Programmed intermittent boluses vs continuous epidural infusion in labor using an adrenaline containing solution: A randomized trial

Traditionally, epidural analgesia has been maintained using a continuous infusion (CEI) with the addition of patient‐controlled boluses (PCEA). In recent years, programmed intermittent boluses (PIEB) has emerged as an alternative showing better efficacy in randomized studies. In this study, the aim was to test PIEB + PCEA vs CEI + PCEA using an epidural solution containing adrenaline.


| INTRODUC TI ON
Epidural analgesia is considered the gold standard of pain relief during labor, and is effective with few side effects. 1 The epidural treatment solution normally consists of a low-dose local anesthesia in combination with a lipophilic opioid. Adrenaline is sometimes used as an additive to enhance the efficacy by producing vasoconstriction in the epidural space, hence lowering systemic absorption of the local anesthetic 2 and, in a randomized trial, we have shown that the addition of adrenaline lowered the serum fentanyl concentrations during the first two hours. 3 30% of all delivery institutions in Norway use an adrenaline containing solution (unpublished data, personal communication, Haidl et al, 2020).
Until recently, the most common strategy for administering epidural labor analgesia is through a continuous epidural infusion (CEI), often in combination with patient controlled epidural boluses (PCEA). In recent years, the concept of programmed intermittent epidural bolus injections (PIEB) has been introduced. Previous studies in the field have recently been summarized in a Cochrane systematic review and meta-analysis. 4 The studies included in the Cochrane analysis were heterogeneous (eg combined spinal-epidural (CSE) vs only epidural, epidural solution mixture, basal infusions, patient-controlled boluses (PCEA) vs only basal infusions etc). However, the meta-analysis concluded that PIEB reduced breakthrough pain, reduced hourly local anesthetic consumption and increased maternal satisfaction compared to a continuous infusion. None of the studies included in the meta-analysis included adrenaline in the solutions studied.
The aim of our study has been to investigate the effect of programmed intermittent boluses using an epidural solution containing adrenaline. We hypothesized that programmed intermittent boluses would require less hourly epidural solution consumption than continuous infusion, while giving equipotent analgesia.

| ME THODS
This study was approved by the regional ethics committee (REK-Sør- The study design was a randomized controlled trial with two parallel groups. Primiparous and multiparous participants were randomized separately to ensure equal distribution of this factor between the two treatment groups. The group assignment was determined by a computer generated algorithm (Randlist®, Datainf, Thübringen, D), and kept in individual sealed envelopes until patients were included.
A researcher who did not participate in data collection performed randomization in two groups (sizes1:1), block randomization with varying block size of 2, 4 and 6. The randomization including the block sizes were concealed until all data were entered into the database and the database was monitored and locked. Participants were randomized to receive the epidural solution by either intermittent epidural bolus (PIEB) or continuous epidural infusion (CEI).
At request for epidural analgesia, a multi-orifice catheter was inserted medially or paramedially in the middle lumbar segments All participants and study personnel assessing patients were blinded to the intervention. The anesthetist including the patient, and starting the treatment, was not blinded to the intervention.
To prevent that any noise from the pump would interfere with the blinding, two parallel pumps were used; one connected to the patient giving the assigned treatment, and one pump connected to itself in a loop with the opposite treatment. Both pumps were kept in an opaque bag. After the end of the study period, the participants and midwifes assessing the patients were asked if they could guess what treatment group the participant was in (with the option of saying "I don't know"). The answers were examined using the Bang index. 5 The Bang index is scaled in the −1; 1 interval, where 1 indicates complete lack of blinding, 0 indicates perfect blinding and −1 complete opposite guessing.
Participants were assessed for baseline characters, including body metrics and pain on contraction before epidural analgesia initiation. Participants were further assessed for pain on the most recent contractions at the start of treatment and once every hour after start using a numeric rating scale (NRS) ranging from 0 to 10, including worst pain experienced during the delivery phase. If the participant was asleep at the time of pain query, the pain score was assigned "0". If labor continued for more than 8 hours, pain was assessed every second hour. At one hour and at 10 cm cervical dilation, motor block was assessed using a modified Bromage scale (0 = able to lift whole extended leg or standing up, 1 = flexion of the knee, 2 = flexion of the ankle, 3 = no flexion of the knee or ankle). 6 If there was a difference in motor block between the legs, the leg with the most intense block was reported. The day after delivery, a short structured interview was performed by an anesthetist blinded for the study, where the patient's satisfaction with the pain treatment was registered. The patient was asked to rate their overall satisfaction with the treatment, considering treatment effect and possible adverse effects, on a 0-10 numeric rating scale.

