Basal Infusion versus Automated Boluses and a Delayed Start Timer for “Continuous” Sciatic Nerve Blocks after Ambulatory Foot and Ankle Surgery: A Randomized Clinical Trial

This parallel group study adhered to Good Clinical Practice quality standards and ethical guidelines defined by the Declaration of Helsinki. Study protocol approval, as well as data and safety oversight, were conducted by the University of California San Diego Institutional Review Board (approval No. 200247). Written informed consent was obtained from all participants. The trial was prospectively registered at clinicaltrials.gov (NCT04458467, Principal Investigator: Brian Ilfeld, date of registration: July 7, 2020) before initiation of enrollment, and the trial protocol is available at http://clinicaltrials.gov. The trial was conducted in accordance with the original protocol.

Enrollment was offered to adults (18 yr and older) who were scheduled to undergo unilateral ambulatory foot and/or ankle surgery associated with pain that the surgeon expected to persist beyond the duration of a single-injection nerve block and require treatment with oral opioid analgesics. Patients eligible for the study were called the night before surgery by an investigator to offer enrollment. If not available by telephone, patients were offered enrollment in person before surgery only if there was sufficient time to fully discuss the study, answer all questions, read the consent form thoroughly, and give patients time to make an informed decision. All participants were enrolled at one of two hospitals or an ambulatory surgery center, all of which are part of a single academic institution in Southern California. Patients were excluded if they had clinically apparent neuropathy of the ipsilateral sciatic nerve and its branches and/or innervating muscles; current daily opioid use within the previous 4 weeks; body mass index greater than 35 kg/m²; surgery outside of ipsilateral sciatic and saphenous nerve distributions (e.g., iliac crest bone graft); pregnancy; or incarceration.

After applying standard monitors, providing oxygen by face mask, and positioning the patient in the prone position, intravenous midazolam and fentanyl were titrated for patient comfort while ensuring that the patient remained responsive to verbal cues. A 13- to 6-MHz, 38-mm linear array ultrasound transducer (Edge II, SonoSite, USA) was used to visualize the sciatic nerve proximal to the bifurcation with a short-axis view. The catheter site was sterilely prepped and draped, and the needle entry site and underlying muscle were anesthetized with 1% lidocaine (3 ml).

A 17-guage Tuohy needle (FlexTip Plus, Teleflex Medical, USA) was inserted on the posterolateral aspect of the leg, from lateral to medial, using an in-plane, short-axis, ultrasound-guided technique, and 1 to 2 ml of normal saline was injected to open the space adjacent to the nerve. A 19-gauge flexible, single-orifice perineural catheter was inserted under ultrasound guidance 2 to 3 cm beyond the needle tip, and the needle was removed over the catheter. Ropivacaine (0.5%) with 5 to 10 µg/ml epinephrine (20 ml) was injected in divided doses through the catheter under ultrasound visualization. The catheter was then secured with clear, occlusive dressings. To facilitate tourniquet placement in the operating room, the catheter was taped up the lateral thigh.

A successful block was defined as sensory (decreased temperature discrimination to ice) and motor (decreased dorsi- and plantar-flexion strength) in the tibial and peroneal nerve distributions within 15 min of injection (unless the foot was not accessible due to a splint) as compared to the contralateral foot. If the planned surgical procedure was anticipated to produce pain in the saphenous nerve distribution, a single-injection saphenous nerve block was performed with ropivacaine (0.5%) with 5 to 10 µg/ml epinephrine (10 to 20 ml).

After confirming a successful block in the sciatic distribution, participants were randomized using a computer-generated list (prepared by the University of California San Diego Investigational Drug Service) and provided to the investigators in opaque, sealed, sequentially numbered security envelopes to one of two treatment groups (1:1 ratio) in blocks of four: (1) automated bolus with a 5-h delayed start or (2) continuous infusion with an immediate start. These envelopes were opened by a member of the care team (e.g., block nurse or regional anesthesiology fellow) who was not a coinvestigator and not involved in study data collection. Thus, both the participants and investigators were blinded to treatment group.

Patients received either intravenous propofol sedation or general anesthesia consisting of inhaled volatile anesthetic with or without nitrous oxide in oxygen. Opioids were administered at the discretion of the intraoperative anesthesia team.

