Perinatal mortality following assisted reproductive technology treatment in Australia and New Zealand, a public health approach for international reporting of perinatal mortality

There is international agreement of the importance of perinatal mortality measures for monitoring pregnancy and childbirth with routine reporting at a country and regional level [1]. This monitoring does not commonly extend to births following assisted reproductive technology (ART) treatment with only Australia and New Zealand, Latin America, Japan and Canada including perinatal mortality measures in their most recent ART reports [2‐5]. In contrast, the ART reports from the United States and Europe present data on the births of almost 150,000 babies but no information on mortality [6, 7]. The lack of perinatal mortality data is a major health information gap in benchmarking the quality of ART policy and the safety of practice.

Internationally, monitoring and comparison of perinatal mortality is limited by the wide variation in the definitions of birth, stillbirth and perinatal death used in countries that practice ART (Table 1) [3, 4, 8, 9]. The World Health Organization (WHO) recommends for international comparison that the definition of births be restricted to 1000 grams and/or 28 weeks gestation for all perinatal measures [10] but the usefulness of this definition for monitoring perinatal mortality is limited for births following ART where there are proportionately high rates of multiple birth and preterm birth [11]. In 2009, the International Committee Monitoring Assisted Reproductive Technologies (ICMART) and the WHO sought to limit the perinatal period for birth following ART treatment to ≥20 completed weeks of gestational age to <7 completed days after birth regardless of birth weight [12]. However, for many countries birth data are not available for gestations of <22 weeks limiting the utility of international comparison based on the ICMART/WHO definition of birth.

Further, when agreeing on the perinatal measures to monitor ART, the impact of ART treatment practices needs to be considered, most notably the number of embryos transferred. The literature show increased risk of multiple gestation with the transfers of two or more embryos [13]. For example in 2010, in Quebec, Canada, the change in policy to elective single embryo transfer (SET) (from 1.6% of cycles to 50%) resulted in a reduction in the multiple pregnancy rate from 25.6% in 2009 to 3.7% in the first three months of implementation of a policy of SET [14].

Reducing the number of embryos per transfer is essential to lower both perinatal morbidity and mortality for ART births [13, 15]. Among the risk factors for perinatal mortality, iatrogenic multiple gestation pregnancy can be considered the most significant preventable risk for women undergoing ART treatment. Research studies and subsequent policy implementation of SET in New Zealand, Sweden, Belgium and Quebec have reduced the proportion of double embryo transfers (DET) and multiple gestations but have not routinely reported embryo-transfer specific perinatal mortality [13, 16‐18]. This lack of reporting of embryo-transfer specific rates of perinatal mortality limits evaluation of the policy of SET [19]. The aim of this study is to develop an approach for standardization of reporting of PMR following ART treatment for international comparison. We applied commonly used American and European perinatal periods to the Australian and New Zealand ART registry data, to investigate the changes in PMRs stratified by plurality and number of embryos transferred.

A review of reporting perinatal mortality of ART births was conducted. The 2004-2008 data from the Australia and New Zealand Assisted Reproductive Database (ANZARD) were used to assess the differences in the definitions of perinatal periods based on internationally used gestational age at birth and number of days after birth. ANZARD is a census of all initiated ART treatment cycles undertaken in Australia and New Zealand. Items on ANZARD are collated annually by the National Perinatal Epidemiology and Statistics Unit (NPESU), in a de-identified format, from all fertility centres within Australia and New Zealand. Fertility centres as part of their quality assurance are required to report ART treatment and outcomes to ANZARD. Birth outcomes collected in ANZARD include birth status, gestational age, birth weight, date of neonatal death and congenital anomalies. The fertility clinic data that makes up ANZARD is audited for completeness and accuracy on an annual basis as part of accreditation for fertility clinics through the Reproductive Technology Accreditation Committee.

This study shows that the wide variation in the measures used for reporting perinatal mortality at a country and regional level limits direct comparison and evaluation of ART policy and best practice internationally. Monitoring of perinatal mortality is an essential component of any quality system that seeks to improve ART services and the related maternal and perinatal outcomes. Unlike natural conception, ART treatment involves elective decision making on the part of the clinician and couple, most notably about the number of embryos transferred. The number of embryos transferred is highly correlated with the risk of multiple gestation pregnancies, which remain prevalent in most countries [13, 14]. Therefore, it is essential that routine perinatal mortality measures following ART are able to monitor the impact of embryo transfer practices among countries.

This study demonstrates the impact different definitions of perinatal mortality have on PMRs. The PMR for Australia and New Zealand was 16.1 per 1000 births (Australian definition ≥ 20 weeks gestation to <28 days post-birth) compared to 5.1 per 1000 births, when the more limited WHO recommended definition for international comparison (≥ 28 weeks gestation to <7 days post-birth) was applied [10, 23]. Application of the WHO definition results in the exclusion of 1,040 (9.5%) preterm births, 402 (64.3%) stillbirths and 143 (78.6%) neonatal deaths. These differences can be overcome by presenting data for three gestational time periods (≥ 20 weeks, ≥ 22 weeks, ≥ 28 weeks and limiting the perinatal period to 7 days post birth) where the majority of neonatal deaths occur. This would allow international comparison. The first of these proposed definitions would be consistent with ICMART / WHO glossary definition of a perinatal death and could be adopted as a standard for international reporting [12]. The inclusion of ≥20 completed weeks’ gestation would allow greater utility to investigate the impact of ART and multiple gestation from the provision of neonatal services.

