Matern Child Health J (2010) 14:705712 DOI 10.

1007/s10995-009-0515-9

Epidural Analgesia and Risks of Cesarean and Operative Vaginal Deliveries in Nulliparous and Multiparous Women
Uyen-Sa D. T. Nguyen Kenneth J. Rothman Serkalem Demissie Debra J. Jackson Janet M. Lang Jeffrey L. Ecker

Published online: 18 September 2009 Springer Science+Business Media, LLC 2009

Abstract Objective is to examine the effect of epidural analgesia in rst stage of labor on occurrence of cesarean and operative vaginal deliveries in nulliparous women and multiparous women without a previous cesarean delivery. Design of the Prospective cohort study. Prenatal care was received at 12 free-standing health centers, 7 private physician ofces, or 2 hospital-based clinics; babies were delivered at a free standing birth center or at 3 hospitals, all in San Diego, CA. This study of 2,052 women used data

U.-S. D. T. Nguyen (&) Institute for Aging Research Hebrew SeniorLife, 1200 Centre Street, Boston 02131, MA, USA e-mail: uyen-sanguyen@hrca.harvard.edu K. J. Rothman RTI Health Solutions, Research Triangle Park, NC, and Department of Epidemiology, Boston University School of Public Health, Boston, MA, USA e-mail: krothman@rti.org S. Demissie Department of Biostatistics, Boston University School of Public Health, Boston, MA, USA e-mail: demissie@bu.edu D. J. Jackson University of the Western Cape School of Public Health, Cape Town, South Africa e-mail: bessrfam@iafrica.com J. M. Lang International Relations at Watson Institute, Brown University, Providence, RI, USA e-mail: Janet_Lang@brown.edu J. L. Ecker Department of Obstetrics, Gynecology and Reproductive Biology, Harvard Medical School, Boston, MA, USA e-mail: JECKER@PARTNERS.ORG

from the San Diego Birth Center Study that enrolled women between 1994 and 1996 to compare the birthing management of the collaborative Certied Nurse MidwifeMedical Doctor Model with that of the traditional Medical Doctor Model. Main Outcome Measures of the Cesarean or operative vaginal deliveries. After adjusting for differences between women who used and those who did not use epidural analgesia in 1st stage of labor, epidural use was associated with a 2.5 relative risk (95% CI: 1.8, 3.4) for operative vaginal delivery in nulliparous women, and a 5.9 relative risk (95% CI: 3.2, 11.1) in multiparous women. Epidural use was associated with a 2.4 relative risk (95% CI: 1.5, 3.7) for cesarean delivery in nulliparous women, and a 1.8 relative risk (95% CI: 0.6, 5.3) in multiparous women. Epidural anesthesia increases the risk for operative vaginal deliveries in both nulliparous and multiparous women, and increases risk for cesarean deliveries in nulliparous more so than in multiparous women. Keywords Epidural Cesarean section Obstetrical extraction Midwifery Birth centers Prospective cohort study

Introduction Whether epidural use increases a womans risk for a cesarean delivery is a question that has received much debate. The American College of Obstetricians and Gynecologists (ACOG) Committee stated in 2002 that there is considerable evidence suggesting that there is in fact an association between the use of epidural analgesia for pain relief during labor and the risk for cesarean delivery [1]. It is thought that epidural analgesia causes dysfunctional labor, prolongs length of labor, relaxes pelvic oor muscle,

123

706

Matern Child Health J (2010) 14:705712

reduces maternal urge to push; thus, increasing the likelihood of a cesarean delivery [2]. In 2006, however, ACOG suggested that more recent studies offered no evidence that epidural analgesia increased the risk for cesarean delivery [1]. Several studies that swayed ACOGs change in opinion were those of Sharma et al. [3] and Wong et al. [4]. Yet, approximately 20% of subjects from the Sharmas study did not comply with the assigned treatment group. Moreover, in Wongs study, the risk for cesarean delivery in the referent group was high, possibly because some women in this group also received epidural during labor. Lack of adherence to the randomized assignment coupled with intent-to-treat analysis rather than as-treated (per protocol), and trial designs that do not distinguish completely between epidural and no epidural anesthesia, may have biased the results of the previous randomized studies in the direction of underestimating an effect of epidural anesthesia on the risk for cesarean delivery [5]. The purpose of the current research is to examine the effect of epidural placement on cesarean sections and operative births in a cohort of nulliparous women, and multiparous women without a previous cesarean delivery.

