).

After the distributions were fitted, model parameters were drawn randomly from their distributions and the results were the joint distributions of costs and outcomes for different strategies. These results were then plotted on the cost-effectiveness plane to graphically inspect the uncertainly around the ICER. The cost-effectiveness plane consists of four quadrants where the horizontal line (i.e. x-axis) represents the incremental level of effectiveness and the vertical line (i.e. y axis) represents the additional cost of undertaking the alternate strategy. This is illustrated in Figure 14.

The left upper quadrant of the cost-effectiveness plane represents the cases in which the alternate strategy is less effective and more costly than the reference strategy (i.e dominated by the reference strategy); the right upper quadrant represents the cases where the alternate strategy is more effective and more costly, left lower quadrant represents the cases where the alternate strategy is less effective and less costly; the right lower quadrant

202

represents the cases where the alternate strategy is more effective but less costly than the reference strategy (i.e dominant).

Figure 14: Cost-effectiveness plane

A cost-effectiveness acceptability curve (CEAC) was used to illustrate the probability that an alternate strategy is cost-effective compared to a reference strategy over a range of cost-effectiveness threshold values (or willingness to pay values) (O’Brien and Briggs,

2002).

b) One-way sensitivity analysis

Besides PSA, one-way sensitivity analysis was also performed due to the paucity of published data and relative uncertainty around the effectiveness of social egg freezing. In the social egg freezing strategy, the number of OPUs varied from one to two to test the

203

uncertainty around the optimal number of OPUs required to generate sufficient oocytes to perform two FET cycles.

All economic and sensitivity analysis were performed in TreeAge Pro Software 2017

(TreeAge Software Inc Williamstown MA) and STATA software version 11.2 (Stata

Corp, College Station, TX).

9.5.9 Ethical approval

This study has Human Research ethics approval from the University of New South Wales,

Human Research Ethics Advisory Panel 1 (Social and Health Research).

9.6 Results

Table 11 summarises the results of the clinical and cost-effectiveness results.

9.6.1 Costs and clinical-effectiveness

The mean cost for the reference autologous ART strategy was AUD 13,681 (AUD 10,578

– AUD 17,403), PGD-A strategy was AUD 14,803 (AUD 11,610– 18,581), social egg freezing strategy was AUD 14,941 (AUD 10,992- 19,368) and AUD 20,773 (AUD

15,182–28,581) for donor ART strategy.

All alternative strategies were associated with a higher CLBR compared with the reference autologous ART strategy (standard ART in women aged 40 years). The PGD-

A strategy resulted in the highest CLBR of 37.5%, followed by 28.0% with donor ART strategy and 25.5% with social egg freezing strategy. Among the four ART strategies, reference autologous ART strategy has the lowest CLBR of 15.5%.

204

9.6.2 Cost-effectiveness (healthcare perspective)

The expected mean cost, live-birth rates and ICER of the three alternative strategies

(PGD-A, social egg freezing and donor ART) were compared with the reference autologous ART strategy, from a healthcare perspective.

205

Table 11: Base case estimates (healthcare perspective) Types of strategy Mean cost AUD Cumulative live- Mean cost Cumulative ICER (AUD) (95%CI) birth rate (95%CI) difference live-birth rate ΔCost/Δeffectiveness (AUD) difference (%) (95%CI)*

Autologous ART (reference strategy) 13,681 (10, 578 – 17,403) 15.5 (9.8 to 24.3)

PGD-A 14,803 (11,610–18,581) 37.5 (19.3 to 53.4) 1,122 22.0 5,100 (53,171 to dominated**) 12,600 (263,234 to dominated Social egg freezing 14,941 (10,992- 19,368) 25.5 (11.5 to 41.4) 1,260 10.0 ***)

Donor ART 20,773 (15,182–28,581) 28.0 (16.1 to 41.8) 7,092 12.5 56,736 (22,664 to 280,046) * 95% confidence interval (CI) is the bootstrap results based on 10,000 simulations ** Dominated: PGD-A strategy was more costly and less effective than the reference autologous ART strategy *** Dominated: Social egg freezing strategy was more costly and less effective than the reference autologous ART strategy

206

1. PGD-A strategy versus autologous ART

Table 11 showed that the PGD-A strategy was more costly (AUD 14,803 with PGD-A strategy versus AUD 13,681 with reference autologous ART) but more than twice as clinically effective compared to the reference autologous ART (CLBR of 37.5% with

PGD-A strategy versus 15.5% with reference autologous ART). This yielded an ICER of

AUD 5,100 for one additional live-birth with the PGD-A strategy which means that if decision makers adopted PGD-A, they would have to invest an additional AUD 5,100 for a live birth. However, as the 95% confidence interval (95%CI) of the ICER for PGD-A strategy ranged from AUD 53,171 to being dominated by the reference autologous ART strategy, this implies that there is a high level of uncertainty with some probability that the PGD-A strategy is more costly and less effective compared to the reference autologous ART strategy.

Probabilistic sensitivity analysis for PGD-A is illustrated in the cost-effectiveness plane in Figure 15 which shows the results from 10,000 simulations of the PGD-A strategy versus the reference autologous ART strategy. Most of simulations are located in the right quadrants of the cost-effectiveness plane which indicates that the PGD-A strategy is likely to be more effective compared to the reference autologous ART strategy. As there are some simulations (1% of total simulations) located in the left upper quadrant, there is a probability that PGD-A is less effective and more costly than the reference ART strategy

(i.e. being dominated by the reference autologous ART strategy). The wide dispersion of simulations above and below the horizontal axis indicates that there is uncertainty about whether the incremental effect (i.e. additional live-birth) from the PGD-A strategy is achieved at a lower or higher cost.

207

Incremental cost-effectiveness plane (PGD-A vs Reference Autologous ART strategy)

6000

4000

2000

Incremental(AUD) cost -2000

-.1 0 .1 .2 .3 .4 .5 Incremental effect (CLBR)

Figure 15: Cost-effectiveness plane for PGD-A versus reference autologous ART strategy

2. Social egg freezing strategy versus autologous ART

Similarly, a strategy with social egg freezing led to a higher live-birth rate compared to

the reference autologous ART (CLBR 25.5% and 15.5% respectively), resulting in an

ICER of AUD 12,600 shown in Table 11. The 95%CI for social egg freezing strategy

ranged from AUD 263,234 to being dominated by the reference autologous ART strategy

which means that there is considerable uncertainty in the ICER resulting in some

probability that social egg freezing strategy is more costly and less effective than the

reference autologous ART strategy.

This is illustrated in the cost-effectiveness plane in Figure 16 which presents the results

from 10,000 simulations of social egg freezing strategy versus the reference autologous

ART strategy. The joint probabilities for social egg freezing are distributed over all four

quadrants, indicating some level of uncertainty around its cost-effectiveness (Fenwick, et

al., 2001). Specifically, 8.3% of simulations are located in the left upper quadrant,

indicating that it is likely that the social egg freezing strategy is less effective and more

costly than the reference autologous ART strategy.

208

Incremental cost-effectiveness plane (Social egg freezing vs Reference Autologous ART strategy)

15000

10000

5000

Incremental(AUD) cost -5000

-.2 0 .2 .4 Incremental effect (CLBR)

Figure 16: Cost-effectiveness plane for social egg freezing strategy versus reference autologous ART strategy

The one-way sensitivity analysis found that when two OPUs were assumed to be required

to obtain sufficient oocytes for two FET cycles at age 40, a strategy of social egg freezing

becomes the most costly strategy in the model, increasing the ICER to AUD 123,090 for

an additional live-birth.

209

3. Donor ART strategy versus autologous ART

Table 11 shows that the donor ART strategy achieved a higher CLBR rate compared to

the reference autologous ART (CLBR 28.0% versus 15.5% respectively) and thus,

resulted in a much higher ICER of AUD 56,736 (95%CI 22,664 to 280,046 AUD) for an

additional live birth. This is likely due to a higher treatment cost associated with the donor

ART strategy.

Figure 17 shows the results from 10,000 simulations of donor ART strategy versus the

reference autologous ART strategy. As all simulations are located in the right upper

quadrant of the cost-effectiveness place, this indicates that the donor ART strategy is

more effective but more costly than the reference autologous ART strategy.

Incremental cost-effectiveness plane (Donor ART vs Reference Autologous ART strategy)

15000

10000

5000

Incremental(AUD) cost 0

-.1 0 .1 .2 .3 .4 .5 Incremental effect (CLBR)

Figure 17: Cost-effectiveness plane for donor ART strategy versus reference autologous ART strategy

210

b) Cost effectiveness acceptability curve

Cost-effectiveness acceptability curve (CEAC) shows the probability that an alternate strategy is cost-effective compared to the reference strategy over a range of threshold values. The shape of the CEAC depends on the data on the CE plane (Fenwick, et al.,

2001). In the study, if the society is willing to pay $50,000 for an ART-conceived child, the model results showed that PGD-A has an 80% probability of being cost-effective while other strategies including the reference strategy in the model have the likelihood of being cost-effective in less than 20% of cases at the same cost-effectiveness threshold

(Figure 18).

Figure 18:Cost-effectiveness acceptability curves over a range of willingness to pay thresholds

211

9.6.3 Cost-effectiveness (patient perspective)

In addition to the healthcare perspective, the cost-effectiveness analysis was repeated based on OOP expenses for the four ART strategies. In Australia, patients pay OOP expenses of approximately one-third of the cost of an autologous ART cycle (fresh and

FET cycle) and all other direct costs associated with alternative strategies (e.g. embryo selection with PGD-A, OPU to retrieve and store oocytes to prevent age-related infertility and donated oocytes) (Chambers, et al., 2012).

Table 12 shows that the reference autologous ART strategy is associated with the lowest

OOP expenses at AUD 7,149 for a CLBR of 15.5%. The PGD-A strategy was associated with a slightly higher OOP expense of AUD 8,903 for a much higher CLBR of 37.5%.

This is followed by the donor ART strategy with an associated OOP expense of $11,100 for a CLBR of 28.0%.

In contrast to the results reported from a healthcare perspective, the social egg freezing strategy incurred the highest OOP expense of AUD 14,703 with a CLBR of 25.5% in the model. The OOP expenses associated with the social egg freezing strategy were almost twofold higher than the reference autologous ART strategy (AUD 14,703 versus AUD

7,149). Consequently, it altered the relative ranking of the ART strategies in Table 12 as the social egg freezing strategy yielded the highest ICER of AUD 75,540 (95%CI AUD

820,599 to being dominated by reference autologous ART strategy for a live birth) compared to the PGD-A strategy (AUD 7,972 95%CI 905 to 63,406 AUD) and the donor

ART strategy (AUD 31,608 95%CI 12,118 to 179,815 AUD).

Thus, from a patient perspective, the PGD-A strategy remains the most cost-effective strategy in the model due to the relatively lower OOP expenses for a live birth whereas the social egg freezing strategy incurred the highest ICER for a live birth.

212

Table 12: Base case estimates (patient perspective)

Types of strategy Mean OOP expenses Cumulative live- Mean OOP Cumulative ICER AUD (95% CI) * (AUD) (95%CI) birth rate (95%CI) expenses live-birth difference rate (AUD) difference (%) Autologous ART (reference strategy) 7,149 (5,495 to 9,163) 15.5 (9.8 to 24.3)

PGD-A 8,903 (7,101 to 11,039) 37.5 (19.3 to 53.4) 1,754 22.0 7,972 (905 to 63,406) Donor ART 11,100 (8,059 to 14,848) 28.0 (16.1 to 41.8) 3,951 12.5 31,608 (12,118 to 179,815) 14,703 (10,900 to Social freezing 18,979) 25.5 (11.5 to 41.4) 7,554 10.0 75,540 (820,599 to dominated**) OOP refers to out-of-pocket expenses associated with different ART strategy *95% confidence interval (CI) is the bootstrap results based on 10,000 simulations **Dominated: Social freezing strategy was more costly and less effective compared to the reference autologous ART strategy

213

9.7 Discussion

The study in this chapter compared four commonly used ART strategies to determine the most clinically and cost-effective ART strategy for women aged 40 years following 6 to

12 months of infertility. Although several studies have evaluated the cost-effectiveness of different ART treatment strategies individually (Collins, et al., 2017, Griffiths, et al.,

2010, van Loendersloot, et al., 2011), this is the first study to use a Markov decision- analytic model to incorporate and synthesise the most contemporary published data from the literature to assess which of the ART strategies (PGD-A, freezing oocytes at age 35 and returning at age 40 years to use the stored oocytes (social egg freezing strategy), ART using donated oocytes, and standard autologous ART) is most clinically and cost- effective strategy.

The results showed that alternate ART strategies – PGD-A strategy, social egg freezing strategy and donor ART strategy were associated with higher treatment costs and higher

CLBR compared with reference autologous ART. The CLBR and mean costs of each strategy were: PGD-A: 37.5% and AUD 14,803; social freezing: 25.5% and AUD 14,941; donor ART: 28.0% and AUD 20,773; and autologous ART: 15.5% and AUD 13,681.

Compared to autologous ART, the ICER for one additional live-birth with PGD-A strategy was AUD 5,100 (95%CI AUD 53,171 to being dominated by the reference autologous ART strategy), social egg freezing strategy was AUD 12,600 (95%CI 263,234 to being dominated by the reference autologous ART strategy) and the donor ART strategy was AUD 56,736 (95 % 22,664 to 280,046 AUD).

The cost-effectiveness from a patient perspective showed that PGD-A incurred the lowest

OOP expenses resulting in an ICER of AUD 7,972 (95%CI 905 to 63,406) to achieve a live birth while the social egg freezing strategy incurred the highest additional cost for a

214

live birth (AUD 75,540 ,95%CI 820,599 to being dominated by reference autologous

ART). As the donor ART strategy incurred a lower OOP expenses and had a higher CLBR compared to the social egg freezing strategy, it had a lower ICER of AUD 31,608 (95%CI

12,118 to 179,815 AUD) for a live birth compared to the social egg freezing strategy.

As described in Chapter 5 on the economics and ART, there is no cost-effectiveness threshold value (or range of values) that represents society’s (or the government’s) willingness to pay for an additional live birth. However, as medical interventions that result in a ICER of up to AUD 50,000 per quality adjusted life years gained are considered cost-effective within the Australian setting (Harris, et al., 2008), PGD-A has the highest probability of being cost-effective compared to the autologous ART strategy, social egg freezing strategy and donor ART strategy for 40-years old women with infertility from both an individual and healthcare perspectives.

Although no published studies have assessed the relative cost-effectiveness of four commonly used ART strategies in one analysis, the results from the present study are consistent with previous findings reported in Chapter 8 which used patient-level data to assess the cost-effectiveness of PGD-A in infertile women aged 37 years or over. The study in Chapter 8 reported that PGD-A for up to three ‘complete ART cycle’ is more cost-effective compared to morphological assessment of embryos alone from a healthcare perspective.

A similar finding was reported in another cost-effectiveness study that concluded that

PGD-A is cost-effective in women aged over 37 years from the US healthcare perspective

(Collins, et al., 2017). This study reported an incremental cost of US$105,489 for an additional live birth with PGD-A compared to using morphological assessment of embryos alone.

215

Although Australia has supportive funding arrangements for ART, the cost of PGD-A, donor oocyte cycle and OPU to retrieve oocytes for future use in the social egg freezing are not publicly funded (i.e. do not attract a Medicare benefit). This means that the cost associated with undertaking these alternate strategies are borne by the patient. The additional OOP expenses will likely have an impact on the assessment of the cost- effectiveness of ART strategy and treatment decisions (De Neubourg, et al., 2010).

This is shown in the present study where the lack of public funding for the alternate strategies increased the OOP expenses and altered the cost-effectiveness ranking of ART strategies between the two perspectives (a patient and healthcare). In the study, as the social egg freezing strategy incurred the highest OOP expenses and had the lowest CLBR compared to other competing ART strategies in the model, it yielded the highest ICER for a live birth from a patient perspective.

The relative higher OOP expenses (due to the lack of public funding) for the costs associated with the social egg freezing strategy means that this strategy may be prohibitive for some women. In a recent US internet-based survey to assess women willingness to pay for oocyte cryopreservation to preserve fertility found that among women who were likely to consider oocyte cryopreservation to preserve fertility (n=216, median age 35.0 years), the amount that these women are willing to pay for the procedure

(US$ 3,811.55) is significantly lower than the average cost of ART treatment in the US

(Milman, et al., 2017).

The present study used the findings from a single RCT comparing PGD-A (day-3 blastomere biopsy and aCGH) with morphological assessment of embryos alone for the model input in the PGD-A strategy. Although some may question the use of clinical results of blastomere biopsy and aCGH as the technique is increasingly being replaced in

216

the clinical setting by trophectoderm biopsy and next-generation sequencing (NGS), it should be highlighted that to date, there remains a paucity of evidence from well-designed trials on the cumulative effectiveness of NGS particularly in older women to inform the model for this study (Sermon, 2017, Sermon, et al., 2016).

9.7.1 Strength of the study

This is the first study to systematically assess the clinical and cost-effectiveness of alternate ART treatment strategies in women aged 40 years following 6 to 12 months of infertility. While an RCT provides the highest level of evidence, they are difficult to undertake in fertility treatment. For example, a single RCT may not be able to compare all available treatment options, incorporate all the relevant inputs (costs and effects) or be carried out over multiple cycles to measure the difference in costs and cumulative live birth rates between the different treatment arms. The Markov model overcomes these limitations by providing an appropriate structure which will reflect the possible outcomes that patients will experience during the entire ART treatment course over multiple cycles.

This study has used the most contemporary evidence from different sources including

RCT and population-based data and undertook extensive sensitivity analysis to facilitate the comparison of different ART strategies in a Markov model. Therefore, it provided decision-makers with the best information regarding the most cost-effective treatment strategy for women aged 40 years. In particular, the social egg freezing strategy utilised outcomes from a recently published study of non-infertile women with a mean age of approximately 40 years who returned to use their stored oocytes that were vitrified at a mean age of 37.7 years to prevent age-related infertility which closely aligns with our study strategy (Cobo, et al., 2015).

217

This overcomes the limitation of previous studies that used live-birth rates of donor oocyte cycles or frozen/thawed oocyte cycles from infertile women treated at fertility clinics for the model input in the social egg freezing strategy in their economic decision models (Devine, et al., 2015, Hirshfeld-Cytron, et al., 2012, van Loendersloot, et al.,

2011). This has been recognised as an important research gap in cost-effectiveness analysis of social egg freezing in a recent editorial commentary (Stoop, 2016). This model also utilised the first RCT results of PGD-A in older women to inform the model (Rubio, et al., 2017). The high quality and applicability of the evidence used in the model strengthened the credibility of the findings.

9.7.2 Limitation of the study

The results of model-based economic evaluation are always dependent on the model structure and assumptions made (Briggs, et al., 2006). Although this model has considered important events such as cancelled cycle and drop-outs based on real-world clinical setting, models are never perfect as they rely on simplifications of real events.

For example, in reality, women may ‘cross-over’ to standard autologous ART cycle for subsequent cycles after failing to achieve a live birth with PGD-A in their initial ART cycle. However, it is assumed that the lack of variation in the model structure is not expected to influence the overall outcomes as the main clinical pathways of the different

ART strategies were included in the model structure.

9.8 Summary

This study addresses a common clinical and policy question regarding which ART treatment approach offers infertile women of advance maternal age, the most clinically and cost-effective strategy to achieve the live birth of a child.

218

Based on an individual and healthcare perspectives, this study found that PGD-A is the most clinically and cost-effective ART strategy compared to other commonly used ART strategies − standard ART autologous, social egg freezing and donor ART strategies, for infertile women aged 40 years with 6 to 12 months of infertility,

As clinical and laboratory techniques for ART continue to evolve and improve, clinicians and infertile couples are increasingly presented with treatment options that are often more costly than those conventionally available. Therefore, considerations of both the clinical and cost-effectiveness of alternative approaches to treatment are important to inform funding decisions by governments and treatment choices for patients. Cost-effectiveness methods provide a useful tool for policy-makers to balance this trade-off between costs and outcomes to ensure that treatment provide value for money.

219

Chapter 10

Discussion

220

Chapter 10- Discussion

10.1 Overview

The overarching aim of this doctoral research program was to contribute to the evidence on the clinical and cost-effectiveness of PGD-A for infertile women of advanced maternal age. This final chapter summarises the main findings of the four studies and describes how these studies have contributed to the overall aim of this doctoral research. The chapter concludes with a general discussion on the direction for future research.

10.2 Main findings of the study

This thesis comprises four related studies. The first study was a systematic review of the current evidence regarding the clinical effectiveness of PGD-A. This review informed the subsequent three empirical studies that generated new evidence on the clinical and cost- effectiveness of PGD-A in infertile women of advanced maternal age. a) Study One: The clinical effectiveness of PGD-A in all 24 chromosomes: systematic review

This review identified 19 studies [three randomised controlled trials (RCTs) on good- prognosis patients and 16 observational studies] that were published up until 2014. The overarching finding from the available evidence was that PGD-A is associated with higher implantation and clinical pregnancy rates in single ART cycle compared to morphological assessment of embryos alone. However, high-level evidence was limited to the effectiveness in young and good-prognosis women measured by implantation rate or pregnancy rate in a single cycle. The findings of the systematic review have been published in Human Reproduction in 2014 (Lee, et al., 2014).

