(951) Prediction of Spontaneous Labor After 39 Weeks in Low-Risk Nulliparas: Secondary Analysis of ARRIVE Trial

Individualized predictions of spontaneous labor after 39 weeks could improve informed shared decision-making around elective induction. Our objective was to understand clinical factors associated with spontaneous labor after 39 weeks in a low-risk, nulliparous population eligible for elective induction and develop and validate a clinical prediction model.

Study Design:

Secondary analysis of the ARRIVE trial including participants in the expectant management arm undelivered at 39w0d and excluding those delivered via induction (elective or medically-indicated) prior to 40w5d (when elective inductions were allowed by study protocol). The primary outcome was spontaneous labor, defined as admission for contractions or augmentation for rupture of membranes. Candidate predictors were identified and the cohort was randomly split into a 70% training and 30% validation set. Discrimination was assessed using AUC and calibration was evaluated using calibration plots.

Results:

The final cohort included 2231 individuals; 1614 (72.3%) experienced spontaneous labor. In the final model, younger age, lower BMI, less weight gain, non-smoking status, lower birthweight, higher Bishop score, and any labor-related triage visit were significantly associated with spontaneous labor (Table 1). The model demonstrated moderate discrimination (AUC = 0.749 training, 0.711 validation; Figure 1). The calibration plot was well calibrated across all predicted probabilities. In the validation set, those in the top quartile ( >87% prediction) went into spontaneous labor 90.2% of the time, compared to 65.2% for all others (P< 0.01). Results were similar when triage visits for labor-related complaints were excluded, as this information would not be available prospectively.

Conclusion:

After 39 weeks, 72% of individuals went into spontaneous labor by 40w5d if not induced prior. Commonly collected clinical variables can be used to refine predictions of spontaneous labor with moderate discrimination. These individualized predictions may better inform patients considering elective induction though need to externally validated.