Early View
RESEARCH ARTICLE
Open Access

Fetal growth after fresh and frozen embryo transfer and natural conception: A population-based register study

Mårten Ageheim

Corresponding Author

Mårten Ageheim

Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden

Correspondence

Mårten Ageheim, Women's and Children's Health, Uppsala University, Akademiska University Hospital, Uppsala SE-751 85, Sweden.

Email: [email protected]

Search for more papers by this author
Alkistis Skalkidou

Alkistis Skalkidou

Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden

Search for more papers by this author
Eva Bergman

Eva Bergman

Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden

Search for more papers by this author
Stavros Iliadis

Stavros Iliadis

Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden

Search for more papers by this author
Erik Lampa

Erik Lampa

Department of Medical Sciences, Uppsala University, Uppsala, Sweden

Search for more papers by this author
Linda Lindström

Linda Lindström

Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden

Search for more papers by this author
Anna Sara Oberg

Anna Sara Oberg

Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden

Search for more papers by this author
First published: 13 February 2024

Linda Lindström and Anna Sara Oberg contributed equally to the work.

Abstract

Objective

To investigate fetal growth trajectories and risks of small and large for gestational age (SGA and LGA), and macrosomia in pregnancies after fresh and frozen embryo transfer (ET), and natural conception (NC).

Design

Longitudinal population-based cohort study.

Setting

Swedish national registers.

Population

A total of 196 008 singleton pregnancies between 2013 and 2017.

Methods

Of all singleton pregnancies resulting in live births in the Swedish Pregnancy Register, 10 970 fresh ET, 6520 frozen ET, and 178 518 NC pregnancies with ultrasound data were included. A general least squares model was used to examine the effect of fresh or frozen ET on fetal growth while adjusting for confounders.

Main Outcome Measures

Fetal growth velocity. SGA, LGA and macrosomia.

Results

At 120 days, fetal weights were lower in fresh ET pregnancies compared with NC pregnancies. Thereafter fresh ET as well as FET fetuses had higher fetal weights than NC fetuses, with no differences between themselves until the second trimester. From 210 days, FET fetuses were heavier than fresh ET fetuses, whereas fresh ET fetuses had lower fetal weights than NC fetuses from 245 days. After fresh ET, SGA was more frequent, whereas LGA and macrosomia were less frequent, than after FET.

Conclusions

This study gives new insights into the differences in fetal growth dynamics between fresh and frozen ET and NC pregnancies. Clinically relevant differences in proportions of SGA, LGA and macrosomia were observed.

1 INTRODUCTION

The use of in vitro fertilisation (IVF) is steadily increasing globally. In Sweden, nearly 5% of children born today originate from IVF pregnancies. Previous studies have reported higher risks of preterm birth and small for gestational age (SGA) in fresh embryo transfer (ET) pregnancies.1, 2 Many SGA fetuses are healthy and genetically small, whereas others suffer from growth restriction, a complication accounting for a large proportion of morbidity and mortality.3, 4 With the improvement of live birth rates after frozen embryo transfer (FET), its use has increased, and in 2019 FET accounted for 48% of treatment cycles leading to ET.5 However, FET is associated with elevated risk of LGA and macrosomia.6-10 Programmed FET cycles, where a corpus luteum is absent, seem to be at particular risk of accelerated fetal growth and hypertensive disorders in pregnancy.11 Being born large for gestational age (LGA) is also associated with increased short- and long-term risks.12 Similar risks are seen with macrosomia, in Sweden defined as birthweight more than 4500 g.13, 14

Although a number of studies have shown differences in birthweight, only a few have studied intrauterine fetal growth velocity.15-17 Little is known about when during gestation these differences develop and if they are independent of confounding factors. These aspects are important for understanding the underlying mechanisms and possible long-term effects.

1.1 Objective

The aim of the present study was to compare fetal growth between pregnancies after fresh ET, FET and natural conception (NC), based on population-based Swedish data on ultrasound measurements throughout pregnancy, as well as birthweights, while adjusting for possible confounders. Secondary outcomes were SGA, LGA and macrosomia.

