Secondary Logo

Journal Logo


The Delivery Rate Accuracy of Portable Infusion Pumps Used for Continuous Regional Analgesia

Ilfeld, Brian M. MD*,; Morey, Timothy E. MD*,; Enneking, F. Kayser MD*†

Author Information
doi: 10.1097/00000539-200211000-00043
  • Free


Ansbro first provided continuous regional anesthesia more than 50 years ago (1). Since then, multiple techniques for local anesthetic delivery have been described, most of which used a polyamide catheter through which local anesthetic was infused. Initially, large, heavy, technically sophisticated infusion pumps were used. Subsequently, many of the procedures amenable to these techniques were moved to an outpatient setting. In the past decade, lightweight, portable pumps designed to infuse narcotics (2) or antibiotics (3) in ambulatory patients were introduced. Several investigators later adapted these portable pumps for regional analgesia. In 1998, Rawal et al. (4) first described using a portable pump to infuse local anesthetic for postoperative analgesia after day surgery. After this report, portable pumps of various designs were described to provide perineural (5–8), wound (9), and intraarticular infusions (10). Although these techniques seem to be gaining acceptance and usage for an increasing number of ambulatory patients, the infusion rate accuracy and reliability of the infusion pumps have not been independently investigated.

We have had ambulatory patients using various portable pumps exhaust their local anesthetic reservoir after 50% to 150% of the expected infusion duration. This is consistent with the experience of other investigators (8). No apparent increase in anesthetic morbidity resulted from these highly variable infusion rates, yet the quality of analgesia lacked consistency, and infusion duration was unpredictable. We were concerned about possible local anesthetic toxicity in patients if infusion rates remained irregular. Many of these pumps regulate the infusion flow rate by using a temperature-dependent device calibrated to skin temperature. Therefore, we performed this laboratory study to define the flow rate accuracy, reliability, and profiles of various portable infusion pumps at expected and higher-than-expected temperatures.


Pump Selection

Six small, lightweight, portable pumps marketed for local anesthetic infusion were selected for testing (Table 1, Fig. 1). The energy source to dispense anesthetic varies between pumps. These sources include elastic- or spring-based, pressure gradient (e.g., vacuum pump), and electrical (e.g., electronic pump) energy. At least one representative from each “category” of infusion drives was selected. All of these pumps are available in various reservoir volume and infusion rate combinations. The largest volume available and the rate most closely approximating 5.0 mL/h were selected for each infusion pump (Table 2). All pumps tested were previously unused and were filled to their recommended capacity with normal saline (NS) immediately before testing. New batteries were inserted in the Microject PCA, the only electronic pump tested, before each infusion.

Table 1:
Infusion Pump Manufacturers
Figure 1:
Portable infusion pump configurations as investigated: Accufuser (A), Sgarlato (B), Pain Pump (C), MedFlo II (D), C-Bloc (E), and Microject PCA (F). Manufacturer information is included in Table 1.
Table 2:
Infusion Pump Attributes (as Tested)

Infusate Selection

Infusate viscosity may influence the infusion flow rate of various pumps. Only McKinley Medical included information with their pumps about the effect that various infusates would have on its pump, stating that the Accufuser was calibrated by using 5% dextrose in water, and that the use of NS would increase the flow rate by 10%. Ropivacaine is one of the most commonly used local anesthetics described in continuous regional anesthetic techniques (5,7,8,10). Because viscosity data for ropivacaine are not available (personal communication with AstraZeneca LP, Wilmington, DE, October 2001), a trial at 32°C using the Accufuser was performed first with ropivacaine, then followed by NS. It demonstrated identical flow profiles for ropivacaine and NS, validating our extrapolation of this data obtained with NS infusions to those involving ropivacaine.

Study Apparatus

To record the infusion rate profiles, the following apparatus was used (Fig. 2). The infusion pump was first attached to a polyamide multiport catheter (B. Braun Medical, Bethlehem, PA). For all pumps, with the exception of the Microject PCA, the flow rate regulator found near the connection between the pump tubing and catheter attachment was placed within an adjustable heating unit (Microplate Incubator; Boekel Scientific, Feasterville, PA). An empty plastic bottle with a removable plastic cap was used to collect the dispensed fluid. One hole in the plastic cap was made with a 19-gauge needle, and the distal end of the catheter was inserted through this hole into the collection bottle.

