Background
It remains uncertain whether clinical factors known prior to delivery can predict which women are more likely to experience shoulder dystocia in the setting of operative vaginal delivery.
Objective
We sought to identify whether shoulder dystocia can be accurately predicted among women undergoing an operative vaginal delivery.
Study Design
This was a case-control study of women undergoing a low or outlet operative vaginal delivery from 2005 through 2014 in a single tertiary care center. Cases were defined as women who experienced a shoulder dystocia at the time of operative vaginal delivery. Controls consisted of women without a shoulder dystocia at the time of operative vaginal delivery. Variables previously identified to be associated with shoulder dystocia that could be known prior to delivery were abstracted from the medical records. Bivariable analyses and multivariable logistic regression were used to identify factors independently associated with shoulder dystocia. A receiver operating characteristic curve was created to evaluate the predictive value of the model for shoulder dystocia.
Results
Of the 4080 women who met inclusion criteria, shoulder dystocia occurred in 162 (4.0%) women. In bivariable analysis, maternal age, parity, body mass index, diabetes, chorioamnionitis, arrest disorder as an indication for an operative vaginal delivery, vacuum use, and estimated fetal weight >4 kg were significantly associated with shoulder dystocia. In multivariable analysis, parity, diabetes, chorioamnionitis, arrest disorder as an indication for operative vaginal delivery, vacuum use, and estimated fetal weight >4 kg remained independently associated with shoulder dystocia. The area under the curve for the generated receiver operating characteristic curve was 0.73 (95% confidence interval, 0.69–0.77), demonstrating only a modest ability to predict shoulder dystocia before performing an operative vaginal delivery.
Conclusion
While risk factors for shoulder dystocia at the time of operative vaginal delivery can be identified, reliable prediction of shoulder dystocia in this setting cannot be attained.
Introduction
Shoulder dystocia is defined as a delivery that requires additional maneuvers for delivery of the fetal shoulders. It occurs in 0.6-1.4% of vaginal births and represents an obstetric emergency as it is associated with birth trauma and risk of permanent disability in the newborn. Some reports have suggested that operative, as opposed to spontaneous, delivery is associated with an increased odds of shoulder dystocia.
Several investigators have described risk factors associated with shoulder dystocia, however none of these have been shown to provide predictive values high enough to provide clinical utility. The vast majority of these studies focused on spontaneous vaginal deliveries, but those that accounted for operative vaginal deliveries were limited in their applicability to contemporary practice as they included midpelvic deliveries, metal vacuum cups, and sequential application of 2 instruments. In addition, these studies included the actual birthweight in their prediction model, even though this variable is unknown at the time of operative vaginal delivery planning and its use would provide falsely increased predictive capability. Therefore, the objective of this study was to identify predictors of shoulder dystocia that could help inform the decision-making process prior to the attempt of an operative vaginal delivery.
Materials and Methods
This was a case-control study of women undergoing an operative vaginal delivery at Northwestern Memorial Hospital from October 2005 through May 2014. Cases were defined as women who experienced a shoulder dystocia at the time of operative vaginal delivery. Shoulder dystocia was defined as an application of additional obstetric maneuvers following failure of gentle downward traction on the fetal head to enable delivery of the fetal shoulders. Women whose deliveries were not complicated by a shoulder dystocia constituted the control group. Shoulder dystocia management during the study period involved a team-level response to each recognized event. Briefly, a physician alerts a nurse that there is a shoulder dystocia, communicates with the patient, and directs the nursing staff to initiate appropriate initial maneuvers (ie, McRoberts or suprapubic pressure). A lead nurse records the events, tracks and communicates to the team the elapsed time, and activates an emergency response call that mobilizes additional obstetric, anesthesia, and nursing staff to the room. Shoulder dystocia drills including each of these steps are routinely conducted on the labor and delivery unit.
All singleton, cephalic, nonanomalous gestations delivered from a low or outlet station were included in the analysis. We excluded women who underwent sequential use of 2 different types of instruments (ie, vacuum and forceps), due to the higher chance of morbidity associated with this practice. In addition, women with midpelvic station instrument application were excluded from the analysis, as this practice is relatively uncommon in contemporary obstetrics, and thus a prediction model using this variable may not be generalizable. Finally, women who had an attempted operative vaginal delivery but ultimately required a cesarean for delivery were excluded, as they could not experience the outcome of interest.