After completion of the surgical procedure, an electronic infusion pump (Nimbus II PainPRO, InfuTronix, USA) with a 500-ml reservoir of 0.2% ropivacaine was attached to the perineural catheter. The pump was programmed based on the participant’s randomization group by an investigator who was not involved in data collection or analysis (table 1). One group received continuous infusion with no delay: the ropivacaine infusion was started immediately on attachment of the pump (6 ml/h basal infusion, 4-ml patient-controlled bolus, 30-min lockout). The other group received automated boluses with a 5-h delay: the infusion pump entered a pause mode after attachment of the pump to delay administration of local anesthetic for 5 h. The pause could be overridden by the participant if the initial peripheral nerve block resolved sooner than anticipated. After the 5-h delay, or when the patient overrode the delay, local anesthetic administration was initiated (automated 8-ml boluses every 2 h, 4-ml patient-controlled bolus with a 30-min lockout).

Participants were discharged with a prescription for the synthetic opioid oxycodone (5-mg tablets) for supplementary analgesia and contacted by telephone daily for 6 days after surgery to collect study outcome measures. Upon ropivacaine reservoir exhaustion, participants or their caretakers were instructed by phone on removing the perineural catheter.

We hypothesized that, compared with a continuous basal infusion initiated before discharge, perineural local anesthetic administered with automated boluses at a lower dose and a 5-h delay after discharge would (1) provide at least noninferior analgesia during the period that both techniques are functioning (primary outcome: average pain score the day after surgery) and (2) result in a longer duration of administration (primary outcome: hours from initiation until reservoir exhaustion) for popliteal–sciatic catheters after ambulatory foot and ankle surgery. We utilized dual primary endpoints with a serial testing strategy, such that superiority for overall duration of local anesthetic infusion would be tested only if the average pain scores on the first postoperative day were found to be at least noninferior.

The participants and investigators collecting the data were masked to treatment group assignment. The first four items of the Brief Pain Inventory were collected daily²³ : worst, average, least, and current surgical pain measured using a numeric rating scale. Additional outcomes included daily opioid consumption, number of sleep disturbances due to pain, degree of sensory block (measured on a 0 to 10 scale, where 0 indicates no deficits, and 10 indicates completely insensate), and satisfaction with postoperative analgesia (where 0 indicates very dissatisfied, and 10 indicates completely satisfied). In addition, for medical purposes, we collected the following information: local anesthetic leakage (binary) and the cause for discontinuation of the infusion (e.g., completion of infusion, accidental dislodgement).

The study was powered for two primary endpoints: (1) the average numeric rating scale queried on postoperative day 1 and (2) the duration of treatment from when the infusion pump was initiated until local anesthetic reservoir exhaustion. The dual hypotheses were tested with a serial testing strategy, such that hypothesis 2 would not be formally tested unless the conclusion of hypothesis 1 was at least “noninferior.” Following the approach described by Althunian et al.,²⁴  noninferiority was assessed by comparing the lower limit of the two-sided 95% CI for the difference (CB minus AB) on the numeric rating scale (range, 0 to 10) to a prespecified noninferiority margin of 1.7 numeric rating scale units (fig. 1, A and B). This provided evidence that the analgesia provided by the novel automated boluses was no worse than 1.7 numeric rating scale units compared to a basal infusion. The same two-sided 95% CI can be compared to 0.0 numeric rating scale to conclude superiority of automated boluses at the 5% significance level.

We assessed the balance of randomized groups on baseline and procedural characteristics using absolute standardized difference, defined as the absolute difference in means, mean ranks, or proportions divided by the pooled SD. Baseline variables with absolute standardized difference greater than 0.47 (i.e., 1.96 ×√(1/n1 + 1/n2)) were noted and included in a linear regression model to obtain an estimate of the treatment group differences adjusted for the imbalanced covariate(s).^25,26  If residuals from the linear regression indicated violations of key assumptions (i.e., homoscedasticity or Gaussian distribution), data transformations and/or alternative generalized linear models were applied, as appropriate.

