Hum. Reprod. Advance Access originally published online on May 20, 2009
Human Reproduction 2009 24(9):2321-2331; doi:10.1093/humrep/dep173
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Twins born following assisted reproductive technology: perinatal outcome and admission to hospital
1 Centre for Child Health Research, Telethon Institute for Child Health Research, The University of Western Australia, West Perth, WA 6872, Australia 2 The National Perinatal Epidemiology Unit, University of Oxford, Oxford OX3 7LF, United Kingdom
3 Correspondence address. Tel: +61-8-9489-7771; Fax: +61-8-9489-7700; E-mail: michele{at}ichr.uwa.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
BACKGROUND: Compared with spontaneously conceived (SC) singletons, adverse perinatal outcome, neonatal intensive care unit (NICU) admission and hospital admission in infancy are more common in those born following Assisted Reproductive Technology (ART). Similar comparisons for twins have shown conflicting results.
METHODS: We investigated perinatal outcome and hospital admission during the first 3 years of life for all twin children born in Western Australia between 1994 and 2000 [700 ART, 4097 SC].
RESULTS: ART twins had a greater risk of adverse perinatal outcome including preterm birth, low birthweight and death compared with SC twins of unlike-sex. In their first year of life, ART twins had a longer birth admission; were 60% more likely to be admitted to a NICU; and had a higher risk of hospital admission. The increased risk of hospital admission continued in the second and third year but was not statistically significant in the third year.
CONCLUSIONS: Couples undertaking ART should be aware that in addition to the known increased perinatal risks associated with a twin birth, ART twins are more likely than SC twins to be admitted to a NICU and hospitalized in the first 3 years of life.
Key words: assisted reproductive technology/hospital admission/IVF/morbidity/twins
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
Between 20 and 30% of deliveries following assisted reproductive technology (ART) in the USA and Europe are twins, compared with approximately 1% following spontaneous conception (Centers for Disease Control and Prevention et al., 2007
; Andersen et al., 2008). In Australia, the most recently published data indicate that the proportion of twin deliveries dropped to 11.7% in 2006, the lowest ever reported (Wang et al., 2008). The well established risks of perinatal and infant mortality and morbidity associated with multiple births (Luke and Keith, 1992) account for a major portion of the poorer perinatal outcomes in ART compared with spontaneously conceived (SC) infants. The use of single embryo transfers (SET) dramatically reduces the multiple birth rate and contingent poor perinatal outcomes following ART and this is a major factor to be considered by parents, clinicians and those determining health policies. However, a further consideration is the health status of ART compared with SC twins both in the perinatal period and the longer term. There is ample evidence that ART singletons are more likely to be born preterm, with low birthweight, and have a major malformation, than their spontaneously conceived counterparts (Helmerhorst et al., 2004; Jackson et al., 2004; Rimm et al., 2004; Hansen et al., 2005; McDonald et al., 2005b), and accumulating evidence suggests that there is also a small increase in their risk of hospital admission during infancy (Ericson et al., 2002; Koivurova et al., 2003; Kallen et al., 2005a; Klemetti et al., 2006; Koivurova et al., 2007; Hansen et al., 2008a, b). Similar comparisons for twins in both the perinatal period (Helmerhorst et al., 2004; Pinborg et al., 2004b; McDonald et al., 2005a; Verstraelen et al., 2005; Boulet et al., 2008) and childhood (Ericson et al., 2002; Pinborg et al., 2004a; Kallen et al., 2005a; Klemetti et al., 2006; Koivurova et al., 2007) show conflicting results but in general little difference between ART and SC infants. Interpretation of these results is complicated by the very different proportions of monozygosity and hence monochorionic placentation in SC and ART twins. A shared chorion, present in about 2/3 of monozygous twins, is associated with increased risks of mortality and morbidity (Derom and Derom, 2005b). In Caucasian populations approximately 22% of SC twins are monochorionic. Although most ART twins are dyzygotic, being the result of transfer and subsequent implantation of more than one embryo, there is an increase in embryo splitting compared with SC conceptions. Studies of chorionicity in IVF twins show that about 2% are monochorionic (Derom and Derom, 2005a).
The aim of this study was to investigate the rate of hospital admission by 3 years of age in twins born following ART (IVF, ICSI or gamete intra-fallopian transfer (GIFT)) compared with their SC counterparts in Western Australia. In order to take account of the differing proportions of monochorionic placentation in the two groups and in the absence of zygosity and chorionicity data, comparisons were made between ART twins and unlike-sex (ULS) spontaneously conceived twins, although comparisons with the total population of SC twins are shown for perinatal outcomes in order to further justify our choice of comparison group. A study of hospital admissions in Western Australian ART and SC singleton births has been published previously (Hansen et al., 2008a, b).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
This cohort study was conducted using record linkage methods to identify hospital admissions up to the age of 3 years for children born as twins in Western Australia between October 1993 and October 2000 following either ART or spontaneous conception. Data relating to Aboriginal children (who account for 6% of all births in Western Australia) were excluded throughout, since Aboriginal women are much less likely to receive ART treatment than other women in the population and their children have high levels of hospitalization at all ages (Hansen et al., 2002
; Carville et al., 2007). Details of the creation of the linked dataset used in this study have been previously described (Hansen et al., 2008a, b).
