Maternal Body Mass Index and Use of Labor Neuraxial Analgesia: A Population-based Retrospective Cohort Study

THE prevalence of maternal obesity in the United States is high, with one in four deliveries occurring to women who are obese (classified as a body mass index greater than or equal to 30 kg/m²).¹  Compared to nonobese women, obese women are at increased risk for pregnancy and peripartum complications, including preeclampsia, gestational diabetes mellitus, fetal macrosomia, stillbirth, postterm pregnancy, prolonged or dysfunctional labor, and cesarean delivery.² 

The anesthetic management of obese women in labor often requires careful multidisciplinary planning and preparation. Performing neuraxial blockade in obese women can be technically challenging because bony landmarks are more difficult to palpate,^3,4  back flexion is more limited, and the depth from skin to lumbar epidural space is increased.⁵  Analgesia may also be suboptimal because of failure to achieve bilateral dermatomal spread or epidural catheter dislodgement, both of which increase the need for catheter replacement.^6–8  Despite these technical concerns, neuraxial labor analgesia offers several important clinical benefits to obese women. From a physiologic standpoint, neuraxial analgesia may optimize maternal respiratory function and attenuate sympathetically mediated cardiovascular responses in obese women.^9,10  Obese parturients are at increased risk for cesarean delivery²  and difficult or failed tracheal intubation (if general anesthesia is performed for emergency cesarean delivery)^11,12 ; a preexisting epidural catheter can mitigate this risk because an epidural top-up dose can extend epidural analgesia to surgical anesthesia. Given these potential benefits, we would expect the rate of neuraxial labor analgesia to be greater in obese women. However, there are limited epidemiologic data examining whether rates of neuraxial labor analgesia differ between obese and nonobese women. In a retrospective cohort study of births in California between 2007 and 2010, Biel et al. found that overweight and obese women were more likely to receive regional labor analgesia.¹³  Data from a contemporary national cohort of women would be important for refining national guidelines for obstetric anesthetic management practices for obese women.

The aim of this study was to test the hypothesis that maternal obesity is associated with a higher rate of neuraxial labor analgesia use compared with nonobese parturients, and to estimate the difference in rate, if it exists.

As U.S. vital statistics data are both publicly available and deidentified, our analysis was exempt from Stanford University Institutional Review Board (Stanford, California) review.

This population-based retrospective cohort study used U.S. vital statistics data from 2009 to 2015 to evaluate the association between maternal obesity and neuraxial labor analgesia. This study period allows us to examine national obstetric anesthetic practices in a contemporary obstetric cohort. Data are based on a 100% sample of birth certificates from states that used the 2003 revision of the birth certificate. The 2003 revised birth certificate contains more detailed demographic, medical, and obstetric data compared to the 1989 version, including information on intrapartum methods of pain relief.¹⁴  The Centers for Disease Control and Prevention provide details of variables available in the revised birth certificate.¹⁵  Specifically, a checkbox exists on the revised birth certificate indicating whether or not “epidural or spinal anesthesia was used during labor.” The exact type of neuraxial block is not described, and no other information is captured for other types or modalities of labor analgesia.

For our analysis, we sourced data from all birth certificates that used the 2003 revision between 2009 and 2015. States using the revised format numbered 28 in 2009 (66% of all births), 33 in 2010 (76% of all births), 36 in 2011 (83% of all births), 38 in 2012 (86% of all births), 41 in 2013 (90% of all births), 47 in 2014 (96% of all births), and 48 in 2015 (96% of all births). Our study cohort consisted of women who underwent labor before vaginal or intrapartum cesarean delivery. Because the birth certificate contains data on whether labor occurred or not before delivery, we were able to identify women who labored. We excluded women with cesarean deliveries without previous labor, women who had a non-U.S. primary residence, out-of-hospital deliveries, and women with missing data for mode of delivery or neuraxial analgesia use.

