Does assisted reproduction technology, per se, increase the risk of preterm birth?


  • I Blickstein

    Corresponding author
    1. Department of Obstetrics and Gynecology, Kaplan Medical Center, Rehovot, Israel
      Prof I Blickstein, Department of Obstetrics and Gynecology, Kaplan Medical Center, 76100 Rehovot, Israel. Email
    Search for more papers by this author

Prof I Blickstein, Department of Obstetrics and Gynecology, Kaplan Medical Center, 76100 Rehovot, Israel. Email


There is little doubt that all methods of assisted reproduction increase the likelihood of multiple pregnancy and, as a result, increase the likelihood of preterm birth. Data from the East Flanders Prospective Twin Study clearly show that the proportion of spontaneous to iatrogenic twins has changed from 25:1 to 1:1 over the past two decades. Data from the very low birthweight (VLBW) Infant Database of the Israel Neonatal Network showed that 10% of VLBW singletons were a result of assisted reproduction compared with 60% of the VLBW twins and 90% of the VLBW triplets. Irrespective of plurality, an association between preterm birth and assisted reproduction has long been suspected and was related to causes such as iatrogenic preterm birth (in the so-called ‘premium’ pregnancies), fertility history, past obstetric performance and to underlying medical conditions of the female partner. With more data available, a clearer picture is defined. Two different, recent meta-analyses showed that singleton pregnancies resulting from in vitro fertilisation (IVF) have increased rates of preterm birth at <33 weeks of gestation (OR 2.99; 95% CI 1.54–5.80), at <37 weeks of gestation (OR 1.93; 95% CI 1.36–2.74) and a relative risk of 1.98 (95% CI 1.77–2.22) for preterm birth in singleton pregnancies resulting from in vitro fertilisation embryo transfer/gamete intra fallopian transfer (IVF-ET/GIFT) compared with naturally conceived pregnancies. Since there is no way to predict which pregnant woman is at increased risk of preterm birth, it may be advisable to consider all pregnancies after assisted reproduction as being at risk. In any case, the most appropriate endpoint after assisted reproduction should also include preterm or term birth as measure of success.


The latest report on births in the USA showed that preterm birth rate (percentage of infants delivered at <37 completed weeks of gestation) increased 0.2%, from 12.1% of all births in 2002 to 12.3% in 2003.1 This figure represents an increase of 16% since 1990 (from 10.6%) and of more than 30% since 1981 (9.4%). Most of the 2003 increase was among moderately preterm births (32–36 weeks of gestation, from 10.12 to 10.37%), whereas the percentage of very preterm infants born at <32 weeks of gestation was essentially unchanged (1.97 compared with 1.96% in 2002) and reflects a moderate rise in the rate of very preterm births from 1.81% since 1981.

Preterm birth is a major contributor to perinatal morbidity and mortality as well as to infant death and long-term neurological morbidity. The increased incidence of multiple births, which are significantly more likely than singletons to be born preterm, has long been considered as a major cause of the increased overall preterm birth rate seen in the past two decades. The data indicate that the preterm birth rate for singletons is also rising steadily, albeit in the moderate preterm birth group.1

It follows that a logical sequence exists between factors which contribute to the increased frequency of multiple births (i.e. infertility treatment) and the overall increased incidence of preterm births in a population.2 At the same time that higher order multiple birth rates decreased—for the first time after the preceding steep four-fold increase since the early 1980s—the incidence of twins continues to escalate; hence, the total number of multiple births is still increasing.1,2 Since actual numbers of higher order multiples are lower than the number of twins and because twins are predictably associated with significant preterm birth rates, the implications of the ever-increasing multiple birth rate is no less alarming, and the decreasing incidence of higher order multiples, per se, is not an indication for the end of the epidemic of multiple births.2

Whereas no arguments exist as to the role of infertility treatment in increasing the rate of preterm birth by increasing the multiple birth rates, some doubts exist as to whether fertility treatment, per se, increases the risk of preterm birth. This article discusses this question.


Preterm birth rates in multiple and singleton pregnancies are different. Figure 1 shows that, in common with the human female, some but not all females in the animal kingdom demonstrate a plurality-dependent reduction of the gestational duration with increasing number of fetuses. In human beings, there is highly significant correlation (Pearson’s r2= 0.99), satisfying a polynomial equation (gestational age = 0.36X2− 4.7X + 43.2, where X = plurality).3 The different distributions of gestational age at birth in different pluralities suggest that the human uterus is simply unable to maintain a multiple pregnancy. As a result, only 1.6% of singletons are born before 32 weeks of gestation, whereas 11.9% of twins and 36.1% of triplets are delivered before this gestational age.3

Figure 1.