Cohort analysis
We analyzed hospital admissions up to 3 years of age. Infants born in the later years had not yet reached the age of 3 years when the datasets were linked, so, to allow for the differential length of follow-up, we initially attempted to analyze the data using Cox proportional hazards modeling. However, the assumption of proportional hazards was violated. Since a greater proportion of ART twins than SC twins were born in the later years of our study (as ART use is increasing over time), and therefore proportionately fewer of the ART group contribute 3 years of follow-up, we proceeded by dividing the duration of follow-up into three 1 year (1st, 2nd and 3rd) age periods in which the children included in the analysis for each period were those children at risk of being admitted for the whole of that age period. Only children alive at the start of an age period were included in the denominator for that period. For these reasons the denominators change between periods. Where an admission spanned two time periods the admission was allocated to the first period to avoid double counting. The length of stay (LOS) was calculated as the number of days from admission to discharge or death. If a child was admitted and discharged on the same day then a LOS of 0.5 days was assigned. Hospital transfers were merged into a single admission and we excluded all admissions for social reasons (ICD9 Chapter 18, V Codes) since our primary interest was to assess the health of the children.
Admissions to a neonatal intensive care unit (NICU) were analyzed separately from other first year admissions and the denominator for this analysis was all live births. All births in hospital, which accounts for 99% of all births, are counted as an admission in the Western Australian hospitalization data. Thus for the purposes of calculating admissions in the first year, we needed to separate ‘normal’ birth admissions from admissions in which the neonate was ‘sick’. Although we were able to identify admissions to NICUs, not all sick neonates are admitted to a NICU. In our analysis of singleton births we defined a birth admission of 10 days or longer as a ‘morbid’ admission; 95.7% of all singleton birth admissions were 9 days or less in duration. This definition is inappropriate for twin births since the median LOS of ART and SC twin births in this study were 12 and 9 days, respectively. As 90.3% of all twin birth admissions were 28 days or less in duration, admissions in the first year were defined as children still in hospital on day 29 or admitted after this day up to the child's first birthday. The denominator for this period was all twin children alive at age 29 days that were born on or before 31st December 1999, that is children with follow-up information available to their first birthday.
Analytical methods
The data analysis was carried out using SPSS (version 15.0) and SAS/STAT (version 9.1). Means were derived for comparison, and medians were used for variables that were not normally distributed. Student's and Welch's t-test were used to compare means, and non-parametric tests were used to compare medians. Generalized Estimating Equation (GEE) Analysis assuming an exchangeable correlation structure was used to derive odds ratios (ORs) and their 95% confidence intervals (95% CI) to compare the relative odds of birth outcomes and hospital admissions in ART children compared with SC children while adjusting for correlations within sibships. The effects of the following covariates on the OR estimates were examined: maternal age, parity, sex, child's year of birth, preterm birth, low birthweight, the presence of major birth defects, admission to a NICU (in analyses of later hospital admission), cigarette smoking, private health insurance status and SEIFA codes (a small area geographical-based score of social disadvantage) (Australian Bureau of Statistics, 2003).
The main perinatal outcomes of interest were preterm birth, low birthweight, mortality and birth defect risk in ART compared with ULS SC twins and all SC twins. The main outcomes of interest in our analyses of hospital admission included odds of NICU admission for ART twins compared with ULS SC twins; odds of later hospital admission in the first, second and third year of life; and odds of admission with principal diagnoses according to ICD-9 chapters where more than five ART children were admitted in at least one of the considered age groups.
In order to justify our choice of a comparison group restricted to ULS SC twins, we present perinatal outcomes for ART twins compared with both ULS SC twins and all SC twins separately. However, subsequent analyses related to hospital admission include only the preferred comparison group of ULS SC twins.
Data access and ethics approval
Access to the data was approved by the Confidentially of Health Information Committee and the Western Australian Reproductive Technology Council on behalf of the Commissioner for Health. Ethics approval for the study was granted by the Princess Margaret Hospital and King Edward Memorial Hospital Ethics Committee.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
During the period of analysis, 700 children were born as twins after ART conception (14.6% of all twins in WA) and 4097 (85.4%) after SC (Table I). ULS pairs comprised 30% of spontaneous conception twins (n = 1240) and 51.7% of ART twins (n = 362). Of the ART twins, 65.4% were born following IVF, 25.7% following ICSI and 8.9% following GIFT. Compared with ULS SC twins and all SC twins, the ART twins were more likely to be born to older, primiparous mothers (Table I). They were more likely to be born preterm or low birthweight when compared with either ULS SC twins or all SC twins, although the difference was greater when the comparison was with ULS twins. The risk of major birth defects was higher in ART twins than ULS SC twins although the difference did not reach statistical significance (OR 1.4, 95% CI 0.9–2.1) and there was little difference in risk between ART and all SC twins (OR 1.1, 95% CI 0.8–1.6). The risk of perinatal death (OR 2.2, 95% CI 1.1–4.6) was increased in ART twins compared with ULS SC twins, but was similar to the risk in all SC twins (OR 1.1, 95% CI 0.6–1.9). The risks of stillbirth (OR 1.9, 95% CI 0.8–4.9) and neonatal death (OR 2.7, 95% CI 0.8–8.5) separately were also increased compared with ULS SC twins, as was the risk of infant death (OR 1.8, 95% CI 0.4–8.8), but these estimates were not statistically significant.
|
The mothers of ART twins had four times the odds of being covered by private health insurance, twice the odds of living in an area of high social advantage, and 70% less odds of smoking during pregnancy. Figure 1 illustrates the odds of a series of perinatal outcomes in the ART group compared with the ULS SC group and all SC twins, having adjusted for maternal age, parity, infant sex and baby's year of birth.
|
Birth admission
The median LOS of the birth admission was 12 days for ART twins compared with 8 days for ULS SC twins. Stratification showed that preterm birth, low birthweight and major birth defects all increased the length of the birth admission, but in each of these groups the median LOS was 2–3 days longer for ART than for ULS SC infants (data not shown). Those with private health insurance had longer birth admissions than public patients; however, in both groups (private and public) born in metropolitan areas the median LOS was still 2 days longer for ART than ULS SC twins.