Figure 2:
The system used to test the infusion flow rates of various portable pumps.

The collection bottle was placed on an electronic scale (Navigator Balance; Ohaus Corp., Florham Park, NJ) that was placed on the same surface as the infusion pumps. Once a test infusion began, data were logged onto an IBM-compatible personal computer (Dimension XPS 400; Dell Computer Corp., Round Rock, TX) by using an RS-232 cable. A software program (Software Wedge; TAL Technologies, Philadelphia, PA) provided entry from the serial port directly into the spreadsheet program (Excel 2000; Microsoft Corp., Seattle, WA). Mass of the infusate was measured every minute over the duration of the infusion by using the tared value of the bottle. The infusion period ended when each pump had exhausted its fluid reservoir. Because the Microject PCA pump does not have a predetermined maximal reservoir volume, as do the other pumps tested, a 1000-mL bag of NS was connected to the pump that was tested for 60 h. Subsequently, the hourly infusion rate was calculated by subtracting the mass at a given hour (Mx) from that obtained after the next hour (M[x+1]). That is, the hourly infusate volume equaled (M[x+1] − Mx), and the rate equaled this volume divided by 60 min.

Although the scale manufacturer reports accuracy of ±0.1 g, we were concerned about time-related drift and possible infusate evaporation over the duration of these experiments. To test for potential evaporation loss and scale drift, 100 g of NS was placed in the collection bottle with the catheter in place for 2 wk. Measurements were taken each minute, and the loss to evaporation was <0.1 g (0.1%) over the testing period. Scale drift over the 2 wk was ±0.4 g (0.4%). The ambient room air temperature was held between 20°–24°C (68°–75°F) during the entire study period. A temperature-monitoring device (Hobo H8; Onset Computer Corp., Bourne, MA) recorded ambient temperature every 5 min during the entire study period to ensure a uniform room temperature for the infusion pumps. Based on these data, we conclude that the apparatus was appropriate to test pump performance over the duration of at least 60 h.

Each infusion pump was tested with the flow rate regulator placed in the heating unit. The temperature of the heating unit was set at the temperature that the manufacturer reported to be skin temperature (the baseline, or expected, temperature). For example, the Accufuser was calibrated by the manufacturer for a flow rate of 5.0 mL/h at 32°C, whereas the I-Flow was calibrated at 31°C. If a manufacturer-recommended temperature was not included with the infusion pump, then 31°C was used.

Each test was performed twice with a new infusion pump unit. If the infusion rate during the second trial differed more than ±10% of the original trial at any point, a third trial was performed. The trials were combined to produce a mean profile for each pump at the baseline (expected) temperature. After this, all pumps (except the Microject PCA) were tested again by using the same protocol, but with the heating unit set 4°C more than the baseline temperature. The Microject PCA infusion rate is controlled electronically and is relatively temperature independent over a small-scale temperature change (e.g., 4°C).

Infusion duration (measured) was considered to end when the measured flow rate decreased <50% of the set rate. Data were reported as mean ± sd. Overall comparisons were made by using analysis of variance on ranks with post hoc Tukey pairwise testing, if appropriate. P < 0.05 was considered to be statistically significant.


The infusion rate profiles for the six portable pumps tested are illustrated in Figure 3.

Figure 3:
Pump performance over time for several portable infusion pumps as noted by title in each panel. Shown is actual infusion rate as a fraction of the set infusion rate. These set infusion rates were: Accufuser (5.0 mL/h), Sgarlato (4.0 mL/h), Stryker (4.16 mL/h), C-Bloc (5.0 mL/h), MedFlo II (5.0 mL/h), and Microject PCA (5.0 mL/h). All pumps, except the Microject PCA, were tested at 30°–32°C (open circles) to simulate normal skin temperature, and 34°–36°C (closed circles) to simulate a 4°C increase above normal temperature. The constant horizontal line represents the expected pump rate at 100% of set flow for 60 h. The constant vertical line represents the expected infusion duration as calculated from the set rate and reservoir volume, except for the Microject PCA, which possesses a variable reservoir amount. Axes labels apply to all panels. Data are expressed as mean ± sd. Representative error bars are inscribed for every fourth data point.


