Perinatal outcomes in 13,626 singleton pregnancies after autologous IVF across three continents over 7 years

Abstract

Purpose

Are trends in singleton autologous IVF perinatal outcomes consistent over time among five international infertility centers?

Methods

This was a retrospective cohort study from January 1, 2012, to December 31, 2018. This study was performed through a large infertility network at five international infertility centers in which patients who had a singleton live birth resulting from fresh and frozen autologous IVF cycles were included. The primary outcome was live birth weight (BW) with secondary outcomes of preterm birth (PTB), large for gestational age (LGA), small for gestational age (SGA), and gestational age at delivery.

Results

The entire cohort (n = 13,626) consisted of 6941 fresh and 6685 frozen autologous IVF cycles leading to singleton deliveries. Maternal age, parity, body mass index, neonatal sex, and GA at delivery were similar for fresh and frozen IVF cycles in the entire cohort and within each infertility center. Four centers had a trend of decreased BW and three centers had decreased rates of PTB before 32 and 28 weeks and LGA newborns annually, although significance was not reached. Three IVF centers had annual increased trends of PTB before 37 weeks and four centers had increased rates of SGA newborns, although significance was not reached.

Conclusion

Similar trends in perinatal outcomes were present across five international infertility centers over 7 years. Additional studies are crucial to further assess and optimize perinatal outcomes at an international level.

Introduction

In vitro fertilization (IVF) is used globally with over 8 million babies having been conceived [1]. Clinical outcomes such as live birth rate, clinical pregnancy rate, and miscarriage rate are the predominant metrics for IVF centers. However, perinatal outcomes such as birth weight (BW), preterm birth (PTB), large for gestational age (LGA), and small for gestational age (SGA) can act as additional surrogate metrics for fetal health in the neonatal period and long-term health of IVF offspring.

Very preterm infants (<32 weeks) have been associated with an increased risk of long-term neurological morbidity [2] and any preterm (<37 weeks) infants may have decreased cognitive performance [3]. SGA newborns may have increased risks of mild cognitive deficits and more learning difficulties in adolescents and adults [4]. In contrast to SGA newborns, LGA newborns have been associated to maternal and neonatal perinatal complications [5, 6], adult chronic kidney disease [7], and adverse arterial profiles [8]. Therefore, understanding how IVF may influence certain perinatal outcomes is important to monitor long-term health of offspring.

For IVF outcomes to improve, the laboratory and clinical aspects of the IVF process are continuously reviewed to optimize outcomes. Examples of improvements, over the last few decades, within the IVF laboratory include the culture media, bench top incubators, time-lapse imaging, and cryopreservation techniques [9–13]. Similarly, clinical changes in IVF stimulation protocols [14], medications, cycle monitoring, day of embryo transfer (ET) [12], and number of embryos transferred [15] also contribute to the improvement in IVF outcomes.

The use of vitrification as the primary cryopreservation technique has become the standard in most IVF laboratories between 2010 and 2012 [16]. With the adoption of vitrification, frozen ETs have become more common compared to fresh IVF cycles in the USA [17] and internationally [18]. When comparing frozen ETs to fresh IVF, prior studies have demonstrated decreased risks of PTB, low birth weight, and SGA newborns but increased risks of LGA [19–21]. However, our institution previously showed a decrease in LGA newborns for both fresh and frozen IVF cycles over a 24-year period [12].

Trends in BW from IVF pregnancies have varied. Our institution based, in the USA, previously demonstrated a gradual decrease in BW per year in fresh and frozen IVF cycles [12] while another study in the UK reported a gradual increase in BW per year in fresh and frozen IVF cycles [22]. Differences in patient ethnicity [23] or IVF clinical and laboratory management [12] could be some reasons for differences in BW trends from clinics in different geographic regions.

Given the constant evolution of IVF laboratorial and clinical strategies worldwide, we investigated how singleton fresh and frozen autologous IVF perinatal outcomes have changed over a period of time in a large sample size and if trends are similar across five international infertility centers, spanning three continents.

