A model-based prediction for transvaginal ultrasonographic identification of early intrauterine pregnancy.

Abstract

Our objective was to develop a mathematic model for the prediction of transvaginal ultrasonographic identification of intrauterine gestation as a function of human chorionic gonadotropin titer or gestational age. In this prospective, descriptive study normal intrauterine pregnancies of 31 patients from the Yale in vitro fertilization and embryo transfer program and infertility clinic were studied. Logistic regression analysis was used to develop a probability curve of transvaginal ultrasonographic detection of intrauterine pregnancy as a function of human chorionic gonadotropin titer or gestational age with a 5 MHz vaginal transducer. A model-based prediction was constructed. It revealed that nearly all intrauterine sacs should be identified by a human chorionic gonadotropin titer of 3000 mIU/ml (first international reference preparation) or by a gestational age of 37 days on the basis of the 5 MHz vaginal transducer. A model-based prediction for transvaginal ultrasonographic identification of early intrauterine pregnancy is presented that can be readily adapted by individual institutions using their own ultrasonographic equipment and human chorionic gonadotropin assay.