Prediction of Fertilization Disorders in the In Vitro Fertilization/Intracytoplasmic Sperm Injection: A Retrospective Study of 106,728 Treatment Cycles

PURPOSE: This study aimed to develop a risk prediction of fertilization disorders during the in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI). METHODS: A retrospective study was performed with 106,728 fresh embryo IVF/ICSI cycles from 2009 to 2019. Basic characteristics of patients, clinical treatment data, and laboratory parameters were involved. The associations between the selected variables and risks for low fertilization rate (LFR) and total fertilization failure (TFF) were investigated. Ordinal logistic regression and the receiver operating characteristic curves (ROCs) were used to construct and evaluate the prediction models. RESULTS: A total of 97,181 controls, 4,343 LFR and 5,204 TFF cases were involved in this study. The model based on clinical characteristics (the ages of the couples, women’s BMI, types of infertility, ART failure history, the diminished ovarian reserve, sperm quality, insemination method, and the number of oocytes retrieved) had an AUC of 0.743 for TFF. The laboratory model showed that primary infertility, ART failure history, minimal-stimulation cycle/natural cycle, numbers of oocyte retrieved < 5, IVF, and Anti-Mullerian hormone (AMH) level < 1.1ng/ml are predictors of TFF, with an AUC of 0.742. CONCLUSION: We established a clinical and a laboratory prediction model for LFR/TFF. Both of the models showed relatively high AUCs.