by Lieberman et al. [9], we also excluded women whose labor was induced with oxytocin and/or prostaglandin and/ or articial rupture of membrane (AROM); or if they had preterm deliveries (gestation \37 weeks), birth weight C 4,500 grams, or non-vertex presentation; or if they had a condition such as placenta previa, vaginal bleeding, prolapsed cord, pre-eclampsia, eclampsia, active herpes, non-reassuring fetal status, placenta abruption, clotting abnormality, or heparin treatment. Basically, to be included in this current analysis, women must have had a trial of labor and had no history of a medical condition that would require immediate cesarean section or preclude epidural placement. Further, the present study was restricted to only women who were nulliparous or were multiparous without a previous cesarean delivery. Of the 2,957 San Diego Birth Center Study women, 2,222 remained eligible for this secondary analysis after applying the above noted exclusions. Of those, we excluded: 109 multiparous women with previous cesarean sections, 35 women with tocolytics or magnesium sulfate that could indicate a difcult pregnancy and could confound the relation under study, 23 women with missing epidural status, and three women with missing parity; thus, our analysis included data from 2,052 women.

Methods Data Study Population Measurements Our study used data from the San Diego Birth Center Study (SDBCS), a prospective longitudinal study that compared safety and resource utilization in a birth center delivery model of collaborative care by certied nurse midwives and medical doctors, with a traditional model of hospital deliveries attended by medical doctors. More detailed information regarding the study and eligibility criteria has already been published [68]. In brief, the study investigators obtained approval by the human subjects committees at all participating institutions, and obtained written informed consent from participants before enrollment. The investigators enrolled a total of 2,957 women from February 1, 1994 to November 1, 1996. Of these, 1,808 were recruited from 12 freestanding health centers in which a birth center delivery was planned, and 1,149 women were recruited from 2 hospital-based prenatal care clinics and 7 private physician ofces. Since only women considered at low-risk for medical complications were permitted to deliver at the birth center, these same criteria were used to determine study eligibility of women in the traditional model of care. The SDBCS excluded women with multiple births, private or military insurance, and gestation at rst prenatal visit [32 weeks. In the current study, following methods For the San Diego Birth Center Study, we abstracted data from medical records including information concerning utilization of resources such as anesthesia. Epidural analgesia was measured as a dichotomous variable with yes indicating that epidural was placed in the rst stage of labor, no otherwise. The main outcome measure was route of delivery: operative vaginal delivery (dened as forceps delivery or vacuum extractions) or cesarean section delivery. Missing Covariate Data In general very little data were missing; a few variables were exceptions, however. Weight and height, for example, had 13.9% and 5.0% of data missing, respectively. To account for missing data, we used SAS PROC MI version 8.2 [10] to perform 10 simultaneous multiple imputations of variables with missing data, based on known demographic, personal characteristics and medical history data in our study population [11]. We included exposure, outcome, and confounding variables as detailed below, in the multiple imputation prediction model [1113]. SAS PROC MIANALYZE was used to summarize odds ratios of

123

Matern Child Health J (2010) 14:705712

707

potential confounders of cesarean and operative vaginal delivery over the 10 imputations for variable selection into the propensity scores models [11, 13].

