We originally designed a randomized, 2-arm clinical trial. However, it became clear during the recruiting process that most women in early labor, and even those in prenatal classes during the third trimester, had already made their decisions about the type of pain medication they wanted. Despite receiving an unbiased presentation of the 2 treatment options, women were reluctant to accept random assignment. After 6 weeks, a change in protocol allowed each subject who had refused randomization to choose either ITNs or epidural analgesia, taking into account the recommendations of her care providers. The response rate for eligible women asked to participate under the revised protocol was 66%.
Analgesia
Experienced anesthesiologists from a large private practice provided the analgesia. For ITNs, 0.25 mg morphine sulfate and 25 to 35 μg fentanyl were injected into the subarachnoid space via the L2-L3 or L3-L4 interspace. To decrease postpartum nausea and itching, naltrexone (6.25 mg sublingually) was given to all subjects in the ITNs group within 30 minutes of vaginal delivery. Epidural analgesia consisted of an 8- to 10-mL bolus of 0.25% bupivacaine with 50 μg fentanyl, followed immediately by a continuous infusion of 0.125% bupivacaine with 2 μg/mL fentanyl. The anesthesiologist selected an initial infusion rate between 8 and 12 mL/h. The infusion was discontinued or significantly decreased during the second stage of labor.
Outcomes
Study participants rated their pain using the visual analog scale (VAS). Research assistants instructed participants to place a mark on the 0- to 100-mm VAS scale at the time of request for analgesia, at placement, at 5, 30, and 60 minutes following placement, and every 60 minutes throughout the first stage of labor. One score was collected during the early portion of the second stage, and a retrospective overall rating was obtained within 24 hours of delivery. An analyst uninvolved with the data collection subsequently measured VAS scores to the nearest millimeter.
On the first postpartum day, each subject participated in a structured interview, using 5-point Likert scales to rate her overall satisfaction with pain control and her ability to walk and push during labor. Women used a standardized 3-point scale to rate the severity of symptoms, including nausea, vomiting, pruritus, urinary retention, headache, and the inability to walk or to push.
Primary outcomes were the mean pain scores for the first and second stages of labor, the duration of adequate pain relief (defined as a VAS score ≤ 30 mm), and the retrospective mean score for overall labor pain. Secondary outcomes included women’s satisfaction ratings and subjective experiences with side effects.
Statistical analysis
Initial sample size calculations indicated that 36 women would be needed, equally divided between 2 groups, for the study to have 85% power to detect a 10-mm difference in the group mean VAS scores in the second stage of labor. A 10-mm change in the VAS score represents 1 standard deviation11 and is considered clinically significant.12 As a result of more rapid enrollment in the ITNs group, the group sizes became unequal. We continued to recruit until there were 19 women in the epidural group.
We calculated a time-weighted mean VAS score for each subject for the first stage of labor. A VAS score of 30 mm or less was considered the “zone of analgesic success.”13 Mean VAS scores, the percentage of women whose scores remained in the analgesic success zone, and satisfaction ratings in the ITNs and epidural groups were compared using multivariate linear and logistic regression methods. The outcomes were adjusted for maternal age, parity, previous spinal analgesic use, cervical dilation at time of placement, oxytocin use prior to placement, and baby’s weight. Side effects were compared using 2 tests for categorical outcomes and Student t tests for the mean severity indices.
The duration of successful analgesia was compared using time-to-event methods—Kaplan-Meier life tables and the log-rank test.14 The event of interest was analgesic failure. The time to the event was the number of minutes after placement of spinal analgesic until the VAS score rose above 30.
Results
Eighty-two women enrolled and completed the study between May 1, 1999, and March 1, 2000; 63 received ITNs and 19 received epidurals. Only 9 women underwent random assignment. Demographic and baseline characteristics for the 2 study groups are shown in Table 1. There were no statistically significant differences between groups, although women in the epidural group were older and had slightly less cervical dilation at the time of analgesic placement. The mean VAS scores prior to analgesic placement were similar (65.3 for ITNs, 67.8 for epidural; P = .73).