Gestational age by plurality (based on data from Martin et al.3).

It would be wrong to compare the outcomes between singletons and multiples to ascertain whether fertility treatment, per se, increases the risk of preterm births, since there is no doubt that twins have a higher perinatal mortality than singletons, and triplets do worse than twins.4

Numerous studies have been conducted to compare the outcomes of singletons conceived by in vitro fertilisation (IVF) compared with the outcomes of those following natural conceptions. Several meta-analyses have been carried out, summarising the available data. Population-based studies for both twins and singletons cohorts have also been reported. In this article, I will consider the most recent (2004 or later) publications, which are more likely to cover the entire relevant literature.


In January 2004, Helmerhorst et al.5 published a systematic review of 25 controlled studies (17 with matched and 8 with nonmatched controls) published during the period 1985–2002. Studies with matched and nonmatched controls yielded similar findings. For singletons, a relative risk of 3.27 (95% CI 2.03–5.28) for very preterm (<32 weeks) and 2.04 (95% CI 1.80–2.32) for preterm (<37 weeks) birth was found in pregnancies after IVF. In twin gestations, relative risks were 0.95 (95% CI 0.78–1.15) for very preterm birth and 1.07 (1.02, 1.13) for preterm birth. Later that year, another meta-analysis was performed by Jackson et al.6 In this study, the authors searched a wide range of available publications to select 15 studies comprising 12 283 IVF and 1.9 million naturally conceived singletons. Compared with natural conceptions, IVF singleton pregnancies were associated with significantly higher odds of preterm birth (OR 2.0; 95% CI 1.7–2.2). In the very preterm births (<32–33 weeks), the authors derived a respective OR of 3.2 (95% CI 2.0–4.8). This finding was consistently observed despite different study designs (i.e. cohort, matched cohort and external comparisons) used for analysis, diverse patient populations and, importantly, different IVF and obstetric protocols. The results did not change after controlling for maternal age and parity, but statistical heterogeneity was noted only for preterm birth and low birthweight (LBW).

In December of the same year, another meta-analysis was published by McGovern et al.7 who performed a systematic review of the literature to determine whether singleton pregnancies resulting from IVF-ET/GIFT were at higher risk for preterm birth. The authors used a MEDLINE search of the years 1965–2000, found 27 articles that met their inclusion/exclusion criteria and performed a summary of relative risks of preterm birth. The random-effects summary relative risk of preterm birth in singleton pregnancies resulting from IVF-ET/GIFT was 1.98 (95% CI 1.77–2.22), almost identical to the finding of the previous meta-analysis performed by Jackson et al.6

The most recent analysis appeared in May 2005 when a Canadian meta-analysis was published by McDonald et al.8 In this study, the authors followed the Meta-Analysis of Observational Studies in Epidemiology guidelines for meta-analysis of observational studies and used comprehensive search strategies to select case–control (i.e. matched for maternal age) and cohort (i.e. controlled for maternal age) studies that compared singleton pregnancies conceived by IVF/intracytoplasmic sperm injection (ICSI) with naturally conceived singletons. This analysis confirmed the findings of the previous studies, namely, that singleton pregnancies resulting from IVF had increased rates of poor obstetric outcome, compared with naturally conceived singletons matched for maternal age, with increases in preterm births at <33 weeks of gestation (OR 2.99; 95% CI 1.54–5.80) and preterm births in total (OR 1.93; 95% CI 1.36–2.74).

Population-based analyses

In contrast to the numerous cohort and case–control studies, there are few population-based studies comparing outcomes of IVF and natural conceptions. Of these, two studies deserve special attention. The first examined the data pertaining to births from assisted reproduction techniques. The authors used the Australian birth records from 1996 to 2000 to study retrospectively the national population data of 18 429 infants conceived through assisted reproduction techniques.9 Preterm birth and LBW were more common among singletons and twins conceived with IVF and born to nulliparae.

The second population-based cohort study examined the East Flanders prospective twin survey to assess gestational length and prevalence of preterm birth among medically and naturally conceived twins.10 A total of 2915 spontaneous twin pairs, 710 twin pairs born after ovarian stimulation and 743 twin pairs born after IVF/ICSI, were evaluated. Compared with naturally conceived twins, twins resulting from subfertility treatment had a slightly decreased gestational age at birth (mean difference 4.0 days, 95% CI 2.7–5.2), corresponding to an odds ratio of 1.6 (95% CI 1.4–1.8) for preterm birth (mainly mild, 32–37 weeks, preterm). The adjusted odds ratios of preterm birth after subfertility treatment were 1.3 (95% CI 1.1–1.5) when controlled for birth year, maternal age and parity and 1.6 (95% CI 1.3–1.8) with additional control for fetal sex, caesarean section, zygosity and chorionicity. The authors reached a similar conclusion to that of Wang et al. that although an increased risk of preterm birth was seen among twins resulting from subfertility treatment, this risk was due to the effect of nulliparity among subfertile couples.

