P-721 Prediction model of pregnancy after intrauterine insemination: accessibility, relevance for clinical use and patient counseling

Abstract

Abstract Study question The aim of this study is to develop a prognostic model for the outcome of intrauterine insemination that is useful for daily clinical practice and to develop an app Summary answer Our predictive model is able to make a good distinction between patients with a good or poor pregnancy chance What is known already IUI is a common, accessible and easily accepted method of treating infertile couples with a success rate varying largely between 5% and 70%. The factors that contribute to the successful IUI rate are numerous including the woman's age, type of infertility, the number of follicles obtained by ovarian stimulation, the estradiol concentration on the day of human chorionic gonadotrophin administration, the sperm parameters (concentration, motility), duration of infertility, endometrial thickness, the number of IUI cycles. Several studies have already tried to establish a success rate formula, but no prediction model is largely used. Study design, size, duration This retrospective cohort study has been conducted at an university level fertility clinic from Oradea, Romania. Patients eligible to participate in the study were 426 infertile couples who underwent IUI treatment between January 2015 and October 2020. Participants/materials, setting, methods The 426 couples were classified into three groups: 1. male infertility factor 2. female infertility factor (minimal/mild endometriosis, ovulation disorders, tubal disease, with a single patented Fallopian tube), 3. Unexplained infertility. All couples underwent IUI with/or without ovarian stimulation and we documented several factors: age, type of infertility, infertility duration, number of procedures, ovarian stimulation, number of follicles, endometrial thickness, ovulation induction, day of induction, associated pathology, tubal patency, post wash sperm count. Main results and the role of chance Our pregnancy rate was 10.3%, according to this the couples were divided intro two groups the negative and the positive result group. We found 4 factors to be significantly associated to a positive result: age of the partners (p-value = 0.012), infertility duration (p-value = 0.018), endometrium thickness (p-value = 0.035) and semen preprocess concentration (p-value = 0.046). We calculated a probability function for a woman to get pregnant after IUI procedure: Prob = 1/(1+e-z) Z = -3.36998 + 0.364*Endometrium (mm) – 0.04567*Female Age (years) – 0.049956*Age of partner (years)– 0.17299*Duration of infertility (years) + 0.018562*prewash count (mil) – 0.9933*A (motility) + 2.29085*B(motility)+ 0.03203*post wash count For example, a patient with a score of 1.507 will have a probability of 81.86% of becoming pregnant. Our predictive model in general has a performance that is close to 0.7 AUROC -0.687 (AUROC between 0.7-0.8 has a fair performance) with a value p-0.001 (high statistical significance, CI confidence interval - 99.9%), meaning it is able to make a good distinction between couples with a good pregnancy chance and those with a poor pregnancy chance after IUI. Limitations, reasons for caution In statistical models fertility treatment chances are never 0 or 100%, and the outcome of IUI is a subject to chance and from this point of view the trick would be in repeating the cycles. Another limitation is the small cohort sample and the fact that it needs external validation. Wider implications of the findings Developing an application program that could be used easily by the clinicians in their daily clinical practice and also by the patients in order to better understand their chances at a successful procedure of assisted human reproduction that could also mean the next step after IUI, the in vitro fertilization. Trial registration number not applicable