Secondary Logo

Journal Logo

Small-Volume Injections: Evaluation of Volume Administration Deviation From Intended Injection Volumes

Muffly, Matthew K. MD*; Chen, Michael I. MD*; Claure, Rebecca E. MD*; Drover, David R. MD*; Efron, Bradley PhD; Fitch, William L. PhD; Hammer, Gregory B. MD*

doi: 10.1213/ANE.0000000000001976
Technology, Computing, and Simulation: Original Laboratory Research Report
Free

BACKGROUND: In the perioperative period, anesthesiologists and postanesthesia care unit (PACU) nurses routinely prepare and administer small-volume IV injections, yet the accuracy of delivered medication volumes in this setting has not been described. In this ex vivo study, we sought to characterize the degree to which small-volume injections (≤0.5 mL) deviated from the intended injection volumes among a group of pediatric anesthesiologists and pediatric postanesthesia care unit (PACU) nurses. We hypothesized that as the intended injection volumes decreased, the deviation from those intended injection volumes would increase.

METHODS: Ten attending pediatric anesthesiologists and 10 pediatric PACU nurses each performed a series of 10 injections into a simulated patient IV setup. Practitioners used separate 1-mL tuberculin syringes with removable 18-gauge needles (Becton-Dickinson & Company, Franklin Lakes, NJ) to aspirate 5 different volumes (0.025, 0.05, 0.1, 0.25, and 0.5 mL) of 0.25 mM Lucifer Yellow (LY) fluorescent dye constituted in saline (Sigma Aldrich, St. Louis, MO) from a rubber-stoppered vial. Each participant then injected the specified volume of LY fluorescent dye via a 3-way stopcock into IV tubing with free-flowing 0.9% sodium chloride (10 mL/min). The injected volume of LY fluorescent dye and 0.9% sodium chloride then drained into a collection vial for laboratory analysis. Microplate fluorescence wavelength detection (Infinite M1000; Tecan, Mannedorf, Switzerland) was used to measure the fluorescence of the collected fluid. Administered injection volumes were calculated based on the fluorescence of the collected fluid using a calibration curve of known LY volumes and associated fluorescence.

To determine whether deviation of the administered volumes from the intended injection volumes increased at lower injection volumes, we compared the proportional injection volume error (loge [administered volume/intended volume]) for each of the 5 injection volumes using a linear regression model. Analysis of variance was used to determine whether the absolute log proportional error differed by the intended injection volume. Interindividual and intraindividual deviation from the intended injection volume was also characterized.

RESULTS: As the intended injection volumes decreased, the absolute log proportional injection volume error increased (analysis of variance, P < .0018). The exploratory analysis revealed no significant difference in the standard deviations of the log proportional errors for injection volumes between physicians and pediatric PACU nurses; however, the difference in absolute bias was significantly higher for nurses with a 2-sided significance of P = .03.

CONCLUSIONS: Clinically significant dose variation occurs when injecting volumes ≤0.5 mL. Administering small volumes of medications may result in unintended medication administration errors.

Published ahead of print March 23, 2017.

From the *Department of Anesthesiology, Perioperative and Pain Medicine, Stanford University Medical Center, Stanford, California; Department of Statistics, Stanford University, Stanford, California; and Division of Biomaterials and Advanced Drug Delivery, Stanford University Medical Center, Stanford, California.

Accepted for publication January 6, 2017.

Published ahead of print March 23, 2017.

Funding: None.

The authors declare no conflicts of interest.

Reprints will not be available from the authors.

Address correspondence to Matthew K. Muffly, MD, Stanford University Medical Center, 300 Pasteur Dr, H3580, Stanford, CA 94305. Address e-mail to mmuffly@stanford.edu.

Perioperative medication errors may result in serious patient harm and are underreported.1,2 One important but underappreciated cause of medication errors is dose inaccuracy with small-volume injections.3–8 The accurate and precise administration of IV medication volumes is a patient safety priority. In the perioperative period, anesthesiologists and postanesthesia care unit (PACU) nurses routinely prepare and administer small-volume IV injections, yet the accuracy of medication volume administration in this setting has not been described.

In this ex vivo study, we sought to characterize the degree to which administered injection volumes deviated from intended injection volumes at and below 0.5 mL among a group of pediatric anesthesiologists and pediatric PACU nurses. We hypothesized that as the intended injection volumes decreased, the proportional deviation from those intended injection volumes would increase. The primary outcome was deviation of the administered volume from the intended injection volume. Secondarily, we characterized interindividual and intraindividual deviation from the intended injection volume.

