Frozen versus fresh embryo transfer on perinatal outcomes-do endometrial preparation methods matter?

study questionAre there differences between perinatal outcomes following frozen versus fresh embryo transfer in IVF, and do endometrium preparation methods contribute to the differences? SUMMARY ANSWER: Compared with fresh embryo transfers, frozen transfers, regardless of hormone replacement treatment or natural treatment cycles, were associated with lower chances of preterm birth, low birth weight, and small for gestational age, but higher chances of caesarean section, high birth weight, and large for gestational age. WHAT IS KNOWN ALREADY: Frozen embryo transfer has been increasing over the past two decades, but its associated perinatal risks and underlying reasons remain controversial. Most existing observational studies have not accounted for multiple cycles from the same couple or known patients' characteristics or treatment protocols, such as endometrial preparation methods, in the analysis. Existing birthweight centile charts are likely to underestimate intra-uterine growth restriction due to the inclusion of deliveries following obstetric interventions. STUDY DESIGN SIZE DURATION: This multicentre retrospective cohort study used routinely collected clinical data of 8081 women undergoing IVF who gave birth to 9243 babies (6125 from the frozen and 3118 from fresh transfer cycles) in 12 IVF clinics across two states in Australia between 2015 and 2021, with follow-up data up to 2023. PARTICIPANTS/MATERIALS SETTING

methodsIndividuals undergoing autologous-oocyte IVF cycles who had singleton live births were included. An individual could have multiple treatment cycles included. Perinatal outcomes included preterm birth, low/high birth weight, small-/large-for-gestational-age, and caesarean section. The birthweight percentiles were calculated based on the New Australian birthweight centiles, where interventions-initiated births were excluded. Multivariable Poisson regression with robust variance was used to analyse all outcomes. Generalized estimating equations (GEEs) were used to account for the cluster effects of multiple embryo transfer cycles of the same individual. Adjusted risk ratios with 95% confidence intervals (CIs) were reported for each outcome. The adjusted model accounted for potential confounding factors, including female age, parity, semen source, ovulatory disorders, preimplantation genetic testing for aneuploidy, blastocyst transfer, number of embryos transferred, and site. In subgroup analysis, frozen transfers with different endometrial preparation methods (hormone replacement and natural cycles) were compared to the fresh transfers. MAIN RESULTS AND THE ROLE OF CHANCE: Compared with births in the fresh group, births in the frozen transfer group were less likely to be preterm (8.9% vs 13.7%, adjusted risk ratio (aRR) 0.66 0.58-0.76), low birth weight (5.3% vs 8.5%, aRR 0.66, 0.55-0.78), and small for gestational age (4.2% vs 7.9%, aRR 0.62, 0.51-0.75), but more likely to be caesarean section (57.5% vs 50.4%, aRR 1.14, 95% CI 1.09-1.19), high birth weight (10.1% vs 7.0%, aRR 1.43, 1.21-1.68), and large-for-gestational-age (20.0% vs 13.6%, aRR 1.37, 1.23-1.53). The differences remained consistent when comparing hormone replacement or natural cycle frozen transfers to fresh transfers. LIMITATIONS REASONS FOR CAUTION: The retrospective nature introduces inherent challenges of residual confounding. WIDER IMPLICATIONS OF THE

findingsThe results indicate that the differences in birth weight related to perinatal outcomes between frozen versus fresh embryo transfer cycles are unlikely due to the differences in the use of endometrial preparation protocols in frozen cycles, but may be attributed to other factors including impaired endometrial characteristics in fresh cycles or the process of vitrification and warming embryos in frozen cycles. STUDY FUNDING/COMPETING INTERESTS: H.Z. was supported by a Monash Research Scholarship. N.A. was supported by an Australian Government Research Training Program Scholarship. R.W. was supported by an NHMRC Emerging Leadership Investigator grant (2009767). D.Z. and M.G. reported that they are employees of the company from which the data were analysed for this study. No conflicts of interest from other authors were reported. TRIAL REGISTRATION NUMBER: N/A.