Predictive modeling of labor epidural catheter replacement or conversion to spinal or general anesthesia: a multicenter retrospective cohort study (2017-2024).

BACKGROUND

Labor epidural analgesia may be inadequate during labor or for cesarean delivery anesthesia. No prediction model identifies cases of labor epidural analgesia with increased risk for epidural catheter replacement or conversion to spinal or general anesthesia.

METHODS

We conducted a retrospective cohort study of patients receiving labor epidural analgesia initiated with standard epidural or dural puncture epidural technique in a large, multihospital healthcare system between 2017 and 2024. Demographic, comorbidity, vital-sign, and procedural data were extracted from the electronic health record. The composite outcome was epidural catheter replacement, subsequent spinal anesthesia, or advanced airway placement indicating general anesthesia for cesarean delivery within the same anesthesia encounter. Statistical and machine-learning models were developed; one hospital was held out for external validation.

RESULTS

The cohort included 82,823 cases of epidural catheter placement among 65,419 unique patients; 5010 cases (6.0%) met the composite outcome. Logistic regression and XGBoost performed similarly on external validation (area under the receiver operating characteristic curve, 0.67). Important predictors included prior replacement or conversion during the same encounter, end-of-pregnancy body mass index, pre-placement pain score, prior cesarean delivery, parity, and delivery hospital. Decision-curve analysis and Monte Carlo simulation suggested possible utility for screening and triage when anesthesia staffing is limited.

CONCLUSIONS

We developed and externally validated Risk Estimation for Anesthesia Catheter Technique replacement or conversion in Obstetrics (REACT-OB) to estimate the risk of this outcome. Prospective validation is required before clinical use.