221

To update these findings, a further systematic review of the literature was undertaken for studies published between 2014 and 2017. This updated review identified one RCT conducted in women of advanced maternal age and six observational studies.

The overall finding from the original and updated review of the literature was that PGD-

A improved implantation rate and per-cycle live-birth rate compared to the morphological assessment of embryos alone in both women of advanced age and young, good-prognosis women. However, Level-1 evidence was only available from four RCTs — three in young and good-prognosis women and one in women of advanced maternal age — and only the latter RCT has assessed the cumulative effectiveness of PGD-A.

The updated review identified only two studies (including one RCT) that reported on the cumulative live-birth rate (CLBR), which includes both fresh and subsequent frozen embryo transfer (FET) cycles, both of which reported comparable CLBR for PGD-A versus conventional morphological assessment of embryos alone (Rubio, et al., 2017,

Ubaldi, et al., 2015). The RCT reported PGD-A reduced the overall time and number of

ART cycles to reach a clinical pregnancy leading to a live birth in older women (Rubio, et al., 2017).

The updated review identified only one study that assessed the cost-effectiveness of PGD-

A (Collins, et al., 2017), concluding that PGD-A is cost-effective compared to morphological assessment of embryos alone in women aged over 37 years from a healthcare perspective. However, the economic model was based on a cost-effectiveness analysis for a single ART cycle and did not account for CLBR.

In conclusion, the available published evidence highlights a paucity of high-quality studies on PGD-A to support the clinical effectiveness of PGD-A, particularly when considered over multiple cycles. Current evidence in the literature shows that PGD-A is

222

associated with better clinical outcomes for a single ART cycle; however, the CLBR is comparable to morphological assessment of embryos alone. The conflicting evidence on the benefit of PGD-A measured by the success rate for a single ART cycle but potentially similar CLBR, presents a challenge for clinicians and patients in deciding whether, overall, PGD-A represents a more effective treatment strategy than conventional embryo selection methods. Furthermore, whether PGD-A is cost-effective and represents ‘good value for money’ from an individual and healthcare perspectives remains largely unknown. The subsequent empirical studies conducted in this doctoral research program aimed to address these research gaps. b) Study Two: ART cumulative live-birth rates following PGD-A or morphological assessment of embryos: a cohort analysis

Study Two is the first of three empirical studies to assess the clinical and cost- effectiveness of PGD-A.

The aim of this retrospective cohort analysis was to assess the ‘real-world’ clinical effectiveness of PGD-A (PGD-A group) compared to the morphological assessment of embryos alone (morphology assessment group) in ART-naïve women aged 37 years or over, attending a large Australian fertility clinic. The primary outcome measure was the

CLBR for up to three ‘complete ART cycles’ (all embryo transfers from a single stimulation). Secondary outcomes included the number of ART cycles and mean time to achieve a clinical pregnancy resulting in a live birth.

Based on the intention-to-treat analytical approach, the study found that women who commenced ART using PGD-A achieved a higher per-cycle live-birth rate (14.47% in the PGD-A group versus 9.12% in the morphology assessment group, p<0.01), a lower rate of pregnancy loss (19.51% in the PGD-A group versus 34.78% in the morphology

223

assessment group, p<0.05), took a shorter mean time to achieve a clinical pregnancy leading to a live birth (104.8 days in the PGD-A group versus 140.6 days in the morphology assessment group, p<0.05) and required fewer ART cycles to achieve a live birth (6.91 cycles in the PGD-A group versus 10.96 cycles in the morphology assessment group, p<0.01). However, a key finding from this study was that after three complete

ART cycles, the CLBR was comparable for the two study groups (30.90% in the PGD-A group versus 26.77% in the morphology assessment group, p=0.34).

This is the first study to provide a substantial longitudinal analysis of the effectiveness of

PGD-A. The study was therefore able to provide clinically important endpoints, i.e.

CLBR, number of cycles needed to treat, and time to pregnancy leading to a live birth, and thus provides a comprehensive analysis of the role of PGD-A for women of advanced maternal age.

A longitudinal analysis of PGD-A is more relevant than assessing its effectiveness in a single cycle because with modern vitrification methods there are compelling arguments that unselected embryos should be sequentially transferred in ART to avoid discarding embryos that are viable but are screened positive (e.g. mosaic embryos and false positive euploid embryos). However, the additional time and costs to undertake repeated embryo transfer procedures with unselected embryos may be detrimental to the overall chance of treatment success, particularly in women of advanced maternal age who have a high risk of miscarriage and limited reproductive time for repeated treatment failure.

While this study was based on real-world observational data and cannot infer causality, it complements evidence obtained from RCTs on the efficacy of PGD-A. It is well recognised that the Level-1 evidence generated by RCTs is the best measure of efficacy and, if appropriately performed, is the best study design to detect an effect if one exists

224

(Barton, 2000, Harper, et al., 2017). However, RCTs are often not generalizable to real- world clinical settings and tend to over-estimate effects (Rothwell, 2005). For this reason, and the fact that women were followed up for up to three ‘complete ART cycles’, this cohort study provides important information for the role of PGD-A in clinical practice.

This study has been published in the Australian and New Zealand Journal of Obstetrics and Gynaecology in 2017 (Lee, et al., 2017). c) Study Three: Longitudinal analysis of PGD-A in women of advanced maternal age: a cost-effectiveness analysis

The aim of this study was to use patient-level data to assess the cost-effectiveness of

PGD-A compared to morphological assessment of embryos alone. This study extended the analyses conducted in Study Two by applying health technology assessment methods to evaluate whether undertaking PGD-A over multiple ART cycles represented value for money compared to the morphological assessment of embryos alone, from an individual and healthcare perspectives.

The economic analysis study found that, from a healthcare perspective, the average treatment cost of up to three ‘complete ART cycles’ for women who commenced ART with PGD-A (PGD-A group) was 22,962 Australian dollars (AUD) compared to AUD

21,801 for women who commenced ART with morphological assessment of embryos alone (morphology assessment group).

The CLBR for up to three ‘complete ART cycles’ reported in Study Two was 30.90% in the PGD-A group and 26.77% in the morphology assessment group. When this CLBR was combined with the corresponding treatment costs, it yielded an incremental cost- effectiveness ratio (ICER) of AUD 28,103 with PGD-A. This represents the additional cost for a live birth with PGD-A compared to morphological assessment of embryos alone

225

from a healthcare perspective. Whether or not an ICER of AUD 28,103 is cost-effective depends on society’s willingness to pay for a live birth from fertility treatment. While there is no such threshold value to determine what society is willing to pay for a live birth, an ICER of up to AUD 50,000 per quality-adjusted life-year (QALY) gained from medical interventions is considered cost-effective within the Australian setting.

Therefore, given that the cost-effectiveness analysis found that a new life from PGD-A resulting in an ICER of AUD 28,103, it is reasonable to conclude the society would consider PGD-A to be highly cost-effective.

The cost-effectiveness analysis was repeated using only the out-of-pocket (OOP) expenses incurred by an individual to provide a patient perspective on the cost- effectiveness of PGD-A. From an individual perspective, women in the PGD-A group incurred a higher OOP expense of AUD 9,913 compared to AUD 7,340 with the morphological assessment group. When the treatment costs are combined with the CLBR for up to three complete ART cycles, it yielded an ICER of AUD 62,270 for an additional live birth using PGD-A. Whether AUD 62,270 is cost-effective, and thus considered good value for money, from a patient’s perspective would depend on their own willingness and ability to pay.

The higher cost associated with PGD-A reported in the study was largely explained by the additional cost of the PGD-A procedure and a higher proportion of fresh ART cycles undertaken by the PGD-A group. Although, overall, the PGD-A group undertook fewer cycles, this did not compensate for the additional cost of fresh ART cycles, nor for the cost of the PGD-A procedure compared to women using morphological assessment of embryos alone.

226

This is the first economic evaluation study to utilise patient-level data to assess the cost- effectiveness of PGD-A in older women. The study used a detailed ‘bottom-up’ costing method and an extensive audit of clinical data to inform the economic analysis. The results provide real-world evidence of the cost-effectiveness of PGD-A. d) Study Four: Which ART treatment strategy is the most clinically and cost- effective for women of advanced maternal age? A Markov Model

Study Four developed and used a Markov model to compare the clinical and cost- effectiveness of four commonly used ART strategies in women aged 40 years following

6–12 months of infertility. The strategies in the Markov model were PGD-A, freezing oocytes at age 35 and returning at age 40 years to use the stored oocytes (social egg freezing), ART using donated oocytes and standard autologous ART.

Although a number of studies have evaluated the cost-effectiveness of individual ART treatment strategies, this is the first study to synthesise the most contemporary published data to specifically address the question of which alternative ART treatment strategies is most cost-effective in women of advanced maternal age.

The results showed that PGD-A, social egg freezing, and donor ART are associated with higher CLBRs and higher treatment costs compared with standard autologous ART. The

CLBRs and mean costs of each strategy were as follows: PGD-A: 37.5% and AUD

14,803; social egg freezing: 25.5% and AUD 14,941; donor ART: 28.0% and AUD

20,773; and autologous ART: 15.5% and AUD 13,681. Compared to autologous ART, the ICER from a healthcare perspective for one additional live birth was AUD 5,100 with

PGD-A, AUD 12,600 with social egg freezing and AUD 56,736 with donor ART.

227

The cost-effectiveness from a patient perspective also found that PGD-A yielded the lowest ICER to achieve a live birth of AUD 7,972, compared to the social egg freezing strategy, which incurred the highest ICER of AUD 75,540 for an additional live birth.

Although the results from patient and healthcare perspectives indicated a considerable degree of uncertainty, they suggest that PGD-A is likely to be considered the most cost- effective strategy compare to standard autologous ART, social egg freezing and donor

ART for infertile women aged 40 years of age.

This study used the most contemporary evidence on the costs and effectiveness of alternative ART strategies, including from the most recent RCT and population-based

ART registry data from Australia. With the increasing use of PGD-A, and social egg freezing in particular, this study can inform clinical practice of these novel treatments and inform policy makers about their relative cost-effectiveness.

The body of work presented within this thesis has expanded the evidence base regarding the clinical and cost-effectiveness of PGD-A for treating infertile women of advanced maternal age. This research found that in women aged 37 years or over, a strategy with

PGD-A led to better clinical outcomes in a single ART cycle, but when successive cycles are considered, the CLBR is comparable to the conventional morphological assessment of embryos alone.

Although PGD-A is a costly procedure, the economic analyses described in this thesis suggest that PGD-A is cost-effective from a healthcare perspective compared to the morphological assessment of embryos alone, as well as other competing strategies commonly used in older women.

228

10.3 Direction of future research a) Clinical and cost-effectiveness of PGD-A

Although the cohort study undertaken in Studies Two and Three provided important findings on the clinical and cost-effectiveness of PGD-A in a ‘real-world’ clinic setting, it remains a retrospective observational study and, as such, can only assess associations rather than causal relationships (Jepsen, et al., 2004). Observational studies are also vulnerable to bias and residual confounding, and thus they need to be interpreted with caution (Grimes and Schulz, 2002, Ioannidis, et al., 2001). To provide the highest level of evidence of the efficacy of PGD-A over successive cycles, a suitably-powered RCT is required (Barton, 2000, Harper, et al., 2017). A number of RCTs which largely evaluate

PGD-A on a single ART cycle are currently underway, but the full results have not been published in peer-reviewed forums. These include the ‘ESHRE Study Into The Evaluation of Oocyte Euploidy by Microarray Analysis’ (ESTEEM) (NCT1532284) and the ‘Single

Embryo TrAnsfeR of Euploid Embryo (STAR)’ study (NCT02268786) (Munne, et al.,

2017, Sermon, et al., 2016).

The scientific community is eagerly awaiting these results; however, initial conference presentations indicates that PGD-A is not superior to conventional morphological assessment of embryos over repeated cycles (Brown, 2017, Munne, et al., 2017).

However, one of the main problems with undertaking RCTs and observational studies evaluating PGD-A is that the technology is advancing so rapidly that the findings of the studies are often out of date by the time they are published given the time required for both recruitment of participants to the study and subsequent follow up over repeated ART cycles. Indeed, the cohort study described in this thesis was based on blastomere biopsy and array comparative genomic hybridization which has now been mostly replaced by

229

trophectoderm biopsy and next-generation sequencing (Sermon, 2017). Thus, future directions of research should include large multi-country, multi-site RCTs to enable rapid recruitment of participants so that Level 1 evidence can be obtained quickly and be generalizable across different settings.

It is also important to incorporate cost-effectiveness analysis and patient-reported outcome measures (PROMS) within future RCT protocols to inform funding decisions in

ART strategies. The incorporation of PROMS is now considered best practice in research that impacts on the patient’s experience of treatment (Black, 2013). This is the case for fertility treatment, which is highly emotive and often involves long periods of treatment for the patient. It is also particularly important for embryo selection techniques, such as

PGD-A, which can result in fewer cycles needed to achieve a clinical pregnancy resulting in a live birth (Lee, et al., 2017, Rubio, et al., 2017).

While RCTs are able to compare PGD-A to other embryo selection techniques, it would not be feasible and ethical to compare PGD-A to social egg freezing or donor ART particularly for multiple ART cycles scenarios and involving older women whose fertility potential decreases with age. In some cases, it will be necessary to use modelling techniques to assess their relative clinical and cost-effectiveness. However, to inform such analyses, comprehensive donor ART and oocyte cryopreservation registries are needed that record the future use of donated embryos and previously stored oocytes.

There are donor registries in some Australian states; however, a national approach that also records information on overseas donations is needed. Furthermore, although

Australian and New Zealand Assisted Reproduction Database (ANZARD) records oocyte cryopreservation cycles, it does not record the indication (medical or non-medical), making interpretation of the outcomes difficult.

230

In Australia, owing to the complexity of ART treatment, more sophisticated modelling techniques could also be used to assess treatment strategies that cannot be evaluated in a trial situation, including machine-learning algorithms and individual microsimulation sampling techniques method. b) Methodological issues in measuring the cost-effectiveness of fertility treatment

There is no cost-effectiveness threshold value (or range of values) that represents society’s (or the government’s) willingness to pay for an additional live birth. The most common unit of measurement in the health economic assessment is the QALY which combines the quality of life and the length of time spent in a particular health state into a single measure. However, the use of the QALY to quantify the benefits of fertility treatment, such as ART, is contentious (Devlin and Parkin, 2003, Klitzman, 2017)

QALYs are designed to capture gains in the quality and quantity of life of those living, whereas fertility treatment is primarily judged by the ability to create new life (McGregor and Caro, 2006). Consequently, this presents methodological challenges for the economic evaluation in fertility treatment and hinders the ability to explicitly compare the benefits of ART with other diseases/conditions to enable the objective comparison of resource allocation within an overall healthcare budget (Baird, et al., 2015).

With increasing pressure to meet higher demand for medical services with constrained healthcare budgets, cost-effectiveness analysis of medical interventions has become a necessary complement to inform stakeholders about whether the intervention demonstrates good value for money (Drummond, et al., 2015). This means that if the management of infertility is to become a more accepted part of public and private-funded healthcare systems, a consensus view on what constitutes ‘good value’ from fertility treatment is needed. In terms of methodological investigation, this is a priority for future

231

research. Techniques that can capture both health (e.g. birth of healthy baby) and non- health outcomes (e.g. family formation, attaining life goals) associated with fertility treatment should be used to quantify the willingness-to-pay thresholds. The most contemporary methodological approaches include the use of discrete choice experiments, which can capture both societal and patient preferences for characteristics of fertility treatment and by extension their willingness to pay for a live birth from fertility treatment.

10.4 Concluding remarks

The body of work presented within this thesis has helped to address the paucity of evidence on the clinical and cost-effectiveness of PGD-A for infertile women of advanced maternal age. This is important because PGD-A is increasingly being adopted in clinical practice with limited information on its indications for use, its value over repeated treatment cycles in terms of live birth rates, time to pregnancy and number of cycles needed to reach a live birth, and whether it represents value for money.

Taking the four studies together, this doctoral research program found that in women of advanced maternal age, PGD-A was clinically effective when considered from a single cycle perspective. When considered over repeated ART cycles, it achieved similar live- birth rates to conventional morphological selection of embryos alone. However, additional benefits of the PGD-A approach included a lower rate of pregnancy loss, shorter time and fewer cycles to reach a clinical pregnancy resulting in a live birth. In terms of cost-effectiveness, PGD-A was found to be cost-effective compared to the morphological assessment of embryos alone, and a cost-effective treatment strategy compared to autologous ART treatment, social egg freezing and donor ART.

232

References

1. 2017 Release of journal citation reports®. source: 2015 Web of Science Data.

2. Abdalla HI, Bhattacharya S, Khalaf Y. Is meaningful reporting of national IVF outcome data possible? Human reproduction 2009;25: 9-13.

3. Adamson GD, de Mouzon J, Lancaster P, Nygren K-G, Sullivan E, Zegers-Hochschild F, Technology ICfMAR. World collaborative report on in vitro fertilization, 2000. Fertility and sterility 2006;85: 1586-1622.

4. Adamson GD, Kennedy S, Hummelshoj L. Creating solutions in endometriosis: global collaboration through the World Endometriosis Research Foundation. J Endometriosis 2010;2: 3-6.

5. Adetoro O, Ebomoyi E. The prevalence of infertility in a rural Nigerian community. African journal of medicine and medical sciences 1991;20: 23-27.

6. Akar M, Oktay K. Restoration of ovarian endocrine function by ovarian transplantation. Trends in Endocrinology & Metabolism 2005;16: 374-380.

7. Alderson P, Chalmers I. Survey of claims of no effect in abstracts of Cochrane reviews. BMJ (Clinical research ed) 2003;326: 475.

8. Anderson K, Nisenblat V, Norman R. Lifestyle factors in people seeking infertility treatment–a review. Australian and New Zealand journal of obstetrics and gynaecology 2010;50: 8-20.

9. Armstrong S, Akande V. What is the best treatment option for infertile women aged 40 and over? Journal of assisted reproduction and genetics 2013;30: 667-671.

10. Australian Bureau of Statistics. 3301.0 Births, Australia, 2015. Accessed on Sept 2017. 2015, Canberra: Australian Bureau of Statistics.

11. Australian Government Department of Health and Ageing (DoHA). Schedule of Pharmaceutical Benefits, effective 1 January 2015 -31 January 2015. 2015. Canberra: DoHA.

12. Australian Institute of Health and Welfare. Australia’s mothers and babies 2015—in brief. Perinatal statistics series no. 33. Cat no. PER 91. . 2017, Canberra: AIHW.

13. Axmon A, Rylander L, Albin M, Hagmar L. Factors affecting time to pregnancy. Human Reproduction 2006;21: 1279-1284.

14. Azziz R, Marin C, Hoq L, Badamgarav E, Song P. Health care-related economic burden of the polycystic ovary syndrome during the reproductive life span. The Journal of Clinical Endocrinology & Metabolism 2005;90: 4650-4658.

15. Babayev SN, Park CW, Bukulmez O. Intracytoplasmic sperm injection indications: how rigorous? Seminars in reproductive medicine. 2014. Thieme Medical Publishers, pp. 283-290.

233

16. Baird DT, Barri PN, Bhattacharya S, Devroey P, Evers JLH, Gianaroli L, Somigliana E, Tapanainen JS, van Wely M, Crosignani PG et al. Economic aspects of infertility care: a challenge for researchers and clinicians. Human Reproduction 2015;30: 2243-2248.

17. Baker T. A quantitative and cytological study of germ cells in human ovaries. Proceedings of the Royal Society of London B: Biological Sciences 1963;158: 417-433.

18. Balasch J, Gratacós E. Delayed childbearing: effects on fertility and the outcome of pregnancy. Current Opinion in Obstetrics and Gynecology 2012;24: 187-193.

19. Baldwin K, Culley L, Hudson N, Mitchell H, Lavery S. Oocyte cryopreservation for social reasons: demographic profile and disposal intentions of UK users. Reprod Biomed Onine 2015;31: 239-245.

20. Barfield WD. Standard terminology for fetal, infant, and perinatal deaths. Pediatrics 2016;137: e20160551.

21. Barnhart KT. Assisted reproductive technologies and perinatal morbidity: interrogating the association. Fertility and sterility 2013;99: 299-302.

22. Barnhart KT. Live birth is the correct outcome for clinical trials evaluating therapy for the infertile couple. Fertility and sterility 2014;101: 1205-1208.

