Obstetric and perinatal outcomes in twin pregnancies conceived via assisted reproductive technology vs. spontaneous conception: a population-based cohort study.

The prevalence of twin pregnancies varies by population; however, the incidence of multiple gestations has markedly increased in recent decades, primarily attributable to advanced maternal age and the widespread utilization of fertility treatments, including assisted reproductive technology (ART) [ 1 – 4 ]. In ART-conceived pregnancies, the likelihood of multiple gestations remains elevated due to the routine use of controlled ovarian hyperstimulation and the transfer of multiple embryos to optimize implantation and pregnancy success rates [ 5 ]. Twin gestations are inherently associated with elevated risks of obstetric complications, such as fetal growth restriction, gestational diabetes, preeclampsia, and preterm birth [ 5 – 9 ]. Although these risks are well recognized in spontaneous twin pregnancies, ART-conceived twins may carry additional risks due to advanced maternal age, infertility-related conditions, and procedural factors related to ART itself [ 10 – 12 ]. In singleton pregnancies, numerous studies have consistently demonstrated that ART conception is associated with higher risks of adverse outcomes, including preterm birth, low birth weight, admission to the neonatal intensive care unit, perinatal death, and placenta previa [ 13 – 16 ]. However, despite the growing use of ART worldwide, relatively few large-scale studies have specifically examined ART-related risks in twin gestations, particularly in Asian populations. In Korea, the number of ART procedures has continued to rise each year, especially since the national health insurance began to cover ART treatments [ 17 ]. As a result, the proportion of twin pregnancies has steadily increased [ 18 ]. However, there remains a lack of large-scale studies comparing the obstetric risks of ART-conceived vs. spontaneously conceived twin pregnancies in the Korean population. Therefore, this study was designed to evaluate and compare obstetric and perinatal outcomes between ART-conceived and spontaneously conceived twin pregnancies using a nationwide Korean cohort, under the hypothesis that ART-related factors, including advanced maternal age and specific procedural characteristics, contribute to an increased risk of complications.

During the study period from October 2017 to December 2021, a total of 36,013 twin pregnancies were included in the final analysis, comprising 14,257 pregnancies conceived via assisted ART and 19,052 conceived spontaneously. As illustrated in Fig. 2 , the annual number of ART-conceived twin pregnancies steadily increased between 2018 and 2020 (from 3,525 to 4,149), while the number of spontaneously conceived twin pregnancies declined (from 2,719 to 2,314). Despite these opposing trends, the total number of twin pregnancies increased over time. Table 1 presents the baseline characteristics of the study population. Women in the ART group were significantly older than those in the spontaneous conception group (mean age, 35.4 vs. 33.3 years; P <0.001) and exhibited a higher prevalence of medical comorbidities, including hypertension, diabetes mellitus, polycystic ovarian syndrome, endometriosis, and autoimmune disorders. Pregnancy outcomes and obstetric/perinatal complications are summarized in Table 2 . For pregnancy outcomes, the analysis was conducted on the entire cohort of twin pregnancies (14,257 in the ART group and 19,052 in the spontaneous conception group). In contrast, obstetric and perinatal complications were evaluated only among women who delivered (12,026 in the ART group and 15,767 in the spontaneous conception group). Compared with spontaneously conceived twin pregnancies, ART-conceived pregnancies were associated with a significantly lower risk of miscarriage (adjusted OR [aOR], 0.63; 95% CI, 0.59–0.67). For live birth, no statistically significant difference was observed between the two groups (aOR, 1.20; 95% CI, 1.00–1.45). However, ART-conceived pregnancies were associated with significantly increased risks of multiple adverse obstetric and perinatal outcomes, including threatened abortion (aOR, 1.75; 95% CI, 1.64–1.87), gestational hypertension (aOR, 1.56; 95% CI, 1.39–1.74), preeclampsia (aOR, 1.31; 95% CI, 1.17–1.47), GDM (aOR, 1.32; 95% CI, 1.22–1.43), placenta previa (aOR, 1.81; 95% CI, 1.57–2.08), placenta accreta (aOR, 1.15; 95% CI, 0.83–1.58), postpartum hemorrhage (aOR, 1.19; 95% CI, 1.07–1.31), preterm labor (aOR, 1.20; 95% CI, 1.14–1.27), PPROM (aOR, 1.19; 95% CI, 1.11–1.28), and emergency cesarean section (aOR, 1.21; 95% CI, 1.08–1.34). In maternal age-stratified analyses, the risk of preeclampsia, preterm labor, and emergency cesarean section was more pronounced in women aged ≥35 years than in those aged <35 years. The aORs for ART vs. spontaneous conception in the ≥35 age group were 1.38 (95% CI, 1.19–1.61) for preeclampsia (vs. 1.26 [95% CI, 1.05–1.50] in <35), 1.19 (95% CI, 1.11–1.28) for preterm labor (vs. 1.17 [95% CI, 1.08–1.26]), and 1.22 (95% CI, 1.06–1.42) for emergency cesarean section (vs. 1.17 [95% CI, 1.00–1.37]), suggesting a clinically meaningful trend of risk amplification with increasing maternal age ( Table 3 ).

