Epidural analgesia has been used to provide labor pain relief for more than 40 years. The technique has been refined over the past 20 years to provide laboring women with higher-quality pain relief, less leg weakness, and more control over the administration of pain relief medication. Since the early 1990s, the combined spinal-epidural (CSE) technique has become popular because it provides more rapid onset pain relief with minimal motor weakness.1,2 Various studies have suggested that CSE compared with epidural analgesia may provide superior overall pain relief during labor and faster rates of cervical dilation.3–5 However, other studies have shown no difference between the techniques in terms of pain relief or the incidence of cesarean deliveries.6,7
Poor study design and low numbers of subjects limit many early studies. Additionally, mostly trainees initiated epidural analgesia. To our knowledge, no large prospective study of CSE has been conducted in a private practice setting by experienced practitioners.
The primary outcome of this randomized clinical research trial was an assessment of pain, using a verbal rating pain scale (VRPS) of scores from 0 to 10, made at the end of the first stage of labor and shortly after delivery. The secondary outcomes were patient-controlled epidural analgesia (PCEA) use, number of epidural supplemental top-up doses administered by an anesthesiologist, epidural catheter replacements, side effects, and labor outcomes. We hypothesized that CSE would provide superior analgesia throughout labor and delivery.
Permission to conduct this study was provided in January 2007 by the IRB of Sharp Health Care. All patients provided written consent after the study had been explained to them. Inclusion criteria for participation included ability to speak and understand English, ASA physical status I to III, uncomplicated term labor, and patient request for neuraxial labor pain relief. Exclusion criteria included inability to speak or understand English, ASA physical status ≥IV, gestational age <37 weeks, malpresentation (e.g., breech), previous cesarean delivery, multiple gestations, prior uterine scar or cervical surgery, and body mass index (BMI) >40 kg/m2.
Upon request for neuraxial labor analgesia, subjects were randomly assigned to receive epidural or CSE analgesia. A statistician (J. B.) created randomization tables for 4 subgroups of 200 women in blocks of 20 using Microsoft Excel’s random number generator. Specifically, for each block of 20, random numbers between 0 and 1 were assigned to the integers 1 to 20, then sorted so that the integers with the smallest 10 random numbers were assigned to epidural group and the largest 10 random numbers to the CSE group. This was repeated with a new seed for each block and performed for each of the 4 subgroups. The subgroups were designed to stratify for parity (nulliparous versus parous) and induction versus spontaneous labor. Hence, the 4 subgroups were (a) nulliparous women in spontaneous labor, (b) nulliparous women in induced labor, (c) parous women in spontaneous labor, and (d) parous women in induced labor.
The patient, attending nurse, and obstetrician were blinded to group assignment. In some, but not all instances, the individual collecting data was blinded to group assignment. After providing consent, patients were assigned to the next available study number in 1 of the 4 subgroups by opening the next sequential sealed, opaque envelope. The anesthesiologist performing the block opened the envelope at the time of request for epidural analgesia.
Demographic data were collected on data entry forms specifically designed for the study. An investigator documented a VRPS score immediately before initiation of neuraxial analgesia. The patient was instructed to provide a numeric score of pain on a scale from 0 to 10, in which 0 represented no pain and 10 represented the worst imaginable pain. After analgesia induction (time 0 = administration of full epidural dose in the epidural group and after the intrathecal dose in the CSE group), the patient was prompted to inform the investigator when she rated her contraction pain as 1 or 0 out of 10. This was recorded as time to complete analgesia (first painless contraction). The end of the first stage of labor was defined as the first time that a vaginal examination indicated complete cervical dilation. At that time or shortly thereafter, the patient was asked by an investigator to rate her pain score during the first stage of labor using the 0 to 10 VRPS providing a range from the least pain felt to the most pain felt, and a “typical” score, i.e., an overall assessment of first-stage pain (e.g., she might respond that her least pain after the epidural was 0, her worst pain was 6, but that the typical [or overall] pain score during that period was 2). This assessment was used also to determine second-stage labor pain and similar questions were asked shortly after delivery (i.e., “From the time you achieved full cervical dilation until you delivered the baby, what was your lowest pain score, your highest pain score, and your typical [or overall] pain score during the second stage of labor?”). Additionally, after delivery, each subject was asked to provide a single number from 0 to 10 to represent the pain felt while delivering the baby (i.e., during the crowning and expulsive phases) and to rate their overall satisfaction with neuraxial analgesia using a 4-point scale (very good, good, fair, or poor). Finally, pain scores were assessed by the labor nurse at hourly intervals as a routine part of care.
