The modern-day vacuum extractor was first introduced in 1954,1 and its use slowly diffused into obstetric practice during the ensuing three decades. However, since the late 1980s the use of the vacuum extractor has increased whereas the use of forceps has decreased.2,3 These trends of obstetric management have occurred despite little evidence regarding which form of assisted vaginal delivery is optimal in different settings.
A recent meta-analysis included all of the prospective randomized trials of forceps compared with vacuum.4 There were only 10 such studies, with sample sizes ranging from 365 to 637.6 This collection of analyses revealed that vacuum extractors have a higher rate of failure than forceps and lead to greater rates of cephalohematoma and retinal hemorrhage. They were also associated with a trend toward a greater rate of 5-minute Apgar scores less than 7. Forceps were associated with greater rates of perineal injury and short-term postdelivery pain. A variety of other outcomes were examined, and no differences were demonstrated. However, most outcomes were too underpowered in the analysis for any conclusions to be drawn.
There have also been several large birth-certificate–based analyses that have demonstrated higher rates of cephalohematoma7–9 and lower rates of perineal laceration,8,9 but no difference in intracranial hemorrhage or perinatal mortality.7,8 However, when intracranial hemorrhage was subdivided further, an increase in subarachnoid hemorrhage was noted in one analysis.9 These studies are limited with respect to causality because little is known about a variety of confounding variables, including type and experience of provider and indication for delivery. One of these analyses also demonstrated a higher rate of shoulder dystocia for vacuum-assisted vaginal delivery when compared with either forceps delivery or vacuum and forceps used sequentially.8 The largest prospective randomized controlled trial of forceps compared with vacuum also identified a trend toward a higher rate of shoulder dystocia among their vacuum-assisted vaginal deliveries.6
The proposed mechanisms for the differences in these outcomes include the increased occupation of space in the vagina, and thus greater distention of the vagina walls and perineum by the forceps, leading to increased perineal and vaginal lacerations. Alternatively, the placement of the vacuum directly on the scalp has been proposed to increase the rates of cephalohematoma, retinal hemorrhage, and intracranial hemorrhage,7 although only the first two have been demonstrated. Finally, the ability to pull harder with forceps is proposed to explain the increased rate of success. It is of note that this last theory should actually support a higher rate of shoulder dystocia among forceps as opposed to vacuum deliveries.
We propose, however, that it is not the total magnitude of the resultant force vector that leads to shoulder dystocia, but rather where it acts and in what direction. The forceps are placed cradling the entire fetal head, thus the resultant force vector applied to the fetal head acts further caudally along the fetal head, approximately half the distance of the blades (Fig. 1), and closer to the fetal shoulders thorax. Moreover, because of the placement around the fetal head, force vectors can be generated ranging from perpendicular to the ground to horizontal. Because the resultant force vector used when delivering by forceps is directed farther down the head and in a more downward direction, it is likely to facilitate in directing the anterior shoulder down and possibly underneath the pubic symphysis. The vacuum, because it is attached directly in front of the posterior fontanelle and the direction of its resultant force vector is at most 45 degrees to the floor, actually may lead to more traction force on the fetal head, pulling the anterior shoulder into the pubic symphysis and lodging it in this location.
Given this background and our proposed mechanism, we generated the hypothesis that forceps should be associated with fewer shoulder dystocias than vacuum. However, because of the generation of greater downward force by the forceps, we hypothesized that they would be associated with a higher rate of perineal laceration.
We designed a retrospective cohort study of all singleton neonates delivered beyond 37 weeks gestational age in the vertex presentation by nonrotational forceps or vacuum from January 1, 1981, to December 1, 2001, at the Moffitt-Long Hospital at the University of California, San Francisco. Exclusion criteria included delivery before 37 weeks of gestation, fetal anomalies, noncephalic presentation, occiput transverse position, and multiple gestation. This study was approved by the Investigational Review Board at the University of California, San Francisco. Our primary outcomes, shoulder dystocia and perineal laceration, were entered into data entry sheets by the delivering clinician and verified by trained abstractors. Shoulder dystocia is defined at our institution as any delivery requiring additional maneuvers to deliver the shoulders of the infant, including McRobert's maneuver, suprapubic pressure, Woods' or Rubin maneuvers, or delivery of the posterior arm. We also collected information on the following secondary outcomes: vaginal lacerations, cervical lacerations, cephalohematoma, intracranial hemorrhage, Erb's palsy, facial nerve palsy, neonatal jaundice, skull, clavicle and humerus fractures, neonatal seizures, Apgar scores, umbilical artery pH, umbilical artery base excess, and neonatal intensive care unit (NICU) admission. The following data were also collected and examined as potential confounders: maternal age, ethnicity, education, body mass index, diabetes mellitus (both preconceptional and gestational), length of labor, station at delivery, position at delivery, obstetric provider(s), year of delivery, parity, anesthesia, and birthweight.