Admission to NICU
Admission to NICU was recorded for 34% of ART twins compared with 12% of ULS SC twins (OR ART:ULS SC 3.8, 95% CI 2.8–5.2). There was no difference in admission to NICU between ICSI and IVF infants (OR ICSI:IVF 1.1, 95% CI 0.7–1.8). Only six GIFT infants were admitted to a NICU (data not shown). Table II shows the results of successively adjusting for potential maternal, infant and socio-economic confounders. Adjusting for maternal age, parity, sex and year of birth reduced the odds but ART twins were still more than twice as likely to be admitted to NICU than ULS SC twins (OR ART:ULS SC 2.7, 95% CI 1.9–4.0). The addition of preterm birth and low birthweight to the model had little effect. The difference was further reduced by adjusting for private insurance at birth (OR ART:ULS SC 1.5, 95% CI 1.0–2.4), but not influenced by the subsequent addition of social disadvantage and major birth defect (or maternal smoking during pregnancy—available for 1997–2000 births only, data not shown). When compared with ULS SC twins there was a residual 60% increase in the odds of admission to NICU for ART twins which could not be accounted for by the potential confounders for which we adjusted.
|
Admissions to 3 years of age
Details of the hospitalization for ART and ULS SC children during their first 3 years of life (excluding birth admissions <29 days in length and social admissions) are shown in Table III. During the first year 35.5% of ART and 25.8% of ULS SC twins were admitted to hospital at least once. Twins admitted during the second year comprised 18.9 and 14.2% and during the third year 13.7 and 12.1% of ART and ULS SC twin births, respectively. The median total LOS was significantly longer for ART than ULS SC twins in the first year but the same in the second and third years. The unadjusted odds of admission to hospital were increased for ART twins in each year, although they did not reach statistical significance in the third year. After adjustment for maternal age, parity, year of birth, child sex, preterm birth and low birthweight, the odds were fairly similar across the 3 years and again were statistically significantly increased in the first 2 years (ART:ULS SC year 1 OR 1.4, 95% CI 1.0–1.9; year 2 OR 1.7, 95% CI 1.2–2.4; year 3 OR 1.3, 95% CI 0.8–2.1). The odds of admission changed only slightly after further adjustment for all covariates but remained statistically significant only in the second year (OR 1.6, 95% CI 1.1–2.5). There was no significant difference in odds of hospital admission between ICSI-conceived twins and IVF or GIFT twins (data not shown).
|
Admissions by principal diagnosis
We examined the principal diagnoses by the ICD-9 chapters where more than five ART children were admitted in at least one of the considered age groups (Table IV). There was an increased risk of admission with diseases of the nervous system and sense organs (Chapter 6) amongst ART twins compared with ULS SC twins in the first and second year, and adjustment increased the odds. ART twins appeared to have a decreased risk of admission for diseases of the respiratory system in the first and second year compared with ULS SC twins. In contrast, the odds of admission for diseases of the digestive system (Chapter 9) in ART twins were increased compared with ULS SC twins in all years, although this increase only reached statistical significance in the second year. Odds of admission with a principal diagnosis of congenital malformations (Chapter 14) were higher for ART compared with ULS SC twins in the first and second year although, again, this increase only reached statistical significance in the second year. The crude odds of admission with a condition originating in the perinatal period (Chapter 15) was increased amongst ART twins in the first year, although adjustment reduced the odds. The crude odds of admission with a principal diagnosis of symptoms, signs and ill-defined conditions (Chapter 16) were around twice as common for ART compared with SC twins in all age groups; however, adjustment for potential confounders reduced the odds to non-significant increases in the first and third year. All other categories in Table IV showed no significant difference between ART twins and ULS SC twins.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
Our study found that twins conceived following ART treatment had a greater risk of adverse perinatal outcome including preterm birth, low birthweight and death compared with spontaneously conceived twins of ULS. Differences in perinatal outcome were reduced but remained statistically significant for preterm birth and low birthweight when comparisons included all SC twins rather than ULS twins only. These findings would predict an increase in hospitalization for ART twins and this is confirmed by our findings. Our results show that in their first year of life, ART twins had a longer birth admission; were more likely to be admitted to a NICU; and had a higher risk of at least one hospital admission in addition to the (<29 day) birth admission. ART twins admitted to hospital during the first year also spent longer in hospital. The increased risk of hospital admission continued in the second and third year of life but was not statistically significant in the third year.