Back to Top | Article Outline

METHODS

Ten attending pediatric anesthesiologists and 10 pediatric PACU nurses participated in this Stanford institutional review board-approved, single-institution study at Lucile Packard Children’s Hospital, a 302-bed tertiary care children’s hospital in Palo Alto, California. Each practitioner performed a series of 10 injections into a simulated patient IV setup (Figure 1).

Figure 1.

Figure 1.

Practitioners used separate 1-mL tuberculin syringes with removable 18-gauge needles (Ref. #309628; Becton-Dickinson & Company, Franklin Lakes, NJ) to aspirate 5 different volumes (0.025, 0.05, 0.1, 0.25, and 0.5 mL) of 0.25 mM Lucifer Yellow (LY) fluorescent dye constituted in saline (Sigma Aldrich, St. Louis, MO) from a rubber-stoppered vial. Each participant then injected the specified volume of LY fluorescent dye via a 3-way stopcock into IV tubing with free-flowing 0.9% sodium chloride (10 mL/min). The injected volume of LY fluorescent dye and 0.9% sodium chloride then drained into a collection vial for laboratory analysis and all collection volumes were normalized to 10 mL. Nurse and physician injection technique was not standardized; practitioners were instructed to inject the specified volume in accordance with their usual clinical practice. Practitioners injected all 5 volumes in duplicate (10 injections per practitioner) using a new syringe, needle, 3-way stopcock, IV tubing, and collection vial for each of the 10 volumes. The order in which practitioners injected each volume was random, determined by selecting injection volume cards from a shuffled deck.

Microplate fluorescence wavelength detection (Infinite M1000; Tecan, Mannedorf, Switzerland) was used to measure the fluorescence of the collected fluid because this method is highly accurate to within ±1 nm.9,10Administered injection volumes were calculated based on the fluorescence of the collected fluid using a calibration curve of known LY volumes and associated fluorescence, according to the principles described by Harris.11

Back to Top | Article Outline

Statistical Methods

The proportional injection volume error of the average of each practitioner’s administered volume replicates was calculated using the Loge (administered volume/intended volume). Calculating the proportional injection volume error on the log scale has 2 advantages. First, symmetry is maintained between injection volume administrations over and under the intended injection volume. For example, if the administered volume is less than the intended volume, the ratio will be between 0 and 1, whereas if the administered volume is greater than the intended volume, the ratio can have any value over 1. The log of this ratio, on the other hand, is equal for under- and overinjection. The second reason to use the log scale is that groups that have larger values (ie, higher injection volume groups) tend to have greater within-group variability. Using a logarithmic transformation makes the within-group variability similar across injection volume groups.

We used a linear regression model to determine if the proportional injection volume error differed with the intended injection volume. The absolute proportional injection volume error was used because volumes above and below the intended injection volume represent potentially clinically significant injection volume errors. A P < .01 was considered strong evidence against the null hypothesis that no difference in deviation from the intended injection volume between groups existed. The median, robust standard deviation, and bootstrap estimate of the standard error of the median (1000 bootstraps) were reported for the log proportional error at each intended injection volume.12 A version of the robust standard deviation was used to account for outliers in the administered injection volumes and was calculated by dividing the interdecile range by 2.56, the value of the interdecile range for a normal distribution.

To characterize the interindividual deviation from the intended injection volume, the robust mean and robust standard deviation of the log proportional errors for each practitioner’s 10 injections were calculated and plotted (robust mean and robust standard deviation values of 0 represent the ideal).13 The robust mean refers to the standard 10% trimmed mean in which the most extreme 10% of the sample from each tail is removed, and the mean of the remaining 80% is reported. The robust standard deviations of the log proportional injection volume errors for physicians relative to PACU nurses were also compared in an exploratory analysis using the Wilcoxon 2-sample test. For a comparison statistic, the ratio (r) of the Loge ([nurse standard deviation]/[physician standard deviation]) is reported along with the 95% bootstrap confidence interval for r.