2 METHODS

2.1 Included sources and study population

This was a longitudinal population-based cohort study based on linkage of several national registers. The target study population was all singleton pregnancies leading to delivery from 22+0 gestational weeks, recorded in the Swedish Pregnancy Register between 2013 and 2017. The register had 90% national coverage, except for the ultrasound data where three regions were missing and therefore the coverage was lower. Data on maternal characteristics and neonatal complications was retrieved from the Medical Birth Register with 97%–99% national coverage. Detailed data on IVF came from the National Quality Register for Assisted Reproduction (Q-IVF), with a national coverage close to 100%. Maternal history of diagnoses was retrieved from the National Patient Register, and education status and country of birth from the Total Population Register. The Prescribed Drug Register provided the anatomic therapeutic chemical code for estrogens (G03C). All registers were linked using the unique personal identity number assigned to all residents in Sweden.

In clinical practice, IVF pregnancies are dated by calculating the expected date of delivery (EDD) by adding 266 days to the date of ET and subtracting the number of culture days. In addition to this EDD (retrieved from the Pregnancy Register), we had access to recordings of ET date and number of culture days (retrieved from the Q-IVF register). Pregnancies where the ET date was missing were excluded (n = 11). Pregnancies with missing embryo culture days were excluded only if the EDD in the antenatal records minus 266 days was less than 2 days or more than 6 days (n = 22). Pregnancies in which the calculated EDD differed from the registered EDD by more than 7 days were excluded (n = 117).

To avoid possible normalisation of early growth differences, we calculated the EDD for NC pregnancies from the last menstrual period (LMP) by adding 280 days. We also calculated the estimated gestational age according to ultrasound for NC pregnancies from the first available measurement of crown–rump length (CRL), 4–85 mm, or biparietal diameter, 21–27 mm, using formulae currently recommended by the Swedish Society of Obstetrics and Gynaecology.18 If LMP date was missing or ultrasound dating and LMP-dating differed by more than 14 days, the pregnancy was excluded from the cohort (n = 19 286). Due to less accuracy in second-trimester dating, and the fact that some pregnancies lacking first-trimester ultrasound only would have two available measurements, i.e. the second-trimester screening and birthweight, we chose to only include NC pregnancies with ultrasound data from the first trimester.

2.2 Exposure and outcomes

Three exposure groups were considered; fresh ET, FET and NC. Conventional IVF as well as intracytoplasmic sperm injection were included in the fresh ET and FET groups. Both single and double embryo transfers were included if the first-trimester ultrasound showed only one fetus. FET where estrogen was prescribed within 70 days before ET were regarded as programmed cycles.

The main outcome was fetal weight. Estimated fetal weight (EFW) was calculated using the formula currently recommended in Sweden.18, 19 A combined fetal weight variable based on EFW and birthweight was evaluated. Using the recommended Swedish fetal growth reference,20 the weight deviance was calculated and expressed as percentages of expected fetal weight for gestational age. In Swedish clinical practice, SGA is defined as weight ≤−2 standard deviations (SD) from the mean, while LGA is defined as weight ≥2 SD from the mean. As secondary outcomes, differences in risks of SGA, LGA and macrosomia were evaluated. CRL was also analysed, because it reflects early fetal growth.

2.3 Confounders

To establish which confounders to adjust for, a causal diagram of the association between mode of conception and fetal growth was created using a directed acyclic graph with all possible common causes of the two identified from subject-matter knowledge. Using DAGitty v3.0, the minimal sufficient adjustment set for estimating the total causal effect of mode of conception on fetal growth trajectory was identified and included maternal age, body mass index, parity, education, smoking, polycystic ovary syndrome, infertility, embryo culture days and oocyte donation (Figure S1).

2.4 Statistical analyses

Longitudinal analyses were performed to examine the effect of fresh ET and FET on fetal weight, where NC was treated as the reference category. Comparisons were also performed directly between fresh ET and FET. Adjustments were made for potential confounding factors identified in the directed acyclic graph, noting that embryo culture days and oocyte donation were only relevant in the direct comparison of fresh ET and FET.

Associations between mode of conception and fetal weight were assessed using generalised least squares models, which are linear regression models allowing non-independent data. An autoregressive correlation of order 1 was assumed for the repeated measurements, meaning that outcome values closer in time are expected to be more correlated than outcome values further apart. All outcomes were log-transformed before model fitting as their variance increased with the mean values. Continuous variables were modelled using restricted cubic splines, with five knots placed at the 5th, 27.5th, 50th, 72.5th and 95th centiles of each variable's marginal distribution. Due to the much skewed distribution of gestational age, the knots were placed at the centiles of the cube-root transformed gestational age. All figures were based on predicted values of the outcome, varying gestational age while holding all confounder variables constant at their median values (continuous) or at their most frequent category (categorical).

