Association between Neuraxial Labor Analgesia and Neonatal Morbidity after Operative Vaginal Delivery: A Retrospective Cross-sectional Study

In this population-based cross-sectional study, we analyzed United States birth certificate data from 2017. Deidentified birth certificate data are made publicly available by the United States Centers for Disease Control and Prevention,¹⁰  thus our analysis was exempt from Stanford University Institutional Review Board review. All states use the 2003 revised United States Standard Certificate of Live Birth format, which comprises detailed demographic, medical, and obstetric information. For this study, we adhered to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Birth facility staff enter data into the birth certificate according to clinical records and maternal surveys. Each birth certificate contains a checkbox to designate whether or not a woman received “epidural or spinal anesthesia during labor.” The birth certificate does not contain information about the specific type of neuraxial analgesia (i.e., epidural, spinal, combined-spinal epidural) or other information regarding other analgesic modalities (e.g., systemic [oral or parenteral] opioids, inhaled nitrous oxide).

Our study population comprised women with nonanomalous singleton pregnancies who underwent vacuum- or forceps-assisted delivery (operative vaginal delivery). We excluded women who had a spontaneous vaginal delivery, cesarean delivery, noncephalic fetal presentation, out-of-hospital birth, as well as those missing data for mode of delivery or neuraxial analgesia use. Although American College of Obstetricians and Gynecologists guidelines during the study period discouraged vacuum extraction at less than 34 weeks’ gestational age,¹¹  there are no contraindications for attempting operative vaginal delivery based on gestational age. Therefore, based on expert obstetric opinion (Y.B.), we set conservative gestational age limits for defining our study cohort (between 36⁺⁰ and 41⁺⁶ weeks’ gestation). We used complete case analysis as the highest percentage of missing values among all covariates was 1.98% for prenatal care.

The main exposure was use of neuraxial labor analgesia. The primary outcome variable was composite neonatal morbidity. The primary outcome was established a priori (before data analysis) and was defined by the presence of any of the following: 5-min Apgar score less than 7, requirement for immediate assisted ventilation, assisted ventilation for more than 6 h, neonatal intensive care unit admission, neonatal transfer to a different facility within 24 h of delivery, and neonatal seizure or serious neurologic dysfunction. Definitions of these birth certificate variables are published by the United States National Center for Health Statistics¹²  and are available in Supplemental Digital Content 1 (https://links.lww.com/ALN/C489). Of note, the definition of a neonatal intensive care admission corresponds with the American Academy of Pediatrics designation of a level III or IV nursery.¹³  We selected these outcomes based on literature review of studies reporting neonatal morbidity outcomes associated with operative vaginal delivery and data availability in the U.S. birth certificate dataset.^4,5,14–17 

The covariates included in our analysis were maternal age (less than 20, 20 to 34, 35 to 39, greater than or equal to 40 yr), education (less than high school, high school, any postsecondary), race/ethnicity (non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, Hispanic, other), prepregnancy body mass index categorized using the World Health Organization classification¹⁸  (underweight, normal, overweight, obese, obesity II, obesity III), trimester when prenatal care commenced (first, second, third, no prenatal care), insurance (Medicaid, private insurance, self-pay, other), prior livebirths (none, 1 or more), prior cesarean, preexisting or gestational diabetes, gestational hypertension or preeclampsia, induction of labor, labor augmentation, gestational age at delivery, and delivery location by U.S. Census region (Northwest, Midwest, South, and West).

A data analysis and statistical plan was written after the data were initially accessed. Before primary data analysis, the main aims and analytic plan were reviewed by all study investigators in September 2019. Statistical analysis was performed with Stata Version 14.0 (StataCorp., USA). We performed descriptive analyses on characteristics of women with and without neuraxial analgesia. An uncorrected chi-square test was used to compare the frequencies of categorical variables between the two groups. All analyses were performed using two-tailed hypothesis testing with a P value of less than 0.05 as statistically significant.

The association between neuraxial analgesia and neonatal morbidity was assessed using relative risk with 95% CI and was estimated by multivariable logistic regression with a binomial distribution and a log-link function.

We performed several a priori sensitivity analyses to assess the robustness of our findings. Because evidence exists that vacuum extraction may be associated with an increased risk of neonatal trauma (e.g., cephalohematoma, intracranial hemorrhage) compared to forceps delivery,¹⁹  we performed stratified analysis to examine risk of neonatal morbidity among women who had forceps only and vacuum only. Additionally, because babies weighing greater than or equal to 4,000 g are at increased risk of newborn complications,²⁰  we performed a prespecified sensitivity analysis to assess the potential association between exposure and outcome in women with nonmacrosomic neonates by restricting the cohort to deliveries with neonates weighing less than 4,000 g.