To characterize the intraindividual deviation from the intended injection volume, we assessed pairwise injection volume differences, which reflect injection volume repeatability by a single practitioner. To do this, the difference in the Loge (administered volume/intended volume) of the 2-injection volume repetitions for each practitioner was used to construct a normal Q-Q plot of the proportional injection volume error. The normal Q–Q, or quantile–quantile, plot is created by graphing the quantiles of the experimental values on the y-axis and a normal quantile distribution on the x-axis. If the experimental values are normally distributed, the points approximate a diagonal straight line. If the points fall along a diagonal straight line in the middle of the graph, but curve off in the extremities, the data have more extreme values than would be expected if they truly came from a normal distribution, meaning that the injection volume differences between a practitioner’s first and second injections were greater than expected.14

Additionally, we modeled the proportional injection volume error at each intended injection volume group if the volume of the syringe hub (0.04 mL) was administered in addition to the intended injection volume (Figure 2). This reflects the scenario in which a practitioner administers contents of the syringe hub in addition to the intended injection volume.

Figure 2.

Figure 2.

We elected to use an injection volume deviation of 10% from the intended injection volume to represent the minimum threshold for a clinically significant error. This value is based on the US Pharmacopeia standard that the concentration of IV infusions must not vary by >10% of the stated concentration; a 10% variation in the stated concentration would result in an equivalent volume administration error.7,15,16 Because no preliminary data were available to conduct a power analysis, we included the maximal number of practitioners given our available resources and estimated the maximal number of injections per practitioner that we felt would be tolerated. To determine the statistical power that we had to detect a 10% deviation from the intended injection volume in each injection volume group, we calculated the loge proportional injection volume error for a 10% deviation from the intended injection volume. We then ran a 2-sample t test of the mean difference for each of the 5 injection volumes to determine the statistical power available to detect a 10% difference and an α level of .05 given the sample size and standard error of the median of the proportional injection volume error in each injection volume group.

Back to Top | Article Outline

RESULTS

As the intended injection volumes decreased, the proportional injection volume error increased significantly (analysis of variance, P < .0018; Figure 3). The median, robust standard deviation, and bootstrap estimate of the standard error of the median (1000 bootstraps) for the log proportional errors are reported in Table 1. For 3 injections, the injection volume was >3 standard deviations from the mean and likely represents practitioner misinterpretation of the prescribed injection volume.

Table 1.

Table 1.

Figure 3.

Figure 3.

Figure 4 plots the robust mean and robust standard deviation of the proportional injection volume error for each practitioner’s 10 injections. A group of 6 practitioners come close to the ideal with a robust mean log proportional error of 0 and robust standard deviation of 0 (represented by the blue asterisk). Two practitioners’ (at the upper left of the chart) injection volumes exhibited large divergence from the ideal both in terms of the robust mean of the proportional injection volume and robust standard deviation. The group of 4 practitioners near the center of the graph represents relatively moderate divergence from the ideal robust mean and robust standard deviation of 0. There was no significant difference in the standard deviations of the log proportional errors for injection volumes between physicians and PACU nurses (Wilcoxon 2-sample test); the ratio (r) of the loge ([nurse standard deviation])/[physician standard deviation]) was 0.36 signifying that the robust standard deviation among nurses is approximately 36% larger than the robust standard deviation among physicians (95% bootstrap confidence interval, −0.365 to 1.018). The difference in the absolute bias of the robust standard deviation, however, was significantly higher for nurses with a 2-sided significance of P = .03.

Figure 4.

Figure 4.

From the pairwise differences of the 2-injection volume repetitions for each practitioner, the distribution on the normal Q–Q plot follows a normal distribution near the center; however, a long tail of more errant values extends in both directions and indicates that the injection volume differences between some practitioners’ first and second injections were greater than expected (Figure 5).

Figure 5.

Figure 5.

The model of the proportional injection volume error at each intended injection volume group when a fixed volume (0.04 mL) is administered in addition to the intended injection volume is shown in Figure 6.

Figure 6.

Figure 6.

Injection volume groups 0.05, 0.1, 0.25, and 0.5 mL had >90% power to detect a difference of 10% from the intended injection volume. The smallest injection volume group did not have the power to detect a 10% difference from the intended injection volume, likely as a result of the large standard error of the median in the log proportional injection volume error in this group. Although we did not have the statistical power to detect the 10% clinical difference at the lowest injection volume group, the difference was so large that the smallest injection volume group did have >90% power to detect the 38% difference that we saw.

Back to Top | Article Outline

DISCUSSION

In this study, we demonstrate the significant injection volume error that may occur when pediatric anesthesiologists and pediatric PACU nurses administer injection volumes ≤0.5 mL into IV systems. Strong evidence exists that as intended injection volumes decrease, deviation from the intended injection volume increases significantly. One in 4 practitioners administered over twice the intended injection volume at 0.025 mL and nearly twice the intended injection volume at 0.05 mL. Although deviation from the intended injection volume was most marked in the lowest injection volume groups, substantial under- and overdoses occurred in all groups. These findings are interesting because they suggest that clinically significant medication errors likely occur frequently in the perioperative setting with small-volume injections.