Maternal body mass index at delivery was calculated based on the recorded weight and height at delivery, and was categorized using the World Health Organization body mass index categories¹⁶ : underweight (body mass index less than 18.5 kg/m²), normal body mass index (body mass index between 18.5 and 24.9 kg/m²), overweight (body mass index between 25 and 29.9 kg/m²), obese class I (body mass index between 30 and 34.9 kg/m²), obese class II (body mass index between 35 and 39.9 kg/m²), and obese class III (body mass index greater than or equal to 40 kg/m²). Women at the extremes of height (height less than 52 inches [1.32 m] or height greater than 76 inches [1.93 m]; data judged likely to be in error) or who had missing height or weight data were excluded.

The outcome variable was use of neuraxial labor analgesia; this outcome was established a priori (before data analysis). Based on literature review^17–21  and our combined clinical experience, we considered the following covariates as potential confounders in our analysis: maternal age, race/ethnicity, highest level of education, insurance type, marital status, prepregnancy hypertension, prepregnancy diabetes, previous cesarean delivery, previous live birth, prenatal care, multiple gestation, gestational diabetes, gestational hypertension, induction of labor, labor augmentation, gestational age at delivery, fetal presentation (classified as cephalic, breech, and other), attendant at birth (classified as doctor of medicine or osteopathy, midwife, and other), state where the delivery occurred, and year of delivery. Data were only available for the state in which each delivery occurred between 2009 and 2013.

Because the anesthetic management becomes increasingly technically complex for morbidly obese women,^6,7  we performed a subgroup analysis of women with a body mass index greater than or equal to 50 kg/m² to examine the unadjusted rates of neuraxial labor analgesia in the following body mass index subgroups: 50 to 60 kg/m², 60 to 70 kg/m², 70 to 80 kg/m², and greater than 80 kg/m².

Before data analyses, the main study aims and analytic plan were reviewed by all study investigators in May 2017. Factors that occur in nine or fewer patients are not reported in accordance with the National Center for Health Statistics data use agreement. We performed descriptive analyses to characterize patient characteristics across body mass index groups. An uncorrected chi-square test was used to compare the frequencies of categorical variables across body mass index classes. The margin command in Stata (StataCorp, USA) was used to determine the adjusted outcome values for each body mass index class and for comparing adjusted rates between body mass index classes. A two-sided P value less than 0.05 was considered to indicate statistical significance.

The associations between maternal body mass index and neuraxial labor analgesia were assessed using relative risk with 95% CIs and were estimated by multivariable log-linear regression analyses (with a Poisson distribution, a log-link function, and a robust variance estimator). The basic model on our analytic sample examined the unadjusted association between body mass index class and neuraxial labor analgesia. We then performed a series of sequential logistic models that grouped covariates into the following categories: sociodemographic factors, antenatal factors, pregnancy/peripartum factors, and attendant at birth. State and year of delivery were included as additional covariates in a final logistic model. Goodness of fit for each model was determined by likelihood ratio test statistics: log-likelihood, Akaike information criterion, and Bayesian information criterion. Additionally, because some women who underwent intrapartum cesarean delivery may have received de novo epidural or spinal blockade for surgical anesthesia instead of labor analgesia, we performed a sensitivity analysis (planned a priori), including only women who underwent vaginal delivery. Post hoc, we examined whether interaction exists between body mass index and a history of live births (categorized as no live births and one or more live births). Last, we examined unadjusted and adjusted risk differences with 95% CIs to present the magnitude of the absolute risk differences. The adjusted risk difference indicates the additional proportion of mothers who would receive neuraxial labor analgesia above a baseline proportion if the same maternal and obstetric characteristics were present in the reference group (women with normal body mass index).

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 2,182,797 women with normal body mass index and 1,442,133 women of obesity class III in our sample, an alpha of 0.05, a power of 80%, and an assumed proportion of nonobese women who received neuraxial labor analgesia of 61%,²²  the minimum detectable relative risk of neuraxial labor analgesia for obese class III compared to nonobese women was 1.002. Therefore, our study was adequately powered to detect a clinically relevant relative risk of 2.0 for use of neuraxial labor analgesia for obese class III women compared to nonobese women. All statistical analyses were performed using Stata Version 14.0.

