Fetal Growth Biometry as Predictors of Shoulder Dystocia in a Low-Risk Obstetrical Population

Abstract

Objective: This study aimed to evaluate fetal biometrics as predictors of shoulder dystocia (SD) in a low-risk obstetrical population.

Study design: Participants were enrolled as part of a U.S.-based prospective cohort study of fetal growth in low-risk singleton gestations (n = 2,802). Eligible women had liveborn singletons ≥2,500 g delivered vaginally. Sociodemographic, anthropometric, and pregnancy outcome data were abstracted by research staff. The diagnosis of SD was based on the recorded clinical impression of the delivering physician. Simple logistic regression models were used to examine associations between fetal biometrics and SD. Fetal biometric cut points, selected by Youden's J and clinical determination, were identified to optimize predictive capability. A final model for SD prediction was constructed using backward selection. Our dataset was randomly divided into training (60%) and test (40%) datasets for model building and internal validation.

Results: A total of 1,691 women (98.7%) had an uncomplicated vaginal delivery, while 23 (1.3%) experienced SD. There were no differences in sociodemographic or maternal anthropometrics between groups. Epidural anesthesia use was significantly more common (100 vs. 82.4%; p = 0.03) among women who experienced SD compared with those who did not. Amniotic fluid maximal vertical pocket was also significantly greater among SD cases (5.8 ± 1.7 vs. 5.1 ± 1.5 cm; odds ratio = 1.32 [95% confidence interval: 1.03,1.69]). Several fetal biometric measures were significantly associated with SD when dichotomized based on clinically selected cut-off points. A final prediction model was internally valid with an area under the curve of 0.90 (95% confidence interval: 0.81, 0.99). At a model probability of 1%, sensitivity (71.4%), specificity (77.5%), positive (3.5%), and negative predictive values (99.6%) did not indicate the ability of the model to predict SD in a clinically meaningful way.

Conclusion: Other than epidural anesthesia use, neither sociodemographic nor maternal anthropometrics were significantly associated with SD in this low-risk population. Both individually and in combination, fetal biometrics had limited ability to predict SD and lack clinical usefulness.

Key points: · SD unpredictable in low-risk women.. · Fetal biometry does not reliably predict SD.. · Epidural use associated with increased SD risk.. · SD prediction models clinically inefficient..