To further exclude confounding influence related to infertility, a sensitivity analysis was performed, where fetal weights of fresh ET and FET pregnancies were compared with NC pregnancies among couples with known infertility (either a diagnosis of female infertility, or a maternal self-report at antenatal care, as taking more than 1 year to conceive).

In another sensitivity analysis, the fetal weight comparison of fresh ET and FET was performed in cleavage and blastocyst-stage transfers separately, to assess whether potential differences depended on embryo culture days.

To evaluate the possible effect of absence of corpus luteum on fetal growth we performed a sensitivity analysis where programmed FET cycles were compared with non-programmed FET cycles.

Crown–rump length was evaluated in the same manner as fetal weight, except for the sensitivity analyses.

Odds ratios (OR) for SGA, LGA and macrosomia, for fresh ET versus FET, as well as gestational hypertension in programmed versus non-programmed FET, were estimated using a multivariable logistic regression with the same set of adjustment variables as the main analysis.

Data management was performed using SAS software, Version 9.4 of the SAS System for Windows.

Statistical analyses were performed with R version 4.0.5. R Core Team (2021).

2.5 Patient involvement

Being a retrospective register-based study, there was no patient or public involvement.

3 RESULTS

The target cohort consisted of 501 545 pregnancies and the exclusion process towards the final study population is described in detail in Figure 1. The majority of exclusions were made because of missing ultrasound data. Pregnancies with EDD later than last date of the data extraction were excluded as they would only represent preterm births, in which growth aberrations are more common.

Details are in the caption following the image
Flow chart leading to the final study population consisting of three groups, fresh embryo transfer (fresh ET), frozen embryo transfer (FET) and natural conception (NC). Analytic sample is a complete case population. 1Last menstrual period, 2estimated fetal weight, 3standard deviations, 4crown–rump length, 5biparietal diameter.

The final study population consisted of 196 008 pregnancies, 39.1% of the target population, 10 970 fresh ET, 6520 FET and 178 518 NC. The distribution of fetal weight observations is described in Figure S2.

3.1 Background characteristics and pregnancy outcomes

Detailed descriptions of maternal background characteristics, infertility treatment and birth outcomes are provided in Table 1, where fresh ET, FET, NC and the subgroup of NC pregnancies of couples with infertility are presented.

