Evidence based quantitative prediction of risk of multiple gestation resulting from the transfer of multiple embryos

Abstract

To develop and evaluate a quantitative model for prediction of risk of multiple gestation after transfer of multiple embryos that can be used with a mobile device application. Cross sectional analysis of clinic based data. We developed a three-step model to quantitatively predict the outcomes of multiple embryo transfers. We used data from three years of autologous cleavage stage and blastocyst embryo transfers at a single academic fertility center. The data set consisted of 760 embryo transfers of a total of 1928 embryos. First, a training set of data was used to develop a best fit model. The data consists of the number of live births that resulted from an embryo transfer, maternal age at oocyte retrieval (<35 years, 35-37 years, 38-40 years, or 41+ years), number of embryos, embryo stage (cleavage or blastocyst), and embryo quality determined by the embryologist (good or fair/poor). Based on the training data a rate of live birth was calculated for each of 16 embryo categories through the use of a specifically designed computer program. Second, the same computer program was utilized to make a quantitative assessment about the likelihood that two embryos transferred concurrently will both implant more often than would be expected by chance alone. This accounts for universal factors that affect all embryos transferred concurrently. Third, means and standard deviations of outcomes for a test sample of embryo transfers were predicted using the best fit model and random number generation. We tested the model with six groups of multiple embryo transfers (Table 1). Z-tests were used to compare actual to predicted outcomes. The predicted and actual rates of multiple birth for each of six embryo transfer groups is shown in Table 1. The differences between predicted and actual rates of multiple birth were not statistically significant and the standard errors were normally distributed on a quantile-quantile plot.Table 1Predicted and actual rates of multiple birth (multiple deliveries / total deliveries)embryos transferredless than 38 years38 years & greater2 blastocystspredicted (95% CI)36% (23-50%)7% (0-30%)actual30% (p = 0.37)17% (p=0.30)2 cleavage stagepredicted (95% CI)9% (0-25%)16% (5-29%) *actual16% (p = 0.31)3 cleavage stagepredicted (95% CI)34% (18-52%)19% (p=0.59)actual29% (p=0.51)4 or 5 cleavage stagepredicted (95% CI)n/a21% (8-36%)actual17% (p=0.59)∗ this group included transfer of 2 or 3 cleavage stage embryos Open table in a new tab ∗ this group included transfer of 2 or 3 cleavage stage embryos Current recommendations for number of embryos to transfer are based on expert opinion. This model can be used with a mobile device application at the point of care for evidence based quantitative prediction of risk of multiple gestation after transfer of multiple embryos.