23. Barton S. Which clinical studies provide the best evidence?: The best RCT still trumps the best observational study. 2000. BMJ: British Medical Journal

24. Bensdorp A, Tjon-Kon-Fat R, Bossuyt P, Koks C, Oosterhuis G, Hoek A, Hompes P, Broekmans F, Verhoeve H, De Bruin J. Prevention of multiple pregnancies in couples with unexplained or mild male subfertility: randomised controlled trial of in vitro fertilisation with single embryo transfer or in vitro fertilisation in modified natural cycle compared with intrauterine insemination with controlled ovarian hyperstimulation. BMJ (Clinical research ed) 2015;350: g7771.

25. Bhattacharya S, Hamilton M, Shaaban M, Khalaf Y, Seddler M, Ghobara T, Braude P, Kennedy R, Rutherford A, Hartshorne G. Conventional in-vitro fertilisation versus intracytoplasmic sperm injection for the treatment of non-male-factor infertility: a randomised controlled trial. The lancet 2001;357: 2075-2079.

26. Bhattacharya S, Harrild K, Mollison J, Wordsworth S, Tay C, Harrold A, McQueen D, Lyall H, Johnston L, Burrage J. Clomifene citrate or unstimulated intrauterine insemination compared with expectant management for unexplained infertility: pragmatic randomised controlled trial. BMJ (Clinical research ed) 2008;337: a716.

27. Bhattacharya S, Porter M, Amalraj E, Templeton A, Hamilton M, Lee A, Kurinczuk J. The epidemiology of infertility in the North East of Scotland. Human reproduction 2009;24: 3096-3107.

28. Bitler MP, Schmidt L. Utilization of infertility treatments: the effects of insurance mandates. Demography 2012;49: 125-149.

29. Black M, Bhattacharya S. Epidemiology of multiple pregnancy and the effect of assisted conception Seminars in Fetal and Neonatal Medicine. 2010. Elsevier, pp. 306- 312.

234

30. Black N. Patient reported outcome measures could help transform healthcare. BMJ: British Medical Journal 2013;346.

31. Blockeel C, Schutyser V, De Vos A, Verpoest W, De Vos M, Staessen C, Haentjens P, Van der Elst J, Devroey P. Prospectively randomized controlled trial of PGS in IVF/ICSI patients with poor implantation. Reproductive biomedicine online 2008;17: 848-854.

32. Boden J, Boden J. When IVF treatment fails. Human Fertility 2007;10: 93-98.

33. Boivin J, Bunting L, Collins JA, Nygren KG. International estimates of infertility prevalence and treatment-seeking: potential need and demand for infertility medical care. Human reproduction 2007;22: 1506-1512.

34. Boulet SL, Mehta A, Kissin DM, Warner L, Kawwass JF, Jamieson DJ. TRends in use of and reproductive outcomes associated with intracytoplasmic sperm injection. Jama 2015;313: 255-263.

35. Braakhekke M, Kamphuis EI, Mol F, Norman RJ, Bhattacharya S, van der Veen F, Mol BWJ. Effectiveness and safety as outcome measures in reproductive medicine. Human Reproduction 2015;30: 2249-2251.

36. Brandes M, Hamilton C, van der Steen J, De Bruin J, Bots R, Nelen W, Kremer J. Unexplained infertility: overall ongoing pregnancy rate and mode of conception. Human Reproduction 2010;26: 360-368.

37. Brandes M, Hamilton CJCM, van der Steen JOM, de Bruin JP, Bots RSGM, Nelen WLDM, Kremer JAM. Unexplained infertility: overall ongoing pregnancy rate and mode of conception. Human Reproduction 2011;26: 360-368.

38. Brandes M, Van Der Steen J, Bokdam S, Hamilton C, De Bruin J, Nelen W, Kremer J. When and why do subfertile couples discontinue their fertility care? A longitudinal cohort study in a secondary care subfertility population. Human reproduction 2009;24: 3127-3135.

39. Briggs AH, Claxton K, Sculpher MJ. Decision modelling for health economic evaluation, 2006. Handbooks in Health Economic E.

40. Briggs AH, O'Brien BJ. The death of cost‐minimization analysis? Health economics 2001;10: 179-184.

41. Brown J, Farquhar C. Clomiphene and other antioestrogens for ovulation induction in polycystic ovarian syndrome. The Cochrane Library 2016.

42. Brown S. Still in high ESTEEM . ESHRE Focus on Reproduction July 2017. Accessed on March 2018. . 2017.

43. Bruinsma F, Brown S, Venn A. The prevalence of prior infertility in recent mothers. Australian and New Zealand journal of public health 1998;22: 841.

44. Bryant J, Suillivan E, Dean J. Assisted reproductive technology in Australia and New Zealand 2015 (Assisted Reproductive Technology Series No 8) 2004, Australian Institute of Health and Welfare National Perinatal Statistics Unit

45. Buchanan JM. Opportunity cost The World of Economics. 1991. Springer, pp. 520-525. 235

46. Bushnik T, Cook JL, Yuzpe AA, Tough S, Collins J. Estimating the prevalence of infertility in Canada. Human Reproduction 2012;27: 738-746.

47. Buxton MJ, Drummond MF, Van Hout BA, Prince RL, Sheldon TA, Szucs T, Vray M. Modelling in ecomomic evaluation: an unavoidable fact of life. Health economics 1997;6: 217-227.

48. Cabrera-León A, Lopez-Villaverde V, Rueda M, Moya-Garrido M. Calibrated prevalence of infertility in 30-to 49-year-old women according to different approaches: a cross-sectional population-based study. Human Reproduction 2015: dev226.

49. CADTH. Canadian Agency for Drugs and Technologies in Health, Ottawa Health Canada. Accessed on December, 2017. 2006.

50. Calhaz-Jorge C, de Geyter C, Kupka MS, de Mouzon J, Erb K, Mocanu E, Motrenko T, Scaravelli G, Wyns C, Goossens V et al. Assisted reproductive technology in Europe, 2012: results generated from European registers by ESHRE†. Human Reproduction 2016;31: 1638-1652.

51. Cantineau AE, Cohlen BJ. Ovarian stimulation protocols (anti‐oestrogens, gonadotrophins with and without GnRH agonists/antagonists) for intrauterine insemination (IUI) in women with subfertility. The Cochrane Library 2007.

52. Carson III CC. Male factor infertility Problem Based Urology. 2013. Springer, pp. 135- 139.

53. Centre for Health Economic Research and Evaluation. Extended Medicare Safety Net Review Report. Australian Government, Department of Health, 2011. Available at http://www.health.gov.au/emsnreview. Accessed on 9 March 2018. 2011.

54. Centres for Disease Control and Prevention. ART success rates. https://www.cdc.gov/art/artdata/index.html Last accessed Jan 2018. 2015.

55. Chambers GM, Adamson GD, Eijkemans MJ. Acceptable cost for the patient and society. Fertil Steril 2013;100: 319-327.

56. Chambers GM, Lee E, Hansen M, Sullivan EA, Bower C, Chapman M. Hospital costs of multiple-birth and singleton-birth children during the first 5 years of life and the role of assisted reproductive technology. JAMA pediatrics 2014;168: 1045-1053.

57. Chambers GM, Paul RC, Harris K, Fitzgerald O, Boothroyd CV, Rombauts L, Chapman MG, Jorm L. Assisted reproductive technology in Australia and New Zealand: cumulative live birth rates as measures of success. The Medical journal of Australia 2017;207: 114-118.

58. Chambers GM, Sullivan EA, Chapman MG, Ishihara O, Zegers-Hochschild F, Nygren KG, Adamson GD. The impact of consumer affordability on access to assisted reproductive technologies and embryo transfer practices: an international analysis. Fertility and sterility 2014;101: 191-198. e194.

59. Chambers GM, Sullivan EA, Ishihara O, Chapman MG, Adamson GD. The economic impact of assisted reproductive technology: a review of selected developed countries. Fertility and sterility 2009;91: 2281-2294.

236

60. Chambers GM, Sullivan EA, Shanahan M, Ho MT, Priester K, Chapman MG. Is in vitro fertilisation more effective than stimulated intrauterine insemination as a first‐line therapy for subfertility? A cohort analysis. Australian and New Zealand Journal of Obstetrics and Gynaecology 2010;50: 280-288.

61. Chambers GM, Wand H, Macaldowie A, Chapman MG, Farquhar CM, Bowman M, Molloy D, Ledger W. Population trends and live birth rates associated with common ART treatment strategies. Human Reproduction 2016;31: 2632-2641.

62. Chambers GM, Zhu R, Illingworth PJ. A reduction in public funding for fertility treatment-an econometric analysis of access to treatment and savings to government. BMC health services research 2012;12: 142.

63. Chambers MG, Paul RC, Harris K, Fitzgerald O, Boothroyd CV, Rombauts L, Chapman MG, Jorm L. Assisted reproductive technology in Australia and New Zealand: cumulative live birth rates as measures of success. The Medical journal of Australia 2017;207: 114-118.

64. Chandra A, Copen CE, Stephen EH. Infertility and impaired fecundity in the United States, 1982-2010: data from the National Survey of Family Growth, 2013. US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics.

65. Chang J, Boulet SL, Jeng G, Flowers L, Kissin DM. Outcomes of in vitro fertilization with preimplantation genetic diagnosis: an analysis of the United States Assisted Reproductive Technology Surveillance Data, 2011–2012. Fertility and sterility 2016;105: 394-400.

66. Chapko MK, Liu CF, Perkins M, Li YF, Fortney JC, Maciejewski ML. Equivalence of two healthcare costing methods: bottom‐up and top‐down. Health economics 2009;18: 1188-1201.

67. Chen M, Wei S, Hu J, Quan S. Can comprehensive chromosome screening technology improve IVF/ICSI outcomes? A meta-analysis. PloS one 2015;10: e0140779.

68. Chua SJ, Akande VA, Mol BWJ. Surgery for tubal infertility. The Cochrane Library 2017.

69. Cissen M, Bensdorp A, Cohlen BJ, Repping S, de Bruin JP, van Wely M. Assisted reproductive technologies for male subfertility. The Cochrane Library 2016.

70. Claxton K, Paulden M, Gravelle H, Brouwer W, Culyer AJ. Discounting and decision making in the economic evaluation of health‐care technologies. Health economics 2011;20: 2-15.

71. Claxton K, Sculpher M, Drummond M. A rational framework for decision making by the National Institute for Clinical Excellence (NICE). The Lancet 2002;360: 711-715.

72. Clement FM, Harris A, Li JJ, Yong K, Lee KM, Manns BJ. Using effectiveness and cost-effectiveness to make drug coverage decisions: a comparison of Britain, Australia, and Canada. Jama 2009;302: 1437-1443.

73. Coates A, Kung A, Mounts E, Hesla J, Bankowski B, Barbieri E, Ata B, Cohen J, Munné S. Optimal euploid embryo transfer strategy, fresh versus frozen, after

237

preimplantation genetic screening with next generation sequencing: a randomized controlled trial. Fertility and sterility 2017;107: 723-730. e723.

74. Cobo A, de los Santos MJ, Castellò D, Gámiz P, Campos P, Remohí J. Outcomes of vitrified early cleavage-stage and blastocyst-stage embryos in a cryopreservation program: evaluation of 3,150 warming cycles. Fertility and sterility 2012;98: 1138- 1146. e1131.

75. Cobo A, García-Velasco JA, Coello A, Domingo J, Pellicer A, Remohí J. Oocytes vitrification as an efficient option for elective fertility preservation. Fertility and sterility 2015: 755-764.

76. Coccia ME, Rizzello F, Cammilli F, Bracco GL, Scarselli G. Endometriosis and infertility: surgery and ART: an integrated approach for successful management. European Journal of Obstetrics & Gynecology and Reproductive Biology 2008;138: 54- 59.

77. Cohen DJ, Reynolds MR. Interpreting the results of cost-effectiveness studies. Journal of the American College of Cardiology 2008;52: 2119-2126.

78. Cohen M. Intrauterine insemination. Int J Fertil 1962;7: 174.

79. Collins JA, Burrows EA, Willan AR. The prognosis for live birth among untreated infertile couples. Fertility and sterility 1995;64: 22-28.

80. Collins SC, Xu X, Mak W. Cost-effectiveness of preimplantation genetic screening for women older than 37 undergoing in vitro fertilization. Journal of assisted reproduction and genetics 2017: 1-8.

81. Collins SC, Xu X, Mak W. Cost-effectiveness of Preimplantation Genetic Screening for Women Older than 37 Years Undergoing IVF [19P]. Obstetrics & Gynecology 2017;129: S168-S169.

82. Connolly MP, Hoorens S, Chambers GM. The costs and consequences of assisted reproductive technology: an economic perspective. Human reproduction update 2010;16: 603-613.

83. Cooper N, Sutton A, Ades A, Paisley S, Jones D. Use of evidence in economic decision models: practical issues and methodological challenges. Health economics 2007;16: 1277-1286.

84. Cramer DW, Walker AM, Schiff I. Statistical methods in evaluating the outcome of infertility therapy. Fertility and sterility 1979;32: 80-86.

85. Culyer A, McCabe C, Briggs A, Claxton K, Buxton M, Akehurst R, Sculpher M, Brazier J. Searching for a threshold, not setting one: the role of the National Institute for Health and Clinical Excellence. Journal of health services research & policy 2007;12: 56-58.

86. Culyer AJ. Four Issues in Cost-Effectiveness Analysis and Health Technology Assessment: a View from the Touch-line Health Technology Assessment and Health Policy Today: A Multifaceted View of their Unstable Crossroads. 2015. Springer, pp. 77-94.

238

87. Custers IM, van Dessel THJHM, Flierman PA, Steures P, van Wely M, van der Veen F, Mol BWJ. Couples dropping out of a reimbursed intrauterine insemination program: what is their prognostic profile and why do they drop out? Fertility and sterility 2013;99: 1294-1298.

88. Custers IM, van Rumste MM, van der Steeg JW, van Wely M, Hompes PG, Bossuyt P, Broekmans FJ, Renckens CN, Eijkemans MJ, van Dessel TJ. Long-term outcome in couples with unexplained subfertility and an intermediate prognosis initially randomized between expectant management and immediate treatment. Human reproduction 2011;27: 444-450.

89. Dahdouh EM, Balayla J, García-Velasco JA. Comprehensive chromosome screening improves embryo selection: a meta-analysis. Fertility and sterility 2015;104: 1503- 1512.

90. Dakin H, Wordsworth S. Cost‐minimisation analysis versus cost‐effectiveness analysis, revisited. Health economics 2013;22: 22-34.

91. Dapuzzo L, Seitz FE, Dodson WC, Stetter C, Kunselman AR, Legro RS. Incomplete and inconsistent reporting of maternal and fetal outcomes in infertility treatment trials. Fertility and sterility 2011;95: 2527-2530.

92. Datta J, Palmer MJ, Tanton C, Gibson LJ, Jones KG, Macdowall W, Glasier A, Sonnenberg P, Field N, Mercer CH et al. Prevalence of infertility and help seeking among 15 000 women and men. Human Reproduction 2016.

93. Davies M, Rumbold A, Marino J, Willson K, Giles L, Whitrow M, Scheil W, Moran L, Thompson J, Lane M. Maternal factors and the risk of birth defects after IVF and ICSI: a whole of population cohort study. BJOG: An International Journal of Obstetrics & Gynaecology 2017;124: 1537-1544.

94. Davies MJ, Moore VM, Willson KJ, Van Essen P, Priest K, Scott H, Haan EA, Chan A. Reproductive technologies and the risk of birth defects. New England Journal of Medicine 2012;366: 1803-1813.

95. Daya S. Life table (survival) analysis to generate cumulative pregnancy rates in assisted reproduction: are we overestimating our success rates? Human Reproduction 2005;20: 1135-1143.

96. De Neubourg D, Bogaerts K, Blockeel C, Coetsier T, Delvigne A, Devreker F, Dubois M, Gillain N, Gordts S, Wyns C. How do cumulative live birth rates and cumulative multiple live birth rates over complete courses of assisted reproductive technology treatment per woman compare among registries? Human Reproduction 2015;31: 93-99.

97. De Neubourg D, Daels C, Elseviers M, Mangelschots K, Vercruyssen M, Van Royen E. Cumulative live-birth delivery after IVF/ICSI since the progressive introduction of single-embryo transfer. Reproductive biomedicine online 2010;20: 836-842.

98. De Rycke M, Goossens V, Kokkali G, Meijer-Hoogeveen M, Coonen E, Moutou C. ESHRE PGD Consortium data collection XIV–XV: cycles from January 2011 to December 2012 with pregnancy follow-up to October 2013. 2017. Oxford University Press.

99. Debrock S, Melotte C, Spiessens C, Peeraer K, Vanneste E, Meeuwis L, Meuleman C, Frijns J-P, Vermeesch JR, D'Hooghe TM. Preimplantation genetic screening for 239

aneuploidy of embryos after in vitro fertilization in women aged at least 35 years: a prospective randomized trial. Fertility and sterility 2010;93: 364-373.

100. Delobel-Ayoub M, Arnaud C, White-Koning M, Casper C, Pierrat V, Garel M, Burguet A, Roze J-C, Matis J, Picaud J-C. Behavioral problems and cognitive performance at 5 years of age after very preterm birth: the EPIPAGE Study. Pediatrics 2009;123: 1485- 1492.

101. Devine K, Mumford SL, Goldman KN, Hodes-Wertz B, Druckenmiller S, Propst AM, Noyes N. Baby budgeting: oocyte cryopreservation in women delaying reproduction can reduce cost per live birth. Fertil Steril 2015;103: 1446-1453. e1442.

102. Devlin N, Parkin D. Funding fertility: issues in the allocation and distribution of resources to assisted reproduction technologies. Human Fertility 2003;6: S2-S6.

103. Dickey RP. The relative contribution of assisted reproductive technologies and ovulation induction to multiple births in the United States 5 years after the Society for Assisted Reproductive Technology/American Society for Reproductive Medicine recommendation to limit the number of embryos transferred. Fertility and sterility 2007;88: 1554-1561.

104. Dickey RP. Strategies to reduce multiple pregnancies due to ovulation stimulation. Fertility and sterility 2009;91: 1-17.

105. Dickey RP. Clomiphene citrate for women with unexplained infertility Unexplained Infertility. 2015. Springer, pp. 261-271.

106. Dickey RP, Holtkamp DE. Development, pharmacology and clinical experience with clomiphene citrate. Human reproduction update 1996;2: 483-506.

107. Dickey RP, Taylor SN, Curole DN, Rye PH, Lu PY, Pyrzak R. Relationship of clomiphene dose and patient weight to successful treatment. Human reproduction (Oxford, England) 1997;12: 449-453.

108. Dodson WC, Whitesides DB, Hughes CL, Easley H, Haney A. Superovulation with intrauterine insemination in the treatment of infertility: a possible alternative to gamete intrafallopian transfer and in vitro fertilization. Fertility and sterility 1987;48: 441-445.

109. Doody KJ. Cryopreservation and delayed embryo transfer–assisted reproductive technology registry and reporting implications. Fertility and sterility 2014;102: 27-31.

110. Doody MC. Drop-out behaviour and fertility table analysis of pregnancy rates. Human Reproduction 1993;8: 886-889.

111. Downs SH, Black N. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. Journal of Epidemiology & Community Health 1998;52: 377-384.

112. Doyle JO, Richter KS, Lim J, Stillman RJ, Graham JR, Tucker MJ. Successful elective and medically indicated oocyte vitrification and warming for autologous in vitro fertilization, with predicted birth probabilities for fertility preservation according to number of cryopreserved oocytes and age at retrieval. Fertility and sterility 2016;105: 459-466. e452.

240

113. Drummond MF, MJ S, GW T, BJ OB, GL S. Methods for the Economic Evaluation of Health Care Programmes. 3rd Edition. 2005. Great Britain, Oxford University Press.

114. Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the economic evaluation of health care programmes, 2015. Oxford university press.

115. Duckitt K, Harrington D. Risk factors for pre-eclampsia at antenatal booking: systematic review of controlled studies. BMJ (Clinical research ed) 2005;330: 565.

116. Duffy J, Arambage K, Correa F, Olive D, Farquhar C, Garry R, Barlow DH, Jacobson TZ. Laparoscopic surgery for endometriosis. Cochrane Database Syst Rev 2014;4.

117. Dumoulin JC, Land JA, Van Montfoort AP, Nelissen EC, Coonen E, Derhaag JG, Schreurs IL, Dunselman GA, Kester AD, Geraedts JP. Effect of in vitro culture of human embryos on birthweight of newborns. Human reproduction 2010;25: 605-612.

118. Dunson DB, Baird DD, Colombo B. Increased infertility with age in men and women. Obstetrics & Gynecology 2004;103: 51-56.

119. Dyer S, Chambers G, De Mouzon J, Nygren K, Zegers-Hochschild F, Mansour R, Ishihara O, Banker M, Adamson G. International Committee for Monitoring Assisted Reproductive Technologies world report: assisted reproductive technology 2008, 2009 and 2010. Human Reproduction 2016;31: 1588-1609.

120. Ebomoyi E, Adetoro O. Socio-biological factors influencing infertility in a rural Nigerian community. International Journal of Gynecology & Obstetrics 1990;33: 41- 47.