Primary outcomes were analyzed using an ordinal regression model for continuous scales to compare “average” pain scores on the first postoperative day (hypothesis 1) and a linear regression model for infusion duration (hypothesis 2).^27,28  Secondary outcomes were analyzed by Wilcoxon–Mann–Whitney test or linear models as appropriate with covariates for any imbalanced covariates. No multiplicity adjustments were applied for these analyses. All analyses were conducted using R version 4.1.1. Treatment group allocation was only revealed to investigators after statistical analysis.

Separate power analyses were conducted for both hypotheses, with hypothesis 2 contingent on hypothesis 1.

Power was simulated based on the distribution of the numeric rating scale observed in previously published studies of pain scores after foot and ankle surgery with a continuous popliteal–sciatic infusion.²⁹  Specifically, numeric rating scale scores were simulated from a discrete distribution that generated an expected interquartile range from 1 to 4, and a median of 3 numeric rating scale units. One thousand trials were simulated in which two groups, n = 35 per group, were assumed to follow the same discrete distribution (fig. 1C). Each trial was submitted to a Wilcoxon–Mann–Whitney test, and 95% CIs were derived.²⁷  Of the 1,000 trials, 792 (79.2%) correctly resulted in a conclusion of noninferiority, suggesting that the probability that the trial correctly concludes noninferiority is about 80% when the groups follow exactly equivalent distributions.

If the test for hypothesis 1 concluded noninferiority or superiority, the difference in overall duration of administration would be tested using the Wilcoxon–Mann–Whitney test. Power was approximated by an independent sample t test calculation. Assuming a SD of 37 h (corresponding to an interquartile range of 50 to 100 h), a sample size of n = 35 would provide 80% power to detect a mean group difference of 25 h with a two-sided α of 5%.

A total of 71 participants were enrolled beginning July 15, 2020, and ending March 10, 2021 (fig. 2). Enrollment was ceased when the target sample size had been obtained, and data collection was finished on March 16, 2021. All but one participant had a successful sciatic nerve block. The remaining 70 participants were randomized and equally divided between the treatment groups, and all randomized participants were included in primary and secondary analyses per the intention-to-treat principle. All participants remained blinded to treatment group for the duration of follow-up. Outcome assessors remained blinded to treatment group allocation until after the statistical analysis was complete. Participants were summarized and compared on potentially confounding baseline and procedural characteristics (table 2). All factors were balanced between the two groups with the exception of body mass index, which was potentially imbalanced between groups with a higher body mass index in the continuous infusion group (absolute standardized difference, 0.693). Therefore, the primary analysis was adjusted for body mass index. A unit increase in body mass index had little effect on the odds of worse pain (odds ratio, 1.0; 95% CI, 0.9 to 1.1). However, it did have an effect on infusion duration such that larger body mass index was associated with shorter infusion duration (–1.1 h per body mass index increase; 95% CI, –2.0 to –0.3; P = 0.008). In a model controlling for the independent effects of height and weight, the continuous infusion group had a shorter mean infusion duration (–46.1 h; 95% CI, –54.6 to –37.5; P < 0.001), higher weight was associated with shorter infusion duration (–0.4 h/kg; 95% CI, –0.7 to –0.1; P = 0.007), and taller heights were associated with longer infusion duration (54.0 h/m; 95% CI, 3.9 to 104.2; P = 0.035). The adjusted analyses of all secondary outcomes were consistent with the unadjusted analyses. There were no missing data for any baseline variables or primary outcomes. Four subjects in the automated bolus group could not be reached by telephone for one of the nonprimary outcome telephone calls versus one subject in the continuous infusion group. Thus, secondary outcome data were incomplete for these days. No data were lost or excluded.

The day after surgery, participants with automated boluses had a median [interquartile range] pain score of 0.0 [0.0 to 3.0] versus 3.0 [1.8 to 4.8] for the continuous infusion group (unadjusted 95% CI, –2.0 to 0.0; P = 0.007; fig. 3A). The odds of worse average pain on day 1 with continuous basal infusion, adjusting for body mass index, were 3.1 (95% CI, 1.2 to 7.8; P = 0.033). Local anesthetic reservoir exhaustion in patients with automated boluses occurred at a median [interquartile range] of 119 h [109 to 125] versus 74 h [57 to 80] for the continuous infusion group (difference, 47; 95% CI, 42 to 53; P < 0.001; fig. 3, B and C). Adjusting for body mass index, the difference was also 47 h (95% CI, 38 to 55; P < 0.001).