Materials and methods

This retrospective cohort study consisted of singleton live births from autologous fresh and frozen IVF cycles from five international infertility clinics in the Eugin network (Boston, USA; Barcelona, Spain; Monza, Italy; Granollers, Spain; Huntington, Brazil). Data was collected from January 1, 2012, to December 31, 2018. The methods section, as described below, has been previously reported (Shah et al. submitted for publication), with the exception that this study was investigating autologous IVF cycles.

Institutional Review Board waiver of approval (2020D000659) was obtained from the Committee on Clinical Investigations at Beth Israel Deaconess Medical Center (Boston), CEIM of EUGIN (CEUGIN-2020-18-BIWE) (Barcelona, Granollers), Ethical Committee of the Province of Monza Brianza (PMA 1-2014) (Monza), and Ethics in Research Committee/National Committee of Ethics in Research (38792222.6.0000.0070) (Huntington). The collated deidentified retrospective dataset included data which was analyzed as an entire cohort and by individual infertility centers. Patients were included if they had autologous IVF (fresh or frozen), a singleton live birth (>22 weeks gestation) from IVF or intracytoplasmic sperm injection (ICSI), a recorded BW, and a recorded GA at delivery. Patients were included if they had a singleton live birth even if there was more than one initial fetal sac.

Baseline maternal and neonatal characteristics, including maternal age, pre-pregnancy body mass index (BMI, kg/m²), gravidity, parity, neonatal sex, GA at delivery, and live BW, were collected. Clinical characteristics, such as mode of fertilization (IVF vs. ICSI), number of embryos transferred, and day of ET, were included. Perinatal outcomes of PTB, SGA, and LGA rates were calculated for all patients [24].

The primary outcome was BW (adjusted for GA at delivery, age at cycle start, pre-pregnancy BMI, parity, neonatal sex, culture duration, number of transferred embryos, and fresh vs. frozen embryo transfer) with secondary outcomes of PTB before 37, 32, and 28 weeks; SGA (BW <10th %); and LGA (BW >90th %).

The unadjusted effects of the year of delivery on several perinatal outcomes were tested by linear (birthweight and gestational age at delivery) or logistic (preterm birth, SGA, and LGA) regression. The adjusted effects were modeled by including as covariates: maternal age at cycle start; BMI at cycle start; maternal parity (0 vs. 1,2,3+); day of ET (cleavage [days 1–3] vs. blastocyst [days 5–7]); fresh versus frozen ET; and number of embryos transferred (1 vs. 2,3,4+). To account for women contributing with more than one cycle (n = 1179), all the above-described models were made hierarchical by including an intercept random effect (multilevel regression).

Mean imputation for BMI was performed in 1965 cycles (Boston: 388, Barcelona: 566, Monza: 0, Granollers: 419, Huntington: 592) by linear regression with age and infertility center as a predictor and using residuals for estimation adjustment with randomness of the missing values checked by Little’s test. Furthermore, neonatal sex was imputed as the mid-effect [25] (male = 1, female = 0, missing = 0.5) in 589 patients (Boston: 38, Barcelona: 8, Monza: 1, Granollers: 0, Huntington: 542). A P value < 0.05 was considered statistically significant.

Results

A total of 13,626 singleton live births resulted from fresh (n = 6941) and frozen (n = 6685) IVF cycles in the entire cohort. Patient demographics, treatment factors, and clinical outcomes are shown in Table 1 for the entire cohort. The number of fresh and frozen IVF cycles at each respective infertility center that resulted in singleton live births is included: Boston (n = 6780), fresh (n = 3269) and frozen (n = 3511); Barcelona (n = 1882), fresh (n = 998) and frozen (n = 884); Monza (n = 2093), fresh (n = 1549) and frozen (n = 544); Granollers (n = 419), fresh (n = 250) and frozen (n = 169); Huntington (n = 2452), fresh (n = 875) and frozen (n = 1577). Maternal age, pre-pregnancy BMI, GA at delivery, parity, and neonatal sex were similar between fresh and frozen IVF cycles in the entire cohort and within each infertility center. Preimplantation genetic testing was performed at 3 centers (Boston, Granollers, Huntington) in 14.2 to 33.0% and 0.2 to 11.2% of frozen and fresh ET cycles with normal results, respectively.