121. Edgar DH, Karani J, Gook DA. Increasing dehydration of human cleavage-stage embryos prior to slow cooling significantly increases cryosurvival. Reproductive biomedicine online 2009;19: 521-525.

122. Edlin R, McCabe C, Hulme C, Hall P, Wright J. Economic Evaluation, Cost Effectiveness Analysis and Health Care Resource Allocation Cost Effectiveness Modelling for Health Technology Assessment. 2015. Springer, pp. 1-13.

123. Eggert J, Theobald H, Engfeldt P. Effects of alcohol consumption on female fertility during an 18-year period. Fertility and sterility 2004;81: 379-383.

124. ESHRE Capri Workshop Group. Fertility and ageing. Human reproduction update 2005;11: 261-276.

125. ESHRE Capri Workshop Group. Intrauterine insemination. 2009. ESHRE.

126. ESHRE Capri Workshop Group. Birth defects and congenital health risks in children conceived through assisted reproduction technology (ART): a meeting report. 2014. Springer.

127. Ethics Committee of the American Society for Reproductive Medicine. Oocyte or embryo donation to women of advanced age: a committee opinion. Fertility and sterility 2013;100: 337-340.

128. European IVF-monitoring Consortium, Reproduction ESoH, Embryology, Calhaz-Jorge C, De Geyter C, Kupka M, de Mouzon J, Erb K, Mocanu E, Motrenko T et al. Assisted

241

reproductive technology in Europe, 2013: results generated from European registers by ESHRE. 2017. Oxford University Press.

129. Farquhar C, Brown J, Marjoribanks J. Laparoscopic drilling by diathermy or laser for ovulation induction in anovulatory polycystic ovary syndrome. The Cochrane Library 2012.

130. Fauser B, Devroey P, Diedrich K, Balaban B, Bonduelle M, Delemarre-Van De Waal H, Estella C, Ezcurra D, Geraedts J, Howles C. Health outcomes of children born after IVF/ICSI: a review of current expert opinion and literature. Reproductive biomedicine online 2014;28: 162-182.

131. Fauser BC, Devroey P, Macklon NS. Multiple birth resulting from ovarian stimulation for subfertility treatment. The Lancet 2005;365: 1807-1816.

132. Federowicz MH, Grossman MN, Hayes BJ, Riggs J. A tutorial on activity-based costing of electronic health records. Quality Management in Healthcare 2010;19: 86-89.

133. Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost‐ effectiveness acceptability curves. Health economics 2001;10: 779-787.

134. Fiorentino F, Bono S, Biricik A, Nuccitelli A, Cotroneo E, Cottone G, Kokocinski F, Michel C-E, Minasi MG, Greco E. Application of next-generation sequencing technology for comprehensive aneuploidy screening of blastocysts in clinical preimplantation genetic screening cycles. Human Reproduction 2014;29: 2802-2813.

135. Fishel S, Craig A, Lynch C, Dowell K, Ndukwe G, Jenner L, Cater E, Brown A, Gordon A, Thornton S. Assessment of 19,803 paired chromosomes and clinical outcome from first 150 cycles using array CGH of the first polar body for embryo selection and transfer. J Fertiliz In Vitro 2011;1: 1-8.

136. Fitzgerald O, Harris K, Paul RC, GM C. Assisted reproductive technology in Australia and New Zealand 2015. 2017. UNSW National Perinatal Epidemiology and Statistics Unit, the University of New South Wales, Sydney.

137. Forman EJ, Tao X, Ferry KM, Taylor D, Treff NR, Scott RT. Single embryo transfer with comprehensive chromosome screening results in improved ongoing pregnancy rates and decreased miscarriage rates. Human Reproduction 2012;27: 1217-1222.

138. Fragouli E, Alfarawati S, Daphnis DD, Goodall NN, Mania A, Griffiths T, Gordon A, Wells D. Cytogenetic analysis of human blastocysts with the use of FISH, CGH and aCGH: scientific data and technical evaluation. Human reproduction (Oxford, England) 2011;26: 480-490.

139. Fragouli E, Alfarawati S, Goodall N-n, Sánchez-García JF, Colls P. The cytogenetics of polar bodies: insights into female meiosis and the diagnosis of aneuploidy. Molecular Human Reproduction 2011;17: 286-295.

140. Fragouli E, Wells D. Aneuploidy screening for embryo selection Seminars in reproductive medicine. 2012, pp. 289-301.

141. Franasiak JM, Forman EJ, Hong KH, Werner MD, Upham KM, Treff NR, Scott RT. The nature of aneuploidy with increasing age of the female partner: a review of 15,169 consecutive trophectoderm biopsies evaluated with comprehensive chromosomal screening. Fertility and sterility 2014;101: 656-663. e651. 242

142. Francavilla, Sciarretta F, Sorgentone S, Necozione S, Santucci R, Barbonetti A, Francavilla S. Intrauterine insemination with or without mild ovarian stimulation in couples with male subfertility due to oligo/astheno-and/or teratozoospermia or antisperm antibodies: a prospective cross-over trial. Fertility and sterility 2009;92: 1009-1011.

143. Franks S, Adams J, Mason H, Polson D. Ovulatory disorders in women with polycystic ovary syndrome. Clinics in obstetrics and gynaecology 1985;12: 605-632.

144. Fujiwara D, Campbell R. Valuation techniques for social cost-benefit analysis: stated preference, revealed preference and subjective well-being approaches: a discussion of the current issues, 2011. HM Treasury.

145. Gad M, Badroui M. M053 Evidence-based Therapy for Infertility associated with early stage Endometriosis International Journal of Gynecology & Obstetrics 2012;119: S548.

146. Gameiro S, Boivin J, Peronace L, Verhaak CM. Why do patients discontinue fertility treatment? A systematic review of reasons and predictors of discontinuation in fertility treatment. Human reproduction update 2012;18: 652-669.

147. Garcia-Velasco JA, Domingo J, Cobo A, Martínez M, Carmona L, Pellicer A. Five years' experience using oocyte vitrification to preserve fertility for medical and nonmedical indications. Fertility and sterility 2013;99: 1994-1999.

148. Garcia-Velasco JA, Fauser B. Preimplantation genetic screening-what a wonderful world it would be! Reproductive biomedicine online 2016;32: 337.

149. Garcia-Velasco JA, Fauser BC. Preimplantation genetic screening–what a wonderful world it would be! Reproductive biomedicine online 2016;32: 337-338.

150. Garry R. The effectiveness of laparoscopic excision of endometriosis. Current Opinion in Obstetrics and Gynecology 2004;16: 299-303.

151. Gebhart MB, Hines RS, Penman A, Holland AC. How do patient perceived determinants influence the decision-making process to accept or decline preimplantation genetic screening? Fertil Steril 2016;105: 188-193.

152. Gianaroli L, Racowsky C, Geraedts J, Cedars M, Makrigiannakis A, Lobo R. Best practices of ASRM and ESHRE: a journey through reproductive medicine. Human Reproduction 2012;27: 3365-3379.

153. Gjönnaess H. Polycystic ovarian syndrome treated by ovarian electrocautery through the laparoscope. Fertility and sterility 1984;41: 20-25.

154. Gleicher N. The irrational attraction of elective single-embryo transfer (eSET). Human Reproduction 2012;28: 294-297.

155. Gleicher N, Barad D. Twin pregnancy, contrary to consensus, is a desirable outcome in infertility. Fertility and sterility 2009;91: 2426-2431.

156. Gleicher N, Weghofer A, Barad D. Preimplantation genetic screening:“established” and ready for prime time? Fertility and sterility 2008;89: 780-788.

243

157. Gnoth C, Godehardt D, Godehardt E, Frank‐Herrmann P, Freundl G. Time to pregnancy: results of the German prospective study and impact on the management of infertility. Human Reproduction 2003;18: 1959-1966.

158. Gnoth C, Godehardt E, Frank-Herrmann P, Friol K, Tigges J, Freundl G. Definition and prevalence of subfertility and infertility. Human reproduction 2005;20: 1144-1147.

159. Goldman MB, Thornton KL, Ryley D, Alper MM, Fung JL, Hornstein MD, Reindollar RH. A randomized clinical trial to determine optimal infertility treatment in older couples: the Forty and Over Treatment Trial (FORT-T). Fertility and sterility 2014;101: 1574-1581. e1572.

160. Gormack AA, Peek JC, Derraik JG, Gluckman PD, Young NL, Cutfield WS. Many women undergoing fertility treatment make poor lifestyle choices that may affect treatment outcome. Human reproduction 2015;30: 1617-1624.

161. Goverde AJ, McDonnell J, Vermeiden JP, Schats R, Rutten FF, Schoemaker J. Intrauterine insemination or in-vitro fertilisation in idiopathic subfertility and male subfertility: a randomised trial and cost-effectiveness analysis. The lancet 2000;355: 13- 18.

162. Greco E, Bono S, Ruberti A, Lobascio AM, Greco P, Biricik A, Spizzichino L, Greco A, Tesarik J, Minasi MG. Comparative genomic hybridization selection of blastocysts for repeated implantation failure treatment: a pilot study. BioMed research international 2014;2014.

163. Greco E, Minasi MG, Fiorentino F. Healthy babies after intrauterine transfer of mosaic aneuploid blastocysts. New England Journal of Medicine 2015;373: 2089-2090.

164. Greenblatt RB. Chemical induction of ovulation. Fertility and sterility 1961;12: 402- 404.

165. Gremeau A-S, Andreadis N, Fatum M, Craig J, Turner K, Mcveigh E, Child T. In vitro maturation or in vitro fertilization for women with polycystic ovaries? A case–control study of 194 treatment cycles. Fertility and sterility 2012;98: 355-360.

166. Griesinger G. Beware of the ‘implantation rate’! Why the outcome parameter ‘implantation rate’ should be abandoned from infertility research. Human Reproduction 2016;31: 249-251.

167. Griffiths A, Dyer SM, Lord SJ, Pardy C, Fraser IS, Eckermann S. A cost-effectiveness analysis of in-vitro fertilization by maternal age and number of treatment attempts. Human Reproduction 2010;25: 924-931.

168. Grimes DA, Schulz KF. Bias and causal associations in observational research. The Lancet 2002;359: 248-252.

169. Gurunath S, Pandian Z, Anderson RA, Bhattacharya S. Defining infertility—a systematic review of prevalence studies. Human reproduction update 2011;17: 575- 588.

170. Guttmacher AF. Factors affecting normal expectancy of conception. Journal of the American Medical Association 1956;161: 855-860.

244

171. Guzick DS. Ovulation induction management of PCOS. Clinical obstetrics and gynecology 2007;50: 255-267.

172. Haas J, Casper RF. In vitro fertilization treatments with the use of clomiphene citrate or letrozole. Fertility and sterility 2017;108: 568-571.

173. Habbema JDF, Collins J, Leridon H, Evers JLH, Lunenfeld B, teVelde ER. Towards less confusing terminology in reproductive medicine: a proposal. Human Reproduction 2004;19: 1497-1501.

174. Haebe J, Martin J, Tekepety F, Tummon I, Shepherd K. Success of intrauterine insemination in women aged 40–42 years. Fertility and sterility 2002;78: 29-33.

175. Halliday JL, Ukoumunne OC, Baker HG, Breheny S, Jaques AM, Garrett C, Healy D, Amor D. Increased risk of blastogenesis birth defects, arising in the first 4 weeks of pregnancy, after assisted reproductive technologies. Human reproduction 2009;25: 59- 65.

176. Hamilton BH, McManus B. The effects of insurance mandates on choices and outcomes in infertility treatment markets. Health Econ 2012;21: 994-1016.

177. Hammarberg K, Kirkman M, Pritchard N, Hickey M, Peate M, McBain J, Agresta F, Bayly C, Fisher J. Reproductive experiences of women who cryopreserved oocytes for non-medical reasons. Human Reproduction 2017.

178. Handyside AH, Kontogianni EH, Hardy K, Winston R. Pregnancies from biopsied human preimplantation embryos sexed by Y-specific DNA amplification. Nature 1990;344: 768-770.

179. Hansen M, Kurinczuk JJ, Milne E, de Klerk N, Bower C. Assisted reproductive technology and birth defects: a systematic review and meta-analysis. Human reproduction update 2013;19: 330-353.

180. Harper J, Jackson E, Sermon K, Aitken RJ, Harbottle S, Mocanu E, Hardarson T, Mathur R, Viville S, Vail A et al. Adjuncts in the IVF laboratory: where is the evidence for ‘add-on’ interventions? Human Reproduction 2017;32: 485-491.

181. Harris AH, Hill SR, Chin G, Li JJ, Walkom E. The role of value for money in public insurance coverage decisions for drugs in Australia: a retrospective analysis 1994-2004. Medical Decision Making 2008;28: 713-722.

182. Harris K, Fitzgerald O, Paul R, Macaldowie A, Lee E, Chambers GM, . Assisted reproductive technology in Australia and New Zealand 2014. 2016. AIHW.

183. Harris K, Fitzgerald O, Repon C Paul, A M, Lee Evelyn, Chambers GM. Assisted reproductive technology in Australia and New Zealand 2014, 2016. AIHW.

184. Harton GL, Munné S, Surrey M, Grifo J, Kaplan B, McCulloh DH, Griffin DK, Wells D, Group PP. Diminished effect of maternal age on implantation after preimplantation genetic diagnosis with array comparative genomic hybridization. Fertility and sterility 2013;100: 1695-1703.

185. Hassold T, Abruzzo M, Adkins K, Griffin D, Merrill M, Millie E, Saker D, Shen J, Zaragoza M. Human aneuploidy: incidence, origin, and etiology. Environmental and molecular mutagenesis 1996;28: 167-175. 245

186. Hassold T, Hall H, Hunt P. The origin of human aneuploidy: where we have been, where we are going. Human molecular genetics 2007;16: R203-R208.

187. Hassold T, Hunt P. To err (meiotically) is human: the genesis of human aneuploidy. Nature Reviews Genetics 2001;2: 280-291.

188. Hasson J, Tulandi T, Son W-Y, Badeghiesh A, Tannus S, Takefman J, Shavit T. Reproductive outcomes of familial oocyte donation. Fertility and sterility 2016;106: 1696-1702.

189. Heijnen EMEW, Macklon NS, Fauser BCJM. What is the most relevant standard of success in assisted reproduction?The next step to improving outcomes of IVF: consider the whole treatment. Human Reproduction 2004;19: 1936-1938.

190. Helmerhorst FM, van Vliet HA, Gornas T, Finken MJ, Grimes DA. Intrauterine insemination versus timed intercourse for cervical hostility in subfertile couples. Obstetrical & gynecological survey 2006;61: 402-414.

191. Henry DA, Hill SR, Harris A. Drug prices and value for money: the Australian Pharmaceutical Benefits Scheme. Jama 2005;294: 2630-2632.

192. Herbert DL, Lucke JC, Dobson AJ. Infertility, medical advice and treatment with fertility hormones and/or in vitro fertilisation: a population perspective from the Australian Longitudinal Study on Women's Health. Australian and New Zealand journal of public health 2009;33: 358-364.

193. Hershkop E, Segal L, Fainaru O, Kol S. ‘Model’versus ‘everyday’patients: can randomized controlled trial data really be applied to the clinic? Reproductive biomedicine online 2017;34: 274-279.

194. Hirshfeld-Cytron J, Grobman WA, Milad MP. Fertility preservation for social indications: a cost-based decision analysis. Fertil Steril 2012;97: 665-670.

195. Hodes-Wertz B, Druckenmiller S, Smith M, Noyes N. What do reproductive-age women who undergo oocyte cryopreservation think about the process as a means to preserve fertility? Fertility and sterility 2013;100: 1343-1349. e1342.

196. Hodes-Wertz B, Grifo J, Ghadir S, Kaplan B, Laskin CA, Glassner M, Munné S. Idiopathic recurrent miscarriage is caused mostly by aneuploid embryos. Fertility and sterility 2012;98: 675-680.

197. Hollier LM, Leveno KJ, Kelly MA, MCIntire DD, Cunningham FG. Maternal age and malformations in singleton births. Obstetrics & Gynecology 2000;96: 701-706.

198. Holzer H, Casper R, Tulandi T. A new era in ovulation induction. Fertility and sterility 2006;85: 277-284.

199. Homburg R. Clomiphene citrate—end of an era? a mini-review. Human Reproduction 2005;20: 2043-2051.

200. Homburg R, Hendriks M-L, König TE, Anderson RA, Balen AH, Brincat M, Child T, Davies M, D'Hooghe T, Martinez A. Clomifene citrate or low-dose FSH for the first- line treatment of infertile women with anovulation associated with polycystic ovary syndrome: a prospective randomized multinational study. Human reproduction 2012;27: 468-473. 246

201. Hughes E, Beecroft M, Wilkie V, Burville L, Claman P, Tummon I, Greenblatt E, Fluker M, Thorpe K. A multicentre randomized controlled trial of expectant management versus IVF in women with Fallopian tube patency. Human Reproduction 2004;19: 1105-1109.

202. Hughes E, Brown J, Collins JJ, Vanderkerchove P. Clomiphene citrate for unexplained subfertility in women. The Cochrane Library 2010.

203. Hughes EG. The effectiveness of ovulation induction and intrauterine insemination in the treatment of persistent infertility: a meta-analysis. Human reproduction (Oxford, England) 1997;12: 1865-1872.

204. Hughes EG. Singleton birth at term: an old alarm or a new debate?†. Human Reproduction 2015;30: 2254-2256.

205. Hull M, Glazener C, Kelly N, Conway D, Foster P, Hinton R, Coulson C, Lambert P, Watt E, Desai K. Population study of causes, treatment, and outcome of infertility. Br Med J (Clin Res Ed) 1985;291: 1693-1697.

206. Hull MG, Fleming CF. Tubal surgery versus assisted reproduction: assessing their role in infertility therapy. Current Opinion in Obstetrics and Gynecology 1995;7: 160-167.

207. Human Fertilisation Embryology Authority. Multiple births after IVF-HFEA. 2009.

208. Hunault C, Habbema J, Eijkemans M, Collins J, Evers J, Te Velde E. Two new prediction rules for spontaneous pregnancy leading to live birth among subfertile couples, based on the synthesis of three previous models. Human Reproduction 2004;19: 2019-2026.

209. Hunter T, Hart R. Endoscopic surgery for female infertility: a review of current management. Australian and New Zealand Journal of Obstetrics and Gynaecology 2009;49: 588-593.

210. Ioannidis JP, Haidich A-B, Lau J. Any casualties in the clash of randomised and observational evidence?: No—recent comparisons have studied selected questions, but we do need more data. BMJ: British Medical Journal 2001;322: 879-880.

211. Jacobs JC, Barnett PG. Emergent challenges in determining costs for economic evaluations. PharmacoEconomics 2017;35: 129-139.

212. Jain T, Harlow BL, Hornstein MD. Insurance coverage and outcomes of in vitro fertilization. New England Journal of Medicine 2002;347: 661-666.

213. Jensen TK, Sobotka T, Hansen MA, Pedersen AT, Lutz W, Skakkebæk NE. Declining trends in conception rates in recent birth cohorts of native Danish women: a possible role of deteriorating male reproductive health. International journal of andrology 2008;31: 81-92.

214. Jepsen P, Johnsen S, Gillman M, Sørensen H. Interpretation of observational studies. Heart 2004;90: 956-960.

215. Jeve Y, Potdar N, Opoku A, Khare M. Donor oocyte conception and pregnancy complications: a systematic review and meta‐analysis. BJOG: An International Journal of Obstetrics & Gynaecology 2016;123: 1471-1480.

247

216. Jo C. Cost-of-illness studies: concepts, scopes, and methods. Clinical and molecular hepatology 2014;20: 327.

217. Joffe M. Time trends in biological fertility in Britain. The Lancet 2000;355: 1961-1965.

218. Joffe M, Key J, Best N, Keiding N, Scheike T, Jensen TK. Studying time to pregnancy by use of a retrospective design. American journal of epidemiology 2005;162: 115-124.

219. Johnson J-A, Tough S. Delayed Child-Bearing. Journal of Obstetrics and Gynaecology Canada 2012;34: 80-93.

220. Johnson NP, Hummelshoj L, Adamson GD, Keckstein J, Taylor HS, Abrao MS, Bush D, Kiesel L, Tamimi R, Sharpe-Timms KL. World Endometriosis Society consensus on the classification of endometriosis. Human reproduction 2017;32: 315-324.

221. Johnson NP, Hummelshoj L, Consortium WESM, Abrao M, Adamson G, Allaire C, Amelung V, Andersson E, Becker C, Birna Árdal K. Consensus on current management of endometriosis. Human Reproduction 2013;28: 1552-1568.

222. Joseph KS, Allen AC, Dodds L, Turner LA, Scott H, Liston R. The perinatal effects of delayed childbearing. Obstetrics & Gynecology 2005;105: 1410-1418.