Assessment of Potential Confounders to Include in Propensity Model We had to address the problem that women with difcult labor may have been more likely to choose epidural analgesia as compared with women with easy labor [14]. We estimated each womans propensity to use epidural analgesia [9, 15, 16], predicting propensity in multivariable logistic models that included potential confounders. We included variables that had OR \ 0.80 or OR [1.25 for the outcome in the propensity model predicting epidural use [17, 18]. One confounder in the analysis of method of delivery was whether the women enrolled in the collaborative Certied Nurse Midwife-Medical Doctor Model or the traditional Medical Doctor Model of care because women in the traditional model of care have easier access to epidural analgesia. Other confounders were age (years), race/ ethnicity, marital status, height (inches), weight (pounds), education level (years), country of birth, language spoken, payment method, prior medical history, major antepartum complications, cervical dilation at admission, cervical station at admission, rupture of membrane (ROM) at presentation, and any complications at presentations including poor labor progress and non-reassuring fetal heart rate, being admitted at presentation, and narcotics use at rst stage of labor, birth weight (grams), sex, and gestation at birth (months). We ranked subjects by their propensity scores and grouped them into quintiles, with women in quintile 1 having the least propensity while those in quintile 5 having the most propensity for epidural analgesia in rst stage of labor. Since nulliparous women had a different baseline risk for cesarean and operative vaginal delivery compared with multiparous women, the effect of epidural placement in rst stage of labor on each of these outcomes may also differ by parity [9, 19]. Therefore, all analyses were stratied by parity for operative vaginal delivery and cesarean delivery. For the analysis of operative vaginal delivery, among nulliparous women, the variables for the propensity model included model of care, cervical dilation at admission \4 cm, major antepartum complication, any complications at presentation, birth weight, gestation at birth, childs sex, maternal age, height, weight, education, marital status, maternal race, and language spoken. For multiparous women, the propensity model included cervical station and maternal country of birth but excluded gestation at birth, childs sex, and maternal age.

For the analysis of cesarean delivery, among nulliparous women, the propensity model was similar to that for operative vaginal deliveries in nulliparous women, it included: model of care, cervical dilation at admission \4 cm, major antepartum complication, any complications at presentation, birth weight, gestation at birth, childs sex, maternal age, height, weight, education, marital status, maternal race, and language spoken; for cesarean delivery, the propensity model also included cervical station, maternal country of birth and maternal narcotics use. For multiparous women without previous history of cesarean sections, the propensity model was similar to that for cesarean delivery outcome in nulliparous women, except that the model also included ROM at presentation and excluded childs sex. Data Analysis Given the observational nature of our study, we controlled for propensity for epidural use, and thus for all variables that were included in the prediction models for the propensity score. We used stratication [20] coupled with standard methods for pooling data across propensity strata to obtain estimates of risk difference and risk ratio, and their respective 95% condence intervals [21]. All analyses were performed separately by parity for each of the methods of delivery.

Results Among women included in this analysis, which excluded women if they had medical conditions that required immediate cesarean sections or precluded the use of epidural, 754 (36.7%) of the women had epidural placement, 252 (12.3%) had an operative vaginal delivery, and 110 (5.4%) had a cesarean section delivery. Distribution of potential confounders by epidural exposure status is shown on Table 1. Women who had epidural were less likely to be Hispanic, Mexican born, or Spanish speaking. Furthermore, there were more women with complications at presentation, more nulliparous women, and more women with latent labor in the epidural exposed group compared with the no epidural group. Crude and propensity adjusted estimates of epidural analgesia on risk for operative vaginal and cesarean deliveries are shown in Tables 2 and 3, respectively. There was a 25.1% increase in absolute risk (95% CI: 19.8, 30.3), or a crude risk ratio of 4.0 (95% CI: 2.9, 5.5), for operative vaginal delivery among nulliparous women who had epidural anesthesia in 1st stage of labor compared to those who did not. After control of confounding by stratication on propensity score, nulliparous women who received epidural anesthesia had a 19.3% higher absolute risk for

123

708 Table 1 Characteristics of potential confounders in the San Diego birth center study cohort between 1994 and 1996, according to epidural status