These findings are consistent with previous studies that describe dose variation with small-volume injections in other settings. Keith et al6 found that insulin syringes were “highly inaccurate” at doses <5 units (0.05 mL). Casella et al3 recommended that insulin doses of 2 units or less be administered in a dilute form, because doses of 0.02 mL result in “unacceptably large” errors. Several authors describe a human bias toward overadministration of small-volume injections using syringes, also consistent with our findings.3,4 In the intensive care unit setting, wide deviation between the intended and actual concentrations of prepared opioid and catecholamine infusions has been reported, likely a result of preparing IV infusions from small volumes of concentrated medications.7,8,17 Lastly, in the preparation of spinal anesthetic mixtures, Dull and Peterfreund18 report that technical factors contribute to large variations in the final concentrations of the drugs added to the solution. The finding that clinically significant variation with small-volume IV injections likely occurs in the perioperative setting among both physicians and nurses adds to the existing literature.

An unexpected but potentially important finding in this study was the high degree of practitioner injection technique variation. The authors witnessed injection technique variation from practitioner to practitioner but also variation in individual practitioner technique from injection to injection. Variation in practitioner injection technique may explain the deviation of administered volumes from intended injection volumes. Witnessed injection technique variation included: (1) connection of syringe to stopcock, open to “patient,” and injection of volume (often with residual LY visible in the stopcock hub); (2) connection of syringe to stopcock, open to IV, aspiration of additional IV fluid into the syringe, then stopcock open to “patient” and injection of volume; (3) injection of the syringe contents into a separate 0.9% saline flush syringe, then connection of flush syringe to the stopcock and injection to “patient”; (4) after aspirating LY, aspiration of additional 0.9% saline from a separate flush syringe, injection of the LY, and 0.9% saline contents back into flush syringe, then connection of flush syringe to stopcock and injection to “patient”; and (5) first changing needle on the syringe, then technique 3 or 4.

Injection technique variation is important because the volume of medication in the syringe hub and needle are not included in the syringe volume markings. Administration of additional volume from the syringe hub and/or needle will result in inadvertent overadministration. The volume of the tuberculin syringe hub used for this study measures 0.04 mL. Administration of the syringe hub contents (0.04 mL), in addition to the intended volume, will result in volume overadministration, proportionally larger at smaller injection volumes as shown in Figure 6. For example, when a hypothetical 10-kg patient receives 0.005 mg/kg of hydromorphone (0.05 mg) from a 2-mg/mL concentration vial, the administered volume should be 0.025 mL. If the volume of the syringe hub (0.04 mL) was given in addition to the intended volume (0.025 mL), the administered dose would be 0.065 mg, nearly a 3-fold overdose. Administration of the contents of the syringe hub (0.04 mL) and attached needle (0.05 mL) will result in even higher injection volumes/doses. As injection volumes increase, the deviation that results from injecting the hub and/or needle contents becomes proportionally less. The safety implications of these findings, especially during transitions of care, are important to consider as medication doses may vary significantly, depending on who prepares and administers that medication. This is a particularly relevant issue in the perioperative period, given the potential number of handoffs that occur intraoperatively, to the postanesthesia unit or to the intensive care unit.

An effective strategy to reduce volume administration deviation with small-volume injections will likely include multiple interventions. Education for practitioners who administer IV medications to increase awareness of this issue is necessary; however, high-reliability interventions that improve systems of medication administration are also needed.19 The use of less concentrated preparations of medications would reduce the need to administer volumes within the range of dramatic variation. Whenever possible, hospital pharmacies should consider replacing medication vials with less concentrated vials, if available, or prefilled syringes of dilute medication. With dilute preparations, however, a greater volume of fluid is administered for each injection and may result in substantial fluid volume in very small patients.

