Regional anaesthesia elastomeric pump performance after
a single use and subsequent refill: a laboratory study


Dr Michael Fredrickson


Ambulatory local anaesthetic delivery systems are often limited by a short effective duration of infusion. Prolonging nerve blockade by substituting a new pump as recommended by the manufacturers, represents a substantial consumable item cost ($US300–500). We therefore evaluated the flow delivery performance of 31 single model elastomeric devices (all with a 2 ml.h−1 background and 5 ml every hour bolus capability) that had been filled, used in clinical practice and then refilled in the laboratory. For the second infusion, there was a pattern of over-infusion (< 10 ml.h−1) in the first hour; however, all pumps depreciated to < 150% of predicted by the second hour. The subsequent performance of all pumps was not only within safe limits, but also predominantly within the range (background infusion ± 15%, bolus +10/−20%) specified by the manufacturer for primary infusion. We conclude that this elastomeric regional anaesthesia pump design performs satisfactorily after having been refilled following a single previous use.

The use of ambulatory portable elastomeric infusion devices in the management of continuous regional analgesic techniques has grown in popularity over the last few years. The lightweight, portable design of these pumps has allowed a number of surgical procedures that previously required inpatient care to be safely undertaken in an ambulatory care setting [1–3]. The cornerstone of these techniques is the highly effective analgesia provided by continuous perineural infusions. This not only controls postoperative pain and minimises the requirement for opioids and their unwanted side effects, but can also accelerate the attainment of early functional rehabilitation targets (through improved range of joint motion), and therefore readiness for hospital discharge [4, 5].

There are now several commercially available elastomeric pumps with a variety of flow rates, many with the facility for additional patient controlled bolus administration. The accuracy and variability of local anaesthetic delivery by these devices has previously been reported by independent clinical investigators [6, 7]. However, one of the limitations of these devices is the manufacturer’s recommendation that they are not refilled for the purpose of prolonging the infusion.

It is recognised that a small but significant group of patients experience moderate to severe postoperative pain lasting in excess of 48 h after certain surgical procedures [1, 8]. Due to inter-individual variability in analgesic requirements postoperatively, elastomeric pumps with a bolus facility may be exhausted within the first 48 h following surgery. The most effective method for managing pain control in these patients would be to continue the local anaesthetic infusion for another 48–72 h. This would require a new pump at a significant increase in the cost of consumable items (US$300–500). An alternative option is to refill and continue to use the existing elastomeric pump, which represents a potentially simple, practical and inexpensive method of providing on-going high quality analgesia in a small number of cases.

The aim of this laboratory study was to independently evaluate the accuracy of elastomeric pump local anaesthetic delivery after refilling of pumps that had previously been used in clinical practice.


Pump selection

We chose to evaluate the On-Q Pain Buster® with On Demand® elastomeric pump (I-Flow corporation, Lake Forest, CA, USA) because it is the most commonly used device by both authors in clinical practice. The On-Q Pain Buster® has a reservoir volume of 270 ml and delivers a constant background infusion of 2 ml.h−1 and a patient controlled bolus facility of an additional 5 ml every hour. The infusion and bolus characteristics of this pump are such that it is suitable for most anatomical locations; it has been used successfully by both authors for continuous interscalene [1], infraclavicular, femoral and sciatic perineural infusions. Theoretically, the duration of local anaesthetic administration with this pump ranges from 38.5 h (7 ml.h−1) up to a maximum of 135 h (2 ml.h−1). In practice, the duration of clinically useful infusion is likely to be shorter than this, as the flow rate will depreciate significantly at the end of the infusion with some infusate remaining in the pump (product information sheet).

We evaluated the performance after refilling of the On-Q Pain Buster® with On Demand® that had been used for continuous interscalene block after elective ambulatory shoulder surgery. Pumps were collected from the surgeon’s clinic where they were returned by patients at the day seven postoperative consultation.

