Objectives: To relate maternal and infant characteristics to newborn shoulder width and to evaluate the predictive value of newborn shoulder width measurement in cases of shoulder dystocia.
Design: Newborn shoulder width was systematically measured at birth during a period of 18 months.
Setting: Department of Obstetrics and Gynaecology of Saint-Antoine University Hospital, Paris, France.
Population: A total of 2222 newborn shoulder width measurements were performed and 22 cases of true shoulder dystocia occurred during the study period.
Methods: Newborn shoulder width measurements were reviewed and correlated with maternal age, parity, nonpregnant weight, weight gain during pregnancy, height, race, fasting glucose and one hour glucose levels, gestational age, birthweight and sex of the neonate. A receiver-operating characteristics curve was constructed to evaluate newborn shoulder width as a test for predicting shoulder dystocia.
Results: The mean newborn shoulder width was 122.06 mm (10.50 SD). Stepwise multiple regression showed that newborn shoulder width was significantly associated with birthweight (P < 0.001), parity (P = 0.04), and nonpregnant weight (P = 0.04). We estimated that the best cut off for shoulder dystocia prediction was a newborn shoulder width measurement with a low false positive rate (< 10%) in association with a high sensitivity rate. Therefore, newborn shoulder width measurement > or = 140 mm was selected. This measurement should have a low sensitivity of 27.27%, a specificity of 91.82%, a positive predictive value of 4.02%, and a negative predictive value of 99.01% for shoulder dystocia prediction. Nevertheless, birthweight > or = 4000 g should have a better predictive value retrospectively for shoulder dystocia.
Conclusions: Newborn shoulder width measurement, which is strongly correlated with birthweight, still remains a poor predictor for shoulder dystocia, even when this evaluation is correct antenatally.