Perinatal outcomes
The literature comparing perinatal outcome in ART and SC twins is conflicting despite clear indications of adverse perinatal outcomes in singleton ART pregnancies compared with SC singletons (Helmerhorst et al., 2004; Jackson et al., 2004; McDonald et al., 2005b). Some studies report no association between ART twins and adverse perinatal outcomes (Bergh et al., 1999; Dhont et al., 1999; Westergaard et al., 1999; Isaksson et al., 2002; Koivurova et al., 2002; Pinborg et al., 2004b), others a small positive association (Moise et al., 1998; Koudstaal et al., 2000; Lambalk and van Hooff, 2001; Zuppa et al., 2001; Nassar et al., 2003; Ombelet et al., 2005b; Verstraelen et al., 2005; Klemetti et al., 2006; Kanat-Pektas et al., 2008) and a few report a protective association (Fitzsimmons et al., 1998; Minakami et al., 1998; Boulet et al., 2008). Two meta-analyses that include many of the twin studies published prior to 2004 present conflicting findings for preterm delivery and mortality (Helmerhorst et al., 2004; McDonald et al., 2005a). Given that monozygotic (MZ) twins are known to have an increased risk of adverse outcome and that there is a much lower proportion of MZ twins following ART conception, it would make sense that the results of studies that have not considered zygosity at all should differ from those that have attempted to do this. But results are conflicting even within those studies that have adjusted or stratified their analyses according to zygosity (Moise et al., 1998; Dhont et al., 1999; Koudstaal et al., 2000; Lambalk and van Hooff, 2001; Pinborg et al., 2004b; Ombelet et al., 2005b; Verstraelen et al., 2005; Boulet et al., 2008).
Our results show a clear influence of zygosity on comparisons for most perinatal outcomes. When we restricted the SC comparison group to ULS twin pairs, thereby excluding monozygous twins, we found that ART twins had increased perinatal risks approaching those reported for ART singletons in our previous study (Hansen et al., 2008a, b). Pinborg et al. (2004b) and Boulet et al. (2008) in contrast did not see any effect on perinatal outcome when they restricted their analyses to ULS twins. In a retrospective cohort study of ICSI births, Ombelet et al. (2005b) found no significant difference in perinatal outcome between 1102 ICSI twin births and 2163 SC twins matched for parity, place of birth, date of birth and fetal sex. However, when they restricted their comparison to ULS twins they found a significantly increased risk of preterm birth, low birthweight, perinatal death and stillbirth. Data from the East Flanders prospective twin survey provide complete information about zygosity and Verstraelen et al. (2005) demonstrated that the crude odds of adverse perinatal outcomes for IVF/ICSI twins were reduced by adjusting for maternal age, parity and birth year but then increased again and were statistically significant when they also adjusted for zygosity. Our crude OR estimates for preterm birth and low birthweight (data not shown) are very similar to those reported by Verstraelen et al. (2005).
Given the clear influence of zygosity on comparisons for most perinatal outcomes in our study, we chose to restrict all further comparisons to ULS SC twins only.
NICU admission
The increased risk of adverse perinatal outcomes for ART twins in this study translated to an almost 4-fold increase in the risk of admission to a NICU, before any adjustments were made. This decreased to a 60% increase following adjustment for year of birth, maternal age, parity, infant sex, length of gestation, birthweight, private health insurance, a measure of social disadvantage and major birth defects. Results in the literature have varied widely ranging from pooled OR estimates of 1.0 (95% CI 1.0–1.1) from five ‘matched’ studies (where the comparison group of SC twins was matched to the ART twins on various confounders such as maternal age and parity) and 1.3 (95% CI 1.2–1.4) from four ‘non-matched’ studies (where the SC twins represented a population or hospital-based sample with or without subsequent adjustment for potential confounders in the analyses) in the meta-analysis by Helmerhorst et al. (2004) to a pooled estimate of 2.2 (95% CI 1.6–3.0) from only two studies in the meta-analysis by McDonald et al. (2005b). Results also vary for two large population-based studies that were published more recently. Data from Finland which compared ART multiple births (including higher order multiples) to SC multiple births and did not account for zygosity, showed no increase in NICU admission risk (OR 1.0 (95% CI 0.9–1.2)) (Klemetti et al., 2006). In contrast, in Denmark Pinborg et al. (2004b) found that the odds of NICU admission were increased in ART twins (OR 1.3, 95% CI 1.2–1.5) after restricting analyses to include only ULS twins. They also found that the frequency of ART twins admitted to NICU for >7 days and >28 days was significantly increased compared with SC twins. Taken together, the results of meta-analyses and more recent studies do suggest a small increase in NICU admission risk in ART twins especially when compared with ULS SC twins. This in turn would predict a small increase in hospital admission at least during infancy.
Admission to hospital in the first 3 years of life
In this study, 35.5% of ART and 25.8% of ULS SC twins were admitted to hospital in their first year (excluding birth admissions <29 days long). The 60% increase was no longer statistically significant after adjustment for maternal, infant and socio-economic confounders but was of the same magnitude as the increase seen in ART singletons in our previous study (Hansen et al., 2008a, b) and may simply reflect the smaller number of twin infants involved and thus the limited power of this comparison. During their second and third year, the risk of hospital admission in ART twins was again increased and adjustment for potential confounders led to a further increase in odds. The slightly lower odds of admission in the third year also reflect what we found for singletons from the same population (Hansen et al., 2008a, b).
The largest population-based studies examining early childhood hospital admission in ART twins or multiples have come from Sweden, Denmark and Finland. A number of these have indicated no difference in admission risk (Pinborg et al., 2004a; Klemetti et al., 2006; Koivurova et al., 2007) and others an increased risk (Ericson et al., 2002; Kallen et al., 2005a). The Danish study by Pinborg et al. (2004a) is the only one to have considered zygosity in comparisons of ART and SC twins by restricting comparisons to ULS twins. In contrast to our results, they found no difference in hospital admission between twin groups when the data were restricted to ULS pairs (OR 1.0 (95% CI 0.8–1.1)) and adjusted for year of birth, maternal age and parity. Results from the large Swedish registry studies do suggest an increased admission risk for ART twins although the magnitude of this risk is lower in the earlier study comparing ART and SC twins (OR 1.2 (95% CI 1.1–1.3)) (Ericson et al., 2002) than the second larger study where all multiples are grouped together (OR 1.9 (95% CI 1.9–2.0)) (Kallen et al., 2005a). No adjustments are made for differences in zygosity between the two groups.
