Health outcomes for Massachusetts infants after fresh versus frozen embryo transfer.

Our study cohort included 14,491 infants. Among these, 12,390 infants were conceived by fresh embryo transfer and 2,101 by FET. ( Figure 1 ). There were significant differences in maternal sociodemographic and clinical characteristics between mothers in the fresh and FET groups. Compared with mothers who underwent fresh embryo transfer, those who underwent FET were less likely to be 40 or older at delivery, non-Hispanic white, and have adequate plus level of prenatal care. For clinical diagnoses, FET-treated mothers had higher incidence of ovarian dysfunction and lower incidence of diminished ovarian reserve. ( Table 1 ). Compared to mothers who underwent fresh embryo transfer, FET-treated mothers were of higher parity, more likely to have one embryo transferred, undergo cesarean section, have pregnancy-induced hypertension, and have prolonged hospital stays. ( Table 2 ) As highlighted in our prior work, compared to Massachusetts mothers of fertile status, mothers who underwent ART had higher prevalence of cesarean section (31.1% vs 44.3%), pregnancy-induced hypertension (9.3% vs 11.9%) and gestational diabetes (5.8% vs 8.1%), but lower prevalence of prolonged hospital stay (7.5% vs 5.0%). There were significant differences in health outcomes between infants conceived by fresh versus FET with FET infants having higher birthweight and LGA status and lower prevalence of SGA and low birthweight status. In addition, FET infants were more likely to have infectious disease, respiratory, and hematologic conditions compared to the fresh embryo transfer group. ( Table 3 ). In the multivariate analysis, compared to infants conceived by fresh embryo transfer, those in the FET group, had higher odds of being LGA (AOR 1.47: 95% CI 1.26–1.70) and lower odds of being SGA (AOR 0.56; 95% CI 0.44–0.70), and low birthweight (AOR 0.72; 95% CI 0.59–0.88). However, infants in the FET group had higher odds of having infectious disease (AOR 1.46; 95% CI 1.03–2.06), respiratory (AOR 1.23; 95% CI 1.07–1.41), and neurologic (AOR 1.32; 95% CI 1.04–1.68) conditions. There was no difference between groups in the outcomes of preterm birth, birth defects, non-chromosomal defects, hematologic and gastrointestinal conditions, and in prolonged length of stay for infants ≥ 35 weeks gestational age. ( Table 3 ) Due to Center for Health Information Analysis (CHIA) at MDPH, there are restrictions on release of data for cell sizes < 11 patients, therefore, adjusted results for the outcomes of mortality, cardiovascular conditions, and chromosomal birth defects could not be reported.

