Clinical Risk Factors Do Not Predict Shoulder Dystocia

To compare 2 different risk factor models for the prediction of shoulder dystocia.

We performed a retrospective study of women with vaginal deliveries at a single institution over an 8-year period. Two distinct multivariable logistic regression models were used to evaluate the occurrence of shoulder dystocia: a traditional model used information based on birthweight and macrosomia, and a clinical model used information based on esti-mated fetal weight and suspected macrosomia.

Of the 13,998 deliveries analyzed, there were 221 cases of shoulder dystocia (1.6%). In addition to the macrosomia or suspected macrosomia variables, the final models included prolonged second stage of labor, diabetes status, and oxytocin use. Neither model was highly sensitive or highly specific, and neither demonstrated a cutoff threshold that yielded a clinically viable PPV.

Despite the presence of 1 or more risk factors for shoulder dystocia, its occurrence remains largely an unpredictable clinical event.