Matern Child Health J (2010) 14:705712

Epidural analgesia in rst stage Yes (n = 754) Collaborative model Age (years) \20 2034 C35 Maternal height (inches) B60 6164 C65 Maternal weight (pounds) \161 161180 C181 Married Education (years) \9 912 [12 Race/ethnicity White, nonhispanic Hispanic African American Other Place of birth U.S. Mexico Other Language English only Bilingual Spanish only Other Perinatal risk factors Major antepartum complications Complications at presentation in labor Parity/delivery history Nullipara Multipara Status of membrane Ruptured Ruptured [=12 h 195 34 26.4 19.0 310 68 24.4 25.6 403 351 53.5 46.6 475 823 36.6 63.4 29 308 3.9 40.9 53 401 4.1 30.9 196 270 256 22 26.3 36.3 34.4 3.0 162 413 666 25 12.8 32.6 52.6 2.0 311 380 48 42.1 51.4 6.5 259 936 73 20.4 73.8 5.8 135 513 72 34 17.9 68.0 9.6 4.5 121 1,098 47 32 9.3 84.6 3.6 2.5 138 468 110 19.2 65.4 15.3 431 682 156 33.7 53.7 12.2 530 65 55 262 81.5 10.0 8.5 35.1 986 86 49 583 88.0 7.7 4.4 45.1 117 394 201 16.4 55.3 28.2 306 708 223 24.7 57.2 18.0 210 511 33 27.9 67.8 4.4 255 976 67 19.7 75.2 5.2 286 % 37.9 No (n = 1,298) 1,044 % 80.4

RD

95% CI

-42.5 8.2 -7.4 -0.8 -8.3 -1.9 10.2 -6.4 2.3 4.1 -10.0 -14.5 11.6 3.1 8.6 -16.6 5.9 2.0 21.7 -22.4 0.7 13.6 3.7 -18.2 1.0 -0.3 10.0

-46.6, -38.4 4.3, 12.1 -11.5, -3.3 -2.7, 1.1 -11.9, -4.7 -6.5, 2.7 6.3, 14.1 -10.0, -2.9 -0.5, 5.1 1.6, 6.5 -14.4, -5.6 -18.4, -10.6 7.2, 16.1 -0.1, 6.3 5.4, 11.7 -20.4,-12.7 3.6, 8.3 0.3, 3.8 17.5, 25.9 -26.7, -18.1 -1.5, 2.9 9.9, 17.2 -0.7, 8.0 -22.6, -13.8 -0.5, 2.4 -2.0. 1.5 5.6, 14.3

16.9 -16.9 2.0 -6.7

12.4, 21.3 -21.3,-12.4 -2.0, 5.9 -14.4, 1.2

operative vaginal deliveries (95% CI: 12.9, 25.8), or a relative risk of 2.5 (95% CI: 1.8, 3.4), as compared with women without epidural. Among multiparous women,

those with epidural anesthesia had a 16.6% increase in absolute risk (95% CI: 12.5, 20.8), or a crude relative risk of 11.5 (95% CI: 6.4, 20.7), for operative vaginal delivery

123

Matern Child Health J (2010) 14:705712 Table 1 continued

709

Epidural analgesia in rst stage Yes (n = 754) % Newborn characteristics C41week gestation C4,000 g Childs sex Female Male Diagnosis at presentation Latent labor Active labor Cervical dilation admission \4 cm (early) Narcotics in rst stage 380 302 52.6 444 41.8 755 54.8 442 54.1 249 33.2 335 35.6 36.5 20.4 25.8 371 382 49.3 653 50.7 645 50.3 49.7 122 77 16.2 192 10.2 106 14.8 8.2 No (n = 1,298) %

RD

95% CI

1.4 -1.9, 4.6 2.0 -0.6, 4.7 -1.0 -5.5, 3.5 1.0 -3.5, 5.5 17.0 12.5, 21.5 18.3 13.7, 22.9 33.7 29.4, 38.0 7.4 3.2, 11.5

60.6 -18.8 -23.3, -14.3

Fetal cervical station -2 or -3 377 379 250

Table 2 Crude effect of epidural on method of delivery in the San Diego birth center study cohort between 1994 and 1996, by parity status Method of delivery Unadjusted estimates of effect and 95% condence intervals P1* (%) Assisted Nulliparous Multiparous Cesarean Nulliparous Multiparous 71/403 (17.6) 10/351 (2.8) 22/475 (4.6) 7/823 (0.8) 13.0 2.0 8.8, 17.2 0.1, 3.9 3.8 3.3 2.4, 6.0 1.3, 8.7 135/403 (33.5) 64/351 (18.2) 40/475 (8.4) 13/823 (1.6) 25.1 16.6 19.8, 30.3 12.5, 20.8 4.0 11.5 2.9, 5.5 6.4, 20.7 P2 (%) RD (%) 95% CI RR 95% CI

