Revisiting the progesterone to oocyte ratio.

OBJECTIVE To critically evaluate the P to oocyte (O) ratio (P/O) in the prediction of live birth in assisted reproductive technology (ART) cycles. DESIGN Retrospective cohort study. SETTING Not applicable. PATIENT(S) A total of 7,608 fresh autologous ART ET cycles. INTERVENTION(S) None. MAIN OUTCOME MEASURE(S) Live birth. RESULT(S) Generalized estimating equation (GEE) models and receiver operating characteristic curves assessed the ability of P, O, and the P/O ratio to predict live birth. In univariate GEE models, P, O, and P/O were each associated with live birth. However, in multivariate GEE models, the P/O ratio was not associated with live birth, but P alone was. This suggested that converting P and O into a ratio of P/O was not more helpful than the two independent variables themselves. Measures of overall model fit further suggested that P/O did not increase the predictive ability of the model over P and O alone. Receiver operating characteristic curves using incremental predictors further demonstrated that the P/O provided no incremental improvement in predicting live birth over P and O separately. CONCLUSION(S) These data suggest that P and O have utility in prediction modeling but demonstrate that additional oocytes were not protective from the negative association of P with live birth. There was no incremental improvement related to the P/O ratio specifically for predicting live birth over each variable independently.