Construction and validation of a prediction model to minimize twin rates at preserved high live birth rates after IVF

Research question: Elective single-embryo transfer (eSET) at blastocyst stage is widely used to reduce the frequency of multiple pregnancies after IVF. There are, however, concerns about increased risks for the offspring with prolonged embryo culture. Is it possible to select embryos for transfer at the early cleavage stage and still achieve low twin rates at preserved high live birth rates? Design: A prediction model (PM) was developed to optimize eSET based on variables known 2 days after oocyte retrieval (fresh day 2 embryo transfers; double-embryo transfers 1999-2002 (n=2846) and SET 1999-2003 (n=945); n total=3791). Seventy-five variables were analysed for association with pregnancy chance and twin risk and combined for PM construction. This PM was validated in 2004-2016 including frozen-thawed transfers (FET), to compare cumulative live birth rate (CLBR) and twin rate before (1999-2002 fresh embryo transfers plus FET from the same oocyte retrievals until the end of 2007, n=3495) and after (2004-2011 fresh embryo transfers plus FET from the same oocyte retrievals until the end of 2016, n=11195) implementing the model. Results: The PM was constructed from four independent variables: female age, embryo score, ovarian sensitivity and treatment history. The calibration, i.e. the fit of observed versus predicted results, was excellent both at construction and at validation. Without compromising CLBR, twin rate was reduced from 25.2% to 3.8%, accompanied by profound improvements in perinatal outcome. Conclusion: The results provide the first successful construction, validation and impact analysis of a day 2 transfer PM to reduce multiple pregnancies.