* P1 is the crude proportion with outcome among those with epidural use during 1st stage, P2 is crude proportion with outcome among those without epidural use

compared to women without epidural. After adjusting for propensity score, the risk difference was 13.7% (95% CI: 9.1, 18.2), and the risk ratio was 5.9 (95% CI: 3.2, 11.1). With regard to cesarean sections, nulliparous women who had epidural placement in rst stage of labor had a 13.0% increase in absolute risk (95% CI: 8.8, 17.2), or a crude relative risk of 3.8 (95% CI: 2.4, 6.0), for this method of delivery compared with women without epidural placement in the rst stage of labor. Adjusting for propensity score, there was a 10.4% increase in absolute risk (95% CI: 5.2, 15.7), or a 2.4 relative risk (95% CI: 1.5, 3.7), for cesarean sections. As for multiparous women, those who had epidural placement in the rst stage of labor had a 2.0% increase in absolute risk (95% CI: 0.1, 3.9), or a relative risk of 3.3 (95% CI: 1.3, 8.7), for cesarean delivery. After adjusting for confounding, the adjusted risk difference was 1.2% (95% CI: -0.9, 3.3), and the adjusted risk ratio was 1.8 (95% CI: 0.6, 5.3).

Discussion The results of our study indicate that epidural analgesia in rst stage of labor is associated with an increased risk for operative vaginal and cesarean deliveries. Our estimated 19.3% increase in absolute risk for operative vaginal delivery in nulliparous women was slightly higher than ndings from Lieberman et al. [9] or Sharma et al. [22], but provided similar inferences. Lieberman and colleagues found that women with epidural analgesia had an operative delivery rate at 19% compared with a 4% operative delivery rate among women not on epidural analgesia, where as a randomized controlled trial by Sharma and colleagues indicated that 12% of women randomized to epidural analgesia had forceps deliveries compared with 3% of women randomized to the comparison group. Regarding cesarean delivery, our estimated 2.4 fold increase in relative risk among nulliparous women were

123

710 * Proportions are stratied by quintiles of propensity score as indicated by the rst subscripted number, followed by the second subscript of 1 indicating the proportion with outcome among those with epidural use during 1st stage in that quintile, and 2 indicating proportion with outcome among those without epidural use in that quintile 3.2,11.1 RRMH 95%CI

Matern Child Health J (2010) 14:705712

lower than those presented in at least two other studies. In the clinical trial study by Thorp et al. [23], 25% of nulliparous women who had epidural analgesia had cesarean sections as compared with only two percent of women who did not have epidural (RR = 10). Results from Liebermans study [9] indicated that women on epidural analgesia had a 3.7 fold increase in risk for cesarean (95% CI: 2.4, 5.7) delivery compared with women not on epidural analgesia. The current literature, however, has conicting results of the effect of epidural analgesia on the risk for cesarean delivery. Several studies, including a natural experiment [24] and several clinical trials [3, 4], reported near-null results of epidural analgesia on risk for cesarean delivery among nulliparous women. Zhang and colleagues [24] compared risk of cesarean delivery between groups of women who delivered before the period in which epidural was made freely available (October 1, 1993) with that of women who delivered after that period. There was a difference of as much as 4 years, however, between the two non-concurrent comparison groups. During that time it is possible that there were changes in clinical practice and management that accompanied the availability of epidural analgesia that could explain the near null result, such as delay of admission until cervical dilatation reached 4 cm [25]. Sharma and colleagues [3] pooled data from ve clinical trials of nulliparous women who were randomized between 1993 and 2000 to epidural analgesia or intravenous opioids for pain relief during labor. They concluded that there was little difference in the risk for cesarean delivery between the two groups. A problem with this interpretation, however, is that 20% of women did not adhere to their assigned method of pain control. Lack of adherence to the assignment makes the results in the compared groups similar; thus, resulting in bias [25, 26]. In another randomized trial, Wong and colleagues [4] found that among nulliparous women who were randomized to receive intrathecal fentanyl at the rst request for analgesia and epidural at the second request, 17.8% had a cesarean delivery. In the comparison group, which was offered systemic hydromorphone at rst or second request for analgesia, followed by an epidural upon the third request or if the womans cervical dilation was at 4 cm, 20.7% had a cesarean delivery. This study design also suffered from bias in the direction of underestimating any difference, because epidural analgesia was offered to some women in both groups. In addition, in a review by Klein [27] of the nine clinical trials conducted from 1974 to 1998 that were cited in a 2000 Cochrane Review which concluded that epidural analgesia did not raise rates of cesarean deliveries (OR = 1.3, 95% CI: 0.93, 1.83), Klein found that ve of those studies had odds ratios ranging from 1.19 to 6.51 but