The PCEA volume, number of requested and administered bolus doses, and the number of physician-administered supplemental top-up epidural bolus doses were recorded after delivery.
An assessment of pruritus during labor was made using a verbal scale from score 0 (no itching) to 10 (most intense itching imaginable). Motor block was assessed at the end of the first stage of labor and shortly after delivery (1 = no block, patient can lift bottom off the bed; 2 = able to straight leg raise; 3 = able to flex knee; 4 = able to flex ankle; 5 = unable to move leg or foot). Ephedrine use for maternal hypotension or fetal bradycardia was documented. The incidence of a nonreassuring fetal heart rate within 30 minutes of initiation of analgesia (fetal heart rate <90 bpm for >60 seconds) was recorded by the registered nurse and confirmed by the principal investigator.
Nurses treated hypotension based on the following guidelines. For systolic blood pressure (BP) <90 mm Hg: IV ephedrine 10 mg every 5 minutes to a maximum of 30 mg until BP was >100 mm Hg, and notify anesthesiologist. Treatment of hypotension was allowed for systolic BP 90 to 100 mm Hg if the patient was symptomatic (nausea or dizziness) or there was a simultaneous fetal bradycardia.
Demographic data collected included patient age, height, weight, ethnicity, gravidity, parity, and gestational age. The following additional information was recorded: cervical dilation at time of analgesia request; induced or spontaneous labor, and use of oxytocin augmentation; reason for induction; type of delivery (spontaneous vaginal, vacuum, forceps, cesarean); use of episiotomy; neonatal birth weight and Apgar scores at 1 and 5 minutes; neonatal resuscitation (intervention by the neonatal intensive care unit team); and admission to the neonatal intensive care unit. When available, umbilical artery blood gas results were recorded.
Neuraxial Analgesia Induction and Maintenance
All subjects had neuraxial analgesia induced in the sitting position while receiving an IV fluid bolus of at least 500 mL lactated Ringer’s solution. An epidural kit (Arrow®) containing a 17-gauge Tuohy needle and 19-gauge FlexTip Plus® single-orifice catheter, and a 26-gauge Gertie Marx® spinal needle (5 in.; IMD, Park City, UT) were used for this study. Loss of resistance was achieved with saline and/or air according to the preference of the anesthesiologist. Women assigned to the epidural analgesia group received 10 mL 0.125% bupivacaine with fentanyl 2 μg/mL in 2 equal divided doses via the epidural needle, followed by 5 mL through the catheter. Women assigned to the CSE group received 2.5 mL of the same solution via the spinal needle before epidural catheter placement. PCEA was used to maintain analgesia for the remainder of labor and delivery using bupivacaine 1.25 mg/mL with fentanyl 2 μg/mL basal infusion 6 mL/h, PCEA bolus dose 5 mL, lockout interval 5 minutes, maximum hourly dose 26 mL.
Sample size was based on a 2-group comparison (epidural analgesia versus CSE) and 2 outcome measures: the primary outcome measure of VRPS score, specifically the typical pain in the first and second stages of labor and at delivery (interval scale variables), and a secondary outcome of spontaneous vaginal delivery (binary variable). From the literature, expected differences in VRPS score ranged from 1 to 2 points on an 11-point scale (0 to 10), with a standard deviation also ranging from 1 to 2. Using the lower value of expected difference and the higher value of standard deviation, 64 subjects in each of 2 groups are required to achieve 80% power at a 5% 2-tailed significance level using 2-sample t test of the means. A second a priori sample-size determination showed that approximately 700 subjects would be required to detect a change in rate of spontaneous vaginal delivery from 65% to 75%, with 80% power and a 5% 2-tailed significance level. Therefore, we planned to recruit 800 subjects (400 per group) to account for possible study dropouts, lost data, and patient crossovers.
Continuous (interval scale) study variables were examined for normality and summarized with means and standard deviations for symmetric distributions; medians and interquartile ranges were used for skewed distributions. These variables include all pain scores, time to analgesia, PCEA volume, pruritus, motor block, cervical dilation, age, height, weight, gestational age, and birth weight. Binary, nominal, and ordinal categorical study variables were summarized using frequencies and percentages. These variables include labor type, delivery type, oxytocin augmentation, epidural top-up boluses, gravidity, parity, ethnicity, and Apgar scores.