Open questions

There is significant difference between the risks of preterm birth found in meta-analyses compared with population-based data. Meta-analyses show a two- to three-fold increased risk, whereas population-based data show only a slightly increased risk. Despite the apparent discrepancy of the results of the different analytical methods, one might conclude that IVF, per se, increases—to a lesser or to a greater extent—the risk of preterm birth. This conclusion approximates the French IVF data from the late 1980s, which observed that multiple pregnancies are not the only risk factor for preterm birth, which is also more frequent among IVF singletons.11 The authors went on to discuss that the quantitative magnitude of this qualitative conclusion remains to be established.

The aetiological explanation for the IVF-associated increased preterm birth rate is also important. The increase in the rate of preterm birth related to a female factor indication for assisted reproduction technique rather than to male factor indications9,12 is important. Confusing results arise following comparison of assisted reproduction technique pregnancies and ‘low technology’ treatments with natural conceptions. Several studies found slightly increased risks of preterm birth following ‘low technology’ conceptions compared with a two-fold increase in the assisted reproduction technique group, suggesting that the influence of infertility technology on increased preterm birth rates cannot be overlooked.9,10 Conversely, when IVF + ICSI cases (i.e. micromanipulation of the gametes primarily performed for male factor infertility) were compared with standard IVF, no statistically significant difference in preterm birth rates were detected.13 This conflicting observation suggests that fertile and infertile women undergoing IVF are at similar risk of preterm birth.

An alternative aetiology for increased preterm birth rates among IVF infants has been suggested by Pinborg et al.14 This Danish group explored the predicting factors of preterm birth and LBW in the survivor of the vanishing twin syndrome following IVF. Of all IVF singletons, 10.4% originated from a twin gestation in early pregnancy. Multiple logistic regression analyses adjusted for maternal age, parity and ICSI treatment showed that survivors of a vanishing co-twin were more likely to be born preterm than singletons originating from a single gestation. This was more pronounced for very early preterm birth, where risks were two- to three-fold. The authors proposed that the >10% prevalence of IVF/ICSI singleton births that started as a twin gestation may partly explain the overall higher rate of preterm birth in IVF/ICSI singletons. The population-based observations of Pinborg et al. were not entirely confirmed by a large, single-centre study.15

A third explanation for the increased preterm birth rate came from the increased likelihood of a more advanced maternal age among IVF pregnancies.16 This problem of iatrogenic pregnancies were derived from major changes in the lifestyle of women during the second half of the 20th century, whereby the age at which childbirth is desired became advanced. At the same time, the naturally reduced fecundity of advanced age, as well as involuntary infertility, were met by effective assisted reproduction technique. With advanced maternal age at conception, an increased likelihood of chronic diseases such as hypertension and diabetes occurred, which, in turn, might be aggravated during pregnancy and result in iatrogenic preterm birth. The converse is also possible, whereby some maternal disease conditions decreased fecundity and increased the need for assisted reproduction technique.

IVF pregnancies are considered by many to be ‘premium’ pregnancies,17 and for some women, the IVF pregnancy might be the first as well as the last pregnancy. Consequently, IVF pregnancies are highly valued by women and their caregivers, and the women are more likely to be hospitalised during pregnancy to undergo labour induction or caesarean section for relatively minor complications.6 It follows that at least a proportion of preterm births in women with IVF are due to treatment bias, but the extent to which this iatrogenic effect increases the rate of preterm birth is unknown. The meta-analysis of Jackson et al.6 found that adverse outcomes such as very early preterm birth were relatively independent of treatment decisions, whereas heterogeneity was detected for outcomes like preterm and abdominal deliveries that may depend on subjective management decisions.


There is an identifiably increased risk of preterm birth in pregnancies following assisted reproduction technique. There is a substantially higher risk of preterm birth derived from meta-analyses (two- to three-fold) compared with that of population cohort studies (30–60%). The reason for the increased risk of preterm birth following assisted reproduction technique is unclear. Some data suggest a method-related cause, whereas other data suggest that infertility rather than its treatment is the cause. The fact that infertile women in need of assisted reproduction technique are older and more likely to have underlying chronic disease may also explain a patient-related increase in preterm birth. At the same time, a potential treatment bias, such as iatrogenic preterm birth, cannot be excluded.