1.8,3.4

1.5,3.7 5.2,15.7 2.4 3/30 (10.0) 10.4% 7/60 (11.7) 26/145 (17.9) 18/116 (15.5) 8/91 (8.8) 7/40 (17.5) 3/136 (2.2) 16/85 (18.8) Nulliparous 4/17 (23.5) 1/158 (0.6) Cesarean

Table 3 Effect of epidural on method of delivery pooled over quintiles of propensity for epidural in the San Diego birth center study cohort between 1994 and 1996, by parity status

Propensity-adjusted mantel haenszel summary statistics

9/60 (15.0) 51/141 (36.2) 10/34 (29.4) 19.3% 12.9,25.8 2.5

9.1,18.2 5.9

RDMH 95%CI

4/69 (5.8)

13.7%

P52 (%)

35/166 (21.1)

5th Quintile

P51 (%)

19/95 (20.0) 4/140 (2.9)

P42 (%)

9/85 (10.6) 38/116 (32.8)

4th Quintile

P41 (%)

8/55 (14.6) 1/180 (0.6)

P32 (%)

Nulliparous 5/18 (27.8) 5/158 (3.2) 12/37 (32.4) 7/138 (5.1) 29/91 (31.9)

3rd Quintile

P31 (%)

2/18 (11.1) 1/217 (0.5)

Quintiles of propensity score for epidural use

2nd Quintile

P22 (%)

P21 (%)

3/217 (1.4)

1st Quintile

P12 (%)

Multiparous 0/17 (0.0)

Method of delivery

123

Assisted

Multiparous 0/13 (0.0)

P11* (%)

0/221 (0.0)

1/24 (4.2)

1/211 (0.5)

1/47 (2.1)

2/188 (1.1)

2/100 (2.0)

3/135 (2.2)

6/167 (3.6)

1/68 (1.5)

1.2% -0.9,3.3

1.8

0.6,5.3

Matern Child Health J (2010) 14:705712

711

four of those studies could not rule out random variation as a possible explanation. When Klein performed a sensitivity analysis of these ve clinical trials by excluding women who were randomized to the epidural arm in their active phase of labor (cervical dilation [4 cm), the odds ratio for cesarean delivery actually doubled for those on epidural (OR = 2.6, 95% CI: 1.3, 5.2). The discordant ndings among published studies and the divergent conclusions reached in reviews and meta-analyses concerning the effect of epidural anesthesia on risk for cesarean or operative vaginal deliveries argue for more careful research in this area. We may be past the point where ethical randomized trials could be conducted to address the issue, and such studies present other challenges, such as enormous expense and non-adherence to protocol. Nonexperimental studies that allow rigorous control of confounding may provide data needed to resolve this debate. Meanwhile, patients should be informed of the continuing controversy. Many will continue to elect epidural anesthesia while in labor, but they should do so with the understanding that their choice may affect the progress of labor, and ultimately, the route of delivery. The limitations and strengths of the current study need to be addressed. One limitation of our study was the lack of data regarding type and dose of epidural analgesia. It is possible that different management or dose of epidural medications could explain some of the divergence among studies. Another limitation is that clinical practice from 1994 to 1996, the period that our study was conducted, may differ from current clinical practice regarding epidural use or cesarean deliveries. Still, Declerq and Chalmers found that epidural use was 71% in the US [28]. In addition, Menacker et al. [29] noted that the rates of cesarean section have increased by over 40% since 1996 and that increases are seen for women of all ages, races and medical conditions. This study has several strengths in comparison with other reports. One is the exclusion of women with major complications to prevent confounding by the presence of these complications, and the use of the propensity score method to efciently handle a multitude of potential confounders that could not be excluded. Furthermore, we stratied all our analyses by parity and past history of cesarean delivery. Another strength is that our study population is mostly of Latina of low medical risk but high socio-economic risk. According to the latest data from the US Census Bureau, Latinos are the largest and fastestgrowing group of immigrants [30], and immigrant Hispanic women have the highest birth rate compared with US-born Hispanic women and non-Hispanic women [31]. The results of this study may have important implication for this population of women, especially in US cities along the USMexico border.