We identified 24,363,280 deliveries that occurred between 2009 and 2015 using the updated birth certificate that resulted in a live birth. We excluded 6,757,394 cases because of residency outside the United States, out-of-hospital delivery, cesarean delivery without labor, and short or tall height. Also excluded were 385,206 records with missing height or weight, mode of delivery, or neuraxial labor analgesia use data. We assumed missing data were missing completely at random. A flow diagram depicting women who met exclusion criteria and our final analytic cohort is presented in figure 1. Our final analytic cohort comprised 17,220,680 deliveries, which accounts for 61.5% of 28 million births that occurred between 2009 and 2015 in the United States. In our analytic sample, 12,275,920 (71.3%) women used neuraxial labor analgesia.

Within our study cohort, 15,128 (0.1%) subjects were underweight, 2,182,797 (12.7%) had a normal body mass index, 6,368,656 (37%) were overweight, 4,881,938 (28.3%) were obesity class I, 2,330,028 (13.5%) were obesity class II, and 1,442,133 (8.4%) were obesity class III. Patient characteristics according to body mass index class and mode of delivery data are presented in table 1. Of note, 88.6% of women in our study cohort had vaginal delivery.

The unadjusted rate of neuraxial labor analgesia use differed according to body mass index class (fig. 2). Compared to normal body mass index women, the rate was lowest among underweight women (59.7%) and highest among women in obesity class III (75.6%; P < 0.001, respectively). After adjusting for sociodemographic, antenatal, obstetric, and peripartum factors, attendant at delivery, and state, the rates of neuraxial labor analgesia remained lowest for underweight women (65.1%) and highest in the obesity class III women (73.1%) compared to normal body mass index women (P < 0.001, respectively; fig. 2).

The results of the unadjusted and adjusted logistic regression analyses are presented in table 2. Using women with normal body mass index as the reference group, the unadjusted relative risk (model 1) of receiving neuraxial labor analgesia was modestly increased for overweight (relative risk, 1.03; 95% CI, 1.03 to 1.03), obesity class I women (relative risk, 1.05; 95% CI, 1.05 to 1.05), obesity class II women (relative risk, 1.08; 95% CI, 1.08 to 1.08), and obesity class III women (relative risk, 1.11; 95% CI, 1.11 to 1.11). In contrast, the unadjusted risk of neuraxial labor analgesia was decreased for underweight women (relative risk, 0.88; 95% CI, 0.86 to 0.89). The sequential addition of each set of covariates marginally altered these risk estimates. In the final model (model 5), women in obesity class III were at slightly increased risk of receiving neuraxial labor analgesia (adjusted relative risk, 1.06; 95% CI, 1.06 to 1.06), and underweight women had a slightly decreased risk (adjusted relative risk, 0.93; 95% CI, 0.92 to 0.94). The point estimates for all variables included in our final model are presented in table A1. In addition, we accounted for clustering according to the U.S. state where the delivery occurred using a generalized estimating equation; this analysis was done post hoc. Because there is a limit to the size of the matrices that can be computed with this approach, we used a 0.3% random sample of our study population using our generalized estimating equation approach; data are presented in table A2. The point estimates for overweight and women in each obese class were similar to those in model 5.

In the sensitivity analysis examining only vaginal deliveries, the risks of receiving neuraxial labor analgesia for each body mass index class after adjusting for all patient-level covariates (model 5) were similar to those observed in primary analyses: underweight women: adjusted relative risk, 0.93 (95% CI, 0.91 to 0.94); overweight women: adjusted relative risk, 1.02 (95% CI, 1.02 to 1.02); obesity class I women: adjusted relative risk, 1.03 (95% CI, 1.03 to 1.04); obesity class II women: adjusted relative risk, 1.04 (95% CI, 1.04 to 1.04); and obesity class III: adjusted relative risk, 1.05 (95% CI, 1.04 to 1.05). We also confirmed that interaction was present between body mass index and a history of live births, and performed stratified analyses according to patients’ history of previous live birth (table A3). The point estimates in each stratified model were similar to those observed in model 5.

The results of our post hoc subgroup analysis examining unadjusted rates of neuraxial labor analgesia among morbidly obese women (body mass index greater than or equal to 50 kg/m²) are presented in the table A4. The unadjusted rates among superobese women were 77.0%, 76.2%, 75.2% and 75.0% for the body mass index 50 to 60 kg/m², 60 to 70 kg/m², 70 to 80 kg/m², and more than 80 kg/m² groups, respectively.