For the logistic models in our primary and sensitivity analyses, we examined crude and adjusted risk differences to quantify the magnitude of any increased or decreased risk associated with neuraxial analgesia.

During the peer review process, several post hoc secondary and sensitivity analyses were requested. First, survey data indicate that an in-house obstetric anesthesia service is more likely to be present in an urban hospital with a high obstetric volume than a low-volume rural hospital.^21,22  Further, perinatal outcomes, such as neonatal mortality and birth asphyxia are less prevalent in high-volume hospitals.^23,24  Therefore, it is plausible that hospitals with a dedicated or in-house obstetric anesthesia service are more likely to have a high-level neonatal intensive care unit and are less likely to arrange neonatal transfer. Based on this assumption, an association between neuraxial analgesia and neonatal morbidity may be explained, in part, by confounding by indication. To address this concern, we excluded neonatal intensive care unit admission and neonatal transport from our composite outcome, and then reanalyzed our estimates of risk associated with neuraxial analgesia using these revised criteria for the composite outcome (5-min Apgar score less than 7, requirement for immediate assisted ventilation, assisted ventilation for more than 6 h, neonatal seizure or serious neurologic dysfunction). Second, because the dataset contains information for the county of birth, we used generalized estimating equation models with a binomial family distribution and a logit link function to account for within-county clustering (1,568 counties in total). Third, we performed propensity score matching for the nonuse of neuraxial labor analgesia to an individual using neuraxial labor analgesia based on covariates included in the primary analysis, with one-to-one matching without replacement using nearest neighbor matching. Balance in maternal characteristics between women receiving versus not receiving neuraxial analgesia was assessed using absolute standardized difference, with less than 0.1 considered as acceptable balance. Within the propensity-matched cohort, we estimated the relative risk and 95% CI using our primary and revised composite outcome of neonatal morbidity.

We did not perform an a priori sample size estimation. However, before performing formal data analysis, we performed a power analysis using baseline data from our analytic sample. With 92,518 women with neuraxial analgesia and 14,327 women without neuraxial analgesia in our sample, an alpha of 0.05, power of 80%, and an assumed neonatal morbidity rate of 10% among women without neuraxial analgesia, the minimum detectable relative risk of neonatal morbidity for women without neuraxial analgesia compared to women with neuraxial analgesia was 1.09. Our sample size analysis did not include a minimum clinically significant effect size.

We identified 3,864,754 livebirths in 2017 (fig. 1). We excluded 3,749,761 out-of-hospital births, cesarean deliveries, spontaneous vaginal deliveries, congenital anomalies, twins or multiple pregnancies, gestational age at delivery less than 36 weeks or greater than or equal to 42 weeks, breech presentation, and other or unknown fetal presentations. We also excluded 8,198 missing values for delivery location, delivery mode, neuraxial analgesia status, covariates, and neonatal morbidity (fig. 1). Our final analytic cohort comprised 106,845 women who underwent operative vaginal delivery of whom 92,518 (86.6%) received neuraxial analgesia.

Table 1 shows the patient characteristics of women who received and did not receive neuraxial analgesia. Compared to women without neuraxial analgesia, women with neuraxial analgesia were more likely to be non-Hispanic White, and have a secondary education, private insurance, no prior live births, preexisting or gestational diabetes, gestational hypertension or preeclampsia, induction of labor, augmentation of labor, initiated prenatal care in the first trimester, and delivered in the North, Midwest, or South census region.

The rate of composite neonatal morbidity was higher among women receiving neuraxial analgesia than those not receiving neuraxial analgesia (10,409 [11.3%] vs. 1,271 [8.9%], respectively; P < 0.001). Table 2 shows the unadjusted and adjusted relative risk of neonatal morbidity. In the initial unadjusted analysis, the relative risk for neonatal morbidity was 1.27 (95% CI, 1.20 to 1.34; P < 0.001) in women with neuraxial analgesia compared to women without. The association was attenuated after adjusting for confounders (adjusted relative risk, 1.19; 95% CI, 1.12 to 1.26; P < 0.001).

Rates and risks of individual components of the composite neonatal morbidity outcome are presented in table 3. The three most frequent neonatal morbidities (in descending order) in both the neuraxial and no neuraxial analgesia groups were neonatal intensive care admission, immediate need for assisted ventilation, and 5-min Apgar scores less than 7.

