Compared with naturally conceived children, singletons born after assisted reproductive technology (ART) have an elevated risk of perinatal mortality, preterm birth, low birth weight, small-for-gestational-age (SGA), and congenital malformations.1–3 These findings have been interpreted as reflecting side effects of infertility treatment but could, in part, reflect an increased risk among infertile couples. Infertility per se has been associated with a number of the same adverse pregnancy outcomes.4–10
Although infertility appears to be associated with decreased birth weight,4,9 the relation between infertility and fetal growth has not been examined in detail. Fetal growth restriction (FGR) reflects a failure to fulfill a fetus’s biologic growth potential but, at present, we have no good measure for this phenomenon.11 The criterion of SGA usually identifies fetuses below the 5th or 10th percentile of the distribution of birth weight at any given gestational age. A large proportion of FGR fetuses will be captured by this definition, although it does not reflect changes in growth rate during fetal life. SGA is associated with an increased risk of perinatal mortality and morbidity.12,13 Using data from the Danish National Birth Cohort,14 we examined the association between infertility, with or without treatment, and fetal growth, as well as perinatal and infant mortality.
MATERIALS AND METHODS
The Danish National Birth Cohort was established to study the importance of environmental and lifestyle factors during pregnancy and early childhood on the health and development of children.14 The first of four interviews was administered during the first and second trimester of pregnancy (median 16th week) between 1997 and 2003. Of the 92,892 pregnancies with available data, we excluded 2,700 from women no longer pregnant at the time of interview, 22,336 unplanned or partly planned pregnancies, and 29 pregnancies with unknown infertility treatment, leading to 67,827 planned pregnancies with a known time to pregnancy. To avoid informative clusters, we only included the first pregnancy contributed by each participant, thus excluding 3,973 pregnancies. We also excluded 66 pregnancies for which infertility treatment not associated with this pregnancy was reported, and 233 conceived after treatment other than intracytoplasmic sperm injection (ICSI), in vitro fertilization (IVF), intrauterine insemination (IUI), or hormonal treatment (HT). Pregnancy outcome was ascertained through linkage to the National Hospital Register and the Medical Birth Register by means of the unique civil registration number assigned to all residents. For this analysis, we excluded pregnancies terminated by spontaneous or induced abortions, ectopic pregnancies, hydatidiform moles (n=752), pregnancies with unknown outcome (n=44), and those resulting in the delivery of twins or triplets (n=1,614). We divided the remaining 61,145 singletons into 3 groups: A) 51,041 born of fertile couples (time to pregnancy 12 months or less), B) 5,787 born of infertile couples conceiving naturally (time to pregnancy more than 12 months), and C) 4,317 born after infertility treatment. While couples in group A were considered fertile, couples in groups B and C were categorized infertile.
In the first interview, women were asked if their pregnancy was planned and, if so, for how long they had tried to become pregnant before succeeding. Response categories for time to pregnancy were “right away,” 1–2, 3–5, 6–12, and more than 12 months. We interpreted “right away” as conceiving within the first cycle and labeled this category as a time to pregnancy of 0. Participants reporting a time to pregnancy of more than 6 months were further asked if they or their male partner had received any infertility treatment, including ICSI, IVF, IUI, and HT. Treatment procedure was classified using the above sequence to establish priority when women reported more than one procedure (21%).
Information on stillbirth, infant death (from birth to 1 year), birth weight, and gestational age were obtained from the Medical Birth Register.15 We defined as perinatal deaths all stillbirths and deaths occurring within the first 7 days after birth. According to the definition used in Denmark during the study period, stillbirth was a fetal death occurring from 28 completed weeks of gestation. A priori, we defined SGA as the lowest 5% of the birth weight distribution by gestational age and sex, using all liveborn singletons from the Danish National Birth Cohort as the standard.