Table 3 presents the unadjusted and the adjusted absolute risk differences in the rates of neuraxial labor analgesia use across body mass index classes. Compared to women with normal body mass index, the adjusted risk difference in neuraxial labor analgesia use increased from 1.4% (95% CI, 1.3 to 1.5) for overweight women to 4.2% (95% CI, 4.0 to 4.2) for obesity class III. In contrast, a negative adjusted risk difference was observed (–3.8%; 95% CI, –4.7 to –2.9) for underweight women compared to women with normal body mass index.

Based on U.S. national vital statistics data from more than 17 million deliveries, we observed only modest differences in the rate of neuraxial labor analgesia according to body mass index class, with a nonadjusted risk difference of 7.4% and an adjusted risk difference of 4.2% between women of obese class III and women with normal body mass index.

Given the perceived benefits of neuraxial labor analgesia among obese women (optimized respiratory function, attenuated sympathetic mediated changes related to labor and painful contractions, and mitigation of the risk of difficult or failed intubation for intrapartum cesarean delivery),^9–12  we expected to observe large differences in rates of neuraxial labor analgesia between obese women and women with normal body mass index. However, the likelihood of receiving neuraxial labor analgesia was only modestly increased in our unadjusted analysis. Furthermore, despite obesity being strongly linked with obstetric comorbidities, such as gestational hypertension, preeclampsia,^23,24  and gestational diabetes,²⁵  the magnitude of the associations between maternal body mass index and neuraxial labor analgesia remained relatively unchanged after accounting for these and other obstetric confounders in our adjusted analyses. These findings suggests that selected confounders minimally influence the relation between body mass index class and neuraxial labor analgesia.

Limited data exist describing the relations between maternal body mass index with neuraxial labor analgesia. Our findings are consistent with those of Biel et al., who examined these relations among a cohort of women who delivered in California between 2007 and 2010.¹³  Among primiparous women, a positive dose-response relation was observed between maternal body mass index class and the likelihood of receiving neuraxial labor analgesia compared to normal body mass index women; for class I, II, and III obesity, the adjusted relative risks were 1.14, 1.18, and 1.21, respectively. Similar findings were observed among parous women. The Biel et al. study did not account for obstetric confounders¹³ ; this may explain why the point estimates were marginally higher than those observed in our analysis.

Several explanations may contribute to our main findings. Because rates of neuraxial labor analgesia were high across all body mass index classes, any difference in rates of neuraxial labor analgesia between normal body mass index and obesity class III women are unlikely to be substantial. Limited differences in labor pain severity or pain sensitivity between obese and nonobese women may explain why there are only small differences in rates and the likelihood of neuraxial labor analgesia use between obese and normal body mass index women. Based on data from two prospective observational studies, Melzack et al. reported only a weak positive correlation (correlation coefficient [r] = 0.23 to 0.36) between body mass index and labor pain severity,²⁶  while Ranta et al. reported no differences in labor pain severity between obese and nonobese women.²⁷  Moreover, studies in nonpregnant humans and animal models have not consistently demonstrated a positive association between obesity and sensitivity to experimentally induced pain.²⁸  Other patient-level factors may explain our findings. For example, obese patients’ labor analgesia preferences may not markedly differ from those of nonobese patients. Last, several studies have reported that obesity is linked to an increased likelihood of epidural catheter failure.^6,7,11,29  Therefore, the small differences in the likelihood of receiving neuraxial labor analgesia between body mass index classes may be due, in part, to some obese women experiencing unsuccessful neuraxial blockade in their index or previous pregnancies.

The observation that underweight women were less likely to receive neuraxial labor analgesia compared to nonobese women was an unexpected finding. Although data from systematic reviews indicate that women who are underweight are at increased risk of perinatal morbidity, including small-for-gestational-age and low-birthweight babies, and preterm birth,^30,31  factors contributing to the lower use of neuraxial labor analgesia in this group are not known. Further research is needed to examine whether biologic or clinical factors (e.g., lower fetal weight, lower labor pain severity, or a shorter duration of labor) explain the moderately lower rates of neuraxial labor analgesia among underweight women.