Table 2 presents the results of our models stratified according to the method of operative vaginal delivery (forceps or vacuum). In the forceps-assisted delivery cohort, compared to no neuraxial analgesia, women who received neuraxial analgesia had a 1.5-fold adjusted risk of neonatal morbidity. In the vacuum-assisted delivery cohort, neuraxial analgesia was associated with a 1.16-fold adjusted risk of neonatal morbidity. When the cohort was restricted to neonates weighing less than 4,000 g, results were similar to our main findings.

Table 4 presents the unadjusted and the adjusted risk differences in the rate of composite neonatal morbidity according to presence or absence of neuraxial analgesia. For the original study cohort, the adjusted risk difference was 1.8 (95% CI, 1.2 to 2.3) per 100 deliveries. In our stratified analyses, the largest adjusted risk difference was observed among forceps deliveries (3.9 [95% CI, 2.5 to 5.2] per 100 deliveries).

After excluding neonatal intensive care unit admission and neonatal transfer from our revised composite outcome for neonatal morbidity, the rate of neonatal morbidity was slightly higher among women receiving versus not receiving neuraxial analgesia (5,843 [6.3%] vs. 785 [5.5%]). Using this revised definition of the composite outcome, the unadjusted risk of neonatal morbidity associated with neuraxial analgesia was 1.15 (95% CI, 1.07 to 1.24; P < 0.001). After adjusting for potential confounders, we observed no statistically significant association between neuraxial analgesia and neonatal morbidity (adjusted relative risk, 1.07; 95% CI, 1.00 to 1.16; P = 0.054).

Results from models which accounted for within-county clustering using generalized estimating equation models are provided in Supplemental Digital Content 2 (https://links.lww.com/ALN/C490). For our primary outcome, neuraxial anesthesia was associated with an increased risk of neonatal morbidity (adjusted relative risk, 1.18; 95% CI, 1.12 to 1.25; P < 0.001). For the revised composite of neonatal morbidity, the adjusted risk was 1.09 (95% CI, 1.02 to 1.17; P = 0.014).

Details of the propensity score matched cohort and results from the propensity score analysis are provided in the Supplemental Digital Content 3 (https://links.lww.com/ALN/C491) and Supplemental Digital Content 4 (https://links.lww.com/ALN/C492). For our primary outcome, the rate and risk of neonatal morbidity was higher in women receiving versus not receiving neuraxial analgesia (10.4% vs. 8.9%, respectively; relative risk, 1.18; 95% CI, 1.10 to 1.26; P < 0.001). For the revised composite outcome, the association was not statistically different (relative risk, 1.03; 95% CI, 0.93 to 1.23; P = 0.606).

In this population-based cross-sectional study of 106,845 women who underwent operative vaginal delivery in the United States in 2017, neuraxial analgesia was not associated with a measurable reduction in neonatal morbidity. Contrary to our hypothesis, we observed a small increased risk of neonatal morbidity. After excluding neonatal intensive care admission and neonatal transfer from our composite neonatal outcome, and accounting for clustering by county, this association was either not or marginally statistically significant. Confounding by indication may explain the observed association between neuraxial analgesia and neonatal morbidity. However, this observational administrative dataset, which is unable to measure the clinician decision-making underlying the use of neuraxial analgesia and mode of delivery cannot definitively address this issue. Given these concerns and the small adjusted risk difference of neonatal morbidity associated with neuraxial analgesia, our findings should not be used as justification to advise women against using neuraxial labor analgesia.

Prior research suggests that neuraxial analgesia is not associated with serious neonatal morbidity or acidosis.^25–27  Yet, sparse data exist regarding the potential effects of neuraxial analgesia on neonatal outcomes among women undergoing operative vaginal delivery. In a Cochrane review of analgesia for forceps delivery,²⁸  only one study (published in 1980) was identified which compared spinal analgesia to a pudendal nerve block; no neonatal morbidity data were reported in this trial. In a more recent single-center study, epidural analgesia was not significantly associated with cephalohematoma among women undergoing forceps delivery (adjusted odds ratio, 0.89 [95% CI, 0.45 to 1.74]) versus no epidural analgesia.⁶  These wide CI are likely explained by the small number of neonates with cephalohematoma (n = 25). By comparison, our study included other important indicators of neonatal morbidity, including low Apgar scores and neonatal intensive care unit admissions, that are available using national birth certificate data.