We had three sources to determine gestational age: last menstrual period reported by the participants in the consent form, expected date of delivery reported in a later interview during pregnancy (at about 30 weeks of gestation), and the gestational age recorded in the Medical Birth Register (the latter two estimates are usually based on ultrasound measures). To avoid systematic misclassification of gestational age by ultrasound measurements if early fetal growth is influenced by the exposure under study,16 we used the estimate based on last menstrual period if it agreed within 2 weeks with either the expected date of delivery or the gestational age recorded in the register. If the difference was more than 2 weeks, we used the one that fit best with the observed birth weight.17 The proportion of gestational age determined by the last menstrual period was 95.4% among fertile couples, 94.8% among infertile couples conceiving spontaneously, and 96.8% among infertile couples receiving treatment.
We estimated odds ratios (ORs) and 95% confidence intervals (CIs) of SGA, perinatal death, and infant death (8th to 365th days) by means of logistic regression models in STATA 9.1 (StataCorp, College Station, TX). By including in the model categories of time to pregnancy as a continuous variable, we tested for a trend between time to pregnancy and outcomes. Potential confounders included maternal age at conception, height, prepregnancy body mass index, smoking, coffee consumption, alcohol intake, occupation, and history of hypertension and metabolic disorders. Maternal age was included in the final logistic models, together with potential confounders that produced a change in estimate of more than 10%. Since parity is an intermediate variable between infertility and pregnancy outcomes, but a potential confounder for the relation between infertility treatment and the outcomes, we estimated the effect of infertility and treatment with and without adjustment for parity. By including an interaction term for parity and infertility (and treatment), we also examined possible interactions by using the Wald test (ratio of the maximum likelihood estimate of the parameter to an estimate of its standard error, which follows a standard normal distribution)18 with a level of significance set at .05.
Constitutionally small fetuses may be wrongly classified as growth-restricted by our criterion. Conversely, some growth-restricted babies will fall outside our criterion. We further examined birth weight as a continuous variable, estimating the effect of infertility and treatment through multiple linear regression models, with and without parity, with birth weight as the dependent variable. We used as predictors gestational age, maternal age at conception, height, prepregnancy body mass index, smoking, coffee consumption, alcohol intake, occupation, sex of child, and a variable indicating the exposure status as independent variables. We modeled gestational age as a third-degree polynomial and the other covariates as categorical variables (as reported in Table 1). The regression model with (and without) parity accounted for 41% (and 40%) of the birth weight variance. By restricting the analyses to only infertile couples, we estimated the effect of treatment. We estimated the association between time to pregnancy and birth weight among couples without treatment.
The establishment of the Danish National Birth Cohort was approved by the regional scientific ethics committee (reference number KF 01-471/94). The Danish Data Protection Agency granted authorization for the implementation of the project (reference number 2005-41-5488), and the Danish National Birth Cohort Steering Committee granted authorization for the use of data from the Danish National Birth Cohort (reference number 2005–10).
Characteristics of the study population are shown in Table 1. Among infertile couples, regardless of treatment, mothers were more likely to be older, be primiparous, and have a higher body mass index.
The study population included 326 perinatal deaths. Compared with singletons born of fertile couples, those born of infertile couples, regardless of treatment, had a higher perinatal mortality (Table 2, reference Group A: crude ORs). After adjustment for maternal age and prepregnancy body mass index, all the estimates were attenuated and no longer statistically significant (Table 2, reference Group A: first column of adjusted ORs). Among infertile couples, singletons conceived after treatment had a perinatal mortality similar to that of naturally conceived babies, regardless of adjustment for maternal age and body mass index (Table 2, reference Group B: crude and first column of adjusted ORs).
Among naturally conceived newborns, we saw no significant trend between time to pregnancy and perinatal mortality (Table 3, perinatal deaths: crude and first column of adjusted ORs).