| Statistical analysis
The predefined, primary endpoint of the study, was the average consumption of the epidural solution per hour of treatment including the starting bolus and any extra manually delivered boluses.
Secondary outcomes included number of successful PCEA boluses, number of blocked PCEA boluses, pain on contraction during labor and at delivery, maternal satisfactions with the treatment, mode of delivery, motor block at one hour and 10 cm cervix dilation, need for additional anesthetic interventions and time from epidural placement to this intervention. Data were assessed for normality by the Shapiro-Wilks test and visual inspection of histograms and QQplots. If variables were found to not be normally distributed, nonparametric statistical tests were used (ie Mann-Whitney U-test), and data were presented by median and percentiles [IQR], rather than Student's t test and mean with standard deviations (SD). In categorical variables, Fisher's exact test was used when the expected count in a cell was below 5 rather than the Chi-squared test. The primary endpoint was assessed by Student's t test. Time-to-event variables were analyzed using Kaplan-Meier survival analysis and the Log-rank test. Statistical differences in pain score changes over time between groups were tested using a mixed model where the interaction between time and treatment group was used as the outcome test. The data development was not linear throughout the treatment period, and we therefore introduced break points at appropriate time points (ie time = 2 hours and time = 7 hours) after visual inspection of the data. The model consisted of treatment group, the different time variables (from the break points) and the interaction between treatment group and time as dependent fixed effects, and with individual random intercepts and slopes for the effect of time.
A significance level of 5% was used. No correction for multiple comparisons was made, and any significant findings in the secondary outcomes must be interpreted with caution. Data were analyzed using SPSS v. 26 (IBM, Chicago, IL) and STATA/SE v.15 (Stata corp. LLC, College Station, TX).
Sample size was calculated by considering data from previous studies that were similar in dosing to our study 7-9 assuming a mean difference of bupivacaine equivalents of 1 mg/h between groups.
With an α = 0.05 and a power = 0.8 we estimated that a total of 100 participants were required. To account for possible differences in effect when adrenaline was used in the solution, an additional 50 participants were included in the study.

| RE SULTS
Of 194 parturients screened, 151 patients were included in the study. One withdrew consent, leaving 150 participants for final analysis ( Figure 1). Baseline characteristics for both groups were similar (Table 1). There was an equal amount of multiparous participants in both groups.
There were no significant differences in total consumption of the epidural solution per hour of treatment (mean consumption of 9.0 ml/h in the CEI group vs 8.1 ml/h in the PIEB, P = .08) ( Table 2).
However, there was a significant difference in number of PCEA boluses given in total; (mean no. 3.9 (SD 4.1) in the CEI group vs 1.9 (SD 2.0) in the PIEB group P < .001), but no significant differences in number of rejected PCEA boluses between groups (P = .44) ( Table 2) There were no significant differences in mode of delivery, or occurrence of side effects (ie hypotension, nausea or pruritus) and no significant differences in overall satisfaction with the treatment (Table 3). There was a non-significant trend toward a higher modified Bromage-score at 60 minutes in group 0, but not at delivery (Table 3).
There were no serious adverse events during the study period.
We found no evidence of a lack of blinding (Table 4).

| D ISCUSS I ON
In this study, there were no major differences in our primary outcome, that is, total consumption of the epidural solution per hour of treatment. However, there were fewer successful PCEA-boluses in the PIEB-group. This is likely explained by the fact that the PIEBgroup had a lock out period after the hourly default bolus. The PIEBgroup also had a tendency toward more bupivacaine rescue boluses.   7,12,13 It is possible that the use of two separate pumps in our study would have resulted in a more even distribution of PCEA boluses between the groups.
However, the use of two pumps per patients may be ideal in a research context but impractical and costly in an ordinary clinical setting. We have compared the two different methods in an ordinary clinical setting, which would give our results more generalizability.
There is no consensus on the optimal volume and interval of the

TA B L E 3 Secondary outcomes
Motor block may lead to increased use of instrumental delivery, an important clinical outcome. 21 While the meta-analysis by Sng concluded with no differences in risk of instrumental delivery, analogous to our findings (motor block was not assessed), a non-randomized impact study have found a reduction in motor block when transitioning from a CEI protocol to a PIEB + PCEA protocol while simultaneously reducing ropivacaine concentration in the solution. 22 Other impact studies have not found the same clear beneficial effects. 23,24 It is hypothesized that the use of boluses would increase the spread of local anesthesia in the epidural space, thus increasing the chance of good analgesia in the sacral segments. In this study, we assessed the maximum experienced pain at delivery. The maximum pain experienced, and the pain at the end of labor is often remembered. 25 We found individual values within a large range, but with the majority of the pain experienced within the "severe range" (≥7 of 10), with no differences between the groups. To the best of our knowledge, this is the first study examining worst pain at delivery when using PIEB.
Previous studies in the field describes that the patient and the person assessing the outcomes were blinded to the intervention. In the initial preparations before this study, we found that the pump we intended to use made a distinct, albeit quiet sound as the bolus was delivered. This sound made it possible in many cases for the midwife caring for the parturient to unveil the treatment the patient received (unpublished pilots). To counteract this, we implemented the use of two pumps as described in the method section (ie one pump to make "placebo noice" There are several limitations to our study. The explicit threshold of pain that mandated contact to the treating anesthetist was not defined, but left to the attending midwife and/or participant to decide. It is possible that an explicit threshold could have resulted in a higher proportion of contacts. However, as a large proportion of participants in both groups were satisfied with the treatment, a mandatory threshold could also have resulted in overtreatment and higher incidence of adverse effects such as extended motor block.
Furthermore, it is possible that a larger sample size would have resulted in a statistically significant difference in the primary outcome.
Nevertheless, our study is among the largest in the field, and possible statistically significant findings between the groups with a larger sample size might not be clinically relevant. Our approach to assessing satisfaction with treatment was similar to previous studies.
However, this field is complex, and satisfaction is often influenced by other factors (such as labor and neonatal outcomes), and these results must therefore be carefully interpreted.
In conclusion, using an epidural solution containing adrenaline, we found no differences in hourly solution consumption, pain scores or maternal satisfaction when using PIEB compared to CEI.