Daily average and worst pain scores were lower for the bolus versus basal groups at nearly all time points through postoperative day 5 (fig. 4, A and B). Automated boluses reduced the median cumulative opioid consumption by 83%, from 9.0 mg [3.5 to 13.0] to 1.5 mg [0.5 to 5.0] (P < 0.001; fig. 4C) and reduced cumulative sleep disturbances by 75%, from 4.0 [2.0 to 10.0] to 1.0 [0.0 to 3.0] (P < 0.001; table 3). Participants receiving automated boluses experienced more numbness at all time points (table 3), although local anesthetic leakage did not differ between treatments (table 3), and satisfaction with analgesia differed only on postoperative days 1 and 4 (table 3).

Four participants in the automated bolus group and six in the continuous infusion group accidentally dislodged their perineural catheters before infusion completion. All participants whose catheters accidentally dislodged were offered catheter replacement; however, only one participant elected to have the catheter replaced.

One participant in the automated bolus group required revision surgery due to a surgical complication on postoperative day 4 from the initial surgery. This participant’s catheter had accidentally dislodged on the third postoperative day. The anesthesia provider caring for this participant on the day of the second surgery placed a new popliteal–sciatic perineural catheter with our institution’s conventional settings (continuous infusion at 6 ml/h with 4-ml patient-controlled bolus and 30-min lockout) without knowing the participant was part of a randomized trial. After surgery, the decision was made to continue with these settings while the participant remained an inpatient to provide our institution’s normal standard of care. The participant was retained within his randomized group per the intent-to-treat principle. No falls, catheter-related infections, nerve injuries, or other treatment-emergent complications were observed in either group.

The main findings of this study are that, compared with a continuous basal infusion initiated before discharge via a popliteal–sciatic catheter, perineural local anesthetic administered with automated bolus doses at a lower volume/dose and a 5-h delay after discharge resulted in (1) superior analgesia during the period that both modalities were functioning and (2) a longer duration of anesthetic administration. Significant benefits were identified for pain scores, opioid consumption, and sleep quality for nearly all of the first 5 postoperative days, when both techniques administered local anesthetic, as well as after the reservoir exhaustion of the basal infusion participants. Our study is unique relative to nearly all previous investigations comparing automated boluses to a basal infusion in that (1) we included an integrated start-delay timer; (2) we did not define a maximum treatment period but rather removed catheters only after reservoir exhaustion; (3) we collected data 6 days after surgery, which is two to three times longer than previous investigations; and (4) we decreased mandatory average hourly local anesthetic delivery, while most others compared equivalent volumes/doses.^{6–20,30–32} 

For popliteal–sciatic catheters after foot and ankle surgery, previous studies suggest that providing a basal infusion maximizes analgesia and other benefits compared with exclusively patient-controlled boluses, presumably due to the observed decrease in anesthetic volume/dose administered when patients must trigger the boluses themselves.^2,33  However, based on findings for epidural catheters,^34–36  it was theorized that increasing the volume of local anesthetic introduced at a single time point—a bolus—might improve perineural spread compared with an equivalent volume/dose provided as a basal infusion providing superior analgesia.²¹  Indeed, the first investigation replacing patient-controlled with automated boluses for sciatic catheters found an analgesic benefit over an equivalent volume/dose administered with a basal infusion.⁶  However, by adding patient-controlled bolus doses to these two regimens, the difference in pain scores disappeared.⁷ 

Our study was somewhat unique in that we intentionally decreased the average volume/dose of local anesthetic administered with automated boluses (8 ml every 2 h) by 33% compared with the basal infusion (12 ml every 2 h). We had hoped to provide noninferior analgesia and conserve local anesthetic, thereby prolonging overall administration and, consequently, analgesia. However, we did not anticipate our finding of superior analgesic effects with the automated boluses while both treatment groups were receiving local anesthetic. Our findings are in contrast to multiple other investigations involving perineural sciatic catheters.^7–9  Taboada et al.^6,7  identified no analgesic or opioid-sparing benefits comparing a 5-ml automated hourly bolus with a 5-ml/h basal infusion, both with available patient-controlled boluses. The difference may be due to our larger bolus volume—8 versus 5 ml—even though it was administered only every 2—versus 1—h, thus providing a lower average hourly volume/dose.

