SEARCH

SEARCH BY CITATION

Keywords:

  • Assisted reproductive techniques;
  • infant mortality;
  • perinatal mortality

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

The increased risk of perinatal and infant mortality observed among in vitro fertilisation (IVF) births and other assisted conception births is thought to be largely attributable to multiplicity. Using mortality statistics and estimates of the proportion of births following infertility treatment, we predicted the excess stillbirths and infant deaths associated with twins and higher order births resulting from assisted conception in England and Wales. According to our results, approximately 73 deaths could have been avoided in 2001 if all IVF infants had been born as singletons or as naturally occurring monozygotic twins, equating to a population attributable risk fraction of around 1% for perinatal and infant deaths. If we include all types of assisted conception, this figure was estimated to be around 4% of deaths—more than 220 perinatal and infant deaths in 2001. We confirm the public health importance of multiple births associated with assisted conception.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

Increased risks of perinatal and infant death have been reported among cohorts of infants born as a result of in vitro fertilisation (IVF). This observed association is thought to be largely attributable to multiple birth, with approximately 35–50% of all infants resulting from IVF in Europe born as a twin, triplet or higher order multiple.1 Little is known about the health of infants resulting from other forms of infertility treatment, but a high multiple birth rate has also been observed among births following ovarian stimulation.2 Extensive literature describes the association between multiple birth and adverse perinatal outcomes such as prematurity and lower birthweight,3 and attempts to reduce the proportion of multiple births is likely to result in improved outcomes for such children.

In the UK, changes in the code of practice issued by the regulatory Human Fertilisation and Embryology Authority (HFEA) have limited the number of embryos that can be transferred during IVF procedures. When using their own eggs or embryos, currently, a maximum of two embryos can be transferred in women younger than 40 years and a maximum of three for women aged 40 or more. Although recent data indicate a decrease in the triplet rate,4 the twinning rate has shown remarkable stability over time. Concern has predominantly been expressed about the impact of IVF-associated triplets and higher order births; however, the twinning rate continues to constitute the major burden of IVF-associated multiple birth and associated sequalae.5

In the UK, data from the HFEA suggest that approximately 7400 births resulted from assisted conception in 2001, 1.3% of all births.1 The HFEA requires information from treating clinics on all IVF (including intracytoplasmic sperm insemination cycles) and their outcomes but not for treatments that exclude manipulation of the gametes. One such treatment is ovarian stimulation (without IVF), and at present, there is no legal requirement to monitor these cycles or to record the outcome of resulting pregnancies. Thus, 1.3% of births is a conservative estimate of the proportion of births in the UK resulting from some form of assisted conception.

Our aim was to predict the contribution of multiple births following assisted conception (IVF and other methods) to overall perinatal and infant mortality in England and Wales. In particular, we aimed to estimate how many deaths could be avoided if all offspring from such conceptions were born as singletons or as spontaneously occurring monozygotic twins.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

We used 2001 statistics (England and Wales) as our source data, capturing the number of total births, together with linked perinatal and infant mortality rates for single and multiple births. All calculations were performed using the infant as the unit of analysis. A major assumption was that—other than risks associated with multiplicity—there was no extra mortality risk for offspring born following IVF or other forms of assisted conception compared with naturally conceived offspring.

We modelled a range of scenarios where IVF-conceived infants or infants conceived following other forms of assisted conception comprised a varying proportion of all births: 1–5%.1 The risk for such infants being part of a multiple birth was estimated to range between 30 and 60%. We did not include quadruplets and other higher order births in the model due to the paucity of data.

Our main outcome was the predicted number of stillbirths and infant deaths that could be avoided (or ‘saved’) if all infants resulting from IVF and other forms of assisted conception were born as singletons or monozygotic twins.

We calculated the number of births and deaths that would be expected in the assisted conception offspring, assuming a higher risk of multiple birth than in the general population of births. We controlled for the effect of maternal age on perinatal and infant mortality, using previously published data to estimate the distribution of maternal age among IVF infants. We then calculated the number of stillbirths and deaths under 1 year that would be expected assuming all IVF infants or infants conceived following other forms of assisted conception were born as singletons (99.6%) or as monozygotic twins (0.4%).

To allow for the increased risk of adverse outcome associated with monozygotic twins relative to dizygotic twins, we estimated a doubling of the Office for National Statistics (ONS) twin perinatal mortality rate in such cases. The difference between these two sets of expected deaths were taken to represent the maximum number of deaths that could be saved. These numbers were then used to estimate the proportion of all stillbirths and deaths under 1 year that could be avoided if IVF offspring or offspring from other forms of assisted conception were born as singletons or as naturally occurring monozygotic twins: this figure being the population attributable risk fraction (PAF).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

ONS data for 2001 (England and Wales) reported 594 634 live births, 3159 stillbirths and 3240 deaths under 1 year of life.

The predicted number of ‘savable’ deaths (stillbirths and deaths up to 1 year) according to the prevalence of births following assisted conception and risk of multiple birth associated with assisted conception are shown in Figure 1. At the lowest level (1% of infants resulting from assisted conception and 30% risk of multiple birth), we estimate there to be 48 avoidable deaths (16 stillbirths and 32 infant deaths). These figures represent the most conservative estimate of the proportion of births in the UK, attributable to IVF alone.1

image

Figure 1. ‘Savable’deaths (stillbirths and deaths under 1 year) according to the proportion of infants in the population born as a result of assisted conception and the proportion of these infants born as part of a multiple birth. Based on 2001 data (England and Wales). AC, assisted conception.