TABLE 1. Background characteristics of the study population, as well as pregnancy and offspring outcomes in natural conception (NC), all and restricted to couples with infertility, fresh embryo transfer (fresh ET) and frozen embryo transfer (FET) pregnancies.
NC NC, infertility Fresh ET FET
(n = 178 518) (n = 17 669) (n = 10 970) (n = 6520)
Maternal characteristics
Maternal age, years, median (IQR) 32 (28–36) 35 (31–38) 34 (30–37) 34 (31–37)
Primipara, n (%) 74 068 (41.5%) 7334 (41.5%) 7954 (72.8%) 3694 (56.7%)
BMI <18.5 kg/m2, n (%) 4240 (2.4%) 347 (2.0%) 207 (1.9%) 129 (2.0%)
BMI ≥30 kg/m2, n (%) 19 207 (10.8%) 2321 (13.1%) 1071 (9.8%) 572 (8.8%)
Missing BMI 4451(2.5%) 473 (2.7%) 561 (5.1%) 342 (5.3%)
Maternal height, m, median (IQR) 1.67 (1.62–1.71) 1.67 (1.62–1.71) 1.67 (1.63–1.72) 1.67 (1.63–1.71)
Missing maternal height 1520 (0.9%) 154 (0.9%) 423 (3.9%) 236 (3.6%)
Higher education ≥2 years, n (%) 106 193 (59.5%) 10 886 (62.5%) 7271 (66.3%) 4476 (68.7%)
Missing education level 6378 (3.6%) 383 (2.2%) 223 (2.0%) 75 (1.2%)
Origin from Nordic country, n (%) 139 572 (78.2%) 13 173 (74.5%) 8817 (80.4%) 5240 (80.4%)
Polycystic ovary syndrome, n (%) 3762 (2.1%) 1675 (9.5%) 773 (7.1%) 546 (8.4%)
Smoking first antenatal visit, n (%) 6391 (3.6%) 523 (3.0%) 141 (1.3%) 64 (1.0%)
Missing smoking 5449 (3.1%) 510 (2.9%) 591 (5.4%) 388 (6.0%)
Chronic hypertension, n (%) 1218 (0.7%) 184 (1.0%) 51 (0.5%) 63 (1.0%)
Pregestational diabetes mellitus, n (%) 1396 (0.8%) 166 (0.9%) 64 (0.6%) 58 (0.9%)
Infertility, n (%) 17 669 (9.9%) 17 423 (100%) 10 119 (92.2%) 6029 (92.5%)
Infertility treatment characteristics
Sperm donation, n (%) 295 (0.2%) 254 (1.4%) 452 (4.1%) 311 (4.8%)
Oocyte donation, n (%) NA NA 163 (1.5%) 145 (2.2%)
ICSI, n (%) NA NA 5169 (47.1%) 2787 (42.8%)
Embryo culture days 2–3, n (%) NA NA 8244 (75.2%) 1870 (28.7%)
Embryo culture days 5–6, n (%) NA NA 2541 (23.2%) 4500 (69.0%)
Missing embryo culture days NA NA 67 (0.6%) 47 (0.7%)
Single embryo transfer, n (%) NA NA 9436 (86.0%) 6280 (96.4%)
Double embryo transfer, n (%) NA NA 1533 (14.0%) 237 (3.6%)
Pregnancy and birth outcomes
Ultrasound examinations, n 484 685 52 581 26 709 16 861
Gestational hypertension, n (%) 5767 (3.2%) 688 (3.9%) 473 (4.3%) 350 (5.4%)
Gestational diabetes, n (%) 2557 (1.4%) 396 (2.2%) 187 (1.7%) 100 (1.5%)
Pregnancy duration, days, median (IQR) 281 (274–287) 280 (272–286) 279 (272–286) 280 (273–287)
Preterm delivery <37 weeks, n (%) 7091 (4.0%) 919 (5.2%) 687 (6.3%) 335 (5.1%)
Caesarean section, n (%) 33 330 (18.7%) 4530 (25.6%) 2712 (24.8%) 1766 (27.1%)
Male sex, n (%) 91 853 (51.5%) 9140 (51.7%) 5604 (51.3%) 3410 (52.4%)
Birthweight, g, median (IQR) 3560 (3420–3890) 3530 (3205–3865) 3445 (3125–3780) 3600 (3272–3935)
Birth length, cm, median (IQR) 50.0 (49–52) 50.0 (49–52) 50.0 (49–52) 51.0 (49–52)
Missing birth length 790 (0.4%) 107 (0.6%) 99 (0.9%) 34 (0.5%)
Birth head circumference, cm, median (IQR) 35.0 (34–36) 35.0 (34–36) 35.0 (34–36) 35.0 (34–36)
Missing birth head circumference 1674 (0.9%) 182 (1.1%) 140 (1.3%) 69 (1.1%)
Birthweight deviance, mediana (IQR) −0.4 (−8.2 to 8.0) −0.4% (−8.4 to 8.0) −2.4% (−10.6 to 5.9) 1.3% (−7.1 to 9.8)
SGA (birthweight <−2 SD), n (%) 5638 (3.2%) 645 (3.7%) 557 (5.1%) 171 (2.6%)
LGA (birthweight >2 SD), n (%) 8545 (4.8%) 941 (5.3%) 414 (3.8%) 424 (6.5%)
Macrosomia (birthweight >4500 g), n (%) 6119 (3.4%) 596 (3.4%) 265 (2.4%) 291 (4.5%)
  • Note: Infertility was defined as either infertility diagnosis, or maternal report at enrolment to antenatal care of taking more than 1 year to conceive.
  • Abbreviations: BMI, body mass index; ICSI, intracytoplasmic sperm injection; IQR, interquartile range; LGA, large for gestational age; NA, not applicable; SD, standard deviation; SGA, small for gestational age.
  • a Compared with the recommended Swedish growth reference.

The IVF groups were different from NC in many aspects, i.e. higher proportions of advanced maternal age, primiparae, polycystic ovary syndrome and infertility, as well as including fewer smokers. Between the two IVF groups the differences were small, the most apparent being more primiparae in the fresh ET group. Compared with the initial NC group, before the exclusion of pregnancies lacking a first-trimester ultrasound, the final NC group had a greater resemblance to the two IVF groups. Details are given in Table S1.