Of the 6 pumps tested, only the MedFlo II had an infusion rate differing >10% between the first 2 trials at each temperature, requiring a third trial. The first and second trials at each temperature differed <10% at any point during the infusion for the Accufuser and Pain Pump units. The first and second trials at each temperature differed <5% at any point during the infusion for the C-Bloc and Microject PCA pumps.


At their “expected” operating temperature, the pumps infused at a rate within ±15% of their set rate to differing degrees (Table 3). The Microject PCA, Accufuser, and C-Bloc pumps infused within this range 100%, 90%, and 86% of their infusion duration, respectively. The MedFlo II, Sgarlato, and Pain Pumps infused within this range 70%, 57%, and 18% of their infusion duration, respectively.

Table 3:
Measured Pump Performance Compared with Ideal Infusion

Flow Profile

To various degrees, all three of the elastomeric pumps (Accufuser, C-Bloc, MedFlo II) infused at a rate faster than expected initially, and then returned closer to their set rates for much of the infusion (Fig. 3). This increased flow rate resulted in a decreased overall infusion duration for the Accufuser and C-Bloc pumps (Table 3). The spring-powered Sgarlato pump had a similar initial period of rapid infusion, but its rate consistently declined and fell below its expected value after approximately 12 h. This resulted in an increased infusion duration of >15%. In contrast, the infusion duration for the vacuum-powered Pain Pump was decreased by >15% as a result of its consistently fast infusion rate for the entire infusion duration. For its entire duration, the electronic Microject PCA pump infused at a slower rate than it was programmed for, and this rate progressively decreased over time (from 5% to 15% below expected).

Temperature Effects

Temperature markedly affected overall pump performance (P = 0.05). Increasing the temperature of the flow rate regulator 4°C affected each pump differently (Fig. 3). With the temperature change, the infusion rates of the MedFlo II and C-Bloc pumps increased between 10% and 33%. This increase in rate resulted in a decreased infusion duration of approximately 25% for each (Table 3). The Accufuser and Pain Pump infusion rates also increased with an increase in temperature, although they were <10%. The infusion rate of the Sgarlato pump was not consistently affected by a change in temperature.


This investigation demonstrates that the infusion rate consistency and accuracy of portable pumps often used to provide postoperative continuous regional analgesia are variable (Table 3). Factors such as pump power source and ambient temperature impact pump infusion rate (Fig. 3). Both the elastomeric- and spring-powered pumps infused at faster-than-expected rates initially, with infusion rates decreasing over the infusion duration. The vacuum pump had consistently faster-than-expected infusion rates whereas the electronic pump infused at a consistently slower-than-expected rate. Increasing flow-regulator temperature increased infusion rates by >10% in 2 of the elastomeric pumps. The faster-than-expected infusion rates led to a decreased total infusion duration (Table 3).


These differences in flow rate accuracy may have significant implications for patient care when applied to continuous regional analgesia. Although there is nothing inherently wrong with an infusion rate that varies over time as these pumps provide, health care providers must be aware of the infusion profile to maximize patient safety and benefit. There are potential advantages and disadvantages to all of the infusion profiles described in this study, and the pump profile must be adequately matched to the situation/indication. For example, because surgical pain generally decreases over time, a pump that provides a declining rate of infusion may be appropriate for an adult patient receiving a perineural local anesthetic infusion for postoperative analgesia. However, this profile may provide a subtherapeutic infusion rate during the latter portion of the infusion period, and, consequently, unsatisfactory analgesia. When choosing the proper infusion pump for a given application, several factors must be accounted for, including, but not limited to, acceptable flow rate accuracy, desired infusion duration, and total local anesthetic volume requirement.

For many of the pumps described in this investigation, temperature influenced the rate of infusion (Table 3, Fig. 3). For the C-Bloc and MedFlo II pumps, an increase of 4°C resulted in an increased flow rate of >10%, whereas the Accufuser and Pain Pump were affected to a lesser extent. Although the Sgarlato pump uses an infusion-regulating device similar in appearance and placement to these other pumps, the manufacturer states that it is temperature independent. This was confirmed in our investigation.

The degree to which the temperature sensitivity of a given pump should influence a decision regarding its use is highly situation dependent. For example, our institution is located in Florida where summer temperatures often reach 39°C, increasing skin and ambient temperature, which affects the flow regulators of various infusion pumps. After this study, we instructed our patients to remain in air-conditioned environments when using a temperature-sensitive infusion pump during the summer months. This investigation only varied temperature with an increase of 4°C. A larger increase should theoretically increase the flow rate more than reported here, whereas a temperature decrease should theoretically result in a flow rate decrease. However, this speculation is based on the physics of the flow regulator technology and requires additional investigation for confirmation.