Overall, 97 infants died between the 8th and the 365th day after birth: 77 born of fertile couples, and 9 and 11 born of infertile couples conceiving without and with infertility treatment, respectively. Compared with infants of fertile couples, the ORs adjusted for maternal age and body mass index were 1.07 (95% CI 0.53–2.14) and 1.83 (95% CI 0.96–3.49) for those born of infertile couples conceiving without and with infertility treatment, respectively. The small number of cases did not permit analysis of infant death by treatment procedures. Among untreated couples, we saw no trend between time to pregnancy and death between the 8th and 365th day of life (Table 3, infant deaths: crude and first column of adjusted ORs).
Further adjustment for parity did not substantially change the above estimates (Tables 2 and 3). We saw no interactions in the risk of perinatal and infant death between parity and categories of infertility or time to pregnancy (P=.48∼.88, Wald test for slope for categories of time to pregnancy as a continuous variable). Restricting the analyses to primiparas yielded estimates similar to those adjusted for parity in Tables 2 and 3. Our analyses included neonates born alive at any gestation but, during the time of the study, delivery of a dead fetus before the 28th week was considered a miscarriage and thus not recorded in the Birth Register. However, the estimates did not change after excluding the 115 neonates born alive before 28 weeks.
Compared with singletons born of fertile couples, those born of infertile couples had an increased risk of SGA, regardless of treatment or adjustment (Table 4, reference Group A: crude and first column of adjusted ORs). Compared with singletons born of infertile couples conceiving naturally, those born after treatment had a slightly higher risk of SGA (Table 4, reference Group B: crude and first column of adjusted ORs).
Analyses among naturally conceived babies showed a statistically significant trend between time to pregnancy and the risk of SGA (Table 3, small-for-gestational-age: crude and first column of adjusted ORs).
Adjustment for parity resulted in attenuated estimates but a similar conclusion, except that the overall difference between treated and untreated couples was no longer statistically significant (Tables 3 and 4). We detected no interaction in the risk of SGA between categories of infertility or time to pregnancy and parity (P=.62∼.87, Wald test for slope for categories of time to pregnancy as a continuous variable), and restricting the analyses to primiparas resulted in estimates similar to those adjusted for parity in Tables 3 and 4.
Before adjustment, we saw a small shift in the birth weight distribution of neonates born to infertile couples, regardless of treatment (Fig. 1). The linear regression model not including parity suggested differences in birth weight of −42.7 grams (95% CI −54.9 to −30.4) and −96.1 grams (95% CI −110.1 to −82.1) among neonates born of infertile couples conceiving naturally and after treatment, respectively. Including parity in the model suggested a more modest reduction in birth weight (−21.2 grams [95% CI −33.3 to −9.1] and −34.5 grams [95% CI: −48.6 to −20.4] for neonates born of infertile couples conceiving naturally and after infertility treatment, respectively). All the following estimates were adjusted for parity. Restricting to infertile couples resulted in a small borderline significant effect of treatment (−17.6 grams [95% CI −36.1 to 0.9]). Compared with neonates born of infertile couples conceiving naturally, neonates born after IUI (−37.8 grams [95% CI −65.2 to −10.5]) and HT (−35.8 grams [95% CI −65.4 to −6.3]) were smaller, but not neonates born after ICSI (−3.6 grams [95% CI −50.0 to 42.9]) or IVF (11.9 grams [95% CI −14.6 to 38.5]).
Figure 2 shows the unadjusted birth weight distributions as a function of time to pregnancy. After adjustment and compared with neonates with time to pregnancy of 0–2 months, the differences in birth weight were 0.6 (95% CI −8.5 to 9.7), −13.2 (95% CI −23.3 to −3.1), and −24.3 grams (95% CI −36.9 to −11.7) for neonates with time to pregnancy of 3–5, 6–12, and more than 12 months, respectively.
Singletons born of infertile couples were slightly more likely to be small-for-gestation than singletons born of fertile couples, and time to pregnancy correlated with the risk among spontaneously conceived babies. Estimates were similar between treated and untreated infertile couples, suggesting that part of the increased risk may be due to the underlying infertility or its determinants, rather than to treatment. The crude increased relative risk of perinatal mortality among babies born of infertile couples was in part explained by older maternal age, higher body mass index, and primiparity.
