Ultrasound examination of the cervix for predicting labor induction success: failed validation in a routine clinical setting of a successful previous pilot study

Background Induction of labor (IL) involves an overload of work in hospitals, as well as increased intervention. Traditionally, the Bishop score (BS) has been used to predict the outcome of a IL, but there is a growing interest in studying the predictive capacity of ultrasound variables. Objective Validate a pilot predictive model performed by a single observer (Alvarez-Colomo C), based on clinical parameters and ultrasound parameters, that showed a significant association with the IL result, obtaining a correct prediction of vaginal delivery in 82.8%, with 15% false positive (FP). This validation was carried out under the usual conditions of clinical practice by four observers without distinction. Methods A prospective, observational study was conducted between September 2010-July 2012, recruiting 231 single pregnancies (Group 2), who were to initiate the IL process, according to the methodology and inclusion criteria of the Alvarez-Colomo study (Group 1151 patients). The outcome variable was the method of delivery. Results Only fetal head-perineal distance (FHPD), cervical length (CL) and BS showed significant association with the result of IL. After applying the logistic regression equation of the pilot study, the model developed by these four observers reached a predictive capacity of 70.74% (FP = 20%). Clinical characteristics were similar in both groups. Statistically significant differences were found between the two groups for: FHPD, posterior cervical angle (PCA) and funnel existence. Conclusion It has not been possible to validate the mathematical model of Alvarez's study in the daily conditions of clinical practice, probably due to differences in the ultrasound measurement of FHPD.