The data source for this study was the Massachusetts Outcome Study of Assisted Reproductive Technology (MOSART) database, a linked data system comprised of data from: 1) the Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS) database which contains cycle-based ART data from nearly all US ART clinics and 2) the Pregnancy to Early Life Longitudinal (PELL) data system, a population-based system that includes birth certificates, death records, and hospital utilization data for Massachusetts mothers and infants. Institutional Review Board approval was obtained from the Massachusetts Department of Public Health, Dartmouth College, and Boston University Medical Center. The SART Research Committee approved the study. Over 90% of ART clinics in the U.S. report data to SART CORS. In compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102–493), data were collected and verified by SART and reported to the Centers for Disease Control and Prevention. The data in the SART CORS are validated annually with some clinics having on-site visits for chart review based on an algorithm for clinic selection. During each visit, data reported by the clinic were compared with information recorded in patients’ charts. Ten out of 11 data fields selected for validation were found to have discrepancy rates of ≤5%.( 15 , 16 ) The following information are included: demographics, ART diagnoses, treatment parameters, and pregnancy outcomes. The PELL system has data on approximately 99% of all births and fetal deaths in Massachusetts since 1998 which are linked to hospital utilization data for women and their children. The Massachusetts Birth Defects Monitoring Program (BDMP), which conducts population-based active surveillance of structural birth defects among Massachusetts residents from birth through the first year of life, provides additional birth defects data for PELL. The SART CORS and PELL data systems were linked to create the MOSART database using a deterministic five phase linkage algorithm based on the baby’s date of birth, mother’s date of birth, mother’s first name and last name; and father/partner’s last name for all children born to Massachusetts resident women between July 1, 2004 and December 31, 2013.( 17 ) The starting date of July 1, 2004 was chosen based on the availability of SART CORS data (January 1, 2004) to allow us to capture any births associated with ART and the end date reflected the latest available data from both SART and PELL when this analysis was initiated. The linkage rate was 90.2% overall and 94.5% for deliveries in which both mother’s zip code and clinic were in Massachusetts. Inclusion criteria included: 1. Mothers’ delivery in MOSART; 2. Infants with inpatient birth hospital records; 3. Maternal age ≥ 18 years; 4. Singleton live births; 5. Conceived by fresh or frozen embryo transfer after ART using autologous oocytes with embryos transferred to the uterus; 6. Cycles in which oocytes or embryos were transferred to the fallopian tubes or those missing information on the number of embryos transferred were excluded ( Figure 1 ). Using ICD-9 codes from birth hospitalization records and birth certificate data, the following infant health outcomes were assessed: preterm birth (< 37 weeks), small for gestational age (SGA), low birth weight (LBW; < 2500 grams), large for gestational age (LGA), neonatal mortality during birth hospitalization (defined as death between day of delivery and the last day of hospitalization), prolonged hospital stay (for infants ≥ 35 weeks GA, > 3 days for infants born vaginally and > 5 days for those born by cesarean section). The outcome of birth defects, obtained from the MBDMP, was categorized as chromosomal or non-chromosomal according to a prior MOSART study( 18 – 20 ) Infants transferred to higher levels of hospital care after birth were included. Specific conditions by the following systems were also assessed: infectious disease, cardiovascular, respiratory, gastrointestinal/nutrition, neurologic, and hematologic. (ICD-9 codes shown in Appendix 1). Length of gestation was calculated based on clinical estimates by first trimester ultrasound and when those were missing, the estimated date of last menstrual period calculated from day of transfer. Birthweight z-scores were calculated to evaluate adequacy of weight-for-age using Massachusetts population-based standards and modeled as continuous and categorical variables. We generated sex-, race/ethnicity-, and gestation-specific birthweight means and standard deviations using Massachusetts data for live births from 2004–2013. Infants with z-scores of ≤1.28 (below the 10 th percentile for gestation) were classified as SGA and those with z-scores of ≥1.28 (above the 10 th percentile for gestation) were classified as LGA. Fresh and frozen embryo transfers were defined on the basis of the SART CORS fields “Fresh” (yes/no) and “Frozen Embryo” (yes/no). Additional independent variables obtained from the birth certificates included maternal age at delivery, race/ethnicity, education, marital status, parity, and insurance status at birth. Other variables were obtained from both birth certificate and hospital discharge records and included chronic and pregnancy-induced hypertension, non-gestational and gestational diabetes. Information on number of embryos transferred, diminished ovarian reserve, and ovarian dysfunction were obtained from SART CORS. We compared birth outcomes of infants born to mothers who underwent fresh versus FET using the Chi-square statistics (alpha = 0.05). Generalized estimating equations were used to assess the independent association between the fresh and FET groups and adverse health outcomes, controlling for maternal age at delivery, race/ethnicity, education, insurance status at birth, pre-existing diabetes, gestational diabetes, pre-existing hypertension, pregnancy-induced hypertension, parity, gender, diminished ovarian reserve, ovarian dysfunction and number of embryos transferred. For the outcome of prolonged infant hospital stay among infants born ≥35 weeks GA, we also adjusted for maternal length of hospital stay. All analyses were performed using the SAS software, version 9.3 (SAS Institute, Cary, NC).

In conclusion, this population-based study of infant outcomes after fresh versus FET demonstrates that infants conceived by FET are of higher birthweight with higher odds of LGA status and lower odds of SGA status, which has been seen in prior studies. New from this work are the results that infants in the FET group have greater odds for infectious disease, respiratory, and neurologic conditions than those in the fresh embryo transfer group. These risks should be considered in the decision-making process of selecting fresh versus FET. Future studies focusing on biologic mechanisms of these risks as well as longer term medical and developmental follow-up of these infants are needed.