Infertile couples not receiving treatment may have a less severe form of infertility than infertile couples recurring to treatment. However, use of this group as the reference for assessing the side effects of treatment reduces the potential for confounding by indication.
In this population-based cohort, information on birth outcome was collected from national registers and was virtually complete. Although time to pregnancy and treatment were retrospectively collected during pregnancy, they were reported before the occurrence of the outcomes under study. Thus, differential recall and differential loss to follow-up are not likely sources of bias in this study. Although participants in the Danish National Birth Cohort were somewhat healthier than mothers in the general population (eg, fewer smokers), the effect of nonparticipation on our estimates is expected to be small, as showed in a validation study.19 The 21,771 singletons excluded from this analysis because of unplanned and partly planned pregnancies had a risk of perinatal and infant (8–365 days) death (5.6 and 1.6 per 1,000 births, respectively) comparable to that of singletons born of fertile couples. The risk of SGA was slightly higher (5.1%), but not significantly different after adjustment for maternal age, parity, and smoking (OR 1.04, 95% CI 0.97–1.12).
The validity of time to pregnancy reported by women has been found to be high even after more than a decade.20 Participants in this cohort had to recall the time to pregnancy of the current pregnancy, thus with a short time lag. Excluding pregnancies reported as conceived right away (labeled as a time to pregnancy of 0) did not change our results (data not shown). Long or irregular cycles may result in a time to pregnancy of longer than 12 months, which could cause bias if associated with the studied outcomes. However, excluding all women reporting long (more than 33 days) or irregular cycles (15%) resulted in slightly stronger estimates (with a change of less than 20%).
Women reporting a time to pregnancy of less than 6 months were not asked about treatment for infertility, but we expect only a small proportion of treated couples to have been wrongly classified as untreated. About 45% of women with a time to pregnancy longer than12 months reported treatment, whereas only 7% of women with a time to pregnancy of 6–12 months reported treatment. Treatment type was also self-reported, but most women would be aware of the treatment they received, especially since the time period of recall was relatively short. Among women reporting more than one procedure, we prioritized the treatment based on an a priori assessment of potential risk to the baby. Analyses restricted to women who reported only one procedure (79%) yielded estimates similar to those presented. Additionally, our previous findings for twinning and congenital malformations in relation to infertility treatment suggest that the reporting of treatment procedures was, overall, reasonably accurate.10,21
The rates of stillbirth and infant mortality in our study population (0.3% and 0.4%) were comparable with the national figures between 1997 and 2001 (0.3–0.4% and 0.4%–0.5%, respectively) (homepage of the Danish Society of Obstetrics and Gynecology: http://www.dsog.dk/). We used SGA defined as the lowest 5% of the distribution of birth weight by sex and gestational age, instead of 10%, to increase specificity. Babies defined as SGA in this study had a rate of infant death (0–365 days of life) of 1.90%, compared with 0.28% in non-SGA babies. Among the several options to determine gestational age, we used the one that we considered the best estimate. Using only gestational age recorded in the Medical Birth Register (likely based on ultrasound estimates) yielded similar results (data not shown).
In a case-control study, Draper and colleagues6 found that history of infertility and untreated infertility were associated with an increased risk of perinatal mortality, but recall bias could not be ruled out. Some authors have reported an increased risk of perinatal mortality among children born after ART compared with naturally conceived children.1,2,22 Our findings suggest that the increased risk may be due to the characteristics of infertile women, consistent with two large studies showing no significant difference between singletons born after IVF or ovarian stimulation and population controls, after adjustment for covariates including maternal age and parity.23,24 Due to the limited number of cases in our study, we cannot, however, rule out a small to moderate increased risk of perinatal mortality.