Table 1.

Characteristics of the entire autologous IVF cohort that achieved a singleton delivery after a fresh or frozen embryo transfer

		IVF (n = 13,626)
	Missing data	Fresh ET	Frozen ET
Singletons	0	6941	6685
Center
Boston, USA		3269 (47.1)	3511 (52.5)
Barcelona, Spain		998 (14.4)	884 (13.2)
Monza, Italy		1549 (22.3)	544 (8.1)
Granollers, Spain		250 (3.6)	169 (2.5)
Huntington, Brazil		875 (12.6)	1577 (23.6)
Patient factors
Maternal age at cycle start (years)	13	35.1 (4.1) [20–46.2]	35.3 (3.9) [19.6–48]
Age > 40 years	13	865 (12.5)	845 (12.6)
Body mass index (kg/m2)	1965	24.7 (5.2) [15.6–48.3]	24.8 (5.5) [15.4–50.3]
Maternal parity	12
0		5460 (78.7)	4989 (74.7)
1		1306 (18.8)	1475 (22.1)
2		146 (2.1)	184 (2.8)
3+		23 (0.3)	31 (0.5)
Treatment factors
Any PGT	0	128 (1.8)	1595 (23.9)
ICSI	1450	4635 (67.3)	1681 (31.8)
Type of embryo transfer	0
Cleavage (days 1–3)		3967 (57.2)	810 (12.1)
Blastocyst (days 5–7)		2974 (42.8)	5875 (87.9)
Total embryos transferred	0
1		2753 (39.7)	4361 (65.2)
2		3384 (48.8)	2103 (31.5)
3		646 (9.3)	204 (3.1)
4+		158 (2.3)	17 (0.3)
Outcomes
Unadjusted birth weight (g)	0	3175 (573) [480–5812]	3329 (575) [510–5443]
Gestational age (weeks)	0	38.4 (2) [23–42.6]	38.6 (1.9) [23.4–43]
Preterm birth	0
<37 weeks		906 (13.1)	738 (11)
<32 weeks		103 (1.5)	73 (1.1)
<28 weeks		28 (0.4)	25 (0.4)
Female sex	589	3294 (48.9)	3054 (48.4)
Male sex		3438 (51.1)	3251 (51.6)
SGA < 10%	0	888 (12.8)	514 (7.7)
LGA > 90%		459 (6.6)	769 (11.5)
Spontaneous fetal reduction	2871	195 (3.4)	88 (1.8)

Continuous variables are presented as mean (standard deviation) [range] and categorical variables as row total (%)

ET, embryo transfer; IVF, in vitro fertilization; PGT, preimplantation genetic testing; ICSI, intracytoplasmic sperm injection; SGA, small for gestational age; LGA, large for gestational age

*Calculated by: delivery date minus transfer date + 15 (day 1), + 16 (day 2), + 17 (day 3), + 19 (day 5), + 20 (day 6), + 21 (day 7)

For fresh ETs, Barcelona, Monza, and Granollers predominately performed cleavage stage double ETs while Boston and Huntington predominately performed blastocyst single and double ETs, respectively. For frozen ETs, Boston, Monza, and Granollers predominately performed blastocyst single ETs, while Barcelona and Huntington predominately performed cleavage single ETs and blastocyst double ETs, respectively. The mean BW for fresh and frozen ETs in the entire cohort was 3175 g (± 573) and 3329 g (± 575), respectively. In all five infertility centers, the BW was higher in frozen versus fresh ETs with a range of 63 to 197 g. For the entire cohort, the adjusted BW for blastocyst versus cleavage stage ET was significantly higher (48.5 g, 95% CI 26.1 g, 70.79 g; P < 0.001). For the entire cohort, the adjusted BW for ICSI versus IVF was significantly lower (−138.8 g, 95% CI −160.7 g, −116.9 g; P < 0.001).