223. Juhl M, Olsen J, Nybo Andersen AM, Grønbæk M. Intake of wine, beer and spirits and waiting time to pregnancy. Human Reproduction 2003;18: 1967-1971.

224. Kafy S, Tulandi T. New advances in ovulation induction. Current Opinion in Obstetrics and Gynecology 2007;19: 248-252.

225. Kawwass JF, Monsour M, Crawford S, Kissin DM, Session DR, Kulkarni AD, Jamieson DJ. Trends and outcomes for donor oocyte cycles in the United States, 2000- 2010. Jama 2013;310: 2426-2434.

226. Keiding N, HØJBJERG HANSEN OK, Sørensen DN, Slama R. The current duration approach to estimating time to pregnancy. Scandinavian Journal of Statistics 2012;39: 185-204.

227. Keiding N, Kvist K, Hartvig H, Tvede M, Juul S. Estimating time to pregnancy from current durations in a cross‐sectional sample. Biostatistics 2002;3: 565-578.

228. Keltz MD, Vega M, Sirota I, Lederman M, Moshier EL, Gonzales E, Stein D. Preimplantation Genetic Screening (PGS) with Comparative Genomic Hybridization (CGH) following day 3 single cell blastomere biopsy markedly improves IVF outcomes while lowering multiple pregnancies and miscarriages. Journal of assisted reproduction and genetics 2013;30: 1333-1339.

229. Kennedy S, Bergqvist A, Chapron C, D’hooghe T, Dunselman G, Greb R, Hummelshoj L, Prentice A, Saridogan E. ESHRE guideline for the diagnosis and treatment of endometriosis. Human reproduction 2005;20: 2698-2704.

230. Kissin DM, Kulkarni AD, Mneimneh A, Warner L, Boulet SL, Crawford S, Jamieson DJ. Embryo transfer practices and multiple births resulting from assisted reproductive technology: an opportunity for prevention. Fertility and sterility 2015;103: 954-961.

231. Kistner R. Induction of ovulation with clomiphene citrate (Clomid). South African Medical Journal 1967;5: 25-35. 248

232. Klipstein S, Regan M, Ryley DA, Goldman MB, Alper MM, Reindollar RH. One last chance for pregnancy: a review of 2,705 in vitro fertilization cycles initiated in women age 40 years and above. Fertility and sterility 2005;84: 435-445.

233. Klitzman R. How much is a child worth? Providers’ and patients’ views and responses concerning ethical and policy challenges in paying for ART. PLoS One 2017;12: e0171939.

234. Koch J, Rowan K, Rombauts L, Yazdani A, Chapman M, Johnson N. Endometriosis and infertility–a consensus statement from ACCEPT (Australasian CREI Consensus Expert Panel on Trial evidence). Australian and New Zealand Journal of Obstetrics and Gynaecology 2012;52: 513-522.

235. Korkmaz C, Tekin YB, Sakinci M, Ercan CM. Effects of maternal ageing on ICSI outcomes and embryo development in relation to oocytes morphological characteristics of birefringent structures. Zygote 2015;23: 550-555.

236. Kousta E, White D, Franks S. Modern use of clomiphene citrate in induction of ovulation. Human reproduction update 1997;3: 359-365.

237. Kovalevsky G, Carney SM, Morrison LS, Boylan CF, Neithardt AB, Feinberg RF. Should embryos developing to blastocysts on day 7 be cryopreserved and transferred: an analysis of pregnancy and implantation rates. Fertility and sterility 2013;100: 1008- 1012.

238. Kulak D, Jindal SK, Oh C, Morelli SS, Kratka S, McGovern PG. Reporting in vitro fertilization cycles to the Society for Assisted Reproductive Technology database: where have all the cycles gone? Fertility and sterility 2016;105: 927-931. e923.

239. Kulkarni AD, Adashi EY, Jamieson DJ, Crawford SB, Sunderam S, Kissin DM. Affordability of Fertility Treatments and Multiple Births in the United States. Paediatric and Perinatal Epidemiology 2017;31: 438-448.

240. Kulkarni AD, Jamieson DJ, Jones Jr HW, Kissin DM, Gallo MF, Macaluso M, Adashi EY. Fertility treatments and multiple births in the United States. New England Journal of Medicine 2013;369: 2218-2225.

241. Kunz R, Oxman AD. The unpredictability paradox: review of empirical comparisons of randomised and non-randomised clinical trials. BMJ (Clinical research ed) 1998;317: 1185-1190.

242. Kushnir VA, Barad DH, Albertini DF, Darmon SK, Gleicher N. Outcomes of fresh and cryopreserved oocyte donation. Jama 2015;314: 623-624.

243. Kushnir VA, Barad DH, Albertini DF, Darmon SK, Gleicher N. Effect of embryo banking on US National assisted reproductive technology live birth rates. PloS one 2016;11: e0154620.

244. Kushnir VA, Darmon SK, Albertini DF, Barad DH, Gleicher N. Effectiveness of in vitro fertilization with preimplantation genetic screening: a reanalysis of United States assisted reproductive technology data 2011–2012. Fertility and sterility 2016;106: 75- 79.

249

245. Kushnir VA, Darmon SK, Albertini DF, Barad DH, Gleicher N. Effectiveness of in vitro fertilization with preimplantation genetic screening: a reanalysis of United States assisted reproductive technology data 2011–2012. Fertil Steril 2016;106: 75-79.

246. Kushnir VA, Frattarelli JL. Aneuploidy in abortuses following IVF and ICSI. Journal of assisted reproduction and genetics 2009;26: 93-97.

247. Lampi KM, Lehtonen L, Tran PL, Suominen A, Lehti V, Banerjee PN, Gissler M, Brown AS, Sourander A. Risk of autism spectrum disorders in low birth weight and small for gestational age infants. The Journal of pediatrics 2012;161: 830-836.

248. Lancaster P, Huang J, Pedisich E. Australia’s mothers and babies 1991 (Perinatal statistics series no.1). 1994. Australian Institute of Health and Welfare, Canberra

249. Lancsar E, Louviere J. Conducting discrete choice experiments to inform healthcare decision making. Pharmacoeconomics 2008;26: 661-677.

250. Land JA, Evers JL. What is the most relevant standard of success in assisted reproduction? Defining outcome in ART: a Gordian knot of safety, efficacy and quality. Human reproduction 2004;19: 1046-1048.

251. Lande Y, Seidman DS, Maman E, Baum M, Hourvitz A. Why do couples discontinue unlimited free IVF treatments? Gynecological Endocrinology 2015;31: 233-236.

252. Landfeldt E, Jablonowska B, Norlander E, Persdotter-Eberg K, Thurin-Kjellberg A, Wramsby M, Ström O. Patient preferences for characteristics differentiating ovarian stimulation treatments. Human Reproduction 2012;27: 760-769.

253. Larsen U. Primary and secondary infertility in sub-Saharan Africa. International Journal of Epidemiology 2000;29: 285-291.

254. Lass A, Croucher C, Duffy S, Dawson K, Margara R, Winston RM. One thousand initiated cycles of in vitro fertilization in women≥ 40 years of age. Fertility and sterility 1998;70: 1030-1034.

255. Ledger WL. Demographics of infertility. Reproductive biomedicine online 2009;18: S11-S14.

256. Lee E, Chambers GM, Hale L, Illingworth P, Wilton L. Assisted reproductive technology (ART) cumulative live birth rates following preimplantation genetic diagnosis for aneuploidy (PGD‐A) or morphological assessment of embryos: A cohort analysis. Australian and New Zealand Journal of Obstetrics and Gynaecology 2017.

257. Lee E, Illingworth P, Wilton L, Chambers GM. The clinical effectiveness of preimplantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A): systematic review. Hum Reprod 2014;30: 473-483.

258. Lee H-L, McCulloh DH, Hodes-Wertz B, Adler A, McCaffrey C, Grifo JA. In vitro fertilization with preimplantation genetic screening improves implantation and live birth in women age 40 through 43. Journal of assisted reproduction and genetics 2015;32: 435-444.

259. Legro, Wu X, Scientific C, KT B, Farquhar C, BC F, B M. Improving the Reporting of Clinical Trials of Infertility Treatments (IMPRINT): modifying the CONSORT statement. Fertility and sterility 2014;102: 952-959. e915. 250

260. Legro RS, Myers E. Surrogate end-points or primary outcomes in clinical trials in women with polycystic ovary syndrome? Human Reproduction 2004;19: 1697-1704.

261. Lepkowski JM, Mosher WD, Davis KE, Groves RM, Van Hoewyk J. The 2006-2010 National Survey of Family Growth: sample design and analysis of a continuous survey. Vital and health statistics Series 2, Data evaluation and methods research 2010: 1-36.

262. Leridon H. Studies of fertility and fecundity: comparative approaches from demography and epidemiology. Comptes rendus biologies 2007;330: 339-346.

263. Leridon H, Slama R. The impact of a decline in fecundity and of pregnancy postponement on final number of children and demand for assisted reproduction technology. Human Reproduction 2008;23: 1312-1319.

264. Levine H, Jørgensen N, Martino-Andrade A, Mendiola J, Weksler-Derri D, Mindlis I, Pinotti R, Swan SH. Temporal trends in sperm count: a systematic review and meta- regression analysis. Human reproduction update 2017: 1-14.

265. Li Z, Zeki R, Hilder L, Sullivan EA. Australia’s mothers and babies 2011 (Perinatal statistics series no. 28). 2013, Canberra: Australian Institute of Health and Welfare.

266. Lie RT, Lyngstadass A, Ørstavik KH, Bakketeig LS, Jacobsen G, Tanbo T. Birth defects in children conceived by ICSI compared with children conceived by other IVF- methods; a meta-analysis. International journal of epidemiology 2004;34: 696-701.

267. López E, Joanne G, Daya S, Parrilla JJ, Abad L, Balasch J. Ovulation induction in women with polycystic ovary syndrome: randomized trial of clomiphene citrate versus low-dose recombinant FSH as first line therapy. Reproductive biomedicine online 2004;9: 382-390.

268. Lorgelly PK, Lawson KD, Fenwick EA, Briggs AH. Outcome measurement in economic evaluations of public health interventions: a role for the capability approach? International journal of environmental research and public health 2010;7: 2274-2289.

269. Louis GB. Fecundity and fertility Reproductive and perinatal epidemiology. 2011. Oxford University Press, New York, pp. 16-61.

270. Luke B, Brown MB, Wantman E, Lederman A, Gibbons W, Schattman GL, Lobo RA, Leach RE, Stern JE. Cumulative birth rates with linked assisted reproductive technology cycles. New England Journal of Medicine 2012;366: 2483-2491.

271. Macaldowie A, Wang A, Chambers GM, EA S. Assisted reproductive technology in Australia and New Zealand 2010. 2012, National Perinatal Epidemiology and Statistics Unit, the University of New South Wales, Sydney.

272. Macklon NS, Stouffer RL, Giudice LC, Fauser BCJM. The Science behind 25 Years of Ovarian Stimulation for in Vitro Fertilization. Endocrine Reviews 2006;27: 170-207.

273. Mahadevan MM, Trounson AO, Leeton JF. The relationship of tubal blockage, infertility of unknown cause, suspected male infertility, and endometriosis to success of in vitro fertilization and embryo transfer. Fertility and sterility 1983;40: 755-762.

274. Maheshwari A, McLernon D, Bhattacharya S. Cumulative live birth rate: time for a consensus? Human Reproduction 2015;30: 2703-2707.

251

275. Maheshwari A, Pandey S, Shetty A, Hamilton M, Bhattacharya S. Obstetric and perinatal outcomes in singleton pregnancies resulting from the transfer of frozen thawed versus fresh embryos generated through in vitro fertilization treatment: a systematic review and meta-analysis. Fertility and sterility 2012;98: 368-377. e369.

276. Malizia BA, Hacker MR, Penzias AS. Cumulative live-birth rates after in vitro fertilization. The New England journal of medicine 2009;2009: 236-243.

277. Mantakas A, Farrell T. The influence of increasing BMI in nulliparous women on pregnancy outcome. European Journal of Obstetrics & Gynecology and Reproductive Biology 2010;153: 43-46.

278. Marcoux S, Maheux R, Bérubé S, Endometriosis CCGo. Laparoscopic surgery in infertile women with minimal or mild endometriosis. New England Journal of Medicine 1997;337: 217-222.

279. Marinakis G, Nikolaou D. What is the role of assisted reproduction technology in the management of age-related infertility? Human Fertility 2011;14: 8-15.

280. Marino JL, Moore VM, Rumbold AR, Davies MJ. Fertility treatments and the young women who use them: an Australian cohort study. Human Reproduction 2010.

281. Marquard K, Westphal LM, Milki AA, Lathi RB. Etiology of recurrent pregnancy loss in women over the age of 35 years. Fertility and sterility 2010;94: 1473-1477.

282. Mascarenhas M, Sunkara SK, Antonisamy B, Kamath MS. Higher risk of preterm birth and low birth weight following oocyte donation: A systematic review and meta- analysis. European Journal of Obstetrics & Gynecology and Reproductive Biology 2017;218: 60-67.

283. Mascarenhas MN, Cheung H, Mathers CD, Stevens GA. Measuring infertility in populations: constructing a standard definition for use with demographic and reproductive health surveys. Population health metrics 2012a;10: 17.

284. Mascarenhas MN, Flaxman SR, Boerma T, Vanderpoel S, Stevens GA. National, regional, and global trends in infertility prevalence since 1990: a systematic analysis of 277 health surveys. PLoS Med 2012;9: e1001356.

285. Mastenbroek S, Repping S. Preimplantation genetic screening: back to the future. Human Reproduction 2014;29: 1846-1850.

286. Matorras R, Matorras F, Mendoza R, Rodríguez M, Remohí J, Rodríguez-Escudero FJ, Simón C. The implantation of every embryo facilitates the chances of the remaining embryos to implant in an IVF programme: a mathematical model to predict pregnancy and multiple pregnancy rates. Human Reproduction 2005;20: 2923-2931.

287. Matthews TJ, Hamilton BE. Delayed childbearing: more women are having their first child later in life. NCHS data brief 2009: 1-8.

288. Mauskopf JA, Paul JE, Grant DM, Stergachis A. The role of cost—consequence analysis in healthcare decision—making. Pharmacoeconomics 1998;13: 277-288.

289. Maxwell SM, Colls P, Hodes-Wertz B, McCulloh DH, McCaffrey C, Wells D, Munne S, Grifo JA. Why do euploid embryos miscarry? A case-control study comparing the rate of aneuploidy within presumed euploid embryos that resulted in miscarriage or live 252

birth using next-generation sequencing. Fertility and sterility 2016;106: 1414- 1419.e1415.

290. McClamrock HD, Jones HW, Adashi EY. Ovarian stimulation and intrauterine insemination at the quarter centennial: implications for the multiple births epidemic. Fertility and sterility 2012;97: 802-809.

291. McGregor M, Caro JJ. QALYs. Pharmacoeconomics 2006;24: 947-952.

292. McIntosh E. Costing methodology for applied cost–benefit analysis in health care. Applied Methods of Cost-Benefit Analysis in Health Care 2010;4: 55.

293. McIntosh E, Luengo-Fernandez R. Economic evaluation. Part 1: Introduction to the concepts of economic evaluation in health care. BMJ Sexual & Reproductive Health 2006;32: 107-112.

294. McLernon DJ, Maheshwari A, Lee AJ, Bhattacharya S. Cumulative live birth rates after one or more complete cycles of IVF: a population-based study of linked cycle data from 178 898 women. Human Reproduction 2016;31: 572-581.

295. Medicare Australia. MBS online. Medicare Australia 2015. Available at https://www.humanservices.gov.au/individuals/enablers/who-medicare-safety-net. Accessed 9 March 2018. 2015.

296. Mégarbané A, Ravel A, Mircher C, Sturtz F, Grattau Y, Rethoré M-O, Delabar J-M, Mobley WC. The 50th anniversary of the discovery of trisomy 21: the past, present, and future of research and treatment of Down syndrome. Genetics in Medicine 2009;11: 611-616.

297. Mertzanidou A, Wilton L, Cheng J, Spits C, Vanneste E, Moreau Y, Vermeesch J, Sermon K. Microarray analysis reveals abnormal chromosomal complements in over 70% of 14 normally developing human embryos. Hum Reprod 2013;28: 256-264.

298. Mesen TB, Mersereau JE, Kane JB, Steiner AZ. Optimal timing for elective egg freezing. Fertil Steril 2015;103: 1551-1556. e1554.

299. Meyer LR, Klipstein S, Hazlett WD, Nasta T, Mangan P, Karande VC. A prospective randomized controlled trial of preimplantation genetic screening in the “good prognosis” patient. Fertility and sterility 2009;91: 1731-1738.

300. Mikkelsen EM, Riis AH, Wise LA, Hatch EE, Rothman KJ, Cueto HT, Sørensen HT. Alcohol consumption and fecundability: prospective Danish cohort study. BMJ (Clinical research ed) 2016;354: i4262.

301. Miller LM, Hodgson R, Wong TY, Merrilees M, Norman RJ, Johnson NP. Single embryo transfer for all? Australian and New Zealand Journal of Obstetrics and Gynaecology 2016;56: 514-517.

302. Mills M, Rindfuss RR, McDonald P, Te Velde E. Why do people postpone parenthood? Reasons and social policy incentives. Human reproduction update 2011;17: 848-860.

303. Milman LW, Senapati S, Sammel MD, Cameron KD, Gracia C. Assessing reproductive choices of women and the likelihood of oocyte cryopreservation in the era of elective oocyte freezing. Fertil Steril 2017;107: 1214-1222. e1213.

253

304. Min JK, Breheny SA, MacLachlan V, Healy DL. What is the most relevant standard of success in assisted reproduction? The singleton, term gestation, live birth rate per cycle initiated: the BESST endpoint for assisted reproduction. Human Reproduction 2004;19: 3-7.

305. Min JK, Hughes E, Young D, Gysler M, Hemmings R, Cheung AP, Goodrow GJ, Min J, Senikas V, Wong BC-M. Elective single embryo transfer following in vitro fertilization. Journal of Obstetrics and Gynaecology Canada 2010;32: 363-377.

306. Mishra GD. Australian Longitudinal Study on Women's Health. 2014.

307. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M, Shekelle P, Stewart LA, Group P-P. Preferred reporting items for systematic review and meta- analysis protocols (PRISMA-P) 2015 statement. Systematic reviews 2015;4.

308. Moini A, Bahar L, Ashrafinia M, Eslami B, Hosseini R, Ashrafinia N. Fertility outcome after operative laparoscopy versus no treatment in infertile women with minimal or mild endometriosis. International journal of fertility & sterility 2012;5: 235.

309. Moolenaar LM, Vijgen SM, Hompes P, van der Veen F, Mol BWJ, Opmeer BC. Economic evaluation studies in reproductive medicine: a systematic review of methodologic quality. Fertility and sterility 2013;99: 1689-1694. e1681.

310. Moragianni VA, Penzias AS. Cumulative live-birth rates after assisted reproductive technology. Current Opinion in Obstetrics and gynecology 2010;22: 189-192.

311. Morris S, Devlin N, Parkin D. Economic analysis in health care, 2007. John Wiley & Sons.

312. Munné S, Cohen J. Advanced maternal age patients benefit from preimplantation genetic diagnosis of aneuploidy. Fertility and sterility 2017;107: 1145-1146.

313. Munné S, Grifo J, Wells D. Mosaicism:“survival of the fittest” versus “no embryo left behind”. Fertility and sterility 2016;105: 1146-1149.

314. Munne S, Kaplan B, Frattarelli J, Gysler M, Child T, Nakhuda G, Shamma F, Silverberg K, Kalista T, Oliver K. Global multicenter randomized controlled trial comparing single embryo transfer with embryo selected by preimplantation genetic screening using next-generation sequencing versus morphologic assessment. Fertility and sterility 2017;108: e19.

315. Munné S, Sultan KM, Weier H-U, Grifo JA, Cohen J, Rosenwaks Z. Assessment of numeric abnormalities of X, Y, 18, and 16 chromosomes in preimplantation human embryos before transfer. American journal of obstetrics and gynecology 1995;172: 1191-1201.

316. Munne S, Wells D, Cohen J. Technology requirements for preimplantation genetic diagnosis to improve assisted reproduction outcomes. Fertility and sterility 2010;94: 408-430.

317. Murray S, Norman J. Multiple pregnancies following assisted reproductive technologies–a happy consequence or double trouble? Seminars in Fetal and Neonatal Medicine. 2014. Elsevier, pp. 222-227.

254

318. Nagaoka SI, Hassold TJ, Hunt PA. Human aneuploidy: mechanisms and new insights into an age-old problem. Nature Reviews Genetics 2012;13: 493-504.

319. Nakayama K, Ueno NT. American Society of Clinical Oncology recommendations on fertility preservation should be implemented regardless of disease status or previous treatments. Journal of Clinical Oncology 2006;24: 5334-5335.

320. Nandi A, Gudi A, Shah A, Homburg R. An online survey of specialists’ opinion on first line management options for unexplained subfertility. Human Fertility 2015;18: 48-53.

