Operative vaginal delivery remains a valuable tool to facilitate vaginal delivery in the setting of prolonged second stage, nonreassuring fetal status, or limited maternal expulsive efforts.1 One of the inherent risks of operative vaginal delivery is the failure to achieve vaginal delivery. Failed operative vaginal delivery is associated with adverse neonatal and maternal outcomes including neonatal metabolic acidosis, seizures, neonatal intracranial hemorrhage, neonatal encephalopathy, maternal wound complications, and higher rates of general anesthesia.2–7 These risks are significantly higher relative to second-stage cesarean delivery without an operative vaginal delivery attempt.7 Thus, the ability to predict failed operative vaginal delivery might improve clinical outcomes.
Although the medical literature contains several reports of factors associated with failed operative vaginal delivery, these studies are limited in their applicability to the contemporary practice of operative vaginal delivery, because they included midpelvic deliveries, metal vacuum cups, and sequential application of two instrument types.8–10 Moreover, prior studies reported associations with rather than prediction of failed operative vaginal delivery and often included factors that are unknown at the time of operative vaginal delivery planning such as birth weight and neonatal head circumference.6,10–12 The aim of this study, therefore, was to identify independent predictors of failed operative vaginal delivery that could be used before the decision to attempt operative vaginal delivery.
MATERIALS AND METHODS
This was a case–control study of women undergoing an attempted operative vaginal delivery from a low or outlet station between October 2005 and May 2014 at Northwestern Memorial Hospital. Women in the case group were defined as women with a failed operative vaginal delivery—women who had an attempted operative vaginal delivery but ultimately required a cesarean for delivery. If forceps blades were positioned around the fetal head but the procedure was abandoned before the application of traction, it was still considered a failed operative vaginal delivery. Women in the control group were defined as women who had a successful operative vaginal delivery. Women were included in this analysis if they were older than 18 years of age and had a singleton, cephalic, nonanomalous gestation. Women who underwent sequential use of vacuum and forceps were excluded from the analysis, because sequential use has been associated with a significantly higher chance of morbidity.2,13 Also, women who delivered from a midpelvic station were excluded, because operative vaginal delivery from this station is uncommon and we wished to determine whether a prediction model could be developed that was most generalizable.
Sociodemographic characteristics, clinical characteristics, and pregnancy outcomes were abstracted from the clinical records and were compared between women in the case group and those in the control group. Sociodemographic data collected included maternal age and race–ethnicity. Baseline clinical characteristics included body mass index at the time of delivery and prior obstetric history. Factors suggested by prior studies to be associated with failed operative vaginal delivery such as maternal diabetes, induction of labor, length of the second stage, indication for operative vaginal delivery, type of instrument, station at the time of instrument application, and occiput-posterior position were abstracted as well.8–10,12 The information about frequency of episiotomy was collected but not included in the analysis because the decision to perform an episiotomy is typically made after there is some certainty regarding an impending vaginal delivery. Accordingly, we felt that this variable would not be of use in determining whether to undertake an operative vaginal delivery. Although birth weight and frequency of macrosomia have been associated with an increased risk of failed operative vaginal delivery,9,10,12 because the actual birth weight is not available at the time of contemplating an operative vaginal delivery, it cannot be used as a predictor of failure or success. Instead, because physicians can and do make an estimate of fetal weight before delivery, we generated an estimate of fetal weight that reflects the documented accuracy of this estimation. Specifically, the estimate of fetal weight was created by randomly varying the actual birth weight along a normal distribution with the actual birth weight as the mean and a preset standard deviation of 20% of the actual birth weight.14–16
Neonatal information was abstracted from the inpatient records to examine adverse neonatal outcomes associated with failed operative vaginal delivery. These included umbilical artery pH 7.0 or less, Apgar score less than 5 at 5 minutes, neonatal intensive care unit admission, fetal injury (scalp and facial lacerations, bone fracture, facial nerve injury, brachial plexus injury), seizures, intracranial hemorrhage (ie, subdural, cerebral, intraventricular, and subarachnoid), and clinical evidence of central nervous system depression.
Women in the case group and those in the control group were compared in bivariable analysis using Student's t test, χ2, Fisher exact test, and Wilcoxon rank-sum (Mann-Whitney U) test as appropriate. Factors that were significantly associated with failed operative vaginal delivery (P<.05) were retained for further analyses, which included multivariable logistic regression for the outcome of failed operative vaginal delivery and receiver operating characteristic curve analysis. The area under the curve (AUC) was calculated to estimate the predictive capacity of the final logistic regression. An AUC of at least 0.80 was considered to represent accurate prediction.17,18 To determine whether there was any effect modification between type of instrument used (vacuum or forceps) and factors that were significantly associated with failed operative vaginal delivery in bivariable analysis, interaction terms were created and entered into logistic regression. All analyses were performed with Stata 12.0. All tests were two-tailed and P<.05 was used to define significance. Approval for this study was obtained from the Northwestern University institutional review board.