However, two additional randomized trials that utilized larger (9.8 to 10 ml) automated boluses every 2 h via popliteal–sciatic catheters—a larger volume than our study—identified no analgesic benefits when compared with a basal infusion of the same average hourly volume/dose (5 ml/h), both with available patient-controlled boluses.^8,9  While their results may appear to contradict our findings, both studies did identify other differences between treatments. Short et al.⁸  reported increased motor block in the automated bolus group, suggesting that bolus doses even in an equivalent volume/dose with a basal infusion have different effects. While the sensory block did not reach statistical significance after correction for multiple comparisons, the use of a 3-point (vs. our 10-point) scale may have provided insufficient dispersion given their sample size.

Similarly, while Breebaart et al.⁹  failed to identify analgesic benefits, their participants receiving automated boluses consumed a smaller volume of local anesthetic, again suggesting a difference between treatments. In contrast, patients receiving automated boluses experienced a 64% incidence of “numbness”—similar to our findings—identical to their basal group. Failing to detect a difference in sensory block between their treatments may be due to their use of binary response (present/absent), while our 0 to 10 sensory block scale may have provided more sensitivity to differences between treatments.

Our analgesic-related results may differ from those of Short et al.⁸  and Breebaart et al.⁹  because of seemingly insignificant protocol variations, such as differing catheter insertion approaches (in- vs. out-of-plane needle advancement), catheter tip locations (subparaneural vs. supraparaneural), or local anesthetic (0.2% ropivacaine vs. 0.125% levobupivacaine). Additionally, participants in the current study had the potential to prolong their local anesthetic infusion—and thus analgesia—by reducing demand bolus use and may have tolerated higher pain scores to prolong the infusion duration and its analgesic effects.

Recent findings of analgesic improvement with automated boluses versus a basal infusion for paravertebral perineural catheters add credence to our results.³⁰  This topic will require additional research to answer the multitude of questions raised by the current study.

Although our study protocol ensured less mandatory local anesthetic delivery in the automated bolus group compared with participants with a basal infusion (8 ml vs. 12 ml every 2 h), we provided both treatment groups with the ability to self-administer additional boluses every 30 min. We had anticipated that participants with the automated boluses would self-administer more boluses to compensate for their scheduled 4-ml local anesthetic deficit every 2 h. However, this did not occur, with automated boluses decreasing local anesthetic consumption and prolonging administration 61% from approximately 3 to 5 days (only 5 h of which was due to the start-delay timer). Previous studies of basal versus bolus dosing of popliteal–sciatic catheters concluded both treatments at the same time regardless of residual reservoir volume (usually after 24 to 48 h).^6–9  This permitted the comparison of local anesthetic consumption and relative benefits during the initial 1 to 2 postoperative days but not subsequent consumption or analgesic effects. For inpatients who have access to reservoir refills, the anesthetic-sparing appears to have little importance given the extraordinarily rare occurrence of local anesthetic toxicity during perineural administration.³⁷  However, for ambulatory patients with a fixed reservoir volume, decreasing local anesthetic consumption allows for an extended administration duration and therefore prolonged analgesia delivery.

Our findings that patients who continued to receive automated bolus doses experienced superior analgesia compared to participants who had previously exhausted their anesthetic reservoir with a basal infusion is unsurprising, yet this has not been previously documented with a prolonged 6-day period of evaluation as in our study. Given our findings of improved analgesia while both administration modalities were delivering local anesthetic, our results are applicable to inpatients as well.

The 5-h start delay was chosen as a conservative estimate to ensure that the first bolus dose occurred before the analgesia provided by the initial preoperative single injection 0.5% ropivacaine block would resolve.³⁸  Our goal was to prolong the duration of the 0.2% ropivacaine infusion as long as possible while ensuring that the participants did not experience pain with the initial block resolution.³⁹  To minimize this likelihood, participants could override the delay if they began experiencing pain before the automatic activation of the first bolus. Start-delay timers are a relatively novel function available on few electronic infusion pumps, and the results of the current study suggest that this feature may be used to prolong the total duration of an ambulatory perineural ropivacaine infusion and the analgesia provided.