For the entire cohort, 88.6% of fresh (83.2–94.8%) and 90.3% of frozen (87.6–93.4%) IVF cycles had term deliveries (≥37 weeks gestation) with 88.9% of fresh (83.3–92.0%) and 92.7% of frozen (91.0–94.5%) IVF cycles having a live BW >2500 g. For the entire cohort, 3.4% of fresh (0.8–8.1%) and 7.8% of frozen (5.3–13.5%) IVF cycles had a macrosomic newborn (BW >4000 g). At each specific infertility center, the percentages of term deliveries (≥37 weeks gestation) for fresh and frozen IVF cycles were, respectively: Boston (85.1%, 87.6%), Barcelona (90.8%, 93.4%), Monza (89.2%, 92.5%), Granollers (94.8%, 89.9%), and Huntington (83.2%, 88.3%). The percentages of live BW >2500 g at each specific infertility center for fresh and frozen IVF cycles were, respectively: Boston (91.2%, 94.5%), Barcelona (88.3%, 92.1%), Monza (89.5%, 92.2%), Granollers (92.0%, 93.5%), and Huntington (83.3%, 91.0%). At each specific infertility center, the percentages of macrosomia (≥4000 g) for fresh and frozen IVF cycles were, respectively: Boston (8.1%, 13.5%), Barcelona (3.6%, 6.6%), Monza (3.4%, 7.9%), Granollers (0.8%, 5.3%), and Huntington (1.3%, 5.5%).

Table 2 shows the incidence of each perinatal outcome for the entire study period by infertility center. Boston and Huntington had the highest incidence of PTB before 37 weeks. The incidence rates of PTB before 32 and 28 weeks ranged from 1.1 to 1.5% and 0.3 to 0.7% for the five centers, respectively. Boston had about half the incidence rate of SGA but more than double the incidence rate of LGA newborns compared to the other four infertility centers.

Table 2.

Incidence of each perinatal outcome by infertility center

Perinatal outcome	Boston, USA n = 6780	Barcelona, Spain n = 1882	Monza, Italy n = 2093	Granollers, Spain n = 419	Huntington, Brazil n = 2452
Preterm birth <37 weeks	923	150	209	30	332
	13.6%	8.0%	10.0%	7.2%	13.5%
Preterm birth <32 weeks	82	20	32	6	36
	1.2%	1.1%	1.5%	1.4%	1.5%
Preterm birth <28 weeks	26	5	8	3	11
	0.4%	0.3%	0.4%	0.7%	0.4%
Small for gestational age	462	267	309	61	303
	6.8%	14.2%	14.8%	14.6%	12.4%
Large for gestational age	868	117	99	13	131
	12.8%	6.2%	4.7%	3.1%	5.3%

Presented as number and percentage of each outcome

Table 3 displays the unadjusted and adjusted analysis estimating annual trends of GA at delivery, BW, PTB, SGA, and LGA by linear regression for each infertility center. There was a general trend for decreased adjusted BW yearly for the entire cohort of IVF cycles in four of the five centers, although significance was only reached in one center (Table 3). There was a non-significant trend for increased annual rates of PTB before 37 weeks but a decreased annual rate of PTB before 32 and 28 weeks in three of the five centers (Table 3). Although not significant, the rate of SGA newborns had an increased annual trend in four centers while the rate of LGA newborns had a decreased annual trend in three centers (Table 3).

Table 3.

Change by year in birthweight, gestational age at delivery, and the odds of perinatal outcomes for each infertility center