The risk of SGA was higher after fresh ET compared with FET (adjusted OR [aOR] 1.65, 95% confidence interval [CI]1.33–2.04), whereas the risk of LGA and macrosomia was lower (aOR 0.66, 95% CI 0.56–0.79 and aOR 0.59, 95% CI 0.48–0.73, respectively).

The risk of gestational hypertension was higher after programmed, compared with non-programmed FET (OR 1.75, 95% CI 1.26–2.44) (Table S2).

3.2 Adjusted relative fetal weight

At 120 days, fetal weights were lower in fresh ET compared with NC. Thereafter fresh ET as well as FET had higher fetal weights than NC, with no differences between themselves until the second trimester. From 210 days, FET fetuses were heavier than fresh ET fetuses, whereas fresh ET fetuses had lower fetal weights than NC fetuses from 245 days. The predicted fetal weight at 40 gestational weeks was lower for fresh ET compared with NC (−1.1%, 95% CI −1.4 to −0.8), and for FET it was higher compared with NC (2.1%, 95% CI 1.7–2.6) (Figure 2, Table 2).

Details are in the caption following the image
Predicted values of the outcome, varying gestational age while holding all confounder variables constant at their median values (continuous) or at their most frequent category (categorical). Conditional effects of gestational age displayed as adjusted relative differences in combined mean fetal weight and birthweight in fresh embryo transfer (fresh ET) and frozen embryo transfer (FET) pregnancies in relation to natural conception (NC), percentages on a logarithmic scale. Model with adjustments for maternal age, body mass index, parity, education, smoking, polycystic ovary syndrome and infertility.
TABLE 2. Predicted values of the outcome, varying gestational age while holding all confounder variables constant at their median values (continuous) or at their most frequent category (categorical).
Day FET vs. fresh ET Fresh ET vs. NC FET vs. NC
Contrast % (95% CI) p value Contrast % (95% CI) p value Contrast % (95% CI) p value
120 0.2 (−0.6 to 0.9) 0.664 −0.7 (−1.3 to −0.1) 0.025 −0.4 (−1.2 to 0.3) 0.251
150 0.0 (−0.7 to 0.8) 0.914 9.0 (8.3–9.6) <0.001 8.4 (7.6–9.2) <0.001
180 0.6 (−0.5 to 1.6) 0.298 8.8 (8.0–9.5) <0.001 8.9 (8.0–9.8) <0.001
210 1.9 (1.2–2.5) <0.001 3.1 (2.6–3.6) <0.001 4.7 (4.1–5.3) <0.001
245 3.1 (2.6–3.7) <0.001 −1.8 (−2.2 to −1.4) <0.001 1.1 (0.6–1.6) <0.001
280 3.5 (3.0–4.0) <0.001 −1.1 (−1.4 to −0.8) <0.001 2.1 (1.7–2.6) <0.001
  • Note: Conditional effects of gestational age displayed as adjusted relative differences in combined mean fetal weight and birthweight in fresh embryo transfer (fresh ET) and frozen embryo transfer (FET) pregnancies in relation to natural conception (NC), as percentages with 95% confidence interval (CI). Adjustments were made for maternal age, body mass index, parity, education, smoking, polycystic ovary syndrome and infertility, and when comparing FET with fresh ET adjustments included oocyte donation and embryo culture days.

Sensitivity analysis showed commensurate results when comparing fetal weights of IVF pregnancies (fresh ET and FET) with NC pregnancies in couples with known infertility (Figure S3, Table S3), indicating that findings were not driven by the underlying infertility.

Holding culture days constant while comparing fetal weights of fresh ET and FET pregnancies also produced consistent results among both cleavage-stage (2–3 days) and blastocyst-stage (5–6 days) transfers (Figures S4 and S5, Table S4).

Finally, comparison of programmed (n = 821) and non-programmed FET cycles (n = 5690) did not show any statistically significant differences in fetal weight (Figure S6).