Ericson et al. (2002) also compared the average number of days spent in hospital in ART and SC twins and found no difference between groups. In our study, we found no difference in LOS in the second and third year, although ART twins were admitted for longer periods than ULS twins in the first year reflecting their increased risk of adverse perinatal outcome and NICU admission. The increased hospital admission risk seen across each year in our study, although not statistically significant in the third year, are in keeping with Kallen et al. (2005a) who found small increased odds of admission in ART children up to the age of 6 years. There is also agreement across studies that no difference exists in hospital admission risk when IVF twins are compared with ICSI twins (Ericson et al., 2002; Pinborg et al., 2004a; Klemetti et al., 2006).
Principal diagnoses
As discussed in our analysis of hospital admission in ART singletons (Hansen et al., 2008a, b), comparisons of hospital diagnoses reported in the literature are complicated by a range of issues. Differences may exist across studies in the specific categories of diagnoses examined, the groups of children included (all infants versus twins or ULS twins only), and by different attitudes to hospitalization between countries (e.g. day treatment versus hospital admission for the same medical condition). Small numbers in many categories also limit confident interpretation and our findings regarding specific diagnoses should be interpreted with caution.
We did not see any increase in diseases of the genitourinary system as we had reported for ART singletons, nor did we see the same pattern of increased risk of admission for birth defects across each age group. Risk of admission with a principal diagnosis of birth defects was significantly increased only in the second year for ART twins. When our data were linked to the Western Australian Birth Defects Registry in order to identify all of the children in our study who were diagnosed with a major birth defect (regardless of hospital admission), we found that ART twins did not have a significant excess risk of birth defects compared with SC twins although comparisons with ULS SC twins showed a 40% excess risk which was not statistically significant (Table I). Birth defect risk in ART multiples was also not significantly increased in the meta-analyses by Rimm et al. (2004) or McDonald et al. (2005a) and the authors suggest this may be because of the different zygosity distributions between ART and SC twins.
ART twins had a decreased risk of admission for diseases of the respiratory system in the first and second year compared with ULS SC twins and this was also reported by Klemetti et al. (2006) in a study of Finnish ART infants. Also in accordance with this study, we found increases in admission for conditions originating in the perinatal period (in the first year) and ‘Symptoms, signs and ill-defined conditions’ referred to as ‘Other’ diagnoses by Klemetti et al. (2006).
Admissions for diseases of the digestive system were increased in comparisons with ULS SC twins in the second year of life. We could not confirm increases in admission for asthma (data not shown) or infection reported in a large Swedish study (Ericson et al., 2002; Kallen et al., 2005a), however diseases of the nervous system were increased compared with ULS SC twins in the first and second year. The number of children with admissions for tumors was too small to report meaningful results.
Strengths and limitations
Our study has a number of methodological advantages including a large population-based data source obtained from record linkage of existing databases, and independent analyses of each of the first 3 years of life so that children who had died before the beginning of each time period were not included in the denominators for that or subsequent periods. A greater proportion of ART twins than SC twins are born in the later years of our study (an observation that probably extends to all studies that examine hospital admission since ART use is increasing over time) and therefore proportionately fewer of the ART group contributes 3 years of follow-up. Grouping admission data across years as others have done, even with adjustment for year of birth, would bias the results towards the null as the ART group would have less exposure time overall than the SC group in which to ‘accrue’ an admission. We used a method of analysis (GEE) that took into account intra-sibling correlations and we were able to compare ART twins (who are largely dizygous by virtue of their method of conception) with ULS SC twin pairs in order to adjust, at least in part, for differences in zygosity distributions between the two groups. We were able to show that comparisons with all SC twins (rather than with ULS SC twins only) produced lower estimates for a number of adverse perinatal outcomes including stillbirths, neonatal deaths and infant death, as well as very low birthweight and very preterm birth.
Validation studies have been undertaken to ensure the quality of routinely collected data in Western Australia's Hospital Morbidity Data System and Midwives’ Notification of Birth System and these have found that data on primary diagnoses and procedures are accurately coded and seldom missing (Gee and Dawes, 1994; Holman et al., 1999). In the case of hospital morbidity data, there are 21 different quality checks built into the provision of data from hospitals and there are periodic audits of random selections of hospital-assigned codes. Although information on up to 20 co-morbidities is recorded we specifically did not analyze this information as there is evidence that co-morbidities are both under-reported and differentially reported between public and private hospitals in Western Australia (Preen et al., 2004).