This study demonstrated that ART-conceived twin pregnancies were significantly associated with an increased risk of various obstetric complications, including placenta previa, preeclampsia, GDM, and emergency cesarean section, compared to spontaneously conceived twin pregnancies. Age-stratified analyses further revealed that certain risks associated with ART became more pronounced in women aged 35 years or older. Notably, for complications such as preeclampsia, preterm labor, and emergency cesarean delivery, there was a clear trend toward risk amplification among women over the age of 35. The underlying factors associated with these adverse outcomes may include the effects of both twin pregnancies and ART. Twin gestations are inherently associated with higher risks of major obstetric complications such as fetal growth restriction or small for gestational age (SGA), GDM, preeclampsia, and preterm delivery [ 6 ]. These adverse outcomes may be attributed to various pathophysiological mechanisms, including excess placental mass (as in GDM and preeclampsia) [ 20 , 21 ], placental crowding, or limited uteroplacental capacity to support the growth of multiple fetuses (e.g., SGA or growth restriction) [ 22 ], and suboptimal placental implantation sites, which increase the risk of placenta previa, velamentous cord insertion, and vasa previa compared with singleton pregnancies [ 23 , 24 ]. Furthermore, singleton pregnancies resulting from ART have been shown to carry a higher risk of adverse perinatal outcomes, including preterm birth, low birth weight, perinatal mortality, and SGA, compared with spontaneously conceived pregnancies [ 7 , 8 ]. These increased risks may be partly attributable to advanced maternal age and underlying parental factors associated with infertility, which are more prevalent among women undergoing ART [ 25 ]. In addition, the manipulation of gametes and embryos during ART procedures may interfere with implantation and placental development, thereby contributing to these adverse outcomes [ 26 ]. Our findings align with previous studies reporting that twin pregnancies conceived via ART are associated with a higher risk of adverse obstetric and perinatal outcomes compared with spontaneously conceived twin pregnancies. Lower birth weight, increased rates of preterm delivery, gestational diabetes, and placental abruption have been more frequently observed in the ART group [ 5 , 27 , 28 ]. Additionally, a recent study has demonstrated that ART pregnancies are associated with a higher risk of placenta accreta spectrum and postpartum hemorrhage [ 27 , 28 ], with several reports indicating increased rates of neonatal intensive care unit admissions and perinatal mortality in ART-conceived twins [ 29 , 30 ]. A recent study also showed that both monochorionic diamniotic (MCDA) and dichorionic diamniotic (DCDA) ART twin pregnancies had higher rates of preterm birth, lower neonatal birth weights, and reduced twin survival compared with naturally conceived twins [ 31 ]. In contrast, a Korean study published in 2011 reported no significant differences in obstetric outcomes between ART and spontaneously conceived dichorionic twin pregnancies [ 32 ]. This discrepancy may be partly explained by important differences in study design. First, the 2011 study included only dichorionic twins, whereas our analysis, due to limitations of the NHIS database, could not distinguish between dichorionic and monochorionic twin pregnancies. It is well established that dichorionic twins generally have more favorable obstetric outcomes compared with monochorionic twins [ 33 , 34 ]. Second, the sample size of the 2011 study was relatively small, with 286 ART-conceived twins and 134 spontaneously conceived twins-approximately one-eighth of the sample size in our study. Finally, the earlier study was a single-center investigation, while our study leveraged a nationwide claims database, allowing for a large-scale, population-based analysis. Taken together, the restricted twin type, small sample size, and single-center design of the 2011 study may have reduced its statistical power, potentially obscuring differences that became evident in our robust nationwide cohort. Furthermore, we observed a higher stillbirth rate in the spontaneously conceived group than in the ART group. This finding may be partially explained by the likely higher proportion of MCDA twins among spontaneously conceived pregnancies [ 35 ]. MCDA twins are known to have significantly elevated risks of stillbirth and perinatal complications due to shared placental circulation and unique complications such as twin-to-twin transfusion syndrome, selective fetal growth restriction, and twin anemia-polycythemia sequence. Previous studies have consistently shown that monochorionicity is more common in spontaneous conception and contributes disproportionately to adverse outcomes in twin pregnancies [ 35 , 36 ]. Although increasing maternal age did not result in statistically significant interactions for most outcomes, stratified analyses revealed higher aOR for preeclampsia, preterm labor, and emergency cesarean section among women aged ≥35 years compared with those aged <35 years. In the context of the rising proportion of women in their late 30s and 40s undergoing childbirth in Korea, this trend warrants clinical attention. The consistent elevation in risk across age strata suggests that maternal age may serve as a potential