An unanticipated finding was that higher weight was associated with shorter infusion duration, while taller height was associated with longer infusion duration. These results suggest a complex relationship between body morphology and local anesthetic requirements that is previously unreported in the literature. Specifically, these data indicate that obesity increases local anesthetic requirements. As such, patients with a higher body mass index may benefit from automated bolus dosing to an even greater degree than their lower–body-mass-index counterparts. Further research is indicated to elucidate this complex relationship between body habitus and local anesthetic requirements.

Although the participants of this investigation were masked to treatment group assignment, the audible pump activation occurring either continuously or at 2-h increments may have allowed some participants to deduce their randomization group. Further, although the infusion settings were not obviously displayed on the pump, participants did have access to the pump controls and could potentially have investigated their treatment group. However, it is improbable that the majority of patients had a preconceived preference for one technique over the other. In addition, the continuous activation of the pump could have interfered with sleep more in the continuous groups compared to the automated bolus group, who would only potentially be awoken by the pump activation every other hour. Regarding study design, two interventions were undertaken to increase the duration of the ropivacaine infusion in the experimental group: a start-delay timer and automated boluses with lower basal dosing. Given the 5-h start delay for the treatment group and median of 45-h overall increase in administration duration for these participants, we assume that the duration difference between treatments would have been 40 h if both groups had been provided with the 5-h start delay. Last, while we used an observer-masked design to minimize outcome assessor bias, an unmasked healthcare provider was required to program the infusion pump, and therefore we could not implement a triple-masked study. However, the unmasked healthcare provider had limited interaction with the participants, and the treatment groups were not identified to the statistician until the analysis was completed.

The results from this study suggest a possible paradigm shift for postoperative perineural catheters, 75 yr after Ansbro⁴⁰  described the first continuous peripheral nerve block: replacing a basal infusion with delayed-start large automated boluses to both improve analgesia potency and prolong analgesia delivery. Evidence may be found in one previously published investigation that—like our study and unlike nearly all previous basal versus automated bolus trials—administered less mandatory local anesthetic for the automated boluses relative to the basal infusion.¹³  Rao Kadam et al.¹³  randomized 20 subjects with bilateral transversus abdominis plane catheters to receive ropivacaine (0.2%) as either provider-administered boluses of 20 ml every 8 h or a pump-administered basal infusion of 8 ml/h for each catheter. Although the boluses administered only 31% as much volume/dose as the basal infusion (2.5 vs. 8.0 ml/h), no differences in pain scores or supplemental opioid consumption were identified during the 48-h study period. This study was possibly underpowered but combined with our own findings suggests that additional research intentionally lowering the average hourly volume/dose for automated boluses may better match the durations of analgesic delivery and surgical pain.¹³ 

To summarize, for popliteal–sciatic catheters after ambulatory foot and ankle surgery, replacing a continuous infusion initiated before discharge with automated boluses and a start-delay timer resulted in both better analgesia and a longer duration of the infusion. Reservoir exhaustion occurred on approximately the third postoperative day with conventional infusion settings and the fifth postoperative day when automated boluses and a start-delay timer were used. Additional research is required to determine whether these results may be replicated for catheters in other anatomic locations and to optimize the various associated factors such as catheter insertion protocol, automated and patient-controlled bolus volumes and frequency, and local anesthetic type and concentration.

InfuTronix (Natick, Massachusetts) provided the electronic pumps used in this study. The company was given the opportunity to review the protocol and suggested minor revisions. The investigators retained full control of the investigation, including study design, protocol implementation, data collection, data analysis, results interpretation, and manuscript preparation.

Drs. Finneran, Said, Swisher, Gabriel, and Ilfeld and the University of California have received funding and product for other investigations from infusion pump manufacturer InfuTronix; the cryoneurolysis device manufacturer Epimed (Farmers Branch, Texas); and the peripheral nerve stimulation device manufacturer SPR Therapeutics (Cleveland, Ohio). Dr. Gabriel has worked as a paid consultant for Avanos (Alpharetta, California). Dr. Donohue has received support for other research projects from the National Institutes of Health (Bethesda, Maryland), Biogen/Eisai (Cambridge, Massachusetts), and Roche (Basel, Switzerland). The other authors declare no competing interests.

Full protocol available at: jfinneran@health.ucsd.edu. Raw data available at: jfinneran@health.ucsd.edu.