The data were then compiled and analyzed using STATA 7 software (StataCorp, College Station, TX). Univariate analyses of the primary predictor, forceps compared with vacuum, were performed examining the primary and secondary outcomes of interest and compared using the χ2 test. Because the rates of many of the neonatal complications were low, a summary variable called “severe complications” was created that included birth trauma (including skull fracture, brachial plexus injury, facial nerve palsy), neonatal seizures, and intracranial hemorrhage. For each of these primary and secondary outcomes a multivariate logistic regression was performed, including the potential confounders. The confounders' contributions to the model were tested using the maximum likelihood ratio test, and they were only kept in the model if they were statistically significant. Statistical significance was defined as P < .05.
During the study period, there were 4,120 women who met the inclusion and exclusion criteria outlined above. Of these, 2,075 (50.4%) delivered by forceps and 2,045 (49.6%) delivered by vacuum-assisted vaginal delivery. The women who underwent forceps-assisted vaginal delivery were more likely to be aged younger than 35 years, nulliparous, and have Medicaid insurance, an episiotomy, and epidural analgesia (Table 1). When the primary outcomes were examined, we found that the rate of shoulder dystocia was lower among women undergoing forceps delivery (1.5%) compared with those undergoing vacuum-assisted vaginal delivery (3.5%, P < .001). The rate of third- or fourth-degree lacerations was higher among forceps deliveries (36.9%) compared with vacuum-assisted vaginal deliveries (26.8%, P < .001). Among the secondary outcomes, the rates of 5-minute Apgar scores less than 7, cephalohematoma, admissions to NICU, and neonatal jaundice were higher among women undergoing vacuum-assisted vaginal delivery (Table 2), whereas vaginal lacerations, cervical lacerations, and facial nerve palsy were higher among women undergoing forceps deliveries. No differences were seen in umbilical artery blood gases, intracranial hemorrhage, or other types of neonatal trauma.
In multivariate analyses, when controlling for potential confounders, most of the significant univariate findings persisted (Table 3), with adjusted odds ratios (AOR) that did not cross unity. Shoulder dystocia was lower among women with a forceps delivery (AOR 0.34, 95% confidence interval [CI] 0.20–0.57), and both vaginal lacerations and third- and fourth-degree perineal lacerations were higher (AOR 1.79, 95% CI 1.52–2.10). A vacuum-assisted vaginal delivery was still associated with higher rates of 5-minute Apgar scores less than 7, cephalohematoma, admission to the NICU, and neonatal jaundice. However, the differences in the rates of cervical laceration and facial nerve palsy were no longer statistically significant.
To examine the effects of forceps compared with vacuum among women of differing parity, subgroup analyses were performed. Among nulliparous patients, the prior observed differences in the primary outcomes persisted, with women undergoing forceps deliveries having a lower rate of shoulder dystocia (1.4% compared with 3.1%, P < .001) and higher rates of third- and fourth-degree perineal laceration (40.9% compared with 30.7%, P < .001). These results were confirmed by multivariate analyses; when comparing forceps to vacuum, the differences were AOR of 0.39 (95% CI 0.23–0.65) for shoulder dystocia and AOR of 1.48 (95% CI 1.28–1.73) for third- and fourth-degree perineal laceration. However, among multiparas the difference persisted only for shoulder dystocia. Women with forceps deliveries had a lower rate of shoulder dystocia (1.7% compared with 4.5%, P = .012), but no statistically significant difference in the rate of third- and fourth-degree lacerations (19.6% compared with 15.7%, P = .107). In multivariate analyses, comparing forceps with vacuum for shoulder dystocia, the outcome was different (AOR 0.41, 95% CI 0.18–0.96), but for third- and fourth-degree perineal laceration it was not (AOR 1.19, 95% CI 0.81–1.69).
We found lower rates of shoulder dystocia among women with forceps deliveries as compared with women undergoing vacuum-assisted vaginal delivery, consistent with our proposed hypothesis. We also found higher rates of perineal and vaginal lacerations among the women who had forceps deliveries. In addition to the differences in our primary outcomes, we also found that women undergoing forceps delivery had infants with lower rates of cephalohematoma, 5-minute Apgar score less than 7, admission to the NICU, and neonatal jaundice.