Two-group comparisons of patient demographic and baseline variables were performed with independent 2-sample t tests for continuous variables and χ2 tests of independence for nominal categorical variables. For continuous variables with non-normal distributions, and for ordinal categorical variables, nonparametric Mann-Whitney U tests were also used. In addition, we calculated Student t test with unequal variances (Welch’s method). For the secondary analysis comparing 4 subgroups, 1-way analysis of variance and Kruskal-Wallis tests were used in place of the t tests and Mann-Whitney tests. All tests were 2-tailed. All analyses were done on an intent-to-treat basis, using SPSS version 18 statistical software (SPSS Inc., Chicago, IL).
The threshold P value of 0.05 for statistical significance was used for the primary and key secondary outcome variables. However, because of the large number of comparisons and the resulting inflated type I error rate, a Bonferroni-type adjustment (i.e., reducing the threshold P value for significance to 0.005 to account for approximately 10 comparisons) was made for secondary and tertiary outcomes. Although statistical significance was assessed from P values, both unadjusted and adjusted 95% confidence intervals (CIs) were calculated. For consistency with the hypothesis tests, the multiplier was set for 99.5% confidence and therefore CIs were calculated using an adjusted multiplier of 2.815. Only the adjusted CIs are reported.
Two exploratory analyses to determine predictors of pain at vaginal delivery were undertaken. Multiple linear regression was used for the interval scale VRPS score at delivery; multiple logistic regression was used to model the presence or absence of pain (VRPS score of zero versus nonzero) at delivery. Potential predictor variables were selected from those variables that showed statistically significant differences between study groups using bivariate analysis. These variables were supplemented by additional (but not statistically significant) variables deemed clinically important from the experience of the lead investigator (DG). Both forward and backward stepwise selection methods were used to identify significant predictors and develop the final models with each method.
The study was conducted within the labor and delivery unit at Sharp Mary Birch Hospital for Women and Newborns from March 26, 2007 until November 10, 2009. Although some protocol failures occurred, data from 398 subjects from the epidural group and 402 subjects from CSE group were included in the final analysis (Fig. 1). Women in both groups were similar in age, height, weight, ethnicity, gestational age, and cervical dilation at the time of epidural insertion (Table 1). The indications for induction of labor did not differ between groups (Table 1). Labor characteristics were similar regarding time from initiation of analgesia to the end of the first stage of labor (Table 2). The duration of the second stage of labor was not significantly different between groups (epidural group 68 minutes versus the CSE group 78 minutes; P = 0.04; 99.5% CI for difference: −3.9, 23.8). There were no significant differences between groups in method of delivery (Table 2) or use of episiotomy (27% epidural group versus 29% CSE group; P = 0.49).
Primary End Point
Baseline pain scores reported just before neuraxial analgesia were similar between groups, but time to complete analgesia was shorter in the CSE group (Table 3). Patients in the CSE group reported better analgesia during the first stage of labor based on a lower typical VRPS score (1.4 vs 1.9; P < 0.001; 99.5% CI of difference: −0.92, −0.14). Pain scores during the second stage of labor (1.7 vs 1.9; P = 0.17; 99.5% CI of difference: −0.82, 0.28) and at delivery (2.0 vs 2.0; P = 0.77; 99.5% CI of difference: −0.73, 0.59) were not different between groups. Scheduled hourly pain scores were similar between groups with the exception of first hour in which better analgesia was observed in the CSE group (Table 4).
Secondary End Points
Delivery type did not differ between groups, with an overall spontaneous vaginal delivery rate of 73% and a cesarean delivery rate of 14% to 16% (Table 2).
The incidence of epidural top-up boluses by an anesthesiologist was significantly less in the CSE group. The need for epidural catheter replacement was between 1% and 2% and not significantly different between groups (Table 3). PCEA variables were not statistically different between groups. Patient satisfaction was high in both groups.
Pruritus scores (mean ± SD) were higher in the CSE compared with the epidural group (2.3 ± 2.5 vs 1.7 ± 2.4; P = 0.002). The requirement for treatment was low in both groups (4.6% CSE vs 3.4% epidural; P = 0.40). No differences were found in the incidence of ephedrine use for hypotension (Table 5). The degree of motor block between groups was similar (data not shown).
The incidence of fetal bradycardia requiring an intervention within 30 minutes of initiation of analgesia was 8.5% in the CSE group versus 4.5% in the epidural group (P = 0.025; 99.5% for difference: −1.0%, 8.8%). All episodes of profound fetal bradycardia resolved with conservative measures alone, and there were no early emergency cesarean deliveries in either group. Neonatal outcomes were similar (Table 5). A comparison of outcomes associated with each subgroup is outlined in Table 6.
