A new approach to estimating cumulative live birth rates using predicted outcomes of all embryos

Objective: Current success estimators for IVF do not take into account unused embryos and may under-predict cumulative live birth rate (CLBR) for some patients. We aimed to develop a new approach that considers predicted outcomes of all embryos.

Materials and Methods: We analyzed data from 2014-2019 in the Society for Assisted Reproductive Technology Clinical Outcomes Reporting System (SART CORS), filtering for first autologous IVF cycles and linked frozen transfers. For reference, we trained a logistic regression model on 250,206 cycles to predict CLBR (probability of at least one live birth) using age, BMI, and diagnosis. We then developed a new approach. First, a negative binomial regression model was trained on 121,509 retrievals to predict the number of total blastocysts (transferred plus cryopreserved) using age, AMH, BMI, and diagnosis. Then, a logistic regression model was trained on 60,393 non-PGT single-blastocyst transfers to predict per-transfer live birth rate using age, BMI, AMH, transfer number, and diagnosis. CLBR was then estimated as: where N is the number of blastocysts and p_i is the predicted probability of live birth for transfer i. We compared this approach to a similar model for euploid blastocysts (using published euploidy rates by age [1]) and per-transfer success rates of PGT transfers.

Results: Compared to the reference model, our model provides similar predictions for CLBR between 0% - 40%. In this range, the average number of unused embryos is 0.6 to 1.5 per cycle. Above 40%, the average number of unused embryos is 1.9 to 7.3, and our model predicts CLBR 9-19% higher than the reference model.

Conclusions: Our model provides CLBR predictions that are comparable to other estimators, but higher for better-prognosis patients. Future work will focus on understanding why certain embryos have remained not transferred, further improving the model, and expanding it for other cycle types.

Impact Statement: We have developed a new approach for estimating CLBR for patients considering IVF that accounts for unused embryos. Results suggest that current estimators may be under-predicting success rates for higher prognosis patients.

References: [1] Goldman RH, Racowsky C, Farland L, et al. A cost of a euploid embryo identified from preimplantation genetic testing for aneuploidy (PGT-A): a counseling tool. J Assist Reprod Genet. 2018 Sep; 35(9): 1641-1650.