Macrosomia is the only reliable predictor of shoulder dystocia in babies weighing 3.5 kg or more.

OBJECTIVE To determine if shoulder dystocia can be predicted in babies born weighing 3.5 kg or more. STUDY DESIGN A case-control study nested in a perinatal database of 899 mothers and their babies who weighed 3.5 kg or more. All were term pregnancies and delivered vaginally. A case was defined as any baby that encountered shoulder dystocia at delivery. Controls were deliveries over the same period that were not complicated by shoulder dystocia. A logistic regression model was created with macrosomia, parity, previous delivery of more than 3.5 kg, diabetes in pregnancy, prolonged labor, prolonged second stage and instrumental delivery as the independent variables. The adjusted odds ratio and the receiver operator characteristics (ROC) curves were used to see if these variables, both individually and as a model, were associated with or were discriminative enough to predict shoulder dystocia; an ROC curve of more than 0.7 showing good prediction. RESULTS There were 36 cases of shoulder dystocia during the study period, an incidence of 4%. Previous delivery of more than 3.5 kg, prolonged labor and prolonged second stage were not associated with shoulder dystocia. Although diabetes and instrumental delivery were independently and significantly associated with shoulder dystocia their importance as a predictor became relevant only in the presence of macrosomia. CONCLUSION Macrosomia is the only reliable predictor of shoulder dystocia.