A multiple linear regression was performed with VRPS score at delivery as the outcome variable and using 9 binary predictor variables (type of analgesia, parity, induction, birth weight >4000 g, episiotomy, vacuum, forceps, duration of stage 2 >120 minutes, delivery within 1 hour of analgesia initiation, epidural top-up bolus) and 2 interval predictor variables (BMI and total PCEA volume at delivery). R2 for the full model was 8.2% and only 4 variables were statistically significant at the 0.05 level. Stepwise regression retained only these 4 variables: PCEA volume at delivery, BMI, duration of stage 2 >120 minutes, and delivery within 1 hour of analgesia initiation. This reduced model had an R2 of 7.4%. Because the VRPS score at delivery is strongly skewed, with a large number of 0 scores, a dichotomous version of the variable (zero versus nonzero) was modeled with stepwise multiple logistic regression. The same 4 predictor variables were retained in the final logistic regression model, with a Cox and Snell pseudo R2 of 6.1%. Results of both models are summarized in Table 7.
This study has demonstrated that a CSE technique is associated with more rapid onset and more first-stage analgesia than a traditional epidural technique. Fewer patients required epidural top-up boluses with CSE analgesia. There were no differences observed between groups in terms of epidural catheter replacement or second-stage and delivery pain scores.
This study is novel for several reasons: (1) a private practice setting extends external validity to prior comparisons; (2) the epidural technique used a bolus through the needle before threading the catheter to achieve the fastest analgesia onset possible for this arm of the study; (3) pain relief during the first and second stages of labor was assessed separately; and (4) pain relief at delivery was determined separately, and independent variables that influence pain at delivery were analyzed.
To our knowledge, no prospective study of CSE has been conducted in a private practice setting. However, one retrospective review has described the use of CSE in labor in a community hospital.8 High-dose intrathecal sufentanil (10–20 μg) was used to initiate CSE, which caused relatively high rates of pruritus, drowsiness, oxygen desaturation, and even dysphagia. Compared with reported rates of epidural catheter failure during labor, they found their CSE technique to have fewer analgesic failures.8 Over the past 10 to 15 years, the dose and composition of the intrathecal component have been reduced so that the incidence and severity of side effects is now significantly less.9,10 Today, many anesthesiologists are using intrathecal bupivacaine 2.5 mg with or without a small dose of opioid (e.g., fentanyl 5–10 μg or sufentanil 2–5 μg). Using 3.125 mg plain bupivacaine and 5 μg fentanyl for our study, we found that the CSE group experienced more first-stage analgesia, with a more rapid onset than a traditional epidural technique designed to produce rapid onset analgesia. Although the practice of injecting 0.125% bupivacaine with 2 μg/mL fentanyl through an epidural needle is controversial, it is a common practice among the experienced anesthesiologists in our hospital. Indeed, no adverse events were observed as a result of this practice, and the safety of this technique has been validated in other studies.11,12
The improvement in first-stage analgesia from a CSE technique may be the result of more accurate placement of the epidural catheter into the posterior epidural space after the confirmation provided by cerebrospinal fluid flow during the spinal component. Some anesthesiologists have suggested that dural puncture alone contributes to improved analgesia,13 although others have disputed this claim.14
In a meta-analysis of 19 trials, CSE techniques required less rescue analgesia than traditional epidural analgesia and obstetric outcomes were similar between groups.15 In our study, there was no difference between groups in duration of first stage of labor after epidural initiation, despite an earlier report that described more rapid cervical dilation with CSE in nulliparous parturients.4
In a large retrospective study,16 the incidence of inadequate analgesia and epidural catheter replacements was lower in women receiving a CSE technique compared with a traditional epidural technique. The epidural catheter replacement rate was 6.8%, significantly higher than the 1.6% rate described in our study. This might indicate that experienced practitioners are more successful at epidural placement than residents in training.