The limitations of this study include its restriction to analysis of hospital admission data. We had no information about the use of other healthcare services. In addition, the Western Australian Reproductive Technology Register does not record information about the use of non-ART treatments such as intrauterine insemination or ovulation induction and births following these treatments will be included in our SC comparison group. Some authors have estimated the proportion of their SC twin group likely to have been conceived following ovulation stimulation alone and results vary from 11 to 17% (Kurinczuk et al., 1995; Pinborg et al., 2003; Verstraelen et al., 2005). It has been suggested that the use of these treatments is also associated with adverse perinatal outcomes (Kallen et al., 2002; Wang et al., 2002; Gaudoin et al., 2003) so that these children may be expected to have higher rates of hospital admission. If this is the case our results for the SC group would be biased towards the null and would thus underestimate the size of the real difference.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
Our results suggest that ART twins, like ART singletons, have an increased risk of NICU admission and are more likely to be admitted to hospital in their first 3 years of life compared with ULS SC twins. Although most of the increase in admission risk can be accounted for by known maternal and infant characteristics, there is still a residual risk, particularly in the second year of life. The increase in adverse perinatal outcomes and subsequent hospital admission may in part be due to underlying causes of parental infertility, to components of the ART procedure or to increased concern about children who are born following a long period of infertility. Various arguments have been made in the literature in support of each of these hypotheses (Klemetti et al., 2002
; Kallen et al., 2005b; Kapiteijn et al., 2006; Romundstad et al., 2008), and all merit further investigation, however we are unable to explore them further with the data available to us.
Estimations of the cost of an ART twin delivery should include the extra 
4 days on average spent in hospital at birth, the almost 4-fold increased risk of admission to a NICU and the increased risk of hospital admission in the first three years of life. In addition, clinicians and couples undertaking ART need to consider the well-documented excess costs of ART twin deliveries compared with ART singletons and the benefits of elective SET for selected patient groups. Recent evidence that these considerations are being taken very seriously include changes to the Human Fertilization and Embryology Authority Code of Practice in the United Kingdom requiring that clinics develop a ‘multiple births minimization strategy’ with the aim of reducing the rate of multiple pregnancies from 24 to 10% in the next 3 years (http://cop.hfea.gov.uk/cop/). In 2001 the Belgian government changed its reimbursement policy for IVF-related procedures to allow for the reimbursement of up to six cycles in a lifetime provided reimbursement of IVF centre costs were linked to a 50% reduction in twin pregnancies and higher order multiple pregnancies were minimized to almost zero (Ombelet et al., 2005a). This method has the advantage of providing an added financial incentive to patients for choosing SET. In many countries where a large proportion of IVF treatment costs are borne by the patient, double embryo transfer with its slightly higher pregnancy rate may unfortunately still appear more attractive regardless of the perinatal and longer term risks associated with an ART twin birth.
|
Author's Role |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
M.H. and L.C. prepared the linked data files for analysis. M.H. conducted the analyses and collated the results. M.H. and B.P. drafted the paper. N.de K. provided statistical advice. C.B. and J.K. contributed to the study design and data analysis. All authors contributed to the interpretation of the data and all were involved in the critical revision of the paper.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
National Health and Medical Research Council of Australia (Project Grant #211930 to M.H., Principal Research Fellowship #353628 to C.B.).
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
We are indebted to all the contributors to, and staff of, the Midwives’ Notification of Birth System, the WA Reproductive Technology Register, the Hospital Morbidity Data System and the WA Birth Defects Registry. Special thanks to Ms Carol Garfield and Mr Peter Cosgrove for providing data linkage between registers.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAuthor's RoleFundingAcknowledgementsReferences |
|---|
Andersen AN, Goossens V, Ferraretti AP, Bhattacharya S, Felberbaum R, de Mouzon J, Nygren KG. Assisted reproductive technology in Europe, 2004: results generated from European registers by ESHRE. Hum Reprod (2008) 23:756–771.
Australian Bureau of Statistics. Socio-Economic Indexes for Areas. In: Information Paper. 2001 Census of Population and Housing (2003) Canberra: Commonwealth of Australia.
Bergh T, Ericson A, Hillensjo T, Nygren KG, Wennerholm UB. Deliveries and children born after in-vitro fertilisation in Sweden 1982–95: a retrospective cohort study.[see comment]. Lancet (1999) 354:1579–1585.[CrossRef][Web of Science][Medline]
Boulet SL, Schieve LA, Nannini A, Ferre C, Devine O, Cohen B, Zhang Z, Wright V, Macaluso M. Perinatal outcomes of twin births conceived using assisted reproduction technology: a population-based study. Hum Reprod (2008) 23:1941–1948.
Carville KS, Lehmann D, Hall G, Moore H, Richmond P, de Klerk N, Burgner D. Infection is the major component of the disease burden in aboriginal and non-aboriginal Australian children: a population-based study. Pediatr Infect Dis J (2007) 26:210–216.[CrossRef][Web of Science][Medline]
Centers for Disease Control and Prevention, American Society for Reproductive Medicine and Society for Assisted Reproductive Technology. 2005 Assisted Reproductive Technology Success Rates: National Summary and Fertility Clinic Reports (2007) Atlanta: Centers for Disease Control and Prevention.
Derom R, Derom C. The east flanders prospective twin survey. In: Multiple Pregnancy: epidemiology, gestation and perinatal outcome—Keith LG, Blickstein I, eds. (2005a) London: Taylor and Francis. 39–47.
Derom R, Derom C. Placentation. In: Multiple pregnancy: epidemiology, gestation and perinatal outcome—Keith LG, Blickstein I, eds. (2005b) London: Taylor and Francis. 157–167.
Dhont M, De Sutter P, Ruyssinck G, Martens G, Bekaert A. Perinatal outcome of pregnancies after assisted reproduction: a case–control study. Am J Obstet Gynecol (1999) 181:688–695.[CrossRef][Web of Science][Medline]
Ericson A, Nygren KG, Olausson PO, Kallen B. Hospital care utilization of infants born after IVF. Hum Reprod (2002) 17:929–932.