This study has several limitations. First, the technique used to measure injection volumes was dependent on measurement of fluorescence concentrations, creation of a calibration curve, and accurate measurement of collected fluid volume. All reasonable steps were taken to ensure accurate measurements, but small variation may be the result of laboratory measurement error. To calculate the median proportional injection volume error for each practitioner, we averaged each practitioner’s two-injection volume repetitions at each intended injection volume, which may conceal extreme variation between injection volume repetitions. Practitioners may have altered their behavior during the observations. If this did indeed occur, the effect would likely be a reduction in deviation from the intended injection volume; therefore, true deviation from the intended injection volume in clinical practice may actually be higher than these results suggest. Lastly, practitioner injection techniques were not controlled, and the study was not powered to determine optimal injection technique. Despite these limitations, we are confident that this work represents an improvement in our understanding of our ability to administer small IV fluid volumes (≤0.5 mL) accurately.

Moving forward, an increased awareness of this important issue and limiting the use of small-volume injections will likely improve medication safety for patients in the perioperative setting.

Back to Top | Article Outline

ACKNOWLEDGMENTS

The authors gratefully acknowledge the input of Margaret Hynds, RN, and Susi Delagnes, RN, in the planning and completion of this research.

Back to Top | Article Outline

DISCLOSURES

Name: Matthew K. Muffly, MD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript

Name: Michael I. Chen, MD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Name: Rebecca E. Claure, MD.

Contribution: This author helped write the manuscript.

Name: David R. Drover, MD.

Contribution: This author helped analyze the data and write the manuscript.

Name: Bradley Efron, PhD.

Contribution: This author helped analyze the data and write the manuscript.

Name: William L. Fitch, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Name: Gregory B. Hammer, MD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

This manuscript was handled by: Maxime Cannesson, MD, PhD.

Back to Top | Article Outline

REFERENCES

1. Merry AF, Webster CS, Hannam J, et al. Multimodal system designed to reduce errors in recording and administration of drugs in anaesthesia: prospective randomised clinical evaluation. BMJ. 2011;343:d5543.
2. Nanji KC, Patel A, Shaikh S, Seger DL, Bates DW. Evaluation of perioperative medication errors and adverse drug events. Anesthesiology. 2016;124:25–34.
3. Casella SJ, Mongilio MK, Plotnick LP, Hesterberg MP, Long CA. Accuracy and precision of low-dose insulin administration. Pediatrics. 1993;91:1155–1157.
4. Gnanalingham MG, Newland P, Smith CP. Accuracy and reproducibility of low dose insulin administration using pen-injectors and syringes. Arch Dis Child. 1998;79:59–62.
5. Lteif AN, Schwenk WF. Accuracy of pen injectors versus insulin syringes in children with type 1 diabetes. Diabetes Care. 1999;22:137–140
6. Keith K, Nicholson D, Rogers D. Accuracy and precision of low-dose insulin administration using syringes, pen injectors, and a pump. Clin Pediatr (Phila). 2004;43:69–74.
7. Aguado-Lorenzo V, Weeks K, Tunstell P, Turnock K, Watts T, Arenas-Lopez S. Accuracy of the concentration of morphine infusions prepared for patients in a neonatal intensive care unit. Arch Dis Child. 2013;98:975–979.
8. Parshuram CS, Ng GY, Ho TK, et al. Discrepancies between ordered and delivered concentrations of opiate infusions in critical care. Crit Care Med. 2003;31:2483–2487.
9. Hanani M. Lucifer yellow—an angel rather than the devil. J Cell Mol Med. 2012;16:22–31.
11. Harris DC. Quantitative Chemical Analysis. 2003.San Francisco, CA: W.H. Freeman.
12. Efron B. Bootstrap Methods: Another Look at the Jackknife. Ann Stat. 1979;7:26.
13. Huber PJ. Ronchetti E. Robust Statistics. 2009.2nd ed. Hoboken, NJ: Wiley
15. Wheeler DW, Degnan BA, Sehmi JS, Burnstein RM, Menon DK, Gupta AK. Variability in the concentrations of intravenous drug infusions prepared in a critical care unit. Intensive Care Med. 2008;34:1441–1447.
16. The United States Pharmacopeial Convention. 1999:Rockville, MD1131–1132.
17. Allen EM, Van Boerum DH, Olsen AF, Dean JM. Difference between the measured and ordered dose of catecholamine infusions. Ann Pharmacother. 1995;29:1095–1100.
18. Dull RO, Peterfreund RA. Variations in the composition of spinal anesthetic solutions: the effects of drug addition order and preparation methods. Anesth Analg. 1998;87:1326–1330.
19. Cafazzo JA, St-Cyr O. From discovery to design: the evolution of human factors in healthcare. Healthc Q. 2012;15:24–29.
Copyright © 2017 International Anesthesia Research Society