Medical records were abstracted for sociodemographic and clinical characteristics, including maternal age, race/ethnicity, body mass index (BMI) at the time of delivery, obstetric history (including history of macrosomia [ie, birthweight >4000 g] or shoulder dystocia in prior pregnancy), presence of diabetes mellitus (pregestational or gestational), induction of labor, length of the second stage, type of instrument used, and fetal gender. For the purpose of this study, women were categorized as having an arrest disorder if the indication for the operative delivery was either arrest of descent, prolonged second stage, or maternal exhaustion. Maternal exhaustion was included in this category given the clinical overlap between these conditions. Although birthweight has been associated with an increased risk of shoulder dystocia, the actual birthweight is not available at the time of operative vaginal delivery planning. Therefore, we did not use this variable but instead, using a previously described methodology, created a variable for estimated fetal weight (EFW) that incorporates the reported inaccuracy of birthweight estimation. Specifically, we generated an EFW for each patient by randomly varying the true birthweight assuming a normal distribution and utilizing the actual birthweight as the mean and 20% of the true birthweight as the SD.
Cases and controls were compared in bivariable analysis using Student t test, χ 2 test, Fisher exact test, and Mann-Whitney U test as appropriate. Factors that were significantly associated with shoulder dystocia at the time of operative vaginal delivery ( P < .05) were retained for further analyses, which included a multivariable logistic regression for the outcome of shoulder dystocia. As the combination of an EFW >4 kg and presence of maternal diabetes mellitus is known to increase the likelihood of shoulder dystocia, we tested for interaction between these 2 risk factors toward the risk of shoulder dystocia with operative vaginal delivery. In a similar manner, we tested for interactions between instrument type and EFW >4 kg as well as instrument type and diabetes. Significant main effect and interaction terms ( P < .05) were included in the final model.
A receiver operating characteristic curve analysis was performed using the final regression model. The area under the curve (AUC) was calculated to estimate the predictive capacity of the logistic regression. An AUC of 0.80 was considered minimally necessary for the model to have clinical applicability. Sensitivity, specificity, and positive and negative likelihood ratios (and their 95% confidence intervals [CI] calculated according to the binomial distribution) were estimated for the number of risk factors present based on the final predictive model that was derived. All analyses were performed with software (Stata, Version 12.0; StataCorp, College Station, TX). All tests were 2-tailed and P < .05 was used to define significance. Approval for this study was obtained from the Northwestern University Institutional Review Board.
Materials and Methods
This was a case-control study of women undergoing an operative vaginal delivery at Northwestern Memorial Hospital from October 2005 through May 2014. Cases were defined as women who experienced a shoulder dystocia at the time of operative vaginal delivery. Shoulder dystocia was defined as an application of additional obstetric maneuvers following failure of gentle downward traction on the fetal head to enable delivery of the fetal shoulders. Women whose deliveries were not complicated by a shoulder dystocia constituted the control group. Shoulder dystocia management during the study period involved a team-level response to each recognized event. Briefly, a physician alerts a nurse that there is a shoulder dystocia, communicates with the patient, and directs the nursing staff to initiate appropriate initial maneuvers (ie, McRoberts or suprapubic pressure). A lead nurse records the events, tracks and communicates to the team the elapsed time, and activates an emergency response call that mobilizes additional obstetric, anesthesia, and nursing staff to the room. Shoulder dystocia drills including each of these steps are routinely conducted on the labor and delivery unit.
All singleton, cephalic, nonanomalous gestations delivered from a low or outlet station were included in the analysis. We excluded women who underwent sequential use of 2 different types of instruments (ie, vacuum and forceps), due to the higher chance of morbidity associated with this practice. In addition, women with midpelvic station instrument application were excluded from the analysis, as this practice is relatively uncommon in contemporary obstetrics, and thus a prediction model using this variable may not be generalizable. Finally, women who had an attempted operative vaginal delivery but ultimately required a cesarean for delivery were excluded, as they could not experience the outcome of interest.