Fitzsimmons BP, Bebbington MW, Fluker MR. Perinatal and neonatal outcomes in multiple gestations: assisted reproduction versus spontaneous conception. Am J Obstet Gynecol (1998) 179:1162–1167.[CrossRef][Web of Science][Medline]
Gaudoin M, Dobbie R, Finlayson A, Chalmers J, Cameron IT, Fleming R. Ovulation induction/intrauterine insemination in infertile couples is associated with low-birth-weight infants.[see comment]. Am J Obstet Gynecol (2003) 188:611–616.[CrossRef][Web of Science][Medline]
Gee V, Dawes V. Validation study of the Western Australian Midwives' Notification System 1992 (1994) Perth: Health Department of Western Australia.
Hansen M, Kurinczuk JJ, Bower C, Webb S. The risk of major birth defects after intracytoplasmic sperm injection and in vitro fertilization. NEJM (2002) 346:725–730.
Hansen M, Bower C, Milne E, de Klerk N, Kurinczuk JJ. Assisted reproductive technologies and the risk of birth defects—a systematic review. Hum Reprod (2005) 20:328–338.
Hansen M, Colvin L, Petterson B, Kurinczuk JJ, de Klerk N, Bower C. Admission to hospital of singleton children born following assisted reproductive technology (ART). Hum Reprod (2008a) 23:1297–1305.
Hansen M, Colvin L, Petterson B, Kurinczuk JJ, de Klerk N, Bower C. Erratum: admission to hospital of singleton children born following assisted reproductive technology (ART). Hum Reprod (2008b) 23:2390.
Helmerhorst FM, Perquin DA, Donker D, Keirse MJ. Perinatal outcome of singletons and twins after assisted conception: a systematic review of controlled studies. BMJ (2004) 328:261.
Holman CD, Bass AJ, Rouse IL, Hobbs MS. Population-based linkage of health records in Western Australia: development of a health services research linked database. Aust N Z J Public Health (1999) 23:453–459.[Web of Science][Medline]
Isaksson R, Gissler M, Tiitinen A. Obstetric outcome among women with unexplained infertility after IVF: a matched case–control study. Hum Reprod (2002) 17:1755–1761.
Jackson RA, Gibson KA, Wu YW, Croughan MS. Perinatal outcomes in singletons following in vitro fertilization: a meta-analysis.[see comment]. Obstet Gynecol (2004) 103:551–563.[Web of Science][Medline]
Kallen B, Olausson PO, Nygren KG. Neonatal outcome in pregnancies from ovarian stimulation. Obstet Gynecol (2002) 100:414–419.[CrossRef][Web of Science][Medline]
Kallen B, Finnstrom O, Nygren KG, Olausson PO. In vitro fertilization in Sweden: child morbidity including cancer risk. Fertil Steril (2005a) 84:605–610.[CrossRef][Web of Science][Medline]
Kallen B, Finnstrom O, Nygren KG, Otterblad Olausson P. In vitro fertilization in Sweden: maternal characteristics. Acta Obstet Gynecol Scand (2005b) 84:1185–1191.[CrossRef][Web of Science][Medline]
Kanat-Pektas M, Kunt C, Gungor T, Mollamahmutoglu L. Perinatal and first year outcomes of spontaneous versus assisted twins: a single center experience. Arch Gynecol Obstet (2008) 278:143–147.[CrossRef][Web of Science][Medline]
Kapiteijn K, de Bruijn CS, de Boer E, de Craen AJ, Burger CW, van Leeuwen FE, Helmerhorst FM. Does subfertility explain the risk of poor perinatal outcome after IVF and ovarian hyperstimulation? Hum Reprod (2006) 21:3228–3234.
Klemetti R, Gissler M, Hemminki E. Comparison of perinatal health of children born from IVF in Finland in the early and late 1990s. Hum Reprod (2002) 17:2192–2198.
Klemetti R, Sevon T, Gissler M, Hemminki E. Health of children born as a result of in vitro fertilization. Pediatrics (2006) 118:1819–1827.
Koivurova S, Hartikainen AL, Gissler M, Hemminki E, Sovio U, Jarvelin MR. Neonatal outcome and congenital malformations in children born after in-vitro fertilization. Hum Reprod (2002) 17:1391–1398.
Koivurova S, Hartikainen AL, Sovio U, Gissler M, Hemminki E, Jarvelin MR. Growth, psychomotor development and morbidity up to 3 years of age in children born after IVF. Hum Reprod (2003) 18:2328–2336.
Koivurova S, Hartikainen AL, Gissler M, Hemminki E, Jarvelin MR. Post-neonatal hospitalization and health care costs among IVF children: a 7-year follow-up study. Hum Reprod (2007) 22:2136–2141.
Koudstaal J, Bruinse HW, Helmerhorst FM, Vermeiden JPW, Willemsen WNP, Visser GHA. Obstetric outcome of twin pregnancies after in-vitro fertilization: a matched control study in four Dutch University hospitals. Hum Reprod (2000) 15:935–940.