Several studies reported an association between a long time to pregnancy and low birth weight or preterm birth.4,7–9 One, however, reported no association with these outcomes and SGA.25 Participants in the study were enrolled between 1959 and 1966, a time period characterized by limited availability of effective contraception. The small proportion of pregnancies with a known time to pregnancy (15%) may reflect a selected low-risk group. Our finding of a higher risk of SGA in singletons born after infertility treatment is in line with previous studies reporting a 40–60% excess risk of SGA in ART children.1,2,22,26 The increased risk, however, may be at least partly due to the underlying infertility or, more likely, to its determinants, although some evidence suggests otherwise.27 Intrauterine insemination was associated with a slightly increased risk of SGA, which may be explained by the indication for treatment (type or severity of infertility. In vitro fertilization, including ICSI, usually involves selection of embryos, which may reduce the excess risk of SGA induced by treatment. When analyzing the entire birth weight distribution, however, we saw a small effect, possibly due to residual confounding. If our results reflect a true association, it is possible that only specific types (or causes) of infertility are associated with elevated risk. Studies with detailed clinical diagnoses are necessary for understanding the mechanisms by which infertility may correlate with fetal growth disruption.
Using a segment of the Danish National Birth Cohort including births between 1998 and 2001, we previously reported an increased risk of neonatal death with increasing time to pregnancy among primiparas.5 However, we could not replicate this finding with the data for the whole period, although the estimates were in the same direction. Compared with primiparas with a time to pregnancy of 0–2 months, we found adjusted ORs of 1.21 (95% CI 0.63–2.35) and 1.46 (95% CI 0.76–2.79) for untreated and treated infertile couples, respectively. The discrepancy with the previous analysis may have resulted from a higher rate of neonatal deaths among women with a time to pregnancy of 0–2 months (used as the reference category) in 2002–2003 compared with the previous years. The number of neonatal deaths was too small to further examine this difference. On the other hand, we still saw an elevated risk of preterm birth among infertile women.4 Compared with primiparas with a time to pregnancy of 0–2 months, we saw adjusted ORs of 1.26 (95% CI 1.05–1.52) and 1.49 (95% CI 1.23–1.81) for untreated and treated infertile couples, respectively.
Infertility is a result of environmental/lifestyle exposures and/or of genetic predisposition, and several of these factors, in women as well as men, may also affect fetal growth. Male factors may be more present in some groups, such as couples treated with ICSI. Maternal factors may, however, play a more direct role during pregnancy. In our data, primiparity, advanced maternal age, and smoking were associated with a higher risk of SGA. The association between time to pregnancy and SGA was reduced after adjustment for these factors. Advanced maternal age, inherently linked to ovarian aging and decreased fertility,28 may explain part of the increased risk of perinatal mortality among infertile couples. Maternal chronic conditions such as preexisting hypertension or metabolic diseases did not affect our estimates, consistent with previous findings.25
A long time to pregnancy may be associated with reduced fetal growth. The reported increased risk of SGA in singletons born after infertility treatment could be, in part, due to the underlying infertility or to its determinants. Treatment procedures per se may have little effect on fetal growth in singletons.
1. 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.
2. Jackson RA, Gibson KA, Wu YW, Croughan MS. Perinatal outcomes in singletons following in vitro fertilization: a meta-analysis. Obstet Gynecol 2004;103:551–63.
3. 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;20:328–38.
4. Basso O, Baird DD. Infertility and preterm delivery, birthweight, and Caesarean section: a study within the Danish National Birth Cohort. Hum Reprod 2003;18:2478–84.
5. Basso O, Olsen J. Subfecundity and neonatal mortality: longitudinal study within the Danish national birth cohort. BMJ 2005;330:393–4.
6. Draper ES, Kurinczuk JJ, Abrams KR, Clarke M. Assessment of separate contributions to perinatal mortality of infertility history and treatment: a case-control analysis. Lancet 1999;353:1746–9.
7. Henriksen TB, Baird DD, Olsen J, Hedegaard M, Secher NJ, Wilcox AJ. Time to pregnancy and preterm delivery. Obstet Gynecol 1997;89:594–9.