Study apparatus and flow measurement

Flow rate was measured with a technique similar to that used by other investigators [6]. Specifically, the weight of the device was measured hourly using an electronic balance (Scout-Pro; Ohaus, Pine Brook, NJ, USA) connected to a laptop computer via an RS-232 serial port connection. The serial weights were recorded in a spreadsheet application (microsoft excel; Microsoft Corporation, Redmond, WA, USA) using a customised data-logging program (Ansutek, Auckland, New Zealand). This method of flow rate measurement assumes a constant fluid density such that 1 g equals 1 ml of pump fluid. Therefore, the depreciating pump weight (g) equated to the volume of saline (ml) delivered per hour. It is recognised that the viscosity of the infusate may affect the infusion flow rate. The On-Q Pain Buster® with On Demand® is calibrated with 0.9% saline at room temperature (product information sheet). In addition, identical flow rates of 0.9% saline and ropivacaine have been demonstrated by independent investigators evaluating primary elastomeric pump performance [6]. On the basis of this information, we selected 0.9% saline as the infusate in this study. The 2 ml.h−1 flow restrictor supplied with the pump is also calibrated and designed for use at room temperature. We therefore maintained the pump and flow restrictor at room temperature (approximately 10–20 °C) during the period of the study. Room temperature was not regulated in order to replicate pump operating conditions in the intended clinical setting.

The date of pump filling was obtained from the filling record and the residual pump weight was obtained using the electronic balance. Each pump was then refilled with 0.9% saline at room temperature. Refilled elastomeric pumps were then placed on the electronic balance and the infusion commenced – the tubing from the pump, the filter and the flow restrictor were connected to a perineural catheter (B. Braun, Melsungen, Germany) – as would happen in clinical practice, however, these items were not placed on the balance.

The initial pump weight after refilling was recorded followed by serial hourly pump weights, which were automatically recorded until each pump was empty. In addition, the accuracy of the bolus facility was examined by intermittently pressing the hand held bolus device, at no specified time point, when the infusion rate of each pump was being tested in order to try and replicate clinical use. The time points at which this occurred were also recorded. The bolus was calculated by subtracting the background rate of infusion from the preceding hour from the time periods over which the bolus was given. A minimum of 1 h separated each bolus activation.

As we measured pump weight directly and not the volume of infusate collected, evaporative losses need not be considered. The actual volume of pump fluid was calculated by subtracting weight of the pump when empty (60 g) from the total weight of pump and fluid as measured by the electronic balance.

Infusion duration was considered to be finished at a flow rate of < 1.0 ml.h−1 (< 50% of specified continuous rate). The repeatability of the electronic balance is quoted at 0.1 g and the linearity is given as ± 0.1 g in the manufacturer’s data sheet.

Statistical analysis

Descriptive statistics were used. Data is presented as median (interquartile range) [range] or mean/SD [range] where appropriate.


A total of 32 elastomeric pumps were refilled for the purposes of this study. One pump was not included in the data analysis as fluid was noted to be leaking from the filling port; therefore data on pump output was collected for 31 pumps. All pumps had been initially filled to 270 ml. The lag between initial filling and refilling, the residual volume, the refilled volume and the duration of the (second) infusion are shown in Table 1. The time lag between initial pump use and pump refilling was due to a number of factors: the number of patients receiving continuous interscalene block, patients returning pumps to clinic at different times and further delay due to being able to study only one device at a time. Residual volume was variable due to some pumps being clamped at a specified postoperative time-point before the pump had depleted. This typically occurred following minor shoulder surgery such as arthroscopic stabilisation.

Table 1.   Pump characteristics at refill (n = 31). Values are median (quartiles) [range].
Time since originally filled; days55 (23–63) [2–151]
Residual volume; ml16 (3–87) [0–171]
Refilled volume; ml255 (234–260) [213–281]
Duration of (second) infusion; h105 (82–114) [60–131]

The flow rate time course observed following the refill and the performance of each individual pump was close to that specified by the manufacturer for primary infusion (Figs 1 and 2). There was a pattern of over-infusion in the first few hours immediately after pump refilling. In 24 pumps (77%) the initial rate did not exceed 150% (≥ 3 ml.h−1) of predicted. Of the remaining seven pumps (33%) the rate of infusion ranged from 3.5 ml.h−1 up to 10 ml.h−1 in the first hour after refilling, however, the rate of infusion depreciated to 3.1 ml.h−1 or less in all seven pumps over the following hour. The finding of relatively high infusion rates was not found at any other point in the study in any pump.