In conclusion, we controlled confounding with restriction criteria, stratication on parity and history of cesarean delivery, and stratication by propensity scores. These methods cannot control for unmeasured factors, but they are proven methods for preventing and removing confounding for measured variables [32, 33]. We believe that this study is the rst prospective cohort study to examine the effect of epidural analgesia on risk for operative vaginal deliveries and cesarean sections with rigorous control of confounding factors in a group of multiparous women without a previous cesarean delivery. Our ndings support the hypothesis that epidural anesthesia increases the risk considerably for operative vaginal deliveries in both nulliparous and multiparous women, and for cesarean deliveries in nulliparous more than in multiparous women to some degree. Receiving epidural analgesia during rst stage of labor could play an important role in a womans risk for cesarean and operative vaginal delivery, and the increased associations we described should be considered as patients and providers consider choices regarding pain management during labor.
Acknowledgments Sincere appreciation goes to William H. Swartz, MD, and the participants of the SDBCS for their time and dedication; and to Marian T. Hannan, DSc, MPH, and Judith T. Fullerton, PhD, CNM, FACNM, for their generous support and review of this manuscript.

References
1. ACOG Committee Opinion: Analgesia and cesarean delivery rates. (2006). Obstet gynecol, 107(6), 1487-8. 2. Thorp, J. A., Parisi, V. M., Boylan, M. B., & Johnston, D. A. (1989). The effect of continuous epidural analgesia on cesarean section for dystocia in nulliparous women. American Journal of Obstetrics and Gynecology, 161, 670675. 3. Sharma, S. K., McIntire, D. D., Wiley, J., & Leveno, K. J. (2004). Labor analgesia and cesarean delivery. Anesthesiology, 100, 142148. 4. Wong, C. A., Scavone, B. M., Peaceman, A. M., et al. (2005). The risk of cesarean delivery with neuraxial analgesia given early versus late in labor. New England Journal of Medicine, 352(7), 655665. 5. Rothman, K. J. (1977). Epidemiologic methods in clinical trials. Cancer, 39, 17711775. 6. Swartz, W., Jackson, D., Lang, J., et al. (1998). The birthplace collaborative practice model: Results from the San Diego birth center study. Primary care update for Ob/Gyns, 5(4), 207. 7. Swartz, W. H. (1999). Final report: A prospective study of an out-of-hospital birth center. Washington DC/San Diego: Agency for Health Care Policy and Research/Primary Care Perinatal Services. 8. Jackson, D. J., Lang, J. M., Dickinson, C., et al. (1994). Use of the nurse-midwifery clinical dataset for classication of subjects in birth center research. Journal of Nurse-Midwifery, 39(4), 197213. 9. Lieberman, E., Lang, J. M., Cohen, A., et al. (1996). Association of epidural analgesia with cesarean delivery in nulliparas. Obstetrics and Gynecology, 88(6), 9931000.