Medical records were abstracted for sociodemographic and clinical characteristics, including maternal age, race/ethnicity, body mass index (BMI) at the time of delivery, obstetric history (including history of macrosomia [ie, birthweight >4000 g] or shoulder dystocia in prior pregnancy), presence of diabetes mellitus (pregestational or gestational), induction of labor, length of the second stage, type of instrument used, and fetal gender. For the purpose of this study, women were categorized as having an arrest disorder if the indication for the operative delivery was either arrest of descent, prolonged second stage, or maternal exhaustion. Maternal exhaustion was included in this category given the clinical overlap between these conditions. Although birthweight has been associated with an increased risk of shoulder dystocia, the actual birthweight is not available at the time of operative vaginal delivery planning. Therefore, we did not use this variable but instead, using a previously described methodology, created a variable for estimated fetal weight (EFW) that incorporates the reported inaccuracy of birthweight estimation. Specifically, we generated an EFW for each patient by randomly varying the true birthweight assuming a normal distribution and utilizing the actual birthweight as the mean and 20% of the true birthweight as the SD.
Cases and controls were compared in bivariable analysis using Student t test, χ 2 test, Fisher exact test, and Mann-Whitney U test as appropriate. Factors that were significantly associated with shoulder dystocia at the time of operative vaginal delivery ( P < .05) were retained for further analyses, which included a multivariable logistic regression for the outcome of shoulder dystocia. As the combination of an EFW >4 kg and presence of maternal diabetes mellitus is known to increase the likelihood of shoulder dystocia, we tested for interaction between these 2 risk factors toward the risk of shoulder dystocia with operative vaginal delivery. In a similar manner, we tested for interactions between instrument type and EFW >4 kg as well as instrument type and diabetes. Significant main effect and interaction terms ( P < .05) were included in the final model.
A receiver operating characteristic curve analysis was performed using the final regression model. The area under the curve (AUC) was calculated to estimate the predictive capacity of the logistic regression. An AUC of 0.80 was considered minimally necessary for the model to have clinical applicability. Sensitivity, specificity, and positive and negative likelihood ratios (and their 95% confidence intervals [CI] calculated according to the binomial distribution) were estimated for the number of risk factors present based on the final predictive model that was derived. All analyses were performed with software (Stata, Version 12.0; StataCorp, College Station, TX). All tests were 2-tailed and P < .05 was used to define significance. Approval for this study was obtained from the Northwestern University Institutional Review Board.
Results
During the study period, operative vaginal delivery was attempted among 4423 women with singleton, cephalic, nonanomalous gestations. Failure occurred in 272 (6.1%) cases and these were excluded from the analysis. Sixty (1.4%) midpelvic operative vaginal deliveries and 11 (0.2%) sequential forceps and vacuum applications were excluded as well. This left 4080 operative vaginal deliveries for the final analysis: 2357 (57.8%) forceps deliveries and 1723 (42.2%) vacuum deliveries. Shoulder dystocia occurred in 162 (4.0%) women. The frequency of shoulder dystocia was 3.1% (N = 72) in the setting of forceps and 5.2% (N = 90) in the setting of vacuum ( P < .001). The rate of brachial plexus injury persistent at the time of discharge was 12.9% (21 of 162) among women whose delivery was complicated by shoulder dystocia, compared to 0.6% (24 of 3918) among women without a documented shouler dystocia.
Maternal and neonatal characteristics stratified by shoulder dystocia occurrence are shown in Table 1 . Women with a shoulder dystocia had a younger mean age and higher BMI, and were more likely to be multiparous and have diabetes. Regarding labor characteristics, chorioamnionitis, an arrest disorder as an indication for operative vaginal delivery, vacuum use, and EFW >4 kg were more common among women with a shoulder dystocia. Four (2.5%) women who had a shoulder dystocia had none of the examined risk factors. When examining fetal station and type of instrument applied, forceps were used more frequently at a low station (83.1% vs 75.6%) and a vacuum was used more frequently at an outlet station (24.4% vs 16.9%) with P < .001.