8. Joffe M, Li Z. Association of time to pregnancy and the outcome of pregnancy. Fertil Steril 1994;62:71–5.
9. Williams MA, Goldman MB, Mittendorf R, Monson RR. Subfertility and the risk of low birth weight. Fertil Steril 1991;56:668–71.
10. Zhu JL, Basso O, Obel C, Bille C, Olsen J. Infertility, infertility treatment, and congenital malformations: Danish national birth cohort. BMJ 2006;333:679.
11. Hay WW. Catz CS, Grave GD, Yaffe SJ. Workshop summary: fetal growth: its regulation and disorders. Pediatrics 1997;99:585–91.
12. Resnik R. Intrauterine growth restriction. Obstet Gynecol 2002;99:490–6.
13. Tan TY, Yeo GS. Intrauterine growth restriction. Curr Opin Obstet Gynecol 2005;17:135–42.
14. Olsen J, Melbye M, Olsen SF, Sørensen TI, Aaby P, Andersen AM, et al. The Danish National Birth Cohort-its background, structure and aim. Scand J Public Health 2001;29:300–7.
15. Knudsen LB, Olsen J. The Danish Medical Birth Registry. Dan Med Bull 1998;45:320–3.
16. Henriksen TB, Wilcox AJ, Hedegaard M, Secher NJ. Bias in studies of preterm and postterm delivery due to ultrasound assessment of gestational age. Epidemiology 1995;6:533–7.
17. Zhu JL, Hjollund NH, Olsen J. Shift work, duration of pregnancy, and birth weight: the National Birth Cohort in Denmark. Am J Obstet Gynecol 2004;191:285–91.
18. Hosmer DW, Lemeshow S. Applied logistic regression.
New York (NY): John Wiley & Sons; 2000.
19. Nohr EA, Frydenberg M, Henriksen TB, Olsen J. Does low participation in cohort studies induce bias? Epidemiology 2006;17:413–8.
20. Joffe M, Villard L, Li Z, Plowman R, Vessey M. A time to pregnancy questionnaire designed for long term recall: validity in Oxford, England. J Epidemiol Community Health 1995;49:314–9.
21. Zhu JL, Basso O, Obel C, Christensen K, Olsen J. Infertility, infertility treatment and twinning: the Danish National Birth Cohort. Hum Reprod 2007;22:1086–90.
22. McDonald SD, Murphy K, Beyene J, Ohlsson A. Perinatel outcomes of singleton pregnancies achieved by in vitro fertilization: a systematic review and meta-analysis. J Obstet Gynaecol Can 2005;27:449–59.
23. Ombelet W, Martens G, De Sutter P, Gerris J, Bosmans E, Ruyssinck G, et al. Perinatal outcome of 12,021 singleton and 3108 twin births after non-IVF-assisted reproduction: a cohort study. Hum Reprod 2006;21:1025–32.
24. Klemetti R, Sevón T, Gissler M, Hemminki E. Health of children born as a result of in vitro fertilization. Pediatrics 2006;118:1819–27.
25. Cooney MA, Buck Louis GM, Sun W, Rice MM, Klebanoff MA. Is conception delay a risk factor for reduced gestation or birthweight? Paediatr Perinat Epidemiol 2006;20:201–9.
26. Westergaard HB, Johansen AM, Erb K, Andersen AN. Danish National IVF Registry 1994 and 1995. Treatment, pregnancy outcome and complications during pregnancy. Acta Obstet Gynecol Scand 2000;79:384–9.
27. Kapiteijn K, de Bruijn CS, de Boer E, de Craen AJ, Burger CW, van Leeuwen FE, et al. Does subfertility explain the risk of poor perinatal outcome after IVF and ovarian hyperstimulation? Hum Reprod 2006;21:3228–34.
28. Swanton A, Child T. Reproduction and ovarian ageing. J Br Menopause Soc 2005;11:126–31.