321. Nasseri A, Mukherjee T, Grifo JA, Noyes N, Krey L, Copperman AB. Elevated day 3 serum follicle stimulating hormone and/or estradiol may predict fetal aneuploidy. Fertility and sterility 1999;71: 715-718.

322. National Institute for Health and Care Excellence. Fertility: assessment and treatment for people with fertility problems. Clinical Guidelines 2013; 156. 2013; .

323. Neumann PJ. What next for QALYs? Jama 2011;305: 1806-1807.

324. Neumann PJ, Johannesson M. The willingness to pay for in vitro fertilization: a pilot study using contingent valuation. Medical care 1994: 686-699.

325. Neumann PJ, Sanders GD, Russell LB, Siegel JE, Ganiats TG. Cost-effectiveness in health and medicine, 2016. Oxford University Press.

326. NICE. Guide to the methods of technology appraisal Process and Methods Guide. Accessed on Feb 2018. 2013.

327. NICE. National Institute for Health and Clinical Excellence. National Collaborating Centre for Women's and Children's Health. Fertility: Assessment and Treatment for People with Fertility Problems. London (UK); 2013 (Clinical guideline; no. 156) 2013.

328. NICE. NICE Guidence. Fertility Problems. Quality statement 7: Intracytoplasmic sperm injection. Available at: https://www.nice.org.uk/guidance/qs73/chapter/Quality- statement-7-Intracytoplasmic-sperm-injection (accessed on 23 Sept 2017). 2014.

329. Niinimäki M, Suikkari A-M, Mäkinen S, Söderström-Anttila V, Martikainen H. Elective single-embryo transfer in women aged 40–44 years. Human Reproduction 2012;28: 331-335.

330. Noyes N, Porcu E, Borini A. Over 900 oocyte cryopreservation babies born with no apparent increase in congenital anomalies. Reproductive biomedicine online 2009;18: 769-776.

331. Nyboe Andersen A, Erb K. Register data on assisted reproductive technology (ART) in Europe including a detailed description of ART in Denmark. International journal of andrology 2006;29: 12-16.

332. O’Brien B, Briggs A. Analysis of uncertainty in health care cost-effectiveness studies: an introduction to statistical issues and methods. Statistical methods in medical research 2002;11: 455-468.

333. Oakley L, Doyle P, Maconochie N. Lifetime prevalence of infertility and infertility treatment in the UK: results from a population-based survey of reproduction. Human Reproduction 2008;23: 447-450. 255

334. Oostenbrink JB, Koopmanschap MA, Rutten FF. Standardisation of costs. Pharmacoeconomics 2002;20: 443-454.

335. Palermo G, Joris H, Devroey P, Van Steirteghem AC. Pregnancies after intracytoplasmic injection of single spermatozoon into an oocyte. The Lancet 1992;340: 17-18.

336. Pandian Z, Gibreel A, Bhattacharya S. In vitro fertilisation for unexplained subfertility. The Cochrane Library 2015.

337. Pandian Z, Marjoribanks J, Ozturk O, Serour G, Bhattacharya S. Number of embryos for transfer following in vitro fertilisation or intra-cytoplasmic sperm injection: summary of a Cochrane review. Fertility and sterility 2014;102: 345-347.

338. Papaleo E, Pagliardini L, Vanni VS, Delprato D, Rubino P, Candiani M, Viganò P. A direct healthcare cost analysis of the cryopreserved versus fresh transfer policy at the blastocyst stage. Reproductive biomedicine online 2017;34: 19-26.

339. PBAC Guidelines. PBAC Guidelines. Overview and rationale of the economic evaluation. https://pbac.pbs.gov.au/section-3a/3a-1-overview-and-rationale-of- economic-evaluation.html. Accessed on Jan 2018. 2016.

340. PCOS Australian Alliance. Evidence-based guidelines for the assessment and management of polycystic ovary syndrome, 2011. Melbourne. Accessed on 24 Nov 17. https://jeanhailes.org.au/contents/documents/Resources/Tools/PCOS_evidence- based_guideline_for_assessment_and_management_pcos.pdf, 2011.

341. Pelinck M, Hoek A, Simons A, Heineman M. Efficacy of natural cycle IVF: a review of the literature. Human reproduction update 2002;8: 129-139.

342. Penzias A, Bendikson K, Butts S, Coutifaris C, Fossum G, Falcone T, Gitlin S, Gracia C, Hansen K, La Barbera A. Guidance on the limits to the number of embryos to transfer: a committee opinion. Fertility and sterility 2017;107: 901-903.

343. Petropanagos A, Cattapan A, Baylis F, Leader A. Social egg freezing: risk, benefits and other considerations. Canadian Medical Association Journal 2015;187: 666-669.

344. Petrou S, Gray A. Economic evaluation using decision analytical modelling: design, conduct, analysis, and reporting. BMJ (Clinical research ed) 2011;342: d1766.

345. Pinborg A. IVF/ICSI twin pregnancies: risks and prevention. Human reproduction update 2005;11: 575-593.

346. Pinborg A, Henningsen A-KA, Malchau SS, Loft A. Congenital anomalies after assisted reproductive technology. Fertility and sterility 2013;99: 327-332.

347. Poder TG, He J, Simard C, Pasquier J-C. Willingness to pay for ovulation induction treatment in case of WHO II anovulation: a study using the contingent valuation method. Patient preference and adherence 2014;8: 1337.

348. Practice Committee of American Society for Reproductive Medicine. Multiple gestation associated with infertility therapy: an American Society for Reproductive Medicine Practice Committee opinion,. Fertility and sterility 2012;97: 825-834.

256

349. Practice Committee of the American Society for Reproductive Medicine. Effectiveness and treatment for unexplained infertility. Fertility and sterility 2006;86: S111-S114.

350. Practice Committee of the American Society for Reproductive Medicine. Age-related fertility decline: a committee opinion. Fertility and sterility 2008;90: S154-S155.

351. Practice Committee of the American Society for Reproductive Medicine. Committee opinion: role of tubal surgery in the era of assisted reproductive technology. Fertility and sterility 2012;97: 539-545.

352. Practice Committee of the American Society for Reproductive Medicine. Diagnostic evaluation of the infertile female: a committee opinion. Fertility and sterility 2012;98: 302-307.

353. Practice Committee of the American Society for Reproductive Medicine. Intracytoplasmic sperm injection (ICSI) for non-male factor infertility: a committee opinion. Fertility and sterility 2012;98: 1395-1399.

354. Practice Committee of the American Society for Reproductive Medicine. Multiple gestation associated with infertility therapy: an American Society for Reproductive Medicine Practice Committee opinion. Fertility and sterility 2012;97: 825-834.

355. Practice Committee of the American Society for Reproductive Medicine. Definitions of infertility and recurrent pregnancy loss: a committee opinion. Fertility and sterility 2013;99: 63.

356. Practice Committee of the American Society for Reproductive Medicine. Diagnostic evaluation of the infertile female: a committee opinion. Fertility and sterility 2015;103: 302-307.

357. Pritts EA. Treatment of the infertile patient with polycystic ovarian syndrome. Obstetrical & gynecological survey 2002;57: 587-597.

358. Puri M, Aggarwal R. Ovarian Stimulation in IUI Principles and Practice of Controlled Ovarian Stimulation in ART. 2015. Springer, pp. 61-70.

359. Rajkhowa M, McConnell A, Thomas G. Reasons for discontinuation of IVF treatment: a questionnaire study. Human reproduction 2005;21: 358-363.

360. Rawlins MD, Culyer AJ. National Institute for Clinical Excellence and its value judgments. BMJ: British Medical Journal 2004;329: 224.

361. Reinblatt SL, Son W-Y, Shalom-Paz E, Holzer H. Controversies in IVM. Journal of assisted reproduction and genetics 2011;28: 525.

362. Reindollar RH, Regan MM, Neumann PJ, Levine B-S, Thornton KL, Alper MM, Goldman MB. A randomized clinical trial to evaluate optimal treatment for unexplained infertility: the fast track and standard treatment (FASTT) trial. Fertility and sterility 2010;94: 888-899.

363. Rimm AA, Katayama AC, Diaz M, Katayama KP. A meta-analysis of controlled studies comparing major malformation rates in IVF and ICSI infants with naturally conceived children. Journal of assisted reproduction and genetics 2004;21: 437-443.

257

364. Rimm AA, Katayama AC, Katayama KP. A meta-analysis of the impact of IVF and ICSI on major malformations after adjusting for the effect of subfertility. Journal of assisted reproduction and genetics 2011;28: 699-705.

365. Rittenberg V, Seshadri S, Sunkara SK, Sobaleva S, Oteng-Ntim E, El-Toukhy T. Effect of body mass index on IVF treatment outcome: an updated systematic review and meta- analysis. Reproductive biomedicine online 2011;23: 421-439.

366. Robinson JN, Forman RG, Nicholson SC, Maciocia LR, Barlow DH. A comparison of intrauterine insemination in superovulated cycles to intercourse in couples where the male is receiving steroids for the treatment of autoimmune infertility. Fertility and sterility 1995;63: 1260-1266.

367. Rock and Menkin, . In vitro fertilization and cleavage of human ovarian eggs. Science 1944;100: 105-107.

368. Romanski P, Goldman R, Farland L, Srouji S, Racowsky C. impact of quality of the entire embryo cohort on implantation potential of the transferred blastocyst. Fertility and sterility 2017;108: e341.

369. Rooney KL, Domar AD. The impact of lifestyle behaviors on infertility treatment outcome. Current Opinion in Obstetrics and Gynecology 2014;26: 181-185.

370. Roque M, Lattes K, Serra S, Solà I, Geber S, Carreras R, Checa MA. Fresh embryo transfer versus frozen embryo transfer in in vitro fertilization cycles: a systematic review and meta-analysis. Fertility and sterility 2013;99: 156-162.

371. Roque M, Valle M, Guimaraes F, Sampaio M, Geber S. Cost-Effectiveness of the Freeze-All Policy. JBRA assisted reproduction 2015;19: 125-130.

372. Rothwell PM. External validity of randomised controlled trials:“to whom do the results of this trial apply?”. The Lancet 2005;365: 82-93.

373. Rubino P, Viganò P, Luddi A, Piomboni P. The ICSI procedure from past to future: a systematic review of the more controversial aspects. Human reproduction update 2015;22: 194-227.

374. Rubio C, Bellver J, Rodrigo L, Bosch E, Mercader A, Vidal C, De los Santos MJ, Giles J, Labarta E, Domingo J. Preimplantation genetic screening using fluorescence in situ hybridization in patients with repetitive implantation failure and advanced maternal age: two randomized trials. Fertility and sterility 2013;99: 1400-1407.

375. Rubio C, Bellver J, Rodrigo L, Castillón G, Guillén A, Vidal C, Giles J, Ferrando M, Cabanillas S, Remohí J. In vitro fertilization with preimplantation genetic diagnosis for aneuploidies in advanced maternal age: a randomized, controlled study. Fertility and sterility 2017: DOI 10.1016/j.fertnstert.2017.1003.1011.

376. Rubio C, Simon C, Vidal F, Rodrigo L, Pehlivan T, Remohi J, Pellicer A. Chromosomal abnormalities and embryo development in recurrent miscarriage couples. Human Reproduction 2003;18: 182-188.

377. Ryan M. Using willingness to pay to assess the benefits of assisted reproductive techniques. Health economics 1996;5: 543-558.

258

378. Ryan M. Using conjoint analysis to take account of patient preferences and go beyond health outcomes: an application to in vitro fertilisation. Social Science & Medicine 1999;48: 535-546.

379. Ryan M. A comparison of stated preference methods for estimating monetary values. Health economics 2004;13: 291-296.

380. Ryan M, Farrar S. Using conjoint analysis to elicit preferences for health care. BMJ: British Medical Journal 2000;320: 1530.

381. Sadeghi MR. Unexplained infertility, the controversial matter in management of infertile couples. Journal of reproduction & infertility 2015;16: 1.

382. Sauer MV. Reproduction at an advanced maternal age and maternal health. Fertility and sterility 2015;103: 1136-1143.

383. Scheike TH, Keiding N. Design and analysis of time-to-pregnancy. Statistical methods in medical research 2006;15: 127-140.

384. Schieve LA, Devine O, Boyle CA, Petrini JR, Warner L. Estimation of the contribution of non–assisted reproductive technology ovulation stimulation fertility treatments to US singleton and multiple births. American Journal of Epidemiology 2009;170: 1396-1407.

385. Schmidt L, Sobotka T, Bentzen JG, Nyboe Andersen A. Demographic and medical consequences of the postponement of parenthood. Human reproduction update 2012;18: 29-43.

386. Schoolcraft WB, Fragouli E, Stevens J, Munne S, Katz-Jaffe MG, Wells D. Clinical application of comprehensive chromosomal screening at the blastocyst stage. Fertility and sterility 2010;94: 1700-1706.

387. Schoolcraft WB, Katz-Jaffe MG. Comprehensive chromosome screening of trophectoderm with vitrification facilitates elective single-embryo transfer for infertile women with advanced maternal age. Fertility and sterility 2013;100: 615-619.

388. Schoolcraft WB, Katz-Jaffe MG, Stevens J, Rawlins M, Munne S. Preimplantation aneuploidy testing for infertile patients of advanced maternal age: a randomized prospective trial. Fertility and sterility 2009;92: 157-162.

389. Schwartz D, Mayaux M. Female fecundity as a function of age: results of artificial insemination in 2193 nulliparous women with azoospermic husbands. Obstetrical & Gynecological Survey 1982;37: 548-550.

390. Scotland G, McLernon D, Kurinczuk J, McNamee P, Harrild K, Lyall H, Rajkhowa M, Hamilton M, Bhattacharya S. Minimising twins in in vitro fertilisation: a modelling study assessing the costs, consequences and cost–utility of elective single versus double embryo transfer over a 20‐year time horizon. BJOG: An International Journal of Obstetrics & Gynaecology 2011;118: 1073-1083.

391. Scott KL, Hong KH, Scott RT. Selecting the optimal time to perform biopsy for preimplantation genetic testing. Fertility and sterility 2013;100: 608-614.

392. Sculpher MJ, Claxton K, Drummond M, McCabe C. Whither trial‐based economic evaluation for health care decision making? Health economics 2006;15: 677-687.

259

393. Sermon K. Novel technologies emerging for preimplantation genetic diagnosis and preimplantation genetic testing for aneuploidy. Expert review of molecular diagnostics 2017;17: 71-82.

394. Sermon K, Capalbo A, Cohen J, Coonen E, De Rycke M, De Vos A, Delhanty J, Fiorentino F, Gleicher N, Griesinger G et al. The why, the how and the when of PGS 2.0: current practices and expert opinions of fertility specialists, molecular biologists, and embryologists. MHR: Basic science of reproductive medicine 2016;22: 845-857.

395. Seyhan A, Ata B, Son W-Y, Dahan MH, Tan SL. Comparison of complication rates and pain scores after transvaginal ultrasound–guided oocyte pickup procedures for in vitro maturation and in vitro fertilization cycles. Fertility and sterility 2014;101: 705-709.

396. Shapiro BS, Daneshmand ST, Garner FC, Aguirre M, Hudson C. Clinical rationale for cryopreservation of entire embryo cohorts in lieu of fresh transfer. Fertility and sterility 2014;102: 3-9.

397. Shapiro BS, Daneshmand ST, Garner FC, Aguirre M, Hudson C, Thomas S. Evidence of impaired endometrial receptivity after ovarian stimulation for in vitro fertilization: a prospective randomized trial comparing fresh and frozen–thawed embryo transfer in normal responders. Fertility and sterility 2011;96: 344-348.

398. Sharma R, Biedenharn KR, Fedor JM, Agarwal A. Lifestyle factors and reproductive health: taking control of your fertility. Reproductive Biology and Endocrinology 2013;11: 66.

399. Sharma R, Harlev A, Agarwal A, Esteves SC. Cigarette smoking and semen quality: a new meta-analysis examining the effect of the 2010 World Health Organization laboratory methods for the examination of human semen. European urology 2016;70: 635-645.

400. Sharma V, Allgar V, Rajkhowa M. Factors influencing the cumulative conception rate and discontinuation of in vitro fertilization treatment for infertility. Fertility and sterility 2002;78: 40-46.

401. Sher G, Keskintepe L, Keskintepe M, Maassarani G, Tortoriello D, Brody S. Genetic analysis of human embryos by metaphase comparative genomic hybridization (mCGH) improves efficiency of IVF by increasing embryo implantation rate and reducing multiple pregnancies and spontaneous miscarriages. Fertility and sterility 2009;92: 1886-1894.

402. Shiroiwa T, Sung YK, Fukuda T, Lang HC, Bae SC, Tsutani K. International survey on willingness‐to‐pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health economics 2010;19: 422-437.

403. Simoens S. Health economic assessment: a methodological primer. International journal of environmental research and public health 2009;6: 2950-2966.

404. Siristatidis C, Sergentanis TN, Vogiatzi P, Kanavidis P, Chrelias C, Papantoniou N, Psaltopoulou T. In vitro maturation in women with vs. without polycystic ovarian syndrome: A systematic review and meta-analysis. PloS one 2015;10: e0134696.

405. Slama R, Hansen OKH, Ducot B, Bohet A, Sorensen D, Giorgis Allemand L, Eijkemans M, Rosetta L, Thalabard J-C, Keiding N. Estimation of the frequency of

260

involuntary infertility on a nation-wide basis. Human reproduction 2012;27: 1489- 1498.

406. Smarr MM, Sapra KJ, Gemmill A, Kahn LG, Wise LA, Lynch CD, Factor-Litvak P, Mumford SL, Skakkebaek NE, Slama R et al. Is human fecundity changing? A discussion of research and data gaps precluding us from having an answer. Human Reproduction 2017;32: 499-504.

407. Smith AD, Tilling K, Nelson SM, Lawlor DA. Live-birth rate associated with repeat in vitro fertilization treatment cycles. Jama 2015;314: 2654-2662.

408. Smith S, Pfeifer SM, Collins JA. Diagnosis and management of female infertility. Jama 2003;290: 1767-1770.

409. Sozou PD, Hartshorne GM. Time to pregnancy: a computational method for using the duration of non-conception for predicting conception. PLoS One 2012;7: e46544.

410. Speroff L. The effect of aging on fertility. Current opinion in obstetrics & gynecology 1994;6: 115-120.

411. Spinella F, Fiorentino F, Biricik A, Bono S, Ruberti A, Cotroneo E, Baldi M, Cursio E, Minasi MG, Greco E. Extent of chromosomal mosaicism influences the clinical outcome of in vitro fertilization treatments. Fertility and sterility 2018;109: 77-83.

412. Stern JE, Brown MB, Luke B, Wantman E, Lederman A, Missmer SA, Hornstein MD. Calculating cumulative live-birth rates from linked cycles of assisted reproductive technology (ART): data from the Massachusetts SART CORS. Fertility and sterility 2010;94: 1334-1340.

413. Stern JE, Hickman TN, Kinzer D, Penzias AS, Ball GD, Gibbons WE. Can the Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS) be used to accurately report clinic total reproductive potential (TRP)? Fertility and sterility 2012;97: 886-889.

414. Steures P, van der Steeg JW, Hompes PG, Bossuyt PM, Habbema JDF, Eijkemans MJ, Schöls WA, Burggraaff JM, van der Veen F, Mol BW. Effectiveness of intrauterine insemination in subfertile couples with an isolated cervical factor: a randomized clinical trial. Fertility and sterility 2007;88: 1692-1696.

415. Steures P, van der Steeg JW, Hompes PG, Habbema JDF, Eijkemans MJ, Broekmans FJ, Verhoeve HR, Bossuyt PM, van der Veen F, Mol BW. Intrauterine insemination with controlled ovarian hyperstimulation versus expectant management for couples with unexplained subfertility and an intermediate prognosis: a randomised clinical trial. The Lancet 2006;368: 216-221.

416. Steures P, van der Steeg JW, Mol BW, Eijkemans MJ, van der Veen F, Habbema JDF, Hompes PG, Bossuyt PM, Verhoeve HR, van Kasteren YM. Prediction of an ongoing pregnancy after intrauterine insemination. Fertility and sterility 2004;82: 45-51.

417. Stolwijk A, Wetzels A, Braat D. Cumulative probability of achieving an ongoing pregnancy after in-vitro fertilization and intracytoplasmic sperm injection according to a woman's age, subfertility diagnosis and primary or secondary subfertility. Human Reproduction 2000;15: 203-209.

418. Stoop D. Social oocyte freezing. Facts, views & vision in ObGyn 2010;2: 31. 261

419. Stoop D. Oocyte vitrification for elective fertility preservation: lessons for patient counseling. Fertility and sterility 2016;105: 603-604.

420. Strathy JH, Molgaard CA, Coulam CB, Melton LJ. Endometriosis and infertility: a laparoscopic study of endometriosis among fertile and infertile women. Fertility and sterility 1982;38: 667-672.