Goodman et al.17 studied breakthrough pain in parous women receiving either CSE or a traditional epidural technique. They found no difference in the number of supplemental top-up boluses required with a top-up rate of 44% for CSE and 51% for traditional epidural analgesia. In our study, parous women had much lower top-up rates of 6.5% for CSE and 15% for epidural. These differences might be explained, in part, by our choice of maintenance analgesia (PCEA 0.125% bupivacaine with 2 μg/mL fentanyl), compared with their use of a continuous infusion of 0.0625% bupivacaine. When observing multiple top-up boluses, our study again demonstrated much lower rates than the Goodman et al.17 study (14% vs 1% CSE and 15% vs 2% epidural). In contrast, a study of nulliparous women6 showed more multiple epidural “rescue doses” with CSE (36%) versus epidural (27%). However, in that study, CSE patients received less maintenance analgesia.6
Using post hoc stepwise multiple regression modeling, 4 factors were predictive of increased pain at vaginal delivery, namely: (1) delivery within an hour of epidural induction, (2) second-stage duration >2 hours, (3) larger BMI, and (4) high PCEA totals at delivery.
Some practitioners object to the use of CSE because epidural catheter function is untested in the first hour or so, and therefore not as reliable in an emergency as a standard epidural technique.18 However, our results agree with those of Norris19 who found CSE catheters to be more reliable than traditional (“stand-alone”) epidural catheters. Indeed, not only were there fewer epidural top-up bolus doses in the CSE group, but also the number of catheters replaced was the same in both groups, and only 2 catheters were reinserted at the time of cesarean delivery, both from the epidural group. Hence, in our practice setting, concerns about inadequate epidural catheter function with a CSE technique have been unfounded.
Another objection to the CSE technique in labor stems from an increased incidence of early fetal bradycardia.9,10,20 The present study has suggested this association but each event was treated successfully by conservative measures alone, and none required emergency cesarean delivery. The neonatal outcomes in these cases, and in those with no fetal bradycardia, were the same in both groups. Our incidence of early fetal bradycardia after the intrathecal dose was 8.5% compared with rates of 4% to 33% in other prospective studies.20,21 The etiology of these events is unknown, although one must exclude maternal hypotension and uterine tachysystole, and treat both immediately. Suggested future research of this phenomenon includes the use of prophylactic ephedrine immediately after the intrathecal dose and early use of sublingual nitroglycerine for uterine tachysystole. Our study was not designed to evaluate those interventions.
The limitations of this study included lack of a consistent double-blind design for each patient. However, the patient and her attending nurse and physician were always blinded to the technique. Concerns that an anesthesiologist giving a top-up bolus might be influenced by group assignment are not difficult to defend. A top-up bolus was required for any patient who had breakthrough pain, despite PCEA use. In our busy hospital, the on-call anesthesiologist would respond to all requests in the same manner. Despite an attempt to conceal upcoming randomization sequences to all investigators, the principal investigator had access to the randomization chart to occasionally check accuracy of recruitment. However, patients were assigned to the next study number in sequence for each subgroup studied by removing a numbered, sealed, opaque envelope from a study binder.
We did not control for total fluid preload because patients had different lengths of stay on the labor floor before requesting an epidural. However, all patients received a 500-mL co-load around the time of the epidural insertion, which reflects our normal clinical practice.
Arguably, the differences in pain scores demonstrated in this study are not clinically significant, even if they achieved statistically significant differences. However, when one considers the fact that fewer top-up doses were required to achieve the improvement in analgesia, it would seem that on balance CSE is the superior technique for first-stage analgesia. This assertion is further strengthened by the finding that more women from the CSE group had a typical score of zero at the end of the first stage of labor. However, there was no difference in second-stage or delivery pain relief between groups, with approximately 55% of all subjects reporting zero pain during delivery.
In conclusion, within the context of a large private practice setting, a CSE technique provides more rapid onset analgesia and more first-stage analgesia compared with a traditional epidural technique. This is achieved despite fewer epidural top-up injections in women receiving CSE analgesia. We found no differences between groups in terms of second-stage or delivery analgesia. Labor, delivery, and neonatal outcomes were similar between groups.
Name: David Gambling, MB, BS.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: David Gambling has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Jonathan Berkowitz, PhD.
Contribution: This author helped design the study, analyze the data, and write the manuscript.
Attestation: Jonathan Berkowitz has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Thomas R. Farrell, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Thomas R. Farrell has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Alex Pue, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Alex Pue has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Dennis Shay, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Dennis Shay has seen the original study data and approved the final manuscript.
This manuscript was handled by: Cynthia A. Wong, MD.
The authors thank all of the labor room nurses and anesthesiologists at Sharp Mary Birch Hospital for Women and Newborns for their assistance with this study. In particular, the authors acknowledge the input of Susan Faron, MN, RNC-OB, CNS (nursing liaison), Marna Morimoto (clerical assistant), and the 2 student research assistants, Sam Mansour and John Hawkins.