A total of 4,423 operative vaginal deliveries were attempted during the study period. After exclusion of 60 (1.4%) midpelvic operative vaginal deliveries and 11 (0.2%) sequential forceps and vacuum application, 4,352 deliveries remained for analysis. Two thousand five hundred twenty-seven (58.1%) women underwent an attempt at operative vaginal delivery using forceps and 1,825 (41.9%) using vacuum. Operative vaginal delivery failure occurred in 170 (6.7%) and 102 (5.6%) of these women, respectively (P=.13).
Maternal and neonatal characteristics were compared between women in the case group and those in the control group (Table 1). In bivariable analysis, women with a failed operative vaginal delivery were more likely to be nulliparous as well as either Hispanic or non-Hispanic white. Induction of labor, chorioamnionitis, second stage 2 hours or greater, higher station at application (low pelvic compared with outlet), fetal occiput-posterior position, and larger estimate of fetal weight were more common among women with a failed operative vaginal delivery. Arrest or exhaustion as an indication for operative vaginal delivery also was associated with an increased risk of failure (Table 1).
Pregnancies complicated by failed operative vaginal delivery had a higher frequency of postpartum hemorrhage and, as expected, a lower frequency of third- or fourth-degree lacerations (Table 2). The frequencies of umbilical arterial pH less than 7.0, neonatal intensive care unit admission, facial and scalp lacerations, neonatal intracranial hemorrhage, and evidence of central nervous system depression also were more common among women with failed operative vaginal delivery. Facial nerve injury and brachial plexus palsy were similar regardless of mode of delivery (Table 2).
After adjusting for potential confounding variables in a multivariable regression, factors that were independently associated with operative vaginal delivery failure were non-Hispanic black race, arrest or exhaustion as an indication for operative vaginal delivery (compared with fetal indications), occiput-posterior position, and low pelvic application (compared with outlet) (Table 3). The corresponding receiver operating characteristic curve has an AUC of 0.74 (95% confidence interval 0.69–0.77; Fig. 1). Estimate of fetal weight was not significantly associated with failed operative vaginal delivery regardless of whether it was entered as a continuous or categorical variable (ie, less than 3.5 kg, 3.5–4 kg, and greater than 4 kg). None of the individual risk factors had a positive predictive value for failed operative vaginal delivery of more than 15%. Furthermore, the combination of each of the risk factors yielded a positive predictive value for failed operative vaginal delivery of only 17.7% (Table 4).
Finally, although the proportions of forceps and vacuum use for operative vaginal delivery did not differ between women in the case group and those in the control group in bivariable analysis, we examined the interaction between type of instrument used and each of the variables described in the logistic regression. We found a significant interaction between the use of vacuum and low station at the time of application (ie, vacuums were significantly more likely to fail at low station) with an interaction odds ratio of 4.54 for failure (95% confidence interval 1.24–16.63).
The underlying goal of this study was to determine whether there are risk factors present that could assist the obstetric provider in candidate selection for operative vaginal delivery to minimize failed operative vaginal delivery attempts. We found that the frequency of failed operative vaginal delivery was 6.3%. Failed operative vaginal delivery was associated with several factors known before the operative vaginal delivery attempt, including race–ethnicity, the indication for operative vaginal delivery, station, and orientation of the vertex at the time of instrument application. Nevertheless, although each of these risk factors was independently associated with failed operative vaginal delivery, when incorporated into a regression, the resulting receiver operating characteristic curve did not show high accuracy in terms of prediction of failure, because the AUC was less than 0.8.17,18
In contrast to prior studies, we did not find an association between the type of instrument used during operative vaginal delivery and risk for failure.8,10,12,13 This may be explained by our institution's high prevalence of utilization of both instruments. In 2014, the rate of operative vaginal delivery in our institution was more than 9%; the national frequency is estimated at approximately 5%.19 Similar to our findings, a randomized prospective trial comparing these two types of instruments did not find a significant difference in their efficacy.20
Another difference between our findings and previous reports is in regard to the association between birth weight and failed operative vaginal delivery.6,9,10,12 When designing this study, our intent was to develop a clinically applicable prediction model and not merely describe associations with failed operative vaginal delivery. Accordingly, actual birth weight could not be used in the model, because this variable is not available to the obstetrician before delivery. Instead, the health care provider estimates the fetal weight based on the imperfect tools at their disposal, including growth ultrasonograms and Leopold's maneuvers. These data are then incorporated into an estimate of fetal weight and can inform their decision-making regarding route of delivery. When an estimate of fetal weight, rather than a birth weight, is used, an independent association with failed operative vaginal delivery was not found. This lack of association suggests that the inaccuracy of weight estimation is sufficiently great to compromise the use of fetal weight in a prediction model. In addition, in real time, estimate of fetal weight is not assessed as an individual factor when deciding to pursue operative vaginal delivery, but rather, its significance comes in the context of additional characteristics such as the shape of the maternal pelvis, fetal station and position, and presence of risk factors for macrosomia such as diabetes. The constellation of all these variables guides the obstetric decision-making regarding a trial of operative vaginal delivery.
