Development and validation of a predictive model to identify the active phase of labor

Abstract

BackgroundThe diagnosis of the active phase of labor is a crucial clinical decision, thus requiring an accurate assessment. This study aimed to build and to validate a predictive model, based on maternal signs and symptoms to identify a cervical dilatation ≥4 cm.MethodsA prospective study was conducted from May to September 2018 in a II Level Maternity Unit (development data), and from May to September 2019 in a I Level Maternity Unit (validation data). Women with singleton, term pregnancy, cephalic presentation and presence of contractions were consecutively enrolled during the initial assessment to diagnose the stage of labor. Women < 18 years old, with language barrier or induction of labor were excluded. A nomogram for the calculation of the predictions of cervical dilatation ≥4 cm on the ground of 11 maternal signs and symptoms was obtained from a multivariate logistic model. The predictive performance of the model was investigated by internal and external validation.ResultsA total of 288 assessments were analyzed. All maternal signs and symptoms showed a significant impact on increasing the probability of cervical dilatation ≥4 cm. In the final logistic model, “Rhythm” (OR 6.26), “Duration” (OR 8.15) of contractions and “Show” (OR 4.29) confirmed their significance while, unexpectedly, “Frequency” of contractions had no impact. The area under the ROC curve in the model of the uterine activity was 0.865 (development data) and 0.927 (validation data), with an increment to 0.905 and 0.956, respectively, when adding maternal signs.The Brier Score error in the model of the uterine activity was 0.140 (development data) and 0.097 (validation data), with a decrement to 0.121 and 0.092, respectively, when adding maternal signs.ConclusionOur predictive model showed a good performance. The introduction of a non-invasive tool might assist midwives in the decision-making process, avoiding interventions and thus offering an evidenced-base care.