Most countries follow the WHO definition (gestational age of 22 weeks or birth weight of 500 grams and <7 days) for national reporting of perinatal mortality. However, some countries only use gestational age and others birth weight [30]. Birth weight is included in the definition of perinatal mortality, particularly in low income countries, as it is easier to measure and less prone to error than calculations of gestational age from the last menstrual period [32]. In contrast, Mohangoo et al. suggested that mortality rates are usually underestimated if the case definition includes birth weight rather than gestational age. This was demonstrated in the Euro-Peristat project which reported comparatively lower fetal death rates when birth weight criteria was used rather than gestational age [33]. Therefore, most of the high income countries prefer to use gestational age, given it is more sensitive predictor for the perinatal outcome [20]. In addition, gestational age of births following ART treatment are calculated based on date of embryo transfer and date of delivery which are usually collected in ART registers [34, 35]. Regardless of which criteria are used, the variation in birth registration practices and in scope of perinatal periods significantly changes the international rankings of countries based on mortality rates [36]. Standardization in perinatal data collection and reporting will help in monitoring ART treatment practices and is consistent with the recommendations from the Global report on preterm birth and stillbirth [20].

The clinical relevance of benchmarking of ART practice with the reporting of embryo-specific- and plurality-specific-rates of perinatal mortality is evident. The transfer of two or more embryos significantly increases the risk of multiple gestation pregnancy [13, 37].Multiple gestation pregnancies are at higher risk of fetal, neonatal and perinatal mortality than singleton pregnancies [15]. In this study, SET was associated with a significantly lower PMR for all births when compared to DET. Double embryo transfers were associated with one extra death per 200 births. In contrast, twin gestations following SET resulted in a significantly higher PMR compared to DET.

Further, there is wide variation in the number of embryos transferred by countries and regions and the policy framework for SET [19]. The international comparison of perinatal mortality is complicated by the variability in the scope and quality of death data between countries. This includes differing registration criteria and denominator populations and variability in data collection practices. The higher PMR for Australia and New Zealand compared to Sweden and Finland [27, 38] is likely related to the earlier adoption and higher proportions of SET in Sweden and Finland compared to Australia and New Zealand. In contrast, the lower PMR for Australia and New Zealand when compared to Canada, the United Kingdom and Latin America likely reflects the higher proportion of SET in Australia and New Zealand than those countries and regions [3, 4, 31]. Additionally, studies have reported better perinatal outcomes of singletons following SET when compared to singletons following DET [39, 40]. It has also been suggested that twins following SET (considering monozygotic twins) had a higher risk of perinatal death than twins following DET (considering dizygotic twins) [41, 42]. Therefore, stratification by number of embryos transferred and plurality would allow making direct assessment of the risk of perinatal mortality due to DET compared to SET, multiples compared to singletons and presumptive monozygotic twins compared to dizygotic twins.

This study used five years of national level Australian and New Zealand data with approximately 10,000 births per year to minimize the effects of random variation and to maximize the stability and reliability of the mortality rates. There are some limitations of this study. First, with all registry studies there is the potential for information biases associated with differential ascertainment of fetal and neonatal deaths from births. However, a recent abstract reported a similar PMR for ART births to ANZARD using the national perinatal data collection [43]. Alternatively, if the quality or completeness of perinatal mortality data in ART registries is limited, data linkage among ART, birth and perinatal death registers is an option. This has the potential to enhance the validity of the death data internationally as well as providing information on cause of death [44]. Secondly, we did not stratify for other patient and treatment related factors even though published research has shown that ART is generally associated with more maternal and fetal complications, which vary according to the type of procedure and type of treatment [45‐47]. Instead, we only included the most critical factors for perinatal mortality, namely the number of embryos transferred and plurality to present a public health approach for reporting perinatal mortality internationally.

International benchmarking of perinatal mortality of ART births provides essential information to improve the safety and quality of ART practice. Due to the persisting high rates of multiple embryo transfers and of multiple birth internationally best practice is to report all mortality measures by number of embryo transfers (SET, DET and all transfers) and plurality (singletons, twins and all). The ICMART / WHO recommended a perinatal period of ≥ 20 completed weeks of gestational age to < 7 completed days after birth which would optimize the standardization of reporting fetal and neonatal deaths following ART treatment. Noting, this is currently not possible in Europe and many other countries. Reporting PMR specifically for three perinatal periods: ≥ 20 weeks gestational age to < 7 days after birth; ≥22 weeks gestational age to <7 days after birth and ≥ 28 weeks gestational age to <7 days after birth would allow international benchmarking of mortality of ART births. This would rapidly provide an international evidence-base to evaluate ART practice and policy against perinatal mortality.

EA Sullivan, MBBS, MPH, MD. Professor and the director of National Perinatal Epidemiology and Statistics Unit, the University of New South Wales.

YA Wang, BMed, MPH, PhD. Biostatistician and Senior Lecturer of National Perinatal Epidemiology and Statistics Unit, the University of New South Wales.

RJ Norman, BSc (Hons), MB ChB (Hons), MD. Professor and the director of Robinson Institute, School of Paediatrics and Reproductive Health, University of Adelaide.

GM Chambers, BAppSci(MLS), MBA, PhD. Senior Lecturer of National Perinatal Epidemiology and Statistics Unit, the University of New South Wales.

AA Chughtai, (MBBS, MPH) Research officer of National Perinatal Epidemiology and Statistics Unit, the University of New South Wales.

CM Farquhar, MBBS, PhD. Professor Department of Obstetrics and Gynaecology and National Women’s Health, University of Auckland.