Eugin Center			Gestational age (weeks)b	Birthweight (grams)a	Preterm birth < 37 weeksb	Preterm birth < 32 weeksb	Preterm birth < 28 weeksb	Small for gestational age < 10%b	Large for gestational age > 90%b
Boston, United Statesf	Unadjusted	Coefficient/OR (95% CI)	−0.02 (−0.04, 0.01)	2.66 (−4.83, 10.15)	1.03 (0.99, 1.07)	1.03 (0.91, 1.16)	1.06 (0.85, 1.31)	0.96 (0.91, 1.01)	0.98 (0.943, 1.018)
	Adjusted		−0.04 (−0.07, −0.01)d	−2.41 (−9.17, 4.36)	1.05 (1.01, 1.10)d	1.03 (0.89, 1.18)	0.93 (0.72, 1.20)	1.02 (0.96, 1.08)	0.93 (0.88, 0.97)d
Barcelona, Spaing	Unadjusted	Coefficient/OR (95% CI)	0.06 (0.01, 0.11)e	−0.47 (−14.16, 13.22)	0.88 (0.81, 0.96)d	0.96 (0.77, 1.21)	0.61 (0.39, 0.97)d	0.98 (0.92, 1.05)	1.03 (0.93, 1.14)
	Adjusted		0.04 (−0.01, 0.10)	−19.53 (−32.37, −6.68)d	0.89 (0.80, 0.98)e	0.96 (0.74, 1.25)	0.65 (0.38, 1.09)	1.06 (0.97, 1.14)	1.00 (0.89, 1.13)
Monza, Italyh	Unadjusted	Coefficient/OR (95% CI)	−0.02 (−0.12, 0.08)	14.64 (−9.67, 38.95)	0.99 (0.92, 1.06)	1.18 (0.76, 1.83)	1.31 (0.67, 2.55)	0.87 (0.76, 1.00)	1.06 (0.79, 1.41)
	Adjusted		0.00 (−0.12, 0.12)	17.05 (−6.60, 40.70)	1.02 (0.80, 1.28)	0.93 (0.53, 1.63)	1.17 (0.51, 2.68)	0.95 (0.80, 1.13)	0.84 (0.58, 1.22)
Granollers, Spaini	Unadjusted	Coefficient/OR (95% CI)	0.00 (−0.05, 0.04)	−0.97 (−12.66, 10.73)	0.99 (0.92, 1.06)	1.06 (0.89, 1.26)	1.03 (0.73, 1.45)	1.04 (0.98, 1.10)	0.96 (0.87, 1.05)
	Adjusted		−0.02 (−0.07, 0.03)	−4.55 (−13.44, 4.34)	1.01 (0.94, 1.09)	1.13 (0.94, 1.37)	1.14 (0.76, 1.71)	1.04 (0.98, 1.11)	0.91 (0.82, 1.01)
Huntington, Brazilj	Unadjusted	Coefficient/OR (95% CI)	0.11 (0.07, 0.15)c	26.90 (15.14, 38.66)c	0.90 (0.85, 0.96)e	0.80 (0.68, 0.96)e	0.71 (0.51, 0.98)d	0.98 (0.91, 1.04)	1.13 (1.02, 1.25)e
	Adjusted		0.07 (0.03, 0.12)e	−8.77 (−19.29, 1.74)	0.95 (0.89, 1.02)	0.83 (0.68, 1.01)	0.78 (0.54, 1.12)	1.05 (0.98, 1.13)	1.01 (0.90, 1.13)

^aAdjusted for gestational age at delivery; age at cycle start; body mass index (kg/m²); parity (0 vs. 1,2,3+); neonatal sex; culture duration (1–3 vs. 5–7); number of transferred embryos (1 vs. 2,3,4+); and fresh vs. frozen embryo transfer

^bAdjusted for age at cycle start; body mass index (kg/m²); parity (0 vs. 1,2,3+); neonatal sex; culture duration (1–3 vs. 5–7); number of transferred embryos (1 vs. 2,3,4+); and fresh vs. frozen embryo transfer

Bold lettering indicates significant outcomes

^cP < 0.001

^dP < 0.05

^eP < 0.002

^fNo missing variables

^gMissing 2 maternal age and 12 parity

^hNo missing variables

ⁱNo missing variables

^jMissing 11 maternal age

Discussion

In this international retrospective cohort study, there were non-significant general trends in lower BW yearly, decreased rates of LGA newborns, and decreased rates of PTB before 32 and 28 weeks for fresh and frozen IVF cycles. However, there were non-significant annual trends of increased rates of PTB before 37 weeks and increased annual rates of SGA newborns for fresh and frozen IVF cycles. Monitoring the trends in these perinatal outcomes are important given the long-term effects of offspring health [26, 27].