When we examined these outcomes by parity, we found the differences in the primary outcomes of shoulder dystocia and third- and fourth-degree perineal laceration persisted for nulliparas. However, among multiparas, although they still had higher rates of shoulder dystocia among women undergoing vacuum-assisted vaginal delivery, there was no statistically significant difference noted in third- and fourth-degree perineal lacerations. This difference could simply be attributed to the overall higher rates of perineal laceration among nulliparas. However, the clinical effect seems to differ, with adjusted odds ratios of 1.5 among nulliparas and only 1.2 among multiparas. Thus, it may be that the downward vector produced with forceps leads to a greater increase in perineal trauma among nulliparas than it does in multiparas.
Our findings of greater shoulder dystocia as well as several measures of short-term neonatal morbidity among the neonates of women delivered by vacuum-assisted vaginal delivery are of concern. Several of these differences have been noted in small randomized trials of forceps compared with vacuum. Despite these findings, the rate of vacuum use seems only to be increasing. Moreover, one recent analysis concluded from its analysis of birth certificates that “Delivery by vacuum extraction is at least as safe as delivery by forceps.”8 Such a conclusive statement is supported neither by our study nor the broader literature.4,6,9
Our study is not without limitations. A nonrandomized, retrospective, cohort study can be compromised by having multiple confounding variables. We attempted to control for these confounders using multivariate techniques in our analysis. Not only did we control for confounders, such as station, birthweight, episiotomy, and parity, that have been associated with the outcomes of analysis, but we also controlled for year of delivery and obstetric provider in an attempt to control for the more intangible factors surrounding forceps and vacuum delivery. Of course, we could not control for the decision-making process that clinicians undergo when choosing which instrument to assist in operative vaginal delivery. Although we were adequately powered to examine our primary outcomes of shoulder dystocia and perineal lacerations, we were underpowered to examine some of our secondary outcomes, such as the more rare neonatal complications. Another possible limitation pertains to the generalizability of our study population to that of all pregnant women. The patients in our study were managed at an academic center in California where the majority of the providers use both instruments, and deliveries are in conjunction with a resident provider. Other than this, our patient population represented a wide spectrum of ethnicities, ages, and educational levels. Furthermore, when we controlled for the various maternal characteristics, our findings were robust.
Determination of which instrument to use for an operative vaginal delivery should entail weighing the risks and benefits of the instrument to both the mother and the fetus. It seems that among nulliparas, the substantial benefit of vacuum to the maternal perineum should be weighed against the increased risk of shoulder dystocia and cephalohematoma. Interestingly, among multiparas, there seems to be no benefit from decreased perineal trauma, whereas the increased risk to the neonate persists. To examine these outcomes, a large, multicenter, randomized trial should be performed. Another important aspect of the decision regarding which instrument to use is provider comfort with these two modes of operative vaginal delivery. If providers are not trained in the use of forceps, they will not be able to provide this form of operative vaginal delivery, nor make a choice between the two types of operative vaginal delivery. Because we do not know which is “better,” and either may be the instrument of choice in particular clinical situations, it is imperative that providers continue to be trained in the use of forceps as well as vacuum.
1. Malmstrom T. The vacuum extractor, an obstetrical instrument. Acta Obstet Gynecol Scand Suppl 1954;33:1–31.
2. Kozak LJ, Weeks JD. U.S. trends in obstetric procedures, 1990-2000. Birth 2002;29:157–61.
3. Roberts CL, Algert CS, Carnegie M, Peat B. Operative delivery during labour: trends and predictive factors. Paediatr Perinat Epidemiol 2002;16:115–23.
4. Johanson RB, Menon BK. Vacuum extraction versus forceps for assisted vaginal delivery. Cochrane Database Syst Rev 2000;(2):CD000224.
5. Fall O, Ryden G, Finnstrom K, Finnstrom O, Leijon I. Forceps or vacuum extraction? A comparison of effects on the newborn infant. Acta Obstet Gynecol Scand 1986;65:75–80.
6. Bofill JA, Rust OA, Schorr SJ, Brown RC, Martin RW, Martin JN Jr, Morrison JC. A randomized prospective trial of the obstetric forceps versus the M-cup vacuum extractor. Am J Obstet Gynecol 1996;175:1325–30.
7. 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.
8. 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 [published erratum appears in BMJ 2004;329:547]. BMJ 2004;329:24–9.
© 2005 The American College of Obstetricians and Gynecologists
9. Wen SW, Liu S, Kramer MS, Marcoux S, Ohlsson A, Sauve R, et al. Comparison of maternal and infant outcomes between vacuum extraction and forceps deliveries. Am J Epidemiol 2001;153:103–7.