3.3 Crown–rump length

Fresh ET and FET fetuses both had larger CRL compared with NC fetuses, most notable at 40 gestational days, but evident also at 60 and 80 days. At 40 days, FET fetuses had larger CRL than fresh ET fetuses, but after that no differences could be seen between the two IVF groups (Figure S7, Table S5). The median difference between dating according to CRL and ET was −0.9 days for fresh and −1.1 days for frozen, a small overestimation of gestational age with CRL compared with ET. The mean difference between gestational age according to LMP and ultrasound in NC pregnancies was larger if first trimester dating was performed using CRL (1.4 ± 4.0 days, ±SD) than when performed using biparietal diameter (−0.8 ± 3.3 days, ±SD).

4 DISCUSSION

4.1 Main findings

Both fresh ET and FET pregnancies had higher estimated fetal weights in the second trimester, compared with NC pregnancies. Fresh ET showed a slower growth rate compared with FET and NC, resulting in lower mean birthweight, whereas FET fetuses remained larger than NC fetuses with higher mean birthweight. Although the differences were small, the larger proportion of newborns being SGA in fresh ET pregnancies and LGA and macrosomia in FET pregnancies, may have clinical implications. SGA is associated with increased risk of impaired cognitive development, as well as metabolic disorders, such as obesity, type 2 diabetes and hypertension.21-23 Also, LGA is associated with increased risks of obesity, type 2 diabetes and hypertension.12, 14, 24 Epigenetic changes, for instance through DNA methylation, have been suggested as an explanation for metabolic alterations associated with assisted reproduction techniques, but further studies are needed.25-27 Evidence points towards increased risks for placental developmental abnormalities associated with FET, including increased risks for pre-eclampsia.28 A prospective study on uterine artery, however, showed lower pulsatility indices in FET pregnancies compared with fresh ET pregnancies, already in the first trimester, indicating a more favourable placental development.29 Another study showed no differences in the first trimester, but lower pulsatility index in FET pregnancies only in the second and third trimester.30

Birthweight differences are in agreement with a study on growth trajectories after IVF, but no differences were seen in estimated fetal weight.31 Another study on both preterm and term birthweights in a large sample of IVF and spontaneous pregnancies, showed more LGA after FET and more SGA in fresh ET pregnancies in the third trimester.32 These results are supported by a retrospective study monitoring fetal growth in all trimesters. Similar to our findings, all types of assisted reproduction had higher EFW compared with the reference in the second trimester, but in the third trimester only FET pregnancies had a higher EFW.17 Our results are also in line with several previous register-based studies.6-8, 33, 32 In addition to these Nordic studies, where the study by Terho, Pelkonen et al. includes an overlap with our study population, there are several international studies showing similar results.34

The study design does not permit a complete analysis of the underlying mechanisms behind the observed differences. However, several sibling studies with different modes of conception have demonstrated differences between fresh ET and FET, suggesting a possible causal effect of the fertility treatment on fetal growth.35, 36 One sibling study showed, like our study, consistent results in a sensitivity analysis including only blastocyst-stage embryos (5–6 days).37 This is a relevant finding because it has been suggested that the differences could at least partly be attributed to the duration of embryo culture. A meta-analysis showed an elevated risk for LGA after blastocyst transfer, irrespective of fresh ET or FET, but in a sensitivity analysis able to adjust for known confounders, the elevated risk was only evident for FET.38 When we evaluated cleavage-stage and blastocyst-stage transfers separately the differences were still evident, further strengthening the theory that the use of fresh ET or FET influences fetal growth. Another meta-analysis showed a higher incidence of LGA and macrosomia in FET pregnancies after programmed cycle, compared with natural cycle FET.39 Two of the included studies also showed elevated risks of gestational hypertensive disorders after programmed cycles compared with natural cycles, possibly by altered levels of vasoactive hormones due to the lack of corpus luteum.11, 40 We could not confirm differences in fetal growth, but gestational hypertension was more common in programmed compared with non-programmed FET. In our study, CRL was larger for both IVF groups compared with NC, contrasting with a previous study where both fresh ET and FET fetuses had negative CRL z scores in weeks 6 and 8, and fresh ET fetuses were smaller than FET ones, suggesting that growth differences are evident as early as the first trimester.41 Our results, on the other hand, are in line with a retrospective cohort study where all ART types showed positive CRL z scores.17

4.2 Strengths and limitations

Strengths of this study include the population-based design, with linkage of several national registers with high coverage also enabling adjustments for multiple possible confounders. The risk for largely deviating measurements was reduced by strict exclusion criteria. In Sweden, as in many European countries, IVF is mainly publicly funded, and although the influence of socio-economic status on the use of IVF is deemed relatively small, it cannot be excluded.