Our findings are consistent with prior studies [12, 28–34] in that BWs for fresh versus frozen ETs were lighter, irrespective of the infertility center, with a range of 63 to 197 g. This is postulated as the supraphysiological environment created during a fresh IVF compared to a frozen ET cycle may play a role in the differing BWs [35]. The ovarian stimulation and gonadotropins in fresh IVF can affect the endometrium, the embryo implantation, and placentation which all may play a role in lower BWs [35]. Consistent with prior literature regarding maternal parity and influence on BW [12, 36, 37], our findings confirm that BW increases with increasing parity.

Our findings demonstrate a significant decrease in mean BW per year for both fresh and frozen IVF cycles of 19.5 g for Barcelona. Although not significant, the mean BW per year decreased in Boston by 2.4 g, in Granollers by 4.6 g, and in Huntington by 8.8 g while Monza had an increase of 17.1 g. A prior study by Castillo et al. [22] demonstrated an increase in BW of 7.4 g per year (1991–2015) for 2780 singletons conceived by autologous IVF. These differences may stem from varied patient populations, geography, and IVF and obstetrical practice patterns. Future analysis looking at differences in demographics and practice patterns may help to discern some of these trends. The similar trend of decreased BW per year correlates to our general trend of decrease rates of LGA newborns in our overall cohort.

Prior literature has shown frozen IVF BWs to be higher compared to BWs with unassisted conception [38]. However, singleton BWs for all modes of conception in the literature for the USA [39], Spain [40], Monza [41], and Huntington [42] have comparable BWs to the frozen IVF BWs in our study. The incidence of macrosomia was similar for fresh and frozen IVF cycles in four of the five infertility centers with the incidence in Boston being double compared to its counterparts. The maternal BMI for the Boston cohort was in the overweight range and was on average two units greater and significantly different (P < 0.001) than the other centers. Historically, the prevalence of obesity and diabetes is higher in the USA compared to other countries [43] which are risk factors for macrosomia [27].

Our results are concordant with two meta-analyses [19, 21] comparing fresh and frozen IVF cycles in demonstrating that frozen ETs result in decreased rates of PTB but our findings differed in that there were decreased rates of LGA newborns annually for three centers for both fresh and frozen IVF cycles. Our study demonstrated a general trend for both fresh and frozen IVF in decreased rates of PTB before 32 and 28 weeks for three centers; however, PTB before 37 demonstrated an increased rate for three centers. With the increased utilization of frozen IVF cycles over the years, future research will be important to continually evaluate BW, PTB, LGA, and SGA to monitor trends over time and correlate them with key changes in laboratory and clinical management.

The worldwide rate of PTB before 37 weeks is 11.1% and 8.6% in developed countries for all modes of conception [44]. Reported rates of PTB before 37 weeks by country for all modes of conception include: 10.2% (USA) [45], 6.7% (Spain), 6.5% (Monza), and 7% (Huntington) [44]. The rate of PTB before 37 weeks for fresh and frozen IVF in the present study is higher compared to the prior published PTB rates for Boston (13.6%), Barcelona (8.0%), Monza (10.0%), Granollers (7.2%), and Huntington (13.5%). The authors propose a theory for the decreased rate of early PTB but increased rate of late PTB possibly due to improvements in obstetrical management. This could potentially be a source of the decreased rate of early PTBs and extending higher risk pregnancies a few weeks further into the late preterm category. Further studies investigating specific factors that may influence PTB are warranted as this information would be helpful with patient counseling.

The incidence of term deliveries in our entire cohort for each specific infertility center was relatively similar with lower rates in fresh versus frozen IVF. Huntington had the lowest percentages of term deliveries and lowest overall BW compared to the other centers for both fresh and frozen IVF cycles; however, they had a larger percentage of blastocyst double ETs for both fresh and frozen IVF cycles. It is unknown how many of these patients from the Huntington cohort had spontaneous fetal reduction, which is known to affect BW [46].