The main strengths of our analysis include a national dataset comprising more than 17 million deliveries between 2009 and 2015. This allowed us to account for obstetric anesthetic practices within a contemporary population-based cohort. We performed sequential log-linear regression analyses and subgroup analyses that account for a large set of demographic and obstetric characteristics. However, several limitations should be acknowledged, including issues inherent to a retrospective observational study design. We could not account for medical contraindications to neuraxial blockade, such as severe thrombocytopenia. However, contraindications to neuraxial blockade are uncommon and are likely to be nondifferential by body mass index class. Additionally, our study does not include physician-level or hospital-level factors that may influence rates of neuraxial labor analgesia, such as availability of anesthesia and obstetric providers, and type of hospital (teaching or community). Future studies should consider models that include hospitals as random effects to account for hospital-level variation in neuraxial labor analgesia rates. However, because the effect of body mass index was small and remained relatively unchanged in our sequential models, any potential differential effect from these patient-level or hospital-level factors would need to be substantial to markedly alter the point estimates in our final model.

Birth certificate data may be less reliable than hospital data with regard to maternal comorbid conditions, with pregnancy- and labor-related complications being underreported on birth certificates.^32–35  The quality of data reporting on birth certificates may also vary by hospital and state.³³  However, any misclassification of covariates is likely to be nondifferential with respect to body mass index class. Although we could not determine the accuracy of the data for neuraxial labor analgesia use, a previous review reported on the accuracy of anesthetic and analgesic data in birth datasets.³⁶  The sensitivity of anesthetic and analgesic data was moderate to good (range, 62 to 97%), with high positive predictive values (range, 83 to 100%) and specificity (range, 96 to 100%).³⁶  Furthermore, in another review of selected medical and health data from birth certificates and hospital medical records among 1,095 births in two U.S. states, the sensitivity for neuraxial labor analgesia use was high (ranging from 85 to 96%).³³ 

We examined the association between body mass index at delivery and neuraxial analgesia because this accounts for gestational weight gain and reflects maternal body habitus of patients at the time of neuraxial block placement. Additionally, body mass index is a commonly used independent variable for assessing risks associated with childbirth and is a readily available indicator of obesity for large observational studies. However, we acknowledge that, by using the World Health Organization body mass index categories,¹⁶  an individual with class III obesity based on pregnancy weight may be different than an individual with class III obesity who is not pregnant. Last, our findings may not be reflective of the entire population of women who delivered in the United States between 2009 and 2015 because states adopted the 2003 revision of the birth certificate at different times.

We did not examine whether, among women receiving neuraxial labor analgesia, maternal or perinatal outcomes differ according to body mass index class. Therefore, there is a pressing need for studies examining whether maternal outcomes are improved in obese women receiving neuraxial labor analgesia compared with nonneuraxial or no analgesia. Biel et al. reported neuraxial labor analgesia was associated with a reduced risk of spontaneous vaginal delivery across all parity and body mass index categories.¹³  However, the validity of these associations is uncertain because of residual confounding and the potential for reverse causation. Future population-wide studies should account for nuanced clinical data, which include the timing of specific events. This information will advance our understanding of the complex relationships between body mass index, anesthetic and obstetric intrapartum interventions, and perinatal outcomes. These data will be important for refining future labor analgesia recommendations from national anesthesia societies for this cohort of parturients.

In conclusion, using national vital statistics data, we observed a small effect of maternal body mass index on the use of neuraxial labor analgesia. Given that neuraxial labor analgesia is perceived to be beneficial for obese parturients, future studies are needed to compare maternal and perinatal outcomes among obese women receiving neuraxial labor analgesia compared with nonneuraxial techniques or no labor analgesia.

This study was supported and funded internally by the Department of Anesthesia and Department of Obstetrics and Gynecology, Stanford University School of Medicine, Stanford, California. Dr. Butwick is supported by an award from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, Rockville, Maryland (1K23HD070972).

The authors declare no competing interests.