Figure 1.

 Flow rate time course for a typical pump.

Figure 2.

 Background flow rate (n = 31). Values are median (horizontal line), quartiles (vertical bars) and range (whiskers). Flow during the first hour and hourly flow rates < 70% of predicted at the conclusion of the infusion were excluded from this graph.

The hourly infusion data was collated for all pumps (Table 2). Refilled pumps operated within the limits specified by the manufacturer (2 ml.h−1 ± 15%) for 69.2% of the infusion duration. Increased and decreased infusion rates occurred at the beginning and end of the infusion period respectively in all pumps, consistent with typical flow curves for this device (product information sheet).

Table 2.   Flow rate relative to the calibrated flow rate (n = 31). Data represents the hourly flow rate for all pumps. Hourly flow rates < 70% of predicted at the conclusion of the infusion were excluded from this analysis.
Flow rate (relative to 2 ml.h−1)Proportion of the infusion period
< 85%26.7%
> 115%4.1%

The bolus function was noted to work effectively and without incident on each pump studied (Table 3). The mean bolus dose delivered across the study was 4.5 ml, which is consistent with the accuracy, stated by the manufacturer, for bolus delivery of 5 ml.h−1 +10/−20%. Larger boluses were given earlier in the infusion study period, the smaller boluses occurred when the pump was nearing the end of its infusion period when it was noted that the bolus device did not refill effectively at these time points.

Table 3.   Pump bolus function (n = 31). Values are median (quartiles) or mean/SD [range].
Total bolus activations for all pumps; n275
Total bolus activations for each pump; n   9 (4–13) [1–19]
Bolus volume; ml  4.5/0.5 [0.7–7.1]


Our study findings are the first, to our knowledge, to demonstrate the infusion characteristics of an elastomeric device that has been filled for a second time. Pumps refilled up to 151 days after initial clinical use appear to maintain the ability to infuse local anaesthetic solutions within safe limits over the duration of subsequent infusion and in keeping with the manufacturers initial use data. Consistent with other investigators findings [6, 7], pump accuracy was variable over time.

The ability of any portable elastomeric device to provide effective analgesia devoid of side effects is fundamental to the success of such analgesic techniques, particularly when an increasing number of these procedures are being performed as ambulatory day-stay procedures. The pump that we chose to study has a relatively low background infusion rate and has been shown to be highly effective for continuous interscalene block for postoperative analgesia following shoulder surgery [1]. Accurate catheter positioning under ultrasound guidance such that the catheter is targeting the most appropriate elements of the brachial plexus (C5–C6 roots/superior trunk) may explain the effectiveness of this relatively low infusion rate [9]. The experience of both authors is that this low background infusion device is also effective for infraclavicular, femoral and sciatic perineural infusions when similarly precise catheter position is achieved.

Deviations from constant flow rates did occur. The majority of pumps studied showed an increased background infusion rate initially which was not in excess of 3 ml.h−1; however, seven pumps did exceed this in the first hour. There was no apparent association between initially high flow rates and length of time since initial use, the residual elastomeric pump volume or refill volume. Nevertheless, the relative under-filling of the pump (median = 255 ml compared with the recommended 270 ml), a phenomenon that is well recognised as being associated with a faster flow rate (product data sheet) may have been a factor. The lower refill volume was to minimise any possible ‘stretching’ of the elastomeric bladder, which might theoretically occur with a second fill. It is notable that the rapid initial hourly rate (maximum 10 ml.h−1) decreased rapidly thereafter. It is standard practice to use a low concentration of local anaesthetic (e.g. ropivacaine 0.2%) for this type of infusion to ensure analgesia while also attempting to minimise motor block. This dose over a 1-hour period is well within the safety margin for this local anaesthetic agent, therefore this rapid initial hour of infusion would be unlikely to represent a significant clinical problem. It is tempting to speculate that this might even be beneficial in patients who have exhausted their supply of local anaesthetic, are experiencing acute pain and therefore require a ‘top-up’ to re-establish effective nerve blockade. However, because we only studied 31 pumps, a more cautious approach is more appropriate and we therefore recommend that patient administration of pump local anaesthetic be avoided during this first hour.