effect modifier in ART-related obstetric outcomes and should be carefully considered in clinical decision-making and counseling. These findings suggest that ART-conceived twin pregnancies should not be regarded merely as multiple gestations but rather classified as high-risk based on the specific characteristics of ART procedures. In particular, for older women with ART-conceived twins, proactive medical interventions-such as intensive antenatal surveillance, preterm birth prevention strategies, and planned cesarean delivery-may be clinically justified. According to data from Statistics Korea, approximately 30% of twin births are monozygotic, while about 70% are dizygotic [ 37 ]. Although our dataset does not distinguish between monozygotic and dizygotic twins, previous studies have indicated that the majority of ART-conceived twin pregnancies are dizygotic [ 38 , 39 ]. This predominance is largely attributed to the practice of transferring multiple embryos during ART procedures, which increases the likelihood of dizygotic twinning. Given these considerations, we cautiously propose that adopting elective single ET during ART treatments could be a viable strategy to enhance maternal and fetal safety by reducing the incidence of high-risk twin pregnancies. In cases where multiple gestation is unavoidable, a multidisciplinary approach involving obstetricians, endocrinologists, and maternal-fetal medicine specialists is essential to optimize perinatal outcomes. This study has several important strengths. To the best of our knowledge, this is the first large-scale, nationwide population-based cohort study in South Korea to specifically examine obstetric and perinatal outcomes in twin pregnancies conceived via ART vs. spontaneous conception using the NHIS database, and one of the few large-scale analyses in an Asian population. By encompassing all women who underwent ART nationwide, the study minimizes selection bias and provides a comprehensive overview of clinical outcomes in a real-world setting. The use of an extensive, unselected dataset allows for the generation of real-world evidence that reflects actual clinical practices. Furthermore, the longitudinal structure of the NHIS database enables the evaluation of long-term reproductive and obstetric outcomes. The scale and robustness of the data enhance its external validity, making the results generalizable to populations with similar healthcare infrastructures and insurance-based registries. In addition to these strengths, our study provides several unique contributions. Specifically, we utilized a comprehensive and recent dataset that captured virtually all ART procedures performed in Korea during the study period, ensuring that the findings are both up to date and nationally representative. Unlike many previous studies, we were able to include both monochorionic and dichorionic twin pregnancies, providing broader insights into the risks associated with ART-conceived twins. Furthermore, by examining a wide spectrum of pregnancy, obstetric, and perinatal outcomes, our study provides a more complete understanding of the potential risks associated with ART-related twin gestations and offers clinically relevant evidence that can inform patient counseling and prenatal care. Nonetheless, this study has certain limitations inherent to the use of administrative claims data. Potential inaccuracies in diagnostic or procedure coding may exist. Moreover, the NHIS database does not capture key clinical details regarding chorionicity (e.g., DCDA vs. MCDA), which limits the ability to distinguish between twin types. Additionally, it lacks information on ovulation induction and stimulation protocols, precluding more detailed analysis of treatment-specific factors. Important clinical and biochemical parameters-such as hormonal profiles, metabolic markers, body mass index, and ovulatory status-were unavailable. Embryo-related factors, including embryo quality and the number of embryos transferred, could not be evaluated. Additionally, as the database includes only reimbursed procedures, certain aspects of clinical care, such as the use of specific formulations of luteal phase support and selective fetal reduction, were not captured. Despite these limitations, the study’s nationwide scope and the use of a comprehensive claims database considerably strengthen its validity and relevance. The inclusion of the entire ART population in South Korea provides valuable insights into clinical outcomes in a real-world setting and contributes important evidence to an area that has not been previously explored at this scale in the Korean population. In conclusion, this large-scale national study highlights that twin pregnancies conceived via ART are associated with significantly higher risks of adverse obstetric and perinatal outcomes compared with spontaneous twin pregnancies. These findings underscore the need for heightened clinical vigilance and tailored management strategies for ART-conceived twins. To improve outcomes, clinicians should consider proactive surveillance. Future prospective studies are warranted to evaluate how specific in vitro ferilization protocols influence outcomes, as well as to assess long-term health trajectories-not only of the offspring, including neonatal intensive care unit admissions, growth, and developmental progress-but also of the mothers, to better understand the extended impact of ART on maternal health.