Battery Charge

Another variable that may influence the infusion rate of electronic pumps is battery power. The continuous decline of the Microject PCA’s infusion rate (a total of approximately 10%) throughout the entire 60-hour test duration may be evidence of this phenomenon. Information included with this pump states that the “battery life” using 2 AA alkaline 1.5-volt batteries is 10 days at a flow rate of 9.9 mL/h. If and when the batteries should be replaced during an infusion deserve further study.

Pump Choice

Although the electronic pump described in this investigation had the most accurate and consistent flow rate of the pumps studied, there are other factors that should be considered when choosing an optimal pump for a given indication. These include patient and health care provider convenience, reliability, cost, ease of use, as well as the clinical factors mentioned previously. For example, for an intraarticular local anesthetic infusion at a desired rate of 2 mL/h in an adult, it may be desirable to use one of the nonelectronic pumps for its simplicity and disposability. In this case, a change of 10%–20% in the flow rate may not be clinically significant, and the other factors that will help determine the optimal device may outweigh infusion rate accuracy and consistency. Whereas it is beyond the scope of this discussion to comment on every variable, health care providers should take all of these into account when choosing an infusion pump for a specific circumstance.

Study Limitation

We included only the infusion rate regulators in the heating unit. This was to simulate average skin temperature for the regulating devices because these are usually taped to the patient’s skin, whereas the infusion pump itself is left at ambient room temperature. To simulate a temperature increase of the flow rate regulator, we increased the temperature of the heating device, but the ambient room temperature of the infusion pumps remained constant. Therefore, it is unknown what the effect of increasing the temperature for both the flow regulator and the pump would have on the flow rate (this includes the Microject PCA).

In conclusion, portable pumps often used for local anesthetic infusion during continuous regional analgesia exhibit varying degrees of delivery rate accuracy and consistency. Furthermore, increases in temperature result in an increased infusion rate of various degrees for many of the infusion pumps investigated. Healthcare providers should take these factors into consideration when choosing and using a portable infusion pump for local anesthetic administration. Controlled clinical studies are needed to investigate how local anesthetic infusion rate variability affects patient analgesia.

The authors thank Jenny Kline Ilfeld, MD, for her valuable editorial contributions.


1. Ansbro FP. A method of continuous brachial plexus block. Am J Surg 1946; 71: 716–22.
2. O’Keefe D, O’Herlihy C, Gross Y, Kelly JG. Patient-controlled analgesia using a miniature electrochemically driven infusion pump. Br J Anaesth 1994; 73: 843–6.
3. Rich DS. Evaluation of a disposable, elastomeric infusion device in the home environment. Am J Hosp Pharm 1992; 49: 1712–6.
4. Rawal N, Axelsson K, Hylander J, et al. Postoperative patient-controlled local anesthetic administration at home. Anesth Analg 1998; 86: 86–9.
5. Klein SM, Greengrass RA, Gleason DH, et al. Major ambulatory surgery with continuous regional anesthesia and a disposable infusion pump. Anesthesiology 1999; 91: 563–5.
6. Klein SM, Grant SA, Greengrass RA, et al. Interscalene brachial plexus block with a continuous catheter insertion system and a disposable infusion pump. Anesth Analg 2000; 91: 1473–8.
7. Ilfeld BM, Enneking FK. A portable mechanical pump providing over four days of patient-controlled analgesia by perineural infusion at home. Reg Anesth Pain Med 2002; 27: 100–4.
8. Ganapathy S, Amendola A, Lichfield R, et al. Elastomeric pumps for ambulatory patient controlled regional analgesia. Can J Anaesth 2000; 47: 897–902.
9. Savoie FH, Field LD, Jenkins RN, et al. The pain control infusion pump for postoperative pain control in shoulder surgery. Arthroscopy 2000; 16: 339–42.
10. Klein SM, Nielsen KC, Martin A, et al. Interscalene brachial plexus block with continuous intraarticular infusion of ropivacaine. Anesth Analg 2001; 93: 601–5.
© 2002 International Anesthesia Research Society