Given similar trends were seen in BW, PTB before 32 and 28 weeks, and LGA newborns across the infertility centers, this provides some reassurance that fresh and frozen IVF perinatal outcomes are improving. However, the increased trend in PTB before 37 weeks and SGA newborns demonstrates the continued need in studying and optimizing perinatal outcomes. The consistency of the trends in our study may lend itself to more international based studies to investigate various infertility topics. This type of analysis can help decipher general effects of the IVF process itself, in comparison to outcomes that are more closely linked to patient demographics specific to a country. The results presented in the current study indicate that certain outcomes, such as the difference in BW between fresh and frozen, can be generalized to the IVF procedure itself.

Our study has several notable strengths. To our knowledge, this is the first cohort study to assess perinatal outcomes internationally across multiple different centers. Given our data is from five international infertility centers, these findings are more generalizable to more geographic locations and patient populations. Our study was able to validate prior findings in the literature regarding BW differences in neonatal sex, cycle type (fresh vs. frozen), and maternal parity. Data from the five infertility centers had a varying number of total IVF patients. However, instead of grouping all centers as one large cohort, a separate linear and logistic regression analysis was performed for each center to give a truer representation of each center’s perinatal outcomes. Our comparison across international regions also allows us to understand factors that are more generalizable to the IVF procedure itself and what may be attributed to regional idiosyncrasies.

Several limitations, however, need to be acknowledged. As a retrospective study, there remains the possibility of unknown confounders limiting conclusions about cause and effect. Changes in obstetrical management could not be accounted for. Obstetrical factors such as gestational diabetes, hypertensive disorders of pregnancy, and history of a prior PTB and patient factors such as tobacco, alcohol use, or patient diagnosis were not available in this dataset. A further limitation is that we have also assumed that the broad range of countries examined in this study represent differences in terms of ethnicity and demographic factors; however, specific data on demographic factors such as income, education level, and marital status were not available in this dataset. Mean imputation of BMI was performed for the cycles with missing data by linear regression with randomness checked by Little’s test. This is relevant as maternal BMI is a known risk factor for LGA newborns. Imputation was performed for neonatal sex for patients with missing data. To account for women contributing with more than one cycle, statistical models were made hierarchical by including an intercept random effect (multilevel regression). Spontaneous fetal reduction was missing in 21.1% of cycles and is a confounding factor for BW, although voluntary reduction is rare in the setting of IVF. A large study by Kamath et al. [47] showed that the vanishing twin rate in frozen ART cycles following transfer of ≥2 embryos was 2.4% (355/14,708) and in fresh transfers was 3.5% (2910/82,244). Although factors such as days of embryo culture and number of embryos transferred were controlled for in the logistic regression, the heterogeneity in practice patterns between clinics still needs to be considered when interpreting our data. Future research should be performed at more international infertility centers to assess perinatal outcome trends to confirm our findings or identify other geographic patterns.

In conclusion, this study showed general trends in annual reductions in BW, rates of PTB before 32 and 28 weeks, and LGA newborns across our international cohort for fresh and frozen IVF cycles. Additionally, there were increased annual rates of PTB before 37 weeks and SGA newborns for fresh and frozen IVF cycles. Similar trends in perinatal outcomes were present across five international infertility centers over 7 years. Additional studies are crucial to further assess and optimize perinatal outcomes at an international level.

Author contributions

All authors contributed to the conceptualization of the idea and study design, data collection, critical revision of the manuscript, data interpretation, and approved the final draft of the manuscript. F.F. performed the analysis and J.S.S. drafted the manuscript.

Declarations

Ethics approval

This research study was conducted retrospectively from data obtained for clinical purposes. We consulted with the IRB at each institution and our study did not require ethical approval. IRB waivers of ethical approval were granted from the Committee on Clinical Investigations at Beth Israel Deaconess Medical Center (Boston), CEIM of EUGIN (Barcelona, FecunMed-Granollers), Ethical Committee of the Province of Monza Brianza (Italy), and Ethics in Research Committee/National Committee of Ethics in Research (Brazil).

Conflict of interest

FF received fees for statistical analysis of the current work. DS is on the speaker panel of EMD Serono and is Deputy Editor of Human Reproduction.