Our study is not without limitations. One is residual confounding by other unavailable factors associated with adverse outcomes in the setting of operative vaginal delivery. One example of this is health care provider experience, which has been shown to be inversely proportional to failed operative vaginal delivery rates and adverse neonatal outcomes.12 However, prior analysis from our institution showed that obstetricians' operative vaginal delivery experience was not associated with adverse maternal or neonatal outcomes21; therefore, we did not feel that this variable should be included in the model. Another limitation is the lack of a confirmatory cohort to validate our results. However, because this prediction model was developed as a best fit for our institutional data, and did not produce an accurate prediction model, application to an independent cohort would be of limited value and likely even less predictive of failed operative vaginal delivery.
Despite these limitations, our study adds to the existing literature by examining characteristics that are available only before delivery, by including deliveries that are most applicable to current obstetric practice, and by including data on indications for operative vaginal delivery, fetal station, and head position, which were not used in many of the studies exploring factors associated with failed operative vaginal delivery.6,9,10,12 In conclusion, although we identified risk factors for failed operative vaginal delivery, these cannot be used to accurately predict outcome before the initiation of the attempt.
1. Shoulder dystocia. ACOG Practice Bulletin No. 40. American College of Obstetricians and Gynecologists. Obstet Gynecol 2002;100:1045–50.
2. Towner D, Castro MA, Eby-Wilkens E, Gilbert WM. Effect of mode of delivery in nulliparous women on neonatal intracranial injury. N Engl J Med 1999;341:1709–14.
3. Johanson RB, Menon BK. Vacuum extraction versus forceps for assisted vaginal delivery. The Cochrane Database of Systematic Reviews 2000, Issue 2. Art. No.: CD000224.
4. Demissie K, Rhoads GG, Smulian JC, Balasubramanian BA, Gandhi K, Joseph KS, et al.. Operative vaginal delivery and neonatal and infant adverse outcomes: population based retrospective analysis. BMJ 2004;329:24–9.
5. Revah A, Ezra Y, Farine D, Ritchie K. Failed trial of vacuum or forceps—maternal and fetal outcome. Am J Obstet Gynecol 1997;176:200–4.
6. Murphy DJ, Liebling RE, Verity L, Swingler R, Patel R. Early maternal and neonatal morbidity associated with operative delivery in second stage of labour: a cohort study. Lancet 2001;358:1203–7.
7. Alexander JM, Leveno KJ, Hauth JC, Landon MB, Gilbert S, Spong CY, et al.. Failed operative vaginal delivery. Obstet Gynecol 2009;114:1017–22.
8. Al-Kadri H, Sabr Y, Al-Saif S, Abulaimoun B, Ba'Aqeel H, Saleh A. Failed individual and sequential instrumental vaginal delivery: contributing risk factors and maternal-neonatal complications. Acta Obstet Gynecol Scand 2003;82:642–8.
9. Gopalani S, Bennett K, Critchlow C. Factors predictive of failed operative vaginal delivery. Am J Obstet Gynecol 2004;191:896–902.
10. Ben-Haroush A, Melamed N, Kaplan B, Yogev Y. Predictors of failed operative vaginal delivery: a single-center experience. Am J Obstet Gynecol 2007;197:308.e1–e5.
11. Ooi PV, Ramphul M, Said S, Burke G, Kennelly MM, Murphy DJ. Ultrasound assessment of fetal head circumference at the onset of labor as a predictor of operative delivery. J Matern Fetal Neonatal Med 2015;28:2182–6.
12. Aiken CE, Aiken AR, Brockelsby JC, Scott JG. Factors influencing the likelihood of instrumental delivery success. Obstet Gynecol 2014;123:796–803.
13. Edozien LC, Williams JL, Chattopadhyay I, Hirsch PJ. Failed instrumental delivery: how safe is the use of a second instrument? J Obstet Gynaecol 1999;19:460–2.
14. Chauhan SP, Hendrix NW, Magann EF, Morrison JC, Kenney SP, Devoe LD. Limitations of clinical and sonographic estimates of birth weight: experience with 1034 parturients. Obstet Gynecol 1998;91:72–7.
15. Sherman DJ, Arieli S, Tovbin J, Siegel G, Caspi E, Bukovsky I. A comparison of clinical and ultrasonic estimation of fetal weight. Obstet Gynecol 1998;91:212–7.
16. Scioscia M, Vimercati A, Ceci O, Vicino M, Selvaggi LE. Estimation of birth weight by two-dimensional ultrasonography: a critical appraisal of its accuracy. Obstet Gynecol 2008;111:57–65.
17. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29–36.
18. McNeil BJ, Hanley JA. Statistical approaches to the analysis of receiver operating characteristic (ROC) curves. Med Decis Making 1984;4:137–50.
19. Martin J, Hamilton B, Ventura S, Osterman M, Wilson E, Matthews T. Births: final data for 2010. Natl Vital Stat Rep 2012;61:1–72.
20. Bofill JA, Rust OA, Schorr SJ, Brown RC, Martin RW, Martin JN Jr, et al.. A randomized prospective trial of the obstetric forceps versus the M-cup vacuum extractor. Am J Obstet Gynecol 1996;175:1325–30.
21. Miller ES, Barber EL, McDonald KD, Gossett DR. Association between obstetrician forceps volume and maternal and neonatal outcomes. Obstet Gynecol 2014;123:248–54.