421. Sunkara SK, LaMarca A, Polyzos NP, Seed PT, Khalaf Y. Live birth and perinatal outcomes following stimulated and unstimulated IVF: analysis of over two decades of a nationwide data. Human Reproduction 2016;31: 2261-2267.

422. Sunkara SK, Rittenberg V, Raine-Fenning N, Bhattacharya S, Zamora J, Coomarasamy A. Association between the number of eggs and live birth in IVF treatment: an analysis of 400 135 treatment cycles. Human reproduction 2011;26: 1768-1774.

423. Svensson A, Connolly M, Gallo F, Hägglund L. Long-term fiscal implications of subsidizing in-vitro fertilization in Sweden: a lifetime tax perspective. Scandinavian journal of public health 2008;36: 841-849.

424. Tannus S, Cohen Y, Son W-Y, Shavit T, Dahan M-H. Cumulative live birth rate following elective single blastocyst transfer compared with double blastocyst transfer in women aged 40 years and over. Reproductive biomedicine online 2017.

425. Tannus S, Son W-Y, Gilman A, Younes G, Shavit T, Dahan M-H. The role of intracytoplasmic sperm injection in non-male factor infertility in advanced maternal age. Human Reproduction 2017;32: 119-124.

426. Tarricone R. Cost-of-illness analysis: what room in health economics? Health policy (Amsterdam, Netherlands) 2006;77: 51-63.

427. Taylor R, Drummond M, Salkeld G, Sullivan S. Inclusion of cost effectiveness in licensing requirements of new drugs: the fourth hurdle. BMJ: British Medical Journal 2004;329: 972.

428. Taylor TH, Gitlin SA, Patrick JL, Crain JL, Wilson JM, Griffin DK. The origin, mechanisms, incidence and clinical consequences of chromosomal mosaicism in humans. Human reproduction update 2014;20: 571-581.

429. Te Velde ER, Eijkemans R, Habbema H. Variation in couple fecundity and time to pregnancy, an essential concept in human reproduction. The Lancet 2000;355: 1928- 1929.

430. Te Velde ER, Pearson PL. The variability of female reproductive ageing. Human reproduction update 2002;8: 141-154.

431. Templeton A, Fraser C, Thompson B. The epidemiology of infertility in Aberdeen. BMJ (Clinical research ed) 1990;301: 148-152.

432. The Thessaloniki ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group. Consensus on infertility treatment related to polycystic ovary syndrome. Fertility and sterility 2008;89: 505-522.

433. Thoma ME, McLain AC, Louis JF, King RB, Trumble AC, Sundaram R, Louis GMB. Prevalence of infertility in the United States as estimated by the current duration

262

approach and a traditional constructed approach. Fertility and sterility 2013;99: 1324- 1331. e1321.

434. Thurin A, Hausken J, Hillensjö T, Jablonowska B, Pinborg A, Strandell A, Bergh C. Elective single-embryo transfer versus double-embryo transfer in in vitro fertilization. New England Journal of Medicine 2004;351: 2392-2402.

435. Tiitinen A, Hydén-Granskog C, Gissler M. What is the most relevant standard of success in assisted reproduction? The value of cryopreservation on cumulative pregnancy rates per single oocyte retrieval should not be forgotten. Human Reproduction 2004;19: 2439-2441.

436. Tjon-Kon-Fat R, Bensdorp A, Bossuyt P, Koks C, Oosterhuis G, Hoek A, Hompes P, Broekmans F, Verhoeve H, De Bruin J. Is IVF—served two different ways—more cost- effective than IUI with controlled ovarian hyperstimulation? Human Reproduction 2015;30: 2331-2339.

437. Tournaye H. Evidence-based management of male subfertility. Current Opinion in Obstetrics and Gynecology 2006;18: 253-259.

438. Troude P, Bailly E, Guibert J, Bouyer J, De La Rochebrochard E. Spontaneous pregnancies among couples previously treated by in vitro fertilization. Fertility and sterility 2012;98: 63-68.

439. Tummon IS, Asher LJ, Martin JS, Tulandi T. Randomized controlled trial of superovulation and insemination for infertility associated with minimal or mild endometriosis. Fertility and sterility 1997;68: 8-12.

440. Ubaldi FM, Capalbo A, Colamaria S, Ferrero S, Maggiulli R, Vajta G, Sapienza F, Cimadomo D, Giuliani M, Gravotta E. Reduction of multiple pregnancies in the advanced maternal age population after implementation of an elective single embryo transfer policy coupled with enhanced embryo selection: pre-and post-intervention study. Human Reproduction 2015;30: 2097-2106.

441. Ubaldi FM, Cimadomo D, Capalbo A, Vaiarelli A, Buffo L, Trabucco E, Ferrero S, Albani E, Rienzi L, Setti PEL. Preimplantation genetic diagnosis for aneuploidy testing in women older than 44 years: a multicenter experience. Fertility and sterility 2017;107: 1173-1180.

442. Usta IM, Nassar AH. Advanced maternal age. Part I: obstetric complications. American journal of perinatology 2008;25: 521-534.

443. van den Boogaard NM, Bensdorp AJ, Rengerink KO, Barnhart K, Bhattacharya S, Custers IM, Coutifaris C, Goverde AJ, Guzick DS, Hughes EC. Prognostic profiles and the effectiveness of assisted conception: secondary analyses of individual patient data. Hum Reprod Update 2013: dmt035.

444. van Loendersloot LL, Moolenaar LM, Mol BWJ, Repping S, van der Veen F, Goddijn M. Expanding reproductive lifespan: a cost-effectiveness study on oocyte freezing. Hum Reprod 2011;26: 3054-3060.

445. van Loendersloot LL, van Wely M, Limpens J, Bossuyt PMM, Repping S, van der Veen F. Predictive factors in in vitro fertilization (IVF): a systematic review and meta- analysis. Human reproduction update 2010;16: 577-589.

263

446. Veltman‐Verhulst SM, Hughes E, Ayeleke RO, Cohlen BJ. Intra‐uterine insemination for unexplained subfertility. The Cochrane Library 2016.

447. Venkatachalam L. The contingent valuation method: a review. Environmental impact assessment review 2004;24: 89-124.

448. Venn A, Epid GD, Lumley J. Births After a Period of Infertility in Victorian Women 1982–199. Australian and New Zealand journal of obstetrics and gynaecology 1993;33: 379-384.

449. Verberg M, Eijkemans M, Heijnen E, Broekmans F, de Klerk C, Fauser B, Macklon N. Why do couples drop-out from IVF treatment? A prospective cohort study. Human Reproduction 2008;23: 2050-2055.

450. Verhaak C, Lintsen A, Evers A, Braat D. Who is at risk of emotional problems and how do you know? Screening of women going for IVF treatment. Human Reproduction 2010;25: 1234-1240.

451. Verhaak C, Smeenk J, Van Minnen A, Kremer J, Kraaimaat F. A longitudinal, prospective study on emotional adjustment before, during and after consecutive fertility treatment cycles. Human reproduction 2005;20: 2253-2260.

452. Verhagen TEM, Dumoulin JCM, Evers JLH, Land JA. What is the most accurate estimate of pregnancy rates in IVF dropouts? Human Reproduction 2008;23: 1793- 1799.

453. Verlinsky Y, Ginsberg N, Lifchez A, Valle J, Moise J, Strom CM. Analysis of the first polar body: preconception genetic diagnosis. Human reproduction 1990;5: 826-829.

454. Vialard F, Boitrelle F, Molina-Gomes D, Selva J. Predisposition to aneuploidy in the oocyte. Cytogenetic and genome research 2011;133: 127-135.

455. Voullaire L, Collins V, Callaghan T, McBain J, Williamson R, Wilton L. High incidence of complex chromosome abnormality in cleavage embryos from patients with repeated implantation failure. Fertility and sterility 2007;87: 1053-1058.

456. Wallace WHB, Anderson RA, Irvine DS. Fertility preservation for young patients with cancer: who is at risk and what can be offered? The lancet oncology 2005;6: 209-218.

457. Walls M, Hunter T, Ryan J, Keelan J, Nathan E, Hart R. In vitro maturation as an alternative to standard in vitro fertilization for patients diagnosed with polycystic ovaries: a comparative analysis of fresh, frozen and cumulative cycle outcomes. Human Reproduction 2014;30: 88-96.

458. Wang J, Sauer MV. In vitro fertilization (IVF): a review of 3 decades of clinical innovation and technological advancement. Therapeutics and clinical risk management 2006;2: 355.

459. Wang R, Kim BV, van Wely M, Johnson NP, Costello MF, Zhang H, Ng EHY, Legro RS, Bhattacharya S, Norman RJ. Treatment strategies for women with WHO group II anovulation: systematic review and network meta-analysis. BMJ (Clinical research ed) 2017;356: j138.

264

460. Wang X, Chen C, Wang L, Chen D, Guang W, French J, Group RHS. Conception, early pregnancy loss, and time to clinical pregnancy: a population-based prospective study. Fertility and sterility 2003;79: 577-584.

461. Wang YA, Farquhar C, Sullivan EA. Donor age is a major determinant of success of oocyte donation/recipient programme. Human reproduction 2011;27: 118-125.

462. Wang YA, Farquhar C, Sullivan EA. Donor age is a major determinant of success of oocyte donation/recipient programme. Hum Reprod 2012;27: 118-125.

463. Webb S, Holman DA. A survey of infertility, surgical sterility and associated reproductive disability in Perth, Western Australia. Australian journal of public health 1992;16: 376-381.

464. Wells D, Levy B. Cytogenetics in reproductive medicine: the contribution of comparative genomic hybridization (CGH). Bioessays 2003;25: 289-300.

465. Wen J, Jiang J, Ding C, Dai J, Liu Y, Xia Y, Liu J, Hu Z. Birth defects in children conceived by in vitro fertilization and intracytoplasmic sperm injection: a meta- analysis. Fertility and sterility 2012;97: 1331-1337.e1334.

466. Wennerholm U, Bergh C. What is the most relevant standard of success in assisted reproduction? Singleton live births should also include preterm births. Human Reproduction 2004;19: 1943-1945.

467. Werbrouck E, Spiessens C, Meuleman C, D’Hooghe T. No difference in cycle pregnancy rate and in cumulative live-birth rate between women with surgically treated minimal to mild endometriosis and women with unexplained infertility after controlled ovarian hyperstimulation and intrauterine insemination. Fertility and sterility 2006;86: 566-571.

468. Werner MD, Leondires MP, Schoolcraft WB, Miller BT, Copperman AB, Robins ED, Arredondo F, Hickman TN, Gutmann J, Schillings WJ. Clinically recognizable error rate after the transfer of comprehensive chromosomal screened euploid embryos is low. Fertility and sterility 2014;102: 1613-1618.

469. Wilkinson J, Roberts S, Vail A. Developments in IVF warrant the adoption of new performance indicators for ART clinics, but do not justify the abandonment of patient- centred measures. Human Reproduction 2017;32: 1155-1159.

470. Wilkinson J, Roberts SA, Showell M, Brison DR, Vail A. No common denominator: a review of outcome measures in IVF RCTs. Hum Reprod 2016;31: 2714-2722.

471. Williams RS, Doody KJ, Schattman GL, Adashi EY. Public reporting of assisted reproductive technology outcomes: past, present, and future. American journal of obstetrics and gynecology 2015;212: 157-162.

472. Wilson C, Pressat R. The Dictionary of Demography, 1985. New York:Blackwell.

473. Wilton L, Wells D. Use of Comparative Genomic Hybridisation (CGH) and Microarray. Human Gametes and Preimplantation Embryos: Assessment and Diagnosis 2013: 147.

474. Wiser A, Shalom-Paz E, Reinblatt SL, Son W-Y, Das M, Tulandi T, Holzer H. Ovarian stimulation and intrauterine insemination in women aged 40years or more. Reproductive biomedicine online 2012;24: 170-173. 265

475. Wiser A, Son W-Y, Shalom-Paz E, Reinblatt SL, Tulandi T, Holzer H. How old is too old for in vitro maturation (IVM) treatment? European Journal of Obstetrics & Gynecology and Reproductive Biology 2011;159: 381-383.

476. Wong KM, Mastenbroek S, Repping S. Cryopreservation of human embryos and its contribution to in vitro fertilization success rates. Fertility and sterility 2014;102: 19-26.

477. Wong KM, Repping S, Mastenbroek S. Limitations of embryo selection methods Seminars in reproductive medicine. 2014. Thieme Medical Publishers, pp. 127-133.

478. Wordsworth S, Buchanan J, Mollison J, Harrild K, Robertson L, Tay C, Harrold A, McQueen D, Lyall H, Johnston L. Clomifene citrate and intrauterine insemination as first-line treatments for unexplained infertility: are they cost-effective? Human reproduction 2011;26: 369-375.

479. World Health Organization. Sexual and reproductive health: infertility is a global public health issue. http://www.who.int/reproductivehealth/topics/infertility/perspective/en/ Last accessed on 1 February 17 2017 2014.

480. Wright VC, Chang J, Jeng G. Assisted reproductive technology surveillance- United States, 2004. MMWR Surveill Summ 2007; 56(6):1-22, 2007.

481. Wyndham N, Figueira PGM, Patrizio P. A persistent misperception: assisted reproductive technology can reverse the “aged biological clock”. Fertility and sterility 2012;97: 1044-1047.

482. Zegers-Hochschild F, Adamson GD, Dyer S, Racowsky C, de Mouzon J, Sokol R, Rienzi L, Sunde A, Schmidt L, Cooke ID. The International Glossary on Infertility and Fertility Care, 2017. 2017. Oxford University Press.

483. Zegers-Hochschild F, Adamson GD, Dyer S, Racowsky C, de Mouzon J, Sokol R, Rienzi L, Sunde A, Schmidt L, Cooke ID et al. The International Glossary on Infertility and Fertility Care, 2017. Human Reproduction 2017;32: 1786-1801.

266

Appendix

Appendix 1 – The clinical effectivenss of PGD-A in all 24 chromosomes: systematic review

267

268

269

270

271

272

273

274

275

276

277

Appendix 2- ART CLBR following PGD-A or morphological assessment of embryos: a cohort analysis

278

279

280

281

282

283

284

285

Appendix 3: Downs and Black checklist

286

287

288

Supplementary Table 1: Detail of each study included in the updated systematic review First Author, Title Study type Population (Selection criteria) Year Lee et al, 2015 In vitro fertilisation with Retrospective Women aged 40 - 43 years at OPU during the period 1 preimplantation genetic cohort study Jan 2011 - 31 Dec 2012. Women using donor oocytes screening improves were excluded. implantation and live birth in women age 40 through 43 *Ubaldi et al, 2015 Reduction of multiple Retrospective Women aged over 35 years and had OPU between Jan pregnancies in the advanced Longitudinal 2010 and Dec 2013. PGD-A was offered to patients maternal age population after cohort study with a high risk of producing aneuploidy embryos, implementation of an elective when female age was over 39 years, patients single embryo transfer policy experiencing pregnancy loss (2 consecutive coupled with enhanced embryo miscarriages) and multiple implantation failures (3 selection: pre-and post- good quality embryos transferred. Cycles involving intervention study PGD for monogenic disease or chromosomal abnormalities were excluded Ubaldi et al, 2017 Preimplantation genetic Longitudinal Only women aged over 43 years with at least three diagnosis for aneuploidy testing observational study antral follicles on the day before starting the in women older than 44 years: a stimulation protocol; no history of previous no multicenter experience." response to the controlled ovarian stimulation; patient undertaking treatment between Apr 2013 and Jan 2016. Exclusion criteria includes positive serology for Hep B and C or human immune deficiency virus, monogenic disease or chromosomal structural abnormalities and maternal diseases

289

First Author, Sample size Population characteristics Types of Intervention Year Lee et al, 2015 Primary analysis: PGD-A group= Primary analysis: PGD-A PGD-A group: TE biopsy using aCGH. (con’t) 170 patients; Control group = 450 group: mean age ± SD: 41.2± Embryos that have developed to the expanded patients. 1.1; Control group: mean age ± blastocysts stage (Grade 2Bc or 2Cb or better) SD: 41.3± 1.1. were biopsied and vitrified for subsequent Subanalysis: Limited to patients FET cycle. (in both PGD-A group and Control Subanalysis: not reported group) with embryos that Control group: embryos were assessed on day developed to the expanded 5 or 6 based on morphologic criteria alone for blastocyst stage with differentiation fresh transfer and excess embryos of suitable of ICM and TE cells (i.e Grade 2Bc good quality (Grades 2Bc or 2Cb or better) or 2Cb or better): PGD-A group = were vitrified for FET. 49 patients, Control group was split into patients with fresh transfer (Control-fresh transfer, n=127 patients) and patients with FET (Control-FET, n= 28 patients)

*Ubaldi et al, PGD-A group: 328 cycles; Control Mean age at oocyte retrieval ± PGD-A group: TE biopsy and vitrified for 2015 group: 1855 cycles SD (years): PGD-A group: subsequent FET; Control group: (con’t) 39.6 ± 2.2 versus Control Morphological assessment of embryos alone group: 39.5 ± 2.2

Ubaldi et al, Study included 137 women and 150 Mean age ± SD (range): 44.7 ± TE biopsy with qPCR for subsequent FET 2017 cycles (No control group) 0.7 (44.0-46.7 years) (con’t)

290

First Author, Number of embryos Implantation rates Pregnancy rates Delivery or live-birth rates Year transferred

Lee et al, 2015 Primary analysis: Primary analysis: Primary analysis: Clinical Primary analysis: Live-birth (con’t) PGD-A group: 1.12 PGD-A group 51% pregnancy per transfer: per transfer: PGD-A group: embryos per FET versus Control group: PGD-A group: 55% per FET 51% per FET versus Control versus Control group: 16.3% (P-value not versus Control group: 36.6% group: 22.7% per fresh embryo 2.67 embryo per reported). per fresh embryo transfer (P- transfer (P-value not reported). fresh embryo transfer Subanalysis: PGD-A value not reported). (P-value not group: 50.9% versus Subanalysis: not reported Subanalysis: Live-delivery per reported). Control- fresh transfer embryo transferred: PGD-A Subanalysis: PGD-A group: 23.8% and group: 45.5% versus Control- group: 1.12 embryos Control-FET group: and fresh group: 15.8 and Control- per FET; Control 25.4%. The implantation FET group: 19%. The live- groups: -Control- rate is significantly delvery rate per embryo fresh transfer group: higher for PGD-A group transferred is significantly 2.38 per fresh compared to control- higher in PGD-A compared to transfer; Control- FET group (p=0.0072). control-FET (p=0.0028). FET group: 2.25 per NS between control FET (PGD-A group groups. versus both control groups, p<.0001)

291

First Author, Year Number of embryos Implantation Pregnancy rates Delivery or live-birth rates transferred rates *Ubaldi et al, 2015 No of embryos per Not reported Clinical pregnancy rate per Cumulative delivery rates per oocyte (con’t) transfer: PGD-A embryo transfer: PGD-A group: retrieval cycle: PGD-A group 24.4% group: 1.05 51.2% versus Control group: versus Control group 20.9%, NS. (180/172) per FET; 25.8%, p=0.01 Live-birth rate per embryo Control group: 2.03 transferred: PGD-A group 45.0% (fresh and FET) versus Control group 10.5%,p=0.01. (4410/2163)

Ubaldi et al, 2017 Single embryo Not reported Clinical pregnancy rate: Per Overall delivery rate: Per cycle (con’t) transfer in FET cycle cycle initiated: 8.7% (95%CI initiated: 8.0 % (95% CI 3.7-12.3); 4.7- 14.3); Per patient started: Per patient started: 8.8% (95%CI 9.5% (95%CI 5.2-15.7) and Per 4.1- 13.5) and Per FET cycle: 57.1% FET cycle: 61.9% (95%CI 41.1- (95%CI 35.9- 78.3). 82.7). Delivery rate per fresh cycle started stratified by age of women at OPU: Women aged 44-44.9 years: 10.6% (95%CI 4.7-16.5); Women aged 45- 45.9 years2.6% (95%CI 0-7.7) and Women aged 46.0- 46.9 years: 0%.