In this population-based analysis of health outcomes of infants conceived by fresh versus FET, we determined that compared with infants conceived from fresh embryo transfer, those born by FET have higher birthweight but increased odds of infectious disease, respiratory and neurologic abnormalities, after adjusting for maternal and infant covariates. Our finding of higher birthweight among FET infants is consistent with previously published observational studies.( 10 , 21 , 22 ) In a retrospective cohort study of approximately 4000 infants conceived with FET, Zheng et al found that full term singletons born after FET had higher birthweights, higher risk for being LGA and lower risk for being SGA.( 23 ) In addition, as in our study, Zheng et al reported no difference in the incidence of congenital malformations. While these prior studies, did not assess clinical conditions beyond preterm birth, birthweight, and birth defects, a recent meta-analysis of 26 studies did include admission to a neonatal intensive care unit (NICU) as one of their outcome measures and found no difference between the fresh and FET groups.( 8 ) However, reasons for NICU admission were not included in these studies, and thus our study, is the first, to our knowledge, to report adverse infectious disease, respiratory, and neurologic outcomes among infants conceived by FET. The biologic mechanisms that underlie these birth outcomes are not well understood. With respect to lower birthweights in infants conceived by fresh embryo transfer, it has been hypothesized that because these women undergo controlled ovarian hyperstimulation for development of multiple oocytes, there may be an increase in abnormal implantation and placentation, causing altered blood flow to the fetus and impaired fetal growth.( 24 ) Thus the uterine environment in FET cycles which is generally induced using estrogen and progesterone, may be more conducive to normal fetal development. In mouse and porcine models, several studies have demonstrated altered genetic and epigenetic changes due to vitrification as well as to differing culture media which may explain the higher birthweights seen in infants conceived by FET.( 25 – 28 ) For instance, Yao et al demonstrated decreased expression of the Grb10 gene in blastocysts after vitrification compared to controls.( 27 ) Grb10 is an imprinted gene that encodes growth factor receptor-bound protein 10, which is an inhibitor in the insulin and/or insulin-like growth factors signal transduction pathway. Thus, decreased expression of this gene may explain the higher birthweights seen in FET infants. While these developmental biologic discoveries continue to be made in animal models, and the roles of some of these genes are understood in humans, the ways in which these genetic alterations may lead to disruptions in various organ systems in the developing fetus is not known. It is possible that the higher odds of infectious disease, respiratory and neurologic conditions among infants in the FET group could be secondary to their higher incidence of LGA status. In an observational cohort study of over 115,000 uncomplicated term deliveries, 8.3% of infants were LGA and compared to their appropriate-for-gestational (AGA) age counterparts, LGA infants had higher risk for traumatic composite neonatal morbidity that included any of the following: Apgar score < 5 at 5 minutes, seizures, cardiopulmonary resuscitation within the first 24 hours, ventilator support within 24 hours, hypoxic ischemic encephalopathy, osseous fracture, intracranial hemorrhage, brachial plexus palsy, facial nerve palsy, or death before discharge.( 29 ) However, it is well-known that SGA infants also have greater risk for adverse birth outcomes compared to their AGA counterparts and given the higher incidence of SGA status among infants born by fresh embryo transfer, we would not have anticipated a notable adjusted difference in the incidence of organ system conditions between the fresh versus FET group. There are several limitations to our study. First, the MOSART database is comprised of vital statistics and hospital-level administrative discharge codes, and thus some parameters, such as body mass index (BMI), which has been shown to associated with adverse infant outcomes, ( 30 – 32 ) were not available. Within the SART CORS data there were limited data on both ovarian hyperstimulation syndrome and the use of pre-implantation genetic testing particularly in the early years of study, limiting our ability to evaluate the influence of these factors on our study population. Data related to pregnancy course such as fetal growth and utero-placental doppler ultrasound data were not available. Data on paternal health including infertility, and BMI were also unavailable. Finally, findings may not be generalizable since our cohort included only Massachusetts-resident births. Moreover, as is that case with observational studies, in general, while we demonstrate a significant association between fresh and FET and infant outcomes, we are unable to establish a direct causal relationship between implantation of fresh versus frozen embryos and infant outcomes. Despite these limitations, this population-based study is the first, to our knowledge, to investigate infant outcomes beyond prematurity, birth weight and birth defects. The linkage of vital statistics data, hospital discharge records for mother and infant, and ART data from SART CORS allows for a methodologically robust approach to defining exposure, outcomes, and key covariates.

The use of assisted reproductive technology (ART), defined as treatments in which both eggs and embryos are handled,( 1 , 2 ) has increased tremendously in the U.S. and worldwide. From 2000 to 2013, ART cycles doubled in the US from 99,629 to 190,773 and more than 5 million babies now have been born from ART worldwide, half within the past 6 years.( 3 – 6 ) In 2015, the proportion of all births across U.S. states from ART ranged from 0.7% to 4.5%.( 7 ) A fresh ART cycle involves stimulation of a woman’s ovaries with gonadotropins, extraction of her eggs, fertilization of her eggs with sperm in the laboratory, and then transfer of the resulting embryo(s) into the uterus. Frozen embryo transfer (FET) ART cycles involve the transfer of embryos that were previously been frozen. Recent evidence has demonstrated higher live birth-rates, higher birth weight, and a smaller proportion of small for gestational age babies, among infants born from FET as compared with fresh transfers.( 8 – 10 ) and this has resulted in increased usage of frozen cycles.( 11 ) However, more recent literature has called into question whether these favorable outcomes are, in fact, generalizable to the heterogeneous population of women who undergo ART. A recent meta-analysis that evaluated perinatal outcomes from frozen versus fresh embryo transfer concluded that data did not support the indiscriminate use of elective FET.( 12 ) In addition, a recent multi-center randomized clinical trial of women undergoing fresh versus FET excluded women with polycystic ovarian syndrome (PCOS), the most common reason for ovarian hyperstimulation syndrome, and found that the live birth-rate did not differ between fresh and frozen transfer.( 13 ) While the prevalence of FET continues to rise,( 14 ) the direction and magnitude of association between infant health outcomes and fresh transfer versus FET is still not clear. Furthermore, research on infant health outcomes beyond live birth rate and infant birth weight is limited. Thus, the objective of this study was to use an extensive Massachusetts data system containing ART clinical data, vital records and hospital data, to extend our understanding of infant health following fresh and FET by comparing a wide array of neonatal health outcomes that went beyond birth weight and gestational age.