One potential source of inaccuracy in our data may have arisen from the combined measurement of background infusion rate and bolus level, after the bolus device had been pressed. The preceding hourly background rate was used as an estimate and subtracted from the hourly total to give the bolus dose, thus assuming the same hourly infusion rate as the previous hour. Furthermore, the time point at which the bolus was administered frequently overlapped between two hourly automatic measuring time points.

A recent study reported on a similar elastomeric device exhibiting frequent periodic cessation of flow possibly related to kinking of the tubing at the site of the tubing clamp [7]. Many of the devices in the current study had been clamped for a period of several days prior to testing; however, periods of flow interruption were not observed with any device in this study.

Elastomeric disposable pumps consist of an elastomeric membrane contained within an outer protective shell. The outer protective shell is either a conformable elastomer (e.g. On-Q Pain Buster® with On Demand®) or a rigid plastic (e.g. Infusor Pump; Baxter Healthcare, Deerfield, IL, USA). Elastomeric pump membranes consist of various elastomers, both natural and synthetic (e.g. latex, silicon and isoprene rubber), and are composed of single or multiple layers. The type of elastomer and the shape of the elastomeric balloon determine the pressure generated on the fluid when the balloon is stretched. Some manufacturers pre-stretch the balloon reservoir to facilitate filling (e.g. Accufuser; Woo Young Medical Co, Jincheon-Gun, Chungcheonbuk-Do, Korea). Because of these design variations, the results of this study are valid only for the specific pump we tested. Although it would be reasonable to expect similar results from pumps with the same elastomer composition (e.g. On-Q C-Bloc; I-Flow, Lake Forrest, CA, USA), the results of this study cannot be used as justification for refilling other elastomeric devices. Similar testing should also be undertaken on these devices. Finally, caution should be exercised with similar pumps delivering a higher background infusion as the relatively high infusion rate delivered during the first hour as seen in this study, may have implications for potential local anaesthetic toxicity.

Although we have demonstrated the accuracy and potential safety of a refilled elastomeric pump for continuous regional anaesthesia, this study does not address two issues with continuous peripheral nerve blockade: contamination of sterile infusate and risk of infective complications arising from prolonged peripheral nerve catheter placement. A recent case report [10] of mediastinitis due to Staphylococcus aureus infection following continuous interscalene block highlighted the importance of asepsis not only for catheter insertion but also whilst preparing local anaesthetic solutions. It is worth noting that breaks in the local anaesthetic circuit/giving set that would occur if a new pump were substituted, may also increase the risk of infection via this route. Although colonisation of peripheral nerve catheters is likely to be a relatively common occurrence [11], abscess or serious infective complications at the site of insertion are exceedingly rare [11–16] with only a few case reports described in the literature over the last decade [17, 18], all without permanent sequelae. Consideration must still be given to the risks and benefits of prolonging these techniques beyond 72 h.

In summary, we have demonstrated that the infusion characteristics of a refilled elastomeric device commonly used for continuous peripheral nerve block is not only safe, but also performs predictably within the limits specified by the manufacturer for the initial use. Our data also confirms the accuracy of the bolus device over and above the background infusion. Increased infusion rates during the first hour were identified in a small number of pumps. We therefore recommend that patient administration of pump local anaesthetic be avoided during this first hour.


The authors thank the Auckland District Health Board Charitable Trust for providing financial support. MJF has received research support from the I-Flow Corporation. This company had no involvement in any aspect of this study including but not limited to study funding, conceptualisation, design, data collection, analysis or preparation of the manuscript.