292

First Author, Miscarriage Multiple gestations Definition Author's Methodological Year Conclusion quality grading

Lee et al, 2015 Primary analysis: Primary analysis: PGD- The clinical pregnancy is PGD-A provides 11. (con’t) Pregnancy loss per CP: A group: 3.7% versus defined as presence of the advantage of Observational PGD-A group: 10.7% Control group: 26.2% intrauterine gestational identifying cohort design versus Control group: (p-value not reported). sac with fetal cardiac euploid embryo with small 38.1% (p-value not activity. Implantation rate for transfer sample size reported). Subanalysis: not is the number of which leads to a which may limit reported. intrauterine gestational higher generalizability. Subanalysis: not sacs/total number of implantation rate Results reported. embryo transferred. LB and live-delivery presented based (babies) per embryo rate. PGD-A on transfer transferred is the number supports single cycles rather of babies born /number of embryo transfer than cycles embryo transferred. and reduce the started may risk of multiple introduced gestations selection bias

293

First Author, Miscarriage Multiple gestations Definition Author's Conclusion Methodological Year quality grading

* Ubaldi et al, Miscarriage per CP: Multiple pregnancies Clinical pregnancy was Although the live-delivery 12. 2015 (con’t) PGD-A group: per delivery: PGD-A defined as the presence of rate per embryo transfer Observational 9.1% versus group: 1.2% versus a gestational sac at 7 and cumulative delivery study design Control group: Control group: 19.8%, weeks of gestation by rate per oocyte retrieval which may limit 30.3%, p=0.01. p<.01 ultrasound. Miscarriage cycle were comparable for generalisability. rate was referred as PGD-A and morphological number of pregnancy loss assessment of embryos per clinical pregnancy alone, PGD-A coupled achieved. The delivery with vitrified elective rate was calculated as the single embryo transfer number of deliveries per resulted in a lower oocyte retrieval or per miscarriage rate transfer.

294

First Author, Miscarriage Multiple Definition Author's Conclusion Methodological Year gestations quality grading

Ubaldi et al, Miscarriage per CP Not reported Clinical pregnancy was Female age and number of 11. 2017 (con’t) (< 20 weeks) defined as the presence of a MII retrieved were Observational overall: 7.7% gestational sac and fetal significantly associated with no control (95%CI 0-22.2) heartbeat at week 7 after with the likelihood of group. transfer. Miscarriage is achieving a live birth per defined as a pregnancy loss started cycle. Specifically, that occurs between weeks 7 female age was a negative and 20 of gestation. predictor and the number of MII oocytes retrieved was a positive predictor for achieving a live-birth per cycle started. The study found that no euploid blastocysts after the age of 45.9, suggesting that the quality of oocyte decreases after 45.9 years.

295

First Author, Title Study type Population (Selection Sample size Year criteria) Kushnir et al, 2016 Effectiveness of IVF with Retrospective The aggregate data were PGD-A group: a total of 5,471 preimplantation genetic cohort study obtained from the Centres fresh autologous cycles started. screening: a re-analysis of for Disease Control and Young women aged < 35 years: US ART data 2011-2012 included only women 1,518 fresh cycles started; undergoing fresh autologous Women aged 35-37 years: 1,166 IVF cycles with OPU fresh cycle started and women between 2011 and 2012. aged > 37 years: 2787 fresh Excludes cycles using donor cycles started. Control group: a eggs and cycles that were total of 97,069 fresh cycles cancelled before OPU. without reported use of PGD-A. Young women aged <35 years: 45,077 fresh cycles started; women aged 35-37 years : 20,177 fresh cycles started and women agd > 37 years: 31,815 cycles started.

296

First Author, Title Study type Population (Selection Sample size Year criteria)

Chang et al, 2015 Outcomes of in vitro Retrospective The aggregate data were PGD-A group: a total of 5,471 fertilization with cohort study obtained from the Centres for fresh autologous cycles. Young preimplantation genetic Disease Control and included women aged < 35 years: 1,518 diagnosis: an analysis of only fresh autologous cycles fresh cycles started; Women aged the United States Assisted with OPU of at least one 35-37 years: 1166 fresh cycle Reproductive Technology oocytes and blastocyst stage started and women aged > 37 Surveillance Data, 2011– embryo (5-6 days after years: 2,787 fresh cycles started. 2012 fertilisation) available for Control group: a total of 97,069 transfer between 2011 and fresh cycles without reported use 2012. Excludes cycles using of PGD-A. Young women aged donor eggs and cycles that <35 years : 45,077 fresh cycles were cancelled before OPU. started; women aged 35-37 years : 20,177 fresh cycles started and women agd > 37 years: 31,815 cycles started.

297

First Author, Year Population characteristics Types of Intervention Number of embryos transferred Kushnir et al, 2016 The data was stratified into 3 age- PGD-A group: aCGH for Not reported (con’t) groups: Young women aged <35 years; blastocyst transfer: Control aged 35-37 years and aged >37 years. group: Cycles without reported use of PGD-A. Biopsy type: Not reported.

Chang et al, 2015 The data was stratified into 3 age- PGD-A group: aCGH for None transferred: 36.7% in the (con’t) groups: Young women aged <35 years; blastocyst transfer: Control PGD-A group versus 37.9% in aged 35-37 years and aged >37 years. group: Cycles without reported the Control group; One embryo use of PGD-A. Biopsy type: Not transferred: PGD-A group reported. 27.8% versus Control group 12.9% in the; When two or more embryos transferred: PGD-A group:36.5% versus Control group: 49.2%. P value not reported

298

First Author, Implantation Pregnancy rates Delivery or live-birth rates Miscarriage Year rates Kushnir et al, Not reported Not reported Live birth rate per fresh cycle Miscarriage rate per pregnancy: 2016 (con’t) started: Young women aged Young women aged <35 years: <35 years: 33.7% in PGD-A 12.9% in the PGD-A group versus 39.8 in the Control versus 10.5%, p=0.06. Women group, p-value <.0001. Women aged 35-37 years: 10.7% in the aged 35-37 years: 32.2% in PGD-A versus 14.0% Control PGD-A group versus 29.5% in group, p=0.05 and women aged the Control group, p<.05) . > 37 years: 16.8% in the PGD-A Women aged > 37 years. versus 26.% in Control group, 17.7% in the PGD-A versus p<.001. 12.7% in the Control group, p<.001).

Chang et al, Not reported Compared to the Control Compared to the Control Compared to the Control group, 2015 (con’t) group, the adjusted odd ratio group, the aOR** of live the aOR** of spontaneous (aOR**) of CP per fresh ET delivery per ET in the PGD-A abortion per pregnancy in the for the PGD-group: Young group: Young women aged<35 PGD-A group: Young Women Women aged<35 years: years: aOR** 0.82 95CI 0.72– aged<35 years: aOR** 1.09 aOR** 0.82 95%CI 0.72 – 0.93); women aged 35–37 95%CI 0.84 – 1.40); Women 0.93); women aged 35 –37 years: aOR** 1.13 95%CI aged 35–37 years: aOR** 0.62 years: aOR** 1.01 95%CI 0.97– 1.31); Women aged >37 95%CI 0.45 – 0.87); Women 0.87– 1.17) and women years (aOR** 1.43, 95%CI aged >37 years (aOR** 0.55, aged >37 years (aOR** 1.26–1.62). 95%CI 0.43 – 0.70). 1.18, 95%CI 1.05–1.34).

299

First Author, Multiple gestations Definition Author's Conclusion Methodological Year quality grading Kushnir et al, Not reported Not Although miscarriage rate in women aged >37 12. Cohort study 2016 (con’t) reported years is lower after PGD-A, the author asserted design using that this is likely due to better patient selection of population-based women who are likely to reach ET (over- data. represented in the results). Older women who are likely to have poor-prognosis are likely to be diverted to embryo banking strategy rather than PGD-A. As there are fewer selection bias in younger women (< 35 years), the lower live- birth rate in the PGD-A group compared to non- PGD-A cycles is likely reflective of the true clinical value of PGD-A. PGD-A in fresh cycles was potentially ineffective and harmful to some patients. Chang et al, 2015 Multiple birth per live Not The lower rates of clinical pregnancy and live 16. Cohort study (con’t) birth:Young women aged reported birth after PGD-A compared to control group design using <35 years: aOR** 1.13 (non-PGD cycles) in women <35 years are likely population-based 95%CI 0.91– 1.40); due to complex infertility problem than the data. Adjusted for women aged 35–37 procedure itself. However, the higher live-birth confounding years: aOR**1.08 95%CI and lower miscarriage rate in older women variables. 0.83–1.42); Women aged following PGD-A suggest the beneficial use of >37 years (aOR** 1.98, PGD-A in older women. But, results presented 95%CI 1.52–2.52). using transfer cycles which may introduce selection bias.

300

First Author, Year Title Study type Population (Selection criteria)

Greco et al, 2014 Comparative genomic hybridization Cohort study Women who were aged < 36 years old and selection of blastocysts for repeated recruited between Mar 2012 and Mar 2013. implantation failure treatment: a pilot Excludes women with abnormal karyotype, study uterine abnormalities; autoimmune conditions, thrombophilia, sever endometriosis and reduced ovarian reserve. Rubio et al, 2017 In vitro fertilization with Multicentre non- Women recruited in the RCT if they were preimplantation genetic diagnosis for blinded RCT aged between 38 and 41 years, had normal aneuploidies in advanced maternal karyotypes, were on their first or second age: a randomized, controlled study ART cycles, BMI < 30 kg/m2, had five or more metaphase II (MII) oocytes obtained from one or two cycles and had sperm concentration ≥ 2 x 10/mL. Exclusion criteria include endocrine or systemic pathologies, a previous PGD-A or PGD cycle, and previous pregnancy or miscarriage due to chromosomal abnormities. Study recruitment was held between May 2012 and Dec 2014. Allocation was carried using computer generated randomisation.

301

First Author, Year Sample size Population characteristics Types of Intervention

Greco et al, 2014 Total sample: 121 patients. PGD-A PGD-A group: Mean age 32.8 ± PGD-A group and Control (con’t) group: 43 patients with repeated 3.1. Control group 1: mean age group 2: TE cells with array implantation failure. Control groups: 31.7 ± 2.9; Control group 2: mean CGH for fully developed Control group 1 (negative control): 33 age 31.7 ± 2.9. Women in both the blastocyst. All surplus patients with repeated implantation PGD-A group and Control group 1 blastocysts and those embryos failures and Control group 2 (positive has an average of 4.9 implantation reached blastocyst stage on control): 45 ART-naive patients without failure in their previous IVF day 6 or 7 are cryopreserved any detected problems of ovarian attempts. Control group 2: ART- using vitrification procedure. reserve and uterine receptivity or sperm naive patients without any detected Control group 1: quality (good prognosis patients) problems of ovarian reserve and Morphological assessment of undertaking PGD-A uterine receptivity or sperm quality. embryos alone

Rubio et al, 2017 PGD-A group: 100 patients started Mean age (years) ± SD: PGD-A PGD-A group: Blastomere (con’t) treatment using PGD-A with day 3 group: 39.1 ± 1.1 versus Control biopsy with array CGH for embryo biopsy and 24 chromosome group: 39.5 ± 1.0 fresh transfer and vitrification screening with aCGH; Control group: of suplus euploid embryos for 105 patients using standard ICSI cycle subsequent FET; Control with morphologic embryo selection at group: Morphological the blastocyst stage assessment of embryo alone

302

First Author, Year Number of embryos Implantation rates Pregnancy rates transferred Greco et al, 2014 No.of embryo per transfer: PGD-A group: 68.3% per Clinical pregnancy rate per ET: PGD-A (con’t) PGD-A group: Single embryo ET (fresh and FET); Control group: 68.3% per ET (fresh and FET transfer in fresh and FET group 1: 22.0% per fresh cycles) ; Control group 1 : 21.2% per fresh cycles; Control group 1: 1.24 embryo transfer; Control embryo transfer ; Control group 2: 70.5% embryo per fresh transfer cycle group 2: 70.5% per ET (fresh per ET (fresh and FET cycles). (no FET cycle) ; Control group and FET) (p<.001 for PGD-A (P-value <.001 between PGD-A group 2: Single embryo in fresh and group versus Control group versus Control group 1 or two control FET cycles 1). NS between PGD-A groups. group and Control group 2.

Rubio et al, 2017 No of embryos per fresh Implantation rate at the initial Clinical pregnancy rate per patient started: (con’t) transfer ± SD: PGD-A: 1.30 ± fresh cycle: PGD-A group: PGD-A group: 37% versus Control group: 0.5 versus Control group: 1.8 ± 52.8% versus Control group: 39%; NS. 0.4 27.6%; p<.001. Cumulative pregnancy rate per patient Implantation rate in started (i.e. pregnancy outcome from initial subsequent FET cycle: Not fresh and subsequent FET cycles): PGD-A reported group: 38% versus Control group: 55.2%, p=0.017

303

First Delivery or live-birth rates Miscarriage Multiple gestations Author, Year Greco et al, Authors reported that all patients with CP No spontaneous abortion in all Not reported 2014 (con’t) in any groups have delivered a healthy groups. Figures not reported. child (no figures presented).

Rubio et al, Live-birth rate per patient started: PGD-A Miscarriage rate per CP (fresh and Not reported 2017 (con’t) group: 44% versus Control group: 24.8%, FET cycles): PGD-A: 2.6% versus p=0.005. Live-birth rate per fresh transfer: Control group: 36.2%, p<.0001 PGD-A group: 64.7% versus Control group: 27.4%, p<.0001. Live-birth rate (FET cycles): Not reported. Cumulative live-birth rate per patient started (i.e live- birth outcome from initial fresh and subsequent FET cycles): PGD-A group: 45% versus Control group: 37.1%, NS.

304

First Author, Definition Author's Conclusion Methodological quality Year grading Greco, et al, Clinical pregnancies were defined as those PGD-A for single blastocyst transfer 11. Cohort Observational 2014 (con’t) showing the presence of an intrauterine provides an efficient tool to improve study design which may gestational sac determined by transvaginal implantation rates in women with limit generalisability ultrasound examination at 7 weeks of repeated implantation failure without gestation. increasing the number of transferred Implantation rate was defined as the embryos number of gestational sacs per transferred embryos Rubio et al, 2017 Implantation rate was defined as the Compared to standard morphological 18. RCT- The RCT (con’t) percentage of embryos transferred that assessment of embryos, PGD-A leads provides clearly defined produced an evolutive implanted embryo in a higher live-birth rate per patient eligibility criteria with up to week 12 of pregnancy. Clinical started and lower miscarriage rate per intervention using pregnancy (CP) rate was calculated clinical pregnancy after the first fresh computer generated according to the presence of a gestational cycle. randomisation. Women sac regardless of the number of embryos were randomised to either transferred. Delivery rates per transfer and However when successive cycles is standard ART with per patient were calculated as the considered, the CLBR for PGD-A is morphologic embryo percentage of clinical pregnancy (i.e. similar to morphological assessment of selection at the blastocyst presence of a gestational sac) that ended in embryos alone stage or PGD-A, with day- a live birth. The miscarriage rate was 3 embryo biopsy and 24- defined as the percentage of intrauterine chromosome screening by clinical pregnancies that were missed aCGH before blastocyst before the 12th week of pregnancy. The transfer. percentage of live births was calculated Study was NOT blinded considering all live births per patient. but sample size calculation was performed

305

Note: * Ubaldi et al (2015): The paper was intended to analyse the outcomes before and after the introduction of an elective single embryo transfer for older women in 2013. However, the authors have analysed the clinical results for PGD-A versus morphological assessment of embryos alone for the whole period between 2010 and 2013 (with one year of follow-up observations). ** adjusted for infertility diagnosis, prior spontaneous abortion, prior ART, number of oocytes retrieved, number of embryos transferred, and embryos cryopreserved CP: Clinical Pregnancy; FET: Frozen embryo transfer; BMI: Body Mass Index; aCGH: array comparative genome hybridization TE biopsy: Trophectoderm biopsy; qPCR: quantitative polymerase chain reaction; NS: Not significant; SD: Standard deviation

306

Supplementary Table 2- Demographic and outcomes according to per-protocol analysis

Demographic characteristic and outcomes of first three ‘single cycles’ (fresh or frozen/thaw cycles) according to per-protocol analysis 1 Morphology PGD-A assessment

Total number of women at start of N= 110 N= 1983 first fresh cycle 2, N p-value

Age at first oocyte pick-up (OPU), mean ± SD, range 40.06 ± 1.97, 37- 45 40.05 ± 2.34, 37-49 p=0.96 ‘Single cycles’3 initiated n, mean per woman ± SD, 188,1.71 ± 0.75 4559,2.30 ± 0.83 p<0.001

Fresh cycle initiated 4, n (% of total ‘single cycles’ initiated) 170 (90.42) 3,280 (71.95) p<0.001

Frozen/thaw cycle initiated 4, n (% of total ‘single cycles’ initiated) 18 (9.58) 1,279 (28.05)

OPU cycles, n, mean per woman ± SD 170, 1.54 ± 0.68 3162, 1.60 ± 0.72 p=0.48

Oocytes collected, mean per OPU ± SD 11.54, ± 6.61 6.88 ± 5.67 p<0.001

Fertilisation rate (two-pronuclei/oocytes collected), % 62.91 56.62 p<0.001

Embryos created per OPU, mean ± SD 7.26 ± 4.55 3.89 ± 3.75 p<0.001

Embryos numbers and quality

Day 2-4 embryos, mean grade ± SD, (1=good, 5=bad) 0 4667, 2.02 ± 0.83 NA

Day 5 & 6 embryos, mean grade ± SD, (1=good, 8=bad) 3.48 ± 1.56 3.43 ± 1.31 p=0.55

Embryos available for transfer or cryopreservation, mean per cycle with OPU ± SD 0.77 ± 1.11 2.60 ±2.65 p<0.001

Embryo transfer procedures

Embryo transfer (ET) procedures, n (% per cycle started) 91 (48.40) 3615 (79.30) p<0.001

Embryos transferred, mean per ET procedure ± SD, range 1.06 ± 0.25, 1-2 1.36 ± 0.48, 1-3 p<0.001

307

Morphology Total number of women at start of first PGD-A assessment fresh cycle ‡, (con’t) n= 110 n= 1983 p-value

Implantation rates, % (gestational sacs/total embryos transferred), % 43.30 13.70 p<0.001 Pregnancy rate 5 Per cycle initiated, % 21.27 14.03 p<0.01

Per fresh ET procedure, % 42.11 17.78 p<0.001

Per frozen/thaw ET procedure, % 53.33 17.54 p<0.001 Percentage of pregnancies failed to reach delivery 6 % 20.00 35.00 p=0.05

Live-birth7, rate Per cycle initiated, % 16.49 9.10 p<.01

Per ET procedure, % 34.44 11.54 p<0.001

Mulitple live-birth rate, (% of live-birth delivery) 6.46 6.75 P=0.96

Total number of SET on day 5/6, n (% of total number of single embryo transferred) 85 (100) 146 (6.31)

Proportion of live births per single embryo transferred on day 5/6 (%) 35.29 10.96 p<.001

Discontinue rate8, % (% of women without a live born) 89.87 63.45 p<.001 1 Per-protocol analysis censors women from the analysis if they cross over to the alternate treatment strategy for subsequent fresh cycle during the study period 2 The cohort was stratified into women whose first OPU used either PGD-A or morphological assessment for selection and transfer of embryos 3 ‘Single cycle’refers to the outcomes of a discrete fresh or frozen/thaw cycle 4 An initiated fresh cycle is defined as commencing with the administration of Follicle Stimulating Hormone (FSH) and initiation of a ‘frozen/thaw cycle’ was defined as the thawing of one or more embryos. 5 Pregnancy rate is defined as the presence of a heart beat per intiaited cycle or ET procedure 6 Pregnancy which failed to reach delivery includes miscarriages, missed abortions, blighted pregnancies, ectopic pregnancies and terminations 7 A live birth is defined as the birth of at least one live-born baby, with twins and triplets counted as one live birth 8 Discontinue rate refers to women who discontinue with treatment without a live birth in first 3 ‘single cycles’. P-values lower than 0.05 were considered statistically significant

308

Supplementary Table 3: Clinical outcomes for up to three ‘complete cycles’

Clinical outcomes for up to three ‘complete ART cycles’ 1 based on per-protocol analysis2 Morphology PGD-A Total number of women at the start assessment p-value n= 110 of first fresh cycle n= 1983

Cumulative live-birth rate (CLBR), 3 % 30.00 26.17 p=0.37

Proportion of women who did not have an embryo transfer (ET) procedure in their first fresh cycle, (% of first fresh cycle initiated) 58.18 20.32 p <0.001

CLBR of women with an ET procedure in first fresh cycle, % 50.00 30.31 p<0.01

Crossed-over rate4, (% of women without a live born) 29.26 5.07 p<0.001 Discontinue rate 5, % 52.44 54.22 p=0.75

Follow-up time6 mean ± SD (days) 505.76 ± 333.90 562.39± 338.75 p=0.08 1 A ‘complete cycle’ refers to the outcomes from all embryos created from an oocyte pick-up (OPU), including fresh and any subsequent frozen/thaw embryo transfer (FET) cycles. 2 Per-protocol analysis censors women from the analysis if they crossed over to the alternate treatment strategy in their subsequent fresh cycle during the study period. 3 Cumulative live-birth rate (CLBR) is defined as the number of live births divided by the total number of women started treatment in either the PGD-A group or the morphology assessment group (i.e. 110 women in PGD-A and 1983 women in the morphology assessment group). 4 Percentage of women who crossed over to the alternate treatment during study period. 5 Discontinue rate refers to women who did not undertake up to three ‘complete cycles’ during the study period. 6 Mean follow-up time is based on the date of first OPU through to 31 Mar 2014 or to the first clinical pregnancy leading to a live born baby. A p-value of less than 0.05 is considered statistically significant.

309