The nearly complete coverage of the Q-IVF register adds to the validity of the available data and reduces the risk of misclassification. Further, sensitivity analyses were performed to account for possible effects of infertility or its underlying risk factors, and embryo culture stage, hence to some degree isolating the effect on fetal growth from fresh ET and FET. Finally, the effect of programmed FET cycles was also evaluated.

There are factors related to the IVF procedures that the registers did not provide data on, for instance culture media, freezing technique, regional differences and possible changes over time. Data on paternal and oocyte donor characteristics were not available. Adjustments were made for several maternal factors, and while attempts were made to adjust for socio-economic status by level of education and smoking, there are still factors that could contribute to confounding.

The register-based design is a limitation in this study, in that it only concerns routinely collected data. Apart from a second-trimester anomaly scan in gestational week 18–20, there are no other routine scans in the standard Swedish maternity healthcare programme. Additional examinations are typically motivated by pre-existing health issues or complications during pregnancy. Consequently, fetal weights between the second-trimester routine scan and birth may not be representative of the overall population. Although possibly over-representing abnormal fetal growth, these are in most cases expected to occur independently of exposure status because IVF use alone is not an indication for additional scans. It is therefore unlikely that the inclusion of non-routine measurements would have biased our findings.

We cannot exclude the possibility that the exclusion of NC pregnancies lacking data on first-trimester ultrasounds could have biased the comparison with IVF pregnancies. The majority of first-trimester ultrasounds are performed within the screening programme for chromosomal abnormalities, and the criteria for offering this examination vary both regionally and over time. Nevertheless, when comparing the background characteristics and outcome of these and the NC pregnancies lacking a first-trimester ultrasound, the included NC pregnancies were more similar to the IVF groups.

Using LMP to calculate EDD has its weaknesses, but the mean difference between dating on LMP and first ultrasound was small.

Finally, estimation of fetal weight with ultrasound is a potential source of error, and operators were not blinded to pregnancy type, but using birthweight as the final observation adds a reliable last measurement.

4.3 Interpretation

This study provides new knowledge concerning different growth patterns associated with fresh ET and FET. While the differences in fetal weight between fresh ET, FET and NC were small at a group level, they impact the proportion of newborns that fall within the SGA, LGA and macrosomia groups, respectively, and negative long-term effects cannot be ruled out.

5 CONCLUSION

Pregnancies from IVF showed different fetal growth trajectories compared with naturally conceived pregnancies, most apparently through decelerating growth in fresh ET pregnancies resulting in a higher incidence of SGA, while FET presented with higher incidence of LGA and macrosomia.

AUTHOR CONTRIBUTIONS

MA, AS, ASO, LL, SI, EB and EL: Study design, and analysis and interpretation of data; MA, ASO and EL: Data management and statistical analyses; MA: First draft and revision of the manuscript; and MA, AS, ASO, LL, SI, EB and EL: Critical revision of the manuscript for important intellectual content and final approval.

ACKNOWLEDGEMENTS

Uppsala University is acknowledged for funding open access publishing.

    FUNDING INFORMATION

    The study was financed by grants from the Swedish state under the agreement between the Swedish Government and the county councils (ALF-agreement).

    CONFLICT OF INTEREST STATEMENT

    Anna Oberg reported receiving personal fees from Abbott, and Erik Lampa reported receiving personal fees from Biogen outside of the submitted work. All other authors: none declared.

    ETHICS APPROVAL

    The study has been approved by the Regional Ethical Review Board in Stockholm, reference number 2013/1849-31/2 (13 November 2013) and 2018/386-32 (26 February 2018).

    DATA AVAILABILITY STATEMENT

    The data used in this study are national register information. The authors had no special privileges in accessing the data. Dissemination of personal information is regulated by the Swedish Secrecy Act. In accordance with Swedish law, researchers seeking access to individual-level data must apply for permission through a Research Ethics Board (etikprovningsmyndigheten.se) and from the primary owners, the National Board of Health and Welfare (https://www.socialstyrelsen.se/en/statistics-and-data/statistics/), Statistics Sweden (https://www.scb.se/en/services/guidance-for-researchers-and-universities), the Swedish Pregnancy Register (https://www.medscinet.com/gr/forskare.aspx) and the National Register for Assisted Reproduction (https://www.medscinet.com/qivf/for-forskare.aspx).