Predicting outcome after emergent cerclage using classification tree analysis

Area of Science:

• Maternal-Fetal Medicine
• Obstetrics
• Predictive Analytics in Healthcare

Background:

• Emergent cerclage is a procedure used to prevent preterm birth.
• Predicting delivery outcomes after emergent cerclage is crucial for clinical management.
• Cervical change in the second trimester necessitates timely intervention.

Purpose of the Study:

• To develop a predictive model for gestational age at delivery following emergent cerclage.
• To identify key clinical factors influencing delivery timing after cerclage placement.
• To provide a tool for better patient and physician information regarding outcomes.

Main Methods:

• Retrospective analysis of 116 women undergoing emergent cerclage (1980-2000).
• Hierarchically optimal Classification Tree Analysis (CTA) employed for prediction.
• Models developed to predict delivery before 24 weeks, 24-27 6/7 weeks, and after 27 6/7 weeks.

Main Results:

• Delivery before 24 weeks predicted by prolapsed membranes and gestational age at cerclage.
• Delivery between 24-27 6/7 weeks primarily predicted by parity.
• Delivery after 27 6/7 weeks best predicted by cervical dilation/length, prolapsed membranes, and parity.
• The model predicting delivery after 27 6/7 weeks showed the highest accuracy.

Conclusions:

• Classification Tree Analysis (CTA) can effectively model outcomes after emergent cerclage.
• Specific clinical variables accurately predict different gestational age windows for delivery.
• The developed predictive model offers valuable insights for clinical decision-making and patient counseling.