First-Trimester Prognosis When an Early Gestational Sac is Seen on Ultrasound Imaging: Logistic Regression Prediction Model.

Abstract

To determine the factors that jointly and independently affect first-trimester outcome of very early intrauterine pregnancies (those whose sonogram shows a gestational sac with no identifiable yolk sac or embryo) and develop a mathematical model and Web-based calculator that computes prognosis based on these factors. Our study population included 590 very early pregnancies scanned between January 1, 2012, and June 30, 2018, with known outcomes (live or spontaneous loss) at 14 weeks. We recorded patient age, mean sac diameter (MSD), human chorionic gonadotropin (hCG) rise, and presence/absence of: vaginal bleeding, history of infertility, prior miscarriage, and pregnancy via assisted reproductive technology. We assessed the correlation between each of these factors and outcome and performed stepwise logistic regression to determine the subset that independently correlated with outcome. Patient age, MSD, hCG rise, vaginal bleeding, history of infertility, and assisted reproductive technology pregnancy were significantly correlated with outcome (P < .05, t test for age and MSD, χ2 for the others). Stepwise logistic regression identified age, MSD, hCG rise, and vaginal bleeding as the subset of factors that independently predicted outcome. The regression model's area under the receiver operating characteristic curve was 0.823. We incorporated the regression model into a Web-based calculator (https://tinyurl.com/Prognosis-PD) that predicts the outcome of an early intrauterine pregnancy based on these 4 key variables. The prognosis of very early intrauterine pregnancies is related to several clinical, biochemical, and sonographic factors. The factors that independently correlate with first-trimester outcome are patient age, MSD, hCG rise, and vaginal bleeding. The logistic regression model predicts outcome based on these variables.