Download figure to PowerPoint

A scenario where births resulting from all forms of assisted conception comprise 5% of births, and where 60% of resulting babies are a twin or a higher order birth, could result in 491 avoidable deaths (167 stillbirths and 324 infant deaths).The majority of savable deaths are concentrated in the perinatal period (data not shown by infant age). These are deaths that could have been prevented if offspring resulting from assisted conception were born as singletons or spontaneously occurring monozygotic twins in the population setting of England and Wales.

At the most conservative estimate of 1% of all births resulting from assisted conception and a 30% risk of multiple birth for resulting offspring, the PAF for perinatal death is 0.80% and for infant deaths 1.03% (Table 1). These deaths can be attributed to IVF conception through the effect of multiplicity alone. In a context where assisted conception births approach 1 in 25 of all births (such as in Denmark),1 assisted conception could be associated with between 3 and 6% of all perinatal deaths and between 4 and 8% of all infant deaths. If 3% of all births in England and Wales resulted from some form of assisted conception, this model predicts that between 2 and 5% of all perinatal deaths plus between 3 and 6% of all infant deaths could have been avoided if these babies had all been born as singletons or, rarely, as monozygotic twins.

Table 1.  Predicting PAFs for perinatal deaths and deaths under 1 year, associated with assisted conception
Proportion of multiple births among assisted conception infants (%)Proportion of infants in the population resulting from assisted conception (%)
1.02.03.04.05.0
Perinatal deaths PAF (%)Infant deaths PAF (%)Perinatal deaths PAF (%)Infant deaths PAF (%)Perinatal deaths PAF (%)Infant deaths PAF (%)Perinatal deaths PAF (%)Infant deaths PAF (%)Perinatal deaths PAF (%)Infant deaths PAF (%)
  1. The percentages illustrate the proportion of deaths that could be saved if all assisted conception infants were born as singletons or as spontaneously occurring monozygotic twins.

300.81.031.592.062.393.093.184.123.985.15
350.931.21.862.412.83.613.734.814.666.01
401.071.382.142.753.214.134.285.515.356.88
451.211.552.413.13.624.654.826.26.037.75
501.341.722.683.454.035.175.376.896.718.62
551.481.92.963.794.445.695.927.597.399.49
601.622.073.234.144.856.216.468.288.0810.36

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

This is the first known attempt to quantify the excess burden of mortality associated with multiplicity following assisted conception. Current estimates suggest that in England and Wales, more than 1% of all infants are born as a result of IVF, and the multiple birth risk with IVF is approximately 45%.1 Our model predicts that approximately 73 deaths (stillbirths and deaths under 1 year) in England and Wales could be avoided if all IVF offspring were born as singletons or spontaneously occurring monozygotic twins. This includes 57 deaths during the perinatal period. These figures equate to an overall PAF for perinatal and infant deaths of 1.2%.

Focussing on IVF only will underestimate the contribution of all types of infertility treatment to perinatal and infant mortality. Ovarian stimulation is an independent risk factor for multiple birth, with research suggesting that these treatments may be associated with around 30% of all multiple births.2 In the UK, the use of ovarian stimulation treatments is not subject to regulation, and treatment may take place outside the scope of HFEA registered clinics. Inadequate data make it difficult to accurately assess the overall use of such techniques, but we can make an informed estimate that the true proportion of babies in the UK resulting from any form of assisted conception is likely to be near 3%. In that event, a much higher proportion of avoidable perinatal and infant deaths seems likely, around 3.5% or more than 220 lives each year in the UK.

Our assumption that both the infants resulting from assisted conception and their naturally conceived peers have the same background risk of mortality may have led to an overestimation of effect. There is some evidence that IVF singletons have a higher risk of perinatal mortality compared with maternal-age- and parity-matched babies,6 although little is known about outcomes associated with other forms of assisted conception. There is also evidence that infertility itself has an independent outcome on perinatal outcome.7 Applying the all-twin-mortality rate to IVF births may also lead to an overestimation of effect since IVF multiples relative to the general population are likely to include a proportionately smaller number of monochorionic twins for whom risks of perinatal mortality are increased. Nonetheless, our conclusions remain robust, and reducing the proportion of multiple births in conceptions following treatment for infertility will reduce the number of perinatal and infant deaths in the UK.

The much increased risk of multiple birth in babies conceived following infertility treatment continues to constitute a public health burden, which is largely preventable. As current fertility trends suggest that the number of people seeking assisted conception is likely to continue to increase, this issue needs to be addressed with some urgency. Reducing the multiple birth rate with IVF can be achieved by restricting the number of embryos that can be transferred in utero. However, the practice of restricting the number of embryos to two (as in the UK) is unlikely to reduce the twinning rate to an acceptable level in the long term and instead the focus is now on the use of single embryo transfer (SET). Sweden was one of the first countries to systematically move towards SET, and their experience indicates that a SET policy does not necessarily compromise the ‘take home baby’ rate but does dramatically reduce the multiple birth rate.8 The evidence suggests that for the majority of women with a good IVF prognosis, SET may be an effective strategy. Reduction in the multiple birth rate associated with other forms of assisted conception—such as ovarian stimulation—is achievable through rigorous clinical monitoring.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

Multiple births resulting from IVF and other methods of assisted conception are likely to be associated with a significant proportion of stillbirths and infant deaths in England and Wales.

A move towards SET, together with better population-based monitoring data for outcomes and unregulated forms of assisted conception, are high public health priorities for the UK.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References

L.O. is funded by a Health Services Research Fellowship, awarded by the NHS Eastern Region.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgement
  9. References