123

712 10. Yuan, Y.C. (2000). Multiple imputation for missing data: concepts and new development. In: Proceedings of the twenty-fth annual SAS users group international conference. SAS institute, Paper No. 267. Available at: http://support.sas.com/rnd/app/ papers/multipleimputation.pdf. 11. Von Hippel, P. T. (2005). How many imputations are needed? A comment on Hershberger and Fisher (2003). Structural Equation Modeling, 12(2), 334335. 12. Norton, N. J., Lipsitz, S. R., & Parzen, M. (2003). A Potential for bias when rounding in multiple imputation. American Statistician, 57(4), 229232. 13. Schafer, J. L., & Olsen, M. K. (1998). Multiple imputation for multivariate missing-data problems: a data analysts perspective. Multivariate Behavioral Research, 33(4), 545571. 14. Camann, W. (2005). Pain Relief during labor. New England Journal of Medicine, 352(7), 718720. 15. DAgostino, R. B., Jr. (1998). Propensity score method for bias reduction in the comparison of a treatment to a non-randomized control group. Statistics in medicine, 17, 22652281. 16. Cepeda, M. S., Boston, R., Farrar, J. T., et al. (2003). Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders. American Journal of Epidemiology, 158, 280287. 17. Hernan, M. A., Hernandez-Diaz, S., Werler, M. M., et al. (2002). Causal knowledge as a prerequisite for confounding evaluation an application to birth defects epidemiology. American Journal of Epidemiology, 155, 176184. 18. Brookhart, M. A., Schneeweiss, S., Rothman, K. J., et al. (2006). Variable selection for propensity score models. American Journal of Epidemiology, 163, 11491156. 19. Jackson, D. J., Lang, J. M., Ecker, J., et al. (2003). Impact of collaborative management and early admission in labor on method of delivery. Journal of Obstetric, Gynecologic, and Neonatal Nursing, 32, 147157. 20. Rothman, K. J., & Greenland, S. (1998). Introduction to stratied analysis. In K. J. Rothman & S. Greenland (Eds.), Modern epidemiology (2nd ed., pp. 253300). Philadelphia, PA: Lippincott Raven. 21. Rothman, K. J. (2002). Episheet. Available at: http://www.ken. rothman.name. 22. Sharma, S. K., Alexander, J. M., Messick, G., et al. (2002). Cesarean delivery a randomized trial of epidural analgesia versus

Matern Child Health J (2010) 14:705712 intravenous meperidine analgesia during labor in nulliparous women. Anesthesiology, 96, 546551. Thorp, J. A., Hu, D. H., Albin, R. M., et al. (1993). The effect of intrapartum epidural analgesia on nulliparous labor: A randomized, controlled, prospective trial. American Journal of Obstetrics and Gynecology, 169(4), 851858. Zhang, J., Yancey, M. K., Klebanoff, M. A., Schwarz, J., & Schweitzer, D. (2001). Does epidural analgesia prolong labor and increase risk of cesarean delivery? A natural experiment. American Journal of Obstetrics and Gynecology, 185, 128134. Lieberman, E. (2004). Epidemiology of epidural analgesia and cesarean delivery. Clinical Obstetrics and Gynecology, 47(2), 317331. Rothman, K. J., & Greenland, S. (1998). Precision and validity in epidemiologic studies. In K. J. Rothman & S. Greenland (Eds.), Modern Epidemiology (2nd ed., pp. 115134). Philadelphia, PA: LippincottRaven. Klein, M. C. (2006). Does epidural analgesia increase rate of cesarean section? Canadian Family Physician, 52, 419421. Declercq, E., & Chalmers, B. (2008). Mothers reports of their maternity experiences in the USA and Canada. Journal of Reproductive and Infant Psychology, 26(4), 295308. Menacker, F., Declercq, E., & Macdorman, M. F. (2006). Cesarean delivery: background, trends and epidemiology. Seminars in Perinatology, 30(5), 235241. Witt, H. Latinos still the largest, fastest-growing minority, census shows. Los Angeles Times May 1 2008. Available at: http:// www.latimes.com/news/nationworld/nation/la-na-census1-2008 may01,1,3417355,print.story. Statistical Portrait of Hispanics in the United States. (2006). Pew Hispanic Center Tabulation of 2006 American Community Survey (1% IPUMS). Available at: http://pewhispanic.org/les/ factsheets/hispanics2006/Table-10.pdf. Accessed May 22, 2008. Kurth, T., Walker, A. M., Glynn, R. J., et al. (2005). Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. American Journal of Epidemiology, 163, 262270. Soledad Cepeda, M., Boston, R., Farrar, J. T., et al. (2003). Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders. American Journal of Epidemiology, 158, 280287.

23.

24.

25.

26.

27. 28.

29.

30.

31.

32.

33.

123

Copyright of Maternal & Child Health Journal is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.