A customized research dataset from the National Health Insurance Service (NHIS) of Korea (NHIS-2023-1616) was used for this study. The NHIS, operating under the Ministry of Health, Welfare, and Family Affairs, prvides medical coverage to approximately 97% of the Korean population, collecting healthcare utilization data from all medical institutions for reimbursement [ 19 ]. This national database includes both inpatient and outpatient claims, coded according to the International Classification of Diseases, 10th revision (ICD-10), along with pharmacy and procedural data recorded using codes designated by the Health Insurance Review and Assessment Service. Since October 2017, infertility treatments have been covered by the national health insurance program. Accordingly, this study included all NHIS claims data between October 2017 and December 2021, a period during which fertility-related treatments were systematically recorded. All data used in this study were fully anonymized prior to analysis to protect personal information. The study protocol was approved by the Institutional Review Board (IRB No. 2304-015-126) and reviewed by the official data access committee of the Korean government. Given the use of de-identified data, the requirement for informed consent was waived in accordance with applicable ethical guidelines. A flowchart outlining the selection of study participants is presented in Fig. 1 . This study included women who were diagnosed with twin pregnancies for the first time during the study period, identified through the NHIS database using the ICD-10 code O30.0. A total of 59,979 women were initially identified based on this diagnostic code. Exclusion criteria included the following: 1) absence of diagnostic or procedural codes related to childbirth (R3133, R3138, R3143, R3148, RA312, RA314, RA316, RA318, R434, R4353, R4358, RA432, RA434, R4516, R4519, R4520, R5001, R5002, Z3720–Z3722, Z3730–Z3732, Z374) within 360 days of the initial pregnancy diagnosis. 2) Absence of a diagnosis of abortion (ICD-10 codes O00–O06) within 360 days of the initial pregnancy diagnosis. And 3) presence of artificial insemination procedure codes (R6560) within 90 days of the initial pregnancy diagnosis. Pregnancies were categorized as ART or spontaneous based on the presence of embryo transfer (ET) procedure codes (R6530–R6533, R6540, R6550) recorded within 90 days of the initial diagnosis. Pregnancy and obstetric outcomes were defined using diagnostic and procedural codes based on the ICD-10 classification system. Pregnancy outcomes were grouped into abortion (O00–O06), live birth (Z3700–Z3702, Z3790–Z3792, and delivery-related procedure codes), and stillbirth (Z371). Analyses of obstetric and fetal complications were limited to women who delivered, excluding those with abortion diagnoses. The obstetric complications examined included threatened abortion (O200), hypertensive disorders of pregnancy (O11, O13–O15)-such as gestational hypertension and preeclampsia-placental disorders (placental insufficiency, abruption, previa, and accreta), postpartum hemorrhage (O72), gestational diabetes mellitus (GDM) (O244), preterm labor (O602), preterm premature rupture of membranes (PPROM) (O42), cesarean section (O82), and emergency cesarean delivery (O821). Fetal and perinatal outcomes of interest included fetal anomalies (O35), intrauterine fetal death (O364), intrauterine growth restriction (O365), macrosomia (O366), and fetal distress (O368, O68). Covariates were assessed to adjust for potential confounding and included maternal age at pregnancy confirmation, gravidity (≥1), and preexisting medical or psychiatric conditions recorded within 1 year prior. These included hypertension (I10–I15), diabetes (E10–E14), endometriosis (N80), dyslipidemia (E78), recurrent pregnancy loss (N96, O26.2), autoimmune disorders (M05–M06, M32–M35, M45, M79, D56, D89, K50–K51, L40), asthma (J45–J46), thyroid dysfunction (E02–E03, E05), and mental health disorders such as depression (F32–F33), anxiety (F40–F41), and bipolar disorder (F30–F31). Group comparisons for categorical variables were performed using the chi-square test. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated via binary logistic regression, adjusted for maternal age and other covariates. All statistical analyses were carried out with SAS Enterprise Guide version 7.15 (SAS Institute Inc., Cary, NC, USA) and R software version 4.3.1 (R Core Team, Vienna, Austria), with relevant supplemental packages for graphical output.