Our a priori hypothesis was that neuraxial labor analgesia facilitates a controlled, atraumatic operative delivery, thereby resulting in less neonatal morbidity. Our main results indicated no reduction in neonatal morbidity associated with neuraxial analgesia. Instead, in our unadjusted analysis, we observed an increased risk of neonatal morbidity. Although the magnitude of this risk was lower after accounting for potential maternal and obstetric confounders, a small increased risk was observed in our primary multivariable model. Confounding by indication is a possible explanation for these findings. Clinical factors that may influence a decision to use neuraxial labor analgesia, such as prolonged or dystocic labor, immediate or potential fetal compromise, and fetal malposition, may also affect neonatal outcomes. Other potential sources of confounding by indication bias are hospital infrastructure factors. For example, a hospital with an obstetric anesthesia service may be more likely to have a neonatal intensive care unit and, thus, less likely to transfer sick infants to another facility. To address this concern, in post hoc analyses, we excluded neonatal intensive care unit and neonatal transfer from our composite of neonatal morbidity and observed no statistically significant association between neuraxial analgesia and the revised composite of neonatal morbidity in our full model and a propensity score matched model. After accounting for clustering by county of birth (the smallest unit of analysis), the adjusted risk for the revised composite outcome was of marginal statistical significance. These findings substantiate why confounding by indication is legitimate concern. Although a randomized trial would limit this and other forms of confounding bias, randomizing women to receive or not receive neuraxial labor analgesia in this context would be inappropriate, unethical, and not feasible. Therefore, prospective studies sourcing clinical data from large delivery hospitals or hospital systems provide the best opportunity to capture data on confounders that are unavailable in administrative or birth certificate datasets.

The observed association between neuraxial analgesia and neonatal outcome could be mediated by a number of obstetric-related factors, including providers’ skill and level of experience performing operative vaginal delivery, fetal position and station (the relation between the fetal presenting part and the ischial spines), type of forceps used (e.g., Simpson, Kjelland), suboptimal instrument placement, and method of instrumental delivery (i.e., direct traction forceps vs. rotational forceps).^16,29–31  As an example, a rotational forceps delivery may not be tolerated in women experiencing obstructed labor without neuraxial analgesia, but these complex operative vaginal deliveries may be associated with worse neonatal outcomes compared to an uncomplicated vacuum delivery in a patient without neuraxial analgesia. A systematic review of perinatal morbidity after rotational forceps delivery lend support to this assertion.³²  In this review, 8.9% of neonates required neonatal intensive care unit admission and 7.5% experienced neonatal trauma.³²  In our sensitivity analyses, the relative risk of neonatal morbidity associated with neuraxial analgesia varied according to instrument type (forceps, vacuum), with the point estimate of relative risk being slightly higher in the cohort undergoing forceps delivery. Although reasons for these different risk estimates are unclear, the choice of instrument may differ according to the anticipated complexity of delivery. Finally, the presence or absence of neuraxial labor analgesia may influence providers’ recommendations for delivery in women with obstructed or prolonged labor. For example, women without a preexisting epidural catheter may be more likely to undergo intrapartum cesarean delivery rather than a difficult operative vaginal delivery without neuraxial analgesia.

Our study has a number of limitations. The major limitation is that the observational nature of our study design means that we cannot determine cause and effect nor whether the decision to perform an operative delivery was made before or after placement of a neuraxial block. However, in our clinical experience, the practice of initiating neuraxial analgesia specifically for facilitating operative delivery is unusual. Second, nondifferential exposure or outcome misclassification may bias the effect of neuraxial analgesia toward the null.³³  For our exposure, previous studies report high sensitivity (85 to 96%) and a low false discovery rate (14 to 20%) of neuraxial analgesia documentation in birth certificates.³⁴  By comparison, in a previous study of 103 California hospitals, neonatal intensive care admissions from birth certificate data were underreported (Lin concordance coefficient = 0.33).³⁵  Studies of national data are lacking on the accuracy of neonatal complications reported on the birth certificate. Third, drug data are not recorded on the birth certificate, therefore we could not assess the effect of neuraxial drug type or dose on the reported association. Although older studies using high-concentration local anesthetic solutions to induce and maintain neuraxial analgesia found an association between neuraxial labor analgesia and operative vaginal delivery, recent investigations of low-concentration solutions used in contemporary practice have not.^26,36  In addition, details were not available for other analgesic modalities, including systemic opioids or nitrous oxide analgesia, which may have been used by women in the nonneuraxial group.

Based on the findings of this national cohort of women who underwent operative vaginal delivery in 2017, a neonatal benefit of neuraxial analgesia for operative vaginal delivery was not observed. Confounding by indication may explain the observed association between neuraxial analgesia and neonatal morbidity.

This study was supported and funded internally by the Department of Anesthesiology, Perioperative, and Pain Medicine, Stanford University School of Medicine, Stanford, California.

The authors declare no competing interests.