Kurinczuk JJ, Pemberton RJ, Binns SC, Parsons DE, Stanley FJ. Singleton and twin confinements associated with infertility treatments. Aust N Z J Obstet Gynaecol (1995) 35:27–31.[Web of Science][Medline]
Lambalk CB, van Hooff M. Natural versus induced twinning and pregnancy outcome: a Dutch nationwide survey of primiparous dizygotic twin deliveries. Fertil Steril (2001) 75:731–736.[CrossRef][Web of Science][Medline]
Luke B, Keith LG. The contribution of singletons, twins and triplets to low birth weight, infant mortality and handicap in the United States. J Reprod Med (1992) 37:661–666.[Web of Science][Medline]
McDonald S, Murphy K, Beyene J, Ohlsson A. Perinatal outcomes of in vitro fertilization twins: a systematic review and meta-analyses. Am J Obstet Gynecol (2005a) 193:141–152.[CrossRef][Web of Science][Medline]
McDonald SD, Murphy K, Beyene J, Ohlsson A. Perinatal outcomes of singleton pregnancies achieved by in vitro fertilization: a systematic review and meta-analysis. J Obstet Gynaecol Can (2005b) 27:449–459.[Medline]
Minakami H, Sayama M, Honma Y, Matsubara S, Koike T, Sato I, Uchida A, Eguchi Y, Momoi M, Araki S. Lower risks of adverse outcome in twins conceived by artificial reproductive techniques compared with spontaneously conceived twins. Hum Reprod (1998) 13:2005–2008.
Moise J, Laor A, Armon Y, Gur I, Gale R. The outcome of twin pregnancies after IVF. Hum Reprod (1998) 13:1702–1705.
Nassar AH, Usta IM, Rechdan JB, Harb TS, Adra AM, Abu-Musa AA. Pregnancy outcome in spontaneous twins versus twins who were conceived through in vitro fertilization.[see comment]. Am J Obstet Gynecol (2003) 189:513–518.[CrossRef][Web of Science][Medline]
Ombelet W, De Sutter P, Van der Elst J, Martens G. Multiple gestation and infertility treatment: registration, reflection and reaction—the Belgian project. Hum Reprod Update (2005a) 11:3–14.
Ombelet W, Peeraer K, De Sutter P, Gerris J, Bosmans E, Martens G, Ruyssinck G, Defoort P, Molenberghs G, Gyselaers W. Perinatal outcome of ICSI pregnancies compared with a matched group of natural conception pregnancies in Flanders (Belgium): a cohort study. Reprod Biomed Online (2005b) 11:244–253.[Web of Science][Medline]
Pinborg A, Loft A, Schmidt L, Andersen AN. Morbidity in a Danish national cohort of 472 IVF/ICSI twins, 1132 non-IVF/ICSI twins and 634 IVF/ICSI singletons: health-related and social implications for the children and their families. Hum Reprod (2003) 18:1234–1243.
Pinborg A, Loft A, Rasmussen S, Nyboe Andersen A. Hospital care utilization of IVF/ICSI twins followed until 2–7 years of age: a controlled Danish national cohort study. Hum Reprod (2004a) 19:2529–2536.
Pinborg A, Loft A, Rasmussen S, Schmidt L, Langhoff-Roos J, Greisen G, Andersen AN. Neonatal outcome in a Danish national cohort of 3438 IVF/ICSI and 10,362 non-IVF/ICSI twins born between 1995 and 2000. Hum Reprod (2004b) 19:435–441.
Preen DB, Holman CD, Lawrence DM, Baynham NJ, Semmens JB. Hospital chart review provided more accurate comorbidity information than data from a general practitioner survey or an administrative database. J Clin Epidemiol (2004) 57:1295–1304.[CrossRef][Web of Science][Medline]
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. J Assist Reprod Genet (2004) 21:437–443.[CrossRef][Web of Science][Medline]
Romundstad LB, Romundstad PR, Sunde A, von During V, Skjaerven R, Gunnell D, Vatten LJ. Effects of technology or maternal factors on perinatal outcome after assisted fertilisation: a population-based cohort study. Lancet (2008) 372:737–743.[CrossRef][Web of Science][Medline]
Verstraelen H, Goetgeluk S, Derom C, Vansteelandt S, Derom R, Goetghebeur E, Temmerman M. Preterm birth in twins after subfertility treatment: population based cohort study. BMJ (2005) 331:1173.
Wang JX, Norman RJ, Kristiansson P. The effect of various infertility treatments on the risk of preterm birth. Hum Reprod (2002) 17:945–949.
Wang YA, Dean JH, Badgery-Parker T, Sullivan EA. Assisted reproduction technology in Australia and New Zealand 2006. AIHW Cat. No. PER 43 (2008) Sydney: AIHW National Perinatal Statistics Unit.
Westergaard HB, Johansen AM, Erb K, Andersen AN. Danish National In-Vitro Fertilization Registry 1994 and 1995: a controlled study of births, malformations and cytogenetic findings. Hum Reprod (1999) 14:1896–1902.
Zuppa AA, Maragliano G, Scapillati ME, Crescimbini B, Tortorolo G. Neonatal outcome of spontaneous and assisted twin pregnancies. Eur J Obstet Gynecol Reprod Biol (2001) 95:68–72.[CrossRef][Web of Science][Medline]
Submitted on January 23, 2009; resubmitted on April 9, 2009; accepted on April 15, 2009.
CiteULike 
Connotea 
Del.icio.us What's this?
Related articles in Hum. Reprod.:
- Editor's Choice
- Johannes L.H. Evers
Hum. Reprod. 2009 24: 2051-2052.[Extract] [FREE Full Text]
This article has been cited by other articles:
|
|
 |
|
  Twins Born Following ART Have Higher Risk for Morbidity Than Spontaneously Conceived Twins Journal Watch Women's Health, October 22, 2009; 2009(1022): 3 - 3. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||